Tyler Vogt co-wrote this blog post with Sharon Karlsberg.
Since their first approval in 2014, PD-L1 and PD-1 checkpoint inhibitors have become the backbone therapy in multiple tumor types: melanoma, bladder cancer and non-small-cell lung cancer (NSCLC). Although only 20 to 35% of patients respond to these therapies, approvals in melanoma, bladder cancer and second-line lung cancer didn’t require an associated diagnostic test for PD-L1 status. Until the failure of BMS’s CheckMate-026 trial in first-line NSCLC, oncologists were asking the question, “Should I test for PD-L1?”
The Purpose of PD-L1 Testing
Since pembrolizumab’s approval in first-line lung cancer in October 2016, the answer to that question has seemingly been “yes” for NSCLC and “maybe” for all other tumors. However, as combination immunotherapies show efficacy in patients regardless of PD-L1 expression, the question is no longer, “Do I prescribe anti-PD-1/anti-PD-L1 therapy?” Instead, the question is, “Do I add a second agent for greater efficacy, yet risk significant toxicities?”
For now, the best answer still lies in the PD-L1 immunohistochemistry (IHC) test to identify those patients who will benefit from anti-PD-1/anti-PD-L1 monotherapy. However, that’s not an answer that inspires full confidence among oncologists. Identifying the right PD-L1 expression levels for response has its own challenges, with multiple IHC tests, protocols and cutoffs, along with the possibility of tumor heterogeneity and the need for interpretation. As Dr. Suzanne Topalian, associate director of the Johns Hopkins Bloomberg-Kimmel Institute for Cancer Immunotherapy, put it when we interviewed her at the Amercian Society of Clinical Oncology's 2017 Annual Meeting (ASCO): “I think it [PD-L1 testing] has been helpful in finding patients who are more likely to respond to anti-PD-1 or PD-L1 drugs, yet we do still see responses in the marker negative group, so that marker has to be used with caution.”
Making Sense of Mutational Load
Where can we find a better answer? The post-mortem of CheckMate-026 showed variability in response rates based on both PD-L1 status and tumor mutational burden (TMB), with “high PD-L1/high TMB” patients showing the best response to nivolumab monotherapy. Other studies, including some shared at ASCO, have reinforced the correlation between TMB and response to immunotherapy. TMB looks promising—at least for melanoma, bladder cancer and lung cancer, which show higher mutational burden than other tumor types.
Most recently, the FDA granted an accelerated approval for pembrolizumab in patients whose cancer carries a biomarker for microsatellite instability (MSI “high” tumors, specifically). Interestingly, this approval was regardless of tumor type—the first of its kind. The idea behind MSI is to identify tumors that are more likely to exhibit an augmented immune response to therapy based on a deficiency in DNA mismatch repair. The MSI biomarker predicts the response to checkpoint inhibition based on tumor genetics, rather than patient genetics.
“We found something [MSI] extremely useful to identify small sub-populations who are very likely to benefit," Dr. Topalian said. “So, hopefully, all the effort that’s going into research about cancer genetics and other intersections with immunotherapy will yield additional biomarkers that can be utilized in that way.”
This may be the first step in what looks to be a very promising area of research: looking at the tumor itself, and the tumor micro-environment, to identify cancers where checkpoint inhibition is best able to activate the immune system. It’s a lot easier for immunotherapies such as anti-PD-1s/anti-PD-L1s to mobilize “hot” tumors, where there’s high activity among tumor-infiltrating lymphocytes, such as T cells, compared to “cold” tumors, where there’s limited immune cell activity. In fact, novel immunotherapies currently under investigation (such as OX40) look to directly stimulate tumor-infiltrating lymphocytes.
The bottom line is that the search for better biomarkers continues. Fortunately, we have some promising leads.