NEW YORK – Strata Oncology is developing a test that can predict whether cancer patients will respond to immunotherapies based on their tumor mutation burden and PD-L1 and PD-L2 expression levels.
The company shared initial research at the American Society of Clinical Oncology's virtual annual meeting last month, which showed that the investigational test may predict patients' responses to the checkpoint inhibitor pembrolizumab (Merck's Keytruda). With that research under its belt, Strata is now validating the findings and the test's capabilities in further studies.
"Figuring out which patients should get checkpoint inhibitors is arguably the most pressing question in precision oncology right now to really maximize the benefits for patients with cancer," said Scott Tomlins, chief medical officer at Strata. "There's not a single diagnostic approach right now that leverages the expression of PD-L1, or other markers of sensitivity, with TMB. Those are two different modalities."
Typically, oncologists use next-generation sequencing to gauge TMB and immunohistochemistry to assess PD-L1 expression in patients' tumors. For example, Foundation Medicine's FoundationOne CDx test uses NGS to identify genomic alternations in hundreds of genes and reports a TMB score. As part of its testing service, Foundation also offers oncologists IHC-based PD-L1 testing of patients' tumor samples as an add-on analysis.
Recognizing that it may be more efficient if oncologists could have a tool that integrates patients' PD-L1 and TMB status when making immunotherapy decisions for them, Strata explored if combined comprehensive genomic and transcriptomic profiling using its NGS assay and an investigational multiplex RNA-seq test could do the job.
The Ann Arbor, Michigan-headquartered firm already has the capability to measure TMB with its StrataNGS assay, which detects alterations in more than 400 genes but does not assess PD-L1 expression. The integrated test in this study combined NGS with a multiplex RNA-seq tissue test to measure both TMB and PD-L1/PD-L2 gene expression in a single workflow.
"The challenge is we need an assay that can do TMB, so it can't be a small NGS panel. It has to be a large enough panel to be able to give you comprehensive genomic profiling," Tomlins said. "Then, we also need to be able to do quantitative gene expression to accurately quantify expression of a single transcriptome like PD-L1 or PD-L2."
Within the study presented at ASCO, Strata wanted to see if its combined assay measuring TMB, PD-L1, and PD-L2 biomarkers could identify patients who will benefit from treatment with pembrolizumab, and potentially, other PD-1 inhibitors. This study did not compare the investigational test to IHC and NGS-based comprehensive genomic profiling tests commonly used to measure PD-L1 expression and TMB, respectively. However, Tomlins said that the study showed that the combination of gene expression profiling and NGS was able to identify TMB, PD-L1, and PD-L2 as independent predictors of treatment benefit.
Working with more than 25 cancer centers for the observational study, Strata collected surplus tumor tissue samples from 610 patients with 24 types of solid tumors who received pembrolizumab either on- or off-label at these centers and collected real-world data on their treatment outcomes. The study also included data from 345 patients who received chemotherapy.
Researchers conducted comprehensive genomic and transcriptomic profiling using an investigational, multiplex RNA-seq assay and StrataNGS on the tumor samples in Strata's CAP-accredited and CLIA-certified laboratory. They analyzed TMB levels and expression of nine biomarkers, including PD-L1, PD-L2, TOP2A, GZMA, PDCD1, and others, to identify those associated with increased time to next treatment, or TTNT. They found that PD-L1, PD-L2, and TMB were the best predictors of patients' TTNT.
Next, researchers developed a multivariate model that could predict TTNT based on patients' TMB, and PD-L1 and PD-L2 expression. The model also factored in gene expression from T cells and T-cell proliferation.
Currently, TMB and PD-L1 expression is used to predict chemotherapy benefit based on cutoffs that separate patients into biomarker high and low groups. However, within the multivariate model, such cutoffs were unnecessary, Tomlins explained, since patients' TMB and gene expression results were integrated into a single score.
Based on that model, patients were assigned an immunoreactivity score, or IRS, and stratified into four groups based on their IRS. The researchers then considered how these scores tracked with their TTNT, which was defined as the time between when patients started their first therapy and when they got on a new therapy or died.
Patients in IRS groups 1 and 2 had the shortest TTNT on pembrolizumab, while IRS groups 3 and 4 had the longest TTNT. In IRS groups 1 and 2, patients took pembrolizumab for a median of seven and 8.7 months, respectively, before progressing and going to the next treatment. In IRS groups 3 and 4, patients were on pembrolizumab for 13 months and more than 24 months, respectively, before receiving their next treatment.
Comparatively, for patients on chemotherapy, the IRS grouping didn't appear to be similarly associated with TTNT, with patients in groups 1, 2, and 3 having a median TTNT of seven months and patients in group 4 having a median TTNT of six months.
Additionally, about 12 percent of patients in IRS groups 3 and 4, who had the best responses to pembrolizumab, were receiving the drug off-label, either outside of approved tumor types or with TMB-low status. This showed that the investigational test can identify patients who could benefit from treatment with pembrolizumab outside of approved indications, the authors noted.
"The major biological findings from our study is that all three of those [biomarkers] are independent predictors of pembrolizumab benefit," Tomlins said. "We showed we can do both the comprehensive genomic profiling required for TMB, as well as clinically relevant quantitative gene-expression profiling, then leverage those [results] along with the treatment data that we captured in the trial to develop a single integrated test."
In practice, Strata hopes that by assessing PD-L1, PD-L2, and TMB in one workflow, physicians who are considering giving patients PD-1 inhibitors off-label will have more information in one tool to help them decide if their patients are likely to respond to immunotherapy, Tomlins said. Currently, using just TMB or PD-L1 status doesn't sufficiently identify all the patients who can derive benefit from immunotherapy. "In basically every study of checkpoint inhibitors to date, there's been patients in either the PD-L1 IHC-low group or the TMB-low group who do respond," he added.
On the heels of the study presented at ASCO, Strata is planning a prospective, blinded trial to validate the earlier findings before considering commercializing the test. This second study will recruit a similar number of patients with solid tumors who have been treated with pembrolizumab, Tomlins said. The company expects to complete this validation study in Q3 2021 but did not share further details about the study design. Strata is also exploring further retrospective and prospective validation studies with the investigational test in patients treated with pembrolizumab or other PD-1 antibodies, the company said.
The integrated test has only been studied in combination with pembrolizumab, but Tomlins believes it could pair with any PD-1 inhibitor. Strata is exploring potential companion diagnostic partnerships with drug companies interested in using a test that evaluates all three biomarkers to identify best responders to their PD-1 inhibitors, though the company didn't provide further details.
Ideally, Strata hopes that pharmaceutical companies will use the investigational test to facilitate a tumor-agnostic indication for their immunotherapies, like pembrolizumab's approval for TMB-high solid tumors last year. If future validation studies are successful, then the company will likely pursue regulatory approval for the test, but it did not provide a timeline for that process.
If successful, Strata believes that this integrated test would provide oncologists certain benefits over current methods gauging PD-L1 and TMB using different technologies. For example, a single-workflow test that works on small formalin-fixed, paraffin-embedded tissue samples may make it easier to incorporate TMB, PD-L1, and PD-L2 screening for patients in the real world, Tomlins said.
The study itself was performed using real-world data, involving patients treated in the community, which gives Tomlins confidence of the test's utility outside of the research setting. He highlighted that the study included patients with a variety of tumor types who were receiving pembrolizumab both on-label and off-label, which Tomlins noted, "represents the full range of tumors that are being seen and how clinicians are prescribing checkpoint inhibitors."
Currently, there is growing demand for immunotherapy, but it is well known that only a minority of patients respond. The study presented at ASCO showed that Strata's investigational test was able to identify not just patients who responded to pembrolizumab but also those who had "such a long response that you really want to be able to extend that [to more patients]," Tomlins said. "It's really important to try to identify all of those patients who benefit, and to do that as broadly across tumor types as possible."