The hidden determinant of IVF success
Constructing an embryo is delicate and artisanal. Can A.I. help make it standardized and accessible?
So much of medicine happens backstage.
During a blood test you feel the snap of the rubber tourniquet on your forearm. You see the rising dark red column fill the collection tube. You don’t feel or see what happens next. Your blood is taken behind the scenes to a lab, time passes, and then, suddenly, mysteriously, you have a result.
There is a particularly unnerving version of this experience at the fertility clinic. A sample from your body is visibly taken and then whisked, out of view, to a lab. But this time, the sample is your gametes and the eventual result, you hope, is a healthy baby.
IVF outcomes vary tremendously, depending on which clinic you go to, and couples in need of fertility care will often shop extensively for the doctor they believe can uniquely help them achieve their dreams. Sponsors of fertility benefits are similarly discerning. Before establishing a contract, they will review dozens of metrics to ensure the fertility doctors in their network are well-credentialed and competent.
But few intended parents or benefits managers ever meet the people who may be even more important to the odds of success. Backstage, in the lab, embryologists are the ones who track, evaluate, handle and combine gametes to make embryos. Their jobs are highly technical. Their work environments are complex and ever-changing.
Tipping the scales
My friend Paxton Maeder-York is a data-minded engineer who founded Alife Health, a company that builds software to optimize care in both the fertility clinic and the embryology lab. I asked him for a bird’s eye view of how much the embryology lab matters, compared to the fertility clinic.
“You can have a great doc and a poor embryology lab, and your success rates will likely be lower than other clinics. If you improve that embryology lab, you're probably going to have best in class success rates. We've seen this over and over again, and it’s why standardization is so critical for the field,” he told me.
Intriguingly, he added, “We've seen clinic groups become best in class in the country by really improving and investing in their embryology team processes.”
He told me to imagine fertility care as a restaurant. The clinic is like the front dining room and the embryology lab is like the back kitchen.
To understand what that means on the ground, I spoke with Dr. Jason Swain, the at once sage and plain-spoken Chief Laboratory Officer of CCRM Fertility. He told me to imagine fertility care as a restaurant. The clinic is like the front dining room and the embryology lab is like the back kitchen.
What makes running a restaurant really hard is that both sides of the house must come together to achieve success. You can have the best ingredients, the best menu, the best staff—but without systems to ensure the right things happen in the right way and at the right time, everything can fall apart.
In the embryology lab, every detail, from selecting the sperm and eggs to preparing the in vitro fertilization dishes, must be done correctly, all without breaking the “chain of custody”— whose gametes and embryos belong to whom.
Dr. Swain’s descriptions remind me of the necessary precision of an operating room, where checklists have been shown to be essential to reducing surgical complications. Much like the operating room, embryology labs use a specialized checklist called a “time out.” At key steps of the IVF process, the embryologists and technicians pause and verify that all identifiers match. Only after a successful time out will any material be moved from one dish or tube to another.
Defining excellence
To be a great embryologist, you must combine an awe for the science of reproduction with exceptional motivation for operational excellence. Unsurprisingly, great embryologists are in high demand—and short supply.
One embryology lab director recently estimated that out of 420 clinics, there were 400 open postings for embryology jobs. For a clinic, finding new talent is competitive: most labs train their own embryologists on the job, or poach them from other labs already-trained.
More surprisingly, the United States lacks a standardized national credentialing system for embryologists. While European embryologists must pursue ESHRE certification, Americans have no equivalent. And few colleges today offer embryology programs. It’s hard to build a pipeline for talent without many signposts. How are you supposed to know what an embryologist does, let alone why you’d like to be one?
Two years ago, Pinnacle Fertility, a rapidly growing fertility clinic chain, decided to keep up with demand for embryologists by starting their own embryology school. To stand it up, they recruited Dr. John Zhang who is among the most senior lab directors in the country.
A large part of training, Dr. Zhang says, is choreography—junior embryologists learn when to use their left hand, and when to use their right…all with the goal of never dropping a dish. The best embryologists must be communicative, calm, and competent. The daily work is both highly collaborative, and high pressure. Perhaps more than anything, though, Dr. Zhang says embryologists need to have integrity. When something goes wrong, they have to own it. After all, no one else can.
An embryologist “can’t be someone who is vibrating at a high frequency all the time,” he told me.
Not everyone is cut out for this kind of work. And by design, not everyone graduates from Pinnacle’s embryology institute. I asked Dr. Zhang how he can tell an embryologist that will drop someone’s eggs, from one that won’t. At a minimum, he emphasized a certain poise. Embryologists “can’t be someone who is vibrating at a high frequency all the time,” he told me.
Preventing errors
Even with a conceptual understanding, embryology felt abstract until I met Alease Daniel Barnes, a senior embryologist and embryology instructor from North Carolina who is on a mission to make her work more accessible. Her videos on Tik Tok have helped me, and millions of others, understand just how much finesse is required. In one video, which has received more than 10 million views, she demonstrates the process of injecting a single sperm into an egg using microscopic manipulators, pipettes and needles.
When Alease trains new embryologists she looks for good eyesight, hand-eye coordination, and fine motor skills. “If you can't stay in the lines in a paint by numbers with really small numbers, you're gonna struggle,” she told me.
The reason is that when things go awry, it can be disastrous. Look no further than recent news headlines littered with mislabeled embryos and unsafe protocols; the implantation of the wrong family’s embryos, or even the implantation of embryos known to be unviable.
In truth, these types of disasters are exceptionally rare. Dr. Denny Sakkas and his colleagues at Boston IVF found that just 0.23% of all 36,654 cycles they performed over the previous decade had a moderate or significant “non-conformance,” the technical term for a breach.
Still, averting disaster is a low bar for defining success. And while essential, adequate selection and training of embryologists only gets you so far. As Dr. Swain at CCRM described, the team must be set up for success with the right policies and procedures. They also need the right infrastructure.
Dr. Steven Katz is the founder of REI Protect—the only malpractice and risk management provider with a specific focus on fertility clinics. Dr. Katz effectively underwrites embryology, which means he’s in the business of touring labs to make IVF safer for patients and clinicians alike. In fact, he may have been inside more embryology labs than anyone else in the country and in his opinion, to get the best outcomes everything matters, from the HVAC systems that circulate air to the quality of the equipment.
Compounding the sense that embryology labs can be a black box, they are not obligated to follow common lab guidelines such as CLIA or CAP (though many do), and public reporting of outcomes to the Society for Assisted Reproductive Technology (SART) is also voluntary. Dr. Katz tells me that when labs make the decision to pursue these certifications and independently report their data it increases his confidence.
He has also developed several other rules of thumb. In making the decision to insure an IVF center, Dr. Katz uses CDC data to look at how many egg retrievals each embryologist is responsible for (no more than 150-180 per year). He looks at whether the lab director is onsite or remote (he insists they must be onsite if there are more than 600 retrievals per year at the facility). He looks at where the embryos are stored to ensure they are unlikely to be disturbed.
At Maven, we’ve started to align our own standards with Dr. Katz’s criteria to identify the best fertility clinics partners. Still, whereas Maven is positioned to ask the right questions, the average couple in need of care often isn’t. What if they didn’t have to be?
Creating the future
Today, constructing an embryo is a delicate and artisanal craft. I wanted to understand what it would take to make it more standardized and accessible. From a policy perspective, regulation may help–perhaps from lawmakers, and certainly within the profession itself (we wrote about that last time). There are also those who feel embryology is in need of a full-on, industrial revolution.
Paxton Maeder-York, founder of Alife, believes that revolution starts with data. From his perspective, with a large enough data set, clear patterns about how to best administer care emerge.
In describing the limitations of human beings looking through a microscope he tells me, “This is well-published: you can take the same embryo to five different embryologists, and you can get five different grades. They rarely agree on the top embryo transfer.”
Algorithms, however, may drive a more accurate consensus.
To cast light inside the black box, Alife’s digital visual capture tool also allows the embryologist to take images of embryos in a secure record, so that there's an audit trail of clean automated reports for every step of the IVF process. Patients know what their embryo looks like, and where it is in its journey back to them.
Serial entrepreneurs and technologists Joshua Abram and Alan Murray are taking a more even more radical approach to bring data, automation, and transparency to the lab.
When the federal government banned embryonic research during the Bush administration, Abram says IVF was “thrown out of the academic temple.” Without federal funding, innovation had to come from the private sector. In this way, it diverged from its natural sibling, cell therapy. Abram and Murray’s company, Conceivable Life Sciences, wants to reunite the field with an IVF lab that borrows from other areas of medicine that also depend on collecting, processing, and transplanting cells.
“IVF today is analog. There is no way to scale an operation that is hundreds of manual steps without making mistakes,” Abrams told me.
If today’s embryology labs are a black box, the labs of the future may quite literally be glass-walled.
Using computer vision and robotics in place of human eyes and hands, Abram and Murray say they’re optimizing every step of constructing an embryo, including the most basic. According to them, 20% of a patient’s eggs can be lost just in the typical process of removing them from the collection media. By using advanced microscopy, they say Conceivable’s automation finds every egg. And even by automating dish prep, a task often left to junior lab techs, researchers have found a nearly double digit improvement in blastocyst formation.
This fall, Conceivable’s Aura (the world's first fully automated IVF lab) will debut in Mexico City, where their clinical partners will offer patients a $10,000 “No Baby, No Fee'' guaranteed service model.
Abram and Murray envision clinic waiting rooms of the near-future that are as sleek and light-filled as an Apple showroom, where intended parents can look directly into the embryology lab to observe their robotics at work, handling and combing their genetic material. If today’s embryology labs are a black box, the labs of the future may quite literally be glass-walled.
I immediately wondered what the human embryologists would think of their idea. Alease Daniel Barnes told me she welcomes a future with more technology and automation in the lab. But she’s also observed the most masterful embryologists achieve consistently high pregnancy rates and isn’t sure the technology is quite there yet.
Like self-driving cars and chatbots, robotic embryologists will have to prove themselves trustworthy. “Can technology make it cheaper for the patient? Can it make it better? If yes, let’s do it,” Alease says. She left me thinking, perhaps the question should not be whether humans or computers are better at handling our embryos—but rather how both can be harnessed to achieve what neither can do alone.
What my team is reading, building, and thinking against:
In 1998, the Department of Health and Human Services held a hearing that indicated they would bring embryology into the fold of governance that all other clinical labs are held to. But then they changed their mind–for reasons that remain unclear. A few months ago, a great industry newsletter (and podcast) that we’d recommend, aptly named Inside Reproductive Health, interviewed some of the people who were there.
To underscore how little we currently know about embryos,
breaks down a provocative scientific study in which 50 women decided to transfer embryos that were previously condemned by testing. The outcome was surprising. 38% became pregnant and 16% went on to have a normal live birth, raising the question–how many perfectly good embryos get discarded because our testing is imprecise?
To better understand where A.I. may augment and even improve precision in embryology, read Alife’s research in Fertility & Sterility. They show in a double-blind and randomized study, machine learning models more often predicted embryos that would result in clinical pregnancy rates than those selected by embryologists alone.
If you’d like to get behind the scenes yourself and see what embryology in action looks like, follow Alease Daniel Barnes on Tik Tok. She is a wealth of embryo knowledge, optimism, and education.
Next week, I’ll be at the ABIM Foundation Forum and will speak on a panel about how we can build systems of medical trust online. Each year, the Forum convenes leaders to work through some of the gnarliest issues facing the medical profession. I always leave with lots to chew on.
Last year the Forum kicked off a recurring theme in the Preprint, which we explored in “Medical Truth and the Fraught Year Ahead.” The way we consume information is evolving, and therefore so has the need to understand and guard against misinformation. We have been increasingly looking to TikTok and Reddit as regular sources for the Preprint, and hope to share results of a peer-reviewed study we conducted soon examining social media trends in fertility.