Outcome of Pregnancy in Women with Mechanical Valves

Abstract

In women who have prosthetic heart valves, pregnancy is risky for mother and fetus. Heparin has been considered safer for the fetus than warfarin, but may not provide adequate anticoagulation for the mother. We examined prospectively gathered data from 100 pregnancies in 67 women with mechanical valves (age range, 19 to 45 years). A subgroup of 20 patients was compared with a control group of relatives and neighbors who conceived at similar ages. Fetal loss occurred in 44 of the 100 pregnancies, due to the following causes: spontaneous abortion (28), intrauterine fetal death (4), stillbirth (3), neonatal death (1), premature birth (2), Rh incompatibility (2), and maternal death (4). Age, parity, atrial fibrillation, and left ventricular enlargement did not affect the outcome. Tricuspid valve disease that required diuretics was associated with a higher rate of fetal loss (17 out of 23 pregnancies, versus 27 out of 77; p = 0.001), but did not affect the mother. Of 66 pregnancies in which the mother was on heparin, 38 (57.6%) resulted in a healthy baby, compared with 18 out of 34 (52.9%) pregnancies in which the mother was on warfarin (p = NS). All thromboembolic complications occurred with heparin therapy (9 cases; p = 0.02). In the control group, fetal loss was 24%, due exclusively to spontaneous abortion.

Women with mechanical valves have higher rates of fetal loss and maternal complications. In our study, tricuspid valve disease adversely affected fetal outcome, which is a new finding that warrants further study. Warfarin was more effective than heparin in preventing thromboembolism in the mothers, and it did not show a significant impact on the babies.

Since the advent of prosthetic cardiac valves, there has been great concern about the outcome of pregnancy in patients who receive them, because mechanical valves require anticoagulation. Orally administered agents achieve anticoagulation in a satisfactory manner, but they cross the placental barrier and have been found to cause congenital malformations in the fetus. ^1–10 Heparin and its fractionated subtypes do not cross the placental barrier; however, they are less effective than oral anticoagulants in preventing thromboembolic events in the mother. ^11–16 In finding a suitable anticoagulant regimen, physicians must concern themselves with the health of both the mother and the fetus.

In Egypt, this problem is of particular importance. An unusually large number of women of childbearing age have mechanical valves, because rheumatic fever and concomitant valvular heart disease are still common and affect the young most severely. Moreover, there is great social pressure to produce children despite the risk of illness and death, because many couples, especially in the lower socioeconomic classes, see childbearing as the sole purpose of marriage. We studied the outcome of pregnancy in Egyptian women who had mechanical valves, and we examined which clinical and therapeutic factors appeared to play a role in maternal and fetal morbidity and mortality.

Patients and Methods

From 1986 to 1999, 348 women were followed up in the prosthetic valve clinic of Cairo University. We fully informed these patients of the risks of pregnancy. Nevertheless, 108 pregnancies occurred in 70 of these patients during the time of this study. We excluded from our prospective analysis 3 patients who were lost to follow-up and 5 pregnancies that were ongoing, which left 67 patients with 100 pregnancies. All patients had mechanical prostheses: 43 patients had valves in the mitral position alone; 12 had valves in the aortic position alone; and 12 had a double-valve replacement (Table I).

Table I. Positions and Types of Valves in the Study Group

We investigated the outcome of pregnancy in terms of fetal and maternal morbidity and mortality. Because no reliable figures exist for spontaneous abortion in Egyptian women, a control group was chosen in the following manner: the patients were asked whether they had any healthy relatives or neighbors (that is, with no mechanical heart valves) who had become pregnant at approximately the same age as the patients. From our study group, we identified a subgroup (Group A) of 20 patients with 33 pregnancies who had relatives or neighbors who had become pregnant at approximately the same age as the patients. The control group (Group B) comprised 20 women with 74 pregnancies. The mean age at pregnancy was 25.32 ± 4.4 years in Group A, versus 23.91 ± 4.65 years in Group B (p = 0.136). We compared the number of pregnancies in both groups, the rate of fetal loss, and the rate of maternal complications. Of necessity, data were gathered retrospectively from the control group.

In the prosthetic valve group, we also examined the effect upon the outcome of pregnancy of such clinical factors as maternal age and the length of elapsed time between the valve operation and the pregnancy (these patients often are unable to carry a fetus to term until their valvular defects have been surgically corrected). Because these women often continue to become pregnant until they have a healthy baby, we examined whether the number of the pregnancy or the fact that the mother was a primipara or multipara played a role in the outcome. We also considered the position of the valve, for we anticipated that a mitral valve prosthesis was more likely to become obstructed than an aortic one, especially in the presence of atrial fibrillation. We also examined left ventricular impairment, but this was a weak point in our investigation because none of the patients was in heart failure prior to conception.

The major points of interest to us were the form of anticoagulation used during pregnancy, its effect on the mother and fetus, and the presence of tricuspid incompetence. In our experience, a substantial portion of rheumatic patients have tricuspid valve involvement, a fact that we had not recognized before. At present, most cardiac surgeons at our institution inspect (and, when possible, repair) the tricuspid valve in each rheumatic patient at the time of mitral or aortic valve replacement. Unfortunately, there are still many patients who have residual tricuspid incompetence severe enough to require the use of diuretics, a therapy much disliked and criticized by obstetricians.

Results

Fetal Outcome

All pregnancies were divided into 2 outcome groups: healthy baby and fetal loss. Of the 100 pregnancies in the study group, 56 resulted in a healthy baby. Fetal loss occurred in 44 pregnancies, due to the following causes: spontaneous abortion in the 1st trimester (28), intrauterine fetal death (4), stillbirth (3, including twins who died of warfarin embryopathy), neonatal death due to meconium aspiration (1), premature birth (2), Rh incompatibility (2), and preterm death of the mother (4).

The following factors were examined in order to determine their effect on fetal outcome:

Anticoagulation during the 1st trimester. In 66 of the 100 pregnancies in the study group, the patients were put on a strict regimen of subcutaneous heparin, 5,000 IU every 8 hours or 10,000 IU every 12 hours. Because most of these patients were from remote rural areas, it was not possible to obtain proper follow-up data on activated partial thromboplastin time. In the other 34 pregnancies, the mother continued to take an oral anticoagulant (warfarin), in order to adjust the prothrombin time to 2.5 of the control, and later, when the international normalized ratio (INR) became available, to adjust the INR to 2-2.5. Of the 66 pregnancies in which the mother was put on heparin, 38 (57.6%) resulted in a healthy baby. Of the 34 pregnancies in which the mother took warfarin, 18 (52.9%) produced healthy babies. The difference was statistically insignificant (Table II).

Table II. Clinical Factors in Association with Fetal Outcome

Position of the valve. There was no statistically significant difference in the outcome of pregnancy in patients with mitral, aortic, or double-valve replacement (Table II).

Parity. The presence of a previous pregnancy did not play a major role in the outcome: the number of healthy babies in primiparae was almost identical to that in multiparae (Table II). We postulated that the order of pregnancy after surgery may play a role in the outcome, as the mothers become more experienced in anticoagulant intake and control. Forty-three women became pregnant only once, 15 became pregnant twice, and 9 ventured a 3rd pregnancy. Although the fetal loss rate declined slightly in the 2nd pregnancy, this did not amount to any statistical significance (Table II).

Age of the mother. We examined maternal age in several ways. The mean age of the mother at the time of pregnancy was 26.05 years (± 5.66) in the healthy-baby group, versus 26.55 years (± 6.43) in the fetal-loss group (p = NS). Age was not, then, a determining factor in the outcome of pregnancy. The time that had elapsed between heart surgery and pregnancy was also examined, because we postulated that a short interval may be significant. The fetal-loss group had a slightly shorter surgery-to-pregnancy time; however, this trend did not amount to a statistically significant difference (4.7 ± 3.5 years, versus 4.2 ± 2.7 years).

Atrial fibrillation. In the outcome of pregnancy, there was no statistically significant difference between patients who were in sinus rhythm and those who were in atrial fibrillation (Table II).

Left ventricular dilatation and failure. None of our patients suffered from left ventricular failure before pregnancy. Left ventricular failure occurred in 1 patient due to pre-eclampsia, and her baby died. In 43 pregnancies, the mothers had dilated hearts, but ventricular contractility was good, and this condition did not adversely affect the outcome of pregnancy (Table II).

Tricuspid incompetence. We found that tricuspid incompetence was a major problem in the 348 women with prosthetic valves who were seen regularly at the follow-up unit of Cairo University during the study period: 38 (11%) had residual tricuspid incompetence that required diuretics. This occurred despite routine surgical inspection and repair by De Vega annuloplasty, when necessary, of the tricuspid valve at the time of mitral or aortic valve replacement. We have found no other reports of the effect of this residual valvular lesion on the outcome of pregnancy. In 23 of the 100 pregnancies that we report here, the mother suffered from tricuspid incompetence severe enough to require diuretic use. These pregnancies did not fare well: 17 (73.9%) ended in spontaneous abortion or intrauterine fetal death. When fetal loss in these pregnancies was compared with fetal loss in pregnancies in which the mother did not exhibit tricuspid incompetence, there was a statistically significant difference (p = 0.001).

Maternal Outcome

Of the 100 pregnancies we studied, 15 developed maternal complications, and 5 ended fatally for the mother. These complications included valve malfunction (thrombosis in 8 pregnancies) and infective endocarditis in 1 pregnancy, postpartum hemorrhage (5 pregnancies), bleeding during pregnancy due to placenta previa (1 pregnancy), and toxemia of pregnancy (1 pregnancy). Caesarean section was required in 2 pregnancies.

The most common and most serious complication was acute valvular malfunction. Except for the 1 patient who developed infective endocarditis, all of the 9 patients who experienced acute valvular obstruction had received heparin in the 1st trimester. Five of these patients died and 4 underwent successful reoperation. We examined the same clinical parameters that we had examined in assessing fetal outcome, but apart from the type of anticoagulant used, none affected maternal outcome significantly (Table III).

Table III. Maternal Complications

Comparison with the Control Group

In Group A (the subgroup of our study group), 13 out of 33 pregnancies (39.4%) resulted in fetal loss. Of these, 7 (21%) ended in spontaneous abortion in the 1st trimester. Other causes included intrauterine fetal death (2 cases), Rh incompatibility (3 cases), meconium aspiration (1 case), and preterm maternal death (1 case). In Group B (the control group), fetal loss was 24% and was due exclusively to spontaneous abortion in the 1st trimester.

In Group A, the number of pregnancies ranged from 1 to 3, with an average of 1.5 pregnancies per patient. In Group B, the number of pregnancies ranged from 1 to 12, with an average of 3 pregnancies per patient. There were no maternal complications in Group B; in Group A, 1 patient developed valvular malfunction and died.

Discussion

Because the best means of handling pregnancy in patients who have mechanical valves remains a controversial issue, we investigated both fetal and maternal outcomes in 100 pregnancies in Egyptian women who had these prostheses in place.

The fetal loss rate of 46% was high. Of these losses, 28% were due to spontaneous abortion in the 1st trimester. Other causes included intrauterine fetal death, meconium aspiration, Rh incompatibility, and maternal death during pregnancy. Warfarin embryopathy was noted in only 1 case—that of the twins who were stillborn. The fetal loss rate reported in the literature for women with prosthetic heart valves varies from 8.5% to 66%. 2,4,9,12–20

Because no reliable figures exist for fetal loss in Egypt, we decided to choose an age-matched control group consisting of healthy women from the families and neighborhoods of our patients. This was possible for 20 of the 67 patients we had studied. The women in the control group had twice as many pregnancies as did their counterparts in the study group. Fetal loss was 24% and was due exclusively to spontaneous abortion. The spontaneous abortion rate was similar in our study group (21% in the subgroup, 28% overall). This suggests that patients with mechanical valves are at higher risk of losing their babies from causes other than abortion in the 1st trimester. To date, there has been no other study that included a control group of healthy pregnant women who had no mechanical heart valves.

We found only 1 study that compared the outcomes of pregnancy in patients with mechanical valves who were on warfarin to those who were on heparin and to patients with bioprostheses who were not receiving any type of anticoagulation. ⁹ In this study, the fetal loss rate was 53% on warfarin, 36% on heparin, and 16.7% without anticoagulant use. However, these statistics are merely descriptive. The patients were not age-matched, which might explain the difference from our results.

We examined maternal characteristics to determine the possible causes of fetal loss. Neither the age of the mother at the time of pregnancy nor the interval between the operation and pregnancy had any effect on the outcome. There was no significant difference between patients in sinus rhythm and those with atrial fibrillation, nor did the position of the valve have any effect. The presence of tricuspid incompetence of a degree necessitating diuretic use, however, was associated with a higher degree of fetal loss. This has not been reported previously in the literature and deserves further investigation. The use of diuretics, low cardiac output, or systemic congestion may explain this correlation. Given that 11% (38 of 348) of all women with mechanical valves who were followed up at Cairo University and 21% (14 of 68) of those who became pregnant had residual tricuspid incompetence warranting diuretic use, this phenomenon should be investigated.

The use of warfarin in the 1st trimester is still controversial. Warfarin is reported to be teratogenic, especially when taken during the 6th to 12th week of gestation. ³ However, recent reports have indicated that the much-feared warfarin embryopathy is extremely rare ^12,14 and that fetal loss may be dose dependent. ^15,21 Furthermore, there have been reports that even full heparinization does not provide reliable and sufficient anticoagulation to the mother, which puts her at risk for thromboembolism and valvular malfunction. ^11–16 When we compared the women in our series who received warfarin during the 1st trimester with those who received heparin, we found no significant difference in fetal outcome. However, mechanical valve obstruction occurred in 60% of the mothers who had complications, and thromboembolism occurred in 53% of mothers with complications. All of the cases of thromboembolism occurred in patients who were put on heparin in the 1st trimester. This appears to confirm the findings of other investigators that heparin does not provide adequate protection against thromboembolism in these patients. Except for postpartum hemorrhage in 1 patient, which could be explained by the oral-anticoagulant therapy, the other maternal complications were all pregnancy related. These included malpositions that required Caesarean section, placenta previa, toxemia of pregnancy, and postpartum hemorrhage due to retained placenta. The maternal complication rate was 14.8%, which compares favorably to international figures that range from 10% to 15%. ^12,17,18

Conclusion

We conclude that patients with mechanical valves have a higher incidence of fetal loss and maternal complications than do healthy women of their age with comparable genetic and environmental backgrounds. Warfarin was more effective than heparin in preventing thromboembolic complications and valvular malfunction in the mother, and it did not seem to have a significantly worse effect on the fetal outcome. Tricuspid valve disease was associated with an extremely high rate of fetal loss, and this association warrants further investigation.