1

Ultrasound studies of the

endometrial-myometrial
junction for the diagnosis of
adenomyosis and endometrial
cancer
Joel Naftalin

The Gynaecology Diagnostic and Outpatient Treatment Unit, Department of Obstetrics and
Gynaecology, Elizabeth Garrett Anderson Wing, University College Hospital, Grafton Way, London
WC1H

Thesis submitted for the degree of MD(Res), University College London

March 2014
 2

STATEMENT OF ORIGINALITY AND PERSONAL CONTRIBUTION TO WORK

I, Joel Naftalin, confirm that the work presented in this thesis is my own. Where information

has been derived from other sources, I confirm that this has been indicated in the thesis. I was

personally involved in the design of all studies, applications for ethical approval and

recruitment of all patients. I performed all ultrasound examinations except where indicated,

and collected all the data. All statistical analyses were done, either by me or by UCL

statistician Paul Bassett.

_____________

Joel Naftalin
 3

ABSTRACT

This thesis investigated the ultrasound assessment of the endometrial-myometrial junction

(EMJ) and its clinical relevance to the uterine diseases, adenomyosis and endometrial cancer.

The inter- and intraobserver variability in the classification of EMJ visualisation using three-

dimensional ultrasound was assessed and a high level of agreement was found. Endometrial

thickness and parity were found to be significantly associated with the quality of EMJ

visualisation.

Seven recognised ultrasound features of adenomyosis, including an irregular endometrial-

myometrial junction, were investigated for their role in the ultrasound diagnosis of

adenomyosis. The other features were an asymmetrically thickened myometrium, parallel

shadowing, linear striations, myometrial cysts, hyperechoic lesions and adenomyomas. The

presence of any of these features was considered diagnostic of adenomyosis. The inter- and

intraobserver variability of ultrasound diagnosis of adenomyosis was also investigated and a

good level of agreement was found. This was the case when real-time ultrasound assessments

were compared with assessments made from stored uterine volumes, as well as when both

assessments were made from stored volumes.

Transvaginal ultrasound was used to assess the prevalence of adenomyosis in women

attending a general gynaecology clinic in a large prospective observational study. Women

were considered to have adenomyosis if one or more ultrasound feature of adenomyosis was

found. Using this criterion, the prevalence was estimated to be 20.9% with 7.6% of women

being excluded from the data analysis. It was also found that age, gravidity and pelvic

endometriosis were all significantly associated with the presence of adenomyosis.

 4

Menorrhagia was evaluated in order to assess if it was associated with adenomyosis.

Multivariable analysis revealed that while adenomyosis was not significantly associated with

menorrhagia when assessed as a binary outcome, when severity of disease was taken into

account, there was a significant association. A similar analysis found that adenomyosis was

significantly associated with dysmenorrhoea.

A second-stage ultrasound test that incorporated assessment of the EMJ was investigated for

its use in the diagnosis of endometrial cancer in women presenting with post-menopausal

bleeding. It was found to significantly increase the specificity of ultrasound in the diagnosis

of endometrial cancer while having a minimal impact on sensitivity.

 5

ACKNOWLEDGEMENTS

First and foremost I would like to thank my supervisor Davor Jurkovic. None of this thesis

would have been possible without his ideas, support and guidance. He is a true inspiration

and someone I feel immensely privileged to have been able to work with.

I would also like to thank Naaila Aslam and Rehan Salim for their clinical support during my

time as a research fellow. Thanks also to Professor Stephenson for her research guidance.

I would like to thank my research fellow colleagues during my time at UCH: Will Hoo,

Michelle Swer, Amna Jamil, Natalie Nunes and Katie Pateman for their practical help and

support. Similarly I would like to thank past research fellows Dimitri Mavrelos, Tom

Holland, Joseph Yazbek and Jara Ben–Nagi for being unfailingly helpful whenever called

upon.

Jacinta O’Neill, sister in charge of the GDOTU provided invaluable craic. And thanks to all

the other staff in the GDOTU: Ogechi, Sarah, Ronke, Lisa, Nancy, Leilani, Sian, Gody, Fumi

and Margaret for assorted chaperoning, appointment scheduling, help with flushings and for

keeping me entertained and well-fed.

To my wonderful parents, I owe everything to you.

Sam, Lyla & Bert – You are the loves of my life. Sam, I could not have done this without

your enormous support. I’m sorry for neglecting you. I hope it’s worth it.
 6

TABLE OF CONTENTS

Page

Statement of originality and personal contribution to work 2

Abstract 3

Acknowledgements 5

Table of contents 6

List of figures 12

List of tables 15

Hypothesis 19

Aims 20

PART I Background 21

1.0 TECHNICAL ASPECTS OF ULTRASOUND 22

1.1 Introduction 22

1.2 Principles 22

1.3 Image formation 23

1.3.1 B-mode image formation 23

1.3.2 Transvaginal ultrasound 23

1.4 Doppler 23

1.4.1 The Doppler effect 23

1.4.2 Colour Doppler Imaging 25

1.4.3 Power Doppler imaging 27

1.5 Three-dimensional ultrasound 27

1.5.1 Three-dimensional ultrasound volume acquisition 28

 7

2.0 ENDOMETRIAL-MYOMETRIAL JUNCTION (EMJ) 30

2.1 Introduction 30

2.2 Embryology of the EMJ 31

2.3 Anatomy and histology of the EMJ 32

2.4 Physiology and hormonal regulation of the EMJ 32

2.5 Imaging of the EMJ 33

2.5.1 Magnetic resonance imaging 33

2.5.2 Ultrasound 35

2.6 Pathology of the EMJ 39

2.6.1 Adenomyosis 39

2.6.1.1 History of adenomyosis 39

2.6.1.2 Pathogenesis of adenomyosis 40

2.6.1.3 Diagnosis of adenomyosis 43

2.6.1.3.1 Histology 43

2.6.1.3.2 Clinical 46

2.6.1.3.3 Ultrasound 47

2.6.1.3.3.1 Diagnostic accuracy of ultrasound 47

2.6.1.3.3.2 Ultrasound features of adenomyosis 47

2.6.1.3.4 MRI 50

2.6.1.3.4.1 Diagnostic accuracy of MRI 50

2.6.1.3.4.2 MRI features of adenomyosis 50

2.6.1.3.4.3 Comparison of diagnostic accuracy of MRI and TVUS 51

2.6.1.3.5 Computed tomography 52

2.6.1.3.6 Hysterosalpingography 53

2.6.1.4 Clinical impact of adenomyosis 54

 8

2.6.1.4.1 Prevalence 54

2.6.1.4.2 Symptoms 54

2.6.1.4.2.1 Abnormal uterine bleeding 54

2.6.1.4.2.2 Pain 56

2.6.1.4.2.3 Reproductive function 56

2.6.2 Endometrial cancer 58

2.6.2.1 Introduction 58

2.6.2.2 Prevalence 58

2.6.2.3 Clinical presentation 58

2.6.2.4 Risk factors 59

2.6.2.5 Precursor lesions 60

2.6.2.6 Diagnosis of endometrial cancer 60

2.6.2.6.1 Blind endometrial biopsy 60

2.6.2.6.2 Hysteroscopy 61

2.6.2.6.3 Transvaginal ultrasound 62

2.6.2.6.3.1 Endometrial thickness 62

2.6.2.6.3.2 Endometrial thickness and morphology 63

2.6.2.6.3.3 Transvaginal ultrasound with Doppler 63

2.6.2.6.3.4 Three-dimensional ultrasound assessment of endometrial volume 64

2.6.2.7 Staging of endometrial cancer 65

3.0 Conclusion to background 66

PART II MATERIALS AND METHODS 68

1.1 Setting 69

1.1.1 The Gynaecology Diagnostic and Outpatient Treatment Unit, University College
Hospital 69
 9

1.2 Ultrasound 69

1.2.1 Two-dimensional transvaginal ultrasound 69

1.2.2 Three-dimensional transvaginal ultrasound 75

1.2.2.1 Off-line assessment of uterine volumes 75

1.3 Pictorial blood loss assessment charts 76

1.4 Numerical rating scale 77

1.5 Endometrial sampling 78

1.6 Statistical analysis 78

1.7 Ethical committee approval 78

PART III RESULTS 79

1.0 Inter- and intraobserver variability in the three-dimensional assessment of the

endometrial-myometrial junction

1.1 Background 80

1.2 Methods 80

1.3 Statistical analysis 81

1.4 Results 81

2.0 Factors affecting visualisation of the endometrial-myometrial junction on three-

dimensional ultrasound

2.1 Background 84

2.2 Methods 84

2.3 Statistical analysis 85

2.4 Results 85

3.0 Inter- and intraobserver variability in the ultrasound diagnosis of adenomyosis

3.1 Background 89

3.2 Methods 90
 10

3.3 Statistical analysis 90

3.4 Results 91

4.0 How common is adenomyosis? A prospective study of prevalence using transvaginal

ultrasound in a gynaecology clinic

4.1 Background 96

4.2 Methods 97

4.3 Statistical analysis 97

4.4 Results 98

5.0 Is adenomyosis associated with menorrhagia?

5.1 Background 104

5.2 Methods 104

5.3 Statistical analysis 105

5.4 Results 106

6.0 Is adenomyosis associated with dysmenorrhoea?

6.1 Background 115

6.2 Methods 115

6.3 Statistical analysis 116

6.4 Results 116

7.0 Evaluation of a second-stage test designed to improve the accuracy of transvaginal

ultrasound in the diagnosis of endometrial cancer

7.1 Background 123

7.2 Methods 124

7.3 Statistical analysis 125

7.4 Results 125

PART IV DISCUSSIONS 132

 11

1.0 Inter- and intraobserver variability in the three-dimensional assessment of the

endometrial-myometrial junction 133

2.0 Factors affecting visualisation of the endometrial-myometrial junction on three-

dimensional ultrasound 135

3.0 Inter- and intraobserver variability in the ultrasound diagnosis of adenomyosis 137

4.0 How common is adenomyosis? A prospective study of prevalence using transvaginal

ultrasound in a gynaecology clinic 140

5.0 Is adenomyosis associated with menorrhagia? 146

6.0 Is adenomyosis associated with dysmenorrhoea? 151

7.0 Evaluation of a second-stage test designed to improve the accuracy of transvaginal

ultrasound in the diagnosis of endometrial cancer 153

PART V CONCLUSIONS & FURTHER RESEARCH 157

REFERENCES 161

APPENDIX 180
 12

List of Figures

Figure 1 – Image capture showing B-mode ultrasound image with Doppler velocimetry of a
blood vessel within a malignant ovary (short white arrow). The graph at the bottom (long
white arrow) is the tracing showing the estimated velocity of blood flow (y-axis) through the
blood vessel over time (x-axis) 25

Figure 2a – B-mode image of uterine corpus and a pelvic mass with appearances suggestive
of a uterine fibroid
Figure 2b – B-mode image of same uterine corpus and pelvic mass, with colour Doppler
illustrating the vascular pedicle connecting the fundus of the uterus and the pedunculated
fibroid 26

Figure 3 – TVUS image showing an endometrial polyp with power Doppler clearly
delineating its ‘feeder’ vessel 27

Figure 4 – A 3D coronal view of a normal uterus 29

Figure 5 - Schematic diagram of a uterus in transverse section showing the endometrial and
myometrial layers and the EMJ (Naftalin and Jurkovic 2009) 31

Figure 6 – Sagittal T2-weighted image of a uterus showing normal zonal anatomy with a high
signal intensity endometrium (arrowhead), low-signal intensity JZ (short arrow) and
intermediate signal intensity outer myometrium (long arrow) (Novellas et al. 2011) 34

Figure 7 – Longitudinal B-mode image of a normal uterus in the proliferative phase of the
menstrual cycle. The callipers are placed on the basal endometrial layer which appears as a
regular continuous hyperechoic line. The surrounding inner myometrium appears hypoechoic
in comparison to both the outer endometrial and outer myometrial layers (Naftalin and
Jurkovic 2009) 36

Figure 8a – A 3D coronal view of a normal uterus showing measurements of the maximal JZ

thickness (callipers 1 and 4), minimal JZ thickness (calliper 3) and maximal myometrial
thickness (calliper 2) (Exacoustos et al. 2011) 38

Figure 8b – A schematic representation of Figure 8a 38

Figure 9 - Hysterosalpingogram showing irregularity of the uterine contour with small

outpouchings of contrast material. Findings consistent with diffuse adenomyosis (Simpson et
al. 2006) 53

Figure 10 – Illustration of PALM COEIN mnemonic for the causes of abnormal uterine
bleeding (Munro et al. 2011) 55

Figure 11 – Asymmetrical myometrial thickening. A longitudinal B-mode image of an

anteverted uterus in which the distance from the endometrium to the anterior serosal surface
(thick white line) is much greater than the distance from the endometrium to the posterior
serosal surface (thin white line) (Naftalin et al. 2012b) 71
 13

Figure 12 – Parallel shadowing. A longitudinal B-mode image of a retroverted uterus in

which parallel hypoechoic lines can be seen running through the anterior myometrium
(yellow s) (Naftalin et al. 2012b) 72

Figure 13 – Linear striations. A transverse B-mode image of a uterus with a linear striation,
seen as thin hyperechoic line (thin white arrow) extending from the endometrium into the
myometrium. (Naftalin et al. 2012b) 72

Figure 14 – Myometrial cysts. A transverse B-mode image of a uterus with several

myometrial cysts (thin white arrows) seen as anechoic lesions within the myometrium
(Naftalin et al. 2012b) 73

Figure 15 - Hyperechoic islands. A coronal view of a uterus where hyperechoic islands of

tissue are seen in the fundal and lateral portions of the myometrium (thin white arrows)
(Naftalin et al. 2012b) 73

Figure 16 - Adenomyomas. A longitudinal B-mode image of a retroverted uterus with an

adenomyoma in the anterior myometrium (between white arrows) (Naftalin et al. 2012b) 74

Figure 17 - Irregular endometrial-myometrial junction. A coronal view of a uterus with an

irregular endometrial-myometrial junction (Naftalin et al. 2012b) 74

Figure 18 – A pictorial blood loss analysis chart 77

Figure 19 – An 11-point numerical rating scale 77

Figure 20 - Coronal view of uterus with EMJ visualisation classified as optimal (Naftalin et
al. 2012a) 82

Figure 21 - Coronal view of uterus with EMJ visualisation classified as satisfactory (Naftalin
et al. 2012a) 82

Figure 22 - Coronal view of uterus with EMJ visualisation classified as unsatisfactory

(Naftalin et al. 2012a) 82

Figure 23 – Is adenomyosis a cause of menorrhagia? Flowchart showing why patients were

excluded from data analysis 108

Figure 24 – A graph illustrating the relationship between increasing numbers of ultrasound

features of adenomyosis and PBAC-assessed menstrual loss (n=304) 113

Figure 25 – Is adenomyosis a cause of dysmenorrhoea? Flowchart showing why women were

excluded from data analysis 118

Figure 26 – A graph showing the relationship between increasing numbers of ultrasound

features of adenomyosis and NRS score (n=304) 121
 14

Figure 27 – A longitudinal B-mode image of a uterus with a heterogenous endometrium. The

patient was subsequently diagnosed with endometrial cancer 127

Figure 28 – Coronal view of a uterus where an endometrial cancer can be seen breaching the
EMJ at the right cornuum 128

Figure 29 – Longitudinal B-mode image of the uterine cavity containing an endometrial

polyp. Colour Doppler shows the polyp’s single feeder vessel. The polyp was found to be
benign 128

Figure 30 – Longitudinal view of a uterus using B-mode ultrasound with colour Doppler in a
woman with endometrial cancer. Multiple vessels can be seen crossing the EMJ at the fundus
(white arrows) 129

Figure 31 – Flowchart illustrating risks of endometrial cancer depending on results of

transvaginal ultrasound 131
 15

List of tables

Table 1 – Ultrasound features that aid differentiation of fibroids and adenomyomas on

transvaginal ultrasound 49

Table 2 – Studies comparing the diagnostic accuracy of TVUS and MRI for the diagnosis of
adenomyosis (Dueholm 2006) 52

Table 3 – Risk factors and protective factors for endometrial cancer (Saso et al. 2011) 59

Table 4 – Revised FIGO staging of endometrial cancer (Creasman et al. 2009) 65

Table 5 – Ultrasound features of adenomyosis used in this thesis (Naftalin et al. 2012b) 71

Table 6 - Inter-observer variability in the assessment of visualisation of the EMJ (Naftalin et

al. 2012a) 83

Table 7 - Intra-observer variability (Operator 1) in the assessment of visualisation of the EMJ

(Naftalin et al. 2012a) 83

Table 8 - Intra-observer variability (Operator 2) in the assessment of visualisation of the EMJ

(Naftalin et al. 2012a) 83

Table 9 - Principal indications for examination of women included in study evaluating factors
affecting EMJ visualisation (Naftalin et al. 2012a) 87

Table 10 – Result of Spearman’s rank correlation in assessing the effect of age, parity and
endometrial thickness on EMJ visualisation (Naftalin et al. 2012a) 87

Table 11 – Result of Kruskal-Wallis test in assessing the effect of stage of menstrual cycle on
EMJ visualisation (Naftalin et al. 2012a) 87

Table 12 – Result of Mann-Whitney test in assessing the effect of menopausal status on EMJ
visualisation (Naftalin et al. 2012a) 88

Table 13 – Results of multivariable analysis assessing the joint effect of the various factors
on EMJ visualisation (Naftalin et al. 2012a) 88

Table 14 - Principal indications for examination of women included in study evaluating inter
and intra-observer variability in the ultrasound diagnosis of adenomyosis 92

Table 15 - Number of ultrasound features of adenomyosis recorded on original real-time

scans 92

Table 16 - Kappa analysis of intra-observer agreement in diagnosis of adenomyosis on real-

time ultrasound scan and on stored uterine volumes (both operator 1) 92
 16

Table 17 - Linear-weighted kappa analysis of intra-observer agreement in the number of

ultrasound features seen on real-time ultrasound scan and on stored uterine volumes (both
operator 1) 93

Table 18 - Kappa analysis of inter-observer agreement in ultrasound diagnosis of

adenomyosis on stored uterine volumes (operators 1 and 2) 93

Table 19 a - Linear-weighted kappa analysis of inter-observer agreement in the number of

ultrasound features seen on stored uterine volumes (operator 1 and 2) 93

Table 19b - Kappa values for inter-observer agreement of the presence of each individual
ultrasound feature of adenomyosis (operator 1 and 2) 94

Table 20 - Linear-weighted kappa analysis of inter-observer agreement in the severity of

adenomyosis on stored uterine volumes (operator 1 and 2) 94

Table 21 - Kappa analysis of intra-observer agreement in ultrasound diagnosis of

adenomyosis on stored uterine volumes (both operator 1) 94

Table 22 - Linear-weighted kappa analysis of intra-observer agreement in the number of

ultrasound features seen on stored uterine volumes (operator 1) 95

Table 23 - Linear-weighted kappa analysis of intra-observer agreement on the severity of

adenomyosis on stored uterine volumes (both operator 1) 95

Table 24 – Principal indications for examination of women included in study assessing the
prevalence of adenomyosis (Naftalin et al. 2012b) 100

Table 25 - Summary of ultrasound diagnoses of the women included in study assessing the
prevalence of adenomyosis (Naftalin et al. 2012b) 100

Table 26 - Concomitant abnormalities found in women diagnosed with adenomyosis in

prevalence study (n=206) (Naftalin et al. 2012b) 101

Table 27 – Results of univariable analysis looking at associations between demographic and

clinical factors and adenomyosis (Naftalin et al. 2012b) 102

Table 28 - Results of multivariate analysis looking at associations between demographic and

clinical variables and adenomyosis (Naftalin et al. 2012b) 103

Table 29 – Principal indications for examination of women included in study assessing

whether adenomyosis is associated with menorrhagia (n=714) 108

Table 30 - Summary of ultrasound diagnoses recorded in the women included in study

assessing whether adenomyosis is associated with menorrhagia (n=714) 108

Table 31 - Concomitant uterine abnormalities found in women diagnosed with adenomyosis

in the study assessing whether adenomyosis is associated with menorrhagia with
adenomyosis (n=157) 109
 17

Table 32 – Kappa analysis assessing agreement between subjective assessment of

menorrhagia and PBAC charts (where PBAC score >100 considered to be consistent with
menorrhagia) 109

Table 33 – Results of univariable analysis looking at associations between demographic and

clinical factors and the subjective assessment of menorrhagia (n=714) 110

Table 34 – Results of multivariable analysis looking at associations between demographic

and clinical factors and the subjective assessment of menorrhagia (n=714) 111

Table 35 – Results of multivariable analyses looking at the associations between

demographic and clinical factors and the subjective assessment of menorrhagia, where the
number of ultrasound features of adenomyosis was treated as a continuous variable (Model 1)
and where the number of ultrasound features was treated as a categorical variable (Model 2)
(n=714) 112

Table 36 –The number of ultrasound features of adenomyosis seen in the women who were
diagnosed with adenomyosis in the study assessing whether adenomyosis was associated with
menorrhagia (n=157) 113

Table 37 – Results of the statistical analysis looking at the effect of the number and type of
ultrasound features on the PBAC assessment of menorrhagia (n=304) 114

Table 38 – Principal indications for examination of women included in study assessing

whether adenomyosis is associated with dysmenorrhoea (n=714) 119

Table 39 - Summary of ultrasound diagnoses recorded in the women included in the study
assessing whether adenomyosis is associated with dysmenorrhoea (n=714) 119

Table 40 - Concomitant uterine abnormalities seen in women diagnosed with adenomyosis

(n=157) 119

Table 41 - Results of univariable analysis looking at associations between demographic and

clinical factors and the subjective assessment of dysmenorrhoea (n=714) 120

Table 42 - Results of multivariable analysis looking at associations between demographic and

clinical factors and the subjective assessment of dysmenorrhoea (n=714) 121

Table 43 - Results of the statistical analysis looking at the effect of the number and type of
ultrasound features on the subjective assessment of dysmenorrhoea (n=304) 122

Table 44 – The number of ultrasound features of adenomyosis seen in the study population
(n=157) 122

Table 45 – Demographics of women included and excluded from the endometrial cancer
study 130
 18

Table 46 – Table showing how the endometria of women subsequently diagnosed with
endometrial cancer were classified at the time of their original ultrasound scan 131
 19

HYPOTHESIS

This thesis will investigate the following hypotheses:

 Three-dimensional transvaginal ultrasound can be used as an accurate and

reproducible method for assessment of the endometrial-myometrial junction.

 Transvaginal ultrasound is an accurate and reproducible imaging modality for the

diagnosis of adenomyosis and for assessment of its severity

 Adenomyosis is a common condition

 Adenomyosis is associated with both menorrhagia and dysmenorrhoea

 Ultrasound assessment of the endometrial-myometrial junction can be used to

improve the accuracy of transvaginal ultrasound in the diagnosis of endometrial

cancer
 20

AIMS

The aims of this thesis are:

 To examine the reproducibility of three-dimensional ultrasound assessment of the

endometrial-myometrial junction

 To examine the reproducibility of ultrasound diagnosis of adenomyosis using three-

dimensional stored uterine volumes

 To determine the prevalence of ultrasound features of adenomyosis in a population of

women attending a general gynaecology clinic, using transvaginal ultrasound.

 To determine if adenomyosis is associated with menorrhagia and dysmenorrhoea

 To evaluate a second-stage test, incorporating assessment of the endometrial-

myometrial junction, designed to improve the accuracy of transvaginal ultrasound in

the diagnosis of endometrial cancer

 21

PART I BACKGROUND
 22

1.0 TECHNICAL ASPECTS OF ULTRASOUND

1.1 Introduction

Ultrasound has transformed many aspects of gynaecological practice. Prior to its use in

medicine, diagnosis of gynaecological conditions was based almost entirely on clinical

examination, surgical and histological findings. The first published clinical use of ultrasound

was in the field of gynaecology when Professor Ian Donald and his registrar John McVicar

used transabdominal ultrasound to differentiate between solid and cystic abdominal masses

(Donald et al. 1958). Professor Donald combined his clinical experience and his knowledge

of SONAR, gleaned during World War II, with the technical expertise of Tom Brown, an

engineer with an interest in the use of ultrasound in metallurgy, to create and use the first

clinically useful ultrasound machine. The ability to provide instant, clinically useful

information using a safe, non-invasive modality was transformational and has revolutionised

gynaecological practice. There have been many significant technical advances, most recently

three-dimensional (3D) ultrasound, yet it is still B-mode ultrasound, first used by Donald and

his team over 50 years ago, that is the primary imaging tool in gynaecology today.

1.2 Principles

Medical ultrasound is based on the principle of passing a current through a piezoelectric

crystal to create ultrasound pulses, which can then be transmitted through the relevant body

tissue. When these pulses encounter an interface between tissues of differing acoustic

impedance or density, a proportion of the emitted soundwaves will be reflected back towards

the piezolelectric element from which they were emitted (echoes). The echoes are then

converted into an electric current, the amount of current produced being dependent on the

number of echoes received. The conversion of this current into an image therefore gives a

visual representation of the varying densities within the body tissue.

 23

1.3 Image formation

1.3.1 Brightness mode image formation

A brightness-mode or B-mode image is a cross-sectional image representing tissues and

organ-boundaries within the body. The image is constructed from echoes generated by the

reflection of ultrasound waves at tissue boundaries. Each echo is displayed at a point in the

image which corresponds to the position of its origin in the tissue being isonated. The

brightness of the image at each point is related to the strength or amplitude of the echo,

giving rise to the term brightness mode or B-mode (Martin 2003).

1.3.2 Transvaginal ultrasound

The frequency of ultrasound pulses used is a compromise between image resolution and the

depth of penetration required. Higher ultrasound frequency results in better image resolution

but there is greater attenuation of the beam within the tissues. Therefore, transabdominal

ultrasound probes, which have to pass through multiple tissue layers before reaching the

abdominal cavity, tend to use lower frequencies (3.5-5MHz) at the expense of image

resolution. The ability of transvaginal ultrasound (TVUS) probes to get much closer to the

organs of interest, enables the use of higher frequencies (8-15MHz) leading to improved

image resolution.

1.4 Doppler

1.4.1 The Doppler effect

The Doppler effect is the apparent difference between the frequency at which waves (sound

or light) leave a source and that at which they reach an observer, caused by the relative

motion of the observer and the wave source (McNay and Fleming 1999). The effect takes its

name from the Austrian physicist Christian Johann Doppler who presented the idea in 1842
 24

(Roguin 2002).When the object emitting the waves is stationary the observed frequency is the

same as the emitted frequency, however if the sound source is moving towards the observer,

the experienced frequency is higher as the sound waves become more compressed and the

opposite happens if the sound source is moving away from the observer. This change in

frequency is called the Doppler shift and is proportional to the relative velocity of the source

to the observer.

This effect can be applied clinically and is widely utilised in both obstetric and

gynaecological ultrasound in order to assess the velocity of blood flow through blood vessels.

When the ultrasound beam is reflected off moving blood, there are two Doppler shifts, one

when the transmitted ultrasound strikes circulating blood cells and a second when circulating

blood cells emit the reflected ultrasound. Evaluation of these Doppler shifts, alongside

knowledge of the transmitted ultrasound frequency, the velocity of sound through the tissue

and the angle of insonation, allow the calculation of the velocity of blood passing through the

vessel (Hoskins et al. 2003). This estimated velocity of blood flow over time can be shown on

the ultrasound machine as a tracing (Figure 1).

 25

Figure 1 – Image capture showing B-mode ultrasound image with Doppler velocimetry of a
blood vessel within a malignant ovary (short white arrow). The graph at the bottom (long
white arrow) is the tracing showing the estimated velocity of blood flow through the blood
vessel (y-axis) over time (x-axis).

1.4.2. Colour Doppler Imaging

In colour Doppler scanning, the same process is applied across an area of tissue rather than a

specific blood vessel. The velocity signals are presented as a colour coded overlay,

superimposed on the real-time B-mode image. This allows production of an angiogram-like

map that provides information on the morphological arrangement of the vascular tree in the

tissue of interest. While its sensitivity is good enough to enable visualisation of vessels

smaller than one millimetre (De Souza and Cosgrove 2003), it is restricted by its reliance on

frequency shifts.
 26

Figure 2a –B-mode image of uterine corpus (long white arrow) and a pelvic mass (short
white arrow) with appearances suggestive of a uterine fibroid.

Figure 2b – B-mode image of same uterine corpus and pelvic mass, with colour Doppler
illustrating the vascular pedicle connecting the uterus and the pedunculated fibroid.
 27

1.4.3 Power Doppler Imaging

Power Doppler has greater sensitivity than colour Doppler imaging, because the display

depends on the amplitude of Doppler signal rather than the frequency shift. Thus, power

Doppler imaging provides information on the concentration of moving blood at the expense

of knowing its velocity and direction. In addition to being more sensitive than colour Doppler

imaging, it is relatively angle independent, more accurate in depicting luminal edges and

better in visualising the continuity of flow (Rubin 1999).

Figure 3 – TVUS image showing an endometrial polyp with its ‘feeder’ vessel delineated by
power Doppler.

1.5 Three-dimensional ultrasound

3D ultrasound is a relatively recent diagnostic modality, which allows detailed evaluation of

pelvic organs by collecting a series of sequential ultrasound images and converting them into

an ultrasound volume. This information is digitally stored as a dataset which is reconstructed

in such a way as to allow visualisation of an organ from any chosen angle and in any arbitrary
 28

plane (Jurkovic 2002). In the field of obstetrics and gynaecology 3D ultrasound has so far had

the greatest impact on the diagnosis of uterine abnormalities.

1.5.1 Three–dimensional ultrasound volume acquisition

3D ultrasound involves the acquisition of a series of consecutive, parallel B-mode images by

sequential movement of either the transducer or the ultrasound beam. This series of images is

then stored in the computer as a volume set which can be examined using three methods:

section reconstruction, surface rendering and volume rendering. The 3D ultrasound

equipment used in this thesis is designed by Kretz Technik, Zipf, Austria (Voluson E8). This

commonly used system combines both short volume acquisition time and fast computing,

enabling almost immediate online data analysis and near real-time surface rendering. Volume

acquisition with this system is performed using mechanical movement of the ultrasound

transducer within the transvaginal probe.

An important aspect of 3D ultrasound is the ability to display and manipulate any chosen

section from within the dataset. This confers many advantages over B-mode imaging. Firstly,

it allows the acquisition of views that are not possible with conventional two-dimensional

sonography, largely as a result of restrictions on probe movement imposed by pelvic anatomy

(Baba et al. 1997). The example of this most relevant to this thesis is the coronal view of the

normal female uterus (Figure 4). Secondly, the ability to view the organ in three orthogonal

planes enables a more detailed analysis of the organ to be performed. Finally, the storage of

images for manipulation enables examination of the volume set by multiple operators at

different times.
 29

Figure 4 – A 3D coronal view of a normal uterus.

 30

2.0 ENDOMETRIAL-MYOMETRIAL JUNCTION (EMJ)

2.1 Introduction

Assessment of the uterus has historically focussed on what were considered to be its two

constituent parts: the endometrium and the myometrium. Magnetic resonance imaging (MRI)

studies in the early 1980s revealed that the myometrium might be further divisible into two

distinct compartments: the inner and outer myometrium (Hricak et al. 1983). Despite being

macroscopically and microscopically indistinct, subsequent research has shown these two

compartments of the myometrium to have quite different embryology, anatomy and

physiology. The inner myometrium has been variously termed the uterine junctional zone

(JZ) (Fusi et al. 2006), the endometrial-subendometrial unit or stratum sub-vasculare (Noe et

al. 1999), the subendometrial myometrium (Lyons et al. 1991) the archimetra (Leyendecker et

al. 1998) or archimyometrium (Leyendecker et al. 2006). While these terms are often used

interchangeably it is not always clear whether they are referring to exactly the same entity.

For the purposes of this thesis, this region will be referred to as the EMJ. It is the transitional

zone between the mucous membrane that is the endometrium and the outer smooth muscle

layer of the myometrium (Figure 5).

 31

Figure 5 - Schematic diagram of a uterus in transverse section showing the endometrial and
myometrial layers and the EMJ (Naftalin and Jurkovic 2009).

2.2 Embryology of the EMJ

The endometrium and the EMJ both arise from the paramesonephric ducts, whereas the outer

myometrium is of mesenchymal origin (Daels 1974; Brosens et al. 1998) The

paramesonephric ducts are formed at about six weeks’ gestation from the coelomic

invagination of mesodermal cells at the level of the third thoracic somite (Koff 1933;

O’Rahilly 1977). The caudal portions of these tubes make contact with each other and fuse in

the midline, giving rise to the uterus and vaginal canal. The septum that initially divides the

uterine cavity is then reabsorbed leaving a single cavity. The endometrial glands originate at

around 19 weeks’ gestation from out-pouchings of the columnar epithelium that lines the

primitive uterine cavity. The smooth muscle cells of the EMJ can be seen from around 21

weeks, while the outer layers of the myometrium develop in the third trimester or even

postnatally (Noe et al. 1998).

 32

2.3 Anatomy and histology of the EMJ

The myometrium consists of bundles of smooth muscle fibres, intermixed with areolar tissue,

blood vessels, lymphatic vessels, and nerves. The muscle fibres of the EMJ have

predominantly circular orientation, in contrast to the longitudinal smooth-muscle orientation

of the outer layers of myometrium (Wetzstein and Renn 1970). The endometrium, the

mucous membrane that lines the uterine cavity, is composed of a single layer of columnar

epithelium.

The EMJ is structurally distinct from other mucosal-muscle interfaces within the human body

in that it lacks a submucosal layer. Most tissues with a mucosa have a subjacent,

histologically recognizable submucosa that protects the underlying tissue from mucosal

invasion, e.g., stomach, intestine, trachea, and bronchi (Marcus 1961, Emge 1962). While the

superficial and basal layers of the endometrium are clearly distinguishable on light

microscopy, there is no histological distinction between the inner myometrium and the outer

myometrium on light microscopy. Tetlow et al. demonstrated increased vascularity of the

EMJ when compared with the rest of the myometrium and also found that the muscle fibres

of this zone were more densely packed than in other zones of the myometrium. They

concluded that these architectural findings would account for the hypoechoic appearance of

the EMJ on both TVUS and MRI (Tetlow et al. 1999).

2.4 Physiology and hormonal regulation of the EMJ

The endometrium and EMJ do not just share an embryological origin. While they are

physiologically distinct, they are both under the cyclical influence of ovarian sex steroids.

The main function of the EMJ appears to be modulation of uterine peristalsis, an area thought

to play an increasingly significant role in fertility. Uterine contractions emanating from the
 33

EMJ were first visualised and subsequently described by Birnholz who visualised them on

transabdominal ultrasound (Birnholz 1984). EMJ contractions vary in orientation, amplitude

and frequency throughout the cycle under the influence of oestradiol and progesterone. In the

menstrual phase these contractions are fundo-cervical in direction (de Ziegler et al. 2001)

facilitating menstruation. The direction reverses during the follicular phase becoming

predominantly cervico-fundal with the amplitude and frequency of contractions increasing

significantly as ovulation approaches. There is evidence that this pattern of contractions in the

late follicular phase facilitates sperm transport (Kunz et al.1996). After ovulation there is a

decrease in overall EMJ contractility under the influence of progesterone. Ijland et al.

suggested that this may help to facilitate implantation of the developing blastocyst, while

improving its supply of oxygen and nutrients (Ijland et al.1996).

2.5 Imaging of the EMJ

Assessment of the EMJ has traditionally been part of the MRI evaluation of the uterus, with

changes in this area being a key component of the diagnosis of adenomyosis. With regard to

ultrasound, although the EMJ is visible, it has not historically played a major part in the

evaluation of uterine pathology. This may be partly due to the fact that in order to obtain clear

images of the EMJ, high resolution ultrasound equipment with 3D imaging facility is often

required, which has only been available in recent years. This may also explain why the

majority of studies addressing visualisation and the appearance of the EMJ have used MRI.

2.5.1 Magnetic Resonance Imaging

Hricak et al. first described the uterus as comprising three distinct zones when visualised on

MRI (Hricak et al 1983). MR imaging of uterine zonal anatomy is best demonstrated on T2-
 34

weighted images. The endometrium is of generally high signal intensity and is therefore

visualised as a thin white stripe. The inner myometrium or JZ, as most MRI studies refer to

the inner myometrium, is of uniformly low signal intensity while the outer myometrium is of

intermediate signal intensity (Lee et al. 1985) (Figure 6).

Figure 6 – Sagittal T2-weighted image of a uterus showing normal zonal anatomy with a high
signal intensity endometrium (arrowhead), low-signal intensity JZ (short arrow) and
intermediate signal intensity outer myometrium (long arrow) (Novellas et al. 2011).

Because of this observed contrast of the three uterine zones, MRI is very good at delineating

and measuring the size of these zones. Many studies have attempted to derive a normal range

for JZ thickness, with the upper limit now considered to be 8mm (Novellas et al. 2011). This

measurement is crucial in the MRI diagnosis of adenomyosis, which is largely dependent on

increased thickness of the JZ (Reinhold et al. 1996). It is worth noting that the thickness and

appearance of the JZ is hormone-dependent and therefore cyclical with the maximal thickness
 35

of the JZ being reached during the menstrual phase of the menstrual cycle (Fusi et al. 2006).

Furthermore, prior to menarche, during pregnancy and after the menopause, uterine zonal

anatomy is less distinct (Demas et al. 1986; Brosens et al. 1998; Willms et al. 1995) with the

JZ and the outer myometrium being poorly distinguished. This finding is reversed in

postmenopausal women by the taking of hormone replacement therapy (McCarthy et al.

1986). These effects mean that the JZ cannot be measured in a significant proportion of

women; up to 30% of premenopausal women (Bazot et al. 2001) and 50% of postmenopausal

women (Fusi et al. 2006).

2.5.3 Ultrasound

Formal ultrasound examination of the uterus includes a detailed assessment of the

morphological characteristics of the myometrium and endometrium. These two main

functional components of the uterus display different acoustic properties, which facilitates

their differentiation on the ultrasound image. On ultrasound, the inner myometrium has been

described as a hypoechoic band or ‘halo’ that regularly encircles the endometrium (Kunz et

al. 2000). While this zone is distinct from both the endometrium and the outer myometrium,

the ultrasonic delineation of uterine zonal anatomy is less clear than on MRI. Ultrasound,

however, has the ability to visualise very clearly the basal endometrial layer, which forms the

actual interface between the inner myometrium and the endometrium. On two-dimensional

ultrasound scan, the EMJ is best seen in the longitudinal section, which enables the

examination of its anterior and posterior aspects. In this view the EMJ is seen as consisting of

two distinctive structures: basal endometrium and inner myometrium. In normal uteri the

basal endometrium is seen as a continuous uninterrupted hyperechoic line and in practical

 36

terms it represents the endometrial–myometrial interface (Naftalin and Jurkovic 2009)

(Figure 7).

Figure 7 – A TVUS image showing a longitudinal view of a normal uterus in the proliferative
phase of the cycle. The callipers are placed on the basal endometrial layer which appears as a
regular continuous hyperechoic line. The surrounding inner myometrium appears hypoechoic
in comparison to both the outer endometrial and outer myometrial layers (Naftalin and
Jurkovic 2009).

On ultrasound, as with MRI, the differences in appearance of the inner and the outer

myometrium are influenced by ovarian sex steroid hormone levels. For example, prior to

menarche when the levels of the ovarian sex steroids are low, just as with MRI, the inner and

outer myometrium are indistinct on ultrasound (Newren 1997). There are, however, no

reported ultrasound studies of the cyclical changes seen within the EMJ (Naftalin and

Jurkovic 2009).

With the development of 3D ultrasound has come the ability to create a 3D reconstruction of

uterine anatomy in the coronal plane. This allows ultrasound assessment of both the lateral
 37

and fundal aspects of the EMJ, which were otherwise impossible to see clearly on standard B-

mode ultrasound imaging. The ability to see the entire lateral borders of the EMJ in a single

view has, literally, added a new dimension to the ability to visualise the EMJ on ultrasound.

Exacoustos et al. applied measurements taken routinely in the MRI assessment of

adenomyosis to ultrasound assessment of the uterus (Figures 8a & 8b). They took various

measurements of the JZ taken using a 3D coronal view of the uterus including the minimal JZ

thickness (JZmin), the maximal JZ thickness (JZmax) and the maximal myometrial thickness,

as a means of diagnosing adenomyosis (Exacoustos et al. 2011). They considered a JZmax

>8mm and a difference between the JZmin and the JZmax of >4mm to have high diagnostic

accuracy however these measurements have not been tested prospectively and furthermore,

no studies have been performed that attempt to describe a normal range for JZ thickness, as

measured by 3D ultrasound.
 38

Figure 8a – A 3D coronal view of a normal uterus showing measurements of the maximal JZ

thickness (callipers 1 and 4), minimal JZ thickness (calliper 3) and maximal myometrial
thickness (calliper 2) (Exacoustos et al. 2011)

Figure 8b – A schematic representation of Figure 8a

 39

2.6 Pathology of the EMJ

2.6.1 Adenomyosis

Adenomyosis is a condition defined by the presence of endometrial glands or stroma within

the myometrium. While there is still much that is poorly understood about the condition, it is

accepted that it often involves the EMJ, while some would consider it a disease of the EMJ.

2.6.1.1 History of adenomyosis

Determining the history of adenomyosis, as with many other aspects of the disease, is fraught

with difficulties. It remains unclear as to when it was first described with Scroen (1690),

Rokitansky (1860) and Cullen (1896) all being credited (Benagiano and Brosens 2006).

Further difficulty stems from the fact that for some time, adenomyosis and endometriosis

were considered to be the same condition. The first detailed descriptions of adenomyosis

came from Cullen in 1896 (Cullen 1896). He subsequently dedicated an entire book to the

condition, entitled ‘Adenomyoma of the uterus’ (Cullen 1908). It was Frankl who first used

the term ‘adenomyosis’ explaining “I have chosen the name of adenomyosis, which does not

suggest any inflammatory genesis as do terms like adenometritis, adenomyositis and

adenomyometritis…..We were never able to find any trace of an inflammatory infiltration,

either in the musculature or the mucosa of this region”. He also noted continuity between

adenomyosis and the endometrium, suggesting an endometrial origin to the disease (Frankl

1925). It was with Sampson’s classical description of peritoneal endometrium and the first

use of the term ‘endometriosis’ two years later, that the conditions started to be considered as

two separate entities (Sampson 1927). The modern definition of adenomyosis – ‘benign

invasion of the endometrium into the myometrium’- was provided by Bird et al. in 1972 (Bird

et al. 1972).
 40

2.6.1.2 Pathogenesis of adenomyosis

The precise pathogenesis of adenomyosis remains unknown with many theories being

proposed. The most widely held theory is that it results from invagination of endometrial

tissue, across the EMJ, into the myometrium. There are numerous histological and ultrasound

reports of endometrial tissue being seen in continuity with adenomyosis within the

myometrium (Frankl 1925; Ferenczy 1998; Naftalin and Jurkovic 2009; Verma et al. 2009)

that would appear to corroborate this theory. Other suggested mechanisms for the disease

include de novo development of endometrial tissue from mullerian remnants, although this is

thought to apply more to adenomyosis found in extra-uterine sites, such as the recto-vaginal

septum (Nisolle and Donnez 1997).

Even if endometrial invagination is the cause, there is still significant uncertainty about what

factors might trigger this invasion. As with most chronic conditions, the aetiology is likely to

be multi-factorial involving some or all of the factors described below. The ‘invasive

potential’ of endometrial tissue has been investigated as a possible cause. In-vitro studies

have shown that endometrial cells have a similar invasive potential to metastatic bladder cell

lines (Gaetje et al. 1995). Others have looked for differences between normal endometrial

glands and ‘adenomyotic’ endometrial glands found in the myometrium. Lei et al. found

increased expression of human chorionic gonadotrophin (hcg) / luteinizing hormone (LH)

receptor mRNA in endometrial glands found in foci of adenomyosis, compared with normal

endometrial glands (Lei et al. 1993). A similarly increased hcg/LH receptor expression has

been found in endometrial cancer cells when compared to normal endometrium (Lei et al.

1992), and in invasive trophoblast when compared to non-invasive trophoblast in

choriocarcinomas (Lin et al. 1994). It is possible therefore that this increased receptor

expression may be related to the potential of endometrial tissue to invade into the

myometrium and form adenomyotic foci (Ferenczy 1998).

 41

Myometrial characteristics may have a role in this endometrial invasion. The EMJ is unusual

in lacking a sub-mucosal layer. Most muscle-mucosa interfaces within the body, having a

distinct subjacent histologically recognizable sub-mucosal layer that protects the underlying

tissue from invasion (Marcus 1961, Emge 1962). The absence of this layer might explain why

invasion can occur, but does not explain why it happens in particular areas of the

myometrium but not others.

Endometrium may cross the EMJ at areas of myometrial ‘weakness’. A few theories of how

myometrial ‘weakness’ might occur have been proposed. Surgery provides a mechanical

model for this. Many studies have looked for an association between adenomyosis and

uterine surgeries, including caesarean section, uterine curettage and termination of

pregnancy. All these operations involve operating across or close to the EMJ and could easily

create defects in the EMJ or myometrium. Indeed, in the case of Caesarean section, it would

be impossible to perform the operation without breaching the EMJ. Defects in the

myometrium following Caesarean section are clearly visible on ultrasound and have been

reported in the literature (Ofili-Yebovi et al. 2008). Studies, however, have not consistently

found an association between Caesarean delivery and adenomyosis (Bergholt et al. 2001;

Harris et al. 1985). Uterine curettage might create defects in the myometrium through which

endometrial tissue might subsequently pass or could even push endometrial tissue through the

EMJ and into the myometrium at the time of surgery. Adenomyosis has been elicited in

pregnant rabbits by curetting one uterine horn and leaving the pregnancy in the other horn

(Lewinski 1931). In human studies, several authors (Levgur et al. 2000; Vavilis et al. 1997;

Curtis et al. 2002) have found an association between adenomyosis and previous surgical

termination of pregnancy. Uterine softening in early pregnancy (Munsick 1985) might

increase the risk of EMJ trauma, perhaps explaining the association reported between
 42

adenomyosis and uterine instrumentation in pregnancy, but not outside of pregnancy (Levgur

et al. 2000; Curtis et al. 2002).

Since adenomyosis is more common in parous women, mechanical events at parturition have

been considered to be the main pathogenenic factors for the development of adenomyosis

(Kitawaki 2006). However there are few data to support that it is parturition rather than

pregnancy per se that is the cause. Certainly the physiological process of trophoblast

invasion, common to all pregnancies, involves invasion up to the EMJ. MR images during a

conception cycle have shown focal changes within the EMJ (Turnbull et al. 1995) and it is

known that trophoblast invasion is able to progress beyond the EMJ, albeit pathologically, in

cases of placenta accreta, increta and percreta. Perhaps the process of trophoblast invasion, in

some women, alters the EMJ leaving it susceptible to adenomyosis in the future.

Uterine autotrauma, mediated by hyperperistalsis of the inner myometrium, has been

suggested as a further mechanism by which both adenomyosis and endometriosis might

spontaneously form. This theory holds that chronic peristalsis or periods of hyperperistalsis in

the inner myometrium cause EMJ microtrauma and that a cycle of permanent

hyperperistalsis, mediated by increased oestrogen production, leads to overt autotrauma. This

combination of hyperperistalsis and overt trauma to the EMJ would physically force basal

endometrial tissue into the myometrium (Leyendecker et al. 2009) causing adenomyosis. The

increased presence of oestrogen plays a key role in this hypothesis. Adenomyosis has been

found to be associated with oestrogen-dependent conditions including uterine fibroids (Azziz

1989), endometriosis (Kunz et al. 2000), and endometrial hyperplasia (Bergholt et al. 2001).

It also appears to be most prevalent in women of reproductive age and regresses after the

menopause (Kitawaki 2006). This has led some researchers to investigate the role of

oestrogen in the pathogenesis of adenomyosis. It has been reported that oestrogen receptors

are always found in adenomyotic tissue (Tamaya et al. 1979). Other investigators have found
 43

that adenomyotic tissue contains aromatase, an enzyme that catalyses the conversion of

androgens to oestrogens, suggesting that local oestrogen production within adenomyotic

tissue may contribute to overall oestrogen concentrations (Urabe et al. 1989). Further

evidence of this local production of oestrogen comes from a study that looked at oestrogen

levels in both peripheral and menstrual blood in women with and without adenomyosis.

While there was no difference in the oestrogen levels in peripheral blood, there were

significantly increased oestrogen levels in the menstrual blood of women with adenomyosis

(Takahashi et al. 1989).

A genetic predisposition to adenomyosis has been suggested from early reports suggesting

heredity (Emge 1962; Arnold et al. 1995) however few studies have investigated the genetics

of adenomyosis. Pandis et al. found a common chromosomal abnormality [7] [q21.q231.2] in

three cases of adenomyosis (Pandis et al. 1995). Similarly Kitawaki found that the PP

genotype is less frequently observed in women with endometriosis and adenomyosis when

compared with women with neither condition (Kitawaki et al. 2001). Wang et al. 2002 found

no chromosomal gain or loss in 25 cases of adenomyosis using comparative genomic

hybridization (Wang et al. 2002).

2.6.1.3 Diagnosis of adenomyosis

2.6.1.3.1 Histology

The histological diagnosis of adenomyosis requires endometrial glands or stroma to be found

beyond the EMJ and up until relatively recently, histopathological examination of the uterus

was the only way to diagnose the condition. While recent advances in imaging make an

accurate, non-invasive diagnosis possible, almost all the published studies of the diagnostic

accuracy of these imaging techniques have used histology as the gold standard. There are,
 44

however, substantial disadvantages to the continued use of histology for the diagnosis of

adenomyosis.

There remains significant disagreement as to the depth of infiltration required for the

diagnostic criteria to be met. It has been suggested that units of microscopic fields should be

used for this depth, however reported depths vary and include half a low-power field of view

(Kurman 1994), one low-power field of view (Rosai 1989), one medium-power field of view

(Gompel and Silverberg 1985) and one high-power field of view (Entmann 1988).

Furthermore, there is significant variation in the size of microscopic fields when different

microscopes are used (Ellis and Whitehead 1981) leading to the suggestion that a different

measure should be used. The depth of infiltration as a proportion of the total uterine wall

thickness has been proposed with depths greater than 25% (Ferenczy 1998) and greater than

one third of the total thickness being used in the literature (Hendrickson & Kempson 1990),

however uterine wall thickness is not uniform throughout the uterus, particularly in the

presence of adenomyosis. Other studies have looked at specific depths of infiltration.

Bergholt et al. re-examined hysterectomy specimens and applied differing depths of

infiltration of endometrial glands into the myometrium (>1mm, >3mm & >5mm) for the

diagnosis of adenomyosis and showed that the reported prevalence varied from 11.5%–18.2%

(Bergholt et al. 2001). Further variation was found depending on whether or not the presence

of myometrial hyperplasia was used as a diagnostic criterion. This followed on from the

study by Bird et al that showed that the reported prevalence of adenomyosis varied depending

on the number of slices of histological tissue examined (Bird et al. 1972).

Histopathological grading systems for severity of adenomyosis have been proposed. Bird &

Molitor both used a histological system to grade the severity of adenomyosis based on depth

of infiltration, whereby infiltration of endometrial tissue to the inner myometrium only was

graded ‘slight’, infiltration to the middle third of the myometrium was termed ‘moderate’ and
 45

infiltration to the outer third was termed ‘extensive’ (Bird et al. 1972; Molitor 1971) . Bird et

al. also used a histological system to grade the degree of myometrial involvement. Here, 1-3

endometrial glands per low-power field was graded ‘slight’, 4-9 endometrial glands per low-

power field was graded ‘moderate’ and 10 or more endometrial glands per low-power field

was termed ‘marked’ (Bird et al. 1972). While these grading systems add an extra level of

subtlety to the histological diagnosis of adenomyosis, they have not been prospectively

evaluated, nor have their inter- and intra-observer variability been evaluated.

This variation in diagnostic criteria in the histological diagnosis of adenomyosis poses

problems not only in the diagnosis of the disease, but also in the investigation of the disease.

Most studies looking at associations with the condition have compared the prevalence of the

condition in different hysterectomized populations eg women with a history of menorrhagia

and those without a history of menorrhagia, and then looked for statistical significance.

Analysis and comparison of different studies is difficult when different diagnostic criteria are

being used, making it difficult to draw firm conclusions about the condition when studied in

this way. Even where standardised diagnostic criteria are used it has been suggested that

awareness of the condition amongst individual pathologists could confer a further source of

variance with the frequency of diagnosis of adenomyosis varying from 12-58% among 15

different hospitals, and from 10-88% among 25 different histopathologists (Seidman and

Kjerulff 1996). One explanation for this might be that histopathologists may see but not

report adenomyosis unless specifically requested to do so, particularly if it is not relevant to

the primary condition. This phenomenon may be exacerbated by the fact that once diagnosed

post-hysterectomy, the condition has already been cured and so a diagnosis of adenomyosis

will not influence future clinical management. This links to another fundamental problem

with the use of histological examination of hysterectomy specimens to diagnose

adenomyosis, which is that it precludes assessment of the success of conservative treatments

 46

for the condition. This drawback of histological diagnosis does not apply to non-invasive

diagnostic techniques, which must therefore be used for assessments of the effectiveness of

conservative treatments for the condition.

Finally, as histology remains the gold standard for diagnosis, much of our current knowledge

of the condition comes from histological examination of the entire uterus. There is great

national, international and temporal variation in the use of, and indications for, hysterectomy

(Bergholt et al. 2001; Reid and Mukri 2005). Furthermore, populations of women undergoing

hysterectomy in different regions are likely to be heterogenous and hence subject to

significant demographic variation. All of these factors, directly linked to the use of histology

as gold standard for diagnosis, explain both the significant variation in reported prevalences

of adenomyosis, as well as the lack of consensus in the literature with regard to its clinical

impact.

2.6.1.3.2 Clinical

There are no symptoms or symptom complexes that are pathognomic of adenomyosis and the

symptoms classically ascribed to adenomyosis; abnormal uterine bleeding, painful periods

and pelvic pain, are non-specific and associated with many other gynaecological pathologies.

Furthermore, adenomyosis is often asymptomatic with 35% of women diagnosed with the

condition in one study not having any symptoms explained by the disease (Benson and

Sneedon 1958). Against this background, it is unsurprising that the accuracy of clinical

diagnosis of adenomyosis is poor. In one study, a putative diagnosis of adenomyosis was

confirmed in under half of women who subsequently underwent a hysterectomy (Lee et al.

1984).
 47

2.6.1.3.3 Ultrasound

2.6.1.3.3.1 Diagnostic accuracy of ultrasound

Many investigators have assessed the accuracy of transvaginal ultrasound in the diagnosis of

adenomyosis, invariably in women undergoing hysterectomy, by comparing pre-operative

ultrasound diagnosis with subsequent histological diagnosis. The reported sensitivity varies

from 53%-89% and the reported specificity varies from 50%-99% (Dueholm 2006). The

prevalence of adenomyosis in these studies varies widely which may reflect the differing

histological criteria used, the different populations studied or the varying degree of patient

selection involved. Generally, the more highly selected the patients are, the higher the

sensitivity of diagnosis (Dueholm 2006). Large studies with no selection will, almost

invariably, include women with either large uterine fibroids, or large numbers of fibroids.

Fibroids of this nature are likely to prevent examination of the entire myometrium, meaning

that ultrasound features of adenomyosis may be missed (Bazot et al. 2001). The difficulty in

differentiating adenomyomas and fibroids may also lead to false positive diagnosis of

adenomyosis. The presence of other pathologies might also reduce the accuracy of ultrasound

in the diagnosis of adenomyosis. Studies of post-menopausal women, and even studies of

pre-menopausal women if large enough, are likely to include some women with endometrial

cancer. The presence of endometrial cancer, particularly if it involves the myometrium,

makes the accurate diagnosis of adenomyosis much more challenging.

2.6.1.3.3.2 Ultrasound features of adenomyosis

Uterine enlargement in the absence of uterine fibroids and asymmetrical myometrial

thickening are indirect signs of adenomyosis originally described on transabdominal

ultrasound (Siedler et al. 1987; Bohlman et al. 1987). Transabdominal ultrasound, however,

does not have sufficient resolution to be able to directly visualise some of the subtle
 48

sonographic features of adenomyosis that have been described on transvaginal ultrasound

(Bazot et al. 2002). The characteristic gross appearance and individual ultrasound features of

adenomyosis on transvaginal ultrasound derive not only from the presence of endometrial

glands and stroma within the myometrium, but also from the associated muscular

hypertrophy and hyperplasia (Ferenczy 1998). Many of these features of adenomyosis are

subtle and may be missed by those not trained to look for or recognise them. Some authors

have suggested that the learning curve for those already familiar with TVUS to be able to

diagnose adenomyosis can be up to 6 months (Dueholm 2006). These difficulties are

compounded by the fact that the features of adenomyosis cannot always be seen on hard-copy

images leading some authors to suggest that it must be diagnosed during real-time

examination, which has implications for teaching (Reinhold et al. 1996).

The asymmetrical myometrial thickening seen on transabdominal ultrasound can also be seen

on TVUS. Another common feature described is heterogenous, poorly defined areas within

the myometrium (Reinhold et al. 1995) or increased myometrial echotexture (Reinhold et al.

1998). These may include or be accompanied by hyperechoic islands or anechoic myometrial

cysts and lacunae (Fedele et al. 1992). Parallel shadowing can sometimes be seen distal to

these myometrial features. Adenomyomas can be seen on TVUS but can be mistaken for

uterine fibroids, with which they will often co-exist. It is particularly important to distinguish

between the two in pre-operative imaging as adenomyomas are less amenable to surgical

resection than uterine fibroids. Adenomyomas tend to be more elliptical than uterine fibroids

and will be less well-circumscribed than uterine fibroids. They will also tend to have less

calcification and less edge-shadowing (Reinhold et al. 1998). The use of Doppler can assist in

distinguishing between adenomyomas and uterine fibroids. Blood flow around fibroids will

be circumferential with minimal or absent blood flow seen within the fibroid, whereas with
 49

adenomyomas, blood vessels will be seen following their normal course through the

myometrium (Table 1).

Table 1 – Ultrasound features of fibroids and adenomyomas that aid differentiation on TVUS.

Fibroid Adenomyomas

Border definition Clear Poor

Echogenicity Mixed with calcifications Hypoechoic with striate

effect

Shape Globular Elliptical

Blood flow on Doppler Circumferential Undisturbed

imaging

Features seen closer to the endometrium include linear striations and an irregular or indistinct

EMJ. Linear striations are seen as fine hyperechoic lines radiating from the endometrium into

the myometrium. An irregular EMJ has been reported as a feature of adenomyosis on both B-

mode (Reinhold et al. 1998) and 3D imaging (Ahmed et al. 2007). Recently, Exacoustos et al.

used the 3D coronal view of the uterus to visualise the EMJ and measure its thickness, to

assess its usefulness in diagnosing adenomyosis (Exacoustos et al. 2011). They found that

alterations in the EMJ had good diagnostic accuracy for diagnosing adenomyosis. While this

is clearly an area of current research focus, there remain no published studies that describe a

normal range for EMJ thickness, when measured in this way.

A few studies have looked at which specific ultrasound features are the most accurate in the

diagnosis of adenomyosis. Bazot et al. found that myometrial cysts were the most specific

and sensitive criterion for diagnosing adenomyosis (Bazot et al. 2001). Kepkep et al. found

that myometrial cysts, along with sub-endometrial linear striations and a regularly enlarged
 50

globular uterus were the ultrasound features with the highest accuracy (74.3%, 71.4% and

80% respectively). Sub-endometrial linear striations were the feature with the highest

specificity (95.5%) (Kepkep et al. 2007). It should be noted that while there was considerable

overlap in the ultrasound features used in these studies, the features used were not identical.

2.6.1.3.4 MRI

2.6.1.3.4.1 Diagnostic accuracy of MRI

There are a small number of studies assessing the performance of MRI in the diagnosis of

adenomyosis, all of which included women undergoing hysterectomy. They compared pre-

operative MRI diagnosis with subsequent histological diagnosis and showed a sensitivity of

70-86% and a specificity of 86-93% with a mean accuracy of 87.5% (Bazot et al. 2001,

Reinhold et al. 1996, Dueholm et al. 2001).

2.6.1.3.4.2 MRI features of adenomyosis

The most widely used feature for diagnosing adenomyosis on MRI is increased JZ thickness.

Focal or diffuse thickening (>12mm) of the JZ on T2-weighted images is considered highly

suggestive of adenomyosis, whereas adenomyosis is unlikely to be present if the JZ is less

than 8mm thick (Reinhold et al. 1998). Focal thickening is considered more specific than

generalised thickening which must be differentiated from the appearance of thickening

caused by physiological inner myometrial contractions. The timing of the MRI must also be

considered before diagnosing adenomyosis by this criterion, as JZ thickness varies

considerably during the menstrual cycle (Masui et al. 2001). The JZ is most clearly visible

during the late secretory phase (Imaoka et al. 2003) and JZ thickness commonly appears to be

>12mm during the menstrual phase, particularly on days 1 and 2. For this reason, some

authors advocate avoiding MRI imaging during the menstrual phase of the menstrual cycle in
 51

order to avoid false positive diagnosis of adenomyosis (Tamai et al. 2006). There are further

limitations in using JZ thickness as the sole criterion to diagnose adenomyosis. One study

found that the JZ was unmeasurable in almost a third of the women in their study, 22% of

whom were subsequently diagnosed with adenomyosis on histology (Bazot et al. 2001).

Less common MRI features of adenomyosis that have been reported include areas of low-

signal intensity within the myometrium with ill-defined borders (Reinhold et al. 1996) and

islands of endometrial tissue visualised as punctate foci of high-signal intensity on T2-

weighted images. If cyclical bleeding occurs within these glands, they become cystic and will

appear as areas of high-signal intensity on T-1 weighted imaging (Outwater et al. 1998).

Rarely, endometrium can be seen to be invading the myometrium, which gives the

appearance on MRI of ‘pseudowidening’ of the endometrium (Reinhold et al. 1999). While

visualisation of these direct signs of adenomyosis has been reported, MRI has less-than-ideal

sensitivity in detecting them, for example small myometrial cysts are only detected in 50% of

cases. It has been suggested that this may be because MRI has insufficient spatial resolution

to identify these smaller features of adenomyosis (Novellas et al. 2011).

2.6.1.3.4.2 Comparison of diagnostic accuracy of MRI and TVUS

Current opinion is divided as to which imaging modality should be the first-line diagnostic

tool in adenomyosis. The diagnostic accuracies of TVUS and MRI have been compared in

three studies of women undergoing hysterectomy for benign reasons (Table 2). While small

advantages were found with MRI, pooled results show the two techniques to be equivalent

and of intermediate accuracy (Dueholm 2006). A significant difference was found in the

inter-observer variability of the two diagnostic modalities with MRI having good inter-

observer agreement (kappa value = 0.73) and TVUS having only fair inter-observer

agreement (kappa value 0.38) (Dueholm et al. 2002). It may be that due to improvements in
 52

the resolution of TVUS machines over the last decade, alongside better training in the

ultrasound diagnosis of adenomyosis, the inter-observer agreement in the ultrasound

diagnosis of adenomyosis is now better, but no published studies have reassessed the inter-

observer variability of TVUS in the diagnosis of adenomyosis.

While TVUS, in the hands of experienced operators had comparable accuracy to MRI, MRI

seemed to be a superior, less observer-dependent diagnostic tool in the diagnosis of

adenomyosis. Nevertheless, a review concluded that where adenomyosis was suspected,

TVUS should be the first-line diagnostic tool, with MRI reserved for cases where TVUS was

inconclusive (Dueholm 2006). This conclusion may stem from the fact that TVUS is well-

tolerated (Bennett and Richards 2000) and is both cheaper and more easily accessible than

MRI.

Table 2 – Studies comparing the diagnostic accuracy of TVUS and MRI for the diagnosis of
adenomyosis (Dueholm 2006).
Reinhold et al. 1996 Bazot et al. 2001 Dueholm et al. 2001
% (95% CI) % (95% CI) % (95% CI)
TVUS:
Sensitivity 89 (71-97) 65 (48-79) 74 (63-82)
Specificity 89 (80-95) 98 (90-100) 87 (81-91)
PPV 71 (54-99) 93 (75-99) 68 (58-77)
NPV 96 (89-99) 85 (75-91) 89 (84-92)
MRI:
Sensitivity 86 (66-95) 78 (61-89) 78 (68-86)
Specificity 86 (76-92) 93 (84-97) 88 (83-92)
PPV 65 (47-79) 84 (67-93) 70 (60-79)
NPV 95 (87-98) 89 (80-95) 92 (87-95)

2.6.1.3.5 CT

There are limited data on the use of computerised tomography (CT) for the diagnosis of

adenomyosis. A retrospective study of 16 women with an MRI diagnosis of adenomyosis, 6

of whom had also had adenomyosis diagnosed on ultrasound scan, suggested that there were
 53

features of adenomyosis that could be seen on CT scan, however the prospectively reported

CT scans failed to diagnose adenomyosis in any of the women (Woodfield et al. 2009).

2.3.1.3.6 Hysterosalpingography

Adenomyosis has been diagnosed on hysterosalpingography, with features suggestive of

adenomyosis including multiple small spicules with saccular endings (‘lollipop-like’

diverticula), extending from the endometrium into the myometrium (Wolf and Spataro 1988)

and irregularity of the uterine contour with small outpouchings of contrast material (Simpson

et al. 2006) (Figure 9). The diagnostic sensitivity is poor, however, and so

hysterosalpingography is no longer used for the diagnosis of adenomyosis (Arnold et al.

1995).

Figure 9 - Hysterosalpingogram showing irregularity of the uterine contour with small

outpouchings of contrast material. Findings consistent with diffuse adenomyosis (Simpson et
al. 2006)
 54

2.6.1.4 Clinical impact of adenomyosis

2.6.1.4.1. Prevalence of adenomyosis

The prevalence of adenomyosis in the general population remains unknown as, so far, it has

only been possible to establish the diagnosis in pathological specimens, which confers a large

selection bias. There is wide variation in the percentage of hysterectomy specimens found to

have adenomyosis with reported prevalences of 5-70% (Azziz 1989). Explanations for this

wide variation include the lack of consensus in the histological criteria used and variation in

the number of tissue blocks examined.

2.6.1.4.2 Symptoms of adenomyosis

The exact clinical importance of adenomyosis remains uncertain (Dueholm 2006). The

common symptoms said to be associated with adenomyosis are menorrhagia and

dysmenorrhoea, with less commonly reported symptoms including chronic pelvic pain and

dyspareunia. These symptoms are common to a number of other gynaecological conditions

including uterine fibroids, endometrial polyps, endometrial hyperplasia and endometriosis.

These conditions frequently co-existent, making assessment of causation of symptoms

difficult. Furthermore, as is the case with endometrial polyps, fibroids and endometriosis,

many women with adenomyosis are asymptomatic (Bird et al. 1972; Benson & Sneedon

1958; Molitor 1971).

2.6.1.4.2.1 Abnormal uterine bleeding

The first detailed description of adenomyosis by Cullen over a century ago described a

condition associated with ‘lengthened menstrual periods’ (Cullen 1908). Many investigators

since then have looked for an association between adenomyosis and abnormal uterine

bleeding with conflicting results. While many studies have found an association between the
 55

two (Emge 1962; Bird et al. 1972; Benson and Sneedon 1958), many others have not

(Bergholt et al. 2001; Parrazzini et al. 1997; Weiss et al. 2009). Despite the uncertainty

regarding an association, adenomyosis is still considered an important cause of abnormal

uterine bleeding as evidenced by its appearance in PALM-COEIN, a mnemonic created by

FIGO in 2011 as an aide-memoire for causes of abnormal uterine bleeding (Figure 10)

(Munro et al. 2011).

Figure 10 – Illustration of PALM COEIN mnemonic for the causes of abnormal uterine
bleeding (Munro et al. 2011).

It is not only the association between adenomyosis and abnormal uterine bleeding that is

uncertain. The mechanism by which adenomyosis might cause bleeding symptoms also

remains unclear. The finding that mefenamic acid can reduce menstrual loss suggests that

prostaglandins may play a role (Azziz 1989). Impaired contractility of the adenomyotic

uterus has been suggested as another causative mechanism (Bergeron et al. 2006). Uterine

enlargement secondary to adenomyosis might result in a larger endometrial surface area and

therefore a greater volume of menstrual loss although a study by Rees et al. found no
 56

correlation between measured menstrual blood loss and endometrial volume-to-surface ratio

(Rees et al. 1984).

2.6.1.4.2.2 Pain

As with abnormal uterine bleeding, pain symptoms featured in early descriptions of

adenomyosis. Cullen described a condition associated with ‘a great deal of pain’ (Cullen

1908). Many authors have subsequently described an association between adenomyosis and

pain symptoms, largely dysmenorrhoea (Emge 1962; Bird et al. 1972; Azziz 1989; Benson

and Sneedon 1958), although once again, not all investigators have replicated these findings

(Bergholt et al. 2001; Weiss et al. 2009). One prospective study found no difference in either

frequency or severity of dysmenorrhoea or pelvic pain between 28 women with adenomyosis

and 157 controls (Kilkku et al. 1984). Despite the lack of consensus as to the relationship

between adenomyosis and pain, it is still considered a recognised cause of dysmenorrhoea

(Peric and Fraser 2006). Dysmenorrhoea may be caused by uterine irritability and

inflammation secondary to bleeding into foci of adenomyosis (Bergeron et al. 2006).

2.6.1.4.2.3 Reproductive function

Historically, adenomyosis has been considered a condition of parous women in their later

reproductive years and so little work has focussed on its effect on reproductive function.

Furthermore, an inability to diagnose the condition pre-hysterectomy has prevented

investigation of the condition in a subfertile population. The advent of pre-operative

diagnosis alongside an increase in maternal age has led to the condition being encountered

more frequently in fertility clinics. Certainly, it is now accepted that the condition can be

found in nulligravid and nulliparous women. Animal studies have suggested a deleterious

effect on reproductive function with induced adenomyosis in baboons being associated with
 57

lifelong infertility (Barrier et al. 2004). This has all contributed to an increasing focus on its

role in reproductive function with some authors proposing that adenomyosis is a significant

contributor to subfertility. As such, its diagnosis or exclusion should be mandatory in fertility

investigations (Leyendecker et al. 2006). The prevalence of adenomyosis in a specific subset

of fertility patients, women below the age of 36 with a confirmed diagnosis of endometriosis

and fertile partners, was found to be as high as 90% (Kunz et al. 2005). It has been proposed

that one mechanism through which adenomyosis might have a deleterious effect on fertility is

through impairment of the inner myometrial contractions previously discussed as having a

crucial role in sperm transport through the uterus (Leyendecker et al. 1996). The finding that

adenomyosis has a negative impact on IVF/ICSI conception rates and live pregnancy rates

(Salim et al. 2012; Tahlluri and Tremellen 2012) suggests that adenomyosis may negatively

impact events after fertilisation and nidation as well.

 58

2.6.2 Endometrial cancer

2.6.2.1 Introduction

Endometrial cancer was recently estimated to be the commonest gynaecological cancer in the

developed world (Sankaranarayanan and Ferlay 2006). While the prognosis is good compared

to other cancers, around 20% of women diagnosed with the condition will not survive beyond

5 years. All but the earliest endometrial cancers, or those developing in foci of adenomyosis,

will involve the EMJ.

2.6.2.2 Prevalence

In 2007, 7536 new endometrial cancers were diagnosed in the UK, making it the fourth

commonest cancer in women (ONS 2009). It is classically, although not exclusively, a

disease of post-menopausal women and its incidence is rising in this population, whereas it

seems to be stable or decreasing in premenopausal or perimenopausal women (Bray et al.

2005). This trend is likely to continue secondary to an ageing population, increasing obesity

and a sharp fall in the number of hysterectomies being performed (Amant et al. 2005;

Sorosky 2008; Reid and Mukri 2005).

2.6.2.3 Clinical presentation

Postmenopausal bleeding, defined as vaginal bleeding occurring at least a year after the last

menstrual period, is the cardinal symptom of endometrial cancer. The probability of

endometrial cancer in women presenting with postmenopausal bleeding is 5-10% (Amant et

al. 2005). Prior to menopause, women with endometrial cancer may present with

intermenstrual bleeding or irregular dysfunctional menstrual bleeding. Pain, vaginal

discharge and pyometra are rarer symptoms and tend to be secondary to advanced cancer

(Saso et al. 2011).

 59

2.6.2.4 Risk factors for endometrial cancer

The development of endometrial cancer has been observed to follow two distinct pathways

(Bohkman 1983). Type I cancers, which account for 80-90%, are usually oestrogen-

dependent, endometrioid adenocarcinomas and generally have a good prognosis. Type II

cancers are largely oestrogen-independent, have a variety of different histologies and tend to

have a higher grade. Type II endometrial cancers therefore tend to have a worse prognosis as

they tend to present later and behave more aggressively. As a significant majority of

endometrial cancers are type I, most of the risk factors for the disease will link in some way

to a relative increase in oestrogen exposure. A list of risk factors and protective factors can be

seen in Table 3 (Saso et al. 2011).

Table 3 – Risk factors and protective factors for endometrial cancer (Saso et al. 2011).
Endogenous risk factors Increasing age
Obesity and physical inactivity
Early menarche and late menopause
Low parity or infertility
Polycystic ovarian syndrome
Family history
Lynch syndrome (hereditary non-polyposis
colorectal cancer)
Oestrogen secreting tumours (granulosa or
thecal cell tumours of the ovary)
Diabetes mellitus
Hypertension
History of breast cancer
Immunodeficiency
Exogenous risk factors Unopposed oestrogen-only hormone
replacement therapy
Tamoxifen therapy
Previous radiotherapy
Dietary factors
Protective factors Cigarette smoking
Combined oral contraceptive use for at least
one year
Grand multiparity
 60

2.6.2.5 Precursor lesions for endometrial cancer

Hyperoestrogenism is also a risk factor for endometrial hyperplasia, a premalignant condition

that predisposes to endometrial cancer. Endometrial hyperplasia involves excessive cellular

proliferation leading to an increased volume of endometrial tissue, with an increased ratio of

endometrial glands to stroma (greater than 1:1). The WHO histological classification

(Silverberg et al. 2003) is used in the UK. It classifies endometrial hyperplasia on the basis of

the complexity of endometrial glands and the presence of any cytological atypia. Women

with simple and complex endometrial hyperplasia have a risk of developing endometrial

cancers of 1% and 3% respectively. Furthermore, 30-40% of women with complex atypical

endometrial hyperplasia will have a concurrent adenocarcinoma, and those that do not, have a

high risk of developing it (Mills and Longacre 2010)

2.6.2.6 Diagnosis of endometrial cancer

In the UK, if women report postmenopausal bleeding, they should be referred urgently to a

gynaecological rapid-access clinic, where they should be seen within two weeks (NICE

2005). A variety of different diagnostic tools have been evaluated in the investigation of

women with suspected endometrial cancer.

2.6.2.6.1 Blind endometrial biopsy

Dilatation and curettage (D&C) used to be the most common method to acquire a sample of

the endometrium for histological analysis. It has fallen out of favour for a variety of reasons,

including its need to be performed under general anaesthetic, its cost, its low sensitivity in

diagnosing intrauterine pathology (Bettocchi et al. 2001) and its relatively high complication

rate. Where a blind endometrial endometrial biopsy is required, D&C has largely been

superceded by outpatient endometrial sampling devices, modelled on the pipelle de cornier

 61

prototype. Endometrial sampling with a pipelle can be performed in an office setting, without

general anaesthetic, and often without local anaesthetic. A meta-analysis has shown that

pipelle biopsy has a detection rate for endometrial cancer of 99.6% in postmenopausal

women and 91% in premenopausal women. It had a sensitivity of 81% for the detection of

endometrial hyperplasia (Dijkhuizen et al. 2000). It appears to be more accurate in

symptomatic and post-menopausal women (Clark et al. 2001). For it to be cost-effective as

the first-line diagnostic tool, the population in which it is used should have a prevalence of

endometrial cancer of at least 15% (Dijkhuizen et al. 2003).

2.6.2.6.2 Hysteroscopy

Hysteroscopy involves the passage of a thin scope through the cervix, allowing direct

visualisation of the uterine cavity and endometrium. It can be performed as a daycase

procedure under general anaesthetic, but is increasingly being performed under local

anaesthetic in an outpatient setting where it has been shown to have equivalent rates of

patient satisfaction but a shorter recovery time (Kremer et al. 2000). The advantage of

hysteroscopy over blind endometrial biopsy is that it allows targeted biopsying of suspicious

areas. Even if the endometrial cavity appears normal, it is recommended that an endometrial

biopsy be performed with one study finding that 50% of the endometrial cancers diagnosed

on endometrial sampling, were not identified at hysteroscopy (Lo et al. 2000). A further

advantage of hysteroscopy is that where endometrial polyps are present, a polypectomy can

accurately be performed, thus treating a potential cause of abnormal bleeding, even if it

subsequently turns out to be benign.

Hysteroscopy under general anaesthetic has the same cost disadvantages as D&C, as well as

being time-consuming for patients. While outpatient hysteroscopy has economic and clinical

benefits it remains an invasive procedure associated with significant patient discomfort (Tahir
 62

et al. 1999). Furthermore, it has a failure rate of up to 10% (Lo et al. 2000). In view of the

invasive nature of outpatient hysteroscopy and its associated costs, transvaginal ultrasound

has been recommended as the preferable first line investigation in women presenting with

abnormal bleeding (Clark 2004).

2.6.2.6.3 Transvaginal ultrasound

In contrast to diagnostic hysteroscopy, TVUS is non-invasive, has no complications and is

well-tolerated by women with postmenopausal bleeding (Nasri et al. 1991). It is therefore

unsurprising that it has been extensively investigated as a tool for screening women with

postmenopausal bleeding.

2.6.2.6.3.1 Endometrial thickness

In populations where the incidence of endometrial cancer is below 15%, using TVUS as the

first line investigation, followed by endometrial biopsy if an abnormality is detected, has

been shown to be more cost-effective than using endometrial biopsy as the first-line

diagnostic tool (Dijkhuizen et al. 2003). In addition to saving money, using TVUS can

potentially save women from undergoing invasive endometrial sampling. A meta-analysis of

35 studies involving 5892 women that evaluated different thresholds to define abnormal

endometrial thickening showed that a cut-off of 5mm had a sensitivity of 96% for

endometrial cancer. Thus a postmenopausal woman presenting with vaginal bleeding with a

pre-test probability of endometrial cancer of 10%, would have a 1% chance of endometrial

cancer with a normal TVUS. The authors concluded that TVUS assessment of endometrial

thickness could reliably identify women with a very low likelihood of endometrial cancer

who could thus avoid endometrial sampling. While the sensitivity of this cut-off for

diagnosing endometrial cancer is very high, the specificity was just 61% (Smith-Bindman et
 63

al. 1998). Therefore the positive predictive value of endometrial thickness measurement in

diagnosing endometrial cancer is poor and many women therefore undergo ultimately

unnecessary invasive procedures. A subsequent meta-analysis had similar results and

concluded that while TVUS assessment of endometrial thickness was a good test for ruling

out endometrial cancer, it was not very good at ruling it in (Gupta et al. 2002). TVUS remains

the standard first-line test in the diagnosis of endometrial cancer, however because of the

limitations of using endometrial thickness alone, many investigators have looked at ways to

improve its specificity.

2.6.2.6.3.2 Endometrial thickness and morphology

Various endometrial ultrasound criteria have been assessed in attempts to improve the

specificity of ultrasound in diagnosing endometrial cancer, with mixed results. While some

have found assessments of endometrial and EMJ morphology useful in differentiating benign

and malignant endometria (Randelzhofer et al. 2002; Epstein and Valentin 2006), others have

not (Epstein et al. 2001; Epstein et al. 2002).

2.6.2.6.3.3 Transvaginal ultrasound with Doppler

There have been similarly contrasting results from studies assessing the use of Doppler to

improve the accuracy of ultrasound in diagnosing endometrial cancer. An early study

reported promising results in the use of colour Doppler to aid diagnosis of endometrial cancer

(Bourne et al. 1991). Subsequent studies of colour Doppler flow in both the uterine and

endometrial arteries were however, unable to show that it was useful in discriminating

between benign and malignant endometrial lesions (Sladkevicius et al. 1994; Sheth et al.

1995). The evidence is no more consistent when Doppler has been used to assess the
 64

morphology of endometrial vasculature. Alcazar et al. reported that a particular power

Doppler pattern of vasculature could be observed in more than 80% of cases of endometrial

cancer (Alcazar et al. 2003). Other investigators used an algorithm including power Doppler

assessment of endometrial vascularity to estimate the risk of endometrial malignancy. While

they concluded that their algorithm was better than subjective assessment, at differentiating

benign and malignant endometria in women with an endometrial thickness of 5-15mm, this

conclusion did not reach statistical significance (Epstein et al. 2002). Subsequent work by the

same research group was unable to show that using colour Doppler to assess the number of

blood vessels within the endometrium aided the diagnosis of endometrial cancer in women

with postmenopausal bleeding (Epstein and Valentin 2006)

2.6.2.6.3.4 Three-dimensional ultrasound assessment of endometrial volume

Gruboeck et al. used compared endometrial thickness and endometrial volume measured by

3D ultrasound in the assessment of women with postmenopausal bleeding (Gruboeck et al.

1996). They found that endometrial volume assessment was a better screening tool than

endometrial thickness measurement with an endometrial volume of 13mls having a

sensitivity of 100% and a specificity of 99.8% for the diagnosis of endometrial cancer. A

subsequent study again showed a higher sensitivity for endometrial volume than endometrial

thickness (Mansour et al. 2007) although interestingly their cut-off value for endometrial

volume was significantly different to that proposed in the study by Gruboeck et al.

Furthermore, they used D&C as their gold standard which has been shown to be unreliable

(Grimes 1982). Perhaps what is most noteworthy is that despite the early promise of 3D

endometrial volume measurement in the screening of women with postmenopausal bleeding,

a relatively small number of papers have subsequently been published on the subject. A

recent review article (Alcazar and Jurado 2011) suggested that most of the published studies
 65

showed that endometrial volume was superior to endometrial thickness. One study

(Opolskiene and Valentin 2010) showed no advantage with 3D endometrial volume

measurement.

2.6.2.7 Staging

Staging of endometrial cancer can be based on surgical, clinical, radiological of

histopathological criteria, however surgical staging has been shown to have better prognostic

value (Creasman et al. 2006). This led to FIGO updating the staging of endometrial cancer to

more accurately reflect stage-for-stage prognosis (Table 4) (Creasman 2009). About 80% of

women present with stage I disease and can be treated with hysterectomy and bilateral

salpingo-oophorectomy.

Table 4 – Revised FIGO staging of endometrial cancer (Creasman et al. 2009).

Stage Ia <50% myometrial invasion

Stage Ib >50% myometrial invasion

Stage II Tumour confined to the uterine corpus but involving cervical

stroma

Stage IIIa Tumour invading the serosa of the uterine corpus and/or adnexa

Stage IIIb Vaginal or parametrial involvement

Stage IIIc Metastasis to pelvic or para-aortic lymph nodes

Stage IVa Invasion of bladder and/or bowel mucosa

Stage IVb Distant metastases including intra-abdominal and/or inguinal

lymph nodes
 66

3.0 Conclusion to Background

It is clear that there is an increasing focus on the EMJ and what its role in various pathologies

might be. Gynaecologists and ultrasonographers are able to look at this area of the uterus in

greater detail and depth and are even starting to make clinical decisions based on their

assessments. Yet there is a danger that the clinical decision-making is out of step with what

we know about the EMJ and its appearance on ultrasound. Obvious issues worth investigating

are whether ultrasound assessment of the EMJ is reproducible and what clinical and

demographic factors influence the appearance of the EMJ.

Adenomyosis, a disease of the EMJ, has been studied for considerably longer than the EMJ

itself and yet still so many aspects of the disease remains unclear or uncertain. While

histology has historically been the primary diagnostic tool for investigating adenomyosis, the

significant reduction in the number of hysterectomies now being performed for benign

disease has blunted its usefulness in investigating the disease. Almost all ultrasound studies

of adenomyosis to date have assessed its diagnostic accuracy. While these studies have

shown that transvaginal ultrasound has good accuracy in diagnosing adenomyosis, few if any

ultrasound studies have attempted to investigate some of the many clinical uncertainties that

remain about the disease. This has perhaps been due to concerns about the reproducibility of

ultrasound diagnosis of adenomyosis. Significant uncertainties remain about the most basic

aspects of the disease including its prevalence and even whether or not it has any clinical

significance. Ultrasound should now be used to investigate these questions.

The role of ultrasound in the screening of women with post-menopausal bleeding is more

established. Transvaginal ultrasound assessment of endometrial thickness remains the

primary determinant of which women can be safely reassured and which women need

endometrial sampling. Using ultrasound in this way has a high sensitivity for endometrial

cancer but a low specificity meaning that many women undergo ultimately unnecessary
 67

invasive investigations. Several investigators have looked at whether more sophisticated

transvaginal ultrasound assessment of these women might be able to improve the diagnostic

accuracy of ultrasound and assessment of the EMJ has been suggested as one aspect of the

ultrasound examination that may aid this.

Given the above, I feel that the aims of this thesis should be to examine the reproducibility of

ultrasound assessment of the EMJ and investigate what factors affect it visualisation, examine

the reproducibility of ultrasound diagnosis of adenomyosis, determine the prevalence of

ultrasound features of adenomyosis in women attending a general gynaecology clinic and

investigate whether their presence is associated with clinical symptoms and lastly, determine

whether assessment of the EMJ can increase the diagnostic accuracy of ultrasound in the

investigation of women with post-menopausal bleeding

 68

PART II MATERIALS AND METHODS

 69

1.0 MATERIALS AND METHODS

1.1 Setting

1.1.1 The Gynaecological Diagnostic and Treatment Unit, University College Hospital

The work contained in this thesis was carried out in the Gynaecology Diagnostic and

Outpatient Treatment Unit (GDOTU) of University College Hospital between October 2008

and March 2011. Situated in central London, University College Hospital is a teaching

hospital and tertiary referral centre which primarily serves the needs of the people of Camden

and Islington. The trust has an annual turnover of more than £769 million, employs over 6000

staff and has 665 inpatient beds. It sees over 789,000 outpatients a year and has around

125,000 inpatient admissions a year. The hospital was formed in 1994 and became an NHS

foundation Trust in 2004. The Women’s Health department at UCH offers specialist

gynaecology services including gynaecological ultrasound scanning, early pregnancy care,

ambulatory gynaecology, urogynaecology, colposcopy, menopause, gynaecological

oncology, paediatric gynaecology, a specialist endometriosis centre, an assisted conception

unit and a specialist clinic for African women. In the financial year 2009/10, there were

11,270 patient attendances at the GDOTU of which 6,487 were women with early pregnancy

problems and 4,783 of which were non-pregnant women with gynaecological problems.

1.2 Ultrasound

1.2.1 Two-dimensional transvaginal ultrasound

In this thesis, all the ultrasound scans were performed by trained gynaecologists on a Voluson

E8 ultrasound machine (GE Medical Systems, Milwaukee, WI, USA) using a 4-9 MHz probe

with three dimensional facility. The scans were all perfomed in lithotomy position using a

standardised protocol. First the uterus was examined in the transverse plane to identify the

cervical canal and the uterine cavity. The probe was then rotated 90o anti–clockwise and the
 70

uterus and endometrium were visualised in the longitudinal plane. Care was taken to identify

the point at which the endometrial thickness was greatest before measuring it in millimetres.

If free fluid was present in the endometrial cavity, the fluid was measured separately and

subtracted from the total thickness (Lee et al. 2005). Fibroids were diagnosed based on direct

visualisation using previously described diagnostic criteria (Anderson 1999). Adenomyosis

was diagnosed if any of the following ultrasound features of adenomyosis were present:

asymmetrical myometrial thickening, myometrial cysts, linear striations, parallel shadowing,

adenomyomas, hyperechoic islands and an irregular EMJ on either B-mode or 3-dimensional

imaging (Table 5, Figures 11-17) (Naftalin et al. 2012b). Previously published ultrasound

features of adenomyosis that were not used to diagnose adenomyosis in this thesis include a

heterogenous myometrial echotexture and a globular uterine configuration. These were both

excluded as they had been found to have a low positive predictive value (Bazot et al. 2001).

The JZ measurements taken from 3D coronal views of the uterus proposed by Exacoustos et

al. were not used as they were published after our study had started. Even so we would not

have included them as their accuracy has yet to be prospectively assessed. Endometrial

polyps were diagnosed based on direct visualisation using previously described criteria

(Timmerman et al. 2003). The examination was then concluded by examining both adnexa

and the pouch of Douglas, identifying both ovaries and any endometriotic nodules where

possible.
 71

Table 5 – Ultrasound features of adenomyosis used in this thesis (Naftalin et al. 2012b).

Asymmetrical myometrial thickening

Parallel shadowing

Linear striations

Myometrial cysts

Hyperechoic islands

Adenomyomas

Irregular EMJ

Figure 11 - Asymmetrical myometrial thickening. A longitudinal view of an anteverted uterus

in which the distance from the endometrium to the anterior serosal surface (thick white line)
is much greater than the distance from the endometrium to the posterior serosal surface (thin
white line) (Naftalin et al. 2012b)
 72

Figure 12 - Parallel shadowing. A longitudinal view of a retroverted uterus in which parallel

hypoechoic lines can be seen running through the anterior myometrium (yellow s) (Naftalin
et al. 2012b)

s s
s

Figure 13 - Linear striations. A transverse view of the uterus showing a linear striation seen
as thin hyperechoic lines (thin white arrow) extending from the endometrium into the
myometrium (Naftalin et al. 2012b)
 73

Figure 14 - Myometrial cysts. A transverse view of the uterus showing several myometrial
cysts (thin white arrows) seen as anechoic lesions within the myometrium (Naftalin et al.
2012b)

Figure 15 - Hyperechoic islands. A coronal view of a uterus where hyperechoic islands of

tissue are seen in the fundal and lateral portions of the myometrium (thin white arrows).
(Naftalin et al. 2012b)
 74

Figure 16 - Adenomyomas. A longitudinal view of a retroverted uterus showing an

adenomyoma in the anterior myometrium (between white arrows). (Naftalin et al. 2012b)

Figure 17 - Irregular endometrial-myometrial junction. A coronal view of a uterus with an

irregular endometrial-myometrial junction. (Naftalin et al. 2012b)
 75

An ultrasound diagnosis of endometriosis was made if ovarian endometriomas or

endometriotic nodules were visualised on ultrasound scan. Ovarian cysts were classified as

endometriomas when they appeared as well-circumscribed thick-walled cysts that contained

homogeneous low-level internal echoes (‘ground glass’) (Tailor et al. 1999). Endometriotic

nodules were typically visualised as stellate hypoechoic or isoechogenic solid masses with

irregular outer margins (Bazot et al. 2007) which were tender on palpation and fixed to the

surrounding pelvic structures.

1.2.2 Three-dimensional transvaginal ultrasound

All 3-D ultrasound volume acquisitions were performed immediately after the 2D TVUS by

the same operator. The volume acquisition technique was performed in a standardised fashion

using a maximum sweep angle of 120 degrees and the medium quality setting. The uterus

was visualised in the longitudinal plane with the angle of the ultrasound beam and the axis of

the endometrial cavity approaching 90 degrees. In order to minimise artefacts, the depth of

the acquired volume was adjusted to cover the entire uterus with minimal inclusion of

parametrial structures. The volumes were generated by automatic rotation of the mechanical

transducer through 360o. The probe was held steady and the patients were asked to hold their

breath whilst volume acquisition was switched on. Following acquisition, the rendered

volumes were saved to the hard drive of the machine for analysis.

1.2.2.1 Off-line assessment of uterine volumes

Assessment of 3D uterine volumes was performed offline. A coronal view was obtained

using render mode, by placing a straight or curved line along the endometrial stripe on the

sagittal and transverse views (Panel A and B of the multi-planar view). The multi-planar view
 76

was then adjusted until a satisfactory coronal image was obtained of both the endometrial

cavity and the external uterine contour, with visualisation of both interstitial portions of the

fallopian tube. The gradient light and sepia settings were used to optimise the view of the

EMJ, which was visualised as a hypoechoic area surrounding the endometrial cavity.

When using off-line evaluation of stored uterine volumes to diagnose adenomyosis, a detailed

assessment of the entire myometrium was performed looking for the presence of each

individual ultrasound feature of adenomyosis. Both operators were required to manipulate the

rendered views, in whichever way they wanted, until they had visualised the entire

myometrium and a diagnosis was reached.

1.3 Pictorial Blood Loss Assessment Charts

Pictorial Blood Loss Assessment charts (PBACs) were given to premenopausal women to

complete in order to obtain semi—quantitative information about the amount of menstrual

loss they had each cycle. PBACs use a scoring system to calculate total menstrual loss that

accounts for both the number of sanitary towels or tampons a woman uses during her period

and also the degree to which each item is soiled (Figure 18). A pictorial chart score of >100,

when used as a diagnostic test for menorrhagia, has been shown to have a high specificity and

sensitivity (Higham et al. 1990).

 77

Figure 18 – A Pictorial blood loss analysis chart.

1.4 Numerical Rating Scale (NRS)

Premenopausal women were asked to complete an 11-point Numerical Rating Scale (NRS)

(Figure 19) in order to obtain a subjective assessment of how painful they found their periods.

The NRS is a scale where the extremes are no pain and pain as bad as it could be and patients

are asked to put a mark on the scale to represent the intensity of the pain being measured.

Figure 19 – An 11-point numerical rating scale.

 78

Women were given the scale and asked to put a mark on the scale to represent the intensity of

the worst pain they experienced with their last period. NRSs have been shown to be sensitive

(Jensen et al. 1986), simple to use with a low failure rate (Van Tubergen et al. 2002) and are

considered to be preferable to traditional visual analog scales for assessment of pain intensity

(Mannion 2007). They have been shown to have sufficient reliability and validity to be used

in clinical research (Lara-Munoz et al. 2004).

1.5 Endometrial sampling (Pipelle & Hysteroscopy)

In women with postmenopausal bleeding found to have an endometrial thickness on TVUS of

>5mm, endometrial sampling was performed. If the endometrium was globally thickened,

and the woman was able to tolerate outpatient endometrial sampling, a pipelle de cornier was

used (Prodimed, Neuillyen-Thelle, France). Women with a focally thickened endometrium

were referred for either outpatient or day-case hysteroscopy.

1.6 Statistical analysis

Data were collected and stored in electronic databases (Microsoft Excel 2003). The Statistical

Package for Social Sciences version 15.0 (Statistical Analysis Systems, Chicago, Illinois) was

used to analyse data. Details of individual tests are documented in the relevant chapters. The

level of statistical significance was chosen as p<0.05.

1.7 Ethical committee approval

The Local Research and Ethics Committee of University College Hospital granted approval

for all studies undertaken. They advised that none of the studies required full formal ethical

approval.
 79

III Results
 80

[1] INTER- AND INTRAOBSERVER VARIABILITY IN THREE-DIMENSIONAL

ULTRASOUND ASSESSMENT OF THE ENDOMETRIAL-MYOMETRIAL
JUNCTION

1.1 BACKGROUND

The endometrial-myometrial junction (EMJ) is thought to play an important role in both

physiological and pathophysiological processes within the uterus. It has been suggested that a

careful ultrasound examination of the EMJ junction can yield clinically useful information in

a number of different pathologies and so should become part of the routine ultrasound

examination of the uterus (Naftalin and Jurkovic 2009).

In order to obtain clear images of the EMJ, particularly at the lateral and fundal aspects of the

uterine cavity, high resolution ultrasound equipment with 3D imaging facilities is often

required, which has only become available in recent years. There are a small number of

published studies involving assessment of 3D views of the EMJ (Exacoustos et al. 2011;

Alcazar et al. 2009), however little is known about the reproducibility of 3D ultrasound

assessment of the EMJ. The aim of this study was to examine inter- and intra-observer

reproducibility of 3D ultrasound visualisation of the EMJ.

1.2 METHODS

Inter- and intraobserver variability in 3D ultrasound assessment of the EMJ was tested on a

group of 30 women. Real time ultrasound examinations were performed and 3D volumes

were obtained in the manner described in the materials and methods section. The investigator

who performed the ultrasound examinations (Dr Will Hoo) was a specialist in gynaecological

ultrasound with a special interest in benign gynaecological conditions. We included in the

study only non-pregnant women with a normal uterus on ultrasound scan. After selecting the

volumes for the study he did not participate in further reproducibility analysis. All saved
 81

volumes were then examined independently by two other investigators (Mr Davor Jurkovic

and Dr Joel Naftalin) who were blinded to each other’s findings. The volumes were examined

using planar reformatted sections and volume rendering. The quality of visualisation of the

EMJ was graded as follows: 1.) Optimal – The EMJ was clearly visible and could be

examined in its entirety, 2.) Satisfactory – Most, but not all parts of the EMJ could be clearly

visualised, 3.) Unsatisfactory – Large parts of the EMJ could not be clearly visualised (Figs

20-22). It was considered that EMJ’s graded as unsatisfactory were inadequate for clinical

evaluation, whereas 1) & 2) could be used for clinical evaluation.

1.3 STATISTICAL ANALYSIS

Statistical analysis was performed, using SPSS software (SPSS Inc., Chicago IL). The inter-

observer and intra-observer agreement for the classification of EMJ visualisation was

assessed by calculating the kappa statistic.

1.4 RESULTS

There was complete agreement between the two operators in classifying the visualisation of

the EMJ as being optimal or not. Both operators classified 16 EMJs as optimal, 8 as

satisfactory and 2 as unsatisfactory. There was a discrepancy between the two observers in 4

cases. In two of these, operator 1 classified two cases as satisfactory which operator 2 had

described as unsatisfactory. In the other two cases, the reverse occurred. The inter-observer

agreement was good with a kappa value of 0.77 (Table 6). The intra-observer agreement was

excellent for observer 1 with a kappa value of 0.83 (Table 7) and good for observer 2 with a

kappa value of 0.70 (Table 8).

 82

Figure 20 - Coronal view of uterus with EMJ visualisation classified as optimal (Naftalin et
al. 2012a)

Figure 21 - Coronal view of uterus with EMJ visualisation classified as satisfactory (Naftalin
et al. 2012a)

Figure 22 - Coronal view of uterus with EMJ visualisation classified as unsatisfactory

(Naftalin et al. 2012a)
 83

Table 6 - Inter-observer variability in the assessment of visualisation of the EMJ (Naftalin et

al. 2012a)

Operator 2
Operator Optimal Satisfactory Unsatisfactory Total
1 Optimal 16 0 0 16
Satisfactory 0 8 2 10
Unsatisfactory 0 2 2 4
Total 16 10 4 30
k=0.77 (good agreement)
95%CI (0.58 – 0.96)
Standard error of kappa = 0.097

Table 7 - Intra-observer variability (Operator 1) in the assessment of visualisation of the EMJ

(Naftalin et al. 2012a)

Operator 1 (2nd assessment)

Operator 1 Optimal Satisfactory Unsatisfactory Total
(1st Optimal 15 1 0 16
assessment) Satisfactory 0 9 1 10
Unsatisfactory 0 1 3 4
Total 15 11 4 30
k=0.83 (excellent agreement)
95%CI (0.66 – 1)
Standard error of kappa = 0.090

Table 8 - Intra-observer variability (Operator 2) in the assessment of visualisation of the EMJ

(Naftalin et al. 2012a)

Operator 2 (2nd assessment)

Operator 2 Optimal Satisfactory Unsatisfactory Total
(1st Optimal 16 0 0 16
assessment) Satisfactory 3 7 0 10
Unsatisfactory 0 2 2 4
Total 19 9 2 30
k=0.70 (good agreement)
95%CI (0.47 – 0.93)
Standard error of kappa = 0.117
 84

[2] FACTORS AFFECTING VISUALISATION OF THE EMJ ON THREE-

DIMENSIONAL ULTRASOUND

2.1 BACKGROUND

The EMJ is thought to play an important role in both physiological and pathophysiological

processes within the uterus. Its appearance on ultrasound is known to be altered in the

presence of certain pathologies, for example adenomyosis (Naftalin and Jurkovic 2009) and

invasive endometrial cancer (Randelzhofer et al. 2002). It is not known, however, whether

physiological factors affect its ultrasound appearance. The aim of this study was to identify

demographic and physiological factors which might affect the quality of visualisation of the

EMJ on 3D ultrasound.

2.2 METHODS

This was a prospective observational study of women attending our gynaecology clinic. In all

women, indication for the scan, age, parity, date of the last menstrual period, menstrual

pattern and the use of exogenous hormones was recorded and stored on a dedicated database

(PIA Fetal Database, version 5.5.4.152, Viewpoint Bildverarbeitung GmbH, Munich,

Germany). Postmenopausal status was defined as the absence of menstruation for ≥ 12

months in women ≥ 40 years old. Ultrasound examinations were performed in the manner

described in the materials and methods section. The examination was then concluded by

examining the adnexa and identifying both ovaries. The presence of a corpus luteum was

determined on B-mode and Doppler examination. Premenopausal women with regular

menstrual cycles, whose ovaries contained a corpus luteum were classified as being in the

luteal phase of the cycle, and the remaining women with regular menstrual cycles were

classified as being in the proliferative phase of the cycle.

 85

We included in the study only women with a normal uterus on ultrasound scan. Those with

evidence of any myometrial or endometrial pathology, pregnant women and those in whom

the ovaries could not be seen clearly were excluded. Various demographic and physiological

variables were recorded and we analysed their effect on the quality of EMJ visualisation on

3D volume rendering.

2.3 STATISTICAL ANALYSIS

Statistical analysis was performed, using SPSS software (SPSS Inc., Chicago IL).

Spearman’s rank correlation was used to examine the association of EMJ visualisation with

age, parity and endometrial thickness. The Kruskal-Wallis test was used to examine the

association between EMJ visualisation and the stage of menstrual cycle and the Mann-

Whitney test was used to examine the association between EMJ visualisation and menopausal

status. Subsequently, the joint effect of the various factors upon EMJ visualisation was

examined in a multivariable analysis. As EMJ visualisation was measured using an ordinal

scale, the analysis was performed using ordinal logistic regression.

2.4 RESULTS

413 consecutive, non-pregnant women were examined by a single operator (JN). 312 women

(75.5%) were excluded due to the presence of uterine pathology and the remaining 101

(24.5%) were included in data analysis. The principal indications for ultrasound examination

for the women included in data analysis are presented in Table 9. In 47/101 (46.5%) of the

women, visualisation of the EMJ was classified as optimal, in 42/101 (41.6%) as satisfactory

and in 12/101 (11.9%) as unsatisfactory. Spearman’s rank correlation showed that the ability
 86

to visualise the EMJ was significantly affected by a women’s age, parity, and endometrial

thickness as measured by ultrasound scan (Table 10).

The Kruskal-Wallis test did not show a significant effect of the stage of the menstrual cycle

or menopausal status on the quality of EMJ visualisation (Table 11). A subsequent analysis

assessing menopausal status only, using the Mann-Whitney test, showed a significant

association between menopausal status and EMJ visualisation (Table 12).

The joint effect of the various factors on EMJ visualisation was examined in a multivariable

analysis. A backwards selection procedure was used to retain only the statistically significant

variables. The results indicated that only parity and endometrial thickness were significantly

associated with EMJ visualisation. After adjusting for the effects of these two variables, there

was no significant difference in EMJ visualisation between pre and post-menopausal women

and the previously noted association between age and EMJ visualisation was no longer found

to be significant.

The size of effect of each variable upon EMJ visualisation, in the form of odds ratios, can be

seen in Table 13. These indicate the odds of being in the next best visualisation category

(optimal visualisation rather than satisfactory visualisation, or satisfactory visualisation rather

than unsatisfactory visualisation) for an increase in parity of one, or a 5mm increase in

endometrial thickness. The results suggest that increasing parity is negatively associated with

EMJ visualisation, whereby for each additional child delivered, the odds of being in the next

best visualisation category were almost halved. The results for endometrial thickness suggest

that increasing endometrial thickness is associated with better EMJ visualisation, a 5-mm
 87

increase in endometrial thickness being associated with a three-fold increased risk of being in

next best visualisation category.

Table 9 - Principal indications for examination (Naftalin et al. 2012a)

Principal indication for examination n (%)

Pelvic pain/dyspareunia 31
Subfertility 22
Irregular/Intermenstrual bleeding 11
Heavy or painful menstrual bleeding 10
Amenorrhoea 7
Postmenopausal bleeding 7
Suspected ovarian cyst 5
Postcoital bleeding 2
Other 6

Table 10 – Result of Spearman’s rank correlation in assessing the effect of age, parity and
endometrial thickness on EMJ visualisation (Naftalin et al. 2012a)

Variable EMJ Category Median (IQR) Correlation P-value

Coefficient

Age Optimal 32 (27, 36) 0.30 0.006

Satisfactory 32 (27, 42)
Unsatisfactory 46 (39, 60)

Parity Optimal 0 (0, 0) 0.33 <0.001

Satisfactory 0 (0, 1)
Unsatisfactory 1.5 (0.5, 3.5)

Endometrial Optimal 6.3 (3.5, 9.5) -0.36 <0.001

Thickness Satisfactory 3.5 (2.5, 6.0)
Unsatisfactory 3.1 (1.9, 4.9)

Table 11 – Result of Kruskal-Wallis test in assessing the effect of stage of menstrual cycle on
EMJ visualisation (Naftalin et al. 2012a)

Stage of cycle Optimal Satisfactory Unsatisfactory P-value

n (%) n (%) n (%)

Proliferative 19 (46%) 20 (49%) 2 (5%) 0.08

Secretory 11 (58%) 5 (26%) 3 (16%)
Irregular / absent 10 (63%) 3 (19%) 3 (19%)
Post-menopausal 2 (14%) 8 (57%) 4 (29%)
Combined pill 5 (45%) 6 (55%) 0 (0%)
 88

Table 12 – Result of Mann-Whitney test in assessing the effect of menopausal status on EMJ
visualisation (Naftalin et al. 2012a)

Menopausal Optimal Satisfactory Unsatisfactory P-value

status n (%) n (%) n (%)

Pre-menopausal 45 (52%) 34 (39%) 8 (9%) 0.004

Post-menopausal 2 (14%) 8 (57%) 4 (29%)

Table 13 – Results of multivariable analysis assessing the joint effect of the various factors
on EMJ visualisation (Naftalin et al. 2012a)

Variable Odds Ratio (95% CI) P-value

Parity 0.53 (0.36, 0.78) 0.001

Endometrial Thickness (*) 3.00 (1.52, 5.88) 0.001

(*) Odds ratio given for a 5mm increase in endometrial thickness

 89

[3] INTER- AND INTRAOBSERVER VARIABILITY IN THE ULTRASOUND

DIAGNOSIS OF ADENOMYOSIS USING STORED 3D VOLUMES

3.1 BACKGROUND

The ability to diagnose adenomyosis without having to perform a hysterectomy has helped

our understanding of the disease. There remains controversy, however, as to which imaging

modality should be the first-line diagnostic tool. Ultrasound is relatively cheap, widely-

accessible and well-tolerated and has been shown to have comparable accuracy to MRI in the

diagnosis of adenomyosis (Dueholm 2006). However, it was found to have a significantly

lower inter-observer variability (kappa=0.38) than MRI (kappa= 0.73) in the only published

study to compare reproducibility of diagnosis in the two modalities (Dueholm et al. 2002).

Furthermore, some authors have claimed that, unlike MRI, the ultrasound diagnosis of

adenomyosis can only be made during a real-time scan, as the ultrasound features of

adenomyosis cannot be reliably identified from static images (Reinhold 1999). The aim of

this study was to assess the inter- and intra-observer variability in ultrasound assessment of

adenomyosis using stored 3D uterine volumes. This assessment included diagnosis of

adenomyosis, the number of ultrasound features of adenomyosis seen and a subjective

assessment of severity of disease. The comparisons were made between:

a) real-time ultrasound scans and evaluation of the uterus from a stored 3D uterine

volume (same operator performing scan and stored volume assessment but 3 years

apart)

b) two different operators evaluating the same stored uterine volumes

c) the same operator evaluating the same stored uterine volumes on two separate

occasions
 90

3.2 METHODS

Out of a total of 985 archived 3D uterine volumes taken from women consecutively scanned

women between January 2009 and January 2010 by the same operator, 36 were selected by

one of the investigators (Dr Kate Pateman) for the examination of reproducibility of

ultrasound diagnosis of adenomyosis. The volumes were selected to ensure a good case mix

including normal uteri and those with adenomyosis. Those with adenomyosis were selected

to ensure a wide range in both the number and type of ultrasound feature of adenomyosis.

Volumes were not selected if it were not possible to adequately assess the entire

myometrium. Reasons for this included the presence of an IUCD, multiple large fibroids and

poor quality of stored volumes. The reproducibility assessments were performed more than 3

years after the original real-time scans in order to minimise bias in the assessment of intra-

observer agreement between diagnosis on real-time ultrasound and on assessment of stored

3D uterine volumes. The first and second assessment of 3D stored uterine volumes (both

assessments by Dr Joel Naftalin) were performed two weeks apart, and in a different order

from the first assessment, (the order being selected by Dr Kate Pateman who played no

further part in ultrasound assessment), again, in order to reduce bias. A third assessment of

the stored volumes was then performed by a third investigator (Mr Davor Jurkovic) who was

blinded to the results of all previous ultrasound assessments of the uteri. This was in order to

assess inter-observer agreement in the ultrasound diagnosis of adenomyosis based on

assessment of stored 3D ultrasound volumes of uteri.

3.3 STATISTICAL ANALYSIS

Statistical analysis was performed, using SPSS software (SPSS Inc., Chicago IL). Intra- and

inter-observer agreement was assessed for the diagnosis of adenomyosis, assessment of the
 91

number of ultrasound features of adenomyosis observed and subjectively assessed severity of

adenomyosis and for the classification of each ultrasound feature of adenomyosis. Weighted

and unweighted kappa analysis was used to assess the level of agreement in all cases.

3.4 RESULTS

The median age of the 36 women included in the data analysis was 41 (inter-quartile range:

35-47.75). 10/36 [27.7% (95% CI: 13.2% to 42.4%)] women were nulligravid and 12/36

[33.3% (95% CI: 17.9% to 48.7%)] were nulliparous. 6/36 [16.7% (95%CI; 4.5% to 28.8%)]

women were menopausal. The indications for ultrasound scans are listed in Table 14. A total

of 22/36 [60.1% (95% CI: 44.1% to 76.1%)] women had at least one ultrasound feature of

adenomyosis at their original real-time ultrasound scan. The total number of ultrasound

features seen on different scans can be seen in Table 15. 12/36 (33.3%) women had uterine

fibroids and 2/36 (5.6%) women had endometrial polyps.

There was good intra-observer agreement, both when the observer was comparing real-time

scan diagnosis of adenomyosis to diagnosis from stored uterine volumes, and when the

observer was comparing two assessments of stored uterine volumes performed on two

separate occasions, with kappa values of 0.67 and 0.83 respectively (Tables 16 and 21). There

remained a good level of intra-observer agreement when comparison was made of the

number of ultrasound features of adenomyosis recorded, with linear-weighted kappa values

of 0.64 and 0.77 (Tables 17 and 22). There was excellent intra-observer agreement (kappa =

0.83) in the subjective assessment of severity of adenomyosis between the two evaluations of

the stored uterine volumes (Table 23).

When two different operators assessed stored uterine volumes, there was good inter-observer

agreement for the diagnosis of adenomyosis (kappa = 0.61) (Table 18), the number of
 92

ultrasound features of adenomyosis seen (linear-weighted kappa = 0.63) (Table 19) and

subjective assessment of severity of adenomyosis (linear-weighted kappa = 0.61) (Table 20).

Table 14 - Indications for ultrasound scan

Indication for ultrasound scan n (%)
Pelvic pain 9
Menstrual symptoms 9
PCB/IMB/PMB 6
Ovarian surveillance 6
Subfertility 4
Other 3

Table 15 - Number of ultrasound features of adenomyosis seen on original real-time scan

Number of features seen n (%)
0 14 (38.9)
1 4 (11.1)
2 3 (8.3)
3 5 (13.9)
4 4 (11.1)
5 4 (11.1)
6 2 (5.6)
7 0 (0)
Total 36 (100)

Table 16 - Kappa analysis of intra-observer agreement between diagnosis of adenomyosis on

real-time ultrasound scan (operator 1) and on stored uterine volumes (operator 1)
Diagnosis on stored volume (operator 1)
Real time Not Adenomyosis Total
ultrasound adenomyosis
diagnosis Not 13 1 14
(operator 1) adenomyosis
Adenomyosis 5 17 22
Total 18 18 36
k=0.667 (good agreement)
95%CI (0.429-0.904)
Standard error of kappa = 0.121
 93

Table 17 - Linear-weighted kappa analysis of agreement on the number of ultrasound features

seen on real-time ultrasound scan (operator 1) and on stored uterine volumes (operator 1)
Number of ultrasound features of adenomyosis seen on stored
uterine volumes (operator 1)
Number of 0 1 2 3 4 5 6 7 Total
ultrasound 0 13 0 1 0 0 0 0 0 14
features of 1 2 1 1 0 0 0 0 0 4
adenomyosis 2 2 0 1 0 0 0 0 0 3
seen on real- 3 0 1 1 2 0 1 0 0 5
time 4 1 0 2 0 1 0 0 0 4
ultrasound 5 0 0 0 2 0 2 0 0 4
scan 6 0 0 0 0 1 0 1 0 2
(operator 1) 7 0 0 0 0 0 0 0 0 0
Total 18 2 6 4 2 3 1 0 36
Linear-weighted kappa = 0.64 (good agreement)
95% CI (0.47-0.80)
Standard error of kappa = 0.082

Table 18 - Kappa analysis of agreement of diagnosis of adenomyosis on stored uterine

volumes (operator 1 and 2)
Diagnosis on stored volume (operator 1)
Real time Not adenomyosis Adenomyosis Total
ultrasound Not adenomyosis 13 5 18
diagnosis Adenomyosis 2 16 18
(operator 1) Total 15 21 36
k=0.61 (good agreement)
95%CI (0.36-0.87)
Standard error of kappa = 0.130

Table 19a - Linear-weighted kappa analysis of inter-observer agreement on the number of

ultrasound features seen on stored uterine volumes (operator 1 and 2)
Number of ultrasound features of adenomyosis seen on stored
uterine volumes (operator 2)
Number of 0 1 2 3 4 5 6 7 Total
ultrasound 0 13 1 3 1 0 0 0 0 18
features of 1 0 1 1 0 0 0 0 0 2
adenomyosis 2 2 0 2 2 0 0 0 0 6
seen on real- 3 0 0 1 3 0 0 0 0 4
time 4 0 0 0 1 1 0 0 0 2
ultrasound 5 0 0 0 1 1 0 1 0 3
scan 6 0 0 0 0 1 0 0 0 1
(operator 1) 7 0 0 0 0 0 0 0 0 0
Total 15 2 7 8 3 0 1 0 36
Linear-weighted kappa = 0.63 (good agreement)
95% CI (0.47 - 0.78)
Standard error of kappa = 0.078
 94

Table 19b – Table showing the kappa values for inter-observer agreement between the
presence of individual ultrasound features of adenomyosis (operator 1 and 2)
Ultrasound feature Kappa value (95% CI)
Asymmetrical thickening 0.771 (0.53-1)
Parallel shadowing 1 (1-1)
Linear striations 0.571 (0.29-0.85)
Myometrial cysts 0.487 (0.22-0.753)
Hyperechoic islands 0.478 (0.135-0.822)
Adenomyomas 0 (0-0)
Irregular EMJ 0.650 (0.398-0.903)

Table 20 - Linear-weighted kappa analysis of inter-observer agreement on the severity of

adenomyosis on stored uterine volumes (operator 1 and 2)
Subjective assessment of severity of adenomyosis on
stored uterine volumes (operator 2)
Subjective Absent Mild Moderate Severe Total
assessment of Absent 13 5 0 0 18
severity of Mild 2 5 2 0 9
adenomyosis Moderate 0 1 2 0 3
on uterine Severe 0 1 3 2 6
volumes Total 15 12 7 2 36
(operator 1)
Linear-weighted kappa = 0.61 (good agreement)
95%CI (0.44-0.78)
Standard error of kappa = 0.087

Table 21 - Kappa analysis of intra-observer agreement of diagnosis of adenomyosis on stored

uterine volumes (operator 1)
Diagnosis on stored volume (operator 1)
Real time Not adenomyosis Adenomyosis Total
ultrasound Not adenomyosis 16 2 18
diagnosis Adenomyosis 1 17 18
(operator 1) Total 17 19 36
k=0.83 (excellent agreement)
95%CI (0.65-1.0)
Standard error of kappa = 0.092
 95

Table 22 - Linear-weighted kappa analysis of intra-observer agreement on the number of

ultrasound features seen on stored uterine volumes (operator 1)
Number of ultrasound features of adenomyosis seen on stored
uterine volumes (operator 2)
Number of 0 1 2 3 4 5 6 7 Total
ultrasound 0 16 1 1 0 0 0 0 0 18
features of 1 0 0 1 1 0 0 0 0 2
adenomyosis 2 1 1 3 0 1 0 0 0 6
seen on real- 3 0 1 1 2 0 0 0 0 4
time 4 0 0 0 0 1 1 0 0 2
ultrasound 5 0 0 0 0 1 1 1 0 3
scan 6 0 0 0 0 0 0 1 0 1
(operator 1) 7 0 0 0 0 0 0 0 0 0
Total 17 3 6 3 3 2 2 0 36
Linear-weighted kappa = 0.77 (good agreement)
95% CI (0.65 - 0.90)
Standard error of kappa = 0.064

Table 23 - Linear-weighted kappa analysis of intra-observer agreement on the severity of

adenomyosis on stored uterine volumes (operator 1)
Subjective assessment of severity of adenomyosis on
stored uterine volumes (operator 2)
Subjective Absent Mild Moderate Severe Total
assessment of Absent 16 1 1 0 18
severity of Mild 1 6 2 0 9
adenomyosis Moderate 0 0 3 0 3
on uterine Severe 0 0 1 5 6
volumes Total 17 7 7 5 36
(operator 1)
Linear-weighted kappa = 0.83 (excellent agreement)
95%CI (0.70-0.97)
Standard error of kappa = 0.067
 96

[3] HOW COMMON IS ADENOMYOSIS? A PROSPECTIVE STUDY OF

PREVALENCE USING TRANSVAGINAL ULTRASOUND IN A GYNAECOLOGY

CLINIC

3.1 BACKGROUND

The reliance on histological assessment of hysterectomy specimens for the diagnosis of

adenomyosis has presented many difficulties in accurately assessing its prevalence. The lack

of clear and uniform histological criteria raises concerns about the reproducibility of

histological diagnosis of adenomyosis and partially explains the variation in the proportions

of women diagnosed with adenomyosis in different studies (Bergholt et al, 2001).

Furthermore, the reliance on histology for diagnosis introduces a heavy selection bias as only

women who have undergone a hysterectomy can be diagnosed with the disease. This bias has

been further amplified by the reduction in the number of hysterectomies being performed for

benign disease (Reid and Mukri, 2005), resulting in the population of women from which

conclusions are drawn becoming increasingly less representative of the general population.

Continuous improvements in the resolution of TVUS have now enabled a more detailed

assessment of uterine architecture. This has facilitated the detection of subtle features of

adenomyosis, which could not have been seen with older equipment. TVUS, along with MRI,

has recently been shown to have good levels of accuracy in the pre-operative diagnosis of

adenomyosis (Dueholm 2006). TVUS is relatively inexpensive, widely available and well-

tolerated by patients. The aim of this study was to assess the prevalence of adenomyosis in a

population of women attending general gynaecological clinic using transvaginal ultrasound.

 97

3.2 METHODS

This was a prospective observational study of women attending our general gynaecology

clinic between January 2009 and January 2010. In all women, demographic data were

recorded and a detailed clinical history was taken prior to undertaking the ultrasound scan.

This included women’s age, ethnicity, body mass index (kg/m2), smoking history, age at

menarche, gravidity and parity (number of all prior pregnancies: miscarriages, terminations

of pregnancy and live births including mode of delivery), breastfeeding history (months of

breastfeeding), contraceptive history (length of time taking exogenous hormones, IUCD

usage) and menopausal status. Body mass index was calculated based on self-reported weight

and height. Women who were not certain of these measurements had their height and weight

measured using a calibrated scale and stadiometer.

Women were considered to have a history of endometriosis, if it had been diagnosed

previously on laparoscopy or on ultrasound scan within our department using the criteria

previously described in the methods section. In women who subsequently underwent

hysterectomy, ultrasound findings were compared with the final histological diagnosis.

Histopathological examinations were performed according to Molitor’s method (Molitor,

1971).

3.3 STATISTICAL ANALYSIS

Statistical analysis was performed, to examine possible associations between various

demographic and clinical variables and the presence of adenomyosis, using logistic

regression. Initially the individual effect of each factor upon the outcome was assessed in a

series of univariable analyses. Subsequently the joint effect of the variables was examined in

a multivariable analysis. Before the multivariable analysis, variance inflation factors (VIFs)
 98

were used to examine the colinearity between the variables. Statistical analysis was

performed using SPSS software (SPSS Inc., Chicago IL). Kappa analysis was used to

determine the level of agreement between ultrasound and histological diagnosis of

adenomyosis.

3.4 RESULTS

A total of 1066 consecutive women attended for clinical visits. None of them were pre-

menarchal or pregnant. 45/1066 [4.2% (95%CI: 3.8%-5.6%)] women were excluded from the

data analysis as they were unable to undergo a transvaginal scan. A further 36/1066 [3.4%

(95%CI: 2.5%-4.6%)] women were excluded as they had previously undergone a

hysterectomy. The median age of the 985 women included in the final data analysis was 40

(inter-quartile range: 32-48). 387/985 [39.3% (95% CI: 36.3%-42.4%)] women were

nulligravid and 529/985 [53.7% (95% CI: 50.6%-56.8%)] were nulliparous. 174/985 [17.7%

(95%CI: 15.4%-20.2%)] women were menopausal. The indications for clinical visits are

listed in Table 24. The diagnoses following ultrasound assessment are listed in Table 25. A

total of 206/985 [20.9% (95% CI: 18.5% - 23.6%)] women were diagnosed with

adenomyosis. In 78/206 (37.9%) of them, adenomyosis was an isolated abnormality, whilst in

128/206 (62.1%) of women concomitant abnormalities were found (Table 26).

The results of the univariable analysis examining associations of demographic and clinical

factors with adenomyosis are shown in Table 27. The multivariable analysis (Table 28)

showed that age, gravidity and pelvic endometriosis were all significantly associated with the

presence of adenomyosis. The prevalence of adenomyosis increased with age, reaching a

peak between the ages of 40 and 59, after which the prevalence decreased. Women with a
 99

previous or current diagnosis of endometriosis were significantly more likely to have

adenomyosis with an odds ratio of 4.06 (95% CI: 2.25-7.33).

Of the 985 women, 45 [4.6% (95% CI: 3.4%-6.1%)] underwent hysterectomy within two

years of their ultrasound scans. Of the 45 women who underwent hysterectomy, 19 did so for

suspected gynaecological malignancies, 9 for menorrhagia, 6 for pelvic organ prolapse, 5 for

urinary symptoms, 2 for pain symptoms and a further 2 had a hysterectomy prophylactically

as they had a genetic predisposition to hereditary non-polyposis colorectal cancer. Women

with uterine cancer (12 women with endometrial cancer and 2 with uterine sarcomas) or very

large fibroids (4 women) were not used for comparison between TVUS and histology as their

pathologies became the primary focus of histological examination. The invasive nature of the

uterine cancers interfered with assessment of the EMJ and the overall uterine size of those

women with very large numbers of fibroids made it difficult to obtain systematic

representative sections from every part of the pathological specimens. The sensitivity and

specificity of ultrasound in diagnosing adenomyosis in the remaining women was 81.8%

(95% CI: 52.3%-94.9%) and 81.3% (95% CI: 57.0%-93.4%). The positive predictive value

was 75% (95% CI: 42.8%-93.3%) and the negative predictaive value was 86.7% (95% CI:

58.4%-97.7%). A kappa analysis showed a good level of agreement between histology and

transvaginal ultrasound in the diagnosis of adenomyosis [Kappa=0.62 (p=0.001), 95% CI

(0.32-0.91)].
 100

Table 24 – Primary indications for scan (Naftalin et al. 2012b)

Indication for TVUS n (%)

Menorrhagia 121 (12.3)

Dysmenorrhoea 18 (1.8)
Menorrhagia / dysmenorrhoea 30 (3.0)
Pelvic pain 187 (18.9)
Infertility 129 (13.1)
Irregular bleeding / amenorrhoea 154 (15.6)
Postmenopausal bleeding 80 (8.1)
Other 266 (27.0)

Table 25 - Summary of ultrasound diagnoses recorded in the study population (n=985)

(Naftalin et al. 2012b)
Ultrasound diagnosis n (%)

No pathology 210 (48.7)

Adenomyosis 206 (20.9)
Fibroids 346 (35.1)
Endometriosis 63 (6.4)
Polycystic ovaries 164 (16.6)
Adnexal tumours 101 (10.3)
Thick endometrium in postmenopause 47 (4.8)
Endometrial polyps 45 (4.6)
Pelvic adhesions 26 (2.6)
Major congenital uterine anomalies 8 (0.8)
Other 9 (0.9)
 101

Table 26 - Concomitant abnormalities found in women with adenomyosis (n=206) (Naftalin

et al. 2012b)

Concomitant abnormalities n (%)

Fibroids 47 (22.8)
Endometriosis 10 (4.9)
Polycystic ovaries 10 (4.9)
Adnexal tumours 17 (8.3)
Thick endometrium in postmenopause 1 (0.5)
Endometrial polyps 2 (1.0)
Pelvic adhesions 1 (0.5)
Major congenital uterine anomalies 0 (0)
Two or more pathologies 40 (19.4)
No additional pathology 78 (37.9)

Total 206 (100)

 102

Table 27 – Results of univariable analysis looking at associations between demographic and clinical factors and
adenomyosis (Naftalin et al. 2012b)
Variable Category/term Adenomyosis, n (%) Odds ratio (95% CI) p-value
Agea Linear term 65.3 (19.4, 220) <0.001
Square term
Ethnicity Caucasian 119/640 (19) 1
Asian 25/80 (31) 1.99 (1.19, 3.32)
Afro-caribbean 35/149 (23) 1.34 (0.88, 2.06)
Oriental 8/38 (21) 1.17 (0.52, 2.61)
Middle eastern 10/40 (25) 1.46 (0.69, 3.07)
Mixed/other 7/38 (18) 0.99 (0.43, 2.30) 0.14
Smoking (pack years) 0 84/426 (20) 1
1-10 56/290 (19) 0.97 (0.67, 1.42)
>10 31/116 (27) 1.48 (0.92, 2.39) 0.21
BMIb (kg/m2) 1.20 (1.03, 1.41) 0.02
Gravidity 0 39/387 (10) 1
>1 165/598 (28) 3.40 (2.33, 4.95) <0.001
Gravidity (detailed) 0 39/387 (10) 1
1 38/187 (20) 2.28 (1.40, 3.70)
2 35/131 (27) 3.25 (1.96, 5.41)
3-5 65/220 (30) 3.74 (2.41, 5.81)
>6 27/60 (45) 7.30 (3.98, 13.4) <0.001
Parity 0 75/529 (14) 1
1 38/170 (22) 1.83 (1.18, 2.83)
2 50/151 (33) 3.14 (2.06, 4.78)
>3 44/135 (33) 3.07 (1.98, 4.75) <0.001
Vaginal deliveries 0 91/607 (15) 1
1 39/150 (26) 1.99 (1.30, 3.06)
2 45/127 (35) 3.11 (2.03, 4.77)
>3 29/101 (29) 2.28 (1.41, 3.71) <0.001
Caesarean deliveries 0 166/869 (19) 1
>1 38/116 (33) 2.06 (1.35, 3.15) 0.001
Miscarriages 0 141/772 (18) 1
>1 63/213 (29) 1.85 (1.30, 2.61) 0.001
ERPC or STOP 0 119/705 (17) 1
>1 85/280 (30) 2.15 (1.56, 2.96) <0.001
Breastfeeding <6 months 32/112 (29) 1
> 6 months 34/102 (33) 1.19 (0.76, 1.87) 0.44
Age at menarche 1.02 (0.93, 1.12) 0.71
Use of intrauterine No 100/519 (19) 1
contraceptive devices Yes 44/151 (29) 1.72 (1.14, 2.60) 0.01
Time on combined oral 0 77/328 (24) 1
contraceptive pill (years) <1 22/162 (14) 0.51 (0.30, 0.86)
1-5 46/215 (21) 0.88 (0.58, 1.34)
5-10 38/174 (22) 0.91 (0.58, 1.41)
>10 21/106 (20) 0.80 (0.47, 1.38) 0.15
Time on progesterone- 0 149/793 (19) 1
only contraception (years) <1 24/75 (32) 2.03 (1.21, 3.40)
1-5 21/87 (24) 1.37 (0.81, 2.31)
>5 10/30 (33) 2.15 (0.99, 4.70) 0.01
Menopausal status No 169/811 (21) 1
Yes 35/174 (20) 0.96 (0.64, 1.44) 0.83
Endometriosis No 179/922 (19) 1
Yes 25/63 (40) 2.73 (1.61, 4.64) <0.001
Fibroids No 127/639 (20) 1
Yes 74/346 (21) 1.08 (0.78, 1.49) 0.63
a
– Odds ratio given for 10-unit increase in explanatory variable
b
– Odds ratio given for 5-unit increase in explanatory variable
 103

Table 28 - Results of multivariate analysis looking at associations between demographic and

clinical variables and adenomyosis (Naftalin et al. 2012b)
Variable Category / term Odds Ratio (95% CI) P-value
Age (**) Linear term 34.3 (9.9, 118) <0.001
Squared term 0.70 (0.62, 0.80)

Gravidity 0 1
1 1.83 (1.09, 3.06)
2 2.46 (1.44, 4.30)
3-5 2.66 (1.62, 4.28)
6+ 4.90 (2.57, 9.35) <0.001

Endometriosis No 1
Yes 4.06 (2.25, 7.33) <0.001

(**) Odds ratios given for a 10-unit increase in explanatory variable

 104

[4] IS ADENOMYOSIS ASSOCIATED WITH MENORRHAGIA?

4.1 BACKGROUND

Original descriptions of adenomyosis reported an association with heavy and prolonged

periods (Cullen 1908). Several later studies reported similar findings (Emge et al. 1962; Bird

et al. 1972; Benson and Sneedon, 1958), but others have not shown significant differences in

the prevalence of adenomyosis between women with and without a history of menorrhagia

(Bergholt et al. 2001; Levgur et al. 2000; Parrazzini et al. 1997; Weiss et al. 2009). Most of

these studies used retrospective histological examination of hysterectomy specimens to

diagnose adenomyosis and compare it to clinical symptoms. The main problem with the use

of histology for the diagnosis of adenomyosis in these studies is the heavy selection bias

incurred (Mehasseb and Habiba 2009). Furthermore, in retrospective studies it is very

difficult to account for confounding variables such as concomitant uterine fibroids which

have an independent adverse effect on menstrual periods.

TVUS has recently been used for the non-invasive diagnosis of adenomyosis (Kepkep et al.

2007) and to study its prevalence (Naftalin et al. 2012b). There have been no large scale

prospective studies using non-invasive techniques to assess the link between adenomyosis

and clinical symptoms. The aim of this study was to investigate the possible association

between adenomyosis and menorrhagia.

4.2 METHODS

This was a prospective observational study of premenopausal women attending our general

gynaecology clinic. In all women, a detailed clinical history was taken prior to undertaking

the ultrasound scan. Women were asked about the frequency and duration of menstrual

periods and about any episodes of intermenstrual or postcoital bleeding. The amount of

menstrual loss was assessed subjectively by asking whether they felt their periods were
 105

excessively heavy or not. We also asked a subgroup of women to complete a PBAC in order

to obtain semi-quantitative information about the amount of menstrual loss. Women were

advised to complete the chart during their next period and then to return it back to the

department by post (a stamped addressed envelope was provided).

Women who were unable to undergo a TVUS or had previously undergone a hysterectomy

and those with a history of amenorrhoea or oligomenorrhoea with a cycle length greater than

60 days were excluded from the study. Women using contraceptives or taking medications

that would affect menstrual flow, women who attended for ultrasound scan during their IVF

cycle and those who underwent hysterectomy prior to their next period following the scan

were not asked to complete a PBAC.

4.3 STATISTICAL ANALYSIS

Statistical analysis was performed in two stages, first for subjective assessment of

menorrhagia and then for the semi-quantitative assessment of menstrual loss. Regression

methods were used to examine possible associations between various demographic and

clinical variables and menorrhagia. The subjective measure was a binary outcome, and so the

analysis of this outcome was performed using logistic regression. The PBAC scores were

measured on a continuous scale and so linear regression was used for the analysis. The scores

were found to have a heavily positively skewed distribution and so were given a log

transformation before analysis.

The analyses themselves were performed in two stages. Initially the separate association of

each factor with the outcome was examined in a series of univariable analyses. Subsequently

the joint effect of the variables upon each outcome was assessed in a multivariable analysis.

To limit the number of variables in this analysis, only factors with a univariable p-value of
 106

<0.2 were considered for this stage of the analysis. A backwards selection procedure was

used to retain only the statistically significant variables in the final model. As menstrual loss

was analysed on a log scale, the results are summarised in the form of ratios whereby the

ratios represent the PBAC values when the symptoms was present relative to when it was

absent. A Kappa analysis was used to analyse the agreement between the subjective

assessment of menorrhagia and the PBACs, where a PBAC score >100 was considered to be

consistent with menorrhagia.

4.4 RESULTS

A total of 892 consecutive premenopausal women attended for clinic visits between January

2009 and January 2010. 178 women were excluded (Figure 23) and 714 were entered into the

data analysis. Their median age was 38 (IQR 30-43). 305 women [42.7% (95% CI. 39.1%-

46.4%)] women were nulligravid and 424 [59.4% (95% CI. 55.7% - 63.0%)] were

nulliparous. Principal indications for ultrasound scans are listed in Table 29. The diagnoses

following ultrasound assessment are listed in Tables 30 & 31. All 714 women were entered

into data analysis for subjective assessment of menorrhagia. 529 women were asked to

complete PBACs and the analysis of the semi-quantitative assessment of menstrual loss was

performed in the 304 women who returned their charts. There was moderately good

agreement between the subjective binary assessment of menorrhagia and thePBAC

assessment of menstrual loss, where a PBAC score of >100 was considered to be consistent

with menorrhagia (k=0.529; 95% CI, 0.435 to 0.623) (Table 32).

The results of the univariable analysis examining the associations of demographic and

clinical variables with subjective assessment of menorrhagia are shown in Table 33. The

multivariable analysis (Table 34) showed that BMI, gravidity, fibroids, in particular sub-
 107

mucous fibroids and endometrial polyps were all significantly associated with menorrhagia.

Adenomyosis, when assessed as a binary outcome, was not significantly associated with

subjective assessment of menorrhagia. There was, however a significant positive correlation

between the number of features of adenomyosis on ultrasound scan and the volume of

menstrual loss expressed as a PBAC score, with each additional feature of adenomyosis being

associated with an average 22% [95% CI. 6% - 42% (p=0.005)] increase in the PBAC value

(Figure 24).

Two further multivariable analyses were performed to compare the number of different

morphological features of adenomyosis recorded on ultrasound in individual women and the

presence of subjectively assessed menorrhagia. The first showed a significant positive

relationship between the number of ultrasound features of adenomyosis and menstrual loss

(Table 35 – Model 1). The second multivariable analysis (Table 35 - Model 2) compared

women with no adenomyosis, women with <4 and those with >4 ultrasound features of

adenomyosis. This analysis showed women with >4 ultrasound features of adenomyosis, but

not women with <4 ultrasound features of adenomyosis, were significantly more likely to

suffer with menorrhagia. A breakdown of the number of ultrasound features of adenomyosis

recorded in the study population can be seen in Table 36. A final analysis looked at the

association between menorrhagia and the specific ultrasound features seen. The outcome

variable in the analysis was PBAC score which was measured on a log scale therefore the

outcomes are expressed as ratios (Table 37). The results for the individual features suggested

that only asymmetrical thickening and irregular EMJ were strongly associated with the PBAC

scores. The presence of these features was associated with a higher PBAC score.
 108

Figure 23 – Flowchart showing why women were excluded from data analysis (Naftalin et al. 2014)
892 women
assessed
Excluded from the study (178)
46 – could not tolerate TVUS
35 – prior hysterectomy
94 - Cycles absent or >60 days
3 - other
714 assessed for subjective menorrhagia

Excluded from PBAC (185)

122 – Using contraceptives that effect menstrual
flow
50 – Taking medications that effect menstrual flow
9 – Had hysterectomies prior to next period 529 women asked to fill in PBACs
4 – Had scans during IVF treatment cycles

225 women did not return PBACs

304 women returned completed PBACs

Table 29 – Principal indications for scan (n=714) (Naftalin et al. 2014)

Principal indication for scan n (%)
Menorrhagia 119 (16.7)
Menorrhagia / dysmenorrhoea 31 (4.3)
Intermenstrual / postcoital bleeding 68 (9.5)
Mild Oligomenorrhoea 66 (9.2)
Pelvic pain 129 (18.1)
Dysmenorrhoea 18 (2.5)
Dyspareunia 12 (1.7)
Subfertility 115 (16.1)
Recurrent miscarriage 9 (1.3)
Other 147 (20.1)

Table 30 - Summary of ultrasound diagnoses recorded in the study population (n=714)

(Naftalin et al. 2014)
Diagnosis n (%)
No uterine pathology 336 (47.1)
Adenomyosis 100 (15.4)
Fibroids 176 (24.6)
Endometrial polyp 32 (4.5)
Two pathologies 69 (9.7)
Three pathologies 1 (0.14)
Total 714 (100)
 109

Table 31 - Concomitant uterine abnormalities found in women with adenomyosis (n=157)

(Naftalin et al. 2014)
n (%)
Intramural/subserous fibroids 43 (27.4)
Submucous fibroids 10 (6.4)
Endometrial polyp 3 (1.9)
Two or more pathologies 1 (0.6)
No additional pathology 100 (63.7)
Total 157 (100)

Table 32 – Kappa analysis assessing agreement between subjective assessment of

menorrhagia and PBAC charts (where PBAC score >100 considered to be consistent with
menorrhagia) (Naftalin et al. 2014)

Subjective Subjectively
menorrhagia normal

PBAC score 119 48 167

>100

PBAC score 24 113 137

<100

143 161 304

k=0.529 (moderate agreement)

95% CI (0.429-0.620)
Standard error of kappa =0.048
 110

Table 33 – Results of univariable analysis looking at associations between demographic and

clinical factors and the subjective assessment of menorrhagia (n=714) (Naftalin et al. 2014)
Variable Category Menorrhagia Odds Ratio P-value
n (%) (95% CI)

Age (*) - 1.31 (0.19, 1.44) <0.001

BMI (*) - 1.50 (1.22, 1.84) <0.001

Ethnicity Caucasian 198/442 (45%) 1 <0.001

Asian 37/66 (56%) 1.57 (0.94, 2.65)
Afro-Caribbean 79/122 (65%) 2.26 (1.49, 3.43)
Oriental 7/26 (27%) 0.45 (0.19, 1.10)
Middle Eastern 15/26 (58%) 1.68 (0.75, 3.74)
Mixed 16/32 (50%) 1.23 (0.60, 2.53)

Gravidity 0 142/305 (47%) 1 <0.001

1 44/133 (33%) 0.57 (0.37, 0.87)
2-3 86/161 (53%) 1.32 (0.90, 1.93)
4+ 80/115 (70%) 2.62 (1.66, 4.14)

Parity 0 180/418 (43%) 1 <0.001

1 47/105 (45%) 1.07 (0.70, 1.65)
2 36/61 (59%) 1.90 (1.10, 3.28)
3+ 89/130 (68%) 2.87 (1.89, 4.36)

Adenomyosis No 256/558 (46%) 1 0.001

Yes 93/157 (59%) 1.89 (1.31, 2.71)

Any fibroids No 197/470 (42%) 1 <0.001

Yes 155/244 (64%) 2.41 (1.75, 3.32)

Submucous No 280/624 (45%) 1 <0.001

fibroids Yes 72/90 (80%) 4.91 (2.86, 8.43)

Fibroids None 197/470 (42%) 1 <0.001

(combined) Any fibroids 83/154 (54%) 1.62 (1.12, 2.33)
Submucous 72/90 (80%) 5.54 (3.20, 9.59)
fibroids

Endometrial No 320/664 (48%) 1 0.03

polyps
Yes 32/50 (64%) 1.91 (1.05, 3.47)

* Odds ratios given for 5-unit increase in explanatory variable

 111

Table 34 – Results of multivariable analysis looking at associations between demographic

and clinical factors and the subjective assessment of menorrhagia (n=714) (Naftalin et al.
2014)
Variable Category Odds Ratio (95% P-value
CI)

BMI (*) - 1.40 (1.13, 1.74) 0.002

Gravidity 0 1 <0.001
1 0.34 (0.18, 0.64)
2-3 1.01 (0.59, 1.74)
4+ 2.33 (1.26, 4.30)

Fibroids None 1 <0.001

(combined)
Any fibroids 1.53 (0.91, 2.58)
Submucous fibroids 5.60 (2.69, 11.6)

Endometrial polyps No 1 0.02

Yes 2.81 (1.15, 11.7)

* Odds ratios given for 5-unit increase in explanatory variable

 112

Table 35 – Results of multivariable analyses looking at the associations between

demographic and clinical factors and the subjective assessment of menorrhagia, where the
number of ultrasound features of adenomyosis was treated as a continuous variable (Model 1)
and where the number of ultrasound features was treated as a categorical variable (Model 2)
(n=714) (Naftalin et al. 2014)
Variable Category Odds Ratio (95% P-value
CI)

Model 1

BMI (*) - 1.38 (1.11, 1.71) 0.004

Gravidity 0 1 <0.001
1 0.31 (0.16, 0.60)
2-3 0.84 (0.48, 1.48)
4+ 1.77 (0.92, 3.39)

Adenomyosis features - 1.21 (1.04, 1.40) 0.01

Fibroids (combined) None 1 <0.001

Any fibroids 1.53 (0.90, 2.58)
Submucous 6.19 (2.96, 12.9)
fibroids

Endometrial polyps No 1 0.02

Yes 2.99 (1.21, 7.40)

Model 2

BMI (*) - 1.39 (1.11, 1.73) 0.003

Gravidity 0 1 <0.001
1 0.31 (0.16, 0.59)
2-3 0.91 (0.51, 1.61)
4+ 2.01 (1.04, 3.92)

Adenomyosis features None 1 0.002

(categorical) 1–3 0.73 (0.39, 1.36)
4+ 3.80 (1.62, 8.91)

Fibroids (combined) None 1 <0.001

Any fibroids 1.49 (0.88, 2.53)
Submucous 6.16 (2.93, 12.9)
fibroids

Endometrial polyps No 1 0.02

Yes 2.87 (1.16, 7.11)

* Odds ratios given for 5-unit increase in explanatory variable

 113

Table 36 –The number of ultrasound features seen in the study population (n=157) (Naftalin
et al. 2014)
Number of ultrasound features of n (%)
adenomyosis seen

1 18 (11.5)
2 37 (23.6)
3 41 (26.1)
4 32 (20.3)
5 16 (2.2)
6 10 (10.2)
7 3 (1.9)

Figure 24 – The relationship between increasing numbers of ultrasound features of

adenomyosis and PBACassessed menstrual loss (n=304) (Naftalin et al. 2014)
350
300
250
PBAC
200
150
100

0 2 4 6 8
Number of adenomyosis features
 114

Table 37 – Results of the statistical analysis looking at the effect of the number and type of
ultrasound features on the PBAC assessment of menorrhagia (n=304) (Naftalin et al. 2014)
Outcome Ratio (95% CI) P-value

Number of features 1.22 (1.06, 1.42) 0.005

Asymmetrical thickening 2.24 (1.36, 3.68) 0.002

Parallel shadowing 1.63 (0.80, 3.37) 0.18
Linear striations 0.96 (0.56, 1.63) 0.88
Myometrial Cysts 1.62 (0.99, 2.68) 0.06
Hyperechoic lesions 0.82 (0.45, 1.51) 0.53
Adenomyomas 1.07 (0.60, 1.89) 0.82
Irregular EMJ 1.83 (1.10, 3.05) 0.02
 115

[5] IS ADENOMYOSIS ASSOCIATED WITH DYSMENORRHOEA?

5.1 BACKGROUND

Original descriptions of adenomyosis reported an association between the disease and ‘a great

deal of pain’ (Cullen, 1908). Several later studies reported similar findings (Emge 1962; Bird

et al. 1972; Benson and Sneedon 1958), but others have not shown significant differences in

the prevalence of adenomyosis between women with and without a history of dysmenorrhoea

(Bergholt et al. 2001; Parrazzini et al. 1997; Weiss et al. 2009). Most of these studies used

retrospective histological examination of hysterectomy specimens to diagnose adenomyosis

and compare it to clinical symptoms. The main problem with the use of histology for the

diagnosis of adenomyosis in these studies is the heavy selection bias incurred (Mehasseb and

Habiba, 2009). Furthermore, in retrospective studies it is very difficult to account for

confounding variables such as endometriosis which are considered to have an independent

adverse effect on menstrual pain.

TVUS has recently been used for the non-invasive diagnosis of adenomyosis (Kepkep et al.

2007) and to study its prevalence (Naftalin et al. 2012b). There have been no large scale

prospective studies using non-invasive techniques to assess the link between adenomyosis

and clinical symptoms. The aim of this study was to investigate the possible association

between adenomyosis and dysmenorrhoea.

5.2 METHODS

This was a prospective observational study of premenopausal women attending our general

gynaecology clinic. In all women, a detailed clinical and demographic history was taken prior

to undertaking the ultrasound scan. In addition, an assessment was made as to how painful

their periods were using an 11-point NRS. Women who were unable to undergo a
 116

transvaginal scan or had previously undergone a hysterectomy and those with a history of

amenorrhoea or oligomenorrhoea with a cycle length greater than 60 days were excluded

from the study.

5.3 STATISTICAL ANALYSIS

Regression methods were used to examine factors associated with the subjective assessment

of dysmenorrhoea. As the NRS scores were measured on a continuous scale, linear regression

was used for the analysis. The scores were found to be approximately normally distributed

and thus no transformation of the scores was given. The analysis was performed in two

stages. Initially the separate association of each factor with the outcome was examined in a

series of univariable analyses. Subsequently the joint effect of the variables upon each

outcome was assessed in a multivariable analysis. To limit the number of variables in this

analysis, only factors with a univariable p-value of <0.2 were considered for this stage of the

analysis. A backwards selection procedure was used to retain only the statistically significant

variables in the final model. Two variables, gravidity and parity, were found to have a small

number of very high values. As a result these two variables were categorised for the purposes

of analysis. Statistical analysis was performed, using SPSS software (SPSS Inc., Chicago IL).

5.4 RESULTS

A total of 892 consecutive premenopausal women attended for clinic visits between January

2009 and January 2010. 178 women were excluded (Figure 25) and 714 were entered into the

data analysis. Their median age was 38 (IQR 30-43). 305/714 [42.7% (95% CI. 39.1%-

46.4%)] women were nulligravid and 424/714 [59.4% (95% CI. 55.7% - 63.0%)] were
 117

nulliparous. Primary indications for ultrasound scans are listed in Table 38. The diagnoses

following ultrasound assessment are listed in Tables 39 & 40.

The results of the univariable analysis examining the associations of demographic and

clinical variables with subjective assessment of dysmenorrhoea showed that there was a

strong positive association between NRS score and the presence of both endometriosis and

adenomyosis (Table 41). Women with adenomyosis had higher NRS scores, on average 1.1

units higher than those without adenomyosis. Women with endometriosis had NRS scores

that were on average 1.6 units higher than. No other variables were significantly associated

with NRS scores.

The multivariable analysis (Table 42) showed that, as in the univariable analyses, both

adenomyosis and endometriosis had a significant association with dysmenorrhoea, as

measured by NRS. The presence of either of these conditions was associated with higher pain

scores. After adjusting for the effects of the other variable, the size of effects of each variable

were slightly reduced from the size of effect observed in the univariable analyses. A linear

regression analysis was performed assessing the association between both the number of

ultrasound features of adenomyosis present and the specific ultrasound features seen. The

outcome variable in the analysis was NRS score which was measured on a continuous scale.

The NRS score was found to be approximately normally distributed. Linear regression was

used to, in turn, to examine the effect of the total number of adenomyosis features, and the

individual features, upon the outcome. The analysis of the NRS score included adjustments

for both treatment and endometriosis, as these were thought to be potentially confounding

factors. The shape of the relationship between the total number of features present and the

outcome was examined. If the relationship was not found to be linear (a straight line), a

curved relationship was assumed, which was achieved by adding a squared term into the
 118

analysis. The results suggest that the number of ultrasound features of adenomyosis present

was significantly associated with the NRS scores (Table 43). The relationship can be seen in

Figure 26 which shows the fitted relationship between the number of ultrasound features of

adenomyosis present and NRS score. The graph suggests that the number of features was not

strongly associated with NRS score below around 4 features, however once the number of

features was above this level, there was an increase in the pain score with increased number

of features. The results for the individual features suggested that only asymmetrical

thickening and irregular EMJ were strongly associated with the NRS scores. The presence of

these features was associated with a higher NRS score. The largest effect was for irregular

EMJ, where the presence of this symptom was associated with an increase of 1.6 units in the

NRS score. A breakdown of the number of ultrasound features of adenomyosis seen in the

study population can be seen in Table 44.

Figure 25 – Flowchart showing why women were excluded from data analysis.
 119

Table 38 – Primary indications for scan (n=714)

Principal indication for scan n (%)
Menorrhagia 119 (16.7)
Menorrhagia / dysmenorrhoea 31 (4.3)
Intermenstrual / postcoital bleeding 68 (9.5)
Mild Oligomenorrhoea 66 (9.2)
Pelvic pain 129 (18.1)
Dysmenorrhoea 18 (2.5)
Dyspareunia 12 (1.7)
Subfertility 115 (16.1)
Recurrent miscarriage 9 (1.3)
Other 147 (20.1)

Table 39 Summary of ultrasound diagnoses recorded in the study population (n=714)

Ultrasound diagnoses n (%)
No uterine pathology 336 (47.1)
Adenomyosis 100 (15.4)
Fibroids 176 (24.6)
Endometrial polyp 32 (4.5)
Two pathologies 69 (9.7)
Three pathologies 1 (0.14)
Total 714 (100)

Table 40 - Concomitant uterine abnormalities found in women with adenomyosis (n=157)

Concomitant uterine abnormalities n (%)
Intramural/subserous fibroids 43 (27.4)
Submucous fibroids 10 (6.4)
Endometrial polyp 3 (1.9)
Two or more pathologies 1 (0.6)
No additional pathology 100 (63.7)
Total 157 (100)
 120

Table 41 - Results of univariable analysis looking at associations between demographic and

clinical factors and the subjective assessment of dysmenorrhoea (n=714)
Variable Category Coefficient (95% CI) P-value

Age (*) - -0.04 (-0.17, 0.09) 0.53

BMI (*) - 0.24 (-0.04, 0.51) 0.09

Ethnicity Caucasian 0 0.10

Asian 0.71 (-0.04, 1.46)
Afro-Caribbean 0.36 (-0.22, 0.93)
Oriental -0.91 (-2.06, 0.24)
Middle Eastern 0.28 (-0.87, 1.43)
Mixed 0.77 (-0.27, 1.82)

Gravidity 0 0 0.99
1 -0.03 (-0.62, 0.57)
2-3 0.02 (-0.53, 0.58)
4+ 0.02 (-0.61, 0.64)

Parity 0 0 0.89
1 -0.20 (-0.82, 0.42)
2 -0.11 (-0.89, 0.68)
3+ -0.17 (-0.75, 0.40)

Adenomyosis No 0 <0.001
Yes 1.07 (0.56, 1.58)

Any fibroids No 0 0.38

Yes 0.20 (-0.25, 0.65)

SM fibroids No 0 0.25
Yes -0.37 (-1.02, 0.27)

Fibroids (combined) None 0 0.11

Non SM fibroids 0.47 (-0.06, 1.00)
SM fibroids -0.26 (-0.92, 0.39)

Endometriosis No 0 <0.001
Yes 1.57 (0.77, 2.37)

(*) Regression coefficients given for a 5-unit increase in explanatory variable

 121

Table 42 - Results of multivariable analysis looking at associations between demographic and

clinical factors and the subjective assessment of dysmenorrhoea (n=714)
Variable Category Coefficient (95% CI) P-value

Adenomyosis No 0 <0.001
Yes 0.94 (0.43, 1.46)

Endometriosis No 0 0.001
Yes 1.36 (0.56, 2.17)

Figure 26 – Relationship between the number of ultrasound features of adenomyosis present

and NRS score.
10
VAS score

8
6

0 2 4 6 8
Number of adenomyosis features
 122

Table 43 - Results of the statistical analysis looking at the effect of the number and type of
ultrasound features on the subjective assessment of dysmenorrhoea (n=304)
Outcome Term Coefficient (95% CI) P-value

Number of features Linear -0.7 (-1.8, 0.4) <0.001

Squared 0.2 (0.0, 0.3)

Asymmetrical thickening - 1.1 (0.2, 2.1) 0.02

Parallel shadowing - 1.2 (-0.1, 2.5) 0.08
Linear striations - 0.4 (-0.6, 1.3) 0.46
Myometrial Cysts - 0.3 (-0.6, 1.3) 0.47
Hyperechoic lesions - 0.3 (-0.8, 1.3) 0.62
Adenomyomas - 0.1 (-1.0, 1.2) 0.86
Irregular EMJ - 1.6 (0.7, 2.6) 0.001

Table 44 – The number of ultrasound features seen in the study population (n=157)
Number of ultrasound features of n (%)
adenomyosis seen
1 18 (11.5)
2 37 (23.6)
3 41 (26.1)
4 31 (19.7)
5 16 (2.2)
6 10 (10.2)
7 3 (1.9)
 123

[6] EVALUATION OF A SECOND-STAGE TEST DESIGNED TO IMPROVE THE

ACCURACY OF TRANSVAGINAL ULTRASOUND IN THE DIAGNOSIS OF

ENDOMETRIAL CANCER

6.1 BACKGROUND

Endometrial cancer is the commonest cancer of the female genital tract in the developed

world (Sankaranarayanan and Ferlay 2006). Ultrasound measurement of endometrial

thickness is commonly used to triage women with post-menopausal bleeding for histological

sampling. The cut-offs for endometrial sampling vary between different authors but >5mm is

a commonly used cut-off. The sensitivity of this cut-off for diagnosing endometrial cancer is

very high but the specificity is just 61% (CI: 59%-63%) (Smith-Bindman et al. 1998).

Therefore the positive predictive value of endometrial thickness measurement in diagnosing

endometrial cancer is poor and many women therefore undergo ultimately unnecessary

invasive procedures.

A number of attempts have been made to improve the specificity of ultrasound in diagnosing

endometrial cancer. Many studies have assessed whether different endometrial morphologies

are more common in malignant lesions (Epstein et al. 2001; Epstein et al. 2002, Epstein and

Valentin 2006, Randelzhofer et al. 2002). Other authors have looked at various Doppler

parameters to assess whether they aided diagnosis of endometrial cancer (Bourne et al. 1991;

Epstein and Valentin 2006).

The aim of our study was to assess whether a test that combined ultrasound examination of

endometrial morphology and the integrity of the EMJ, with Doppler examination of the

vasculature supplying endometrial lesions, could improve the specificity of ultrasound in

 124

diagnosing endometrial cancer in women with postmenopausal bleeding and increased

endometrial thickness on TVUS.

6.2 METHODS

This was a prospective observational study of women with a history of postmenopausal

bleeding who attended our rapid-access clinic. In all women, a clinical history was taken and

documented on a dedicated database (PIA Fetal Database, version 5.5.4.152, Viewpoint

Bildverarbeitung GmbH, Munich, Germany). They all underwent clinical assessment which

included speculum examination to exclude local vaginal and cervical abnormalities as a cause

of bleeding prior to a TVUS. A smear test was performed only in women younger than 65

who were not up-to-date with their smears. Women who had not been sexually active and

those with severe atrophic vaginal changes were offered a transrectal scan. Women in whom

transvaginal or transrectal scans were not possible and those in whom ultrasound images of

the uterine cavity were sub-optimal were excluded from the study and advised to undergo

hysteroscopy. Women were also excluded from the study if they were taking hormone

replacement therapy (HRT) or if they had had endometrial sampling performed within the six

months prior to attending the clinic.

In women with suspected endometrial polyps or an endometrial thickness >5mm, a second-

stage test was performed at the same time as the initial ultrasound examination. The second

stage test involved a detailed subjective assessment of endometrial morphology, the integrity

of the EMJ alongside a Doppler assessment of the endometrium. The endometrium was

classified as being either homogeneous or heterogeneous in the absence of focal pathology. If

focal pathology was present it was classified as being either regular or irregular. The EMJ

was assessed and classified as being either regular or irregular.

 125

For the purposes of the study, the assessment of endometrial morphology was classified as

suspicious if two criteria were met. The first criterion was that either focal pathology was

present and irregular or, if no focal pathology was present, that the endometrium was

heterogenous in echotexture (Figure 27). The second criterion was that there must be

evidence that any lesion breached the EMJ (Figure 28). With regard to the Doppler

assessment, lesions with no detectable blood supply and those with a single feeding vessel

were classified as benign (Figure 29) whilst those supplied by multiple vessels crossing the

EMJ were considered suspicious (Figure 30).

Only women in whom both the endometrial morphology and the colour Doppler assessment

were considered suspicious were classified, for the purposes of the study, as high risk for

endometrial cancer. Patient management continued as per our clinical protocol and the

ultrasound diagnosis was compared with the final histological diagnosis.

6.3 STATISTICAL ANALYSIS

A database file was set up using Microsoft Excel (Redmond, WA, US) for Windows to

facilitate data entry and retrieval. Data were analysed using 2x2 tables to assess sensitivity

and specificity.

6.4 RESULTS

125 women were assessed between September 2009 and November 2010. Eighteen women

were excluded as they were using hormone replacement therapy. A further 7 women were

excluded from statistical analysis; the endometrium could not be fully assessed on ultrasound

in 4 of them, 2 of them declined surgery and 1 transferred her care to her local hospital

following the ultrasound scan. Thus, 100 women were included in the study. Demographic
 126

characteristics of the women included and excluded are shown in Table 45. The median age

of women studied was 57.5 (inter-quartile range: 53-64). 20/100 [20% (95% CI: 12.2%-

27.8%)] women were nulliparous. There was a high burden of co-morbidity with 32% (95%

CI: 22.9%-41.1%) having hypertension, 12% (95% CI: 5.6%-18.4%) being hypothyroid, 12%

(95% CI: 5.6%-18.4%) taking anti-coagulants and 12% (95% CI: 5.6%-18.4%) being

diabetic.

Of the 100 women included, 49 had an endometrial thickness <5mm and no focal pathology.

These women were reassured, discharged and advised to return should they experience any

further vaginal bleeding. A histological diagnosis was obtained in all of the 51 women with

either an endometrial thickness >5mm or focal pathology. The final diagnosis was obtained

by pipelle in 15 cases and in a further 5 cases, pipelle sampling was inadequate and the

women had to undergo an outpatient hysteroscopy to achieve a final diagnosis.

10/11 (90.9%) women eventually diagnosed with endometrial cancer were correctly

identified as high risk for endometrial cancer using the second-stage test (Table 46). The one

false negative was a woman with a carcinosarcomatous endometrial polyp. While this had an

unusual appearance on ultrasound, it did not fit the criteria for ultrasound diagnosis of

endometrial cancer according to the study protocol, as the EMJ was intact and only a single

blood vessel could be seen crossing the basal endometrium on colour Doppler assessment.

There were three false positives. All three of these women had fibroids and two of them had

adenomyosis.

The sensitivity and specificity of the second-stage test in diagnosing endometrial cancer were

90.9% (95% CI: 62.3% - 98.4%) and 96.6% (95% CI: 90.1% - 98.9%) respectively. 38

women were classified as low risk for endometrial cancer of whom 37 had benign disease,

giving a negative likelihood ratio of 0.094. Of the 13 women classified as high risk for
 127

endometrial cancer, 10 were subsequently diagnosed with endometrial cancer, giving a

positive likelihood ratio of 27.0. The effect of the second-stage test on the risk of endometrial

cancer can be seen in Figure 31.

When endometrial morphology alone was assessed, alongside endometrial thickness, the

sensitivity and specificity in diagnosing endometrial cancer were 90.9% (95% CI: 62.3% -

98.4%) and 95.5% (95% CI: 89% - 98.2%). The sensitivity was unchanged when Doppler

assessment was analysed alone, alongside endometrial thickness and the specificity dropped

slightly to 94.4% (95% CI: 87.5% - 97.6%).

Figure 27 – Longitudinal view of a uterus with a heterogenous endometrium. The patient was
subsequently diagnosed with endometrial cancer.
 128

Figure 28 – Coronal view of a uterus where a hyperechoic tumour can be seen filling the right
fundal portion of the endometrial cavity. A breach in the EMJ can be seen in the right
cornuum (white arrow)

Figure 29 – Longitudinal view of a uterus using B-mode ultrasound with colour Doppler
showing an endometrial polyp with single feeder vessel. The polyp was found to be benign.
 129

Figure 30 - Longitudinal view of a uterus using B-mode ultrasound with colour Doppler
showing multiple vessels crossing the EMJ at the fundus. A pipelle biopsy subsequently
revealed endometrial cancer.
 130

Table 45 – Demographics of women included and excluded from the study

Characteristic Included Excluded Total
(n=100) (n=25) (n =125)
Age in years (median; interquartile 57.5 (53-64) 57 (53-63) 57 (53-64)
range)

Endometrial thickness mm (range) 6.3 (0.8-56.3) 7.5 (1.4-10.5) 7.4 (0.8-56.3)

CI (4.8-7.8) (1.9-13.1) (5.9-8.9)

Nulliparous n (%) 20 (20.0) 7 (28.0) 27 (21.6)

Hypertension n (%) 32 (32.0) 7 (28.0) 39 (31.2)
Diabetes n (%) 12 (12.0) 2 (8.0) 14 (11.2)
Use of anticoagulants n (%) 12 (12.0) 1 (4.0) 13 (10.4)
Use of levothyroxine n (%) 12 (12.0) 1 (4.0) 13 (10.4)

Final diagnosis
Malignant n (%) 11 (11.0) 1 (4) 12 (9.6)

Histological diagnosis obtained 51 19 70

Pipelle 15 3 18
Outpatient hysteroscopy 12 2 14
Inpatient hysteroscopy 23 8 31
Hysterectomy 1 2 3

Final histological diagnosis

Atrophic/benign endometrium 16
Benign endometrial polyp 21
Simple hyperplasia 3
Complex hyperplasia 0
Cancer 11

Histological subtype
Endometrioid adenocarcinoma 10
Carcinosarcoma 1

Stage of Malignancies
I 7
II 1
III 2
IV 1
 131

Table 46 – Table showing how the endometria of women subsequently diagnosed with
endometrial cancer were classified at the time of their original ultrasound scan.

Stage Endometrial Endometrium EMJ Number of Ultrasound

thickness (mm) vessels classification
Ia 8.3 Irregular Intact >2 Malignant
Ia 12.6 Irregular Disrupted >2 Malignant
Ia 12.9 Irregular Disrupted >2 Malignant
Ia 14.5 Irregular Intact >2 Malignant
Ia 17.7 Irregular Intact >2 Malignant
Ib 13 Regular Disrupted >2 Malignant
Ib 26.1 Regular Disrupted >2 Malignant
IIb 13.6 Irregular Disrupted >2 Malignant
IIIa 5 Irregular Disrupted >2 Malignant
IIIb 31.4 Regular Intact 1 Benign
IV 56.3 Irregular Disrupted >2 Malignant

Figure 31 – Flowchart showing risks of endometrial cancer depending on results of TVUS

 132

IV Discussions
 133

[1] INTER- AND INTRAOBSERVER VARIABILITY IN THREE-DIMENSIONAL

ULTRASOUND ASSESSMENT OF THE ENDOMETRIAL-MYOMETRIAL

JUNCTION

1.0 DISCUSSION

In this study, there was complete agreement between operators in classifying how well the

EMJ could be visualised as optimal or sub-optimal, where sub-optimal included both

satisfactory and unsatisfactory views. The discrepancies between operators occurred where

EMJs were classified as either satisfactory or unsatisfactory. This is perhaps unsurprising as

judging whether most of the EMJ can be seen or not is more subjective and hence more likely

to suffer from inter-observer variability, than judging whether all of the EMJ can be seen.

Despite the observed variability, the overall reproducibility of examination was very good

with kappa values between 0.70 and 0.83. This indicates that assessment of the EMJ on 3D

scan is reproducible enough to be used in clinical practice. This is important because

assessment of the regularity or integrity of the EMJ is increasingly being used to make

clinical diagnoses, mostly for the ultrasound diagnosis of adenomyosis (Ahmed et al. 2007;

Exacoustos et al. 2011) but also for the diagnosis of endometrial cancer (Randelzhofer et al.

2002). A reasonable reproducibility in the assessment of the EMJ should be considered a

prerequisite if clinicians are to base clinical decisions on ultrasound findings. A criticism

could be that the classification system used in this study was entirely arbitrary however there

was no pre-existing classification system for assessment of the EMJ that could have been

used. Furthermore, a similar approach has been used in the past when assessing

reproducibility of ultrasound diagnosis of adnexal masses based on subjective evaluation

(Yazbek 2010). It must be remembered that these results apply only to women with normal
 134

uteri. Further work is needed to see if our findings are reproducible in different settings and

different populations, more specifically, women with endometrial or myometrial pathology.

In conclusion, this study shows that assessment of EMJ visualisation has both good inter- and

intra-observer variability in women with normal uteri.

 135

[2] FACTORS AFFECTING VISUALISATION OF THE EMJ ON THREE-

DIMENSIONAL ULTRASOUND

2.0 DISCUSSION

This study showed that the ability to analyse the EMJ improves with increasing endometrial

thickness. This finding is consistent with MRI studies that show that EMJ thickness increases

in parallel with, although not to the same extent as, endometrial thickness (Novellas 2011).

Ultrasound studies have shown that endometrial echogenicity also increases in parallel with

endometrial thickness, particularly in the luteal phase of the cycle (Fleischer 1986). This

would increase the visual contrast between the endometrium and the myometrium, which

could further facilitate visualisation of the EMJ.

A more practical reason why increasing endometrial thickness enables better EMJ

visualisation is that it allows greater flexibility in producing a clear rendered 3D coronal

image. With a thin endometrium, curvature or asymmetry of the cavity can make it harder to

find a plane through which the EMJ can be seen continuously. A thicker endometrium, and

hence endometrial cavity, minimises this effect.

Other work, including MRI studies, has shown that the EMJ changes in both appearance and

size at different stages of the menstrual cycle (McCarthy et al. 1986; Janus et al. 1988;

Wiczyk et al. 1988). While our results were unable to show that stage of cycle had a

significant effect on EMJ visualisation, this may be because the numbers were too small to

determine a difference between the proliferative stage of the cycle and the luteal phase of the

cycle, once post-menopausal women, women taking the pill and women with irregular cycles

were excluded from the analysis.

 136

We also found that the ability to see the EMJ decreases with increasing parity. The finding

that parity has an effect on the EMJ is not unexpected. We know that the EMJ is involved in

and changed by placentation (Brosens et al. 2010) and that there are differences in the

appearance of the sub-endometria of nulliparous and multiparous women, as measured on 3D

power Doppler angiography (Raine-Fenning et al. 2004). One might surmise that this effect

was due to the presence of early or undiagnosed adenomyosis, a condition for which

increased parity is thought to be a risk factor and which is known to affect the EMJ. This is

purely speculative however, as none of the women included in the statistical analysis had any

ultrasound features of adenomyosis, despite detailed assessment of the myometrium.

The study by Raine-Fenning et al also showed a statistically significant negative correlation

between age and sub-endometrial vascularity (Raine-Fenning et al. 2004). MRI studies have

reported a change in EMJ size as age increases (Novellas 2011). We found that age did not

have a statistically significant effect on EMJ visualisation. Comparisons between these

studies are difficult however, as they either used different imaging modalities or assessed

subtly different aspects of the EMJ. These two difficulties endure throughout the literature

and are compounded by a paucity of published material assessing EMJ visualisation.

Further work is needed to look at what factors, both physiological and pathological, affect the

EMJ.
 137

[3] INTER- AND INTRAOBSERVER VARIABILITY IN THE ULTRASOUND

DIAGNOSIS OF ADENOMYOSIS

3.0 DISCUSSION

There was good intra-observer variability when comparing real-time B-mode ultrasound scan

assessment and assessment of stored 3D uterine volumes for the diagnosis of adenomyosis,

with a kappa value of 0.67. There was also good inter-observer variability when two different

operators independently used assessment of stored 3D uterine volumes to diagnose

adenomyosis, with a kappa value of 0.61. The intra- and inter-observer variability remained

good or better when assessing agreement in the subjective classification of severity of

adenomyosis on stored 3D uterine volumes, or in evaluating the number of ultrasound

features of adenomyosis seen on stored 3D uterine volumes. When inter-observer variability

was assessed there were seven disagreements in diagnosis. In all seven of these cases, the

adenomyosis was subjectively classified as mild meaning that there was complete agreement

in the diagnosis of adenomyosis, where adenomyosis was considered to be moderate or

severe. Given that data from this thesis suggests that adenomyosis only becomes clinically

significant, at least with regard to menorrhagia and dysmenorrhoea, until more than 2

ultrasound features are present, this means that there was complete agreement in the

diagnosis of clinically significant adenomyosis.

The main strength of this study is that it was a prospective study with pre-defined objectives

that assessed both inter- and intra-observer agreement. Furthermore, it not only assessed

agreement of ultrasound diagnosis of adenomyosis, but also agreement in the number of

ultrasound features seen as well as agreement in classification of severity of adenomyosis.

A potential weakness is that the numbers are relatively small. Another criticism might be that

the prevalence of adenomyosis in the population studied was much higher (61.1%) than is
 138

likely to be present in a general gynaecology clinic. This was done in order to ensure

inclusion of stored 3D uterine volumes with a wide range in the number of ultrasound

features seen in the original real-time scans so as to ensure that reproducibility of

classification of severity of adenomyosis could be evaluated and to see whether subjective

ultrasound classification of severity of adenomyosis influenced reproducibility of diagnosis

of adenomyosis, which it did.

There has only been one other study assessing the inter-observer variability in diagnosis of

adenomyosis on transvaginal ultrasound. This study found a considerably lower level of

inter-observer agreement (kappa=0.38) than we found in our study (Dueholm et al. 2002).

The study in question collected their data around ten years prior to data collection in our

study and it may be that technical improvements in TVUS during that time period have led to

increased resolution and therefore better visualisation of the ultrasound features of

adenomyosis.

Consistent with this, is our finding that the intra-observer agreement was better when

comparing two assessments using stored 3D volumes, then when comparing real-time

ultrasound scan and an assessment using a stored uterine volume. The higher resolution of

real-time ultrasound scan may have enabled visualisation of some of the more subtle

ultrasound features that were not able to be visualised subsequently on 3D stored volumes.

This may be one of the reasons why some authors have stated that real-time ultrasound

scanning is mandatory for the accurate diagnosis of adenomyosis (Reinhold et al. 1996). Our

finding that there was good intra-observer agreement between real-time ultrasound scan and

stored 3D uterine volume assessment, performed over 3 years apart suggests that accurate

diagnosis is possible without real-time ultrasound. This has clinical relevance as significant

expertise in gynaecological ultrasound is required in order to make a diagnosis of

adenomyosis. If the diagnosis can be made from assessment of stored 3D uterine volumes,
 139

then less experienced sonographers who are uncertain of a diagnosis could store a 3D uterine

volume for assessment by an experienced gynaecological sonographer at a later date. This has

implications for service provision as it means that experienced sonographers do not have to

be constantly available to supervise all scans in order for a diagnosis of adenomyosis to be

made. Furthermore, patients would not need to undergo a second transvaginal ultrasound scan

in order to confirm the presence of adenomyosis, if the diagnosis can be made on stored

uterine volumes.

The inter-observer variability of visualisation of individual ultrasound features of

adenomyosis was generally good with 3 of the seven features displaying moderate agreement,

2 displaying good agreement and one displaying perfect agreement. Two of the three features

with the highest inter-observer variability were asymmetrical myometrial thickening (kappa

value 0.77) and an irregular EMJ (kappa value 0.65). Data later on in this thesis shows that

the presence of these two ultrasound features is independently associated with both

dysmenorrhoea and increasing PBAC-assessed menstrual loss. Future studies should assess

the real-time inter-observer agreement of ultrasound diagnosis of adenomyosis as well as the

real-time inter-observer agreement of the individual ultrasound features of adenomyosis.

 140

[4] HOW COMMON IS ADENOMYOSIS? A PROSPECTIVE STUDY OF

PREVALENCE USING TRANSVAGINAL ULTRASOUND IN A GYNAECOLOGY

CLINIC

4.0 DISCUSSION

The overall prevalence of adenomyosis in our study population was 20.9%. The prevalence

increased with age reaching a peak of 32% in women aged 40-49. The prevalence was not

significantly different in pre- and postmenopausal women. On multivariate analysis we found

that the presence of adenomyosis was significantly associated with age, gravidity and

endometriosis.

The main strength of this study is that it was a large prospective study with pre-defined

objectives and clear diagnostic criteria. All examinations were performed by a single

observer using high quality, top of the range equipment, which ensured a consistent approach

to data collection and ultrasound examination, eliminating concerns regarding inter-observer

variability. We have also recorded a large number of demographic and clinical variables

which has enabled us to assess their possible effects on the prevalence of adenomyosis. A

potential weakness is the lack of histological confirmation of diagnosis in the majority of

women. In women who underwent a hysterectomy however, histological examination was

also performed in a detailed, standardised fashion. The good agreement between pre-

operative ultrasound diagnosis of adenomyosis and histological findings is reassuring and it

confirms that targeted ultrasound examination can be used in clinical practice for non-

invasive diagnosis of adenomyosis.

The majority of previous studies evaluating the prevalence of adenomyosis assessed

populations of women who underwent hysterectomy. The sample sizes were typically very
 141

small and they included mainly women with severe symptoms, who were more likely to have

adenomyosis than asymptomatic women and those with mild symptoms. In view of this it is

likely that the prevalence of adenomyosis in these studies was overestimated. Our estimate of

prevalence of adenomyosis is likely to be closer to the true prevalence in the general

population of women. As expected the prevalence of adenomyosis in our study was less than

that in most recent studies (Weiss et al. 2008; Yeniel et al. 2007; Parazzini et al. 2009). In

order to establish a true prevalence, a screening study involving a large number of unselected

women would be required, but even then self-selection of women would provide a source of

bias as those with past or present concerns about their gynaecological health would be more

likely to volunteer for screening. Our estimate is therefore likely to be slightly higher than the

genuine prevalence of the condition as we studied symptomatic women attending a

gynaecology clinic.

Our finding of good agreement between ultrasound and histology has significant implications

for clinical practice. Hysterectomy is rarely performed in modern practice for benign

indications and it is neither practical nor realistic to use histology as a gold standard, in future

clinical studies of adenomyosis. TVUS and MRI are the only non-invasive diagnostic

techniques that could be used to investigate the prevalence of adenomyosis in a population of

women attending outpatient clinics. MRI is much more expensive, time consuming and less

accessible than ultrasound. In addition, previous studies have not demonstrated that MRI

provides superior diagnostic accuracy compared to ultrasound (Dueholm 2006). Recent

studies have shown that ultrasound diagnosis of adenomyosis is sensitive and specific enough

to employ this technique for in-vivo studies of adenomyosis. The diagnostic criteria we used

have been assessed previously and they were found to be satisfactory (Kepkep et al. 2007). It

is therefore likely that in the future, TVUS will be the most common and often the only
 142

method, which will be used to make the diagnosis of adenomyosis and monitor its response to

conservative treatment.

We found a strong, statistically significant association between adenomyosis and

endometriosis. Early descriptions of adenomyosis also described an association with

endometriosis. Indeed, up until the 1920s they were considered to be part of the same entity.

One recent study (Kunz et al. 2005) suggested that adenomyosis and endometriosis may be

different expressions of the same pathological process; a process involving abnormal

peristaltic activity within the inner myometrium. A number of studies have confirmed an

association between adenomyosis and endometriosis (Bird et al. 1972; Emge 1972), but many

other have not (Vavilis et al. 1997; Kilkku et al. 1984; Vercellini et al. 1995; Weiss et al.

2009). A possible explanation for these differences is a tendency to look for the association

between adenomyosis and endometriosis at the time of hysterectomy. A number of women

affected by both adenomyosis and endometriosis may have been successfully treated for

endometriosis in the past. As a result they could be free of active endometriosis at the time of

hysterectomy, which would typically take place sometime later. Furthermore, three of the

studies that found no link between adenomyosis and endometriosis were retrospective. There

are many potential methodological problems with retrospective studies, which could lead to

under-reporting of endometriosis in study populations. These include the inability to detect

endometriosis on histological examination of the uterus alone and the risk of under-reporting

incidental endometriosis at the time of surgery and subsequent histological examination, even

if the ovaries and fallopian tubes had been removed at the operation. We tried to overcome

these difficulties by performing a prospective study and by taking into account both a past

history of endometriosis confirmed at diagnostic laparoscopy and objective evidence of

endometriosis on the TVUS.

 143

Increasing age up to the menopause has long been considered a risk factor for adenomyosis.

Molitor reported that adenomyosis becomes more common in later reproductive years with a

decline in the frequency of diagnosis after menopause, however age has not consistently been

shown to be associated with the disease and more recent studies evaluating age as an

independent variable have not found an association (Bergholt et al. 2001; Vercellini et al.

1995; Weiss et al. 2009). These studies, however, are limited in their ability to assess age as

an independent variable as they draw conclusions from women undergoing hysterectomy,

who tend to present a relatively narrow age range. The advantage of using a non-invasive

technique like ultrasound is that it can be applied to a much broader population of women.

For example, in Bergholt’s study just 31.2% of women were under the age of 45, whereas in

our study 65.6% of women were under the age of 45. Our findings are therefore more likely

to provide reliable information about the link between age and prevalence of adenomyosis.

Many pregnancy-related factors have been linked with adenomyosis including parity

(Parazzini et al. 1997; Kilkku et al. 1984; Vavilis et al. 1997; Vercellini et al. 1995),

caesarean delivery (Vavilis et al. 1997; Whitted et al. 2000) and terminations of pregnancy

(Curtis et al. 2002; Levgur et al. 2000; Vavilis et al. 1997). Again, other studies have not

replicated these findings (Bergholt et al. 2001; Parazzini et al. 1997; Panganamamula et al.

2004; Weiss et al. 2009). The univariate analysis in our study suggested that gravidity, parity,

uterine instrumentation during pregnancy, spontaneous miscarriage, vaginal delivery and

caesarean delivery were all associated with adenomyosis. After multivariate analysis only

gravidity was found to be significantly associated with adenomyosis, although parity was not

included in the multivariate analysis as analysis suggested it was co-linear with the other

variables. This finding suggests that it may be the presence of a pregnancy rather than the
 144

specifics of what occurs in any given pregnancy that may play a role in the development of

adenomyosis.

Body mass index, intrauterine contraceptive device use, use of the progesterone-only

contraceptive pill, vaginal or caesarean section delivery, miscarriage and either evacuation of

retained products of conception or surgical termination of pregnancy all appeared to be

significantly associated with adenomyosis prior to the multivariate analysis. It is likely that

age accounted for most of these trends and that once the data were adjusted for age the

significance was no longer seen. An important finding was the absence of a significant

difference in the prevalence of adenomyosis between pre- and post-menopausal women. The

prevalence of adenomyosis in postmenopausal women, however was less when compared to

the prevalence in the group of women aged 40-49, when the condition was found to be most

prevalent. This indicates that adenomyosis tends to regress after menopause, but still remains

detectable in a significant number of women. Others have described a similar relationship

(Kitawaki 2006). Kitawaki also noted that no studies have shown that use of the combined

oral contraceptive pill causes regression of adenomyosis. Our data adds further weight to this

and showed that there was no association between use of the combined oral contraceptive pill

and the presence of adenomyosis.

Adenomyosis is a common condition and yet little is known about its aetiology, natural

history and clinical significance. This was largely due to the inability to make a conclusive

and accurate diagnosis of adenomyosis in the past using non-invasive diagnostic methods.

Our study, and other recent studies, has shown that transvaginal ultrasound is an accurate

method to diagnose adenomyosis. Due to its wide availability and acceptability, ultrasound

could be used to facilitate large scale studies of adenomyosis in different populations of

women. Future research should try to elucidate its natural history and to examine possible
 145

associations between adenomyosis and different clinical symptoms including subfertility.

This information should help to develop more effective preventative and treatment strategies

in women affected by adenomyosis.

 146

[5] IS ADENOMYOSIS ASSOCIATED WITH MENORRHAGIA?

5.0 DISCUSSION

Our results did not show a significant association between the presence of adenomyosis and

menorrhagia on multivariable analysis. There was, however, a statistically significant positive

correlation between amount of menstrual loss and the number of ultrasound features of

adenomyosis seen. Other independent variables associated with menorrhagia included

fibroids, in particular sub-mucous fibroids, endometrial polyps, gravidity and BMI. Sub-

mucous fibroids were found to have the strongest association with menorrhagia.

The main strength of our study was that ultrasound rather than histology was used to

diagnosis uterine pathology which helped to reduce selection bias. This also allowed the

inclusion of a relatively large number of women with a greater range of symptoms and

varying degrees of severity of adenomyosis who are likely, therefore, to be more

representative of the population of women attending gynaecology clinics. In addition, this

was a prospective study with clearly defined inclusion criteria and a standardised approach to

the ultrasound examinations which were all performed by a single highly-trained operator

using advanced, modern ultrasound equipment.

While two different assessments of menstrual loss were used, both methods had limitations.

The subjective binary assessment of menorrhagia is simple to measure and easy to assess in a

large population, however studies have shown that self-assessed menorrhagia may not be an

accurate measure of excessive menstrual blood loss (Chimbira et al. 1980). Nevertheless, the

use of a subjective assessment of menorrhagia is consistent with recent national guidance

(NICE 2007) for clinicians. PBACs offer a relatively simple alternative to the more arduous

and time-consuming laboratory methods of objectively measuring menstrual blood loss. It

must be remembered however, that they are not as accurate and remain a subjective
 147

assessment of menstrual blood loss. Furthermore, there are limitations to the way they were

used in our study. It has been estimated that women may experience 20%-40% variation in

menstrual loss between different periods (Hallberg & Nilsson 1964). The fact that PBACs

were only used for one menstrual cycle in our study may therefore have limited the validity

of the PBAC assessment. Another limitation in our use of PBACs is the fact that women

filled in the charts based on usage with their own sanitary wear, which will be of variable

absorbency across the study population. Studies have generally addressed this issue by

providing standardised sanitary wear. We chose not do so for reasons of practicality and cost.

Despite these limitations the ease of use of PBACs enabled us to make a semi-quantitative

assessment of menstrual loss in a large population. Furthermore, by making a semi-

quantitative assessment of menstrual blood loss, their use allowed us to take account of the

severity of menorrhagia in the study population.There was a fair level of agreement between

both assessments of menorrhagia so we used both methods in data analysis.

There is a lack of consensus in the literature regarding the relationship between adenomyosis

and menorrhagia. This is not surprising bearing in mind that the majority of studies were

retrospective in nature and mainly included populations of women undergoing hysterectomy.

These studies used differing criteria for the diagnosis of adenomyosis and few of them

attempted to quantify severity of disease. In addition, none of the studies controlled for the

presence of concomitant pathology and their potential effect on the volume of menstrual loss.

Our study has clearly shown that severity of adenomyosis, as measured by the number of

different ultrasound features of adenomyosis seen, correlates with the amount of menstrual

loss. The severity of adenomyosis is difficult to express in quantitative terms as the lesions

are often poorly defined and they may be disseminated throughout different parts of the

myometrium. Levgur et al. used depth of foci of adenomyosis on histological examination to

 148

assess severity of disease in their retrospective study (Levgur et al. 2000). While this study

had relatively small numbers and used retrospective case note analysis to define menorrhagia,

it was still able to uncover an association between severity of adenomyosis and menorrhagia.

Not only did our study show a positive association between the number of ultrasound features

of adenomyosis seen and increasing PBAC-assessed menstrual loss, it also showed that some

of the individual ultrasound features of adenomyosis were independently associated with

increasing PBAC-assessed menstrual loss. Asymmetrical myometrial thickening and an

irregular EMJ were both independently associated with PBAC assessed menstrual loss. It is

not surprising that asymmetrical myometrial thickening has a greater effect on symptoms

than other ultrasound features, as it is by definition, only seen when the disease is severe

enough to have affected an entire wall of the uterus. This is in contrast to some of the other

features, such as linear striations, that may affect only a small volume of the uterus.

Irregularity of the EMJ may have a greater impact on menstrual loss than other features

because any pathological effect is occurring at the interface where blood loss occurs, in

contrast to myometrial cysts, for example, that by definition occur some distance from the

endometrium. Despite the fact that some features appear to have a greater impact on

menstrual loss than others, we think that it is better to express severity of adenomyosis semi-

quantitatively using the number of different morphological features alone in an individual

woman as an indirect measure of severity of disease. Any proposed system to formally grade

severity of adenomyosis based on the number or type of features seen needs to be

prospectively assessed before it can be brought into widespread clinical use.

The recognition that the severity of a condition is important when assessing clinical impact

has already been adopted in clinical practice when studying the effect of fibroids on

menstrual loss. It has been generally accepted that fibroids cause menorrhagia and that their
 149

size and location determines their clinical significance (NICE 2007). Our study has also

confirmed that the location of fibroids is a critical diagnostic feature as submucous fibroids

were found to have a much more severe effect on menstrual loss than fibroids in other

locations. It is also known however, that the majority of fibroids are asymptomatic (Divakar

2008). Furthermore, there is a surprisingly small amount of evidence showing an association

between fibroids and menorrhagia (Parker 2007). While early reports (Miller et al. 1953)

ascribed heavy periods to submucous fibroids only, others have found that neither the

number, volume or location of fibroids were related to menstrual characteristics (Marino et

al. 2004). This study was however, relatively small and did not have enough numbers to

assess the effect of submucous fibroids. A larger study in which the population had a higher

prevalence of fibroids found an association between both length of menses and “gushing

bleeding” and the size of fibroids but not the number or their location (Wegienka et al. 2003).

This study however, used telephone interviews as a means of assessing patient symptoms.

Our study had a large sample size, used high quality ultrasound to assess the exact location of

the fibroids and used both objective and semi-quantitative methods to measure menstrual

loss. It is likely therefore, that our findings that fibroids are associated with heavy periods and

that this association is stronger with sub-mucous fibroids, are accurate. A long-standing

difficulty in assessing the association of fibroids and adenomyosis on menorrhagia is the

potentially confounding effect that both conditions will have on each other, with both the

prevalence and the burden of both conditions being positively associated with advancing age

(Naftalin et al. 2012b; Mavrelos et al. 2010). By combining a thorough ultrasound assessment

looking for all uterine pathologies with multiple demographic and clinical factors before

using multivariate logistic regression analysis, our study was able to account for this

potentially confounding effect.

 150

Other variables found to be associated with menorrhagia in our study include endometrial

polyps, increasing gravidity and increasing BMI. While endometrial polyps are frequently

cited as a cause of menorrhagia (Munro et al. 2011), many studies have merely found a

higher than expected prevalence of endometrial polyps in symptomatic populations of women

and extrapolated causation. The only study that compared symptomatic women to

asymptomatic women found a significantly higher prevalence of polyps in women with

abnormal menstrual bleeding, however this study included women with prolonged periods

and polymenorrhoea as well as women with menorrhagia. Furthermore, the authors

concluded that small endometrial polyps are frequently asymptomatic (Clevenger-Hoeft et al.

1999). One study used PBACs to measure menstrual loss before and after performing

hysteroscopic polypectomy and found that post-procedure, there was a significant reduction

in menstrual loss (van Dongen et al. 2009). This suggests that endometrial polyps contribute

significantly to menstrual loss and is in keeping with our finding that they are significantly

associated with both PBAC-assessedand subjectively measured menorrhagia.

While there are little data assessing the effect of increasing BMI on menorrhagia, it is known

that the peripheral conversion of adrenal and gonadal androgens to oestrogens in peripheral

tissue is greater in obese women (Siiteri 1987). The increased endometrial proliferation

caused by this oestrogen excess is known to progress to endometrial hyperplasia and

endometrial cancer. Endometrial proliferation would also be expected to cause increased

menstrual flow. Our study cannot explain the association between increasing gravidity and

menorrhagia. It could be postulated that the increased blood loss may be due to the increase

in uterine size seen with increasing gravidity (Verguts et al. 2013), however Rees et al. found

no correlation between measured menstrual blood loss and endometrial volume-to-surface

ratio (Rees et al. 1984).

 151

[6] IS ADENOMYOSIS ASSOCIATED WITH DYSMENORRHOEA?

6.0 DISCUSSION

On multivariate analysis we found adenomyosis and endometriosis both had a significant

independent positive correlation with dysmenorrhoea. There was also a positive correlation

between the number of ultrasound features of adenomyosis seen and NRS score for painful

periods.

The main strengths of our study were that ultrasound was used to diagnosis adenomyosis,

minimising the heavy selection bias seen when histology is used to assess symptomatology in

the past. This also allowed for the inclusion of a relatively large number of women. All the

scans were performed using top of the range equipment by a single operator who was

prospectively looking for the presence of adenomyosis. A further strength of the study was

the use of a validated, reproducible pain scoring system to assess the extent of

dysmenorrhoea. A potential weakness is the lack of histological confirmation of diagnosis in

the majority of women studied however, in those women in the study population who

underwent hysterectomy there was good agreement between ultrasound and histological

diagnosis (Naftalin et al. 2012b).

Our finding that adenomyosis was significantly associated with pain symptoms, is neither

new, nor surprising. The first detailed description of adenomyosis by Cullen over a century

ago described one of the main symptoms as being a ‘great deal of pain’ (Cullen 1908). Many

subsequent authors have described a similar clinical picture involving painful periods in

women with adenomyosis (Emge 1962; Bird et al. 1972; Benson and Sneedon 1958).

However, several authors have found no association between the presence of adenomyosis

and pain-related symptoms (Bergholt et al. 2001; Parazzini et al. 1997; Weiss et al. 2009).
 152

An important contributor to these conflicting reports is the use of histology as the means to

diagnose adenomyosis. Pain symptoms have historically been a common indication for

hysterectomy. Thus the hysterectomized population are significantly more likely to have pain

symptoms than those women with an intact uterus. This makes it harder to show a significant

difference in symptomatology between two groups undergoing hysterectomy. A recent study

looking only at perimenopausal women undergoing hysterectomy (Weiss et al. 2009) having

failed to show a significant difference in symptomatology between women found on

histological examination to have adenomyosis and those who did not, concluded that

adenomyosis was not a “disease per se, but rather a normal variant”. This study, as with most

other studies on the subject, used retrospective case note analysis to assess pain symptoms

and assessed them as a binary outcome only. By using a quantitative assessment of pain our

study was able to take account of severity of dysmenorrhoea in the data analysis and hence

our findings are likely to be more accurate.

A further obstacle in assessing the association between adenomyosis and dysmenorrhoea is

dealing with the confounding effect of endometriosis. Both conditions are a recognised cause

of dysmenorrhoea and studies have revealed a strong association between the two conditions

(Naftalin et al. 2012b). By performing a thorough pelvic ultrasound assessment looking for

the presence of ultrasound features of adenomyosis as well as direct and indirect features of

endometriosis, prior to using multivariable logistic regression analysis, our study was able to

account for this potentially confounding effect.

 153

[7] EVALUATION OF A SECOND-STAGE TEST DESIGNED TO IMPROVE THE

ACCURACY OF TRANSVAGINAL ULTRASOUND IN THEDIAGNOSIS OF

ENDOMETRIAL CANCER

7.0 DISCUSSION

We have shown that a second-stage ultrasound test applied to women presenting with

postmenopausal bleeding and either an endometrial thickness >5mm or focal pathology, can

significantly improve the specificity of ultrasound diagnosis of endometrial cancer. There are

a number of advantages to being able to more accurately diagnose endometrial cancer in

women with postmenopausal bleeding. It allows more efficient triaging of women with

postmenopausal bleeding for invasive diagnostic testing. Without application of the second-

stage test, women identified with focal endometrial lesions would normally have to wait for a

hysteroscopy, which might take weeks, before reaching a tissue diagnosis. The identification

of women with focal endometrial lesions with appearances suggesting a higher risk of

endometrial cancer, facilitates same-day outpatient endometrial sampling, for example pipelle

biopsy, which has been shown to have a sensitivity of 99.6% in detecting endometrial cancer

(Dijkhuizen et al. 2000). This would speed up the diagnostic process and reduce the time

from presentation to treatment in those women, while reducing the total number of

hysteroscopies required. This would both reduce the overall risk associated with diagnostic

invasive procedures while also reducing service costs and waiting times.

The post-test odds of endometrial cancer in women with homogenous, avascular endometria

and an intact EMJ are low (0.025). While this study is insufficient to conclude that an

endometrial sample would not be required in this group, it could provide significant

reassurance to patients and reduce the time-pressure for endometrial sampling in women
 154

found to be low risk for endometrial cancer, despite having a thickened endometrium.

Endometrial sampling could even be avoided in this group if, due to other co-morbidities, the

invasive procedures required to get one represented too significant a risk.

Various endometrial ultrasound criteria have been assessed in attempts to improve the

specificity of ultrasound in diagnosing endometrial cancer with mixed results. A study by

Randelzhofer et al. agreed with our finding that endometrial morphology and regularity of the

EMJ were useful criteria to differentiate benign and malignant endometria (Randelzhofer et

al. 2002). In contrast, other investigators (Epstein et al. 2001; Epstein et al. 2002) concluded

in two separate studies that there was significant overlap of ultrasound morphology between

benign and malignant endometrial lesions. A later study by the same research group (Epstein

and Valentin 2006) did find that heterogeneous echogenicity and an irregular surface of

endometrial lesions on B-mode ultrasound scan were useful criteria for diagnosing

endometrial malignancy in women with thickened endometria, but only when there was fluid

within the uterine cavity. Differences in study design and population may explain why their

conclusions differ from ours. Our population (mean age 59 years v 65 years) was younger

and had a greater proportion of malignancies staged 2 or above (36.4% v 9.1%). Their study

population also included women taking HRT whereas both our study and the study by

Randelzhofer excluded patients taking HRT (Randelzhofer et al. 2002). Lastly, in their

analysis, heterogeneous endometrium was close to statistical significance (P=0.08) and as the

authors themselves concluded, it may just have been that their study was underpowered to

detect a true difference as being statistically significant.

There have been similarly contrasting results from studies assessing the use of Doppler as a

means to improve the accuracy of ultrasound in diagnosing endometrial cancer. Early studies

using colour Doppler reported promising results in the diagnosis of endometrial cancer.
 155

Bourne et al. used colour Doppler to measure and compare the pulsatility indices (PI) of the

uterine arteries in 34 women with postmenopausal bleeding. They found that the PI was

reduced in the 17 women subsequently found to have endometrial cancer and concluded that

Doppler had great potential in the assessment of uterine pathology (Bourne et al. 1991). It is

unclear however, from their methods, how women were selected for the study. Furthermore,

the prevalence of endometrial cancer in the study population (50%) was much greater than in

most other studies of women with postmenopausal bleeding. Subsequent studies of Doppler

flow both in the uterine and endometrial arteries (Sladkevicius et al. 1994; Sheth et al. 1995)

have been unable to show that Doppler is useful in discriminating between benign and

malignant endometrial lesions.

Timmerman et al. showed that using colour Doppler to assess the distribution of blood

vessels supplying focal endometrial lesions was helpful in the diagnosis of endometrial

polyps (Timmerman et al. 2003), but they excluded women with a suspicion or confirmed

endometrial cancer. Alcazar et al reported that in more than 80% of cases of endometrial

cancer, a particular power Doppler pattern of vasculature could be observed (Alcazar et al.

2003). Epstein et al. used an algorithm including power Doppler assessment of endometrial

vascularity to estimate the risk of endometrial malignancy. They concluded that their

algorithm was better than subjective assessment at differentiating benign and malignant

endometria in women with an endometrial thickness of 5-15mm, although this conclusion did

not reach statistical significance (Epstein et al. 2002). Subsequent work by the same research

group was unable to show that using colour Doppler to assess the number of blood vessels

within the endometrium aided the diagnosis of malignancy in women with postmenopausal

bleeding (Epstein and Valentin 2006). In contrast to our study however, they assessed the

number of vessels seen within the endometrium as a whole. It may be that in order for colour
 156

Doppler assessment of vascular morphology to be helpful, it needs to be limited to

assessment of the number of vessels crossing the EMJ at the site of any endometrial

irregularity or focal lesion.

Any test including assessment of endometrial morphology and/or Doppler examination will

be limited by the fact that these methods are subjective. Furthermore, there were a relatively

small number of women included our study. While, these results would need to be

reproduced in larger numbers and in a different population prior to being introduced in a

clinical setting, we have nevertheless shown that a large increase in the specificity of

ultrasound in diagnosing endometrial cancer is possible without a large drop in sensitivity.

 157

PART V Conclusions and further research

 158

This thesis has explored aspects of the ultrasound assessment of the EMJ. It has shown that a

classification of EMJ visualisation based on 3D ultrasound is reproducible and is therefore

feasible to use in clinical practice. Endometrial thickness and parity have been shown to be

significantly associated with quality of EMJ visualisation. These findings however, only

relate to normal uteri and further work is required to assess what pathological variables affect

EMJ visualisation.

In contrast to other work, it was found that there was a good level of intra- and interobserver

agreement in the ultrasound diagnosis of adenomyosis. This was the case when real-time

ultrasound assessments were compared with assessments made from stored uterine volumes,

as well as when both assessments were made from stored volumes. Arguments against the

use of TVUS for the pre-operative diagnosis of adenomyosis have centred on inadequate

levels of inter-observer agreement and the reported necessity of real-time images to make the

diagnosis. This finding strengthens the case for TVUS to be the pre-operative imaging

modality of choice. It is also helpful to the clinician and to the patient, allowing a second

opinion to be sought, in cases of uncertainty, without recourse to a second scan. Further work

should focus on assessing the inter-observer agreement of TVUS in the diagnosis of

adenomyosis when two real-time scans are performedon the same occasion.

TVUS was used to estimate the prevalence of adenomyosis in women attending a general

gynaecology clinic in a large prospective observational study. The prevalence was estimated

to be 20.9%. It was also found that age, gravidity and pelvic endometriosis were all

significantly associated with the presence of adenomyosis. Better estimates of the prevalence

of adenomyosis using TVUS could help improve understanding of the burden of the disease,

in particular with regard to menstrual symptoms, pain symptoms and subfertility.

Furthermore, more accurate estimates of the prevalence of adenomyosis will lead to more
 159

accurate assessment of the associations of the disease and will therefore aid studies in the

future looking to identify women at increased risk of developing the condition.

TVUS was also used to investigate whether adenomyosis was associated with menstrual

symptoms. Subjective assessments of dysmenorrhoea were used to look for an association

between adenomyosis and dysmenorrhoea. Multivariable analysis revealed that both

endometriosis and adenomyosis were independently associated with dysmenorrhoea. Both

subjective and semi-quantitative assessments of menorrhagia were used in the same

population to investigate if adenomyosis was associated with menorrhagia. Multivariable

analysis showed that while adenomyosis was not significantly associated with menorrhagia

when assessed as a binary outcome, when severity of disease was taken into account, there

was a significant association. In both studies, severity of disease was based on the number of

ultrasound features of adenomyosis seen. While this classification system has revealed

insights into the relationship between adenomyosis and menstrual symptoms, its validity

remains uncertain. A prospective evaluation is required.

Lastly, the role of a second-stage ultrasound test incorporating assessment of the EMJ in the

diagnosis of endometrial cancer in women presenting with post-menopausal bleeding was

investigated. Its use was found to significantly increase the specificity of ultrasound while

having a minimal effect on sensitivity. It must however be noted that the sample size was

small.

This thesis has, I hope, set a new benchmark for the ultrasound assessment of both the EMJ

and the uterine diseases adenomyosis and endometrial cancer. By incorporating 3D

ultrasound assessment with a more detailed analysis of the EMJ and the individual ultrasound

features of adenomyosis, this thesis has increased our understanding of the EMJ,

adenomyosis and paved the way for a new way of triaging women with post-menopausal
 160

bleeding. It should also facilitate future work on adenomyosis and the EMJ. Key areas for

further investigation are the prospective development of a clinically relevant ultrasound

grading system for the severity of adenomyosis, the role of adenomyosis in subfertility and a

comparison of the current model of assessing women with post-menopausal bleeding with the

the second-stage test used in this thesis to assess women with post-menopausal bleeding.

Application of the ultrasound assessment used in this thesis to subfertile women could help

determine how severity of adenomyosis affects fertility and also the impact of individual

ultrasound features of adenomyosis on fertility. Continuation of this work in subfertile

women who then conceive could provide novel information on the impact of adenomyosis on

obstetric outcomes.
 161

References

Ahmed AI, Mahmoud AEA, Fadiel AA, Frederick N. Comparison of 2-, 3D and Doppler
ultrasound with histological findings in adenomyosis. Fertil Steril 2007; 88: S82.

Alcázar JL, Castillo G, Mínguez JA, Galán MJ. Endometrial blood flow mapping using
transvaginal power Doppler sonography in women with postmenopausal bleeding and
thickened endometrium. Ultrasound Obstet Gynecol 2003 Jun;21(6):583-8.

Alcázar JL, Galván R, Albela S, Martinez S, Pahisa J, Jurado M, López-García G.

Assessing myometrial infiltration by endometrial cancer: uterine virtual navigation with
three-dimensional US. Radiology 2009 Mar;250(3):776-83.

Alcázar JL, Jurado, M. Three-dimensional ultrasound for assessing women with

gynecological cancer:A systematic review. Gynecologic Oncology 2011;120:340–346.

Anderson J. The myometrium. In: Anderson J (ed.). Gynecologic Imaging. Philadelphia:

Churchill Livingstone, 1999, 243.

Arnold LL, Meck JM & Simon JA. Adenomyosis: evidence for genetic cause. Am J Med
Genet 1995; 55:505-506.

Azziz R. Adenomyosis: Current Perspectives. Obstet Gynecol Clin North Am 1989 Mar;
16(1): 221-235.

Baba, K, Okai T. Basis and principles of three - dimensional ultrasound. Three - dimensional
ultrasound in Obstetrics and Gynaecology. K. Baba and D. Jurkovic. 1997, London,
Parthenon.

Barrier BF, Malinowski MJ, Dick EJ, Hubbard GB, Bates GW. Adenomyosis in the baboon
is associated with primary infertility. Fertil Steril 2004 Oct;82 Suppl 3:1091-4.
 162

Bazot M, Cortez A, Darai E, Rouger J, Chopier J, Antoine J, Uzan S. Ultrasonography

compared with magnetic resonance imaging for the diagnosis of adenomyosis: correlation
with histopathology. Hum Reprod 2001;16:2427–2433.

Bazot M, Malzy P, Cortez A, Roseau G, Amouyal P, Darai E. Accuracy of transvaginal

sonography and rectal endoscopic sonography in the diagnosis of deep infiltrating
endometriosis. Ultrasound Obstet Gynecol 2007;30:994–1001.

Bazot M, Darai E, Rouger J, Detchev R, Cortez A, Uzan S. Limitations Of Transvaginal

Sonography For The Diagnosis Of Adenomyosis With Histopathological Correlation.
Ultrasound Obstet Gynecol 2002;20(6): 605-611.

Benagiano G, Brosens I. History of adenomyosis. Best Practice & Research Clinical

Obstetrics and Gynaecology 2006; 20(4): 449-463.

Bennett CC, Richards DS. Patient acceptance of endovaginal ultrasound. Ultrasound Obstet
Gynecol 2000 Jan;15(1):52-5.

Benson RC & Sneeden VD. Adenomyosis: A Reappraisal Of Symptomatology. Am J Obstet

Gynecol 1958;76: 1044.

Bergeron C, Amant C, Ferenczy A. Physiology and physiopathology of adenomyosis. Best

Practice & Research Clinical Obstetrics and Gynaecology 2006;20(4):511-521.

Bergholt T, Eriksen L, Jacobsen M, Hertz JB. Prevalence and risk factors of adenomyosis at
hysterectomy. Hum Reprod 2001;16:2418–2421.

Bettocchi S, Ceci O, Vicino M, Marello F, Impedovo L, Selvaggi L. Diagnostic inadequacy

of dilatation and curettage. Fertil Steril 2001; 75(4): 803-5.

Bird CC, McElin TW & Manalo-Estrella P. The elusive adenomyosis of the uterus. Am J
Obstet Gynecol 1972; 112: 583-593.

Birnholz JC. Ultrasonic visualization of endometrial movements. Fertil Steril 1984; 41:157-
158.
 163

Bohlman ME, Ensor RE, Sanders RC. Sonographic Findings In Adenomyosis Of The Uterus.
Am J Roentgenol 1987; 148(4): 765-766.

Bourne TH, Campbell S, Steer CV, Royston P, Whitehead MI, Collins WP. Detection of
endometrial cancer by transvaginal ultrasonography with color flow imaging and blood flow
analysis: a preliminary report. Gynecol Oncol 1991 Mar;40(3):253-9.

Bray F, Dos Santos Silva I, Moller H, Weiderpass E. Endometrial cancer incidence trends in
Europe: underlying determinants and prospects for prevention. Cancer Epidemiol Biomarkers
Prev2005;14:1132-42.

Brosens JJ, Barker FG, de Souza NM. Myometrial zonal differentiation and uterine junctional
zone hyperplasia in the non-pregnant uterus. Hum Reprod Update 1998; 4: 496-502.

Brosens I, Derwig I, Brosens J, Fusi L, Benagiano G, Pijnenborg R. The enigmatic uterine

junctional zone: the missing link between reproductive disorders and major obstetrical
disorders? Hum Reprod 2010; 25: 569–574.

Chimbira TH, Anderson AB, Turnbull AC. Relation between measured menstrual blood loss
and patient's subjective assessment of loss, duration of bleeding, number of sanitary towels
used, uterine weight and endometrial surface area. BJOG 1980; 87(7): 603-9.

Clark TJ, Mann CH, Shah N, Khan KS, Song F, Gupta JK. Accuracy of outpatient
endometrial biopsy in the diagnosis of endometrial hyperplasia. Acta Obstet Gynecol Scand
2001; 80:784–93.

Clark TJ. Outpatient hysteroscopy and ultrasonography in the management of endometrial

disease. Curr Opin Obstet Gynecol 2004; 16: 305–11.

Clevenger-Hoeft M, Syrop CH, Stovall DW, Van Voorhis BJ. Sonohysterography in

premenopausal women with and without abnormal bleeding. Obstet Gynecol 1999
Oct;94(4):516-20.

Creasman W. Revised FIGO staging for carcinoma of the endometrium. Int J Gynaecol
Obstet 2009;105:109.
 164

Creasman WT, Odicino F, Maisonneuve P, Quinn MA, Beller U, Benedet JL, Heintz AP,
Ngan HY, Pecorelli S. Carcinoma of the corpus uteri. FIGO 26th Annual Report on the
Results of Treatment in Gynecological Cancer. Int J Gynaecol Obstet 2006 Nov;95 Suppl
1:S105-43.

Cullen TS. Adeno-myoma Uteri diffusum benignum. Johns Hopkins Hosp Rep 1896; 6: 133.

Cullen TS. Adenomyoma of the uterus. Philadelphia & London: W.B. Saunders; 1908.

Curtis KM, Hillis SD, Marchbanks PA, Peterson HB. Disruption of the endometrial-
myometrial border during pregnancy as a risk factor for adenomyosis. Am J Obstet Gynecol
2002;187:543–544.

Daels J. Uterine contractility patterns of the outer and inner zones of the myometrium. Obstet
Gynecol 1974; 44: 315-326.

Demas BE, Hricak H, Jaffe RB. Uterine MR imaging: Effects of hormonal stimulation.
Radiology 1986; 159: 123-126.

De Souza N, Cosgrove D (2003). Imaging techniques in Gynaecology. Gynaecology. Shaw

RW, Soutter WP, Stanton SL. London, Elsevier Science Ltd.

De Ziegler D, Bulletti C, Fanchin R, Epiney M, Brioschi PA. Contractility of the nonpregnant

uterus: the follicular phase. Ann NY Acad Sci 2001; 943: 172-84.

Dijkhuizen FP, Mol BW, Brölmann HA, Heintz AP. The accuracy of endometrial sampling
in the diagnosis of patients with endometrial carcinoma and hyperplasia: a meta analysis.
Cancer 2000; 89: 1765–72.

Dijkhuizen FP, Mol BW, Brolmann HA, Heintz AP. Costeffectiveness of the use of
transvaginal sonography in the evaluation of postmenopausal bleeding. Maturitas 2003;
45:275–82.

Divakar H. Asymptomatic uterine fibroids. Best Practice & Research Clinical Obstetrics and
Gynaecology 2008. ;22(4):643–654.
 165

Donald I, MacVicar D, Brown TG. Investigation of abdominal masses by pulsed ultrasound.

Lancet 1958;1(7032): 1188-95.

Dueholm M, Lundorf E, Hansen ES, Ledertoug S, Sørensen JS, Olesen F. Magnetic

Resonance Imaging And Transvaginal Ultrasonography For Diagnosis Of Adenomyosis.
Fertil Steril 2001; 76(3): 588-594.

Dueholm M, Lundorf E, Sorensen JS, Ledertoug S, Olesen F & Laursen H. Reproducibility

Of Evaluation Of The Uterus By Transvaginal Sonography, Hysterosonographic
Examination, Hysteroscopy And Magnetic Resonance Imaging. Hum Reprod 2002; 17(1):
195-200.
Dueholm M. Transvaginal ultrasound for diagnosis of adenomyosis: a review. Best Pract Res
Clin Obstet Gynaecol 2006;20:569–582.

Emge L. The elusive adenomyosis of the uterus. Am J Obstet Gynecol 1962; 83:1541–63

Ellis PS, Whitehead R. Mitosis counting--a need for reappraisal. Hum Pathol 1981
Jan;12(1):3-4.

Entmann SS. Uterine Leiomyoma And Adenomyosis. In Jones HW,Wentz AC, Burnett LS
(eds.). Novak’s Textbook Of Gynecology. Baltimore: William & Wilkins; 1988, p. 443.

Epstein E, Ramirez A, L. Skoog L, Valentin L. Transvaginal sonography, saline contrast

sonohysterography and hysteroscopy for the investigation of women with postmenopausal
bleeding and endometrium > 5 mm. Ultrasound Obstet Gynecol 2001;18:157-162.

Epstein E, Skoog L, Isberg PE, De Smet F, DeMoor B, Olofsson PA, Gudmundsson S,

Valentin L. An algorithm including results of gray-scale and power Doppler ultrasound
examination to predict endometrial malignancy in women with postmenopausal bleeding.
Ultrasound Obstet Gynecol 2002; 20: 370–376.

Epstein E, Valentin, L. Gray-scale ultrasound morphology in the presence or absence of

intrauterine fluid and vascularity as assessed by color Doppler for discrimination between
benign and malignant endometrium in women with postmenopausal bleeding. Ultrasound
Obstet Gynecol 2006; 28: 89–95.
 166

Exacoustos C, Brienza L, Di Giovanni A, Szabolcs B, Romanini ME, Zupi E, Arduini D.

Adenomyosis: three-dimensional sonographic findings of the junctional zone and correlation
with histology. Ultrasound Obstet Gynecol 2011; 37: 471–479.

Fedele L, Bianchi S, Dorta M, Zanotti F, Brioschi D & Carinelli S. Transvaginal

Ultrasonography In The Differential Diagnosis Of Adenomyoma Versus Leiomyoma. Am J
Obstet Gynecol 1992; 167(3): 603-606.

Ferenczy A. Pathophysiology of adenomyosis. Hum Reprod Update 1998; 4: 312-322.

Fleischer AC, Kalemeris G, Entman S. Sonographic depiction of the endometrium during

normal cycles. Ultrasound Med Biol 1986; 12: 271–277.

Frankl O. Adenomyosis uteri. Am J Obstet Gynecol 1925; 10: 680-684.

Fusi L, Cloke B, Brosens JJ. The uterine junctional zone. Best Prac Res Clin Obstet
Gynaecol 2006; 20: 479-491.

Gaetje R, Kotzian S, Hermann G et al. Invasiveness of endometriotic cells in vitro. Lancet

1995; 346:1463-1464.

Gompel C & Silverberg SG. The Corpus Uteri. Pathology In Gynecology And Obstetrics.
Philadelphia: J.B. Lippincott;1985, pp. 208-209.

Grimes, DA. Diagnostic dilation and curettage: a reappraisal. Am J Obstet Gynecol 1982;
142(1): 1-6.

Gruboeck K, Jurkovic D, Lawton F, Savvas M, Tailor A, Campbell S. The diagnostic value

of endometrial thickness and volume measurements by three-dimensional ultrasound in
patients with postmenopausal bleeding. Ultrasound Obstet Gynecol 1996 Oct;8(4):272-6.
Gupta JK, Chien PF, Voit D, Clark TJ, Khan KS. Ultrasonographic endometrial thickness for
diagnosing endometrial pathology in women with postmenopausal bleeding: a meta-analysis.
Acta Obstet Gynecol Scand 2002; 81: 799–816.
 167

Hallberg L, Nilsson L. Constancy of individual menstrual blood loss. Acta Obstet Gynecol
Scand 1964; 43: 352-9.

Harris WJ, Daniell JF, Baxter JW. Prior cesarean section. A risk factor for adenomyosis? J
Reprod Med 1985 Mar;30(3):173-5.

Hendrickson MR, Kempson RL. Nonneoplastic Conditions Of The Myometrium And Uterine
Serosa, Surgical Pathology Of The Uterine Corpus. Philadelphia: W.B.Saunders; 1990, pp.
452-453.

Higham J, O’Brien PMS, Shaw RW. Assessment of menstrual blood loss using a pictorial
chart. Br J Obstet Gynecol 1990;97:734-739.

Hoskins PR, Criton A (2003). Colour Flow Imaging. Diagnostic Ultrasound. P. R. Hoskins,
A. Thrush, K. Martin and T. A. Whittingham. London, Greenwich Medical Media.

Hricak H, Alpers C, Crooks LE, Sheldon PE. Magnetic resonance imaging of the female
pelvis: initial experience. Am J Roentgenol 1983 Dec;141(6):1119-28.

Ijland MM, Evers JL, Dunselman GA, van Katwijk C, Lo CR, Hoogland HJ. Endometrial
wavelike movements during the menstrual cycle. Fertil Steril 1996;65:746-749.

Imaoka I, Wada A, Matsuo M et al. MR imaging of disorders associated with female

infertility: use in diagnosis, treatment and management. Radiographics 2003; 23: 1401-1421.

Janus CL, Wiczyk HP, Laufer N. Magnetic resonance imaging of the menstrual cycle. Magn
Reson Imaging 1988; 6: 669–674.

Jensen MP, Karoly P, Braver S. The measurement of clinical pain intensity: a comparison of
six methods. Pain 1986 Oct;27(1):117-26.
Jurkovic D. Three-dimensional ultrasound in gynecology: a critical evaluation. Ultrasound
Obstet Gynecol 2002;19:109-117.
 168

Kepkep K, Tuncay YA, Goynumer G, Tutal E. Transvaginal sonography in the diagnosis of

adenomyosis: which findings are most accurate? Ultrasound Gynecol Obstet 2007;30:341–
345.

Kilkku P, Erkkola R, Gro¨nroos M. Non-specificity of symptoms related to adenomyosis. A

prospective comparative survey. Acta Obstet Gynecol Scand 1984;63:229–223.

Kitawaki J, Obayashi H, Ishihara H, Koshiba H, Kusuki I, Kado N, Tsukamoto K, Hasegawa

G, Nakamura N & Honjo H. Oestrogen receptor-alpha gene polymorphism is associated with
endometriosis, adenomyosis and leiomyomata. Hum Reprod 2001; 16: 51-55.

Kitawaki J. Adenomyosis: the pathology of an oestrogen-dependent disease. Best Pract Res

Clin Obstet Gynaecol 2006;20:493–502.

Koff AK. Development of the vagina in the human fetus. Contrib Embryol 1933; 24: 59-91.

Kremer C, Duffy S, Moroney M. Patient satisfaction with outpatient hysteroscopy versus day
case hysteroscopy: randomised controlled trial. BMJ 2000 Jan 29;320(7230):279-82.

Kurman RJ. Mesenchymal Tumors Of The Uterus. 4th edn. New York: Springer Verlag;
1994.

Kunz G, Beil D, Deininger H, Wildt L, Leyendecker G. The dynamics of rapid sperm

transport through the female genital tract. Evidence from vaginal sonography of uterine
peristalsis (VSUP) and hysterosalpingoscintigraphy (HSSG). Human Reproduction
1996;11:627-632.

Kunz G, Beil D, Huppert P, Leyendecker G. Structural abnormalities of the uterine wall in

women with endometriosis and infertility visualized by vaginal sonography and magnetic
resonance imaging. Human Reproduction 2000;15 1:76-82.

Kunz G, Beil D, Huppert P, Noe M, Kissler S, Leyendecker G. Adenomyosis in

endometriosis--prevalence and impact on fertility. Evidence from magnetic resonance
imaging. Hum Reprod 2005 Aug;20(8):2309-16.
 169

Lara-Muñoz C, De Leon SP, Feinstein AR, Puente A, Wells CK. Comparison of three rating
scales for measuring subjective phenomena in clinical research. I. Use of experimentally
controlled auditory stimuli. Arch Med Res 2004 Jan-Feb;35(1):43-8.

Lee NC, Dicker RC, Rubin GL & Ory HW. Confirmation Of The Preoperative Diagnoses For
Hysterectomy. Am J Obstet Gynecol 1984; 150(3): 283-287.

Lee JK, Gersell DM, Balfe DM, Worthington JL, Picus D, Gapp G. The uterus: in vitro MR-
anatomic correlation of normal and abnormal specimens. Radiology 1985; 157: 175-9.

Lee C, Jurkovic D. Clinical Presentation and Investigation: The ‘‘One-Stop’’ Clinic,

Endometrial Sampling, Hysteroscopy, Vaginal Ultrasound. In Uterine Cancer, Luesley DM,
Lawton FG, Berchuck AB (eds). Taylor and Francis: New York, 2005.

Lei ZM, Reshef E, Rao V. The expression of human chorionic gonadotrophin/luteinizing

hormone receptors in human endometrial and myometrial blood vessels 1992. J Clin
Endocrinol Metab;75:651-659.

Lei ZM, Rao CV, Lincoln SR, Ackermann DM. Increased expression of human chorionic
gonadotropin/human luteinizing hormone receptors in adenomyosis. J Clin Endocrinol Metab
1993 Mar;76(3):763-8.

Levgur M, Abadi M, Tucker A. Adenomyosis: Symptoms, Histology, and Pregnancy

Terminations. Obstet Gynecol 2000;95(5):688-691.

Lewinski H. Beitrag Zur Frage Der Adenomyosis. Zentralbl Gynakol 1931; 55: 2163.

Leyendecker G, Kunz G, Wildt L, Beil D, Deininger H. Uterine hyperperistalsis and

dysperistalsis as dysfunctions of the mechanism of rapid sperm transport in patients with
endometriosis and infertility. Hum Reprod 1996 Jul;11(7):1542-51

Leyendecker G, Kunz G, Noe M, Herbertz M, Mall G. Endometriosis: a dysfunction and

disease of the archimetra. Hum Reprod Update 1998; 4: 752-762.

Leyendecker G, Kunz G, Kissler S, Wildt L. Adenomyosis and reproduction. Best Prac Res
Clin Obstet Gynaecol 2006; 20: 523-546.
 170

Leyendecker G, Wildt L, Mall G. The pathophysiology of endometriosis and adenomyosis:

tissue injury and repair. Arch Gynecol Obstet 2009 Oct;280(4):529-38.

Lieng M, Qvigstad E, Dahl GF, Istre O. Flow differences between endometrial polyps and
cancer: a prospective study using intravenous contrast-enhanced transvaginal color flow
Doppler and three-dimensional power Doppler ultrasound. Ultrasound Obstet Gynecol 2008
Dec;32(7):935-40.

Lin J, Lei ZM, Lojun S, Rao CV, Satyaswaroop PG, Day TG. Increased expression of
luteinizing hormone/human chorionic gonadotropin receptor gene in human endometrial
carcinomas. J Clin Endocrinol Metab 1994 Nov;79(5):1483-91.

Lo KW, Yuen PM. The role of outpatient diagnostic hysteroscopy in identifying anatomic
pathology and histopathology in the endometrial cavity. J Am Assoc Gynecol Laparosc 2000
Aug;7(3):381-5.

Lyons EA, Taylor PJ, Zheng XH, Ballard G, Levi CS, Kredentser JV. Characterization of
subendometrial myometrial contractions throughout the menstrual cycle in normal fertile
women. Fertil Steril 1991; 55: 771-4.

Mannion AF, Balagué F, Pellisé F, Cedraschi C. Pain measurement in patients with low back
pain. Nat Clin Pract Rheumatol 2007 Nov;3(11):610-8.

Marcus C. Relationship of adenomyosis uteri to endometrial hyperplasia and endometrial

carcinoma. Am J Obstet Gynecol 1961; 82:408–16.
Martin, K. (2003). B- Mode Instrumentation. Diagnostic Ultrasound. P. R. Hoskins, A.
Thrush, K. Martin and T. A. Whittingham. London, Greenwich Medical Media Ltd.

Marino JL, Eskenazi B, Warner M, Samuels S, Vercellini P, Gavoni N, Olive D. Uterine

leiomyoma and menstrual cycle characteristics in a population-based cohort study. Hum
Reprod 2004 Oct;19(10):2350-5.

Mansour GM, El-Lamie IK, El-Kady MA, El-Mekkawi SF, Laban M, Abou-Gabal AI.
Endometrial volume as predictor of malignancy in women with postmenopausal bleeding. Int
J Gynaecol Obstet 2007 Dec;99(3):206-10.
 171

Masui T, Katayama M, Kobayashi S et al. Changes in myometrial and junctional zone

thickness and signal intensity: demonstration with kinematic T2-weighted MR imaging.
Radiology 2001; 221: 75-85.

McCarthy S, Tauber C, Gore J. Female pelvic anatomy: MR assessment of variations during

the menstrual cycle and with use of oral contraceptives. Radiology 1986; 160:119–123.

McNay M, Fleming J. Forty years of obstetric ultrasound 1957–1997: from A-scope to three
dimensions. Ultrasound Med Biol 1999;25:3–56.

Mehasseb M, Habiba M. Adenomyosis uteri: an update. The Obstetrician & Gynaecologist

2009;11:41-47.

Miller NF, Ludovici PP, Dontas E. The problem of the uterine fibroid. Am J Obstet Gynecol
1953 Oct;66(4):734-46.

Mills AM, Longacre TA. Semin Diagn Pathol 2010 Nov;27(4):199-214.

Molitor JJ. Adenomyosis: a clinical and pathological appraisal. Am J Obstet Gynecol

1971;110:275–284.

Munro MG, Critchley HO, Fraser IS; FIGO Menstrual Disorders Working Group. The FIGO
classification of causes of abnormal uterine bleeding in the reproductive years. Fertil Steril
2011 Jun;95(7):2204-8.
Munsick RA. Dickinson's sign: focal uterine softening in early pregnancy and its correlation
with the placental site. Am J Obstet Gynecol 1985 Aug 1;152(7 Pt 1):799-802.

Naftalin J, Jurkovic D. The endometrial-myometrial junction: a fresh look at a busy crossing.

Ultrasound Obstet Gynecol 2009;34(1):1-11.

Naftalin J, Hoo W, Nunes N, Mavrelos D, Nicks H, Jurkovic D. Inter- and intraobserver

variability in three-dimensional ultrasound assessment of the endometrial-myometrial
junction and factors affecting its visualization.Ultrasound Obstet Gynecol 2012
May;39(5):587-91.
 172

Naftalin J, Hoo W, Pateman K, Mavrelos D, Holland T, Jurkovic D. How common is

adenomyosis? A prospective study of prevalence using transvaginal ultrasound in a
gynaecology clinic. Hum Reprod 2012;27(12):3432-9.

Naftalin J, Hoo W, Pateman K, Mavrelos D, Foo X, Jurkovic D. Is adenomyosis associated

with menorrhagia? Hum Reprod 2014;29(3):473-479

Nasri MN, Shepherd JH, Setchell ME, Lowe DG, Chard T. The role of vaginal scan in
measurement of endometrial thickness in postmenopausal women. Br J Obstet Gynaecol
1991 May;98(5):470-5

National Institute for Health and Clinical Excellence. Referral guidelines for suspected
cancer. 2005. www.nice.org.uk/nicemedia/pdf/cg027niceguideline.pdf.

National Institute for Health and Care Excellence Heavy menstrual bleeding (CG44).
London: National Institute for Health and Care Excellence;2007.

Newren L. Abdominopelvic cavity. In Diagnostic Medical Sonography – A Guide to Clinical

Practice (2nd edn). Kawamura DM, ed. Lippincott Williams and Wilkins Publishing:
Philadelphia, 1997; 525–536.
Nisolle M, Donnez J. Peritoneal endometriosis, ovarian endometriosis, and adenomyotic
nodules of the recto-vaginal septum are three different entities. Fertil Steril 1997:68;585-596.

Noe M, Kunz G, Herbertz M, Mall G, Leyendecker G. The cyclic pattern of the

immunocytochemical expression of oestrogen and progesterone receptors in human
myometrial and endometrial layers: characterization of the endometrial-subendometrial unit.
Hum Reprod 1999;14 1:190-197.

Novellas S, Chassang M, Delotte J, Toullalan O, Chevallier A, Bouaziz J, Chevallier P. MRI

characteristics of the uterine junctional zone: from normal to the diagnosis of adenomyosis.
Am J Roentgenol 2011 May;196(5):1206-13.

Ofili-Yebovi D, Ben-Nagi J, Sawyer E, Yazbeck J, Lee C, Gonzalez J, Jurkovic D. Deficient

lower-segment Cesarean section scars: prevalence and risk factors. Ultrasound Obstet
Gynecol 2008; 31: 72-77.
 173

Office for National Statistics. Cancer registration statistics England 2007. 2009.
www.statistics.gov.uk/StatBase/Product.asp?vlnk=7720.

O’Rahilly R. Prenatal human development. In Wynn R (ed.). Biology of the Uterus. New
York: Plenum Press; 1977, 35-57.

Outwater EK, Siegelman ES & Van Deerlin V. Adenomyosis: current concepts and imaging
considerations. Am J Roentgenol 1998; 170: 437- 441.

Panganamamula UR, Harmanli OH, Isik-Akbay EF, Grotegut CA, Dandolu V, Gaughan JP.
Is prior uterine surgery a risk factor for adenomyosis? Obstet Gynecol 2004;104:1034–1038.

Pandis N, Karaiskos C, Bardi G, Sfikas K, Tserkezoglou A, Fotiou S, Heim S. Chromosome

analysis of uterine adenomyosis. Detection of the leiomyoma-associated del(7q) in three
cases. Cancer Genet Cytogenet 1995; 80: 118-120.

Parazzini F, Vercellini P, Panazza S, Chatenoud L, Oldani S, Crosignani PG. Risk factors for
adenomyosis. Hum Reprod 1997;12:1275–1279.

Parazzini F, Mais V, Cipriani S, Busacca M, Venturini P. GISE. Determinants of

adenomyosis in women who underwent hysterectomy for benign gynecological conditions:
results from a prospective multicentric study in Italy. Eur J Obstet Gynecol Reprod Biol
2009;143:103–106.
Parker WH. Etiology, symptomatology, and diagnosis of uterine myomas. Fertil Steril 2007
Apr;87(4):725-36.

Peric H, Fraser IS. The symptomatology of adenomyosis. Best Practice & Research Clinical
Obstetrics and Gynaecology 2006 ;20(4):547-555.

Raine-Fenning NJ, Campbell BK, Kendall NR, Clewes JS, Johnson IR. Quantifying the
changes in endometrial vascularity throughout the normal menstrual cycle with three-
dimensional power Doppler angiography. Hum Reprod 2004; 19: 330–338.
 174

Randelzhofer B, Prompeler HJ, Sauerbrei W, Madjar H, Emons G. Value of

sonomorphological criteria of the endometrium in women with postmenopausal bleeding: a
multivariate analysis. Ultrasound Obstet Gynecol 2002; 19: 62–68.

Rees MC, Dunnill MS, Anderson AB, Turnbull AC. Quantitative uterine histology during the
menstrual cycle in relation to measured menstrual blood loss. BJOG 1984; 91(7): 662-6.

Reid PC, Mukri F. Trends in number of hysterectomies performed in England for

menorrhagia: examination of health episode statistics 1989 to 2002–3. BMJ 2005;330:938–
939.

Reinhold C, Atri M, Mehio A, Zakarian R, Aldis AE, Bret PM. Diffuse Uterine
Adenomyosis:Morphologic Criteria And Diagnostic Accuracy Of Endovaginal Sonography.
Radiology 1995; 197(3):609-614.

Reinhold C, McCarthy S, Bret PM, Mehio A, Atri M, Zakarian R, Glaude Y, Liang L,

Seymour RJ. Diffuse adenomyosis: Comparison of endovaginal US and MR imaging with
histopathologic correlation. Radiology 1996; 199: 151-158.

Reinhold C, Tafazoli F, Wang L. Imaging Features Of Adenomyosis. Hum Reprod Update

1998; 4(4):337-349.
Reinhold C, Tafazoli F, Mehio A, Wang L, Atri M, Siegelman ES, Rohoman L. Uterine
adenomyosis: endovaginal US and MR imaging features with histopathologic correlation.
Radiographics 1999; 19:S147–S160.

Roguin A. Christian Johann Doppler: the man behind the effect. Br J Radiol 2002
Jul;75(895):615-9.

Rosai J. Female Reproductive System. Ackerman’s Surgical Pathology. St Louis: Mosby;

1989, pp. 1058-1060.

Rubin JM. Power doppler. Eur Radiol 1999;9 Suppl 3: S318-22.

Salim R, Riris S, Saab W, Abramov B, Khadum I, Serhal P. Adenomyosis reduces pregnancy

rates in infertile women undergoing IVF. Reprod Biomed Online 2012 Sep;25(3):273-7.
 175

Sampson JA. Peritoneal endometriosis due to the menstrual dissemination of endometrial

tissue into the peritoneal cavity. Am J Obstet Gynecol 1927; 14: 422-469.

Sankaranarayanan R, Ferlay J. Worldwide burden of gynaecological cancer: the size of the

problem. Best Pract Res Clin Obstet Gynaecol 2006;20:207-25.

Seidman JD, Kjerulff KH. Pathologic findings from the Maryland Women's Health Study:
practice patterns in the diagnosis of adenomyosis. Int J Gynecol Pathol 1996 Jul;15(3):217-
21.

Sheth S, Hamper UM, McCollum ME, Caskey CI, Rosenshein NB, Kurman RJ. Endometrial
blood flow analysis in postmenopausal women: can it help differentiate benign from
malignant causes of endometrial thickening? Radiology 1995 Jun;195(3):661-5.

Siedler D, Laing FC, Jeffrey RB, Wing VW. Uterine Adenomyosis. A Difficult Sonographic
Diagnosis. J Ultrasound Med 1987; 6(7): 345-349.

Siiteri P. Adipose tissue as a source of hormones. Am J Clin Nutr 1987;45(1):277-82.

Simpson WL, Beitia LG, Mester J. Hysterosalpingography: a re-emerging study.

Radiographics 2006 Mar-Apr;26(2):419-31.

Silverberg SG, Kurman RJ, Nogales F, Mutter GL, Kubik-Huch RA, Tavassoli FA. Tumors
of the uterine corpus. In: Tavassoli FA, Devilee P, editors. Pathology and genetics of tumours
of the breast and female genital organs. World Health Organization classification of tumours.
Lyon, France: IARC Press; 2003. p. 217-32.

Sladkevicius P, Valentin L, Marsal K. Endometrial thickness and Doppler velocimetry of the

uterine arteries as discriminators of endometrial status in women with postmenopausal
bleeding: a comparative study. Am J Obstet Gynecol 1994; 171: 722-8.

Smith-Bindman R, Kerlikowske K, Feldstein VA, et al. Endovaginal ultrasound to exclude

endometrial cancer and other endometrial abnormalities. JAMA 1998; 280: 1510–17.

Sorosky JI. Endometrial cancer. Obstet Gynecol 2008 Feb;111(2 Pt 1):436-47

 176

Tahir MM, Bigrigg MA, Browning JJ, Brookes ST, Smith PA. A randomised controlled trial
comparing transvaginal ultrasound, outpatient hysteroscopy and endometrial biopsy with
inpatient hysteroscopy and curettage. Br J Obstet Gynaecol 1999 Dec;106(12):1259-64.

Tailor A, Jurkovic D, Bourne TH, Collins WP, Campbell S. Sonographic prediction of

malignancy in adnexal masses using an artificial neural network. Br J Obstet Gynaecol
1999;106:21–30.

Takahashi K, Nagata H & Kitao M. Clinical usefulness of determination of estradiol level in

the menstrual blood for patients with endometriosis. Nippon Sanka Fujinka Gakkai Zasshi
1989; 41: 1849-1850.

Tamai K, Koyama T, Umeoka S, Saga T, Fujii S, Togashi K. Spectrum of MR features in

adenomyosis. Best Practice & Research Clinical Obstetrics and Gynaecology 2006; 23(4):
4583-602.

Tamaya T, Motoyama T, Ohono Y, Ide N, Tsurusaki T & Okada H. Steroid receptor levels
and histology of endometriosis and adenomyosis. Fertil Steril 1979; 31: 396-400.

Tetlow RL, Richmond I, Manton DJ, Greenman J, Turnbull LW, Killick SR. Histological
analysis of the uterine junctional zone as seen by transvaginal ultrasound. Ultrasound Obstet
Gynecol 1999;14:188-193.
Thalluri V, Tremellen KP. Ultrasound diagnosed adenomyosis has a negative impact on
successful implantation following GnRH antagonist IVF treatment. Hum Reprod 2012
Dec;27(12):3487-92.

Timmerman D, Verguts J, Konstantinovic ML, Moerman P, Van Schoubroeck D, Deprest J,

van Huffel S. The pedicle artery sign based on sonography with color Doppler imaging can
Gynecol 2003;22(2):166–171.

Turnbull LW, Manton DJ, Horsman A, Killick SR. Magnetic resonance imaging changes in
uterine zonal anatomy during a conception cycle. Br J Obstet Gynaecol 1995; 102: 330-331.

Urabe M, Yamamoto T, Kitawaki J, Honjo H & Okada H. Estrogen biosynthesis in human

uterine adenomyosis. Acta Endocrinol (Copenh) 1989; 121: 259-264.
 177

Van Dongen H, Janssen C, Smeets M, Emanuel M, Jansen F. The clinical relevance of

hysteroscopic polypectomy in premenopausal women with abnormal uterine bleeding. Br J
Obstet Gynecol 2009;116(10):1387-90.

Van Tubergen A, Debats I, Ryser L, Londoño J, Burgos-Vargas R, Cardiel MH, Landewé R,

Stucki G, Van Der Heijde D. Use of a numerical rating scale as an answer modality in
ankylosing spondylitis-specific questionnaires. Arthritis Rheum 2002 Jun 15;47(3):242-8.

Vavilis D, Agorastos T, Tzafetas J, Loufopoulos A, Vakiani M,Constantinidis T, Patsiaoura

KK, Bontis JJ. Adenomyosis at hysterectomy: prevalence and relationship to operative
findings and reproductive and menstrual factors. Clin Exp Obstet Gynecol 1997;24:36–38.

Vercellini P, Parazzini F, Oldani S, Panazza S, Bramante T, Crosignani PG. Adenomyosis at

hysterectomy: a study on frequency distribution and patient characteristics. Hum Reprod
1995;10:1160–1162.

Verguts J, Lieveke A, Bourne T, Timmerman D. Normative data for uterine size according to
age and gravidity and the possible role of the classical “golden ratio” for defining optimal
uterine proportions. Ultrasound Obstet Gynecol 2013:10. doi: 10.1002/uog.12538. [Epub
ahead of print]

Verma SK, Lev-Toaff AS, Baltarowich OH, Bergin D, Verma M, Mitchell DG.
Adenomyosis: Sonohysterography with MRI correlation. Am J Roent 2009;192:1112-1116.

Wang PH, Shyong WY, Lin CH, Chen YJ, Li YF, Chao HT & Yuan CC. Analysis of genetic
aberrations in uterine adenomyosis using comparative genomic hybridization. Anal Quant
Cytol Histol 2002; 24: 1-6.

Wegienka G, Baird DD, Hertz-Picciotto I, Harlow SD, Steege JF, Hill MC, Schectman JM,
Hartmann KE. Self-reported heavy bleeding associated with uterine leiomyomata. Obstet
Gynecol 2003 Mar;101(3):431-7.

Weiss G, Maseelall P, Schott LL, Brockwell SE, Schocken M, Johnston JM. Adenomyosis a
variant, not a disease? Evidence from hysterectomized menopausal women in the Study of
Women’s Health across the Nation (SWAN). Fertil Steril 2009;1:201–206.

Wetzstein R, Renn KH. Zur Anordnung der glatten Muskulatur im Corpus uteri des
Menschen. Verh Anat Ges 1970;64:461-468.
 178

Whitted R, Verma U, Voigl B, Mendez L. Does cesarean delivery increase the prevalence of
adenomyosis? A retrospective review. Obstet Gynecol 2000;95:S83.

Wiczyk HP, Janus CL, Richards CJ, Graf MJ, Gendal ES, Rabinowitz JG, Laufer N.
Comparison of magnetic resonance imaging and ultrasound in evaluating follicular and
endometrial development throughout the normal cycle. Fertil Steril 1988; 49:969–972.

Willms AB, Brown ED, Kettritz UI, Kuller JA, Semelka RC. Anatomic changes in the pelvis
after uncomplicated vaginal delivery: evaluation with serial MR imaging. Radiology 1995;
195:91–94

Wolf DM, Spataro RF. The current state of hysterosalpingography. Radiographics 1988
Nov;8(6):1041-58.

Woodfield CA, Siegelman ES, Coleman BG, Torigian DA. CT features of adenomyosis. Eur
J Radiol 2009 Dec;72(3):464-9.

Yazbek J, Ameye L, Testa AC, Valentin L, Timmerman D, Holland TK, Van Holsbeke C,
Jurkovic D. Confidence of expert ultrasound operators in making a diagnosis of adnexal
tumor: effect on diagnostic accuracy and interobserver agreement.Ultrasound Obstet Gynecol
2010; 35: 89–93.

Yeniel O, Cirpan T, Ulukus M, Ozbal A, Gundem G, Ozsener S, Zekioglu O, Yilmaz H.

Adenomyosis: prevalence, risk factors, symptoms and clinical findings. Clin Exp Obstet
Gynecol 2007;34:163–167.
 179

Appendix

This thesis has resulted in the following publications:

 Naftalin J, Jurkovic D. The endometrial-myometrial junction: a fresh look at a busy

crossing. Ultrasound Obstet Gynecol 2009;34(1):1-11

 Naftalin J, Hoo W, Nunes N, Mavrelos D, Nicks H, Jurkovic D. Inter- and

intraobserver variability in three-dimensional ultrasound assessment of the
endometrial-myometrial junction and factors affecting its visualization.Ultrasound
Obstet Gynecol 2012;39(5):587-91

 Naftalin J, Hoo W, Pateman K, Mavrelos D, Holland T, Jurkovic D. How common is

adenomyosis? A prospective study of prevalence using transvaginal ultrasound in a
gynaecology clinic. Hum Reprod 2012;27(12);3432-9

 Naftalin J, Hoo W, Pateman K, Mavrelos D, Foo X, Jurkovic D. Is adenomyosis a

cause of menorrhagia? Hum Reprod 2014;29(3) 473-479

Work stemming from, but not included in this thesis, resulted in the following publication:

 Naftalin J, Nunes N, Hoo W, Arora R, Jurkovic D. Endometrial cancer and

ultrasound: why measuring endometrial thickness is sometimes not enough.
Ultrasound Obstet Gynecol 2012 Jan;39(1):106-9