NEW YORK – There has been a keen interest in developing pan-cancer or tissue-agnostic biomarkers for cancer treatments in recent years, with many experts anticipating the field to shift from organ-based cancer treatments to ones that are based on molecular alterations. But a new study by researchers at Memorial Sloan Kettering Cancer Center questions whether certain mutations always behave the same way across tissue types.
The 2017 expansion of the indication for pembrolizumab (Merck's Keytruda) for any microsatellite instability-high tumor regardless of its tissue of origin and the 2018 approval of larotrectinib (Loxo Oncology's Vitrakvi) for tumors with NTRK gene fusions by the US Food and Drug Administration were together seen as a signal that the field was shifting toward a tissue-agnostic treatment approach.
But there are questions about the assumptions behind such approaches.
"[A tissue-agnostic approach] is based on [the] presumption that tumors that have the same abnormality will respond in the same way to a given intervention," said Richard Schilsky, senior vice president and chief medical officer of the American Society of Clinical Oncology. "I think what is becoming clear is that that's not the case."
In their study, the MSK researchers, led by Barry Taylor, specifically examined the effect of BRCA mutations on homologous repair deficiency and response to PARP inhibitors in a range of cancers, not just the ones typically associated with BRCA mutations, to gauge whether they always acted the same.
"Is the impact of mutant BRCA the same no matter where it arises?" said Taylor, who is associate director of the Center for Molecular Oncology at MSK.
He noted that with increased clinical tumor sequencing, BRCA mutations have been found more often in cancers other than breast, ovarian, pancreatic, and prostate cancers where germline BRCA1 or BRCA2 mutations are known to increase disease risk.
This suggested to some researchers and clinicians that targeting those mutations even in those other tumor types could be an effective treatment.
BRCA1 and BRCA2 proteins are a key part of the homologous recombination process through which double-stranded DNA breaks are repaired, and mutations in the genes encoding those proteins can hamstring the repair process and lead to some types of cancer. A handful of PARP inhibitors have been approved in the US to treat advanced breast or ovarian cancer and a few other tumor types in patients with germline BRCA1 or BRCA2 mutations.
Their off-label use in other cancer types assumes BRCA1 and BRCA2 mutations are tumor-agnostic biomarkers, which may not be the case, as some previous work has hinted.
In 2015, for example, the University of Pennsylvania's Susan Domchek and her colleagues performed a basket trial of the PARP inhibitor olaparib (AstraZeneca's Lynparza) in patients with germline BRCA1 and BRCA2 mutations. The trial mainly focused on patients who had breast, ovarian, prostate, and pancreatic cancers and largely found some response to PARP inhibitor treatment in patients with those tumor types, as they reported in the Journal of Clinical Oncology at the time.
But the study also included some patients with other cancers that are not usually associated with BRCA1 or BRCA2 mutations, and the researchers found little indication those tumors responded to PARP inhibitor treatment.
"There's always been this concern that some people can have mutations and it may have nothing to do with their cancer," said Domchek, a medical oncologist at UPenn. "And if it's not the reason someone developed cancer, then it's unlikely that a PARP inhibitor will lead to a response."
In their recent study, Taylor and colleagues began to tease out whether BRCA1 and BRCA2 mutations were driver mutations in a range of tumor types. For their analysis, published earlier this month in Nature, they sequenced germline blood and matched tumor tissue samples from 17,152 cancer patients, some with BRCA-associated cancers and some with other types of cancer not typically associated with BRCA mutations.
Overall, 2.7 percent of patients had a germline pathogenic BRCA1 or BRCA2 allele and of the remaining 16,690 patients with wild-type germline BRCA1 or BRCA2 genes, 5.5 percent had a somatic BRCA1 or BRCA2 mutation. Most commonly, patients with germline pathogenic or somatic loss-of-function BRCA1 or BRCA2 mutations did have a BRCA-linked cancer type, like breast, ovarian, prostate, or pancreatic cancer. But some patients with BRCA1 or BRCA2 mutations in this cohort had other cancer types.
Taylor and his colleagues found that these BRCA1 or BRCA2 mutations had different effects on homologous recombination repair, based on the tumor type in which they were found. Cancers typically linked to BRCA1 or BRCA2 mutations that had those alterations were associated with high levels of homologous recombination deficiency. However, only a few other cancer types in which BRCA1 or BRCA2 mutations occurred also exhibited homologous recombination deficiency.
This suggested to them that in these other cancer types, mutations in BRCA1 or BRCA2 genes might be passenger mutations rather than drivers of the disease.
This further means that patients with such BRCA passenger mutations in non-BRCA-cancers types might not benefit from PARP inhibitor therapy, Taylor said.
In their paper, he and his colleagues reported that patients with BRCA1 or BRCA2 mutations in BRCA-linked cancers exhibited greater benefit from PARP inhibitor therapy than patients without those mutations. Meanwhile, patients with BRCA1 or BRCA2 mutations in non-BRCA-linked cancers did not show greater benefit from PARP inhibitor therapy, as compared to patients without those mutations.
This finding, Domchek said, is "striking," though she noted that the study was "a first pass." She added that the results should strike a cautious note to clinicians considering treating patients with BRCA1 or BRCA2 mutations in non-BRCA-linked cancers with a PARP inhibitor.
"I think that this really makes us say we're really not sure that this will help people," she said.
The MSK researchers also found that in many — though not all — cases, inherited germline and somatic BRCA1 or BRCA2 mutations in BRCA-associated cancers acted similarly. Both, for instance, had similar responses to PARP inhibitors.
That, too, could have consequences for treatment approaches. For instance, in breast cancer, Taylor noted that currently only inherited BRCA mutations, not somatic ones, are indicators for PARP inhibitor therapy.
"It was exciting to see that in these cancer types that are very clearly BRCA-dependent by every measure, somatic mutations seem to act pretty similarly to germline mutations," he said.
He cautioned, though, that that wasn't always the case. "We did see individual cancers that violated that pleasing rule where the somatic mutations seem to be the consequence, rather than the cause of tumorigenesis in that patient," he noted.
Still, these findings, Domchek said, seem to set up tiers of evidence for PARP inhibitor use in BRCA-mutated cancers. At the top, with the most evidence of benefit, would be germline mutations in canonical BRCA tumor types. Those would then be followed by somatic mutations in those typical BRCA cancer types where there is some evidence they may respond to treatment, and lastly would be BRCA alterations in the non-canonical tumor types where the response seems to be lower.
But as ASCO's Schilsky noted, more research is needed to guide clinical decisions. "I don't know that the data in this particular paper is strong enough yet," he said.
Taylor agreed, noting that the findings need to be validated prospectively in a clinical trial.
Some additional evidence may come from the Targeted Agent and Profiling Utilization Registry (TAPUR) study, for which Schilsky is the principal investigator. Part of that large basket trial is assessing whether patients with BRCA1 or BRCA2 mutations who are treated with the PARP inhibitor olaparib — even when their cancer is not one of the drug's indications — show signs of response to the drug. Currently, the trial includes patients with seven tumor types, all of which have germline BRCA mutations or somatic BRCA alterations detected in the tumor, who are being treated with olaparib.
So far, Schilsky said all those tumor types have exhibited some preliminary evidence of response following olaparib treatment. These tumor types, he noted, include not only breast, pancreatic, and prostate cancer, but also non-small cell lung cancer, gallbladder and biliary tree cancer, and uterine cancer — tumor types not always associated with BRCA mutations. He and his colleagues have recently expanded the number of patients in this part of the trial, with the hope of getting more definitive results.
"Over time, I think we will gather some information as to whether or not the olaparib sensitivity is the same or different in these various tumor types," he said, estimating that his team may have some answers in about a year.
Likewise, Taylor and his colleagues are pursing additional studies aimed at gathering additional evidence, including designing clinical trials to examine what their findings might mean for clinical PARP inhibitor use.
"We obviously don't want to deny any patient a therapy from which they might get benefit," Taylor said. "But I think these are the kinds of questions we need to drive back into the clinical trial setting to validate prospectively in ways that can drive changes to clinical practice."