NEW YORK – Strata Oncology published new data this week on the development and validation of a proprietary pan-solid tumor predictive biomarker for anti-PD-1/PD-L1 checkpoint inhibitor benefit.
Dubbed the Immunotherapy Response Score (IRS), the predictor combines gene expression of four genes with tumor mutational burden, potentially offering the opportunity to expand immunotherapy benefit to more patients than TMB alone, the study authors wrote in a paper published in Communications Medicine on Tuesday.
Strata Oncology developed the IRS using treatment and outcome data matched to clinical genomic and transcriptomic profiling results that it has been collecting under the ongoing, observational Strata Trial.
According to Strata CEO Dan Rhodes, the rationale behind exploring a combined gene-expression and tumor mutation burden (TMB) approach was that a significant number of patients with low TMB still respond to immunotherapy.
"These drugs work really well when mutation burden is high," he said. "That's not debatable. But when you dig into the tumor mutation burden papers from Merck … what you see is, there are also an equal number or more tumor mutation burden-low patients who benefit."
The company has been sequencing DNA and quantifying RNA in the Strata Trial for some time, so the idea of looking at this combination for potential improvement of immunotherapy benefit prediction made sense.
Rhodes said that his team was "pleasantly surprised" when its model identified a collection of biomarkers that went beyond TMB as the best factors to explain immunotherapy outcome.
In their development of IRS, Strata investigators used a training set of 648 pembrolizumab-treated patients, representing 26 solid tumor types, to train a predictor that ultimately combined TMB with quantitative gene expression analysis of PD-L1, PD-1, ADAM12, and TOP2A.
They then validated the locked algorithm in an independent set of 248 patients with 24 solid tumor types treated with other anti-PD-(L)1 drugs — all in a monotherapy setting.
In this validation cohort, patients with high IRS had significantly longer real-world progression-free survival compared to those with low scores (23.1 versus 10.2 months, respectively). Overall survival was also significantly longer in immunotherapy-treated patients if they were IRS-high, about 40 months compared to just 21 months in the IRS-low group.
"In those patients that we predict, it's not just that they're benefiting. A majority of them are having substantial real-world progression-free survival or overall survival benefit," said Scott Tomlins, Strata's chief medical officer.
According to the study authors, TMB alone was not a significant predictor of either measure in the cohort. Also, when Strata looked at IRS status across the whole observational trial cohort, the IRS-high population was nearly twice the size of the TMB-high population (21 versus 11 percent).
Although high IRS was more common in tumor types that are already known to benefit from immunotherapy, and for whom these drugs are frequently used already, it was also high in a subset of patients, 7.6 percent, who wouldn't currently be candidates (no tumor-specific approval and low-TMB status). This group would be expected to benefit from IO monotherapy, assuming the results seen in the study population held up.
Rhodes said that Strata recognizes the study was limited because it was retrospective. But he argued that the publication still represents a robust clinical validation of the biomarker in the pan-tumor setting.
He said Strata has additional retrospective studies ongoing and will continue to share additional evidence around the new biomarker throughout the year. "We think we hit the bar for clinical use, but we recognize that where folks draw that set that bar varies, and more evidence will only continue to build the case," he added.
Patients with late-stage cancer that have failed prior treatment and have limited options could benefit from the test, he said. "We see that in about 8 percent of patients outside of approved indications that you wouldn't have found with TMB, our test is going to be positive."
"You might say 8 percent is a smallish number, but when you multiply that by this population size, we're talking about tens of thousands of patients in the United States alone that our test would recommend immunotherapy [for]."
Benefit from immunotherapy is not usually measured in a few months, Tomlins added. "It's a profound benefit. So that's why that decision is so important."
Among other limitations, the study authors reported that the Strata trial population was biased toward tumor types for which PD-(L)1 therapy is already indicated. As such, the cohort was enriched for immunotherapy responders. But the fact that so many tumor types were represented in both the discovery and validation cohort was a boost, they argued, as was the fact that the magnitude of benefit in non-indicated patients in the trial was similar to what was seen in the TMB-high patients in Merck's pivotal trial of pembrolizumab.
Additional studies will be needed to establish the clinical utility of IRS in patients with conflicting biomarker results, for example IRS-high but TMB-low patients, or where both monotherapy and combination therapy are indicated, the authors added.
According to Rhodes, Strata also believes that the IRS score could have utility in the care of patients who already fit approved indications for immunotherapy, namely in guiding decisions around combination chemo-immunotherapy.
"In most cases now, immunotherapy in the first line is combined with chemotherapy," Rhodes said. "Studies have shown that it's additive, so why not combine them to get some of the benefit of both modalities across the population?"
But for patients who stand to have a profound benefit from immunotherapy alone, adding chemo arguably just adds toxicity. In their paper, the investigators performed an exploratory analysis of lung cancer patients in the cohort, demonstrating that IRS-high patients had similar outcomes on immunotherapy alone compared to combination chemo-immunotherapy.
"It's secondary, maybe, to the prospect of finding new, responsive patients, but for that first-line lung cancer patient who could be spared chemotherapy and have a great outcome on immunotherapy alone, we think that's a real benefit as well," Rhodes said. "And as we're talking to clinicians, they're gravitating to both of those use cases about equally in terms of how impactful they feel [this test] will be in their practice and for their patients."
Another exploratory analysis in the combined discovery and validation cohorts showed an ultra-low IRS subset with particularly poor PD-(L)1 response, suggesting the biomarker could also have utility in guiding patients away from immunotherapy when an alternative is available.
Rhodes added that Strata believes there are other therapeutic modalities where a combined genomic and gene-expression approach could yield new therapy response predictors.
"Obviously there would be a different set of genes and a different multivariate algorithm for different classes of drugs, but we believe that with classes of drugs like angiogenesis inhibitors, antibody-drug conjugates, [or] CDK inhibitors, a combination DNA/quantitative RNA approach is likely to bear fruit," he said.