LONG-TERM SURVIVORS of pancreatic cancer display evidence of enhanced tumor-specific T-cell responses that are associated with unique neoepitope quality but not quantity, according to Steven D. Leach, MD, Director of the Norris Cotton Cancer Center and the Preston T. and Virginia R. Kelsey Distinguished Chair in Cancer at the Geisel School of Medicine at Dartmouth.
This finding suggests there may be a particular immunogenic profile that could help predict which patients may respond to immunotherapeutics, Dr. Leach said at the 2018 Gastrointestinal Cancers Symposium.1 In the Keynote Lecture, he described the complexity of the immune landscape in pancreatic cancer and the recent discoveries made at the David M. Rubenstein Center for Pancreatic Cancer Research at Memorial Sloan Kettering Cancer Center, when he was Director.
Immunogenic or Not Immunogenic?
ALTHOUGH PANCREATIC cancer has appeared to be resistant to immunotherapeutic approaches, this could change with a better understanding of the mechanisms of resistance, Dr. Leach said.
T cells interact with epitopes, the part of the antigen that is recognized by the immune system; they recognize tumors by spotting their tumor-specific neoantigens. T cells appear able to recognize both mutational neoepitopes and self-epitopes, which are not normally expressed in a way that activates the immune system. The potential for somatic mutations to provide neoepitopes that can be recognized by T cells is of great interest in pancreatic cancer, he said.
The quantity of neoepitopes is conventionally thought to be a potential biomarker of response to checkpoint inhibitors. However, work from Dr. Leach and his colleagues has shown that neoepitope quality, not quantity, is the important determinant.
Conventional thinking is that pancreatic cancer is a nonimmunogenic tumor, based on its relatively low mutation burden—about 30 somatic mutations per tumor, or one mutation per megabase. This would result in only occasional neoantigen formation. But work by Dr. Leach’s lab has indicated these tumors may have at least 2.5 mutations per megabase, which indicates they are “regularly and frequently” generating an immune response, he said.
“So it’s not that pancreatic cancer is a mutationally silent tumor that lacks the potential to activate the immune system. There appear to be plenty of mutations present,” he indicated. There must be other reasons, he suggested, why pancreatic cancer is largely resistant to checkpoint inhibitors.
Unique Neoantigen Quality in Long-Term Survivors
DR. LEACH RECENTLY TEAMED UP with researchers from other institutions to look for unique neoantigen qualities in long-term survivors of pancreatic cancer.2 They asked whether there are mutational neoepitopes in pancreatic adenocarcinoma; whether T cells in this cancer are responding to those mutational epitopes; and what is the association between mutational neoepitopes, T cells, and survival.
Their tools included multiplexed immunohistochemistry, T-cell receptor V-beta (TCRVβ) sequencing, RNA sequencing, and whole-exome sequencing. This approach aimed to provide detailed immunophenotyping, to identify the clonal repertoire of T cells in each tumor, to characterize the intensity of the immune response, and to identify the burden of somatic mutations in patients.
“It’s not that pancreatic cancer is a mutationally silent tumor that lacks the potential to activate the immune system. There appear to be plenty of mutations present.”— Steven D. Leach, MD
Tweet this quote
They compared findings between 82 patients with surgically resected primary tumors (no metastatic tumors, no neoadjuvant chemotherapy) still alive at a median time of 6 years (long-term survivors) and 68 patients with a median survival of 0.8 years (short-term survivors). Patients were matched by stage, age, and gender.
Multiplexed immunohistochemistry revealed no difference in the absolute number of CD3-positive T cells between long-term and short-term survivors. A slight increase in the number of CD3-positive, FoxP3-positive T-regulatory cells was seen in long-term survivors, as was a slight increase in dendritic cell infiltration and myeloid cells, but the overall number of T cells was not different.
“But when we looked at cytotoxic CD8-positive T cells, we saw a threefold increase among long-term survivors,” he said. “We also saw a 12-fold increase in the number of cytolytic CD8-positive T cells, including CD3-positive, CD8-positive, and granzyme B– positive cells. This suggests the long-term survivors did, indeed, display enhanced intratumoral T-cell immunity.”
The next question was whether the intratumoral T cells that were more abundant in long-term survivors were tumor-specific. TCRVβ sequencing of infiltrating T cells in tumor and in adjacent normal tissue revealed that, for both groups of patients, T-cell clones were highly unique to the tumor. More than 95% of T-cell clones resided uniquely in the tumor and were not shared with the adjacent normal pancreatic tissue. This suggested that T-cell infiltration may be a specific T-cell response to tumor-specific antigens—not a reflection of nonspecific inflammation in the setting of pancreatitis.
In long-term survivors, they did show a more diverse (polyclonal) tumor-specific T-cell receptor repertoire. RNA sequencing also confirmed a transcriptional signature of immune activation in long-term survivors, including greater upregulation of programmed cell death protein 1 (PD-1) and dendritic cell markers and lower expression of immunosuppression pathways such as STAT-3.
“In summary, what we saw in this extraordinary cohort of long-term pancreatic cancer survivors is a 12-fold increase in infiltration of cytotoxic T cells, a highly immunogenic microenvironment, and a polyclonal tumor-specific T-cell repertoire, which raises the question, ‘What are these T cells responding to?’” Dr. Leach asked.
What Are T Cells Responding to?
THROUGH WHOLE-EXOME sequencing, mutation identification, and neoantigen prediction, the researchers confirmed the finding of a “reasonable mutational load” in both short-term and long-term survivors. In particular, they found an average of 72 nonsynonymous mutations per patient, or about 2 per megabase.
They also found no differences in the mutation or neoepitope burden between the patient populations, suggesting “it’s not neoepitope quantity that is driving the differential immune response in short- vs long-term survivors,” he said. Looked at another way, survival was no better for patients whose tumors were high in neoantigens than in those low in neoantigens or for those high in neoantigen number plus high density of CD4 helper T cells.
“Only when we combined neoantigen quantity with the number of cytotoxic T cells, defined as CD3-positive/CD8-positive T cells or CD3-positive/granzyme B–positive T cells, did we see a remarkable delineation of long- vs short-term survivors,” he reported. “This suggests that some of these neoantigens are capable of driving an effective immune response, and when that happens, patients enjoy long-term survival.”
Infectious Pathogens Enter the Lab
THE NEXT CLUE that quality, not quantity, of the neoantigens was more important came from complex experiments based on discoveries in other malignancies—that effective antigens bore homology to antigens known to be presented by infectious microbial pathogens. The researchers hypothesized that neoantigens homologous to microbial peptides might provide a surrogate marker for immunogenicity; they tested this theory by creating a neoantigen quantity model (ie, immunogenicity is determined by total neoantigen number) and a neoantigen quality model (immunogenicitiy is determined by immunodominant neoantigens with maximal microbial homology).
Again, they showed that the simple use of neoantigen quantity was not predictive of survival. The characterization of the neoepitope’s homology to infectious pathogens, on the other hand, proved to be a remarkable discriminating predictor of long-term survival. Quality, in fact, proved to be the most significant predictor of long-term survival, in the multivariate analysis of this cohort (hazard ratio [HR] = 0.03; P = .008).
Is Neoantigen Quality Also Predictive in Nonselected Populations?
DR. LEACH and his team further showed that neoantigen quality is prognostic of survival in a nonselected population of 166 patients from the International Cancer Genome Consortium. High neoantigen quality was defined as homology to antigens known to be presented by infectious pathogens (ie, patients have mutations that generate proteins resembling those expressed by infectious microbes).
In pancreatic cancer “all-comers” with high-quality neoantigens, median survival was 30 months, compared with 14 months for those with low-quality neoantigens (HR = 0.4; P < .0001). Neoantigen quantity had no predictive value (P = 0.7). This remained highly significant in the multivariate analysis (HR = 0.5; P = .01).
Evidence of Ongoing Persistent Immunity
FINALLY, THE RESEARCHERS showed that long-term survivors display persistent T-cell clones that cross-react with both tumor neoepitopes and homologous microbial antigens. In fact, this persistence can be seen more than a decade after tumor resection, according to their examination of seven patients, who are still alive 7 to 12 years later.
TCRVβ sequencing on current blood samples and resected tumors revealed persistent reactive T-cell clones in five patients. The same clones that were present and reactive in the original tumor had persisted for a decade and were reactive to antigens from the tumor and from an associated microbial pathogen.
“The microbial pathogens to which mutations in long-term survivors of pancreatic cancer bore resemblance is a pretty exotic list, including things like yellow fever virus and dengue virus— diseases not prevalent on the Upper East Side of Manhattan,” he said. He believes this is not due to protective immunity from prior exposure, but more likely a result of a T-cell recombination repertoire, which “was evolutionarily sculpted in an ancestral environment, where T cells evolved not to protect against cancer but against infectious pathogens.”Error loading Partial View script (file: ~/Views/MacroPartials/TAP Article Portrait Widget.cshtml)
“The implications of this work are that it represents one strategy for a biomarker-driven personalized approach to immunotherapy in pancreatic cancer and potentially in other diseases,” he said. “This could guide the selection of patients for immuno-oncology treatment protocols and for the design of individualized peptide-based vaccines, selecting peptides that are predicted by this computation of neoepitope quality to be the most likely to generate an effective immune response.”
Vinod Balachandran, MD, is the first author on the article in Nature. This was a collaborative effort involving investigators from Memorial Sloan Kettering Cancer Center, Mt. Sinai, the Institute for Advanced Studies at Princeton, and the University of Glasgow.
DISCLOSURE: Dr. Leach reported no conflicts of interest.
1. Leach SD: Mapping the immune landscape in pancreatic cancer. 2018 Gastrointestinal Cancers Symposium. Keynote Lecture. Presented January 20, 2018.
2. Balachandran VP, Luksza M, Zhao JN, et al: Identification of unique neoantigen qualities in long-term survivors of pancreatic cancer. Nature 551:512-516, 2017.