Novel Electrocardiographic Screening Criterion for Hypertrophic Cardiomyopathy in Children Matthew B. Brothers, MD, Matthew E. Oster, MD, MPH, Alexandra Ehrlich, MPH, Margaret J. Strieper, DO, and William T. Mahle, MD* Electrocardiography is often advocated as a screening tool in children for hypertrophic cardiomyopathy (HC). We sought to establish an electrocardiographic screening tool to identify children with HC. We hypothesized that a pediatric-specific electrocardiographic criterion would perform better than the popular criteria used for screening children for left ventricular hypertrophy and HC. The earliest available electrocardiogram for children (n [ 108) with HC (ages 7 to 21 yrs) was reviewed. We sought to compare the diagnostic accuracy of 4 screening algorithms: (1) Sokolow-Lyon criterion (SV1 D RV5/RV6 >35 mm), (2) Cornell criterion (RaVL D SV3 >28 mm in men, 20 mm in women), (3) total 12-lead voltage criterion (R wave to the nadir of Q/S wave >175 mm), and (4) pediatric-specific criterion (RaVL D SV2 >23 mm). The same criteria were applied to a cohort of agematched and gender-matched controls without cardiac disease. Statistically significant correlations were found between children with HC and positive screen using all 4 criteria. However, comparison of receiver operating characteristic demonstrated an area under the curve of 0.67 for Sokolow-Lyon criterion, 0.70 for Cornell criterion, 0.83 for total 12-lead criterion, and 0.82 for pediatric-specific criterion. Pediatric-specific criterion had superior sensitivity in gene-positive children and superior overall specificity than total 12-lead criterion. In conclusion, our study demonstrates that the pediatric-specific criterion employing leads RaVL D SV2 is more accurate in identifying children with HC in comparison with other popular screening criteria. Ó 2014 Elsevier Inc. All rights reserved. (Am J Cardiol 2014;113:1246e1249)
Hypertrophic cardiomyopathy (HC) is the leading cause of sudden cardiac death in the young (2 standard deviations above the normal mean for body-surface area or localized ventricular hypertrophy: such as, septal thickness >1.5 left ventricular posterior wall thickness with at least normal left ventricular posterior wall thickness, with or without dynamic outflow obstruction. Genetic testing was offered to all patients with phenotypic findings and patients with family history of HC. A selection of patients, however, did not have genetic testing results for a variety of reasons including failure of insurance to cover testing, family refusal, and pending test results at the time of publication. Patients aged 3.5 mV14; (2) Cornell voltage criterion: RaVL þ SV3, with 0.8 mV added in women, >2.8 mV15; and (3) total www.ajconline.org
Cardiomyopathy/Pediatric-Specific Criterion for HC
1247
Table 1 Definition of electrocardiographic criteria ECG Criterion
Definition
Sokolow-Lyon
S wave in V1 þ R wave in V5/R wave in V6 >35 mm R wave in aVL þ S wave in V3 >28 mm in men, 20 mm in women Sum of total 12-lead voltage (R wave to the nadir of Q/S wave) >175 mm R wave in aVL þ S wave in V2 >23 mm
Cornell Total 12-lead voltage Pediatric-specific
ECG ¼ electrocardiographic.
Table 2 Demographic and clinical findings of study groups Characteristic Age (yrs) Male patients Race/ethnicity White Black Latino Asian Other Voltage criteria Sokolow-Lyon Cornell Total 12-lead voltage Pediatric-specific Echocardiographic Interventricular septal thickness in diastole (mm) Interventricular septal thickness in diastole z score LV posterior wall dimension in diastole (mm) LV posterior wall dimension in diastole z score Presence of LV outflow tract obstruction Genotype positive
Cases (n ¼ 108)
Controls (n ¼ 107)
12.2 3.6 12.3 3.1 77 (71) 57 (71) 66 29 10 1 2
(61) (27) (9) (1) (2)
58 40 4 5 0
(54) (37) (4) (6) (0)
45 40 96 77
(42) (37) (89) (71)
6 3 61 8
(93) (97) (43) (93)
17.2 8.9
—
3.8 3.3
—
12.5 5.4 2.5 2.4
— —
27 (25) 37 (54)
— —
Data are presented as mean SD or n (%).
12-lead voltage, R wave to the nadir of Q/S wave, >175 mm.9 Our modified pediatric-specific criterion was created based on the observation that V3 is frequently replaced with V3R or V4R and, therefore, may vary as a child ages. Additionally, Panza et al16 described patients with abnormalities of V1 or V2 before developing significant ventricular wall thickening, and Montgomery et al8 reported maximal voltages in leads V1 and V2 in patients with HC with phenotypic expression. Therefore, we used a model with RaVL þ SV2. A 2-sample t test was used to compare case-control differences. Chi-square analysis was used to determine the accuracy of each screening test in identifying HC. A receiver operating characteristic curve was created for each electrocardiographic screening criterion. Chi-square analysis was used to compare area under the curve (AUC) values. Data were analyzed using the SAS statistical software package (version 9.4; SAS Institute, Cary, North Carolina).
Figure 1. Areas under the receiver operating characteristic curves of Sokolow-Lyon, Cornell, total 12-lead voltage, and pediatric-specific criteria. AUC: Dashed line ¼ Sokolow-Lyon (0.67); Interrupted line ¼ Cornell (0.70); Broken line ¼ total 12 lead (0.83); Solid line ¼ pediatric specific (0.82).
Table 3 Comparison of sensitivity, specificity, and areas under the receiver operating characteristic curves of electrocardiographic criteria ECG Criterion Sokolow-Lyon Cornell Total 12-lead voltage Pediatric-Specific
Sensitivity (%) Specificity (%) AUC 42 37 89 71
94 97 43 93
p Value
0.67 0.0001 0.70 0.0019 0.83 0.8 0.82 Reference
ECG ¼ electrocardiographic.
Results Table 2 lists the clinical demographics, echocardiographic data, and electrocardiographic analysis. Male patients accounted for a slightly higher proportion of our cohort than previous reports. The mean diameter (expressed as a Z score) of the interventricular septum was more hypertrophied than the posterior wall of the LV (3.78 3.31 vs 2.52 2.44). LV outflow tract obstruction was present in 25% of patients with HC. Using the definition in the 2011 American College of Cardiology Foundation/American Heart Association guidelines for HC in children, 78% of patients met the clinical definition of HC of wall thickness 2 SDs.17 Of those who underwent genetic testing, 54% tested positive. The analysis of electrocardiographic recordings demonstrated that the Cornell criterion has the lowest sensitivity, whereas total 12-lead voltage had the highest sensitivity. When analyzing for patients who are phenotype positive, the sensitivities for Sokolow-Lyon, Cornell, total 12-lead voltage, and pediatric-specific criteria were 38%, 46%, 70%, and 73%, respectively. As demonstrated in Figure 1, pediatric-specific criterion and total 12lead voltage criterion had very similar AUC of 0.82 and 0.83, respectively. The AUC for pediatric-specific electrocardiographic criterion was statistically significant in comparison with Sokolow-Lyon and Cornell criteria (Table 3).
1248
The American Journal of Cardiology (www.ajconline.org)
Discussion In this study, which is the largest pediatric single-center study to determine and examine electrocardiographic criteria for HC screening, we found that pediatric-specific criterion for HC performs better than standard commonly used electrocardiographic screening methods. In a retrospective analysis, we demonstrated that voltage criterion derived from pediatric data is a better discriminatory test than criteria based on adult data. With comparable specificity, criterion derived from pediatric data proved to have superior sensitivity. Our findings of poor sensitivity of existing electrocardiographic criteria have been previously described.18e22 In fact, our reported sensitivity for Sokolow-Lyon and Cornell criteria is actually greater than some previous reports. However, as depicted by the AUC of the receiver operating characteristic curve, testing using pediatric-specific criterion is more accurate than both commonly used tests. Although total 12-lead voltage proved to have a greater AUC, this difference is not significant. Additionally, using the commonly reported cutoff of 175 mV,9 the specificity of the total voltage criterion would lead to a large number of false-positive results and unnecessary additional testing. Because HC is an unpredictable progressive disease, the need to diagnose as early as possible is imperative. However, no previous study has focused on children. Konno et al analyzed a cohort of patients aged
Hypertrophic cardiomyopathy (HC) is the leading cause of sudden cardiac death in the young (2 standard deviations above the normal mean for body-surface area or localized ventricular hypertrophy: such as, septal thickness >1.5 left ventricular posterior wall thickness with at least normal left ventricular posterior wall thickness, with or without dynamic outflow obstruction. Genetic testing was offered to all patients with phenotypic findings and patients with family history of HC. A selection of patients, however, did not have genetic testing results for a variety of reasons including failure of insurance to cover testing, family refusal, and pending test results at the time of publication. Patients aged 3.5 mV14; (2) Cornell voltage criterion: RaVL þ SV3, with 0.8 mV added in women, >2.8 mV15; and (3) total www.ajconline.org
Cardiomyopathy/Pediatric-Specific Criterion for HC
1247
Table 1 Definition of electrocardiographic criteria ECG Criterion
Definition
Sokolow-Lyon
S wave in V1 þ R wave in V5/R wave in V6 >35 mm R wave in aVL þ S wave in V3 >28 mm in men, 20 mm in women Sum of total 12-lead voltage (R wave to the nadir of Q/S wave) >175 mm R wave in aVL þ S wave in V2 >23 mm
Cornell Total 12-lead voltage Pediatric-specific
ECG ¼ electrocardiographic.
Table 2 Demographic and clinical findings of study groups Characteristic Age (yrs) Male patients Race/ethnicity White Black Latino Asian Other Voltage criteria Sokolow-Lyon Cornell Total 12-lead voltage Pediatric-specific Echocardiographic Interventricular septal thickness in diastole (mm) Interventricular septal thickness in diastole z score LV posterior wall dimension in diastole (mm) LV posterior wall dimension in diastole z score Presence of LV outflow tract obstruction Genotype positive
Cases (n ¼ 108)
Controls (n ¼ 107)
12.2 3.6 12.3 3.1 77 (71) 57 (71) 66 29 10 1 2
(61) (27) (9) (1) (2)
58 40 4 5 0
(54) (37) (4) (6) (0)
45 40 96 77
(42) (37) (89) (71)
6 3 61 8
(93) (97) (43) (93)
17.2 8.9
—
3.8 3.3
—
12.5 5.4 2.5 2.4
— —
27 (25) 37 (54)
— —
Data are presented as mean SD or n (%).
12-lead voltage, R wave to the nadir of Q/S wave, >175 mm.9 Our modified pediatric-specific criterion was created based on the observation that V3 is frequently replaced with V3R or V4R and, therefore, may vary as a child ages. Additionally, Panza et al16 described patients with abnormalities of V1 or V2 before developing significant ventricular wall thickening, and Montgomery et al8 reported maximal voltages in leads V1 and V2 in patients with HC with phenotypic expression. Therefore, we used a model with RaVL þ SV2. A 2-sample t test was used to compare case-control differences. Chi-square analysis was used to determine the accuracy of each screening test in identifying HC. A receiver operating characteristic curve was created for each electrocardiographic screening criterion. Chi-square analysis was used to compare area under the curve (AUC) values. Data were analyzed using the SAS statistical software package (version 9.4; SAS Institute, Cary, North Carolina).
Figure 1. Areas under the receiver operating characteristic curves of Sokolow-Lyon, Cornell, total 12-lead voltage, and pediatric-specific criteria. AUC: Dashed line ¼ Sokolow-Lyon (0.67); Interrupted line ¼ Cornell (0.70); Broken line ¼ total 12 lead (0.83); Solid line ¼ pediatric specific (0.82).
Table 3 Comparison of sensitivity, specificity, and areas under the receiver operating characteristic curves of electrocardiographic criteria ECG Criterion Sokolow-Lyon Cornell Total 12-lead voltage Pediatric-Specific
Sensitivity (%) Specificity (%) AUC 42 37 89 71
94 97 43 93
p Value
0.67 0.0001 0.70 0.0019 0.83 0.8 0.82 Reference
ECG ¼ electrocardiographic.
Results Table 2 lists the clinical demographics, echocardiographic data, and electrocardiographic analysis. Male patients accounted for a slightly higher proportion of our cohort than previous reports. The mean diameter (expressed as a Z score) of the interventricular septum was more hypertrophied than the posterior wall of the LV (3.78 3.31 vs 2.52 2.44). LV outflow tract obstruction was present in 25% of patients with HC. Using the definition in the 2011 American College of Cardiology Foundation/American Heart Association guidelines for HC in children, 78% of patients met the clinical definition of HC of wall thickness 2 SDs.17 Of those who underwent genetic testing, 54% tested positive. The analysis of electrocardiographic recordings demonstrated that the Cornell criterion has the lowest sensitivity, whereas total 12-lead voltage had the highest sensitivity. When analyzing for patients who are phenotype positive, the sensitivities for Sokolow-Lyon, Cornell, total 12-lead voltage, and pediatric-specific criteria were 38%, 46%, 70%, and 73%, respectively. As demonstrated in Figure 1, pediatric-specific criterion and total 12lead voltage criterion had very similar AUC of 0.82 and 0.83, respectively. The AUC for pediatric-specific electrocardiographic criterion was statistically significant in comparison with Sokolow-Lyon and Cornell criteria (Table 3).
1248
The American Journal of Cardiology (www.ajconline.org)
Discussion In this study, which is the largest pediatric single-center study to determine and examine electrocardiographic criteria for HC screening, we found that pediatric-specific criterion for HC performs better than standard commonly used electrocardiographic screening methods. In a retrospective analysis, we demonstrated that voltage criterion derived from pediatric data is a better discriminatory test than criteria based on adult data. With comparable specificity, criterion derived from pediatric data proved to have superior sensitivity. Our findings of poor sensitivity of existing electrocardiographic criteria have been previously described.18e22 In fact, our reported sensitivity for Sokolow-Lyon and Cornell criteria is actually greater than some previous reports. However, as depicted by the AUC of the receiver operating characteristic curve, testing using pediatric-specific criterion is more accurate than both commonly used tests. Although total 12-lead voltage proved to have a greater AUC, this difference is not significant. Additionally, using the commonly reported cutoff of 175 mV,9 the specificity of the total voltage criterion would lead to a large number of false-positive results and unnecessary additional testing. Because HC is an unpredictable progressive disease, the need to diagnose as early as possible is imperative. However, no previous study has focused on children. Konno et al analyzed a cohort of patients aged
