If we give just a dose or two of an anti–PD-1 agent and look at the early on-treatment biopsy, we potentially could tell with relative certainty who will become a responder.— Jennifer A. Wargo, MD, MMSc
Tweet this quote
A key challenge in advancing immunotherapies is to understand mechanisms of response and resistance. Emerging research in this area—including evidence that early on-treatment biopsies can predict response—was discussed at the 2017 ASCO-SITC (Society for Immunotherapy of Cancer) Clinical Immuno-Oncology Symposium by Jennifer A. Wargo, MD, MMSc, Associate Professor in the Departments of Surgical Oncology and Genomic Medicine at The University of Texas MD Anderson Cancer Center, Houston.1
“We’ve made tremendous progress in cancer through the use of targeted therapies and immunotherapy. Advanced melanoma is a prime example, with the approval of more than 10 regimens in the past 5 years. We have seen a huge benefit to patients, first with a doubling and now a near tripling of 1-year survival rates,” she said.
Dr. Wargo continued: “However, despite these advances, responses are heterogeneous and are not always durable. Toxicity can also be an issue for checkpoint inhibitors as monotherapy and in combination. There is still a critical need to better understand who will benefit from these treatments, as well as their proper timing, sequence, and use in combination.”
Emerging Mechanisms of Resistance
It is becoming clear that how tumors grow and respond to cancer therapy depends on a number of different tumor-intrinsic factors, including the tumor genome, epigenome, and microenvironment. “Tumors are complex ecosystems. Cellular and soluble components of the tumor microenvironment have dramatic implications for how each tumor responds,” Dr. Wargo explained.
In addition, host immunity and the “local and extended environment,” including the microbiome, play a key role in tumor growth and response to therapy. “On top of this, all of these factors are tightly interrelated and potentially interdependent,” she noted.
Within the tumor and its microenvironment, several factors stand out. One is mutational load—patients with the most mutations tend to derive the most treatment benefit. Another is density and distribution of infiltrating T cells. These factors can be predictive of response, but they are difficult to quantify and certainly not perfect in this regard. The presence of immune-related genes in the tumor microenvironment also correlates with outcome; however, there is tremendous overlap among these signatures between responders and nonreponders to therapy, added Dr. Wargo.
Early Responses May Be Predictive
“We wanted to build on this elegant work by doing deep tissue–based analysis in patients with metastatic melanoma going onto checkpoint blockade,” she revealed. The 53 patients were initially treated with cytotoxic T-lymphocyte–associated protein 4 (CTLA-4) blockade, and those who failed to respond or whose disease progressed then received monoclonal antibodies designed to block the programmed cell death protein 1 (PD-1) inhibitors. Tissue was sampled and molecularly profiled at multiple time points during treatment and at time of disease progression, when applicable.
Using a 12-marker immunohistochemistry panel, they found immune “signatures” to be modestly different between responders to PD-1 blockade and nonresponders, in baseline tumor biopsies, but there was a considerable overlap and no clear cutoff over which a patient could be predicted to be a responder. On the other hand, biopsies taken after one or two treatment doses demonstrated “something profound,” Dr. Wargo shared. There was a highly statistically significant nonoverlapping difference between responders and nonresponders in these “early on-treatment” samples, she reported.
“In nonresponders, the tumor remains bland, but in responders, you see an immune infiltrate that’s highly predictive of later response. This suggests that if we give just a dose or two of an anti–PD-1 agent and look at the early on-treatment biopsy, we potentially could tell with relative certainty who will become a responder, which is critically important; however, this needs to be tested in the context of carefully designed clinical trials,” she said.
The same findings were made using a 795-gene panel. “In pretreatment biopsies, we could not predict response using 12 markers, and we can’t even do it with 795 genes,” she said. However, in early on-treatment biopsies of patients receiving anti–PD-1 agents (and less so with ipilimumab [Yervoy]), “you see a night-and-day difference in the usual suspects” (such as CD8-positive cells) between responders and nonresponders, explained Dr. Wargo.
“The initial data suggest, at least to me, that we may actually have acceptable predictive biomarkers, but we are simply looking at the wrong time point. We have been fishing in the shallow pretreatment waters, when we should be looking at the early adaptive immune responses on therapy, where the waters are rich and deep,” she maintained. Other investigators have recapitulated these findings—including a group from the Melanoma Institute of Australia, led by Dr. Georgina Long and Dr. Richard Scolyer.
Her laboratory also compared genetic profiles from paired biopsies from pretreatment and on-treatment samples, examining the 37 most upregulated or downregulated genes for responders vs nonresponders to PD-1 blockade. “We found the profile of responders clearly separates from that of the nonresponders,” she reported. For example, genes related to defects in interferon signaling and antigen processing were common in the nonresponders. A follow-up study analyzing additional molecular markers was recently published by the group in Science Translational Medicine and demonstrated that differences in T-cell clonality and copy number loss for relevant genes were also observed in nonresponders to therapy.2
Takeaway Message
“What we can take away from this is that responses to immunotherapy depend on a number of factors. The ‘pillars of response’ to checkpoint blockade include the tumor genome and epigenome, the tumor microenvironment, systemic immunity, and the overall environment. And onto these pillars of response, you can onlay hallmarks of response to checkpoint blockade, including mutational load among several other factors. In summary, Dr. Wargo contends that, “The optimal monitoring and targeting of therapeutic resistance require a global integrated approach of all these factors and ideally should include on-treatment assessment of responses—at least until we identify better pretreatment biomarkers of response and therapeutic resistance.” ■
Disclosure: Dr. Wargo has received hononaria from the speakers bureau of Dava Oncology, Bristol-Myers Squibb, and Illumina and is an advisory board member for GlaxoSmithKline, Roche/Genentech, and Novartis.
References
