Early repolarization pattern (ERP) has recently been associated with idiopathic ventricular fibrillation and with cardiovascular mortality in the general population. We aimed to identify electrocardiographic tools to differentiate the “malignant” form of ERP from benign ERP in a population-based study. We retrospectively assessed the prevalence of ERP by recording electrocardiograms in 1,161 southwestern French subjects 35 to 64 years old. ERP was defined by an elevation of the J point ≥1 mm in 2 consecutive leads excluding leads V 1 through V 3 . We categorized ERP as notching or slurring pattern as located in inferior and/or lateral leads and measured the J-point elevation amplitude. ST segment after ERP was categorized as ascendant or horizontal/nonascendant and T waves as negative or positive. Association of ERP with all-cause and cardiovascular mortalities was assessed by adjusted Cox proportional hazard models. ERP was found in 157 subjects (13.3%). During a mean follow-up of 14.2 ± 2 years, 77 subjects died (6.6%), of whom 24 (2.1%) died from cardiovascular causes. Subjects with ERP had an increased hazard ratios for all-cause mortality (2.45, 95% confidence interval [CI] 1.44 to 4.15, p = 0.001) and cardiovascular mortality (5.60, 95% CI 2.27 to 11.8, p = 0.001). The highest risk was found for notching ERP and ERP with a nonascendant/horizontal ST segment, yielding when associated increased hazard ratios of 3.84 (95% CI 2.14 to 6.92, p = 0.001) and 8.75 (95% CI 3.48 to 22.0, p = 0.001) for all-cause and cardiovascular mortalities, respectively. Conversely, a slurring ERP or ascendant ST segment was not associated with increased mortality. ERP localization, J-point elevation amplitude, or T-wave morphology did not distinguish benign from malignant forms of ERP. In conclusion, ERP with notching pattern and horizontal/descendant ST segments was associated with the highest risk of all-cause and cardiovascular deaths. These electrocardiographic patterns may be used for risk stratification in subjects with ERP.
Early repolarization pattern (ERP) is a common electrocardiographic (ECG) pattern, especially in young athletic populations, but results are conflicting regarding the ERP prevalence in different populations, estimated at 1% to 14%. Evaluated arrhythmic mortality in healthy subjects with ERP is very low. Thus, the current problem is to identify the very few malignant variants of ERP from the vast majority of benign variants. Some ECG characteristics of ERP have been recently proposed to have greater predictive for cardiovascular or arrhythmic mortality, but results have been conflicting. The aim of this study was to assess the prevalence of ERP in a population of middle-aged healthy subjects and the prognostic significance of ERP characteristics on total and cardiovascular mortalities.
Methods
The study population was composed of subjects randomly recruited from the general population to participate in the third French Monitoring Trends and Determinants in Cardiovascular Disease (MONICA) Project. The objectives of the MONICA Project were to measure trends in cardiovascular mortality, coronary heart disease, and cerebrovascular disease morbidity and to assess the extent to which these trends are related to changes in known risk factors, daily living habits, health care, and major socioeconomic features measured at the same time in defined communities in different countries. The study was approved by an institutional ethic committee in agreement with French law on human biomedical research and the Declaration of Helsinki. The population-based sample was randomly selected using pooling lists from December 1994 to July 1997 in middle-aged men and women (35 to 64 years old) living in the southwestern France. The participation rate was 66%. The initial sample was comprised of 1,182 subjects.
Extensive questionnaires were completed by the participants with the help of trained and certified medical staff to collect data on age, socioeconomic status, occupational activity, personal and parental medical histories, physical activity, drug intake, and cardiovascular risk factors. Research nurses specially trained in agreement with the MONICA protocol performed clinical and anthropometric measurements and recorded standard 12-lead surface electrocardiograms while various biological variables were collected.
Presence of ERP was retrospectively assessed on standard surface electrocardiograms recorded previously (post hoc analysis). Electrocardiograms were recorded in each case at rest during the same conditions from 8:30 to 11:30 a . m . (12 leads, paper speed 25 mm/s, 1 mV = 10 mm). ERP was defined by an elevation (≥1 mm) of the QRS takeoff (J point) in ≥2 ECG leads in the inferior (II, III, VF) and/or lateral leads (I, VL, V 4 , V 5 , V 6 ). When ERP was present, we coded the subtype of ERP as “notching,” defined by a positive J deflection inscribed on the terminal QRS complex, or “slurring,” defined as a smooth transition from the terminal QRS complex to the ST segment. J-wave amplitude was measured at leads showing the highest J-point elevation. ST aspect was coded as “concave/rapidly ascending,” defined by a persistently elevated ST segment ≥0.1 mV throughout the ST segment or an ascending elevation ≥0.1 mV within 100 ms after the J point, or “horizontal/descending,” defined as ST elevation <0.1 mV within 100 ms after the J point according to the proposed criteria. ECG interpretation was done by 2 independent trained cardiologists (P.M. and A.R.) blinded to clinical and follow-up data.
Vital status was obtained in June 2011 for each participant through the French national database recording all deaths occurring in the French population. All-cause mortality and specific causes of death were available until December 2010. Death certificates were checked for supporting clinical and postmortem information on cause of death. A medical committee including 2 medical doctors and 2 epidemiologists was established to examine medical data relating to all deaths. The first outcome analyzed was all-cause mortality and the second outcomes were cardiovascular mortality (including sudden cardiac death [SCD]) and noncardiovascular mortality.
Interobserver agreement for ERP determination was assessed using kappa statistic. When results of ERP determination conflicted between the 2 cardiologists, a new reading of electrocardiograms was performed and a consensus was sought. ERP prevalence was presented with 95% confidence interval for age and gender. Baseline characteristics and mortality rates were compared according to presence or absence of ERP. Qualitative variables were compared between groups using chi-square test or Fisher’s exact test in case of small expected numbers. Student’s t test was used to compare the distribution of quantitative data. Variables were logarithmically transformed when distribution departed from normality or when homoscedasticity was rejected. The relation between baseline variables and mortality was assessed using Cox proportional hazards regression analysis. Adjustment was made systematically for variables tested in an initial analysis before ERP was considered. Validity of the proportional hazards ratios (HRs) assumption for covariates was verified. We tested the proportionality assumption using cumulative sums of martingale-based residuals.
Likelihood ratio test of 2 consecutive nested models was used to assess the statistical contribution of adding ERP. As a complementary discriminating analysis, contribution of ERP was tested in reclassification approaches. Net reclassification improvement and integrated discrimination improvement were calculated. Reclassification was assessed after introducing ERP to the basic model. Predicted risk of death was categorized in 3 levels: 0% to 6%, 7% to 15%, and >15%.
Analyses were conducted for all-cause mortality and for cardiovascular and noncardiovascular mortalities. Interactions between gender and ERP and between age and ERP were tested systematically for each final model. All statistical analyses were carried out using SAS (SAS Institute, Cary, North Carolina). All tests were considered statistically significant at a p value ≤0.05.
Results
Twenty-one subjects were excluded from the initial sample (n = 1,182) because electrocardiograms were unreadable (n = 13) or missing (n = 1) or subjects presented with complete bundle branch block (n = 6) or paced cardiac beats (n = 1). Therefore, 1,161 subjects (599 men and 562 women) were considered in the present study. Interobserver kappa coefficient for the determination of ERP was 0.89. Clinical and biological characteristics of the population are presented in Table 1 .
| Variables | Entire Population | ERP | p Value | |
|---|---|---|---|---|
| No | Yes | |||
| (n = 1,002) | (n = 159) | |||
| Men | 599 (51.6%) | 473 (47.2%) | 126 (79.3%) | 0.001 |
| Age (years) | 49.8 ± 8.6 | 50.0 ± 8.5 | 48.3 ± 8.9 | 0.02 |
| Diabetes mellitus | 36 (3.1%) | 28 (2.8%) | 8 (5.0%) | 0.14 |
| Dyslipidemia | 160 (13.8%) | 140 (14.0%) | 20 (12.6%) | 0.64 |
| Hypertension | 169 (14.6%) | 148 (14.8%) | 21 (13.2%) | 0.65 |
| Vascular diseases ⁎ | ||||
| Personal history | 33 (2.8%) | 24 (2.4%) | 9 (5.7%) | 0.03 |
| Parental history | 344 (29.6%) | 302 (30.1%) | 42 (26.4%) | 0.34 |
| Intense physical activity † | 542 (46.7%) | 467 (46.6%) | 75 (47.2%) | 0.90 |
| Current smoker | 248 (21.4%) | 214 (21.4%) | 34 (21.4%) | 0.99 |
| Pack-year | 13.4 ± 20.7 | 13.1 ± 20.5 | 15.3 ± 21.3 | 0.22 |
| Alcohol (g/day) | 17.7 ± 23.8 | 17.1 ± 23.6 | 21.5 ± 24.7 | 0.04 |
| Waist (cm) | 87.2 ± 12.9 | 87.0 ± 13.3 | 88.5 ± 10.4 | 0.17 |
| Body mass index (kg/m 2 ) | 25.5 ± 4.2 | 25.5 ± 4.3 | 25.3 ± 3.7 | 0.56 |
| Systolic blood pressure (mm Hg) | 131.2 ± 18.0 | 131.3 ± 18.3 | 130.2 ± 15.6 | 0.48 |
| Heart rate (beats/min) | 67.4 ± 9.0 | 67.6 ± 9.0 | 65.1 ± 8.8 | 0.001 |
| Blood glucose (mg/dl) | 103 ± 22 | 102 ± 21 | 104 ± 25 | 0.53 |
| C-reactive protein (mg/L) | 1.7 ± 3.3 | 1.7 ± 3.4 | 1.4 ± 2.3 | 0.28 |
| Total cholesterol | ||||
| mmol/L | 5.83 ± 1.01 | 5.84 ± 0.99 | 5.77 ± 1.13 | 0.38 |
| mg/dl | 226 ± 39 | 226 ± 38 | 223 ± 44 | |
| High-density lipoprotein cholesterol | ||||
| mmol/L | 1.46 ± 0.44 | 1.48 ± 0.44 | 1.35 ± 0.41 | 0.001 |
| mg/dl | 57 ± 17 | 57 ± 17 | 52 ± 16 | |
| Triglyceride | ||||
| mmol/L | 1.18 ± 0.73 | 1.18 ± 0.75 | 1.16 ± 0.58 | 0.88 ⁎§ |
| mg/dl | 103 ± 64 | 103 ± 66 | 102 ± 51 | |
| Apolipoprotein A1 (mg/dl) | 162 ± 26 | 1.66 ± 26 | 159 ± 24 | 0.006 |
| Apolipoprotein B (mg/dl) | 120 ± 27 | 120 ± 27 | 122 ± 28 | 0.43 |
| Fibrinogen (g/L) | 3.42 ± 0.71 | 3.44 ± 0.71 | 3.29 ± 0.65 | 0.02 |
| γ-Glutamyl transferase (IU/L) | 48.1 ± 60.0 | 48.8 ± 63.1 | 43.8 ± 34.2 | 0.75 ⁎§ |
| Hemoglobin (g/dl) | 13.8 ± 1.2 | 13.8 ± 1.2 | 14.2 ± 1.0 | 0.001 |
| Renal insufficiency ‡ | 2 (0.2%) | 1 (0.1%) | 1 (0.6%) | 0.26 †□ |
| Cancer | 3 (0.3%) | 2 (0.2%) | 1 (0.6%) | 0.36 †□ |
| β Blockers | 75 (6.5%) | 64 (6.4%) | 11 (6.9%) | 0.89 |
| Diuretics | 41 (3.5%) | 36 (3.6%) | 5 (3.1%) | 0.78 |
| Angiotensin-converting enzyme inhibitor | 43 (3.7%) | 39 (3.9%) | 4 (2.5%) | 0.39 |
| Calcium inhibitor | 26 (2.2%) | 23 (2.3%) | 3 (1.9%) | 0.89 †□ |
| Antiplatelet agents | 15 (1.3%) | 12 (1.2%) | 3 (1.9%) | 0.45 †□ |
| Fibrates | 73 (6.3%) | 61 (6.1%) | 12 (7.6%) | 0.48 |
| Statins | 65 (5.6%) | 60 (6.0%) | 5 (3.1%) | 0.15 |
⁎ Ischemic heart disease (n = 16), stroke (n = 1), or peripheral artery disease (n = 6).
† At least 20 minutes per session 2 times a week.
‡ Renal transplantation or dialysis.
ERP was present in 159 subjects (13.7%, 95% confidence interval [CI] 11.8 to 15.8). Prevalence of ERP was higher in men (126 of 599, 21.0%, 95% CI 17.8 to 24.5) than in women (33 of 652, 5.9%, 95% CI 4.1 to 8.2, p <0.001). Prevalence of ERP significantly decreased with age: 62 of 368 35- to 44-year-old subjects (16.9%, 95% CI 13.2 to 21.1) compared to 51 of 395 45- to 54-year-old subjects (12.9%, 95% CI (9.8 to 16.6) and 46 of 398 55- to 64-year-old subjects (11.6%, 95% CI 8.6 to 15.1, p for trend = 0.04). This decrease, however, depended on gender: in men, ERP frequency significantly decreased with age (35 to 44 years old, 27.1%; 45 to 54 years old, 21.4%; 55 to 64 years old, 15.1%, p for trend = 0.004) but remained relatively stable in women (35 to 44 year old, 5.7%; 45 to 54 years old, 4.1%; 55 to 64 years old, 7.8%, p for trend = 0.37).
ERP was located in the inferior leads only in 79 subjects (49.7%), in the lateral leads only in 42 subjects (26.4%), and in the inferior and lateral leads in 38 subjects (23.9%). Seventy-one subjects (44.6%) presented a slurring pattern only and 77 (48.4%) a notching pattern only, whereas 11 subjects (6.9%) present with slurring and notching patterns. ST segment was concave/ascending in 56 subjects (35.2%) and horizontal/descending in 103 subjects (64.8%). Some subjects had concave/ascending and horizontal/descending patterns in different ECG leads: these subjects were classified as having the horizontal/descending pattern because of the poorer prognosis associated with this form (see further) as also used by others. T wave after ERP was negative in 58 subjects with ERP (36.5%). J-point elevations were ≥2 mm in 74 subjects (46.6%) and 1 to 2 mm in 85 subjects (53.4%).
Baseline characteristics of the study population according to ERP status are presented in Table 1 .
During a mean follow-up of 14.2 ± 2 years (median 15.0 years 45 days to 16 years), 77 subjects died (6.6%), 2 times as often in men (n = 51, 8.5%) than in women (n = 26, 4.6%). Twenty-four subjects (2.1%) died from cardiovascular causes (including SCD). Mean ages at death were 55.0 ± 8.2 years for all-cause mortality and 58.6 ± 6.5 years for cardiovascular mortality.
The following variables were maintained in the model because they were significantly associated with all-cause mortality: age, smoking habits, treatment for diabetes and hypertension, and γ-glutamyl transferase (p <0.05). Gender, which was not significantly related to mortality, was forced into and maintained in the model. In all further models adjustments were done for these variables. Further adjustment for heart rate at rest, alcohol consumption, and apolipoprotein A-I did not modify the results.
Mortality was 2 times higher in subjects with ERP (12.6%) than in subjects without ERP (5.7%, p = 0.002). The association between all-cause mortality and ERP was much greater in women (HR 5.34, p = 0.003) than in men (4 times higher, p = 0.001), whereas the difference was not significant when unadjusted (HR 1.3, p = 0.28; Table 2 ). Likelihood ratio test between the basic Cox regression model and the basic model plus ERP was chi-square equal to 9.57 (p = 0.002). Compared to the basic Cox regression model, the contributions of ERP were 1.6 ± 0.7% (p = 0.023) and 12.1 ± 5.5% (p = 0.027) for integrated discrimination improvement and net reclassification improvement, respectively.
| Without ERP | With ERP | p Value | HR | 95% CI | |||
|---|---|---|---|---|---|---|---|
| Subjects | Percentage (95% CI) | Subjects | Percentage (95% CI) | ||||
| Entire sample | 57/1,002 | 5.7 (4.3–7.1) | 20/159 | 12.6 (7.9–18.8) | 0.002 | 2.31 | 1.39–3.84 |
| Men | 37/473 | 7.8 (5.6–10.6) | 14/126 | 11.1 (5.6–16.6) | 0.28 | 1.45 | 0.78–2.68 |
| Women | 20/529 | 3.8 (2.3–5.8) | 6/33 | 18.2 (7.0–35.5) | 0.003 | 5.34 | 2.14–13.3 |
| 35–44 years old | 9/306 | 2.9 (1.4–5.5) | 1/62 | 1.6 (0.4–8.7) | 0.99 | 0.55 | 0.07–4.32 |
| 45–54 years old | 15/344 | 4.4 (2.5–7.1) | 5/51 | 9.8 (3.3–21.4) | 0.16 | 2.33 | 0.85–6.42 |
| 55–64 years old | 33/352 | 9.4 (6.5–12.9) | 14/46 | 30.4 (17.7–45.8) | 0.001 | 3.71 | 1.98–6.93 |
The strong positive association between ERP and all-cause mortality increased with cardiovascular mortality ( Table 3 ). Actuarial survival curves for all-cause and cardiovascular mortalities according to presence of ERP are shown in Figure 1 . There was a significant interaction between ERP and gender in all-cause mortality (chi-square = 2.79, p = 0.10) and noncardiovascular mortality (chi-square = 3.58, p = 0.06). Stratification analyses after multivariate adjustment yielded HRs of 1.94 (95% CI 1.03 to 3.66, p = 0.04) in men and 4.77 (95% CI 1.87 to 12.2, p = 0.002) in women for all-cause mortality and 1.16 (95% CI 0.46 to 2.90, p = 0.76) and 4.09 (95% CI 1.26 to 13.2, p = 0.02) in men and women for noncardiovascular mortality. No significant interaction was found between gender and ERP for cardiovascular mortality (chi-square = 0.30, p = 0.64). After adjustment, HRs for cardiovascular mortality were 4.68 (95% CI 1.67 to 13.1, p = 0.004) in men and 7.11 (95% CI 1.29 to 39.2, p = 0.02) in women.
| HR | 95% CI | p Value | |
|---|---|---|---|
| Total mortality | 2.31 | 1.39–3.84 | 0.002 ⁎ |
| 2.45 ‡ | 1.44–4.15 | 0.001 † | |
| Noncardiovascular mortality | 1.73 | 0.86–3.47 | 0.13 ⁎ |
| 1.83 § | 0.89–3.77 | 0.10 † | |
| Cardiovascular mortality | 4.91 | 2.07–11.6 | 0.001 ⁎ |
| 5.60 | 2.27–11.8 | 0.001 † |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
