A new study of sick infants sees potential in DNA sequencing, but also a need for better ways to interpret the genome

There’s growing evidence that DNA sequencing can help diagnose the health care system’s youngest patients — babies in their first year of life. But a new report resurfaces a thorny challenge in researchers’ quest to turn long strings of A’s, T’s, G’s, and C’s into information doctors and patients can use: Reading the genome is one challenge, interpreting it is another.

The recent results come from the Genomic Medicine in Ill Infants and Newborns, or GEMINI, a prospective study in which 400 hospitalized infants had their entire genomes sequenced and were also tested with a targeted gene panel. Whole-genome sequencing, while slower, outperformed the targeted test in pinpointing the cause of an infant’s symptoms, with a 49% diagnosis rate compared to 27% for the narrower panel. For a fifth of participants, that led to changes in their treatment, such as doctors opting to use or avoid certain medicines and surgical procedures.

But the study also found that labs often interpreted genetic information differently, despite working with identical patient samples and medical records. In 43% of cases, labs differed in how they interpreted genetic variants, a discrepancy the authors say points to the need for more standardized ways to deduce which mutations could explain a patient’s symptoms.

The findings, published on Tuesday in the Journal of the American Medical Association, come at a time when the cost of genetic sequencing is plummeting. But two decades after the initial completion of the Human Genome Project, the potential of DNA sequencing to revolutionize health care remains largely untapped amid open questions about the total cost and usefulness of these tools. The study’s authors say their findings add to growing evidence that sequencing has a role to play in diagnosing and treating seriously ill infants.

“This is yet another study pointing to the effectiveness of genomic sequencing when used in the right population,” said Jill Maron, a pediatrician at Women & Infants Hospital of Rhode Island and the study’s first author. “Any test that can diagnose over 50% of babies is huge.”

The GEMINI study, which ran from 2019 to 2021 and was led by Tufts Medical Center, brought together six U.S. hospitals across the country. The study broadly admitted babies suspected of having a disorder, often because of serious and otherwise unexplained symptoms such as bleeding or seizures. Roughly 58% of the infants in the study were male, 31% were Hispanic, 12% were Black, and 6% were Asian.

Doctors collected teeny amounts of blood from the infants — less than a teaspoon — that were sent to either Rady Children’s Hospital in San Diego, where infants had their whole genome sequenced, or Athena Diagnostics, a diagnostic testing business owned by Quest Diagnostics where samples were tested on a panel of more than 1,700 genes.

While it typically took around six days to return whole-genome results compared to four days for the panel, the former more routinely diagnosed the infants. That’s largely because scanning the billions of chemical letters in a person’s genome allowed researchers to analyze genes not included in the targeted panel and to look at larger-scale changes in which a segment of DNA gets copied, flipped, or moved around.

“I can tell you even as a practicing clinician, what isn’t helpful is when I do [gene] panel after panel after panel after panel, because insurers will pay for a panel,” Maron said. “If you’ve done four panels and what you really need to do the whole time was just a whole genome, you’ve actually wasted money.”

In many cases, sequencing diagnosed diseases that were otherwise not on a physician’s radar. Only a third of pretest reports from geneticists listed the disorder later identified by testing as a potential culprit.

And the benefits of sequencing went beyond an initial diagnosis, Maron said. For instance, in one case, researchers diagnosed a baby with hemophilia, but a look at the child’s genome also uncovered that the infant was prone to a potentially deadly reaction to certain forms of anesthesia. That led to a last-minute phone call while the infant was being wheeled into surgery, Maron said, to ensure doctors used different anesthetics. In other cases, sequencing infants actually helped diagnose parents who had suffered from that same disorder for years, an unexpected benefit noted in other studies, too.

Those are impressive benefits, said Jason Vassy, a researcher at Harvard Medical School not involved in this study but whose work lies at the intersection of primary care and genomics. The big question? Whether sequencing would produce the same benefits in other health systems.

“These are still academic medical centers. And so these are probably some pretty astute clinicians,” Vassy said. “In the hands of a real-world [neonatal intensive care unit], you might see a different kind of performance.”

Researchers also quickly realized that generating this genetic information was just part of the challenge. Scientists found more than 80 cases in which both labs (Rady Children’s and Athena Diagnostics) detected the same variant but interpreted it differently. That meant, for example, that one lab reported that a variant was disease-causing while the other dubbed it a variant of unknown significance, a catch-all term for genetic variation that researchers don’t fully understand.

In the current study, researchers erred on the side of caution and went with whichever lab reported greater concern. But the study’s authors say these discrepancies point to the need for better standardization in genome interpretation. Currently, scientists rely on algorithms that take into account a patient’s symptoms, their genetics, and what’s currently known about the function of genes to spit out ranked lists of variants suspected to cause disease. Those lists are then manually reviewed by geneticists who use their own experience to finalize a report.

“It’s not the sequencing that gives you the answer. It’s the interpretation,” Maron said. “You think, ‘Well, if you sequence it, you must know.’ And that’s not the case, and we have to be cognizant of that.”

The study’s authors are in the process of doing a follow-up analysis to test whether the benefits of whole-genome sequencing in GEMINI outweigh its costs. A prior study in California found that spending $2 million to sequence critically ill newborns essentially paid for itself, resulting in $2.5 million in savings on shorter hospital stays and other cost-saving measures.

Researchers are also hoping to diagnose the nearly 200 participants who didn’t get a clear answer for the cause of their disease. In some cases, that answer may not be genetic, but the authors are now planning a future study to see if there were some genetic disorders they missed initially. To do so, they plan to use long-read sequencing, a strategy that reads the genome in larger chunks and can decipher more complex mutations, and artificial intelligence to more efficiently scan medical records for symptom information that can help researchers make sense of genetic variants.

The latest study uncovered 134 genetic variants that have never been reported, information that will help researchers who encounter those same variants in the future better understand them. To Maron, it’s a clear sign that this field, much like the tiny patients who doctors and scientists are looking to help, is still in its infancy.