Prevention of Recurrent Preterm Delivery by 17 Alpha-Hydroxyprogesterone Caproate

Medical records of women presenting for prenatal care at the 19 participating centers were screened for eligibility to participate in the trial; criteria for eligibility included a history of spontaneous preterm delivery in a previous pregnancy and a current pregnancy between 15 weeks and 20 weeks 3 days of gestation. Reasons for exclusion were multifetal gestation, known fetal anomaly, progesterone or heparin treatment during the current pregnancy, current or planned cervical cerclage, hypertension requiring medication, a seizure disorder, or a plan to deliver elsewhere. An ultrasonographic examination was required between 14 weeks and 20 weeks 6 days of gestation to confirm the duration of gestation and to identify any major fetal anomalies. The duration of gestation at the time of randomization was determined according to a previously described algorithm¹² on the basis of the last menstrual period and the results of ultrasonography.

Candidates for the trial were approached by a research nurse, who explained the study and asked prospective participants to sign a form for the release of medical records to permit the research nurse to obtain a copy of the chart from the previous pregnancy ending in preterm delivery. If the previous preterm delivery was of a liveborn singleton infant between 20 weeks of gestation and 36 weeks 6 days of gestation and was due to spontaneous preterm labor or preterm premature rupture of the fetal membranes, and if no criteria for exclusion were present, the woman was deemed to be eligible for the study. Each eligible woman was then invited to participate and to sign a consent form approved by the local institutional review board.

The trial started in April 1998 but was stopped in February 1999 because the Food and Drug Administration had ordered the pharmaceutical company that supplied the active study drug to shut down and had mandated a total recall of all the company's drugs, including the study drug, because of poor quality control and documentation. Patient safety was not considered to have been compromised, but the potency of the product that had been supplied was thought to be questionable. At the time the study was stopped, 150 women had been enrolled, but none of the data had been analyzed. The trial was started anew with the study drug and placebo supplied by a company that manages investigational drugs (Eminent Services), and the data that had been collected previously were not included in the analyses.

Consenting women were given a trial intramuscular injection of the inert oil placebo and asked to return in one week for randomization. If a woman did not return for a randomization visit between 16 weeks and 20 weeks 6 days of gestation, she was not permitted to participate in the trial. Returning eligible patients were then assigned to receive identically appearing active (17P) or placebo (castor oil) injections prepared by a research pharmacy. The women, their caregivers, and research personnel were not informed of the study-group assignment.

The boxes of 17P or placebo were packaged for each center according to a randomization sequence prepared by the George Washington University Biostatistical Coordinating Center. The urn method of randomization,¹³ with stratification according to clinical center, was used to create the computer-generated randomization sequence. A 2:1 ratio was used for the assignment of women to 17P or to placebo, because it was known that patients assigned to placebo would be receiving painful injections on a weekly basis with no possibility of direct benefit.

After entering the study, the subjects returned for weekly injections of 17P or placebo given by a study nurse; the injections continued until 36 weeks of gestation or delivery, whichever occurred first. In addition to the weekly visits for study injections, the women received prenatal care at their institutions, as judged appropriate by their caregivers for their known level of risk of preterm delivery.

After delivery, study personnel reviewed all prenatal, delivery, newborn, and postpartum records and documented the date of delivery, birth weight of the infant, and neonatal course, as well as the occurrence of complications of pregnancy. Infants were followed until discharge from the hospital where they were born or, if they were transferred elsewhere, from the hospital to which they were transferred. Preterm delivery was defined as delivery at less than 37 completed weeks (259 days) of gestation, calculated as delineated above.

The analysis was performed according to the intention-to-treat principle. Continuous variables were compared with the use of the Wilcoxon rank-sum test, and categorical variables were compared with the use of the chi-square or Fisher's exact test (the latter when there was an expected value of less than five for any cell). Prolongation of pregnancy was assessed by life-table methods, with the duration considered being that between the time of randomization and the time a woman gave birth, was lost to follow-up, or reached 40 weeks of gestation, whichever came first. Curves for event-free survival were estimated with use of the Kaplan–Meier method, with adjustment to account for differing durations of gestation at entry, and were tested with the log-rank test.

On the basis of data from a previous study by the Maternal–Fetal Medicine Units Network,¹⁴ we estimated that 37 percent of the women in the placebo group would deliver before 37 weeks of gestation. With the use of this estimate, a total sample size of 500 women (334 in the 17P group and 166 in the placebo group) was deemed to be sufficient for the detection of a reduction of 33 percent in the rate of preterm delivery (from 37 percent to 25 percent), under the assumptions of a type I error (two-sided) of 5 percent and a power of at least 80 percent. Before the study began, it was decided that the independent data and safety monitoring committee would use the group sequential method of Lan and DeMets,¹⁵ with a spending function for the type I error corresponding to the O'Brien–Fleming boundary, for interim monitoring and adjustment of the type I error. At the second interim analysis, conducted when 463 patients had undergone randomization, outcome data were available for 351 patients (70 percent of the planned sample). The boundary (P=0.015) for the test of significance of the primary outcome, preterm delivery, was found to have been crossed, and enrollment in the trial was halted.

A total of 2980 women were identified as potentially eligible for the study on the basis of a review of medical records from September 1999 to February 2002. Of these women, 1039 were found to be eligible, and 463 eligible women gave consent for the trial and underwent random assignment to 17P or placebo. The main reasons for ineligibility included lack of documentation of the qualifying preterm delivery (in the cases of 549 women), a gestational age of more than 20 weeks (482 women), and current or planned cervical cerclage (241 women).

The characteristics of the 310 women in the 17P group and the 153 women in the placebo group are shown in Table 1. The women in the two groups were similar in terms of the mean duration of gestation in the qualifying delivery, the mean duration of gestation at the time of randomization, race or ethnic group, marital status, body-mass index, educational level, smoking status, and substance use during pregnancy. The women in the placebo group had had more previous preterm deliveries (mean, 1.6 vs. 1.4; P=0.007).

Noncompliance was defined by a gap of 10 days or more between any two injections. According to this definition, 91.5 percent of the women were compliant with all of their injections. There was no difference in the rate of compliance between the two groups. A total of 231 women (50 percent) reported at least one adverse effect. The most common side effects were local injection-site reactions, including soreness (in 34.2 percent of the women), swelling (in 14.1 percent), itching (in 11.3 percent), and bruising (in 6.7 percent). More women in the 17P group than in the placebo group had swelling at the injection site (17.2 percent vs. 7.8 percent, P=0.007) or a lump at the injection site (5.5 percent vs. 1.3 percent, P=0.03).

Outcome data were available for 459 of the 463 women (99.1 percent) (Table 2). The frequency of delivery before 37 weeks of gestation was 36.3 percent in the 17P group, as compared with 54.9 percent in the placebo group (P<0.001). Delivery before 35 weeks of gestation was also less frequent in the 17P group (20.6 percent vs. 30.7 percent, P=0.02). There was a 42 percent reduction in the rate of delivery before 32 weeks of gestation in the 17P group (11.4 percent vs. 19.6 percent, P=0.02). Rates of preterm delivery in the 17P group did not differ according to the week of gestation at the time of the qualifying delivery. Survival analysis showed a significant prolongation of pregnancy with 17P as compared with placebo (P=0.01). Because there was an imbalance between the 17P and placebo groups with regard to the number of previous preterm deliveries, we performed an analysis with adjustment for this variable. The adjusted relative risk of delivery before 37 weeks of gestation in the 17P group as compared with the placebo group was 0.70 (95 percent confidence interval, 0.57 to 0.85). There were no significant differences between the two groups in the rates of hospital visits for preterm labor, use of tocolytic drugs, corticosteroid use, cesarean delivery, or chorioamnionitis (Table 2).

More than half the women enrolled were black. The reduction in the rate of preterm delivery with 17P among the black women was very similar to that among nonblack women (Table 2).

The effectiveness of 17P in this study suggests that only 5 to 6 women (95 percent confidence interval, 3.6 to 11.1) with a level of risk for preterm delivery similar to that among these women would need to be treated in order to prevent one preterm delivery before 37 weeks of gestation. Similarly, 12 women (95 percent confidence interval, 6.3 to 74.6) with a similar level of risk would need to be treated in order to prevent one delivery before 32 weeks of gestation.

Rates of spontaneous miscarriage between 16 weeks of gestation and 19 weeks 6 days of gestation, and rates of fetal death after 19 weeks 6 days of gestation are shown in Table 2 and Table 3. There was a small and nonsignificant increase in the rate of miscarriages and stillbirths in the 17P group as compared with the placebo group. With one exception, all stillbirths occurred before 24 weeks of gestation.

There was a significant reduction in the risk of a birth weight of less than 2500 g in the 17P group as compared with the placebo group (relative risk, 0.66; P=0.003) and a nonsignificant reduction in the risk of a birth weight of less than 1500 g (relative risk, 0.62; P=0.08) (Table 3). Treatment with 17P led to significant reductions in the rates of necrotizing enterocolitis (P=0.01), need for supplemental oxygen, and intraventricular hemorrhage of any grade. However, there was no significant difference between groups in the rate of intraventricular hemorrhage of grade 3 to 4 specifically. The rates of infant death, transient tachypnea in the newborn, respiratory distress syndrome, bronchopulmonary dysplasia, need for ventilatory support, retinopathy of prematurity, and patent ductus arteriosus were slightly but not significantly lower in the 17P group. Of the 17 neonatal deaths, 16 were due to complications of prematurity and 1 to intrapartum hypoxia subsequent to uterine rupture.

Nine of the infants were found to have congenital malformations (2.0 percent in each group). There was no consistent pattern to these defects, and none involved genital organs. One infant of a woman in the 17P group had torsion of the testicles in utero, with subsequent infarction.