NEW YORK (GenomeWeb) – Biodesix presented two studies last weekend at the Society for Immunotherapy of Cancer annual meeting in National Harbor, Maryland detailing its work to develop proteomic tests for predicting patient response to immunotherapy.
The studies looked at the ability of MALDI mass spec-based proteomic signatures to identify melanoma patients likely to respond to Bristol-Myers Squibb's programmed death 1 (PD-1) checkpoint inhibitor nivolumab (Opdivo); to predict progression-free survival of melanoma patients after HD IL-2 therapy; and to identify patients who might benefit from combinations of immune-checkpoint inhibitors.
This last effort addresses one of the more urgent needs in immunotherapy treatment, Jeffrey Weber, deputy director of New York University's Perlmutter Cancer Center and collaborator with Biodesix on the studies, told GenomeWeb.
"The key thing is, who can you choose to have just one drug and who needs the combination [of multiple drugs], which has a lot of toxicity," Weber said. "I think that is the ultimate application of this kind of test. It would be hugely useful."
In that study, the researchers looked at a cohort of 119 patients with stage III or IV unresectable melanoma who had progressed on at least one prior therapy and had been treated with nivolumab but no other anti-PD-1 or PD-L1 drugs.
Using Bruker's MALDI-based discovery system, the researchers profiled the proteomes of these patients, developing a classifier that divided the cohort into two groups. The first group, consisting of 34 patients, responded significantly better to treatment, exhibiting a median progression-free survival of 16.6 months with 62 percent progression-free at six months and 67 percent alive at two years. The second group, which consisted of 85 patients, responded less well, with a median progression-free survival of 3.1 months, 33 percent progression-free at 6 months and 33 percent alive at two years.
They then applied the classifier to two validation cohorts. The first consisted of 30 patient with advanced unresectable melanoma treated with anti-PD-1 agents. In this case, the 13 patients assigned by the test to group one had significantly better overall survival, with 84 percent still alive at two years compared to 12 percent for the 17 patients assigned to group two.
The second validation cohort consisted of 25 patients with advanced melanoma who likewise had been treated with an anti-PD1 agent, primarily Merck's pembrolizumab (Keytruda). Here, too, the 11 patients assigned to group one had improved overall survival, with 100 percent still alive at one year compared to 58 percent for the 14 patients in group two.
The researchers then looked at an investigational cohort consisting of 21 patients with advanced unresectable melanoma who had been treated with a combination of nivolumab and Bristol-Myers Squibb's ipilimumab (Yervoy). The study found no statistically significant difference in overall survival between group one and group two patients in this cohort, with 83 percent of the 13 patients assigned to group one still alive at two years compared to 63 percent of the patients in group two.
Looking across all the studies, patients classified as group one fared similarly whether on monotherapy or multiple checkpoint inhibitor therapies, while patients classified as group two had better outcomes when treated with the nivolumab and ipilimumab combination, indicating that the test is able to identify patients who will benefit from combination therapy.
Beyond the test performance, a look at the proteins significantly associated with the classifier highlight some potentially interesting biology underlying cancer immunotherapy. Specifically, the test found that high levels of complement proteins like c-reactive protein and C3a were associated with the more poorly responding group two.
This, Weber noted, ties in with recent research indicating that high levels of complement proteins inhibit T-cell reactivity, which hampers immunotherapy performance.
While the work establishing this relationship is new, it tracks with what Weber said is actually a well-established phenomenon in T-cell research.
"For years and years, whenever scientists used serum, either fetal calf serum or human serum, to grow T cells, you had to heat-inactivate it," he said. "If you don't do that it screws up the T cell growth. Everyone tells you that you have to do that to get rid of the complement [proteins]."
"When you think about it, complement inhibits T-cell reactivity, and that is why we heat-inactivate the serum. I never connected the dots… but if the complement [level] is high, it is going to inhibit T-cell reactivity, and that means the immunotherapy won't work well."
"So this work confirms that," Weber said, adding that the apparent role of complement proteins in hampering immunotherapies also suggests a potential therapeutic intervention.
"There is an antibody that blocks complement that is used as a [US Food and Drug Administration]-approved drug for paroxysmal nocturnal hemoglobinuria and some other complement-related diseases," he said. "It is a very expensive drug, but it is out there, and technically you could try to apply that antibody with checkpoint inhibition."
"If you buy into the data, which I do, that high levels of complement inhibits T-cell reactivity and the ability of the PD-1 blockade to promote T-cell activity, then a complement-binding antibody like this might boost the effect," Weber said.
He said that he and his colleagues are now preparing for a larger validation study of the test with retrospective samples they have already obtained, and a prospective adjuvant study in which they plan to look at samples from many hundreds of melanoma patients.
"Then, at the end of the day, if you want to validate it as a marker where you intervene [based on its results] you will have to do a prospective study to randomly allocate patients based on their test results to decide whether you give them single or multiple agents," Weber said.
He added that he is collecting samples from other kinds of cancers including lung cancer to look into whether the test will be effective at guiding therapy across different histologies.
"I would assume that this is a phenomenon that is generalizable across other cancers, but you have to subject it to the acid test, and that is why you do the trials," he said.