Researchers from the University of California San Diego School of Medicine say a new method can better predict the quality of embryos used for in vitro fertilization, potentially raising the odds of a successful pregnancy for those relying on assisted reproductive technology.

For people struggling with infertility, IVF — which involves retrieving eggs and fertilizing them in the lab before implanting embryos in the uterus — often provides hope. But the process can be draining, at times requiring people to undergo multiple rounds of treatment.

“This is often a really trying experience, where we often say: ‘We don’t know why an embryo is good. We don’t know why an embryo doesn’t result in a baby.’ And we can’t know why until we better understand what’s happening in the embryo,” said H. Irene Su, one of the senior scientists involved in the study, who works as a professor at UC San Diego School of Medicine and as a reproductive endocrinologist and epidemiologist at UC San Diego Health.

A woman under age 35 has a 51% chance of having a live birth after one IVF cycle, with live birth rates dropping to 38% for women between ages 35-37 and 25% for women between 38-40, according to a 2019 report from the Society for Assisted Reproductive Technology. Clinicians wish success rates were higher, but one issue is that it’s difficult for doctors to know which lab-grown embryos will result in a successful pregnancy. That can mean multiple rounds of treatment, which get expensive at an average $12,400 per cycle.

The new method developed by Su and her colleagues seeks to assess embryo quality by looking at pieces of genetic material called extracellular RNA (exRNA) left behind in the liquid that embryos are grown in. Their findings were published last week in Cell Genomics. With this non-invasive approach, Su hopes that one day couples receiving assisted reproductive technology will have better outcomes.

Currently, embryo quality is assessed based on morphology (examining their physical characteristics) or by taking a biopsy of the cells. “Both morphology and chromosomal testing from a biopsy are limited in terms of predicting which embryos will become a healthy baby,” Su said.

The new technique could be meaningful for couples going through IVF in a number of ways, according to Victoria Jiang, a reproductive endocrinology and infertility fellow at Massachusetts General Hospital who was not involved with the study. 

Replacing biopsies with the exRNA technique “could be an incredible advancement as it is less invasive, less expensive, and less dependent on operator technical skill” compared to biopsies, Jiang said via email. That “can ultimately improve outcomes, both from a financial perspective and a pregnancy outcomes perspective,” Jiang said. Genetic testing for embryos can cost up to $5,000, according to Jiang.

The new method leaves the embryo completely undisturbed and looks at the culture medium left behind. When culturing a fertilized egg into an embryo, it is grown in a very small droplet of culture media, which helps the embryo develop before it is used for implantation. In this study, the scientists took the leftover liquid media, which was donated by 29 participants. Then, researchers isolated exRNA, converted it, and then mapped it to the human genome.

The researchers analyzed a total of 295 droplets, and created an atlas based on the type of exRNA they found. Su and her colleagues were able to characterize the RNA into five distinct genetic profiles or time periods of embryonic development: the oocyte or egg, zygote or fertilized egg, cleavage, morula and blastocyst stages. By identifying the different stages in each droplet, the researchers were able to track which embryos grew to the blastocyst stage, which is when they are ready for implantation, and which embryos were “arrested” or just stopped growing.

If this form of non-invasive testing “can lead to easier identification of those arrested embryos, then it will significantly help with selection, particularly when the morphology is the same,” said Jiang.

Su and her colleagues also identified certain genes in the embryos that stopped growing, but more research needs to be done to figure out those genes’ specific role in the process.

Putting these elements together, the researchers used the exRNA profiles to train a machine-learning model to identify higher-quality embryos. The researchers first graded 26 embryos and gave each a morphology-based score. Then they put them through the machine learning model and found that the predicted score correlated with the morphology-based score the embryologists came up with.

The researchers who spoke with STAT all agreed that this doesn’t mean that embryologists will be going away any time soon. But it could mean that one day, clinicians will be less reliant on biopsies.

“Most of these advances, even in the machine learning/AI realm, will be helpful adjunctive tools to streamline the work flow of embryologists, and help with the shortage of embryologists that our field is facing, but will not be replacing embryologists by any means in the near or distant future,” Jiang said.

Throughout this study, the researchers encountered some challenges that limited what data could be collected. They decided to use the leftover media from a group of embryos since embryos grow better when cultured together. As a result, it was also difficult to compare results from genetic testing of embryos, since embryos are usually individually tested and with the shared liquid researchers couldn’t tell from which embryo the exRNA was coming from.

“This model may help identify genes that are valuable in development, but I would want to see the model applied to individual embryos and it is far from being a tool for embryo specific selection,” Jiang said.

Su told STAT that it will take additional research to confirm whether the test can be used to successfully predict positive IVF outcomes. She said the reproductive field may be able to learn other valuable lessons from the technique as well.

“Mostly until now, how we understand parts of early human embryo development is actually looking at, you know, other mammals,” Su said. “But if you don’t have to perturb a normal human embryo, and be able to look at the spent media to understand its physiology, that’s really cool.”