Preventing Preterm Birth

Progesterone use fell out of favor, however, after studies linked diethylstilbestrol (DES) to uterine malformations and cervical cancer in the offspring of treated women. Even though progesterone's actions differ from those of estrogen, hormones in general were deemed to be worrisome.

The net result of this brief period of progesterone use, however, was a series of observational studies tracking the outcomes and health of individuals who were treated in the late 1970s and early 1980s as fetuses.

Although they were not rigorously scientific, the studies provided reassuring findings about the long-term safety of progesterone, as discussed in a thorough review of 17P by Dr. Paul Meis (Obstet. Gynecol. 2005;105:1128–35).

In 1990, Dr. Marc Keirse revived the idea that progesterone could be effective in protecting against preterm birth with a meta-analysis of trials employing 17P. He found “no support for the view that 17 alpha-hydroxyprogesterone caproate protects against miscarriage,” but said that trials did collectively “suggest that that [the therapy] does reduce the occurrence of preterm birth.” (Br. J. Obstet. Gynaecol. 1990;97:149–54).

His review prompted investigators to start looking at progesterone again. Perhaps it had been discarded a little too early, they thought.

More Recent Studies

The study that eventually led to the reintroduction of progesterone for the prevention of spontaneous preterm birth—and the study that led to the cautious endorsement of progesterone therapy by ACOG—was a much larger, randomized, double-blind trial conducted by the Maternal-Fetal Medicine Units (MFMU) Network of the National Institute of Child Health and Human Development, and published in the New England Journal of Medicine in 2003.

Investigators enrolled women at 19 clinical centers who had had at least one previous spontaneous preterm delivery and randomized them, using a 2:1 ratio, to receive weekly injections of 17P (250 mg) or a placebo. Treatment was begun between 16 and 20 weeks and was continued until either delivery or 37 weeks' gestation. (N. Engl. J. Med. 2003;348:2379–85).

The study planned to enroll 500 women, but enrollment was stopped at approximately 450 women by an independent data-monitoring panel when data showed that the rate of preterm delivery (defined as fewer than 37 weeks' gestation) was almost 55% in the one group but just over 36% in the other group. When the data were unmasked, the treatment group was found to have the lower rate of recurrent preterm birth.

This reduction in preterm birth of about one-third took members of the MFMU Network by surprise. When Dr. Meis, the lead author on this paper, first proposed in the mid-1990s that the network study 17P, network investigators—myself included—expected to find minimal, if any, benefit of 17P prophylaxis.

The network, in existence since the late 1980s, is known for conducting unbiased research on interventions that are part of routine obstetric care but are not yet backed by rigorous study. Rarely had the network published a study showing benefit for interventions or processes that many had hoped would be proven beneficial. (For this reason, study designers did not incorporate a full array of outcomes measures—particularly long-term outcomes measures—in the study of 17P.)

Also surprising was a secondary finding that progesterone was more effective in preventing premature births in women whose previous premature delivery occurred earlier than 32 weeks' gestation than in women whose previous delivery occurred later.

Our original theory was based on the expectation that progesterone would act primarily as a uterine relaxant, so—if progesterone worked—we thought it would likely be most effective in women whose previous premature delivery occurred later in gestation (after 32–34 weeks). To our surprise, the network study found that the earlier the previous preterm birth occurred, the more likely progesterone would be to prevent another preterm birth.

Some criticized the study, usually for reasons related to misunderstandings of the rigorous rules under which the MFMU Network operates. The study was stopped and restarted, for instance, because the initial supplier of 17P could not guarantee its purity and therapeutic effect. This required us to set these data aside and start all over again with study medication provided by a new supplier whose product was consistently prepared.

Critics have also said that the high rates of preterm birth raise suspicions about the study's design. What has not been appreciated, however, is that the mean gestational age of previous preterm deliveries was 31 weeks in both the treatment and placebo groups. Moreover, a significant portion (30%–40%) of women in both groups had more than one previous preterm birth, and almost 60% of the women enrolled were black.

The rate of premature birth in the placebo group was thus exactly what one would expect to see in such a study population, based on existing epidemiologic literature. These women were exactly the kind of patients one would expect to sign up for a research trial: women whose index pregnancy resulted in an early preterm birth, an experience they were anxious to avoid.