Despite its original description in the medical literature during the nineteenth century by Ritchie and Virchow, peripartum cardiomyopathy remains a diagnosis associated with great uncertainty.1 The incidence is estimated to be 1 in 3,000–4,000 deliveries in the United States, with 1,000–1,350 new cases reported annually,2,3 although tertiary centers such as ours often report much higher incidences.4,5 Although a precise etiology for peripartum cardiomyopathy is unknown, there is much current interest in infectious, autoimmune, and genetic factors that may play a role in the inciting event.6 Epidemiologic risk factors, well known to the obstetrician, include age greater than 30 years, black race, multiple gestation, obesity, preeclampsia, and chronic hypertension.7 Peripartum cardiomyopathy historically has had a 25–50% mortality rate, a rate that may still persist at some centers, with nearly one-half of all related deaths occurring within the first 3 months after delivery.8 Despite advances in medical therapy and the development of cardiac transplantation, the condition remains a significant cause of maternal mortality, accounting for 6% of all maternal deaths among liveborn gestations in the United States.9
Recently, we described specific echocardiographic criteria to define the disorder,10 providing a more objective definition than the traditional subjective parameters for peripartum cardiomyopathy,11 because echocardiography is now the standard noninvasive tool for measuring cardiac function, quantifying left ventricular performance and providing a definitive diagnosis of left ventricular dysfunction. Similar to idiopathic dilated cardiomyopathy, O'Connell et al12 suggested that survivors of peripartum cardiomyopathy have better echocardiographic ventricular function at the time of diagnosis compared with those who succumb. Witlin et al,13 in a series of only 9 patients, suggested that fractional shortening and end diastolic dimension at the time of diagnosis were prognostic. Although most patients with persistent cardiac dysfunction after the index pregnancy are advised to avoid subsequent pregnancies, those parturients who experience prompt resolution of their symptoms and normalization of cardiac function may be more problematic to counsel.
After recently reviewing and analyzing existing data, the National Heart, Lung, and Blood Institute of the National Institutes of Health concluded that uncertainty regarding many aspects of peripartum cardiomyopathy warrants further investigation and reporting of data.14 We have undertaken the current work to review our own experience with peripartum cardiomyopathy and to determine whether echocardiographic parameters at the time of diagnosis are predictive of outcome in our cohort of patients. Furthermore, we wished to evaluate outcomes in those women undertaking a subsequent pregnancy after a diagnosis of peripartum cardiomyopathy.
MATERIALS AND METHODS
We performed an analysis of all patients with the diagnosis of peripartum cardiomyopathy who were cared for at the University of Chicago from May 1988 to March 2001. Institutional review board approval was obtained before beginning this investigation. Using International Classification of Diseases, 9th Revision, discharge summary codes from cardiology outpatient clinics, echocardiographic laboratory records, and a computerized obstetric research database, we identified women with the potential diagnosis of peripartum cardiomyopathy. The inpatient and outpatient medical records of these patients, including echocardiographic data, were reviewed to confirm the diagnosis. In each case, other potential causes of heart failure in late gestation, including severe preeclampsia, fluid overload, and amniotic fluid embolism, were determined not to be present. We used our echocardiographic-based precise definition for peripartum cardiomyopathy, which requires a fractional shortening calculation of less than 30% and left ventricular end diastolic dimension 4.8 cm or greater at the time of diagnosis.10Fractional shortening is defined as left ventricular end diastolic dimension minus left ventricular end systolic dimension divided by left ventricular end diastolic dimension (LVEDD − LVESD/LVEDD). Patients who did not meet these criteria were excluded from our analysis.
Epidemiologic data, including age, race, and parity, were recorded, as well as information on timing of presentation in relation to gestation, presenting signs and symptoms, risk factors, and subsequent pregnancy outcome. Patients were monitored with echocardiography after delivery. Those who demonstrated either a return of fractional shortening to more than 30% or left ventricular end diastolic dimension to less than 4.8 cm at the time of follow-up were classified as recovered. Patients who did not meet either of these criteria at follow-up were classified as having persistent dysfunction. If available, data from long-term assessment of patients with left ventricular dysfunction were collected, with particular emphasis on outcomes such as need for transplantation or patient death. For those who undertook a subsequent pregnancy, the same criteria for follow-up were used.
Statistical analysis included comparison between patients with recovery of left ventricular function and those with residual dysfunction using t tests for continuous variables and Fisher exact test for categorical variables. For left ventricular end diastolic dimension and fractional shortening results, the distribution of values for both the recovered and persistent dysfunction subgroups were assessed with the Lilliefors test for normality. Values for skewness and kurtosis were determined, and there was little evidence found against a normal distribution for these results among either of the subgroups. Relative risk (RR) and 95% confidence intervals for recovery of left ventricular function were calculated, as were P values for significance. We selected the values of left ventricular end diastolic dimension greater than 6.0 cm and fractional shortening less than 20% at the time of initial diagnosis as risk factors for persistent dysfunction based on data from a previous study.13
During the study period, 40,200 live births occurred at the University of Chicago. Of the 56 patients identified as potentially suffering peripartum cardiomyopathy, 35 met our criteria for diagnosis and were included in this investigation, corresponding to an incidence of 1 case per 1,149 live births. Three patients with peripartum cardiomyopathy were lost to follow-up. Twenty-one patients who did not meet the criteria for this study were excluded. Preexisting disease, such as idiopathic dilated cardiomyopathy or concurrent cardiovascular risk factors, including hypertension and alcoholism, were the primary reasons for patient exclusion, as well as 6 cases with insufficient data.
Eighty percent of study patients were African American, while the remainder were white, corresponding to the overall obstetric population at our institution. The mean age at the time of diagnosis of peripartum cardiomyopathy was 27 ± 6 years, with a range of 16–38 years. Parity in the index pregnancy ranged from 1 to 6, with a median value of 2. The diagnosis of peripartum cardiomyopathy was made in the antepartum period in 7 gravidas and in the postpartum period in 28 women. Included in our analysis are 2 patients who presented 2 and 3 months before delivery and one who presented at 7 months postpartum. Four women had multifetal gestations (3 twin, 1 triplet). Five gravidas had preeclampsia and an additional five had been treated with tocolytics for preterm labor with oral terbutaline for 6–56 days (mean 24.5 days). The most common presenting symptoms and signs were dyspnea (90%), tachycardia (62%), or peripheral edema (59%). One patient had a concomitant diagnosis of cerebral vascular accident. Only 2 women underwent endomyocardial biopsy, but neither procedure was contributory to the diagnosis because one revealed no abnormality and the other sample was insufficient for diagnostic purposes.
At diagnosis, the mean fractional shortening of the entire study group was 17.3 ± 7.0%, with a mean left ventricular end diastolic dimension of 6.4 ± 0.8 cm. Of the remaining 32 patients, 13 (41%) had recovered left ventricular function at a median follow-up period of 3 months, while 19 (59%) had persistent ventricular dysfunction at a median of 46 months follow-up. Mean fractional shortening and left ventricular end diastolic dimension values, at the time of diagnosis and at follow-up, are presented for both of these groups in Table 1. Fractional shortening and left ventricular end diastolic dimension are significantly different between the group that recovered left ventricular function and the group with persistent ventricular dysfunction. An initial fractional shortening below a threshold value of 20% at the time of diagnosis was associated with a 3-fold increased risk of persistent left ventricular dysfunction at follow-up (Table 2). Further, a left ventricular end diastolic dimension of 6.0 cm or greater at the time of the initial diagnosis was associated with more than a 3.5 times increased risk for persistent ventricular dysfunction compared with those who recovered (Table 2). With regard to demographic characteristics and risk factors, there were no significant differences between women whose cardiac function improved and returned to normal and those who remained with significant impairment (Table 1). Outcomes following the index pregnancy are presented in Figure 1. Two patients underwent cardiac transplantation (6.5%), and 3 died from heart failure (9.6%), including one of the transplant recipients.
Regarding outcomes with subsequent pregnancy after the index pregnancy, 6 of 35 patients had 8 documented subsequent gestations (Figure 2). Four of the patients with recovered left ventricular function had one additional pregnancy each, while the 2 patients with persistent left ventricular dysfunction had 1 and 3 subsequent pregnancies, respectively. All 4 patients with recovered left ventricular function became symptomatic and demonstrated echocardiographic evidence of recurrent left ventricular dysfunction in the third trimester. Three of these gravidas with recurrence of peripartum cardiomyopathy then had persistent left ventricular dysfunction after the subsequent pregnancy, while one again demonstrated full cardiac recovery. The 2 patients with persistent left ventricular dysfunction after the index pregnancy had no recurrence of symptoms in their subsequent gestations, but continued to have stable left ventricular dysfunction, both during and after the pregnancies (Figure 2).
Peripartum cardiomyopathy remains a significant cause of maternal morbidity and mortality in the United States. The increased incidence of peripartum cardiomyopathy at our center is likely due in part to our role as a tertiary care referral center for a large perinatal network, which serves an urban, primarily African-American population.
Echocardiography is an important tool in properly diagnosing peripartum cardiomyopathy and assessing the degree of cardiac dysfunction. In this work we have confirmed that the specific parameters of fractional shortening and left ventricular end diastolic dimension at the time of diagnosis are predictive of the degree of recovery of cardiac function at follow-up. Those women with a fractional shortening less than 20% or left ventricular end diastolic diameter of 6.0 cm or greater on initial echocardiogram incurred more than a 3 times increased risk of not fully recovering their left ventricular function, and in fact, 3 patients in this group suffered mortality. No demographic risk factors differed between those women who recovered and those who continued with ventricular impairment. The prognostic value of fractional shortening and left ventricular end diastolic dimension may be beneficial in guiding treatment for and counseling these patients.
Our findings are in accord with the observations of O'Connell et al12 who initially suggested that echocardiographic parameters, such as markedly reduced ejection fraction and increased left ventricular cavity size at the time of diagnosis, are associated with increased maternal mortality. More recently, Witlin et al13 studied a cohort of 9 peripartum cardiomyopathy patients and noted that left ventricular end diastolic dimension of 6.0 cm or greater and fractional shortening of 21% or less were risk factors for persistent left ventricular dysfunction. The results of this study confirm these findings in a significantly larger cohort of peripartum cardiomyopathy patients. The morbidity in our patients, measured in terms of persistent left ventricular dysfunction, is similar to previous reports, but the mortality rate was lower than that documented by other investigators.8,15,16 Advances in cardiac critical care, the increasing availability of cardiac transplantation, and the exclusion of other cardiac disease with echocardiography are likely causes of the improvement in overall mortality figures.
Women with a history of peripartum cardiomyopathy remain at high risk for recurrence of cardiac dysfunction in subsequent pregnancies, despite seemingly full recovery. Our data, although limited by the small number of patients embarking on a new conception, suggest that recovery of left ventricular function after the index gestation, as documented by echocardiography, is not predictive of recurrence or outcome in subsequent pregnancies in our population. We were surprised that all 4 of the completely recovered women had recurrent left ventricular dysfunction during pregnancy. Sutton et al17 and DeSouza et al18 describe much better success with subsequent pregnancy in their recovered peripartum cardiomyopathy patients compared with our cohort, although each report includes only 4 and 7 patients, respectively. In contrast, Elkayam et al,19 in the largest series to date, have documented significant deleterious effects of subsequent pregnancies in women with a history of peripartum cardiomyopathy who have echocardiographic evidence of residual dysfunction.
Perhaps additional factors are better indicative of the heart's ability to compensate for the increased hemodynamic demands of a subsequent pregnancy. Lampert et al20 demonstrated that patients diagnosed with peripartum cardiomyopathy and completely recovered left ventricular function actually had evidence of decreased left ventricular contractile reserve, as demonstrated by a dobutamine challenge test, suggesting that these women may indeed be at risk for recurrence of disease in future pregnancies. Because peripartum cardiomyopathy is associated with multiparity, it may be that each subsequent pregnancy in susceptible individuals further compromises contractile reserve and is more likely to lead to persistent cardiac dysfunction. Still, a number of other factors, as yet undefined, including genetic and autoimmune, may underlie and contribute to outcomes in these women. Surprisingly, in our experience the gestations that occurred in women with residual cardiac dysfunction after the index pregnancy were uncomplicated and not marked by a significant worsening of symptoms or cardiac function. These cases, however, are only 2 in number, and both patients were receiving ongoing care and maintained on appropriate medications. Thus, we caution strongly against the extrapolation of this limited data to other similar cases.
Improved outcomes in index cases of peripartum cardiomyopathy and in subsequent pregnancies are primarily due to advances in cardiac care, both medical and surgical. Pharmacologic therapy for heart failure has evolved tremendously and is now based on sound therapeutic principles that include afterload reduction and increasing myocardial contractility. The former can be attained effectively during gestation with careful use of a potent vasodilator, such as hydralazine, whereas the latter aim may be achieved with digoxin and other pressor agents. Angiotensin-converting enzyme inhibitors are also an excellent choice in the postpartum setting for afterload reduction, but these drugs can lead to embryopathy when used during pregnancy. Fluid restriction and judicious use of diuretics may also alleviate some of the symptoms associated with congestive heart failure in the gravida. Finally, improvements in cardiac critical care and the emergence of cardiac transplantation have provided hope to patients who previously would not have survived.
We have chosen to include 3 patients in our cohort whose timing of disease is outside the previously suggested time period for diagnosis.11 These women met all the stringent echocardiographic criteria, as well as all subjective criteria, with the exception of timing. We believe the time element is no longer relevant, and in view of current diagnostic tools, medications, and cardiac intensive care, a diagnosis of peripartum cardiomyopathy antenatally does not automatically imply immediate delivery, particularly if the fetus is quite premature. Thus, a gravida diagnosed with peripartum cardiomyopathy in the second or third trimester, more than 1 month before delivery, may indeed carry the gestation much closer to term, allowing for improved fetal maturity. Similarly, the patient diagnosed with cardiomyopathy 7 months after delivery may have had echocardiographic evidence of cardiac dysfunction long before her presentation. Thus, we believe that timing relative to gestation may be a somewhat arbitrary consideration, and inclusion of gravidas who otherwise meet all diagnostic criteria is warranted.
We acknowledge that because of the small sample sizes in this study, which is due to the rarity of this disease, some of our observations may have limited statistical power. Additionally, the cases occurred over a 13-year time period, during which significant advances in cardiac care have been made. Thus, the prognosis and outcome for patients who presented in the more recent past would be expected to be improved over those patients presenting earlier. However, further subanalysis of the data, according to year of diagnosis during the study period, was not able to be performed because of the small sample sizes. These limitations emphasize the need for larger, prospective multicenter studies.
Echocardiography appears to be extremely valuable in diagnosing peripartum cardiomyopathy, formulating prognosis for recovery, and following the course of the disease. Specific echocardiographic parameters, including fractional shortening and left ventricular end diastolic dimension at the time of diagnosis, may be predictive of long-term cardiac dysfunction. Subsequent pregnancy in women who have been diagnosed with peripartum cardiomyopathy should be approached with extreme caution. Because our understanding of the underlying etiology and natural history of peripartum cardiomyopathy is currently incomplete, management of these patients can prove to be difficult. We believe an international registry of cases would greatly aid progress in unraveling the complexities of this disease.
1. Veille JC. Peripartum cardiomyopathies: a review. Am J Obstet Gynecol 1984;148:805–18.
2. Ventura SJ, Peters KD, Martin JA, Maurer JD. Births and deaths: United States, 1996. Monthly vital statistics report; Vol. 46, no. 1, suppl 2. Hyattsville (MD): National Center for Health Statistics; 1997.
3. Guyer B, Strobino DM, Ventura SJ, Singh GK. Annual summary of vital statistics—1994. Pediatrics 1995;96:1029–39.
4. Desai D, Moodley J, Naidoo D. Peripartum cardiomyopathy: experiences at King Edward VIII Hospital, Durban, South Africa and a review of the literature. Trop Doct 1995;25:118–23.
5. Fett JD, Carraway RD, Dowell DL, King ME, Pierre R. Peripartum cardiomyopathy in the Hospital Albert Schweitzer District of Haiti. Am J Obstet Gynecol 2002;186:1005–10.
6. Brown CS, Bertolet BD. Peripartum cardiomyopathy: a comprehensive review. Am J Obstet Gynecol 1998;178:409–14.
7. Heider AL, Kuller JA, Strauss RA, Wells SR. Peripartum cardiomyopathy: a review of the literature. Obstet Gynecol Surv 1999;54:526–31.
8. Demakis JG, Rahimtoola SH, Sutton GC, Meadows WR, Szanto PB, Tobin JR, et al. Natural course of peripartum cardiomyopathy. Circulation 1971;44:1053–61.
9. Koonin LM, MacKay AP, Berg CJ, Atrash HK, Smith JC. Pregnancy-related mortality surveillance—United States, 1987–1990. MMWR CDC Surveill Summ 1997;46:17–36.
10. Hibbard JU, Lindheimer M, Lang RM. A modified definition for peripartum cardiomyopathy and prognosis based on echocardiography. Obstet Gynecol 1999;94:311–16.
11. Demakis JG, Rahimtoola SH. Peripartum cardiomyopathy. Circulation 1971;44:964–8.
12. O'Connell JB, Costanzo-Nordin MR, Subramanian R, Robinson JA, Wallis DE, Scanlon PJ, et al. Peripartum cardiomyopathy: clinical, hemodynamic, histologic and prognostic characteristics. J Am Coll Cardiol 1986;8:52–6.
13. Witlin AG, Mabie WC, Sibai BM. Peripartum cardiomyopathy: a longitudinal echocardiographic study. Am J Obstet Gynecol 1997;177:1129–32.
14. Pearson GD, Veille JC, Rahimtoola S, Hsia J, Oakley C, Hosenpud JD, et al. Peripartum cardiomyopathy: National Heart, Lung, and Blood Institute and Office of Rare Diseases (National Institutes of Health) workshop recommendations and review. JAMA 2000;283:1183–8.
15. Witlin AG, Mabie WC, Sibai BM. Peripartum cardiomyopathy: an ominous diagnosis. Am J Obstet Gynecol 1997;176:182–8.
16. Costanzo-Nordin MR, O'Connell JB. Peripartum cardiomyopathy in the 1980s: etiologic and prognostic consideration and review of the literature. Prog Cardiol 1989;2:225–39.
17. Sutton MS, Cole P, Plappert M, Saltzman D, Goldhaber S. Effects of subsequent pregnancy on left ventricular function in peripartum cardiomyopathy. Am Heart J 1991;121:1776–8.
18. de Souza JL Jr, de Carvalho Frimm C, Nastari L, Mady C. Left ventricular function after a new pregnancy in patients with peripartum cardiomyopathy. J Card Fail 2001;7:30–5.
19. Elkayam U, Tummala PP, Rao K, Akhter MW, Karaalp IS, Wani OR, et al. Maternal and fetal outcomes of subsequent pregnancies in women with peripartum cardiomyopathy. N Engl J Med 2001;344:1567–71.
20. Lampert M, Weinert L, Hibbard J, Korcarz C, Lindheimer M, Lang RM. Contractile reserve in patients with peripartum cardiomyopathy and recovered left ventricular function. Am J Obstet Gynecol 1997;176:189–95.