Choosing embryos for in vitro fertilization (IVF) is fraught with uncertainty. The process has only a 35 percent success rate, and transferring more than one embryo to increase the odds of successful implantation increases a woman’s chances of having multiple births or requiring a selective abortion.
By analyzing time-lapse videos of fertilized eggs as they develop, a team of researchers from Stanford University School of Medicine has identified three specific milestones that eggs must reach in order to form a blastocyst, a critical stage. The team created an algorithm for monitoring the embryos’ progress, and says this algorithm can predict which will develop to a blastocyst and which will die with 93 percent certainty. The technique could be used to enhance decision-making in the clinic, which currently relies on subjective evaluation of an embryo’s quality.
For reasons unknown, human embryos have a particularly high failure rate, and between half and more than two-thirds of IVF embryos fail to reach the blastocyst stage. Currently, embryos are usually chosen based on their visual appearance, which is checked at specific points in their development. Renee Reijo Pera, from the Institute for Stem Cell Biology and Regenerative Medicine at Stanford, and lead author of the new work, which is published in Nature Biotechnology, says that even though visual cues are used to evaluate embryos in the clinic, there was a paucity of imaging data about the complete course of their progress.
To better understand how embryos look over time, her team studied a set of 242 one-celled embryos that had been frozen 12 to 18 hours after fertilization. Embryos are now usually frozen after they have had a chance to develop for three to five days; thawing and observing embryos while still in the one-celled stage allowed the researchers to track their progress from the earliest stages of development.
The team took time-lapse images of the cells using dark-field microscopy, a technique that is useful for imaging live cells. First author Connie Wong led an analysis of the images to identify visual parameters that corresponded to survival and successful formation of a blastocyst at day five. Out of 10 parameters studied, the researchers found three of them to be critical to success. All of them related to timing: the final stage of cell division, in which the cell physically separates into two; the interval between the first and second cell division cycle; and the interval between the second and third cell division.
Kevin Loewke, a postdoctoral fellow now at Auxogyn in Menlo Park, California, developed a method of automatically measuring the three parameters using an image-analysis algorithm. Loewke says that the challenge in tracking the progress of the cells is converting two-dimensional microscopic images into information about the three-dimensional structure of the embryo, especially as multiple cells can block one another from view. His solution was to create a predictive model of each cell of the embryo and use the model to determine the timing of each milestone. The algorithm has been licensed by Auxogen, which plans to develop a product that would use the approach to determine embryo viability.
All of the parameters discovered by the team occurred before day two of the embryo’s development. Many clinics grow the embryo until it reaches the blastocyst stage at day five, but some scientists believe that spending too long in culture might raise the risk of altered gene expression in the infant. “The advantage is, if you can predict blastocyst formation by day two, you can get it out of the artificial environment sooner,” says Marcelle Cedars, director of the Division of Reproductive Endocrinology and Infertility at the University of California, San Francisco , who was not involved in the study.
But Cedars says that, “even in the youngest patients, 50 percent of embryos that make it to the blastocyst stage won’t implant.” The question is whether predicting the development into a blastocyst will also predict successful implantation. “It’s halfway there, but it’s not telling you if this is going to make a baby,” she says. Other methods of screening embryos include genetically analyzing a biopsy, or measuring molecular factors in the culture medium that surrounds the embryos. Cedars says that none of these have so far proven successful at boosting pregnancy rates.
As part of their study, the Stanford researchers also analyzed gene expression in the cells at different stages of embryonic development. The parameters they discovered all occurred at a time when the embryos had not yet activated their own genes, and were relying on genetic instruction from the egg. Cedars says this finding reinforces the idea that “the biggest driver of having a good embyro is the egg.”
The scientists also discovered that at these early stages of development, the cells in the embryo are acting autonomously, expressing genes at different times. This lack of coordination could help explain why some embryos fare so poorly.
DeepMind’s cofounder: Generative AI is just a phase. What’s next is interactive AI.
“This is a profound moment in the history of technology,” says Mustafa Suleyman.
What to know about this autumn’s covid vaccines
New variants will pose a challenge, but early signs suggest the shots will still boost antibody responses.
Human-plus-AI solutions mitigate security threats
With the right human oversight, emerging technologies like artificial intelligence can help keep business and customer data secure
Next slide, please: A brief history of the corporate presentation
From million-dollar slide shows to Steve Jobs’s introduction of the iPhone, a bit of show business never hurt plain old business.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.