BRCA1 is one of the best-studied cancer genes in the world. Still, on occasion, doctors will test a patient and find a BRCA1 mutation no one has ever seen before. This creates a dilemma. The newly discovered “variant of unknown significance,” or VUS, could be harmful, making a woman’s risk of developing breast cancer as high as 72 percent. Or it could be nothing to worry about at all.
“You hear these horror stories about people who have a VUS,” says Fergus Couch, a breast-cancer researcher at Mayo Clinic. “They have the surgery”—to preemptively remove their breasts—“because they’re convinced the VUS is actually pathogenic, but then we find out later that it is neutral.” The invasive surgery was never necessary at all. The opposite can also be true: A variant that looks neutral could turn out to be pathogenic.
For this reason, researchers have spent the decades since BRCA1’s discovery trying to catalog its many variants. BRCA1 is a tumor-suppressing gene, and only certain mutated forms are associated with high cancer risk. Theoretically, there could be thousands of variants. And even once a new variant has been identified, it may be so rare that doctors never gather enough data in patients to understand its associated breast-cancer risk.
When Lea Starita was a graduate student, her advisor would talk about analyzing BRCA1 piece by piece—creating a blueprint for the effects of every possible mutation before it’s even seen in a patient. “I remember rolling my eyes,” says Starita. “I didn’t think it was possible.” At the time, it wasn’t: Scientists were generating mutations one by one and doing a set of experiments for each one.
Fifteen years later, Starita and her colleague Jay Shendure at the University of Washington and the Brotman Baty Institute for Precision Medicine have figured out a way to study BRCA1 at scale Starita never imagined. In a study published in Nature, they assess the cancer risk of a whopping 3,893 BRCA1 mutations. It’s made possible by CRISPR, a gene-editing tool that allowed them to engineer all 3,893 mutations essentially at once.
“This work was really tour de force,” says Susan Domchek, an oncologist at the University of Pennsylvania who was not involved in the study. It’s still too early to base patient decisions on the study alone, but it provides a road map for researching all the unknown variants of BRCA1 and other cancer genes.
To introduce different variants into BRCA1, Starita and Shendure ordered snippets of DNA—scientists can just do this online now—that corresponded with every possible single-letter mutation along approximately 1,300 letters of the gene. (Since DNA has four different bases—A, T, C, and G—there are three possible mutations at each location to give the 3,893 mutations reported in this study.) They then used CRISPR to paste those segments into the DNA of human cells.
But they didn’t use any ordinary human cell. They specifically found a type of human cell that dies when its copy of BRCA1 does not work. This was clever because they simply had to wait for the cells with different BRCA1 variants to grow, noting which variants allowed cells to thrive (meaning its BRCA1 was functional) and which caused them to die (non-functional). Since BRCA1 normally suppresses cancer, a nonfunctional BRCA1 mutation is likely to put human breast tissue at a high risk for cancer.
The team compared their results to a database of known BRCA1 variants to test their findings’ accuracy. Of the variants they engineered, 169 turned out to be known pathogenic variants: 162 were non-functional, two functional, and five somewhere in between. Of 22 known benign mutations, their test deemed 20 functional, one non-functional, and one in between. Not perfect, but close. The work, in total, took about six months.
In contrast, says Couch, who was not involved in the study, his frequent research collaborator Alvaro Monteiro at Moffitt Cancer Center has been analyzing mutations in BRCA1 one by one in cells growing in a lab. Over 15 years, he’s looked at 300 to 350 variants. His analysis has been more carefully validated against actual patient data, so it carries a little more heft.
In general, doctors are hesitant to base patient decisions only on data from cells growing in a lab. “There’s a degree of caution,” says Couch. But for some rare VUS, that may be the best they have. And large-scale methods like those described in this Nature paper could start generating a lot more data a lot more quickly.
Starita and Shendure aren’t done with BRCA1 yet. The entire coding sequence of the gene is about 5,600 letters long, and this study only covered about 1,300 of those. “This is an initial stab at this,” says Shendure. “With further scaling, you can imagine much more ambitious efforts to engineer tens of thousands or hundreds of thousands of mutations into the genome.” And they’re looking to use the same technique to study other cancer genes such as BRCA2, PALB2, and BARD16.
Because BRCA1 is an especially well-studied cancer gene, fewer than five out of 100 patients will get variants of unknown significance when tested today. Many other cancer genes are even less well understood, though. And for those, this technique could rapidly fill in the gaps.