The US Food and Drug Administration (FDA) recently licensed for the first time use of an immunotherapy solely on the basis of the presence of a biomarker. The labeling was approved just 10 days before the 2017 annual meeting of the American Society of Clinical Oncology (ASCO), where several experts at a symposium on biomarkers for immunotherapy explained that this approval, although the first, will not be the last.
The biomarker is mismatched repair deficiency (dMMR), which documents microsatellite instability and generally correlates with tumor mutational burden. According to the new FDA labeling, the presence of dMMR is a sufficient indication for the use of the checkpoint inhibitor pembrolizumab in any unresectable or metastatic solid tumor in an adult or child.
This relatively broad indication represents "the first site-agnostic indication for any cancer therapy," according to Kurt A. Schalper, MD, PhD, director of the translational immuno-oncology laboratory at Yale University Cancer Center, New Haven, Connecticut. Speaking at the ASCO symposium, Dr Schalper explained that dMMR is the first FDA-approved biomarker for an immunotherapy across any solid tumor type, but it is not the first biomarker approved for a checkpoint inhibitor.
The Example of PD-L1
Rather, the checkpoints themselves, such as programmed death ligand 1 (PD-L1) or cytotoxic T-lymphocyte–associated protein 4 (CTLA-4), might be expected to the best biomarkers for guiding immunotherapy developed for this target, and the FDA has approved assays for PD-L1. However, although a correlation between PD-L1 expression and benefit has been demonstrated repeatedly in many solid tumors, the correlation is highly imperfect.
"PD-L1 expression can identify an enriched population likely to benefit although the assays have not been standardized, and it is not a definitive biomarker for all patients," Karen L. Reckamp, MD, medical director of clinical research operations at City of Hope in Duarte, California, who also participated in the symposium. Speaking specifically about the role of the PD-L1 biomarker in lung cancer, Dr Reckamp outlined a very mixed clinical experience.
There are three assays for PD-L1 that have been granted FDA approval. Others are in advanced testing. The experience with atezolizumab using the immunohistochemistry (IHC) SP142 antibody assay provides a telling example. The third anti–PD-1 therapy approved for metastatic non-small cell lung cancer (NSCLC), atezolizumab has showed far more clinical activity in patients with the highest PD-L1 expression on tumors and immune cells than in those with the lowest, but the relationship is anything but absolute.
Clearly, PD-L1 is a marker, but it is not selecting all of the patients who will benefit.
"Clearly, PD-L1 is a marker, but it is not selecting all of the patients who will benefit, and some of the patients with high levels turn out to be patients who will not necessarily benefit," said Dr Reckamp, reviewing the atezolizumab data. While she indicated that similar statements could be made about other IHC antibody assays used to predict the clinical activity of other checkpoint inhibitors, the labeling of checkpoint inhibitors differs on the basis of the phase 3 trials on which approval is based.
Specifically, the labeling of atezolizumab, which is approved for metastatic NSCLC that has progressed during or after conventional chemotherapy or appropriate targeted therapies, does not require demonstration of the PD-L1 target. Neither does the labeling for nivolumab, which is similar for NSCLC. The labeling for pembrolizumab, however, does include a requirement for demonstrating the presence of the biomarker both for the second-line indications granted to atezolizumab and nivolumab as well as one of the more recently granted first-line indications.
On the basis of a tumor proportion score (TPS) provided by biomarker IHC assays that quantify the presence of PD-L1 in NSCLC and other solid tumors, the indication for pembrolizumab as a monotherapy in second-line NSCLC is a TPS ≥ 1%, as determined by a FDA-approved test. In a subsequently granted indication for first-line therapy in metastatic NSCLC, the labeling for pembrolizumab calls for a TPS ≥ 50%. However, demonstration of PD-L1 expression is not in the labeling of the latest indication, which is first-line treatment of NSCLC in combination with pemetrexed and carboplatin.
The Problems With PD-L1
For clinicians, several issues further complicate PD-L1 testing as a tool to guide the use of checkpoint inhibitors, according to Dr Reckamp. Although she conceded that the presence of this biomarker "can identify and enrich a population likely to benefit," the currently available assays are not standardized and not interchangeable.
Moreover, the best approach to tissue sampling remains unresolved, with several confounding variables. This includes the fact that there is evidence that PD-L1 expression fluctuates, down- or upregulating over time and in response to therapy, and that there is a lack of clarity regarding the value of archived versus fresh tissue, an issue that is relevant across biomarkers and not just in quantifying PD-L1.
The biggest challenge in using these biomarkers is tissue heterogeneity.
"Then we come to the biggest challenge in using these biomarkers, which is tissue heterogeneity," said Dr Reckamp, referring now to the variability in the expression of any biomarker, including mutational burden, PD-L1, and biomarkers in development within any given tumor. In one of several examples cited by Dr Reckamp, multiregional whole-exome sequencing in NSCLC tissue resections obtained before systemic therapy produced varying levels of chromosome instability and driver mutations across 327 regions evaluated.
"What we know from these studies is that if we are looking at one part of the biopsy, we cannot be assured that it will be same everywhere," Dr Reckamp said. This concern is relevant to methods of tissue sampling, such as fine-needle and core biopsies. Optimal protocols remain unresolved. Liquid biopsies pose their own challenges for evaluating biomarkers. Not least of these is that greater or lesser tumor volume has the potential to change the quantity of the measured biomarker, such as dMMR or PD-L1, to a degree that might be misleading.
However, the complex interrelationship between oncogenic pathways may explain why the two biomarkers approved so far for immunotherapy, dMMR and PD-L1, have relative rather than absolute predictive value for a therapeutic response. It is likely that the value of all biomarkers will be improved with greater context. According to Dr Schalper, a more detailed understanding of the tumor microenvironment will permit both better individualization of therapy and new insights into cancer pathophysiology.
At Yale, tissue immunoprofiling with mass spectrometry technology, which Dr Schalper described as "one of many multiplex platforms that are being developed," is now capable of evaluating 26 biomarkers simultaneously from tissue mounted on a conventional slide. More markers are being added at a rapid pace at many institutions pursuing this vein of research. Dr Schalper suggested that the work in understanding the tumor microenvironment can and will be integrated with other relevant information for understanding the specific pathology of any given tumor.
"Integration of genomic, molecular, phenotypic, and clinical factors can be used and will probably be used to develop signatures that are much more powerful than single markers and to understand resistance to therapy," Dr Schalper said.
With dMMR and PD-L1, biomarkers have an established, if not fully defined, role in guiding immunotherapy. Whereas Dr Reckamp explained that the value of each has the potential to improve significantly as more information standardizes their application, Dr Schalper outlined a future in which new biomarkers augment rather than displace the current options to further individualize therapy.