Jason J. Luke, MD, FACP
At the 2015 annual meeting for ASCO we witnessed an extraordinary explosion into the general oncology consciousness of the Programmed Death-1: Programmed Death Ligand-1 (PD-1:PD-L1) axis and drugs that target that interaction. These drugs have changed the field of melanoma over the past few years, but at ASCO the presentations highlighted advances in non-small cell lung cancer (NSCLC), urothelial and renal cancers, mismatch repair (MMR) colon cancer and hepatocellular cancer, head and neck cancers as well as triple negative breast cancer. In fact, clinical trial results were presented that will immediately change the standard of care in NSCLC with the anti-PD-1 antibody nivolumab1
and possibly in melanoma with the combination of nivolumab with the anti-cytotoxic T-lymphocyte-associated protein 4 (CTLA4) antibody ipilimumab.2
Beyond that however, pembrolizumab (anti-PD-1), atezolizumab (formerly MPDL3280a, anti-PD-L1), and durvalumab (formerly MEDI-4736, anti-PD-L1) with and without tremelimumab (anti-CTLA4) all demonstrated groundbreaking results in various tumors. Each of these agents is on a rapid track to approval by the Food and Drug Administration.
There is tremendous optimism about PD-1:PD-L1 immunotherapy and perhaps we will eventually see a time when this approach is used in nearly all cancers. That being said, it is important to note that the response rate to these drugs as single agents remains in the 20% range in most tumors and that there is significant heterogeneity of response. A great deal of work remains to develop biomarkers of treatment benefit and to learn to apply these to the clinical management of our patients.
Where is the field at currently in the development of biomarkers of activity? Much further than you would likely expect. An obvious biomarker for PD-1:PD-L1 benefit has been staining by immunohistochemistry (IHC) for PD-L1 in the tumor, infiltrating immune cells, and/or stroma/vasculature. There are a number of problems with PD-L1 IHC, however. As you can tell from the previous statement about PD-L1, there is disagreement about the most appropriate cell type to test, what the cut off for positivity should be (1%, 5%, 50%?), when the testing needs to be done (fresh tissue at the time of each line of therapy?) and several other questions. Probably most importantly however, PD-L1 testing is inadequate because approximately 10% of patients of all cancer types who have had responses to PD-1:PD-L1 antibodies have stained negative by IHC. There are various reasons for this but pointedly, if we were to select patients by PD-L1 IHC alone, we would effectively deny at least 10% of the cancer patient population access to a drug that might have otherwise led to durable disease control.
There are other biomarkers that have been recently advanced as well. One of these is the number of mutations in the tumor being treated. Influential studies have suggested that higher mutational load corresponds with response to ipilimumab in melanoma3
and to pembrolizumab in NSCLC.4
Similarly, the approximately 60% response rate in MMR-deficient colon cancer to pembrolizumab5
(0% in MMR-proficient colon cancer) advances this logical hypothesis that a higher number of mutations in a cancer will lead to an increased frequency of neoantigen (non-self molecules that can be seen by the immune system) and therefore a higher likelihood of an anti-tumor immune response. While appealing as a hypothesis, this correlation is not perfect. Melanoma and NSCLC commonly do have high mutational load and high response to PD-1:PD-L1, however, renal cancer also has high response to PD-1:PD-L1 but demonstrates much lower mutational load.
Another approach to biomarker development is to more specifically examine the tumor microenvironment. Elegant translational work has demonstrated that those patients who are most likely to respond to PD-1 treatment are those whose tumors have the presence of both tumor infiltrating lymphocytes and PD-L1 at the invasive tumor margin.6
Further, work over several years has suggested that a productive anti-tumor immune response is driven broadly by the cytokine interferon-γ7
and multiple abstracts at ASCO this year demonstrated that the presence of an interferon-γ gene expression signature correlates with response and progression-free survival.8,9
Perhaps more importantly however, the lack of the interferon-γ gene signature correlated almost exactly with those patients who had no benefit to PD-1 immunotherapy. Tumors that demonstrate an interferon-γ gene expression signature have been previously described as “T-cell inflamed” tumors and are contrasted “non-T-cell inflamed” tumors which lack the signature.10