NEW YORK (360Dx) – Biodesix last month presented data on a proteomic test for predicting lung cancer patients' response to nivolumab (marketed by Bristol-Myers Squibb as Opdivo), as the company continues its efforts in immune-oncology.
In the study, which the company presented at the European Society for Medical Oncology (ESMO) annual meeting in Munich, the test showed some ability to distinguish between non-small cell lung cancer (NSCLC) patients who responded to nivolumab treatment and those who did not. However, the test did not outperform commonly used prognostic measures like the Eastern Cooperative Oncology Group Performance Status.
With immunotherapies becoming more commonly used to treat NSCLC, Biodesix is looking to grow its NSCLC portfolio by adding a test that could help predict patients' responses to these agents.
While immunotherapies have achieved excellent results in some NSCLC patients, many patients do not benefit from the drugs and some experience severe side effects. Currently, tests measuring levels of the protein PD-L1 are the primary approach for assessing the likelihood a patient will respond to checkpoint inhibitor immunotherapies like nivolumab, with patients with higher expression levels more likely to receive benefit. However, many patients with high PD-L1 levels do not respond, while some patients with low PD-L1 levels do.
Given this state of affairs, identification of markers for better determining patients likely to respond to immunotherapy has became a major area of interest within the field.
Last year Biodesix pushed back its immunotherapy test development timeline. After saying in April that it planned to launch a test around the middle of 2017, the company shifted course, with Heinrich Röder, Biodesix's chief technology officer, saying in November that it had decided to take more time to better understand the clinical need and market for such a test.
Röder noted that one factor driving this decision was the rapidly changing immunotherapy landscape.
This week, Egbert Smit, professor of pulmonary medicine at the Netherlands Cancer Institute and first author on the ESMO study, noted that the test remains in its early stages with more data and validation needed.
In the ESMO study, Biodesix used its MALDI mass spec platform to profile the serum proteome of 116 second-line NSCLC patients, generating mass spectral profiles that it used to stratify patients into three groups based on their treatment outcomes — good, intermediate, or poor. Of the 116 patients, 41 (35 percent) were classified as poor, 43 (37 percent) were classified as intermediate, and 32 (28 percent) were classified as good.
The researchers then tested the classifier in two validation sets: one composed of 58 second-, third-, fourth-, and fifth-line NSCLC patients treated with nivolumab, and another composed of 75 second-line NSCLC patients treated with nivolumab. They also ran the test on a set of 68 second-line patients treated with the chemotherapeutic doxatacel.
In the first validation set, patients with a test result of good or intermediate had median overall survival (OS) of 12.6 months versus 4.8 months for patients testing poor. In the second validation set, median overall survival was 9.1 months in patients testing poor, while patients testing good or intermediate did not hit this endpoint during the study.
The researchers also looked at whether the test tracked with PD-L1 status, finding that it did not. Patients with low PD-L1 (under 1 percent) made up 59 percent of the poor group and 60 percent of the good and intermediate group, while patients with high PD-L1 (greater or equal to 1 percent), made up 41 percent of the poor group and 40 percent of the good and intermediate group.
This suggests the test provides information on patient response beyond that provided by PD-L1 expression.
A multivariate analysis found that the hazard ratio for death in the good-intermediate group versus the poor group was .56. However, that analysis also found that the hazard ratio for performance status of 2 or greater versus zero was 4.08, suggesting that this commonly used clinical measure was more effective at distinguishing between responders and non-responders than was the Biodesix test.
There is also the question of whether the test is predictive of a patient's response to nivolumab (and potentially other immunotherapies) specifically, or if it is more broadly a prognostic measure reflecting things like disease aggressiveness and the underlying health of the patient.
Biodesix has in recent years transitioned its lead product Veristrat from a test for predicting patient response to EGFR inhibitors to a prognostic assay as questions have arisen about its predictive utility and changes in drug indications have reduced the need for tests identifying likely responders to EGFR inhibitors among patients negative for EGRF mutations.
Last year, Wendy Blosser, then the chief commercial officer at Biodesix (now chief commercial officer at Animated Dynamics) said that patients scoring poor on Veristrat are "identified with an aggressive disease state that is associated with poor prognosis," and that the company sees "utility in the test for planning, monitoring, as well as for patient counseling."
Smit pointed to the fact that in the 68-patient cohort treated with doxatacel, the new test did not effectively distinguish between responders and non-responders, noting that this suggested it was predictive of response to nivolumab as opposed to being broadly prognostic. However, in the PROSE trial, one of the major studies supporting use of Veristrat to guide EGFR inhibitor therapy, Veristrat was similarly ineffective in distinguishing between good and poor responders to chemotherapy (pemetrexed or docetaxel).
Veristrat has also shown some ability to separate nivulomab responders from non-responders in NSCLC. In a 2017 study in the Journal of Thoracic Oncology, the company ran Veristrat and a new MALDI mass spec test on samples from 60 advanced NSCLC patients and found that both were able to distinguish between responders on non-responders to nivolumab based on overall survival. For the experimental test, patients testing poor had median OS of 5.5 months while median OS for patients testing good was not reached. For Veristrat, patients testing poor had median OS of 4.1 months, with median OS not reached for patients testing good.
Smit, who has no financial relationship with Biodesix, said that additional data is needed to determine what the test's clinical utility might be. In particular, he said, more data is needed on how the test performs in first-line patients, where immunotherapy is becoming more common and where he said the test could help identify patients who could be treated with immunotherapy alone rather than immunotherapy combined with chemotherapy.
In second line patients, such a test would be most useful for identifying patients in whom immunotherapy is unlikely to work, he said.
"I don't think [the test] is clinically applicable immediately," Smit added. "But I think it is worthwhile to put it through further clinical testing."