Wing, Deborah A. MD; Miller, Hugh MD; Parker, Lamar MD; Powers, Barbara L. MSN, PhD; Rayburn, William F. MD; for the Misoprostol Vaginal Insert Miso-Obs-204 Investigators
Rates of induction of labor continue to rise, with current levels exceeding 22% of U.S. pregnancies.1 The most common reasons for induction include postterm pregnancy, preeclampsia, and diabetes mellitus.2 Elective inductions of labor contribute to the increased frequency. Often induced labors require cervical ripening, and in these circumstances prostaglandins, in a variety of forms, have been used to successfully induce labor.
The misoprostol vaginal insert (MVI) contains misoprostol as the active ingredient, but is otherwise identical to the Cervidil vaginal insert, the 10 mg dinoprostone product licensed and marketed worldwide for cervical ripening and induction of labor. This vaginal delivery system utilizes misoprostol and may improve safety by the controlled release of the drug reservoir and retrieval tape technologies, which offer the ability to remove the product at the onset of labor or if an adverse reaction occurs. The dose reservoir releases approximately 1/24th of the total dose per hour of insertion.
Misoprostol is a synthetic prostaglandin currently marketed in oral tablet form for prophylaxis against nonsteroidal antiinflammatory drug–induced gastric ulcers. Off-label use of misoprostol tablets (Cytotec), given either by mouth or as intravaginal tablet fragments, has been shown to ripen the cervix and induce labor.3
Several previous studies have demonstrated the dose-response, efficacy, and safety of multiple doses of the misoprostol vaginal insert in nonpregnant women4 and in pregnant women at term.5–7 In this report, we describe the results of a phase II dose-ranging study of the MVI 100, MVI 150, and MVI 200 dose reservoirs in participants who required cervical ripening before the induction of labor.
MATERIALS AND METHODS
The study was reviewed and approved by the institutional review boards of the 11 participating institutions (see the Appendix online at http://links.lww.com/AOG/A219). Of these, seven (64%) were university-affiliated and the remaining four were community-based sites. All participants were informed and provided written consent before participation in the study.
Participants were at least 18 years of age, of low parity (three or lower) with singleton pregnancies, and at least 36 0/7 weeks of gestation. They had no conditions requiring urgent delivery. The participants required cervical ripening defined as a baseline modified Bishop score of 4 or lower. Additionally, participants were excluded for active labor; presence of uterine or cervical scar or uterine abnormality; receipt of oxytocin or any cervical ripening or labor inducing agents or a tocolytic drug within 7 days before enrollment; severe preeclampsia with either central nervous system involvement or hemolysis, elevated liver enzymes, low platelets syndrome; suspected or confirmed cephalopelvic disproportion or fetal malpresentation or both; diagnosed fetal abnormalities; any evidence of fetal compromise at baseline (eg, nonreassuring fetal heart rate [FHR] pattern or meconium staining); receipt of a nonsteroidal antiinflammatory drug within 4 hours of study treatment; ruptured membranes 48 hours or more before the start of treatment or suspected chorioamnionitis; fever higher than 37.5°C; any condition in which vaginal delivery was contraindicated, such as placenta previa; known or suspected allergy to prostaglandins; or inability to comply with the protocol.
After obtaining informed consent, baseline information was collected including demographic information, a medical and pregnancy history, and maternal height and weight. An examination was performed to confirm vertex fetal presentation and determine baseline modified Bishop score. Each participant underwent at least 15 minutes of cardiotocographic assessment to ensure reassuring fetal status and confirm no uterine pattern indicative of active labor.
The study was conducted in a randomized, double-blind fashion and the MVI 100, MVI 150, and MVI 200 vaginal inserts were identical in appearance. Random assignment was ensured by providing sequentially numbered kits for each parity cohort (nulliparous or parous). After completion of baseline procedures, the participants were assigned the next kit according to their parity cohort. The randomization sequence was created using a computer-generated randomization scheme and was stratified by center, ensuring random allocation to each arm of the study. Enrollment was also stratified by parity to ensure a population of approximately 60% nulliparous women and 40% parous women to allow interstudy results comparisons for time-to-event and rate of cesarean delivery analyses.
The study drug was placed high in the vaginal fornix using water-soluble gel as needed. The participants remained in bed for at least 2 hours after insertion. Participants were continually monitored for uterine activity and FHR while in the study, except for brief trips to the bathroom and short periods of ambulation.
The vaginal insert was to be discontinued for onset of active labor, study drug falling out of the vagina, completion of the 24-hours dosing period, maternal-fetal complications including uterine contractile abnormalities or nonreassuring FHR patterns, and maternal request. Uniform definitions were used for uterine contractile abnormalities and other adverse events as in our previous study with the exception of the FHR and uterine activity patterns.5 Hyperstimulation syndrome was defined as the combination of any nonreassuring FHR with tachysystole or hypertonus. The FHR patterns were defined for this protocol using the Eunice Kennedy Shriver National Institute of Child Health and Human Development categorizations.8 The term “non reassuring FHR pattern that prompts clinical intervention” was recorded as the adverse event for category II and III events and the pattern of concern was to be identified. Conventional interventions to treat nonreassuring FHR were applied where appropriate at the discretion of the managing physician.
The primary efficacy measure was the proportion of women with vaginal delivery within 24 hours after insertion of the study drug. Secondary efficacy end points included time to vaginal delivery of the neonate with an additional analysis of time to any delivery mode; adverse events as defined previously5; proportion of women undergoing cesarean delivery; proportion of women at 12 hours with successful cervical ripening based on achievement of the composite end points; proportion of women delivering vaginally within 12 hours; proportion of women exposed to predelivery oxytocin; and time to onset of active labor.
Active labor was defined as at least three firm, rhythmic contractions with duration at least 45 seconds occurring within a 10-minute period, or achievement of 4 cm dilatation. After delivery, both the mother and neonate were followed for 48 hours or until hospital discharge, whichever came sooner. Time-to-event analyses such as time to active labor and time to vaginal delivery were based on the initial induction attempt (first hospitalization). Participants failing to deliver in the first hospitalization were censored from these analyses. Participants who had cesarean delivery during the first hospitalization were censored using the longest time interval from study drug administration to cesarean delivery during the first hospitalization independent of treatment group.
MVI 100 was considered the comparison group. Based on the primary efficacy end point, sample size estimates were prepared to compare the MVI 100 and MVI 200 treatment arms. As this was a dose-ranging study, the MVI 150 arm was included to ensure that the lowest effective dose would be identified. Sample size estimates were prepared to compare the MVI 100 and MVI 200 treatment arms based on the primary efficacy end point of proportion of women delivering vaginally within 24 hours. A minimum sample size of approximately 120 individuals per arm would provide 84 individuals per arm with vaginal delivery. This would ensure more than 80% power (P=.05, two-sided) to detect an improvement of 20 percentage points in the proportion of women delivering vaginally within 24 hours for MVI 200 compared with an expected rate of 60% for MVI 100. Based on these assumptions, we used a total sample size of 360 participants to be distributed equally among the three treatment arms and stratified by site to obtain an approximate distribution of 60% nulliparous and 40% parous women.
Adverse event details included start and stop times, severity, and causality relationship to study drug. The principal investigator at each site was responsible for reviewing the electronic fetal monitoring and cardiotocographic strips after delivery to identify any missed adverse events. The categories for cesarean indication included possible fetal compromise; arrest of dilatation; arrest of descent; failure to progress; or lack of efficacy (24 hours of drug exposure with no active labor). At the discretion of the treating clinician, participants with lack of efficacy could have additional interventions such as transcervical Foley bulb placement or administration of the dinoprostone vaginal insert (Cervidil) if additional cervical ripening was necessary, or oxytocin could be initiated. A Data and Safety Monitoring Board was convened to oversee the safety and general conduct of the study. The Board conducted one preplanned blinded review as to the identity of the treatment groups to determine whether the rate of cesarean births in any of the treatment arms was unsafe, defined as being significantly greater than the 30% expected rate defined a priori by the U.S. Food and Drug Administration's (FDA's) Division of Reproductive and Urologic Drug Products. This review was conducted after approximately 150 participants had been enrolled, and the Data and Safety Monitoring Board recommended that there were no significant safety signals and the study could continue with no changes to the protocol. The blinded safety review was performed using participants who had data available for both study drug insertion and delivery mode.
Statistical analyses were performed using SAS 8.2 statistical software and all tests were two-sided tests assessed at the .05 significance level. Comparisons are reported using the MVI 100 group as the referent. For baseline characteristics, continuous variables were evaluated using a one-way analysis of variance categorical variables evaluated using Fisher exact or χ2 tests. For the analysis of proportion of participants with vaginal delivery by hour 24, exact confidence intervals were reported and treatment groups were compared using a Cochran-Mantel Haenszel test adjusted for site. For the analysis of oxytocin use, a two-way analysis of variance was used with factors of treatment and site. Kaplan-Meier estimates and log-rank tests were used to compare even time distribution functions, for example, times to delivery and times to onset of active labor. Participants who had cesarean delivery during the first hospitalization were censored using the longest time interval from study drug administration to cesarean delivery during the first hospitalization independent of treatment group. Rates of cesarean delivery and adverse events were analyzed using Fisher exact tests.
A total of 374 women were enrolled in this investigation from April 22, 2009, through December 19, 2009 (Fig. 1). The demographic and baseline characteristics of the participants were comparable between groups (Table 1) with the exception of indications for induction: more MVI 200–treated women underwent elective induction and more MVI 100–treated women were induced for hypertension than the other groups. The median modified Bishop score at enrollment was 3 (range 0–4) for all three treatment groups.
The median duration of drug exposure was 861 (range 24–1,480) minutes for the MVI 100 group, 675 (range 80–1,516) minutes for the MVI 150 group (P=.001), and 515 (range 43–1,493) minutes for the MVI 200 group (P<.001, generalized linear model). The most common reason for removal of the study drug was onset of active labor (39.4% of participants), which occurred with similar frequencies among treatment groups (Table 2). The second most common reason for removal of the MVI 150 and MVI 200 was maternal-fetal complications, which included abnormal FHR patterns and vaginal bleeding, among others (P<.001 for each dose reservoir compared with MVI 100). Study drug in situ for 24 hours was the third most common removal reason for the MVI 100 group (22.2%); this reason was significantly less frequent for the MVI 200 group (8.4%, P=.002).
Successful cervical ripening at 12 hours was not different among the groups, with 77.78%, 77.6%, and 80.15% in the MVI 100, MVI 150, and MVI 200 groups, respectively (P=.98 and .65 for MVI 150 and MVI 200 compared with MVI 100, respectively).
Participants treated with the MVI 200 entered active labor faster than those treated with the two other doses. Median time to active labor was 1,069 minutes (range 885–1,153) for the MVI 100 group compared with 775 minutes (range 724–977) for the MVI 150 (P=.16) and 701 minutes (range 550–759) for the MVI 200 (P=.01). Women treated with the MVI 200 also had the shortest median time to any delivery mode (P<.001 MVI 100 compared with MVI 200) (Table 3).
Vaginal delivery within 24 hours was achieved for 63.8%, 66.7%, and 76.0% of participants exposed to MVI 100, MVI 150, and MVI 200, respectively (P=.54 for MVI 100 compared with MVI 150, and P=.057 for MVI 100 compared with MVI 200) (Table 3 and Fig. 2). Twenty-four percent of women receiving MVI 200 failed to achieve vaginal delivery within 24 hours compared with 36.3% of those receiving MVI 100 (P=.057, relative risk [RR] 0.66, 95% confidence interval [CI] 0.42–1.04). For nulliparous women, 67.3% treated with MVI 200 had vaginal delivery within 24 hours compared with 52.4% treated with MVI 100 (P=.14). Approximately 75% (99/136) of parous individuals delivered vaginally within 24 hours. Results by parity are presented in Table 3. Median time to vaginal delivery was more than 9 hours shorter for the MVI 200 group than for the MVI 100 group (1,181 [range 1,035–1,443] minutes, P=.02). A significantly higher proportion of MVI 200 participants delivered by any mode within 24 hours (72.5%) compared with 52.14% of MVI 100 participants (P<.001); this finding was also true for MVI 200 nulliparous and parous women separately as well as for the combined group (P=.012 and P=.016, respectively, by parity).
Oxytocin augmentation was significantly reduced for those treated with MVI 200 compared with MVI 100 (48.9% compared with 70.9%, P<.001, RR 0.70, 95% CI 0.56–0.85) (Table 4) with less than half of the MVI 200 participants requiring any oxytocin compared with more than 70% of MVI 100 participants.
Fetal heart rate patterns were found to be Eunice Kennedy Shriver National Institute of Child Health and Human Development category II or III in the majority of participants (58.0% overall) (Table 4) at some stage of their labors; most were deemed unrelated to the misoprostol vaginal insert. Category II or III patterns judged related to the use of the misoprostol vaginal insert occurred in 9 of 75 (7.6%), 27 of 125 (21.6%), and 22 of 131 (16.8%) of the MVI 100, 150, and 200 participants, respectively (P=.03 for MVI 100 compared with MVI 200).
MVI 200 was associated with an increased rate of tachysystole (54/131, 41.2%) compared with MVI 100 (23/118, 19.5%) (P<.001, RR 2.11, 95% CI 1.39–3.22), and MVI 150 (32/135, 25.6%) (P=.26, RR 1.31, 95% CI 0.82–2.11; Table 4). Tachysystole occurred with the drug in situ in 17 (14.4%) and 50 (32.8%) of MVI 100 and 200 participants, respectively (P<.001). Category II or III FHR patterns were encountered after tachysystole had occurred in nine (7.6%) and 26 (19.8%) women in the MVI 100 and 200 groups, respectively (P=.006).
Cesareans were performed in 28.1% of participants overall, with 31.4%, 30.4%, and 22.9% of individuals undergoing cesarean delivery in the MVI 100, 150, and 200 groups, respectively (P=.15, RR 0.73, 95% CI 0.48–1.10 for MVI 100 compared with MVI 200). Nonreassuring FHR was the most common reason for cesarean delivery (11.0% of participants overall) with arrested dilatation as the second most common reason (10.2%) (Table 5). Cesarean deliveries performed for nonreassuring FHR related to the study drug occurred in 2.7% of participants (1.7%, 2.4%, and 3.8% for the MVI 100,150, and 200, respectively). Uterine contractile events resulted in a cesarean delivery in one participants; this was an MVI 100 individual (0.8%) and was attributed to hyperstimulation syndrome. Seventeen participants who had the study drug removed after achieving onset of active labor ultimately went on to have a cesarean delivery (4.6%); these were distributed equally among the groups (6.0%, 3.2%, and 4.6% for MVI 100, 150, and 200, respectively). In contrast, having the inserts in situ for more than 24 hours led to cesarean delivery in a disproportionate percentage of cases, with cesarean deliveries in 13 (11.1%) and eight (6.4%) participants exposed to MVI 100 and MVI 150, respectively, compared with three (2.3%) of the individuals treated with the MVI 200.
No maternal or neonatal deaths occurred during the study and no adverse events led to premature withdrawal of any participants. Systemic adverse events such as nausea, vomiting, and diarrhea in mother or neonate occurred with less than 1% incidence in any treatment group (data not shown). The mean birth weights and Apgar scores were similar among the groups. There were no differences in other neonatal outcomes between groups (Table 6).
With induction of labor becoming increasingly more common, it is important to identify an agent that can safely and effectively ripen cervices and induce labor. Ideally, a cervical ripening and labor-inducing agent would mimic the natural cascade of events of human parturition and reduce overall time to delivery without increasing the likelihood of cesarean. However, induction of labor has been associated with increased rates of cesarean deliveries by some,9 although the increase is not universally observed.10 It is important that any induction method demonstrate no increase in this important safety parameter.
We demonstrated that the MVI 200 significantly reduced time to any delivery mode compared with the MVI 100. The MVI 200 also significantly reduced the time to vaginal delivery. Use of the MVI 200 resulted in significantly more vaginal deliveries within 12 hours (P=.02) compared with the MVI 100 dose reservoir. A total of 76.0% of MVI 200–treated women delivered vaginally by 24 hours compared with 63.8% of MVI 100–treated women (P=.057). Use of the MVI 200 also reduced the necessity for oxytocin augmentation of labor. These results were associated with a nonsignificant reduced RR of cesarean delivery for women treated with MVI 200 compared with MVI 100 patients.
Removal of the study drug insert owing to adverse event was more frequent for the MVI 200 and MVI 150 groups compared with the MVI 100 group. Invariably, we found these adverse events leading to removal were almost always associated with retrospectively diagnosed tachysystole. We found that this method often led to observations that were discrepant from the bedside assessment, and resulted in tabulation of adverse events that were often well tolerated and did not result in discontinuation of the study drug. However, cesarean deliveries attributable to nonreassuring FHR occurred at a higher rate with MVI 200 than with MVI 100 (Table 5).
We acknowledge the limitations of this randomized controlled trial, which include a potential lack of generalizability due to restrictions imposed by our inclusion criteria, and a limited sample size by which to be able to detect rare events in laboring mothers or neonates.
All three dose reservoirs tested in this study were well tolerated. There was a dose-dependent increase in the number of participants with treatment-related adverse events with the most reported in the MVI 200 treatment group. These were mostly uterine contractile abnormalities that were retrospectively identified. Although no assessment was made of contraction duration, intensity, or quality,8 these findings provide reassurance that the misoprostol vaginal insert did not appear to potentiate the effects of oxytocin, in contrast to use of some other labor induction agents.11–13
The results of this phase II investigation are comparable with results of Miso-Obs-004, a phase III investigation5 insofar as the 118 women treated with the MVI 100 in the phase II investigation had similar efficacy outcomes to the 428 MVI 100–treated women and the 436 Cervidil–treated women in the phase III study.
In summary, there was a dose response in favor of the efficacy of the MVI 200 among the three treatment arms. Mothers treated with MVI 200 went into labor sooner and delivered more rapidly, all with less use of oxytocin. However, this was accompanied by a substantively higher rate of uterine tachysystole and more cesarean deliveries for nonreassuring FHR patterns than the MVI 100, although the overall rate of cesarean delivery was lowest for women exposed to the MVI 200. The safety of MVI 200 is of critical importance, and the balance between safety and efficacy can be determined only with additional large-scale investigations in various clinical settings. Further development is planned.