J. Anat. (1986), 145, pp.

87-96 87
With 6 figures
Printed in Great Britain

Schmorl's nodes (intravertebral herniations of intervertebral

disc tissue) in two historic British populations
G. SALUJA, K. FITZPATRICK, M. BRUCE
AND J. CROSS
Department of Anatomy, Marischal College, University of Aberdeen,
Aberdeen AB9 lAS, Scotland
(Accepted 20 June 1985)
INTRODUCTION
The human vertebral column may be affected by herniation of intervertebral disc
tissue into the vertebral bodies, forming ectopic deposits of nucleus pulposus
material. These are called Schmorl's nodes (Fig. 1) after G. Schmorl whose extensive
work included descriptions of the lesion (Schmorl & Junghanns, 1971). Schmorl's
nodes are most frequent in the lumbar and lower thoracic regions (Resnick &
Niwayama, 1978).
Collins (1949) indicated the sequence of events liable to follow the formation of
Schmorl's nodes. The displacement of disc substance may produce narrowing of
the space between vertebral bodies with consequent diminution of spinal move-
ments. Forward tilting of the vertebral bodies may occur, with the production of
abnormal stresses leading to arthropathies such as marginal osteophytosis, ankylosis
of adjacent vertebrae and osteoarthritis of apophyseal joints. According to Resnick &
Niwayama (1978), the occurrence of multiple Schmorl's nodes in early life may lead
to Scheuermann's disease (juvenile kyphosis).
The aetiology of Schmorl's nodes is largely unknown. Occasionally their develop-
ment is associated with specific diseases which produce weakening of the subchondral
bone and resultant disruption of the cartilaginous endplate, for example, hyper-
parathyroidism and metastatic deposits in the vertebral column (Schmorl &
Junghanns, 1971; Resnick & Niwayama, 1978). In some cases, their occurrence can
be linked with trauma such as vertebral fractures (Schmorl & Junghanns, 1971).
However, the majority of Schmorl's nodes are of unknown cause ('idiopathic'). It
is possible that comparative studies of populations of different ethnic and/or
temporal origins may help to clarify the cause of these lesions.
Whilst often undetectable in radiographs of the living (Schmorl & Junghanns,
1971), Schmorl's nodes are readily identified in dried bone because they produce
characteristic deformations on the superior and/or inferior surfaces of vertebral
bodies (Fig. 2). The present report describes the occurrence of Schmorl's nodes in
two adult historic British populations. The vertebral sequence Thoracic 8 - Sacral
1 (TV8-SVI) was selected for examination in this study for two reasons: (1) within
this region there are specific anatomical features enabling precise identification of
individual vertebrae, a cogent factor when confronted with cases where the remains
were incomplete or partially damaged; (2) Schmorl's nodes occur most frequently in
this zone.
88 G. SALUJA AND OTHERS
vb
cp

af
np
al PI
A

Fig. 1. Diagrammatic illustration showing production of Schmorl's nodes. af, annulus fibrosus;
al, anterior longitudinal ligament; cp, cartilaginous endplate; np, nucleus pulposus; pi, posterior
longitudinal ligament; vb, vertebral body. A, upward herniation of nucleus pulposus producing
a node on the inferior surface of the vertebral body above; B, downward nuclear herniation
producing a node on the superior surface of the vertebral body below.

Fig. 2. Photograph showing an inferior view of a lumbar vertebra. The arrow

indicates a Schmorl's node.
 Schmorl's nodes 89

(RP CP LP
mca RC-CC-LC
\ RA CALA

msa
Fig. 3. Diagram used for recording the position of Schmorl's nodes on the vertebral surface.
*, the lateral attachments of the pedicles to the edge of the vertebral body; nsa, midsagittal axis,
defined by a line running through the midpoint between * and *; mca, midcoronal axis, defined
by a line perpendicular to the midsagittal^axis, through the midpoint on the midsagittal axis.
The upper case letters indicate the nine possible locations for nodes: RA Right quadrant,
anterior to the midcoronal axis. RP Right quadrant, posterior to the midcoronal axis. LA Left
quadrant, anterior to the midcoronal axis. LP Left quadrant, posterior to the midcoronal axis.
RC On the midcoronal axis, to the right of the midsagittal axis. LC On the midcoronal axis, to
the left of the midsagittal axis. CA On the midsagittal axis, anterior to the midcoronal axis. CP
On the midsaggital axis, posterior to the midcoronal axis. CC On intersection of the midcoronal
and midsagittal axes.

MATERIALS AND METHODS

Population groups
The sample representing the chronologically earlier population comprised remains
excavated from a thirteenth to sixteenth century cemetery associated with a Car-
melite Friary in Aberdeen. For this sample, age at death and sex were estimated
using conventional osteological methods -(Krogman, 1973). These remains yielded
24 individuals in which the vertebral sequence TV8-SVI was complete (n = 7) or
nearly complete (n = 17). These individuals comprised 18 males and 6 females. The
age range was from about 19 years to over 45 years.
The chronologically later series represented a London population of the eighteenth
to nineteenth centuries. Their skeletons were housed in the crypt of St Bride's Church
and age at death and sex were documented for this group. These remains yielded
53 adult individuals in which the vertebral sequence TV8-SV1 was complete (n = 14)
or nearly complete (n = 39). These individuals comprised 28 males and 25 females.
The age range was from 21 to 77 years.
For convenience, the two populations are designated simply as 'Aberdeen' and
'London' throughout this report.
Assessment of Schmorl's nodes
For each skeleton, the presence or absence of Schmorl's nodes was noted for each
available superior and inferior vertebral body surface within the TV8-SV1 sequence.
Whenever a node was found, its position on the vertebral surface was recorded on
a diagram where the surface was divided into four quadrants by two lines passing
through the centre of the surface, one in the sagittal plane and the other in the coronal
plane. Using this diagram, a Schmorl's node was assigned to one of nine possible
positions (Fig. 3).
RESULTS
None of the individuals showed evidence of pathological conditions known to be
associated with Schmorl's nodes. Table 1 shows the number of individuals showing
4 A NA 145
90 G. SALUJA AND OTHERS

Table 1. Number of individuals affected with Schmorl's nodes

Aberdeen Aberdeen London London
males females males females
Number of individuals 18 6 28 25
Number of individuals with 13 4 21 5
Schmorl's nodes
Y. of individuals with Schmorl's 72 67 75 20
nodes
Xs = 8-48; P < 0-05.

Table 2. Number of vertebral surfaces affected with Schmorl's nodes

Aberdeen: number of London: number of
affected surfaces/total affected surfaces/total
number of surfaces number of surfaces
Males and females 141/399 88/866
Males 107/298 80/435
Females 34/101 8/431

osteological evidence of Schinorl's nodes. High incidence rates were found in

Aberdeen males, Aberdeen females and London males (72, 67 and 75 %, respectively).
A significantly lower frequency was seen in London females where only 20% of
individuals were affected.
The incidence of node-affected vertebral surfaces in the two groups is shown in
Table 2.
The incidence of node-affected surfaces was significantly greater in the Aberdeen
material than in the London material when the sexes were combined (2 = 99-23,
P < 0-001). Considering the sexes separately, Aberdeen males had significantly more
node-bearing surfaces than London males (2 = 20-96, P < 0 001) and, similarly,
Aberdeen females had significantly more affected surfaces than London females
(2 = 103 84, P < 0 001). In the London group, a significant sex difference was
found in the frequency of affected surfaces, with males showing a greater incidence
than females (2 = 64-1 1, P < 0001). In contrast, in the Aberdeen group, there
was no significant sex difference in the occurrence of node-bearing vertebral surfaces.
Comparison of Tables 1 and 2 demonstrates that although similarly high proportions
of node-bearing individuals occurred among Aberdeen males, Aberdeen females and
London males, the number of affected vertebral surfaces was distinctly lower in
London males, indicating that the extent of the lesion was less severe in individuals
of the latter group. London females showed not only low incidence rates but also
mild severity in terms of fewer affected surfaces per individual as shown by a mere
2 % of affected vertebral surfaces.
The question of severity of Schmorl's nodes in individual skeletons was investi-
gated further by considering separately those individuals with complete TV8-SV1
columns (Table 3).
This confirmed the greater severity of the lesion in the Aberdeen group where the
majority of the individuals (6/7) had over 25 % of their vertebral surfaces affected
with Schmorl's nodes. This contrasted sharply with the London group where over
half the individuals had nodes on less than 5 % of their surfaces.
 Schmorl's nodes 91
Table 3. Proportion of node-affected vertebral surfaces in skeletons
with complete TV8-SV1 columns
Aberdeen (n = 7) London (n = 14)
Node-affected A _____ _____
surfaces (%) Males (n = 5) Females (n =2) Males (n = 8) Females (n = 6)
5- - 4 6
6-25 1 2
26-50 2 1 1
51-75 1 1 1
> 75 1 -

, 100
0
, Thoracic 0.2 Lumbar
80 -~
0
E
c 60 - Males Females
3: P < 0 05
8 40- N.S.D. Males Females

E 20 .
* ~~~~N.S.D

Aberdeen London
Fig. 4. The proportion of node-bearing vertebral surfaces in the thoracic and lumbar regions
of Aberdeen and London males and females. P denotes significance of difference between groups
calculated by thex-square test; N.S.D. indicates no significant difference between groups accord-
ing to the X-square test.

Table 4 shows the incidence of Schmorl's nodes associated with specific vertebral
surfaces.
These data showed that the frequency of nodes declined markedly below the level
of the superior border of LV4, i.e. below the third lumbar disc space. However,
applying the X-square test, no statistically significant differences in incidence were
found among the surfaces. The superior and inferior surfaces of the vertebral bodies
did not differ significantly in frequency of Schmorl's nodes in either sex or either
locality or in all the data combined.
There was, however, a significant sex difference in the distribution of Schmorl's
nodes within the thoracic and lumbosacral portions of the column, irrespective of
geographical provenance (Fig. 4). In the males, the thoracic region contained a
significantly greater proportion of affected vertebrae than the lumbosacral region
(Aberdeen: X2 = 4.44, P < 005; London: %2 = 4-48, P < 005). In the females,
however, nodes were more evenly distributed along the column. Neither the
Aberdeen nor the London females showed a significant difference in node frequency
between the thoracic and lumbar zones.
The observed positions of Schmorl's nodes on the vertebrae are illustrated in
Figure 5. The vast majority of nodes occurred on, or posterior to, the midcoronal
4-2
92 G. SALUJA AND OTHERS

Table 4. Frequencies of Schmorl's nodes on individual vertebral surfaces

Number of surfaces affected/number of surfaces observed:
Aberdeen London
Vertebral ,- A_i- _ _ _ ,_
_ _

surface Males Females Males Females

TV8
S* 3/14 2/4 4/20 0/23
It 6/14 0/4 8/20 0/20
TV9
S 5/12 1/4 4/19 0/20
I 7/14 0/4 8/20 0/20
TV10
S 7/14 0/4 3/20 0/21
I 8/15 1/4 8/20 1/21
TV1l
S 5/13 1/3 3/21 1/19
I 7/13 3/4 8/23 1/20
TV12
S 7/13 3/4 2/20 0/19
I 6/15 2/4 4/19 0/19
LV1
S 5/13 1/5 2/17 0/23
I 5/14 2/6 3/18 1/23
LV2
S 8/14 3/6 5/24 1/22
I 6/16 3/6 3/22 0/23
LV3
S 9/17 3/6 4/20 0/20
I 7/17 2/6 5/20 0/21
LV4
S 4/15 3/6 4/22 2/20
I 1/14 1/6 2/22 0/19
LV5
S 1/16 1/6 0/23 1/20
I 0/15 1/6 0/23 0/20
SV1
S 0/10 1/3 0/22 0/18
* S, superior surface; t I, inferior surface.

axis in both groups (Aberdeen: 99 %; London: 97 %). Furthermore, in both groups,

over 60 % of the lesions were confined to the midsagittal axis on, and posterior to,
its intersection with the midcoronal axis.
In both groups, the nodes often occurred on two or more sequential vertebral
surfaces to form discrete series of remarkably similar shape and position (Fig. 6).
Out of 38 node series observed, 20 (i.e. 53 %) occurred in this peculiar replicated
form.
In the London group, there was no significant correlation between age and the
proportion of node-affected vertebral surfaces. In the Aberdeen group, correlation
analysis was precluded by lack of documentation of age. However, the material was
divided into two broad categories, namely, under 30 years and over 30 years. No
significant difference was found between these two groups.
 Schmorl's nodes 93

<4
9 41 21
6-21-0-7
6 11 8
+955_8
006 7 0 L1

Aberdeen London
Fig. 5. The location of Schmorl's nodes on thevertebral surface. The numerals denote the
percentage of nodes occupying each of the nine positions described in Figure 3.

Fig. 6. Photograph showing Schmorl's nodes on two sequential vertebral surfaces. Note the
virtually identical shape, size and position of the nodes.

DISCUSSION
This study reports the occurrence of Schmorl's nodes in the thoracolumbosacral
(TV8-SVl) vertebral columns of adult skeletal remains derived from two historic
British populations, one from thirteenth to sixteenth century Aberdeen and the other
from eighteenth to nineteenth century London.
High incidences are found in Aberdeen males, Aberdeen females and London
males (72, 67 and 75 %, respectively). A significantly lower proportion of London
females are affected (20%). Hilton, Ball & Benn (1976) reported a high incidence
(76 %) in a modern cadaveric population derived from Manchester. Batts (1939)
found a relatively low incidence (20 %) in a cadaveric group in Ann Arbor, Michigan.
However, these studies did not specify the incidence according to sex. An inter-
mediate rate of 38% was reported by Schmorl & Junghanns (1971) in a modem
cadaveric population from Germany, with similar incidences in males and females
(39-9 and 34-3 %, respectively).
94 G. SALUJA AND OTHERS
The lesions are more severe in Aberdeen than in London, in terms of the pro-
portion of affected vertebral surfaces in the skeletons. Aberdeen males andfemales
had 37 and 34 % affected surfaces, respectively whilst significantly fewer node-
bearing surfaces were found in London males (19 %) and London females (2 %). In
London, females had significantly fewer affected surfaces than males, whereas in
Aberdeen the sexes did not differ significantly in this respect. McWhirr, Viner &
Wells (1982) found that Schmorl's nodes affected 12 and 9 % of male and female
vertebral surfaces (TV4-LV5), respectively, in a Romano-British population ex-
cavated at Cirencester. Compared to the present study, these rates are rather low,
except for the London females in whom the lesion is particularly scarce.
Thus although methodological differences may preclude precise comparisons, it
appears that different populations exhibit considerable variation in the incidence
and severity of Schmorl's nodes.
No statistically significant differences in node frequency are associated with any
specific vertebral surface. McWhirr et al. (1982) recorded maximal incidence on the
inferior surface of TV 1, but gave no statistical confirmation. In the present work,
upper and lower vertebral surfaces exhibit similar incidences of Schmorl's nodes.
Reporting on a series of routine autopsies in San Diego, Resnick & Niwayama (1978)
claimed that the lesion favours the lower vertebral surface but presented no quanti-
tative data.
In the present material, the distribution of Schmorl's nodes within the column is
markedly different in the sexes, regardless of geographical origin. In males, the
lesions are significantly more frequent in the thoracic than in the lumbosacral
portion, whereas in females they are spread evenly along the thoracolumbosacral
column. No mention of this sex difference has been found in previous reports.
The Schmorl's nodes found in the present study are considered to be of the
idiopathic type in view of the absence of evidence of any condition known to induce
the lesions. However, the findings of this investigation may contain clues to the
aetiology of Schmorl's nodes. Firstly certain hypothetical causal factors are not
supported by this study. For example, Dent's (1955) implication of osteoporosis is
contradicted by the lack of relationship between age and incidence of Schmorl's
nodes. The same objection applies to the association with degenerative disc disease
by Hilton et al. (1976).
Owsley & Bradtmiller (1983) have suggested that stresses associated with pregnancy
and childbirth may induce Schmorl's nodes. If reproductive history were a major
causal factor, it is perplexing to find such marked differences between Aberdeen and
London females. A definitive test of this hypothesis would require a study based on
subjects with documented obstetrical history.
Abnormalities of the vertebral blood vessels have been invoked in the genesis of
Schmorl's nodes. In the fetus, the intervertebral discs are well supplied with blood
vessels. These subsequently degenerate so that the normal adult intervertebral disc
is virtually avascular. According to Schmorl & Junghanns (1971), anomalies in this
process could result in the persistence of vascular channels in the cartilaginous end-
plate. The presence of such channels would weaken the endplate and thereby
facilitate the herniation of the nucleus pulposus into the subchondral bone. In post
mortem adult lumbar columns, Nachemson, Lewin, Maroudas & Freeman (1970)
and Maroudas, Stockwell, Nachemson & Urban (1975) found that vascular channels
perforating the subchondral bone and penetrating the cartilaginous endplate
occurred mainly in the central region of the vertebral surface. This site corresponds
 Schmorl's nodes 95
closely with the localisation of Schmorl's nodes in the present study. The role of
residual vascular channels might be clarified by quantitative studies on their degree
of concurrence with Schmorl's nodes.
Some authors have suggested that Schmorl's nodes may form at foci of endplate
weakness arising from incomplete resorption of the notochord (Schmorl & Jung-
hanns, 1971; Resnick & Niwayama, 1978; McWhirr et al. 1982). The present study
supports the notochordal remnant hypothesis by virtue of the clustering of Schmorl's
nodes in the central/central-posterior part of the vertebral surface, since this locality
represents the site originally occupied by notochord (Williams & Warwick, 1980).
Furthermore, the tendency for node sequences to assume not only the same position,
but also the same shape, is suggestive of flawed notochordal regression. The fact
that ageing did not bring about an increased frequency of Schmorl's nodes may
reflect the inability of the nucleus pulposus to herniate through weak spots in the
endplate once it has lost its youthful elastic turgor (Resnick & Niwayama, 1978).
It has been suggested that trauma and strenuous activity, particularly during
adolescence, may contribute to the formation of Schmorl's nodes (Schmorl &
Junghanns, 1971; McWhirr et al. 1982). When considering skeletal remains, it is
difficult to assess such factors in the absence of information on the life style of the
populations concerned. Such information was not available for this study. It might
have helped to explain the marked differences between Aberdeen and London
females, particularly in view of Brugsch's (1957) claim that housework is the most
common cause of Schmorl's nodes in females.
The observed sex differences in the regional distribution of Schmorl's nodes is
puzzling and requires further investigation. It may reflect sex related differences in
prenatal axial skeleton development. Williams & Warwick (1980) recorded that
ossification of the centra commences in the lower thoracic region and extends
cranially and caudally along the column. However, in a study of unsexed fetuses
Bagnall, Harris & Jones (1977) reported that the centra begin to ossify in both the
lower thoracic and upper lumbar regions. Further work on fetal series of known sex
may disclose developmental differences that might be related to the different male
and female nodal distribution patterns. Alternatively, sex differences in node distri-
bution may be associated with different stress patterns arising from sex differences
in skeletal structure and/or physical activity.

SUMMARY
The herniation of the nucleus pulposus into the vertebral body produces ectopic
deposits of disc material which are known as Schmorl's nodes. This prolapsed disc
tissue leaves characteristic deformations on the surface of the vertebral body and
hence the incidence of this lesion can be studied in skeletal remains. This report
describes the occurrence of Schmorl's nodes in TV8-SV1 in two historic adult
British populations, one from Aberdeen and the other from London.
In the Aberdeen group, both males and females showed a high incidence rate and
severity of Schmorl's nodes. In the London group, the males had a similarly high
affliction whereas the females were nearly free of the condition. The lesion had no
significant predilection for any one particular vertebral surface. However, in males
in both localities, the frequency of Schmorl's nodes was significantly higher in the
thoracic region than in the lumbosacral region. In contrast, both groups of females
showed similar node frequency in these two zones. The majority of Schmorl's nodes
96 G. SALUJA AND OTHERS
were localised in the central and central-posterior regions of the vertebral surface.
When nodes occurred on successive vertebral surfaces, they often formed sequences
showing similar shape and position.
The aetiology of Schmorl's nodes is unclear. Various hypothetical causal factors
were appraised in relation to the findings of this study. It was suggested that
anomalies in vascular and/or notochordal regression may be related to the
development of the lesion.
We gratefully acknowledge the financial support of the Scottish Development
Department and the Scottish Home and Health Department. We are indebted to the
Aberdeen District archaeologists and to Dr Dewi Morgan for availability of the
skeletal remains examined in this work. Thanks are also due to Miss Carole Ross for
photographic assistance, Mrs Margaret Moir for secretarial help and Professor E. J.
Clegg for critical reading of the manuscript.

REFERENCES
BAGNALL, K. M., HARRIS, P. F. & JONES, P. R. M. (1977). A radiographic study of the human fetal spine.
2. The sequence of development of ossification centres in the vertebral column. Journal ofAnatomy 124,
791-802.
BATrS, M. (1939). Rupture of the nucleus pulposus: an anatomical study. Journal of Bone and Joint
Surgery 21, 121-126.
BRUGSCH, H. G. (1957). Rheumatic Diseases, Rheumatism and Arthritis. Philadelphia: Lippincott.
COLLINS, D. H. (1949). Pathology of Articular and Spinal Diseases. London: Edward Arnold & Co.
DENT, C. E. (1955). Idiopathic osteoporosis. Proceedings of the Royal Society of Medicine 48, 574-578.
HILTON, R. C., BALL, J. & BENN, R. T. (1976). Vertebral end-plate lesions (Schmorl's nodes) in the
dorsolumbar spine. Annals of the Rheumatic Diseases 35, 127-132.
KROGMAN, W. M. (1973). The Human Skeleton in Forensic Medicine. Springfield, Illinois: Charles C.
Thomas.
MAROUDAS, A., STOCKWELL, R. A., NACHEMSON, A. & URBAN, J. (1975). Factors involved in the nutrition
of the human lumbar intravertebral disc: cellularity and diffusion of glucose in vitro. Journal of
Anatomy 120, 113-130.
MCWHIRR, A., VINER, L. & WELLS, C. (1982). British Cemeteries at Cirencester. Cirencester: Excavation
Committee, Corinium Museum.
NACHEMSON, A., LEWIN, T., MAROUDAS, A. & FREEMAN, M. A. R. (1970). In vitro diffusion of dye
through the end-plates and the annulus fibrosus of human lumbar inter-vertebral discs. Acta ortho-
paedica scandinavica 41, 589-607.
OWLEY, D. W. & BRADTMILLER, B. (1983). Mortality of pregnant females in Arikara villages: osteological
evidence. American Journal ofPhysical Anthropology 61, 331-336.
RESNICK, D. & NIWAYAMA, G. (1978). Intravertebral disk herniations: cartilaginous (Schmorl's)
nodes. Radiology 126, 57-65.
SCHMORL, G. & JUNGHANNS, H. (1971). The Human Spine in Ilealth and Disease. 2nd American edition.
New York and London: Grune & Stratton.
WILLIAMS, P. L. & WARWICK, R. (ed.). (1980). Gray's Anatomy, 36th ed. Edinburgh, London, Melbourne,
New York: Churchill Livingstone.