Taking some of the doubt out of IVF
Complex algorithm predicts probability of in vitro fertilization success better than standard age model
Couples contemplating a second attempt at having a child through in vitro fertilization after failing the first time may now have a tool that takes some of the guesswork out of the decision to try again.
By incorporating dozens of factors pertaining to a couple’s fertility, age and health, a new algorithm more accurately predicts the probability that IVF will succeed compared with a currently used standard calculation, researchers report online July 19 in the Proceedings of the National Academy of Sciences.
“This is excellent scientific work,” says Andrew La Barbera, scientific director at the American Society for Reproductive Medicine. The scientists deployed “an unusual and in-depth statistical analysis of all these factors related to assisted reproductive technologies,” says La Barbera, a physiologist also affiliated with the University of Cincinnati and the University of Alabama at Birmingham.
In IVF, eggs are retrieved from a woman and fertilized in a lab dish with sperm from a male donor. If a healthy embryo develops, it is reinserted into the woman’s uterus. In the best-case scenario, this embryo attaches to the uterine wall and a normal pregnancy ensues. This happens only about 25 to 30 percent of the time.
What’s more, IVF can cost $7,000 to $15,000 per attempt and often is not covered by insurance. Thus, apart from emotional stress, a failed first attempt can leave couples with hard financial choices.
Aside from knowing that young women have better success rates at IVF than older ones do, researchers know very little about what factors increase a woman’s chances of becoming pregnant and giving birth through IVF. A predictive model that uses factors in addition to a woman’s age would provide many couples — or single women trying to conceive — with a better idea of the probability that a second try will succeed, says Mylene Yao, an obstetric gynecologist at Stanford University School of Medicine who coauthored the new study.
To develop such a predictive model, Yao and her colleagues used information from 1,676 IVF attempts, 29 percent of which resulted in a live birth. The researchers analyzed 52 variables that might affect IVF, including drugs given to boost fertility in the mother, the number of embryos with more than four cells reinserted, the woman’s age and weight, the age of the sperm provider and so on. The scientists then devised a predictive algorithm that weighed the effects of these factors on the chances of success.
The research team then applied this new algorithm to data from 230 other women who had already tried IVF once and failed. Results from a second round of IVF procedures in those women showed that the new algorithm predicted the outcome more accurately than did the standard model more than 99 percent of the time. In about 60 percent of cases, the probability of success generated by the new algorithm differed substantially from that of the standard model, Yao says. Because the technique is experimental, this analysis was completed after the second IVF round and was not used to predict outcomes for these 230 women in real time.
La Barbera says that while the finding constitutes a “definite improvement” in IVF prediction, it has limitations. For one, it can apply only to second attempts, since it relies on a history of the first IVF procedure to generate the data needed for the calculation. Also, it provides only a probability. Even with a high probability of success, the treatment could fail, he notes.
Yakoub Khalaf, an IVF physician at King’s College London, says that while the finding is “interesting” it leaves many questions unanswered. In particular, he says, it’s unclear whether this algorithm would be better than a simple clinical assessment using the woman’s age plus a few key facts obtainable from a first IVF attempt, such as the number of eggs retrieved, the number of embryos available for reinsertion and the woman’s previous history of live births.
Yao maintains that the algorithm could nevertheless help doctors in counseling a woman trying to decide on a second try at IVF. The researchers have cofounded a company to develop the technology.
“We want to make this accessible and affordable nationwide,” Yao says, but it will require regulatory clearance first.