Is the success of future immunotherapies dependent on biomarkers?
Biomarkers will be essential for certain combinations.
PD-L1 and tumor mutational burden are two more mature biomarkers that have been assessed in clinical trials and correlated with outcomes. Many other biomarkers are under development — such as T-cell receptor repertoire, circulating T cells, infiltrating T cells, neoantigens, interferon gamma and inflammatory gene signatures, PD-L2, and other markers in the peripheral blood and at the tumor site. Although a lot of work is ongoing, these research efforts are early, and biomarkers to predict or select response, survival or even toxicity are in exploration only.
Biomarkers are helpful to attain early approval of a drug by selecting a population that has a high response rate. However, the problem with current biomarkers such as PD-L1 is that they are not entirely reliable. PD-L1 levels can be heterogenic; they can fluctuate or change with prior or ongoing therapy; and they can depend on biopsy timing and sampling. PD-1 checkpoint blockade therapy generally has a more favorable toxicity profile with respect to cytotoxic therapy, and the ability to produce sustained responses. Using a biomarker to select immunotherapy may be an even larger issue for diseases with no effective therapy, as it can deny patients from receiving a potentially beneficial therapy. In fact, data from second-line trials in advanced NSCLC have shown that PD-L1-negative status does not necessarily exclude people from having a response to immunotherapy.
Further, many other diseases — such as head and neck cancers — do not have many effective well-tolerated standard therapies that improve survival substantially. In these situations, biomarkers are less helpful, as anti-PD-1 therapy is mostly well tolerated and patients would prefer to be treated with immunotherapy due to limited treatment options. Moreover, biomarkers in this setting have been less reliable in excluding people who would still benefit.
Biomarkers would be more effectively developed to predict response to combination immunotherapy rather than single-agent anti-PD-1 therapy. For example, in melanoma, the response rates are substantially higher for the combination of ipilimumab and nivolumab than single-agent anti-PD-1 therapy.
This combination is the most effective available therapy for melanoma, but it has an exceptionally high toxicity rate. As this and other combinations are being developed for other tumor types, there is a strong need for better biomarkers to predict who we should and should not treat. For single-agent anti-PD-1 therapy across the various cancer types, biomarkers will not have as much of a role in selecting patients.
Laura Q.M. Chow, MD, is a medical oncologist at Seattle Cancer Care Alliance, associate professor of medical oncology at University of Washington School of Medicine and associate member of clinical research at Fred Hutchinson Cancer Research Center. She can be reached at email@example.com. Disclosure: Chow reports consultant/advisory board roles with Genentech and Takeda, and research funding paid to her institution from AstraZeneca, Bristol-Myers Squibb, Genentech, Incyte, Merck and Seattle Genetics.
Biomarkers are not neccessarily needed for the initial success of future therapies. We already see responses across the board regardless of histology in a small percentage of patients treated. However, the optimization of future immunotherapy use is dependent on biomarker discovery. To get to the Holy Grail — for a patient to walk into their doctor’s office and have the oncologist look at their tumor samples, perform diagnostic analyses and then select their ideal therapy with high likelihood of benefit — is dependent on ideal biomarker discovery.
We are getting closer. There is a lot of very interesting information out there. Most people recognize that immunotherapy response depends on three things: factors that are specific to the tumor, factors specific to the host or patient, and factors specific to immune microenvironment.
PD-L1 testing on tumor samples is being incorporated into clinical practice, but it is not an ideal biomarker in many settings. Other factors, such as the mutation load, gene expression profile for specific tumors, expression of certain viruses and even features of the microbiome are being studied. Perhaps there will ultimately be a composite diagnostic assay, or several tests that we use together to help us identify who will respond best to immunotherapy.
In the absence of firm biomarker-driven data, in certain clinical settings where effective standard treatments are lacking, it is reasonable to treat first and ask questions later. For instance, in melanoma our team does not test routinely for PD-L1. This is a different clinical approach from that employed in the treatment of patients with lung cancer, for whom testing is often done upfront based on evidence from randomized clinical trials. However, we are getting closer to “precision immunotherapy,” with a lot of the new advances being made.
Janice Mehnert, MD, is regional phase I clinical program director at Rutgers Cancer Institute of New Jersey and associate professor of medicine at Rutgers Robert Wood Johnson Medical School. She can be reached at firstname.lastname@example.org. Disclosure: Mehnert reports consultant/advisory roles with or research funding from Amgen, AstraZeneca, Boehringer Ingelheim, EMD Serono, Genentech, Immunocore, Incyte, Macrogenics, Merck, Novartis and Polynoma.