Electrocardiographic findings in peripartum cardiomyopathy

Background There is limited data on electrocardiographic (ECG) abnormalities and their prognostic significance in women with peripartum cardiomyopathy (PPCM). We sought to characterize ECG findings in PPCM and explore the association of ECG findings with myocardial recovery and clinical outcomes. Hypothesis We hypothesized that ECG indicators of myocardial remodeling would portend worse systolic function and outcomes. Methods Standard 12‐lead ECGs were obtained at enrollment in the Investigations of Pregnancy‐Associated Cardiomyopathy study and analyzed for 88 women. Left ventricular ejection fraction (LVEF) was measured by echocardiography at baseline, 6 months, and 12 months. Women were followed for clinical events (death, mechanical circulatory support, and/or cardiac transplantation) until 1 year. Results Half of women had an “abnormal” ECG, defined as atrial abnormality, ventricular hypertrophy, ST‐segment deviation, and/or bundle branch block. Women with left atrial abnormality (LAA) had lower LVEF at 6 months (44% vs 52%, P = 0.02) and 12 months (46% vs 54%, P = 0.03). LAA also predicted decreased event‐free survival at 1 year (76% vs 97%, P = 0.008). Neither left ventricular hypertrophy by ECG nor T‐wave abnormalities predicted outcomes. A normal ECG was associated with recovery in LVEF to ≥50% (84% vs 49%, P = 0.001) and event‐free survival at 1 year (100% vs 85%, P = 0.01). Conclusions ECG abnormalities are common in women with PPCM, but a normal ECG does not rule out the presence of PPCM. LAA predicted lower likelihood of myocardial recovery and event‐free survival, and a normal ECG predicted favorable event‐free survival.


| INTRODUCTION
Peripartum cardiomyopathy (PPCM) is a dilated cardiomyopathy marked by systolic dysfunction occurring at the end of pregnancy or, more commonly, in the early postpartum period. 1,2 Globally, incidence appears to be highest in Nigeria (up to 1 in 100 live births) 3

and Haiti
(1 in 300 live births). 4 In the United States, the incidence of PPCM is rising 5 and is approximately 4-fold higher in black women (1 in 1000-1500 live births) than in Caucasian women (1 in 4000 live births). 2,6 Other risk factors include older maternal age, multiple gestation, and preeclampsia. 7,8 The etiology of PPCM remains unclear; proposed mechanisms have included angiogenic factor imbalance, abnormal prolactin cleavage, inflammation, selenium deficiency, and genetic susceptibility. [9][10][11][12][13][14] More than half of women recover left ventricular (LV) function after PPCM, but a significant proportion is left with chronic heart failure, and some women require ventricular assist device (VAD) implantation or cardiac transplantation. In a recent North American series of 100 women, the Investigations of Pregnancy-Associated Cardiomyopathy (IPAC) study, 72% of women experienced recovery in LV ejection fraction (LVEF) to >50% by 12 months. 15 Predictors of persistent LV dysfunction included LVEF <30% or LV end-diastolic diameter > 6 cm at diagnosis, black race, and late presentation. 15 Elevation in troponin, B-type natriuretic peptide, and soluble fms-like tyrosine kinase-1 (sFlt-1) may also portend adverse cardiac remodeling and outcomes. [16][17][18] There is limited and conflicting data on ECG abnormalities and their significance in women with PPCM. In a Nigerian series of ECGs for 54 cases of PPCM and 77 postpartum women without PPCM, women with PPCM had faster heart rate, longer QRS and QTc intervals, and a higher frequency of ST-T-wave abnormalities than controls. 19 Of note, the authors did not examine the prognostic significance of ECG findings in this cohort. In a series of 78 South African women, 59% had T-wave abnormalities, 12% had a bundle branch block (BBB), 10% had left atrial abnormality, and 6% had ST-segment changes on baseline ECG. 20 Follow-up was available on only 56% of this cohort, but among women with follow-up, T-wave inversions and ST depressions on the presenting ECG were associated with lower LVEF at 6 months. 20 In a series of 77 women in Beijing with PPCM, neither QRS nor QTc interval nor the frequency of ST depressions differed between women who did and did not recover LV function. 17 As the phenotypic presentation of PPCM differs across continents, 21 ECG changes observed in Africa and Asia may not match those seen in North American PPCM patients, for whom there has been no data published on ECG findings to date.
Thus, we sought to characterize ECG findings at PPCM presentation in patients from the United States and Canada in the IPAC cohort and to explore the potential prognostic significance of specific ECG findings in this population. We hypothesized that ECG indicators of myocardial remodeling, such as ventricular hypertrophy and atrial abnormalities, would portend less recovery of systolic function and worse outcomes.

| METHODS
Between 2009 and 2012, 100 women at 30 participating sites in North America with newly diagnosed PPCM were enrolled in the IPAC study up to 13 weeks postpartum. Eligible women were 18 years of age or older, lacked underlying cardiac disease, had an LVEF <45% at enrollment, and had been ruled out for alternate etiologies of cardiomyopathy. All women had an echocardiogram at enrollment, 6, and 12 months, and these studies were reviewed by a core laboratory at the University of Pittsburgh for assessment of ventricular volumes and calculation of ejection fraction. Clinical events, including hospitalizations, mechanical circulatory support, cardiac transplantation, and death, were followed to 12 months postpartum. Institutional review boards at all participating centers approved the protocol and all patients signed informed consent.
Standard left-sided 12-lead electrocardiograms (ECGs) were obtained at the time of enrollment and were available for review in 88 of 100 women in the IPAC cohort. For 10 women, a written report of the enrollment ECG was available but not the ECG tracing itself, and two women had tracings of poor quality that were deemed uninterpretable; these 12 subjects were excluded. Features of each ECG tracing (eg, rate, rhythm, intervals, and amplitudes) were systemically analyzed in a blinded fashion by one investigator (M.C.H.), and a second investigator (M.M.G.) validated a random subset of ECG interpretations. The investigators reviewing ECGs were blinded to subject demographics, clinical presentation, echocardiograms, and outcomes.
A normal QRS axis was −30 to +90 . BBBs were defined per American Heart Association, American College of Cardiology, and Heart Rhythm Society guidelines. 22 Left atrial abnormality (LAA) was defined as terminal negative deflection of the P-wave in V 1 > 40 ms wide and > 1 mm deep. Right atrial abnormality was defined as a Pwave >2.5 mm tall in II and/or positive initial deflection of the Pwave in V 1 > 1.5 mm. 23 Left ventricular hypertrophy (LVH) was defined using the Sokolow-Lyon criteria (S in V 1 plus R in V 5 or Student t tests and Fisher exact tests were used to compare continuous and categorical variables between groups, respectively. The Kaplan-Meier log-rank analysis was used to estimate event-free survival, which was defined as survival free from death, mechanical circulatory support, and/or cardiac transplantation. Event-free survival was compared by characteristics of the ECG at entry, including LAA, LVH, ST segment depression, and a "normal" ECG by the exact log-rank test. In addition, the LVEF at 6 and 12 months postpartum was compared by ECG characteristics at entry.

| RESULTS
Of the 100 women in the IPAC cohort, 88 had available baseline ECGs. Demographic and clinical characteristics of these women are summarized in Table 1. Mean age was 30 ± 6 years, and 15 (17%) presented with multiple gestation. Diabetes was present in 10 (11%) and hypertension in 38 (43%). At baseline, mean systolic and diastolic blood pressures were 111 ± 17 mm Hg and 70 ± 13 mm Hg, respectively. LVEF at entry was 34% ± 10% and LV end-diastolic dimension was 5.6 ± 0.7 cm. 15 By 6 months, LVEF had increased to 51% ± 11%, and by 12 months to 53 ± 11%. Women with an "abnormal ECG" at study enrollment (defined as presence of BBB, ventricular hypertrophy, atrial abnormality, and/or ST-segment deviation) were more likely to receive inotropes and had a larger LV end-diastolic dimension (LVEDD) on baseline echocardiogram (58 vs 54 mm, P = 0.002). Six women experienced nine major events: four deaths, four LVAD implantations, and one cardiac transplantation. Of women who required an LVAD, two died and one later underwent cardiac transplantation.

| Electrocardiographic findings
Findings of 12-lead ECGs at study enrollment are summarized in Table 2. One woman was in atrial fibrillation, and all other subjects were in sinus rhythm; 45 (51%) had a normal sinus rhythm, 37 (42%) showed sinus tachycardia, and 5 (6%) showed sinus bradycardia. Ventricular ectopic beats were observed in 3 subjects. QRS axis was normal in 74 (84%). There was no first-, second-, or third-degree atrioventricular block. Two subjects had a QRS duration >120 ms; one met criteria for left BBB, and the other right BBB. LAA was observed in 15 (17%), right atrial abnormality in 5 (6%), and LVH in 8 (9%) ECGs.
A "normal" ECG, defined as absence of atrial abnormality, ventricular hypertrophy, ST-segment deviation, or BBB, was present in 43 (49%) of women. Tables 3 and 4 show the trajectory of LV function for women with and without various ECG and echocardiographic findings. As reported previously, 15 heart rate at enrollment was not correlated with LV recovery (P = 0.40 at 6 months and P = 0.26 at 12 months).

| Prediction of left ventricular recovery and event-free survival
LAA by ECG was specific (96%) but not sensitive (38%) for left atrial enlargement by echocardiogram. The presence of LAA on ECG was associated with lower LVEF at 6 months (44% vs 52%, P = 0.02) and 12 months (46% vs 54%, P = 0.03); these findings are almost identical to those comparing women with and without left atrial enlargement by echocardiogram (LVEF 46% vs 54% at 6 months (P = 0.003), and 47% vs 56% at 12 months (P = 0.001). As shown in Figure 1A, the presence of LAA by ECG additionally predicted decreased event-free survival at 1 year (76% vs 97%, P = 0.008).

Event-Free Survival (%)
FIGURE 1 (A) Survival free from mechanical circulatory support, cardiac transplantation, and/or death for women with and without left atrial abnormality (LAA) on electrocardiogram (ECG) at study enrollment. (B) Survival free from mechanical circulatory support, cardiac transplantation, and/or death for women with a "normal" ECG (no atrial abnormality, ventricular hypertrophy, ST-segment deviation, or bundle branch block) or "abnormal" ECG at study enrollment