Materials and Methods
Subjects and screening
This open-label, multicenter, randomized controlled trial was performed in 11 university hospitals across France. Women were potentially eligible for inclusion if they presented an asymptomatic singleton pregnancy considered at high risk for PTB because of a history of spontaneous PTB or of cervical surgery, a uterine malformation, or prenatal exposure to DES. In this asymptomatic high-risk group, the follow-up included serial CL measurements by TVU at 2 week intervals, starting at 16 +0 weeks of gestation.
Enrollment in the trial was offered to women between 20 +0 and 31 +6 weeks of gestation who had a TVU CL measurement (in the sagittal plane according to the standard technique ) less than 25 mm. No further CL measurements were taken once one was found to be less than 25 mm. Women also had to be at least 18 years old, agree to regular follow-up, and provide written informed consent.
Women with any of the following characteristics were ineligible: cervical dilatation greater than 3 cm, chorioamnionitis, premature rupture of the membranes, placenta previa, twin pregnancy, severe intrauterine growth restriction, any known major structural or chromosomal fetal abnormality, any maternal or fetal disease requiring induced PTB, progestogen therapy before inclusion, ongoing anticonvulsant treatment, or participation in any other treatment trial.
A first-trimester ultrasound scan, routinely performed in France, was used to determine gestational age. The placement of a cerclage was left to the physician’s discretion when the CL was less than 25 mm before 23 weeks. Abdominal ultrasound was performed before inclusion to assess both fetal well-being and the quantity of amniotic fluid, and vaginal sonography to confirm the CL (<25 mm). The Ethics Committee of Poissy Saint-Germain Hospital (Comité de Protection des Personnes), Saint-Germain en Laye, France, approved the study protocol for all centers. The trial is registered at ClinicalTrials.gov (no. NCT00331695 ).
The 17OHP-C used in our trial was Progestérone Retard Pharlon (Bayer Pharma AG, Berlin, Germany), the only 17OHP-C commercially available in France. It is licensed by the Agence Nationale de Sécurité du Médicament and has been marketed in France since 1958. It is currently manufactured by Bayer Pharma AG, which produces it in Germany in accordance with the detailed guidelines for starting materials established in part II of the European Union Good Manufacturing Practices. The manufacturer of the active substance is included in Bayer’s audit program.
Randomization and follow-up
After verification of the inclusion and exclusion criteria, eligible consenting women were randomly assigned in a 1:1 ratio to receive 500 mg of intramuscular 17OHP-C weekly until 36 weeks or PTB, whichever occurred first, or to no treatment with 17OHP-C (control group). Additional management in both arms was determined by the attending physician, except that progestogens were not allowed in the control group. An independent, centralized, computer-generated randomization sequence (CleanWeb; Télémedecine Technologies, Boulogne, France) was used for this allocation, based on a randomization list established by the study statistician, according to a permuted block method, balanced, and stratified by center.
If an episode of preterm labor occurred before 34 weeks, the woman was admitted and received tocolysis and a course of betamethasone 12 mg, given intramuscularly and repeated after 24 hours. The attending physician determined the type, duration, and regimen of tocolysis and decided about any maintenance tocolysis. Patients in the 17OHP-C arm continued treatment with 17OHP-C while hospitalized if delivery did not occur. In addition to the monthly visits for the study, the women received prenatal care at their institutions, as judged appropriate by their caregivers for their known level of PTB risk.
Study outcomes
The primary outcome was the time from randomization to delivery. Prespecified secondary outcomes were as follows: (1) obstetric criteria: rates of PTB before 37, 34, and 32 weeks and the number of readmissions for preterm labor; (2) neonatal criteria: birthweight, transfer to a neonatal intensive care unit (NICU), respiratory distress syndrome, bronchopulmonary dysplasia, necrotizing enterocolitis, periventricular leukomalacia, and death; and (3) safety criteria: any severe maternal or neonatal adverse effects (congenital anomalies or other ill effects).
Statistical analysis
A sample size of 240 women (120 per group) was deemed sufficient to demonstrate a mean prolongation of pregnancy of 14 days with 17OHP-C with at least 90% power and a 2-sided 5% type I error rate, assuming an SE of 33 days for the prolongation of pregnancy.
The main analysis was performed according to the intention-to-treat principle (ie, all randomized women were analyzed in the group to which they were allocated, regardless of protocol deviations). Prolongation of pregnancy was assessed by life table methods that defined duration as the period between inclusion and delivery or loss to follow-up, whichever occurred first. Patients lost to follow-up were censored at that time.
The distribution of the prolongation of pregnancy was compared across groups with the Gehan-Wilcoxon test. An additional comparison, adjusted for gestational age at randomization, used the Cox proportional hazards model. An adjusted estimate of the mean difference in the prolongation of pregnancy was derived from a Weibull model.
For the secondary outcomes of PTB rates before 37, 34, and 32 weeks, missing data because of the loss to follow-up were imputed according to 2 different strategies: all missing data were first treated as failures for all subjects and then as failures for patients in the experimental arm and as successes for those in the control arm (worst-case scenario). A secondary analysis of the complete cases compared continuous variables with the Wilcoxon rank-sum test and categorical variables with a Fisher exact test. Secondary pregnancy, fetal, and neonatal outcomes were assessed with Fisher exact tests. All tests were 2 sided, at the P = .05 significance level. The R statistical package version 2.15.2 (R Foundation for Statistical Computing, Vienna, Austria) was used to analyze the data.
Two simultaneous studies assessed the efficacy of 17OHP-C among the following: (1) women admitted for an episode of preterm labor with intact membranes, successfully arrested by tocolytic treatment and a CL less than 25 mm, and (2) asymptomatic twin pregnancies with a CL less than 25 mm. An interim analysis of the results in this study was planned at their conclusion if recruitment in the trial reported here had not been completed.
The main objective of this interim analysis was 2-fold: to enable us either to stop the study early or to reassess the sample size according to the adaptive design methodology described by Bauer and Kohne. Because the other 2 studies did not find 17OHP-C to be superior, we used conditional power at the interim analysis to assess whether either stopping for futility or sample size reassessment was appropriate.
Results
Patient characteristics
An interim analysis was performed after 105 women (51 in the 17OHP-C group and 54 in the control group) had been enrolled between August 2006 and December 2009 at 11 study locations. The conditional power was 68% for the unadjusted analysis and 51% for the analysis adjusted for gestational age at randomization. This means that, given the interim results and even if 17OHP-C was found to prolong pregnancy by a mean of 14 days in all the future cases, the unadjusted analysis of the trial after inclusion of the 135 remaining patients would have only a 68% probability of showing a significant effect of 17OHP-C and the adjusted analysis only slightly more than a 50% chance.
To reach the preplanned 90% power, 264 additional patients would have had to be included for the unadjusted analysis and 320 for the adjusted analysis, for a total of 369 or 425 patients in the trial. Given the nonsignificant results of the other 2 trials and the accrual rate for the current trial, the investigators did not consider it feasible to continue the trial, which was stopped for reasons of futility.
All women met the inclusion criteria and none of the exclusion criteria. Two patients were lost to follow-up immediately after inclusion, 1 in the each group ( Figure 1 ). Table 1 summarizes the patient’s characteristics according to group. The groups were similar for maternal age, body mass index, parity, gestational age at inclusion, history of uterine anomalies, DES syndrome, previous preterm delivery or midtrimester abortion, and CL at randomization. All cervical cerclages were performed before randomization. The median (interquartile range: first quartile [Q 1 ] to third quartile [Q 3 ]) gestational age at randomization was 24 +6 (22 +6 to 27 +1 ) in the 17OHP-C group and 25 +4 (22 +6 to 27 +1 ) in the control group.
| Characteristic | 17OHP-C (n = 51) | No 17OHP-C (n = 54) |
|---|---|---|
| Maternal age, median (Q 1 –Q 3 ), y | 30 (26–36) | 32 (27–36) |
| Body mass index, median (Q 1 –Q 3 ), kg/m² | 22.9 (21.0–26.4) | 22.3 (20.1–26.4) |
| Gestational age, median (Q 1 –Q 3 ), wks | 24 +6 (22 +6 to 27 +1 ) | 25 +4 (22 +6 to 27 +1 ) |
| Smoking during pregnancy, n/total (%) | 7/48 (15) | 7/51 (14) |
| Previous pregnancies, n | ||
| Patients, n | 49 | 53 |
| Mean ± SD | 1.8 ± 1.5 | 1.8 ± 1.5 |
| One or more previous midtrimester abortions, n/total (%) | 23/49 (47) | 20/53 (38) |
| One or more previous spontaneous PTBs, n/total (%) | 27/49 (55) | 30/53 (57) |
| One or more previous term deliveries, n/total (%) | 13/49 (27) | 16/53 (30) |
| Uterine malformation, n/total (%) | 10/49 (20) | 10/53 (19) |
| Diethylstilbestrol syndrome, n/total (%) | 4/49 (8) | 6/53 (11) |
| History of cervical surgery, n/total (%) | 2/49 (4) | 4/53 (8) |
| Cerclage before randomization, n/total (%) | 22/49 (45) | 20/53 (38) |
| Cervical length at admission, median (Q 1 –Q 3 ), mm | 19 (10–20) | 17 (11–21) |
Primary outcome and preterm delivery
Figure 2 and Table 2 summarize the obstetric outcomes. The intention-to-treat analysis with censoring at last follow-up showed no significant difference between the 17OHP-C and control groups for median [Q 1– Q 3 )] time to delivery (77 [54–103] and 74 [52–99] days, respectively; mean difference, 4; 95% confidence interval [CI], –9 to +17) ( Table 2 ); nor did they differ significantly in their rates of preterm delivery before 37 (45% vs 44%, P > .99), 34 (24% vs 30%, P = .50), or 32 (14% vs 20%, P = .44) weeks of gestation, when missing data were considered as failure ( Table 2 ). After adjustment for gestational age at inclusion, time to delivery in the 17OHP-C group did not increase at all: 0 days (95% CI, −7 to +8).
| Characteristics | 17P | No 17P | Mean difference (95% CI) | Relative risk (95% CI) | P value |
|---|---|---|---|---|---|
| Patients randomized, n | 51 | 54 | |||
| Intention-to-treat analysis | |||||
| Analysis with censoring at last follow-up | |||||
| Time to delivery (d), mean (SE) a | 76 (5) | 72 (5) | 4 (−9 to 17) | .48 b | |
| Median (Q 1 –Q 3 ) | 77 (54–103) | 74 (52–99) | |||
| Analysis of imputed data: missing data considered as failure | |||||
| Delivery <37 wks, n (%) | 23 (45) | 24 (44) | +1% (−18 to 19) | 1.01 (0.66–1.55) | > .99 |
| Delivery <34 wks, n (%) | 12 (24) | 16 (30) | −6% (−22 to 11) | 0.79 (0.42–1.51) | .51 |
| Delivery <32 wks, n (%) | 7 (14) | 11 (20) | −7% (−21 to 8) | 0.67 (0.28–1.60) | .44 |
| Analysis of imputed data: missing data considered as failure if experimental/success if control | |||||
| Delivery <37 wks, n (%) | 23 (45) | 23 (43) | +3% (−16 to 21) | 1.06 (0.69–1.63) | .85 |
| Delivery <34 wks, n (%) | 12 (24) | 15 (28) | −4% (−20 to 12) | 0.85 (0.44–1.63) | .66 |
| Delivery <32 wks, n (%) | 7 (14) | 10 (19) | −5% (−19 to 10) | 0.74 (0.31–1.80) | .60 |
| Analysis of complete cases | |||||
| Time to delivery | |||||
| Patients, n | 50 | 53 | |||
| Median (Q 1 –Q 3 ), d | 77 (52–102) | 74 (46–99) | 4 (−10 to 18) | .50 | |
| Delivery <37 wks, n/total (%) | 22/50 (44) | 23/53 (43) | +1% (−18 to 19) | 1.01 (0.65–1.57) | > .99 |
| Delivery <34 wks, n/total (%) | 11/50 (22) | 15/53 (28) | −6% (−22 to 10) | 0.78 (0.40–1.53) | .50 |
| Delivery <32 wks, n/total (%) | 6/50 (12) | 10/53 (19) | −7% (−21 to 8) | 0.64 (0.25–1.62) | .42 |
a Estimated mean for censored data
We also performed analyses adjusted for cerclage placement at baseline. Results did not differ significantly from the main analysis: there was an adjusted mean difference of +4 days in favor of the 17OHP-C group (95% CI, –9 to +18).
Finally, we conducted 2 post-hoc subgroup analyses: by gestational age categories at inclusion ( Table 3 ) and by CL categories at inclusion ( Table 4 ). These categories did not differ significantly between the 17OHP-C and control groups for time from randomization to delivery. No patient (with known outcome) had a midtrimester abortion in the current pregnancy. The rates of corticosteroid administration for fetal lung maturity, acute tocolysis, and maintenance of tocolysis did not differ significantly in the 17OHP-C and control groups (57% vs 63%, P = .54; 17% vs 22%, P = .80; 27% vs 32%, P = .65, respectively) ( Table 5 ). The mode of delivery was also similar. No patient in the control group received progestagens.
| Subgroup Arm | Gestational age <25 weeks | Gestational age ≥25 weeks | ||
|---|---|---|---|---|
| 17P | No 17P | 17P | No 17P | |
| Patients, n | 27 | 25 | 24 | 29 |
| Time to delivery, d | ||||
| Mean (SE) | 94 (6) | 83 (8) | 56 (5) | 62 (5) |
| Mean difference (95% CI) | +11 (−11 to +29) | −5 (−17 to +9) | ||
| P value | .37 | .47 | ||
| Tests for interaction | Quantitative interaction: P = .18; qualitative interaction a : P > .20 | |||
| Subgroup Arm | Cervical length <20 mm | Cervical length ≥20 mm | ||
|---|---|---|---|---|
| 17P | No 17P | 17P | No 17P | |
| Patients, n | 29 | 34 | 22 | 20 |
| Time to delivery, d | ||||
| Mean (SE) | 74 (7) | 71 (6) | 79 (7) | 73 (6) |
| Mean difference (95% CI) | +3 (–18 to +19) | +6 (–13 to +26) | ||
| P value | .75 | .50 | ||
| Tests for interaction | Quantitative interaction: P = .82 a | |||
a No test for qualitative interactions because both effects are in the same direction.
| Characteristics | 17OHP-C | No 17OHP-C | P value |
|---|---|---|---|
| Weeks of gestation at delivery | |||
| Patients, n | 50 | 53 | |
| Median (Q 1 –Q 3 ) | 37 +3 (35 +0 to 38 +5 ) | 37 +2 (33 +2 to 39 +0 ) | .76 |
| Delivery before 37 wks’ gestation | .18 | ||
| Patients, n | 22 | 23 | |
| Preterm labor, n (%) | 11 (50) | 9 (39) | |
| Preterm PROM, n (%) | 7 (32) | 13 (57) | |
| Elective delivery, n (%) | 4 (18) | 1 (4) | |
| Corticosteroids for fetal lung maturity, n/total (%) | 26/46 (57) | 32/51 (63) | .54 |
| Acute tocolysis, n/total (%) | 8/46 (17) | 11/51 (22) | .80 |
| Maintenance tocolysis, n/total (%) | 12/45 (27) | 16/50 (32) | .65 |
| Patients, n | 46 | 51 | |
| Mode of delivery, n (%) | .85 | ||
| Vaginal delivery | 30 (65) | 36 (71) | |
| Cesarean delivery during labor | 11 (24) | 11 (22) | |
| Elective cesarean delivery | 5 (11) | 4 (8) |
Fetal and neonatal outcome
The median (Q 1 –Q 3 ) birthweight was similar in the 17OHP-C (2875 [2011–3298] g) and control (2860 [2280–3111] g; P = .55) ( Table 6 ). The groups did not differ significantly for rates of NICU admission, respiratory distress syndrome, bronchopulmonary dysplasia, necrotizing enterocolitis, periventricular leukomalacia, neonatal death, or adverse maternal or fetal events ( Table 6 ). There were 4 intrauterine deaths and 2 per partum deaths in the control group and no perinatal death in the 17OHP-C group. Nonetheless, the difference was not significant.
| Characteristic | 17OHP-C (n = 51) | No 17OHP-C (n = 54) | P value |
|---|---|---|---|
| Outcome, n (%) | |||
| Born alive | 49 (96) | 50 (93) | |
| Intrauterine death | 0 (0) | 2 (4) | |
| Peripartum death | 0 (0) | 1 (2) | |
| Unknown issue | 2 (4) | 1 (2) | |
| Birthweight, median (Q 1 –Q 3 ), g | 2875 (2011–3298) | 2860 (2280–3111) | .55 |
| NICU transfer, n/total (%) | 12/43 (28) | 13/50 (26) | > .99 |
| Respiratory distress syndrome, n/total (%) | 8/43 (19) | 12/49 (24) | .61 |
| Bronchopulmonary dysplasia, n/total (%) | 1/43 (2) | 3/49 (6) | .62 |
| Intraventricular hemorrhage, n/total (%) | 0/43 (0) | 1/49 (2) | > .99 |
| Sepsis, n/total (%) | 2/43 (5) | 1/49 (2) | .60 |
| Neonatal death, n/total (%) | 1/43 (2) | 1/50 (2) | > .99 |
Adverse events
There was no difference in the rate of congenital anomalies or other potential adverse effects between mothers or infants exposed to 17OHP-C 500 mg weekly and the control group.
Results
Patient characteristics
An interim analysis was performed after 105 women (51 in the 17OHP-C group and 54 in the control group) had been enrolled between August 2006 and December 2009 at 11 study locations. The conditional power was 68% for the unadjusted analysis and 51% for the analysis adjusted for gestational age at randomization. This means that, given the interim results and even if 17OHP-C was found to prolong pregnancy by a mean of 14 days in all the future cases, the unadjusted analysis of the trial after inclusion of the 135 remaining patients would have only a 68% probability of showing a significant effect of 17OHP-C and the adjusted analysis only slightly more than a 50% chance.
To reach the preplanned 90% power, 264 additional patients would have had to be included for the unadjusted analysis and 320 for the adjusted analysis, for a total of 369 or 425 patients in the trial. Given the nonsignificant results of the other 2 trials and the accrual rate for the current trial, the investigators did not consider it feasible to continue the trial, which was stopped for reasons of futility.
All women met the inclusion criteria and none of the exclusion criteria. Two patients were lost to follow-up immediately after inclusion, 1 in the each group ( Figure 1 ). Table 1 summarizes the patient’s characteristics according to group. The groups were similar for maternal age, body mass index, parity, gestational age at inclusion, history of uterine anomalies, DES syndrome, previous preterm delivery or midtrimester abortion, and CL at randomization. All cervical cerclages were performed before randomization. The median (interquartile range: first quartile [Q 1 ] to third quartile [Q 3 ]) gestational age at randomization was 24 +6 (22 +6 to 27 +1 ) in the 17OHP-C group and 25 +4 (22 +6 to 27 +1 ) in the control group.
