Premature birth remains a mystery. A drug that seemed to prevent it is now being reevaluated.

Eight years ago, the first drug specifically approved to prevent premature birth was hailed as a public health victory, a way to put a dent in the leading cause of newborn death and disability.

But that drug, a synthetic version of the hormone progesterone, has a long, strange history that just got even stranger.

Makena, a weekly shot, was no better than a placebo at preventing preterm births in AMAG Pharmaceuticals’ recently completed clinical trial – one that the U.S. Food and Drug Administration ordered as a condition of approval in 2011. So on Oct. 29, an FDA advisory panel will consider whether the agency should withdraw its blessing.

It won’t be an easy call. The Makena study enrolled women who were at much lower risk of preterm birth than the women in the government-funded study that led to approval.

“Despite having the same eligibility criteria, the [two studies] ended up having very dissimilar patient populations, which may explain the marked difference" in results, wrote the leaders of the Makena study, published Friday in the American Journal of Perinatology.

Meanwhile, the consumer advocacy group Public Citizen this month petitioned the FDA to withdraw approval of all medications containing the preemie-prevention drug — Makena, newer generics, and custom formulations made by compounding pharmacies.

“The drug never should have been approved in the first place because the data that were relied on to establish efficacy were seriously flawed,” the petition declares.

This latest twist in the 65-year-old story of hydroxyprogesterone caproate reflects how little is known about the causes of preterm labor and delivery, despite major advances in saving babies born too early.

“Isn’t that discouraging?” lamented Einstein Medical Center obstetrician-gynecologist Arnold W. Cohen, who has been dealing with the conundrum for almost 50 years.

Natural progesterone is needed to maintain a pregnancy, but it’s not clear how the synthetic version helps. Developed in the 1950s, it was first approved as Delalutin for purposes including preventing miscarriage. In 1999, the maker, Bristol-Myers-Squibb, voluntarily ended U.S. sales for business reasons.

But doctors could still get the drug from compounding pharmacies for $10 a dose, and it was also available in vaginal gels and suppositories.

» READ MORE: As more pregnant women use marijuana, a new study finds risk to babies

In 2003, a large, government-sponsored study at 19 medical centers — including Jefferson University Hospital in Philadelphia — renewed interest in the compound. Women carrying a single fetus who had a history of preterm birth — which increases the chance of a recurrence — were given a placebo or shots. The drug dramatically cut the chance of delivering before 37 weeks, the threshold for prematurity, as well as before 32 weeks, when newborns need intensive care. Babies of treated mothers also had lower rates of brain bleeding, devastating intestinal disease, and supplemental oxygen.

Based on that seminal, taxpayer-funded study, Makena got accelerated approval for use in such women. But as the Public Citizen petition points out, FDA advisers had some safety questions and found flaws in the study. For example, the Alabama hospital that enrolled the most women and had the best results may have biased the overall conclusion.

To confirm that Makena really was safe and effective, the FDA ordered the post-approval study.

But ob-gyn groups didn’t wait to embrace hydroxyprogesterone.

At the time, one in eight babies were born prematurely — one in six for black babies — and the rate had been creeping up since 1981, driven by older mothers, teen mothers, obesity, infertility treatment, poverty, and unknown factors. Doctors prescribed bed rest, home monitoring machines, antibiotics, labor-suppressing drugs, stitching the cervix shut. Still, half of women who went into labor too early delivered too early.

» READ MORE: Bed rest during pregnancy doesn't help. It hurts.

Despite an eager market, Makena’s 2011 launch sparked outrage. Then-owner KV Pharmaceuticals set an exorbitant price of $1,400 per dose while trying to squelch compounding pharmacies.

Cohen, at Einstein, was among specialists who wrote articles decrying the “price gouging” as “unethical.”

“Makena is down to $495 a dose now, but it’s still a pricey drug,” Cohen said, noting that Einstein doctors use a $20-per-dose compounded injection.

In May, AMAG reported that Makena’s first-quarter revenues had fallen by two-thirds because of new generic competition. Even so, Makena still brought in $31 million for those three months.

Now, some market analysts say Makena’s future is in doubt, given that it didn’t work in the confirmatory trial.

But medical experts say what that trial really shows is that even among women with a history of preterm birth, risk profiles vary. Less than half the women in the Makena trial had additional risk factors — such as smoking, being unmarried, being black, and having multiple previous preterm deliveries — compared to 91 percent of women in the 2003 government trial.

The U.S. preterm birth rate has fallen to one in 10 babies, and many experts say hydroxyprogesterone has played at least a small role in the improvement.

“I think vaginal progesterone has made a difference,” said Vincenzo Berghella, director of maternal-fetal medicine at Jefferson.

But studies of the compound — there have been many since 2003 — have added to the uncertainty. Consider that the American College of Obstetricians and Gynecologists currently recommends the drug, but doesn’t specify shots or vaginal forms. The Society of Maternal-Fetal Medicine, in contrast, says shots are superior and vaginal forms should be reserved for women who have a shortened cervix, measured by ultrasound.

Berghella, who helped develop the society’s guidelines, prescribes vaginal gel exclusively, because “the vast majority of studies" show it is better.

Mary Norton, director of maternal-fetal medicine at the University of California, San Francisco, said, “It’s like so many things. People can look at the same evidence and draw different conclusions, especially because preterm labor is complicated.”

“To be honest,” she added, “we don’t have a good grasp on exactly what is happening. We don’t know how to tell which women are going to respond to which treatments.”