CARDIOMYOPATHY

longitudinal Changes in left Ventricular Architecture and Function in Peripartum Cardiomyopathy

PATRICIA COLE, MD, FRANCIS COOK, ScD, TED PLAPPERT, CVT, DANIEL SALTZMAN, MD, and MARTIN St. JOHN SUTTON, MRCP

Left ventricular (LV) function was quantitated in 14 patients with peripartum cardiomyopathy, in 10 us- ing 2-dimensional (2-D) echocardiography and in 4 radlonuclide ventriculography, and values were compared with those in 11 normal women in the immediate postpartum period. LV end-diastolic and end-systolic volume indexes, LV wall mass index and ejection fraction were calculated during the acute phase of the illness and serially through long- term follow-up (mean 24 months). During the acute phase of illness there was marked LV dilatation: mean end-diastolic volume index was 95 f 22 ml/m*, vs 67 f 9 ml/m* in control subjects (p <O.OOS), and mean end-systolic volume index was 66 f 16 ml/m*, compared to 27 f 5 ml/m* in control subjects (p <O.OOl). Mean heart rates and mean systolic pressures in the patients with peripar- turn cardiomyopathy and the control subjects were

similar, 91 f 24 vs 79 f 14 beats/min and 120 f 14 vs 117 f 10 mm Hg, respectively. LV wall mass index was higher, 139 f 36 vs 96 f 6 g/m* (p <0.005), and ejection fraction much lower, 29 f 5% vs 67 f 5% (p <O.OOl), in control sub- jects. Five patients underwent endomyocardial biop- sy during the acute illness, which showed interstitial fibrosis but no evidence of active inflammation. There was rapid and early improvement in LV func- tion in 13 of 14 patients. Changes in LV volume, mass and ejection fraction all followed an exponen- tial time course during LV remodeling. By late fol- low-up, LV size and function had returned to normal in more than half the patients. The severity of early compromise in LV function did not predict long-term functional outcome.

(Am J Cardiol 1967;60:671-676)

P eripartum cardiomyopathy is a rare and poorly characterized syndrome of heart failure that develops between 3 months before and 5 months after parturi- tion in women without preexisting heart disease.1 Epi- demiologic studies estimate that peripartum cardiomy- opathy has an incidence between 1 in 4,000 and 1 in 15,000 gestations.2r3 This disease is reported to predom- inate in older, black, multiparous women and in pa- tients with pre-eclampsia or with twin pregnancies.@ There is no consensus regarding the pathogenesis of the left ventricular (LV) dysfunction in peripartum

From the Cardiovascular Division, Department of Medicine, and the Department of Obstetrics & Gynecology, Brigham & Women’s Hospital, Boston, Massachusetts. Manuscript received March 31,1%7; revised manuscript received and accepted June 8, 1987.

Address for reprints: Martin St. John Sutton, MRCP, Cardio- vascular Division, Brigham and Women’s Hospital, 75 Francis Street, Boston, Massachusetts 02115.

cardiomyopathy, although it has been suggested that viral myocarditis may play a role in some patients”87

This clinical entity was first described in 1849,8 and many subsequent reports describe the symptoms, physical findings and clinical outcome in women with peripartum cardiomyopathy2~gJ0; however, few studies have assessed the severity of LV dysfunction7 Mea- surements of LV function are mostly confined to the acute phase of the illness and have produced conflict- ing results. Ejection fraction has been variously report- ed to be elevated, normal and reduced.7J1-15 These inconsistencies may have arisen in part because LV function in peripartum cardiomyopathy has been com- pared with that of normal nonpregnant women instead of an appropriate normal postpartum population and partly because of the imprecise criteria by which this syndrome is diagnosed. No systematic attempt has been made to evaluate serially changes in LV function through long-term follow-up.

We performed a longitudinal study on 14 patients with peripartum cardiomyopathy using %dimensional

a72 PERIPARTUM CARDIOMYOPATHY

(Z-D] echocardiography or radionuclide ventriculog- raphy, and compared them with women in the imme- diate postpartum period. Our goals were to assess LV chamber architecture and quantitate pump function during the acute phase of the illness and to quantitate changes in chamber architecture and function with time by serial studies through long-term follow-up in an attempt to characterize the natural history of this disease.

Methods Patients: Peripartum cardiomyopathy: Our popu-

lation consisted of 14 patients with peripartum cardio- myopathy, aged 17 to 46 years (mean 301, all of whom had been in normal health before pregnancy and had no history or symptoms of systemic hypertension or heart disease. Thirteen were white, 1 was black, all were from the middle socioeconomic class, 9 were multigravida and none had pre-eclampsia or eclamp- sia. Peripartum cardiomyopathy was diagnosed using the conventionally accepted criteria recommended by Demakis et al,l which included: (1) development of heart failure in the last 3 months of pregnancy or in the first 5 postpartum months; (2) absence of a determina- ble cause for cardiac failure; and (3) absence of de- monstrable heart disease before the last trimester of pregnancy.

All patients presented to medical attention in New York Heart Association symptom class III or IV from 61 days before parturition to 106 days after parturition (mean 3 days after parturition]. Patients were followed for a mean of 24 months. Four patients had spontane- ous vaginal deliveries and 10 had cesarean sections. Indications for cesarean section included previous ce- sarean section in 1 patient, severe maternal distress from congestive heart failure in 3 and cephalopelvic disproportion in 6. Thirteen patients had singleton pregnancies and 1 patient had twins. Two patients underwent 2-D echocardiography in the second tri- mester for suspected mitral valve prolapse, which showed a normal mitral valve and normal LV cavity size and function in both patients. There were 3 peri- natal deaths: one due to prematurity (27 weeks), one due to nonimmune hydrops fetalis and one unex- plained intrauterine fetal death. In 2 deaths, maternal heart failure developed early in the third trimester.

Electrocardiograms at presentation showed that all patients were in sinus rhythm, but 1 patient had inter- mittent atria1 bigeminy. Nonspecific ST-segment and T-wave changes were present in 89% and incomplete right bundle branch block occurred in 1 patient. Spe- cifically, there .were no infarction patterns and no oth- er conduction abnormalities.

All patients had cardiomegaly with a cardiotho- racic ratio greater than 0.55 at pr,esentation. Radio- graphic signs of LV failure including pulmonary ve- nous redistribution or frank pulmonary edema were present in all patients.

Five patients had endomyocardial biopsies in the acute stage of the illness after initial presentation. All showed mild nonspecific interstitial fibrosis and in- creased myocardial fiber size but no round cell infil- tration to suggest an inflammatory etiology.

Treatment regimens in 12 patients during the acute illness included diuretics and digoxin; 2 patients also received short-term steriods and azathioprine imme- diately after endomyocardial biopsy because the cause was initially believed to be myocarditis. In both pa- tients these drugs were discontinued when endo- myocardial biopsy results showed no evidence of inflammation.

Normal control subjects: The normal population consisted of 11 women, aged 24 to 38 years (mean 32), who were studied 1 to 3 days (mean 2) postpartum. Ten were white, 5 were multigravida and all were from the middle socioeconomic class. All had uncomplicated pregnancies and normal live births at term. None had symptoms or signs of previous heart disease, diabetes or hypertension and none was taking any medications. All had normal cardiovascular findings and normal 2- D echocardiograms. Ten had spontaneous vaginal de- liveries and 1 woman had a cesarean section. The indication for cesarean section was fetal distress.

Data acquisition: Two-dimensional echocardio- grams were recorded in the 11 normal control patients in the immediate postpartum period and in 10 of the 14 patients with peripartum cardiomyopathy at initial presentation, through the acute illness and serially over longitudinal follow-up, which ranged from 1 to 5 years (mean 2.21, yielding a mean of 3.6 studies per patient. Two-dimensional echocardiographic images were recorded with simultaneous electrocardiograms of the LV short axis at the level of the tips of the papillary muscles from the left parasternal region and the 4-chamber view from the apex.

Radionuclide ventriculograms were recorded in the 4 remaining peripartum cardiomyopathy patients at presentation, during the acute illness and serially during longitudinal follow-up, yielding a mean of 4.5 studies per patient.

Heart rates, systolic blood pressures, electrocardio- grams and chest x-rays were obtained at presentation and repeated during the acute illness and throughout the long term follow-up.

Data analysis: Short-axis left ventricular echocar- diograms: Endocardial and epicardial surfaces of the short-axis images of the left ventricle from 5 high- quality stop action end-diastolic and end-systolic frames were digitized with their calibration factors using a Diasonics Cardio Revue Center. From these digitized data the following were computed: total cross-sectional area [At] enclosed by the LV epicardi- um and the right side of the septum, cavity area (A,), myocardial area (Am), determined by subtracting A, from At, i.e., (At - A,).

Apical d-chamber echocardiogram: LV lengths (L) were digitized from 5 end-diastolic and 5 end-systolic stop action frames from the apical 4-chamber view of the left ventricle using the Diasonics Cardio Revue Center. Cavity length was defined as the distance be- tween the apical endocardium and the midpoint of the plane of the mitral valve anulus.

These data were used to calculate LV short- and long-axis ratio at end-diastole; end-systolic volume and end-diastolic volume using the short-axis area- length method (V = 5/6 A,L,); ejection fraction; and

October

LVmuscle mass r1.055 X 5/6 (AtLt - A&,], where 1.055 = density of myocardial muscle. LV volume and mass calculations were normalized to unit body surface area to calculate end-systolic and end-diastolic vol- ume indexes and LV wall mass index.

The reproducibility and interobserver variabil- ity from our laboratory of 2-D echocardiographic determinations of LV volume and mass have been reported.l@J7

Radionuclide ventriculograms: After ‘injection of technetium-99m-labeled red blood cells, cardiac up- take was measured in 46 stop frames per cycle, for a total of 300 cardiac cycles and an ejection fraction was generated by computer.

Statistical analysis: Heart rates, systolic blood pres- sures and echocardiographic indexes of LV function during the acute presentation were compared in 10 peripartum cardiomyopathy patients and 11 normal control subjects using the Student unpaired t test.18 In addition, the same test was used to compare ejection fractions in patients with peripartum cardiomyopathy during the acute presentation (pooled echocardio- grams and radionuclide ventriculograms) with those in the normal control subjects. For the 7 patients who had a follow-up echocardiogram within 4 months, a Stu- dent paired t test was used to compare acute with early follow-up results.

Because subsequent serial studies were performed at varying intervals from the time of presentation in each patient, changes in the measurements of LV func- tion with time were examined using a linear regression model that expressed each measurement as a function of the natural logarithm of time from presentation, For individual patients, a model for each of the indexes of LV function was developed that contained baseline values to express the behavior of that patient over time. To describe the “average” behavior of all pa- tients, a summary curve (of the form index = B. f B1 [In time]) was obtained by averaging each patient’s baseline values by the number of studies performed on the individual patient. In this manner, a summary curve for each index was obtained that took into ac- count both the number of studies in each patient and the time of each study from presentation. One patient who died of progressive heart failure was-eliminated from the analysis as an outlier.

Results Acute studies: Mean heart rates and systolic blood

pressures at initial presentation in the patients with peripartum cardiomyopathy were not significantly dif- ferent from those of the normal subjects, 91 f 24 vs 79 f 14 beats/min and 120 f 14 vs 117 f 10 mm Hg, respectively. Two-dimensional echocardiography showed that the left ventricles in patients with peripar- turn cardiomyopathy were significantly dilated, with a mean end-diastolic volume index of 95 f 22 ml/m2 and a mean end-systolic volume index of 66 f 18 ml/m2, compared with 67 f 9 and 27 f 5 ml/m2, re- spectively, in the postpartum control subjects (p <0.005 and p <O.OOl, respectively] (Fig, 1). Despite the LV dilatation in the peripartum cardiomyopathy patients, LV shape expressed as short-/long-axis ratio was simi-

1987 THE AMERICAN JOURMAL OF CARDlOLOGY Voiume 60 78

lar to that in normal women, 0.58 f 0.09 vs 0.53 f 0.04. This was important because major differences in cavi- ty geometry between the populations preclude use of the 51'6 ellipsoidal formulations for LV volumes and mass estimations. LV wall mass index was also signifi- cantly increased in patients with peripartum cardio- myopathy, 139 f 38 vs 96 f 8 g//m2 in the control subjects (p <O.OOS) (Fig. 21. However, the mass to end- diastolic volume ratio in peripartum cardiomyopathy and control subjects was similar, 1.45 f 0.16 vs 1.43 f 0.12, indicating that despite cavity dilatation, LV mus- cle mass and volume increased in such a way as to maintain normal LV cavity architecture. LV function was severely depressed in patients with peripartum cardiomyopathy, with a mean ejection fraction of 29 f 5%, compared with 60 f 5% (p <O.OOl] in the control subjects, with no overlap between the 2 populations (Fig. 2). In addition, there were no segmental LV wall motion abnormalities by 2-D echocardiography in any of the peripartum cardiomyopathy patients or in the control subjects.

In the 4 patients with peripartum cardiomyopathy who underwent radionuclide ventriculography, initial ejection fraction was similarly decreased, with a mean of 25 f 1.0 with global reduction in function and no segmental wall motion abnormalities. The ejection fraction during the acute presentation for the peripar- turn cardiomyopathy patients as a group (echocardiog- raphy and radionuclide ventriculography, n = 14) was significantly lower than that of the control population, 28 f 6% vs 67 f 5% (p <O.OOl).

Two-dimensional echocardiography was repeated in 7 patients during the acute phase of the illness at a

160

20

0

End-dlastole .

, p<o.005 ,

NLS PPCM

t .

6

p<O.O01 ,

M-S PPCM

FIGURE 1. End-diastolic and end-systolic volume index for patients with peripartum cardiomyopathy (PPCM) during the acute presen- tation compared with control subjects (NLS).

074 PERIPARTUM CARDIOMYOPATHY

mean of 1 month after initial presentation. In 5 of these 7 patients ejection fraction improved by at least 25%.

By 4 months there was a significant reduction in LV chamber size; mean values for end-diastolic volume index and end-systolic volume index had decreased from 90 f 9 to 63 f 11 ml/m2 (p <0.007) and from 62 f 8 to 33 f 12 ml/m2 (p <0.003), respectively. LV wall mass index decreased from 130 f 15 to 100 f 13 g/m2 (p <O.O3]. These changes in LV volume and mass were accompanied by an increase in ejection fraction in all but 1 patient, from a mean of 31 f 6% to 49 f 11% (p <O.Ol]. However, during this LV remodeling the LV mass/end-diastolic volume ratio did not change.

Long-term follow-up: One patient showed no im- provement in LV function either during the acute phase or at long-term follow-up; she was so markedly different from the remaining population that she was regarded as an outlier and excluded from subsequent analysis. Another patient died from a pulmonary em- bolism 3% weeks after presentation, but by this time LV volume and mass had decreased and ejection frac- tion had increased from 15% to 53%.

Sequential changes in LV end-diastolic and end- systolic volume indexes, LV mass index and ejection fraction during longitudinal follow-up were best de- scribed by the simple exponential function index = Bo+B1 (In time) (Fig. 3 and 41. Linear regression analy sis revealed that there was major LV remodeling with substantial reduction in end-diastolic (r = -0.72) and end-systolic volumes (r = -0.79) and in LV muscle

300-

280 -

“$ 240 -

2 E

2 200 - .r % 2 160 -

: 3

z .c 120 - L

80 -

70- l

60- I

+

. 3

:

”

. 0

:

.

.

1 ’ p<o.o05 , p<O.OOl ,

0 NLS PPCM 0 NLS PPCM FIGURE 2. Left ventricular wall mass index and ejection fraction for patients with perlpartum cardiomyopathy (PPCM) during the acute presentation compared with controls (NLS).

mass (r = -0.82) early after presentation, which continued more slowly through late follow-up (Fig. 3 and 4).

These changes in LV mass and volume occurred in such a way that the ratio of mass to end-diastolic vol- ume remained similar to that of controls throughout. This LV remodeling was accompanied by significant improvement in contractile function, shown by the ini- tial rapid but continued increase in ejection fraction at late follow-up (r = 0.79) (Fig. 41. Ejection fraction in- creased in all but one of the long-term survivors, but in only 57% did ejection fraction return to more than 50%. In the patients who undewrwent only radionu- elide ventriculography, ejection fraction showed the same exponential increase with time, but in only 2 of the 4 patients did values return to normal. The changes in these LV function indexes over time for each patient were similar to the behavior of the group as a whole.

Discussion All of our patients presented in New York Heart

Association symptom class III or IV with clinical signs of severe LV failure with cardiomegaly and pulmo- nary venous congestion on chest radiographs. Two- dimensional echocardiograms showed LV dilatation with increased end-diastolic and end-systolic volumes

t:/* ,-• , , ,

100 200 300 400 Time From Presentation In Days

FIGURE 3. End-diastolic and end-systolic volume Index as a func- tion of time in days for patlents with perlpartum cardiomyopathy at long-term follow-up.

October i, 1967 THE AMERICAN JOURNAL OF CARDIOLOGY Volume 60

and increased LV mass compared with normal women in the immediate postpartum period. LV function was severely depressed, with ejection fractions varying from 17 to 39%, which completely separated the peri- partum cardiomyopathy patients from control sub- jects. Initial radionuclide ventriculogram ejection fractions were similarly depressed. In the few isolated case reports in which echocardiograms were recorded in peripartum cardiomyopathy patients, the left ventri- cle was mildly dilated and contractile function im- paired.ll-l3

Most of our patients showed early rapid improve- ment in clinical symptoms, resolution of electrocardio- graphic abnormalities, reduction in heart size on chest x-ray and improvement in echocardiographic mea- sures of LV function by 1 month from presentation. Many studies suggest that improvement in LV func- tion, when it occurs, does so early, although isolated instances of late recovery have been reported.lge21

We were initially concerned that changes in end- diastolic and end-systolic LV volumes detected by 2-D echocardiography in the peripartum cardiomyopathy patients during the acute phase of the disease may simply reflect changes in preload or intravascular vol- ume status resulting from the acute administration of diuretics. However, heart rates and systolic blood pressures during this acute phase were not significant- ly different from that in normal control subjects, and furthermore the changes in LV mass, which are insen- sitive to such transitory perturbations in preload and volume status, followed the same exponential time course as the end-systolic and end-diastolic volumes. The similar pattern of regression of LV mass and vol- umes with time indicated that the acute changes in LV loading conditions during therapy were not major de- terminants of LV remodeling. The increased LV mass in peripartum cardiomyopathy is unexplained be- cause none of our patients had hypertension or preeclampsia. However, ventricular hypertrophy on histologic examination of endomyocardial biopsy specimens, present in our patients, has also been de- scribed in 2 recent large series of patients.7r22

The time course of the changes in volume and mass and improvement in ejection fraction was best de- scribed by an exponential function, in that LV remod- eling slowly continued through late follow-up after its initial rapid early onset. There was no relation be- tween LV volume, mass or ejection fraction at initial presentation and LV function at late follow-up. This was exemplified by several patients with severe LV dysfunction and initial ejection fractions of less than 3O%, in whom LV volumes and ejection fraction re- turned to normal at late follow-up, and by the 1 patient who died from heart failure due to progressive myopa- thy whose initial ejection fraction was in the middle of the range for the population. The time trend analysis of the changes in LV architecture and function showed improvement in most patients with progressive LV re- modeling, and in 57% of the patients LV function re- turned to normal.

Our patients with peripartum cardiomyopathy dif- fered from those in previous reports in which patients showed no symptomatic or functional improvement or

even deterioration in clinical studies, with a poor long- term prognosis and high mortality.1~7,21~23 The extent of LV remodeling and improvement in contractile func- tion did not appear to be directly related to treatment, which in 12 of the 14 consisted only of diuretics and digoxin, while 2 patients received a single short course of steroids and azathioprine.

Although there are case reports of peripartum car- diomyopathy after cesarean section,24 this high inci- dence of cesarean section has only once been previ- ously described in peripartum cardiomyopathy.7 It is unlikely that cesarean section was causally related to development of peripartum cardiomyopathy, because the frequency of cesarean section is 23% in our institu- tion, which exceeds the incidence of peripartum car- diomyopathy by 2 orders of magnitude. One patient in our control group had delivery by cesarean section, and LV function in the postpartum period was normal.

petal prognosis is believed to relate to the timing of presentation of the maternal heart failure,25 and is re- ported to be good when maternal heart failure devel- ops postpartum but poor when heart failure develops during the third trimester of pregnancy. There were 3

. 60 . .

.

Summary Curve: x=27.2 + 4.5 (htime) r=0.79

0’ , 1 I J 100 200 300 400

Time From Presentation In Days

FIGURE 4. Left ventricular wall mass index and ejection fraction as a function of time in days for patients with peripartum cardiomyopa- thy at long term follow-up.

876 PERIPARTUM CARDIOMYOPATHY

fetal deaths in our series; in 2 of these 3 maternal heart failure developed postpartum. Of the remaining 11

References

patients, in whom all fetuses survived, maternal heart 1. Demakis JG, Rahimtoola SH, Sutton GC, Meadows WR, Szanto PB, Tobin

failure presented in the postpartum period in 8 pa- JR, Gunnar RM. Natural course of peripartum cardiomyopathy. Circulation 1971;44:1o53-1061.

tients and during pregnancy in only 3. 2. Meadows WR. Idiopathic myocordial failure in the last trimester of preg-

Little is known regarding the status of LV function nancy and the puerperium. Circulation 1957;15:993-914.

in patients with peripartum cardiomyopathy during 3. Cunningham FG, Pritchard JA, Hankins GD, Anderson PL, Lucas MJ, Armstrong KF. Peripartum heart failure: idiopathic cardiomyopathy or com-

subsequent pregnancies despite reports of clinical ex- pounding cardiovascular events? Obstet GynecoJ 1966;67:157-166.

acerbation of symptoms. 3~2,5 This has clinical impor- 4. Susser MW, Stein EA. Postpartum cardiomyopathy. J Obstet Gynaecol Br Emp 1956;65:769.

tance, because sterilization has been recommended 5. Seftel H, Susser M. Maternity and myocardial failure in African women.

for women with peripartum cardiomyopathy, based on i.! zrt ’ 1g61’2F43-52’ individual case reports of maternal death or exacerba-

e vm KR, Rmhardson PJ, Olsen EG, Daly K, Jackson G. Peripartum cardiomyopathy due to myocarditis. N EngJ J Med 1962;307:731-734.

tion of congestive heart failure. Two of our patients 7. O’Connell JB, Constanzo-Nordin MR, Subrumanian R, Robinson JA, Wallis

had subsequent pregnancies that were uncomplicated, DE, Scanlon PJ, Gumnar RM. Peripartum cardiomyopathy: clinical, hemody- namic, histologic and prognostic characteristics, JACC 1986;6:52-56.

and resulted in 2 normal live births at term with no 8. Ritchie C. Clinical contributions to the pathology, diagnosis and treatment

deterioration in maternal LV function. In both of these of certain chronic diseases of the heart. Edinburgh Med Surg J 1649;3:333.

patients LV function had returned to normal before 9. Knobel B, Melamud E, Kishon Y. Peripartum cardiomyopathy. Isr J Med Sci 1984;2o:1o61-1os3.

their subsequent pregnancy. The patients reported to 10. Meadows WR. Postpartum heart disease. Am J’CardioJ 1960:6:786-602.

have a poor prognosis during subsequent pregnancy 11. Weitz C, Spence MR. Peripartal cardiomyopathy. Obstet GynecoJ 1983;62:555-575, may have been those in whom LV function did not 12. Cepin D, James F, Carabello B. Left ventricular function in peripartum

return to normal1 cardiomyopathy. Chest 1963;63:701-704.

The pathogenesis of peripartum cardiomyopathy 18. Hodgman MT, Pessin MS, Homans DC, Panis W, Prager RJ, Lathi ES, c .

remains unknown, although recent studies of patients rrscitello MG. Cerebral embolism as the initial manifestation of peripartum

cardiomyopathy. Neurology 1962;32:668-671.

with peripartum cardiomyopathy in whom endomyo- 14. Sanderson JE, Adesanya CO, Anjorin FI, Parry EH. Postpartum cardiac

cardial biopsies were performed have suggested a role failure - heart failure due to volume overload? Am Heart J 1979;97:613-621. 15. Johnson JB, Mir GH, Flores P, Mann M. Idiopathic heart disease associ-

for myocarditis .6JoJ6 Five patients in our series ated with pregnancy and the puerperium. Am Heart J 1966;72:609-616.

had right ventricular biopsies during the symptomatic 18. Helak JW, Reichek N. Quantitation of human left ventricular mass and

phase of their illness, but microscopy failed to show volume two-dimensional echocardiography: in vitro anatomic validation. Circulation 1961;63:1396.

active myocarditis. Other mechanisms that have been 17. Reichek N, Helak J, Plappert T, St. John Sutton M, Weber KT. Anatomic

proposed in peripartum cardiomyopathy include validation of left ventricular mass estimates from clinical two-dimensional echocardiography: initial results, Circulation 1963:67:346.

exacerbation of underlying LV dysfunction, ex- 18. Colton T. Statistics in Medicine. Boston: Little, Brown, 1974:545-47.

cessive salt intake in Nigerians and autoimmune 18. Brockington IF. Postpartum hypertensive heart failure. Am J Cardiol

abnormalities.27 1971;27:650-656. 28. Stuart KL. Cardiomyopathy of pregnancy and the puerperium. Q J Med

We conclude that LV dilatation and severe LV dys- 1g68;37:463-478. function are present at initial presentation; significant 21. Demakis JG, Rahimtoola SH. Peripartum cardiomyopathy. CircuJation 1971;44:g64-g6& early improvement in LV function occurs in most pa- 22. Sanderson JE, Olsen EGJ, Batei D. Peripartum heart disease: an endomyo-

tients; changes in chamber volume, mass and ejection cardial biopsy study. Br Heart J 1966;56:265-291.

fraction during LV remodeling follow an CXpOIIeDkl 23. Walsh JJ, Burch GE, Black WC, Ferrans VJ, Hibbs RG. Idiopathic myocar- diopathy of the puerperium. Circulation 1965;32:19-31.

time course, and at late follow-up, chamber size, archi- 24. Malinow AM, Butterworth JF, Johnson MD, et al. Peripartum cardiomy-

tecture and function have returned to normal in more opathy presenting at caesarean delivery. Anesthesiology 1965;63:545-547.

than half the patients; neither LV size nor function 25. Homans DC. Peripartum cardiomyopathy. N EngJ J Med 1965;312:1432- 1437.

during the acute phase predict long-term functional 28. Huerta EM, Erice A, Espiro RF, Navascues I, Martin de Dios R. Postpar-

outcome; and return to normal LV function may not turn cardiomyopathy and acute myocarditis. Am Heart J 1965;110:1079-1061. 27.~

preclude further pregnancy. ossman CE. Peripartum cardiomyopathy. J Tenn Med Assoc 1964;77:29-

31,33.