Original article

Peguero–Lo Presti criteria for diagnosis of left ventricular

hypertrophy: a cardiac magnetic resonance validation study
Cláudio Guerreiroa,, Pedro Azevedob,, Ricardo Ladeiras-Lopesa,
Nuno Ferreiraa, Ana Raquel Barbosaa, Rita Fariaa, João Almeidaa,
João Primoa, Bruno Melicaa and Pedro Bragaa

Aims The diagnostic performance of the new Peguero–Lo outperformed both the Cornell (47 vs. 29%, P < 0.001) and
Presti ECG criteria for left ventricular hypertrophy (LVH) has Sokolow–Lyon voltage criteria (vs. 25%, P < 0.001). The
not been validated by cardiac magnetic resonance (CMR). specificities of all the criteria were above 94%, without
The aim of this study was to evaluate and compare the significant differences between them.
diagnostic performance of Peguero–Lo Presti, Cornell and
Conclusion In an all-comers European population with LVH
Sokolow–Lyon voltage criteria for LVH as defined by CMR in
Downloaded from http://journals.lww.com/jcardiovascularmedicine by BhDMf5ePHKbH4TTImqenVCbF2BRbWAbdN3TNAM3x1Sx1b5NgOMHx03/HCwUA60OcM2qsSa3ewZc= on 08/27/2020

defined by CMR, the criteria of Peguero–Lo Presti showed

an all-comers European population.
increased sensitivity for this diagnosis, when compared
Methods A total of 240 consecutive patients referred for with the Sokolow–Lyon and Cornell voltage criteria. As such,
CMR whohada concomitantelectrocardiogramfor reviewwere they could become the preferred ECG diagnostic tool when
evaluated. LVH group patients were defined according to the evaluating patients at risk for LVH.
reference values for sex and age of left ventricular mass index
J Cardiovasc Med 2020, 21:437–443
(LVMi). A control group, adjusted by sex, was randomly selected
from a population without LVH. We applied the ECG voltage Keywords: cardiac magnetic resonance, electrocardiography, left ventricular
criteria to both groups and evaluated their diagnostic accuracy. hypertrophy, left ventricular mass, Peguero–Lo Presti criteria
Diagnostic sensitivity and specificity were compared. a
Department of Cardiology, Centro Hospitalar de Vila Nova de Gaia and
b
Department of Cardiology, Centro Hospitalar Universitário do Algarve, Portugal
Results Two hundred and forty patients (mean age 63 years;
65% men) were divided into two groups (LVH n U 149; Correspondence to Cláudio Guerreiro, Rua Conceição Fernandes, 4434-502
Vila Nova de Gaia, Portugal
control n U 91). The main causes of LVH were hypertension Tel: +35 1924337705; fax: +35 1227830209;
(24.8%) and hypertrophic cardiomyopathy (21.5%). The e-mail: claudioeguerreiro@gmail.com
remaining patients of this group had a diagnosis of dilated
cardiomyopathy (14.8%), valvular heart disease (7.4%) and Received 11 September 2019 Revised 15 November 2019
Accepted 25 January 2020
infiltrative cardiomyopathy (2.0%). Overall, the sensitivity for
LVH diagnosis of the Peguero–Lo Presti criteria

Introduction the most common of which being the Cornell10 and the
Left ventricular hypertrophy (LVH) is defined by an Sokolow–Lyon voltage criteria.11–13 Several factors other
increase in left ventricular mass (LVM). The technique than left ventricular size or mass can influence QRS
of choice to make the definitive diagnosis of LVH is echo- voltages, such as age, sex, race, and body habitus.
cardiography because of its widespread use.1 However, this
imaging method has some limitations, such as the interob- Despite the high specificity (above 90%), one of the
server variability and the dependence on an appropriate pitfalls of all ECG criteria for LVH screening is their
acoustic window for accurate interpretation of the acquisi- low sensitivity.14 In a receiver-operating characteristic
tions.2 Cardiac magnetic resonance (CMR) has emerged as a (ROC) curve analysis, Cornell showed higher perfor-
highly precise modality for assessment of LVM because of mance compared with that of Sokolow–Lyon,15 with a
its higher accuracy and reproducibility,3,4 providing mea- diagnosis sensitivity that is still low, at only 20–40%
surements of LVM without the need for any geometric depending on the studies.15,16
assumptions, and the less interobserver and intra-observer Developing new criteria that could reduce the prevalence
variability compared with echocardiography (ECG).5–8 of false negative screenings was, therefore, needed.
Nevertheless, ECG still remains the most commonly Peguero et al.17 proposed a novel ECG voltage criterion,
used method for LVH screening owing to its low cost defined by the summation of the deepest S-wave in any
and wide availability.9 Several ECG criteria for LVH lead and the S-wave in lead V4 (SDþSV4) to diagnose
have been studied and applied in routine daily practice, LVH. In a Caucasian American population, screened with
transthoracic echocardiography and ECG, the designated

Cláudio Guerreiro and Pedro Azevedo equally contributed to this work. Peguero–Lo Presti voltage criteria showed the highest
1558-2027 ß 2020 Italian Federation of Cardiology - I.F.C. All rights reserved. DOI:10.2459/JCM.0000000000000964

© 2020 Italian Federation of Cardiology - I.F.C. All rights reserved.

438 Journal of Cardiovascular Medicine 2020, Vol 21 No 6

sensitivity when compared with Cornell and Sokolow– 2018, selecting 459 patients who were above 34 years old and
Lyon criteria.17 Another interesting study that was had a concomitant electrocardiogram for review. Patients
designed to further validate the Peguero–Lo Presti cri- younger than 35 years were excluded, as QRS voltage criteria
teria in an Asian population did not demonstrate the are not as well established for the 16-to-35-year age group.23
assumption that Peguero–Lo Presti criteria would pro- Patients were further selected according to the following
vide higher overall accuracy in LVH diagnosis in this exclusioncriteria:complete leftorright bundlebranch block,
patient population and, in fact Cornell voltage outper- ventricularpacedrhythm orQSwavesinthe precordial leads.
formed Peguero–Lo Presti as a screening test for LVH.18 One hundred and forty-nine patients had increased LVMi
(LVH group) and 310 had normal LVMi. From the latter
Several studies have shown an association between
population of 310consecutive patients referred for CMR and
Peguero–Lo Presti criteria and increased cardiovascular
without LVH, we randomly selected 91 patients (control
disease events and death,19–21 although there seem to be
group), after matching for sex with the LVH group. The
different performances according to the ethnicity of the
study flowchart can be seen in Fig. 1.
screened population.20 This illustrates the potential util-
ity of the newly proposed Peguero–Lo Presti voltage LVH was defined by CMR according to the LVM/body
criteria, not only for LVH screening but also as a predictor surface area, using reference values adjusted for age and
marker of worse outcome. Our aims were, therefore, to sex as presented by Kawel-Boehm et al.22 (LVM/BSA
further validate the higher sensitivity of Peguero–Lo >91 and >90 g/m2 for men under and above 60 years,
Presti voltage criteria for LVH screening using CMR, respectively; and LVM/BSA >77 and >78 g/m2 for
the gold-standard method for LV mass evaluation, and to women under and above 60 years, respectively).
compare its overall predictive performance with the most
Written informed consent was obtained from all patients
widely used Cornell and Sokolow–Lyon voltage criteria,
in accordance with requirements of the local institutional
in a Caucasian European population above 34 years old.
ethics committee.
Methods
Patients ECG analysis
We conducted a retrospective study of 1155 patients con- A 12-lead standard ECG (10 mm ¼ 1 mV, 25 mm/s) was
secutively referred for CMR fromJanuary 2016 to December acquired in supine position. The maximum time interval
Fig. 1

1155 consecutive patients referred to CMR center from 01/2016 to 12/2018

Excluded: < 35 years, absence of ECG for review,

complete left or right ventricular branch block, paced
rhythm, QS waves in precordial leads

459 consecutive patients referred to CMR center with > 34 years and an
electrocardiogram for review

149 patients with increased LVMi 310 patients with normal LVMi

LVH Group

91 randomly selected patients

matched for gender with LVH
group

Control Group

Flowchart of the study.

© 2020 Italian Federation of Cardiology - I.F.C. All rights reserved.

 Peguero–Lo Presti criteria for diagnosis of LVH Guerreiro et al. 439

between ECG and CMR was 6 months (mean 77 days). found in lead V4, the S-wave amplitude would be dou-
All 12-lead ECG were independently reviewed by two bled to obtain the value SD þ SV4.17
cardiologists and the diagnosis of LVH on ECG under-
Cornell and Sokolow–Lyon voltage criteria were used for
taken was blinded to CMR analysis.
comparison as they are the most commonly used electro-
The criteria recently described by Peguero–Lo Presti cardiographic criteria for LVH. Sex-specific Cornell voltage
were used.17 Individual leads of ECG were analyzed by criteria were determined by the amplitude of R in aVL plus
measuring the tallest R or R0 and the deepest S or QS (SD) the amplitude of S or QS complex in V3 (RaVL þ SV3),
complex in all the precordial and limb leads using the PR with a cutoff for LVH of greater than 2.8 mV in men and
segment as baseline. Next, we added SD to the S ampli- greater than 2.0 mV in women.10 Sokolow–Lyon voltage
tude in V4 (SD þ SV4). LVH was defined when SD þ SV4 criteria were obtained by adding the amplitude of S in V1
at least 2.3 mV (23 mm) for female subjects and at least and the amplitude of R in V5 or V6, with a cutoff for LVH of
2.8 mV (28 mm) for male subjects (Fig. 2). If the SD was at least 3.5 mV (SV1 þ RV5 or RV6).23

Fig. 2

Example of electrocardiogram and cardiac magnetic resonance analysis. (a) Sixty-seven-year-old male patient who met Peguero–Lo Presti
electrocardiographic criteria for left ventricular hypertrophy (deepest S-wave in any lead, which corresponds to SV3 þ S-wave in V4 2.8 mV in male
subjects): 1.9 þ 1.7 mV ¼ 3.6 mV. The criteria of Cornell voltage [aVL þ SV3 ¼ 2.5 mV (0.6 þ 1.9 mV); male subjects >2.8 mV] and Sokolow–Lyon
voltage [SV1 þ (RV5 or RV6) ¼ 1.4 mV (0.8 þ 0.6 mV); male subjects  3.5 mV] were not met. (b) Cardiac magnetic resonance showing increased
left ventricular mass index (98 g/m2).

© 2020 Italian Federation of Cardiology - I.F.C. All rights reserved.

440 Journal of Cardiovascular Medicine 2020, Vol 21 No 6

Cardiac magnetic resonance analysis Table 1 Characteristics of the study population

CMR studies were performed in the supine position on a Variable LVH (n ¼ 149) Non-LVH (n ¼ 91) P value
1.5 Tesla Siemens MRI Symphony TIM System (Siemens
Age (years) 64.1  12.9 61.1  13.2 0.09
Medical Solutions, Erlangen, Germany), with a phased- Male gender) 101 (67.8) 55 (60.4) 0.27
array receiver coil and breath-hold acquisitions with ECG Height (cm) 166.2  8.2 165.5  9.8 0.55
Weight (kg) 77.4  13.5 75.0  13.0 0.19
gating. Cine images were acquired in multiple short-axis BMI (kg/cm2) 27.9  4.3 27.4  4.1 0.30
and long-axis views with fast imaging with steady-state- BSA (m2) 1.85  0.18 1.82  0.19 0.28
free-precession (SSFP). Slice thickness was 5 mm with no Hypertension 113 (75.8) 60 (65.9) 0.10
Diabetes 44 (29.5) 26 (28.6) 1.00
gap between slices. The temporal resolution was 21  1 ms. Smoker 51 (34.2) 28 (30.8) 0.67
Sequence parameters included repetition time/echo time COPD 9 (6.0) 4 (4.4) 0.77
Chronic kidney disease 12 (8.1) 4 (4.4) 0.42
of 3.2/1.6 ms, in-plane pixel size of 2.1  1.3 mm, flip angle Atrial fibrillation 19 (12.8) 4 (4.4) 0.04
608, and acquisition time of 18 heartbeats. Image analysis Previous MI 20 (13.4) 6 (6.6) 0.13
was performed off-line using the semiautomatic CMR42 Heart disease cause
Hypertrophic CMP 37 (24.8) 0
program. Semiautomated tracking of the endocardial and Ischemic heart disease 32 (21.5) 2 (2.2) <0.001
epicardial borders of short-axis slices was performed. All Dilated cardiomyopathy 22 (14.8) 1 (1.1)
measurements were performed by a single observer. LVM Valvular heart diseasea 11 (7.4) 1 (1.1)
Infiltrative CMPb 3 (2.0) 0
was measured at end-diastole, which was defined as the Antihypertensive drugs 142 (96.6) 76 (83.5) 0.001
phase with the largest intraventricular volume. LVM was Number of antihypertensive drugs 2.5  1.1 1.8  1.2 <0.001
SBP (mmHg) 148  21 146  26 0.62
calculated after additional detection of epicardial borders of DBP (mmHg) 81  13 81  11 0.59
the LV, including the papillary muscles, by the subtraction Heart rate (bpm) 69  12 67  12 0.28
of endocardial volume from epicardial volume multiplied
Data are presented as mean  standard deviation or n (%). BSA, body surface
by the specific gravity of the myocardium (1.05 g/cm3). LV area; CABG, coronary artery bypass graft; CMP, cardiomyopathy; COPD, chronic
ejection fraction was calculated as (LVEDV  LVESV)/ obstructive lung disease; LVH, left ventricular hypertrophy; LVM, left ventricular
LVEDV. Both LVM and LVEDV were indexed to body mass; LVMI, left ventricular mass index; MI, myocardial infarction; PCI, percutane-
ous coronary intervention. a Aortic stenosis. b Amyloidosis.
surface area. Left atrium maximum volume, end-systolic
diameters and areas were measured after delineation of n ¼ 149; control n ¼ 91). As shown in Table 1, both groups
atrial endocardial borders, including atrial appendage, in all were well balanced, without significant differences
planes in all cardiac phases. The systolic descent and twist between baseline variables. The main causes of LVH
of the mitral valve was calculated from tracking of the valve were hypertension (24.8 vs. 0% control) and hypertrophic
motion on the long-axis cines, and used to correct for an cardiomyopathy (21.5 vs. 0%). The remaining patients of
increase in atrial volume because of AV ring descent. the LVH group had a diagnosis of dilated cardiomyopathy
(14.8 vs. 1.1%), valvular heart disease (7.4 vs. 1.1%) and
Statistical analysis infiltrative cardiomyopathy (2.0 vs. 0%) (P < 0.001 for all
Normal distribution of continuous variables was deter- the comparisons). This diseased group had also higher
mined using the Shapiro–Wilk test and they are expressed prevalence of atrial fibrillation (12.8 vs. 4.4%, P ¼ 0.04).
as mean  SD or as median  inter quartile range (IQR),
There was no significant difference in SBP and DBP
wherever appropriate. Categorical variables were reported
values when the CMR was performed, although LVH
as frequencies and percentages. Continuous variables were
patients were being treated with more antihypertensive
compared using the unpaired Student t test or the Wil-
drugs (2.5  1.1 vs. 1.8  1.2, P ¼ 0.007).
coxon rank sum test, as appropriate. Chi-square test or
Fisher exact test was used to compare categorical variables. Patients in the LVH group had significantly higher volt-
age values for Peguero–Lo Presti, Cornell and Sokolow–
Diagnostic sensitivity, specificity, positive-predictive value
Lyon criteria (Table 2), reflecting higher LVMi (80  23
(PPV), negative-predictive value (NPV) and accuracy
vs. 54  11 g/m2, P < 0.001). There were higher left atrial
(ACC) for the Peguero–Lo Presti, Cornell and Sokolow–
volumes in the LVH group (98  36 vs. 73  23 ml,
Lyon voltage criteria were calculated. McNemar x2-test
P ¼ 0.006). It is also noteworthy that LVH patients had
was used to compare the performance of the three ECG
higher left ventricular volumes (LVEDVI 92  37 vs.
criteria, regarding their sensitivities and specificities.
76  13, P < 0.001; LVESVI 42  35 vs. 27  7,
A P value less than 0.05 was considered statistically P < 0.001) and lower ejection fraction (58  14 vs.
significant. All analyses were performed using StataCorp. 65  5, P < 0.001). The morphological characteristics of
2017 Stata Statistical Software: Release 15 (College Station, both groups on CMR are summarized in Table 3.
Texas, USA: StataCorp LLC).
Sensitivity and specificity analysis of the different
Results electrocardiographic criteria for left ventricular
Characteristics of the study population hypertrophy
A total of 240 patients were enrolled (65% men), with a As shown in Table 4a, the newly proposed Peguero–Lo
mean age of 63 years and divided into two groups (LVH Presti criteria for LVH had the highest sensitivity,

© 2020 Italian Federation of Cardiology - I.F.C. All rights reserved.

 Peguero–Lo Presti criteria for diagnosis of LVH Guerreiro et al. 441

Table 2 Electrocardiographic characteristics of the study 19.8–37.2) vs. Sokolow–Lyon 18.0% (95% CI 11.3–26.4)]
population
(data not shown).
Variable LVH (n¼88) Non-LVH (n ¼ 87) P value

SD (mm) 17.7  8.2 11.1  3.2 <0.001

SV4 (mm) 9.3  6.8 4.7  3.3 <0.001
Discussion
PLP (mm) 27.1  13.3 15.9  5.5 <0.001 In this study, we describe for the first time the improved
SV3 (mm) 14.7  8.4 7.9  4.0 <0.001 sensitivity and similar specificity of Peguero–Lo Presti
RaVL (mm) 7.7  5.0 5.1  3.3 <0.001
CL (mm) 22.4  9.7 13.0  5.7 <0.001 criteria to diagnose LVH using CMR, the gold-standard
SV1 (mm) 10.9  5.2 8.0  3.1 <0.001 imaging method to evaluate LV morphology. A particu-
RV5 or V6 (mm) 16.7  7.7 14.9  5.9 0.047 larly important aspect of our study concerns also the
SL (mm) 27.6  9.9 23.0  7.4 <0.001
demonstration of the superiority of these criteria in an
Data are presented as mean  standard deviation or n (%). CL, Cornell voltage all-comers European population with high prevalence of
criteria ¼ SV3 þ RaVL; LVH, left ventricular hypertrophy; PLP, Peguero–Lo Presti
criteria ¼ SD þ SV4; SD, deepest S deflection; SL, Sokolow–Lyon voltage criteria
heart disease.
¼ SV1 þ RV5 or V6.
ECG remains an important screening exam for patients at
risk for LVH because of its widespread availability, ease
outperforming the Cornell [47.0%; 95% confidence inter- of use and its demonstrated independent clinical prog-
val (CI 38.8–55.3) vs. 28.8% (21.7–36.8), P < 0.001] and nostic impact.12,16–19
the Sokolow–Lyon voltage criteria [vs. 24.8% (95% CI:
The most commonly used diagnostic criteria for LVH are
18.1–32.5%) P < 0.001], without losing specificity. In
based on measurements of QRS voltages, particularly
fact, the specificities of all criteria were above 94%. In
Sokolow–Lyon and Cornell voltage criteria. Many more
the subgroup analysis according to sex, Peguero–Lo
have been developed, incorporating abnormalities in P
Presti criteria remained the one with nominally the best
wave, QRS axis and duration and ST-T segment mor-
sensitivity for both men [Peguero–Lo Presti 42.6% (95%
phology, including computerized recording methods, for
CI 32.8–52.8) vs. Cornell 20.8% (95% CI 13.4–30.9); vs.
example, for performing a composite analysis of several
Sokolow–Lyon 25.7% (95% CI 17.6–35.4)] and women
criteria.15,24 Nevertheless, no single ECG criterion is
[Peguero–Lo Presti 56.3% (95% CI 41.2–70.5) vs. Cor-
recommended over the other,23 as their sensitivity
nell 45.8% (95% CI 31.4–60.8); vs. Sokolow–Lyon 22.9%
remains low (less than 50%), albeit with high specificity
(95% CI 12.0–37.3)], while maintaining high specificity
(above 85–90%).15
when compared with the Cornell and Sokolow–Lyon
voltage criteria. Sensitivity and diagnostic accuracy of It remains, therefore, important to identify new criteria
Peguero–Lo Presti criteria were higher for women than that could increase the diagnostic sensitivity, while main-
for men. Those results are presented in Table 4b and c, taining high specificity for LVH. Recently proposed
respectively, for male and female sex. Also, in our study, Peguero–Lo Presti voltage criteria have been shown to
in obese patients (BMI above 25 kg/m2) the Peguero–Lo increase sensitivity (57%), while maintaining high speci-
Presti criteria remained the one with the highest sensi- ficity (90%), for the diagnosis of LVH when compared to
tivity [47.8 (95% CI 38.8–57.4) vs. Cornell 27.9% (95% CI the widely used Cornell (sensitivity 31%, specificity 93%)
and Sokolow–Lyon voltage criteria (sensitivity 14%,
specificity 99%), using echocardiography as the diagnosis
Table 3 Morphological characteristics on cardiac magnetic method for LVH.17 This was demonstrated in a Cauca-
resonance sian American population, but Sun et al.18 showed that its
Variable LVH (n ¼ 88) Non-LVH (n ¼ 87) P value applicability has a heterogeneous performance as com-
pared with Cornell voltage; it had significantly lower
LVM (g) 148  44 99  24 <0.001
LVMI (g/m2) 80  23 54  11 <0.001 specificity under recommended cutoffs in Asian popula-
LA volume (ml) 98  36 73  23 <0.001 tions and lower AUC for diagnosing LVH. One reason
LA area (cm2) 28  6 24  5 <0.001
RA area (cm2) 21  5 20  4 0.05
postulated by the authors could be the different ECG
LVEDV (ml) 170  72 138  30 <0.001 characteristics among races. In our study, with a popula-
LVEDVI (ml/m2) 92  37 76  13 <0.001 tion more similar to the Caucasian American, we con-
LVESV (ml) 79  67 49  16 <0.001
LVESVI (ml/m2) 42  35 27  7 <0.001 firmed the superiority of the Peguero–Lo Presti criteria
LVEF (%) 58  14 65  5 <0.001 for the diagnosis of LVH using CMR, with higher sensi-
CO (l/min) 6.1  1.5 6.0  1.5 0.421 tivities in the global cohort and particularly in the female
COI (l/min/m2) 3.0  1.3 3.0  1.1 0.976
SV (ml) 90.5  23.7 88.6  17.7 0.459 subgroup, while maintaining high specificity (above
SVI (ml/m2) 49.2  13.0 48.6  8.1 0.621 97%), even after adjusting for increased BMI. Another
Data are presented as mean  standard deviation or n (%). CO, cardiac output;
proposed reason was the fact that the hypertensive sub-
COI, cardiac output index; LA, left atrium; LAI, left atrium index; LVEDV, left group had significantly lower BP levels compared with
ventricular end-diastolic volume; LVEDVI, left ventricular end-diastolic volume the Peguero–Lo Presti foundational study. However,
index; LVESV, left ventricular end-systolic volume; LVESVI, left ventricular end-
systolic volume index; LVH, left ventricular hypertrophy; LVM, left ventricular mass; when compared with our study population, overall the
LVMI, left ventricular mass index; SV, stroke volume; SVI, stroke volume index. LVH group had lower SBP (148  21 vs. 158  19 mmHg)

© 2020 Italian Federation of Cardiology - I.F.C. All rights reserved.

442 Journal of Cardiovascular Medicine 2020, Vol 21 No 6

Table 4 Predictive performance of electrocardiographic indices for left ventricular hypertrophy in the global cohort (a) and according to male
(b) and female (c) sex
(a)

Criteria – global cohort (n ¼ 240) Sensitivity (%), 95% CI Specificity (%), 95% CI PPV (%), 95% CI NPV (%), 95% CI ACC (%), 95% CI

Peguero–Lo Presti voltage 47.0 (38.8–55.3) 95.6 (89.1–98.8) 94.6 (86.9–97.9) 52.4 (48.5–56.3) 65.4 (59.0–71.4)
Cornell voltage 28.8 (21.7–36.8) 97.8 (92.3–99.7) 95.6 (84.2–98.9) 45.6 (43.0–48.3) 55.0 (48.4–61.4)
Sokolow–Lyon voltage 24.8 (18.1–32.5) 94.5 (87.6–98.1) 88.1 (75.1–94.7) 43.4 (40.8–46.0) 51.2 (44.7–57.7)

(b)

Criteria – male (n ¼ 156) Sensitivity (%), 95% CI Specificity (%), 95% CI PPV (%), 95% CI NPV (%), 95% CI ACC (%), 95% CI

Peguero–Lo Presti voltage 42.6 (32.8–52.8) 94.6 (84.9–98.9) 93.5 (82.3–97.8) 47.3 (42.8–51.8) 60.9 (52.8–68.6)
Cornell voltage 20.8 (13.4–30.0) 100 (93.5–100) 100 40.7 (38.4–43.2) 48.7 (40.6–56.8)
Sokolow–Lyon voltage 25.7 (17.6–35.4) 92.7 (82.4–97.9) 86.7 (70.5–94.6) 40.5 (37.2–43.8) 49.4 (41.3–57.5)

(c)

Criteria – female (n ¼ 84) Sensitivity (%), 95% CI Specificity (%), 95% CI PPV (%), 95% CI NPV (%), 95% CI ACC (%), 95% CI

Peguero–Lo Presti voltage 56.3 (41.2–70.5) 97.2 (95.5–99.9) 96.4 (79.4–99.5) 62.5 (54.6–69.8) 73.8 (63.1–82.8)
Cornell voltage 45.8 (31.4–60.8) 94.4 (81.3–99.3) 91.7 (73.4–97.8) 56.7 (49.9–63.2) 66.7 (55.5–76.6)
Sokolow–Lyon voltage 22.9 (12.0–37.3) 97.2 (85.5–99.9) 91.7 (59.8–98.8) 48.6 (44.5–52.7) 54.8 (43.5–65.7)

ACC, accuracy; CI, confidence interval; NPV, negative-predictive value; PPV, positive-predictive value.

and DBP (81  13 vs. 88  11 mmHg) values. Also, we Peguero–Lo Presti could be better for female patients.
could find a slightly lower LVM (148  44 vs. 149  39) This result needs further investigation. Also, another
and LVMi (80  23 vs. 87  20) in our LVH group. important aspect to mention is that, although it is known
Probably, the heterogeneity inherent to the diagnostic that higher BMI is associated with lower ECG amplitudes
method used (echocardiography) and the ECG interpre- in patients with similar LV mass, decreasing the sensi-
tation by different cardiologists to account for the wide tivity of ECG voltage criteria for LVH,25 in our study, in
studied population (n ¼ 14 016) could have influenced the obese patients (BMI above 25 kg/m2) the Peguero–Lo
results negatively for the Peguero–Lo Presti criteria. Presti criteria remained the one with the highest sensi-
tivity (47.8% vs. Cornell 27.9% vs. Sokolow–Lyon
Our study brings a new input, as we applied the widely
18.0%), highlighting an important utility of these new
used Cornell and Sokolow–Lyon voltage criteria and
criteria in this specific patient population that need
compared with the recently proposed Peguero–Lo Presti
further analysis.
criteria, using CMR, the most accurate exam for LVM
evaluation and, consequently LVH diagnosis, obviating It should be noted that overall ECG voltage criteria
the interobserver variability on the acquisition of images sensitivity for LVH remains sub-optimal; however, based
and measures and also the unpredictable acoustic window on our findings, when applied to an all-comers European
that limits image quality acquisition. In our patient population, above 34 years old, Peguero–Lo Presti crite-
population, for the current LVH diagnostic cutoffs of ria improved it, and it therefore seems to be a better
the proposed Peguero–Lo Presti, Cornell and Sokolow– screening tool for LVH when compared with Cornell and
Lyon criteria, sensitivity analysis demonstrated overall Sokolow–Lyon voltage criteria.
better performance of Peguero–Lo Presti for diagnosing
LVH, particularly for female sex. This might be Study limitations
explained by differences in blood pressure (BP) and In this study, data were collected retrospectively. Nev-
LVMi distribution between both sexes with and without ertheless, we randomly selected a control group of age-
LVH, with female patients with LVH having a signifi- matched and sex-matched patients to mitigate for that
cantly higher SBP (154  31 vs. male 143  24 mmHg). bias. Every measurement made with CMR was per-
There was also a higher LVMi delta between female formed by the same observer, obviating a potential inter-
patients with and without LVH (DLVMI, female 28  17 observer bias. Second, not all patients underwent an
vs. male 24  11 g/m2), potentially indicating a more electrocardiogram on the same day as the CMR,
severe hypertrophy in female sex (data not shown). although, we excluded patients with more than 6 months’
Another explanation could be attributed to the small difference between both diagnostic exams to ensure their
sample size, which could have weakened the results accuracy. We also performed a subgroup analysis adjusted
for Peguero–Lo Presti in male patients or simply that for BMI, a known important confounder for ECG

© 2020 Italian Federation of Cardiology - I.F.C. All rights reserved.

 Peguero–Lo Presti criteria for diagnosis of LVH Guerreiro et al. 443

interpretation, which increased the consistency of our 8 Bellenger NG, Davies LC, Francis JM, Coats AJ, Pennell DJ. Reduction in
sample size for studies of remodeling in heart failure by the use of
results. We also only compared Peguero–Lo Presti to cardiovascular magnetic resonance. J Cardiovasc Magn Reson 2000;
Cornell and Sokolow–Lyon voltage criteria as those are 2:271–278.
9 Estes EH Jr, Jackson KP. The electrocardiogram in left ventricular
the most commonly used criteria and current guidelines hypertrophy: past and future. J Electrocardiol 2009; 42:589–592.
do not establish a definitive recommendation about the 10 Casale PN, Devereux RB, Kligfield P, et al. Electrocardiographic detection
use of one criterion over the other.23 of left ventricular hypertrophy: development and prospective validation of
improved criteria. J Am Coll Cardiol 1985; 6:572–580.
Further studies are still needed to confirm the results. 11 Sokolow M, Lyon TP. The ventricular complex in left ventricular hypertrophy
as obtained by unipolar precordial and limb leads. American Heart Journal
1949; 37:161–186.
Conclusion 12 Levy D, Salomon M, D’Agostino RB, Belanger AJ, Kannel WB. Prognostic
In an all-comers European population above 34 years old, implications of baseline electrocardiographic features and their serial
changes in subjects with left ventricular hypertrophy. Circulation 1994;
the recently proposed criteria of Peguero–Lo Presti 90:1786–1793.
showed increased sensitivity for the diagnosis of LVH, 13 Devereux RB, Casale PN, Eisenberg RR, Miller DH, Kligfield P.
when compared with the Sokolow–Lyon and Cornell Electrocardiographic detection of left ventricular hypertrophy using
echocardiographic determination of left ventricular mass as the reference
voltage criteria, using CMR as the gold-standard method standard. Comparison of standard criteria, computer diagnosis and
for LVM evaluation. As such, they could become the physician interpretation. J Am Coll Cardiol 1984; 3:82–87.
14 Rider OJ, Ntusi N, Bull SC, et al. Improvements in ECG accuracy for
preferred ECG diagnostic tool when evaluating patients diagnosis of left ventricular hypertrophy in obesity. Heart (British Cardiac
at risk for LVH. Society) 2016; 102:1566–1572.
15 Schillaci G, Verdecchia P, Borgioni C, et al. Improved electrocardiographic
diagnosis of left ventricular hypertrophy. Am J Cardiol 1994; 74:714–719.
Acknowledgements 16 Pewsner D, Juni P, Egger M, Battaglia M, Sundstrom J, Bachmann LM.
C.G. and P.A.: study concept, design and drafting of the Accuracy of electrocardiography in diagnosis of left ventricular hypertrophy
manuscript; A.B., R.L.L., N.F., J.A.: analysis and inter- in arterial hypertension: systematic review. BMJ 2007; 335:711.
17 Peguero JG, Lo Presti S, Perez J, Issa O, Brenes JC, Tolentino A.
pretation of data; B.M., R.L.L., R.F., N.F., J.P., P.B.: Electrocardiographic criteria for the diagnosis of left ventricular
critical revision of the manuscript for important hypertrophy. J Am Coll Cardiol 2017; 69:1694–1703.
intellectual content. 18 Sun G-Z, Wang H-Y, Ye N, Sun Y-X. Assessment of novel Peguero-Lo
Presti electrocardiographic left ventricular hypertrophy criteria in a large
Asian population: newer may not be better. Can J Cardiol 2018; 34:1153–
Conflicts of interest 1157.
19 Ramchand J, Sampaio Rodrigues T, Kearney LG, Patel SK, Srivastava PM,
There are no conflicts of interest. Burrell LM. The Peguero-Lo Presti electrocardiographic criteria predict
all-cause mortality in patients with aortic stenosis. J Am Coll Cardiol 2017;
References 70:1831–1832.
1 Marwick TH, Gillebert TC, Aurigemma G, et al. Recommendations on the 20 Jain A, Tandri H, Dalal D, et al. Diagnostic and prognostic utility of
use of echocardiography in adult hypertension: a report from the European electrocardiography for left ventricular hypertrophy defined by magnetic
Association of Cardiovascular Imaging (EACVI) and the American Society resonance imaging in relationship to ethnicity: the Multi-Ethnic Study of
of Echocardiography (ASE). J Am Soc Echocardiogr 2015; 28:727–754. Atherosclerosis (MESA). Am Heart J 2010; 159:652–658.
2 Dai S, Ayres NA, Harrist RB, Bricker JT, Labarthe DR. Validity of 21 Ha LD, Elbadawi A, Froelicher VF. Limited relationship of voltage criteria for
echocardiographic measurement in an epidemiological study. Project electrocardiogram left ventricular hypertrophy to cardiovascular mortality.
HeartBeat! Hypertension 1999; 34:236–241. Am J Med 2018; 131:101.e1–101.e8.
3 Hudsmith LE, Petersen SE, Francis JM, Robson MD, Neubauer S. Normal 22 Kawel-Boehm N, Maceira A, Valsangiacomo-Buechel ER, et al. Normal
human left and right ventricular and left atrial dimensions using steady state values for cardiovascular magnetic resonance in adults and children.
free precession magnetic resonance imaging. J Cardiovasc Magn Reson J Cardiovasc Magn Reson 2015; 17:29.
2005; 7:775–782. 23 Hancock EW, Deal BJ, Mirvis DM, et al. AHA/ACCF/HRS
4 Myerson SG, Bellenger NG, Pennell DJ. Assessment of left ventricular recommendations for the standardization and interpretation of the
mass by cardiovascular magnetic resonance. Hypertension 2002; electrocardiogram: part V: electrocardiogram changes associated with
39:750–755. cardiac chamber hypertrophy: a scientific statement from the American
5 Buchner S, Debl K, Haimerl J, et al. Electrocardiographic diagnosis of left Heart Association Electrocardiography and Arrhythmias Committee,
ventricular hypertrophy in aortic valve disease: evaluation of ECG criteria by Council on Clinical Cardiology; the American College of Cardiology
cardiovascular magnetic resonance. J Cardiovasc Magn Reson 2009; Foundation; and the Heart Rhythm Society. Endorsed by the International
11:18. Society for Computerized Electrocardiology. J Am Coll Cardiol 2009;
6 Grothues F, Smith GC, Moon JC, et al. Comparison of interstudy 53:992–1002.
reproducibility of cardiovascular magnetic resonance with two-dimensional 24 Okin PM, Roman MJ, Devereux RB, Kligfield P. Time-voltage area of the
echocardiography in normal subjects and in patients with heart failure or left QRS for the identification of left ventricular hypertrophy. Hypertension
ventricular hypertrophy. Am J Cardiol 2002; 90:29–34. 1996; 27:251–258.
7 Mor-Avi V, Sugeng L, Weinert L, et al. Fast measurement of left ventricular 25 Nasir JM, Rubal BJ, Jones SO, Shah AD. The effects of body mass index on
mass with real-time three-dimensional echocardiography: comparison with surface electrocardiograms in young adults. Journal of Electrocardiology
magnetic resonance imaging. Circulation 2004; 110:1814–1818. 2012; 45:646–651.

© 2020 Italian Federation of Cardiology - I.F.C. All rights reserved.