Although colorectal cancers generally do not show a strong response to anti-PD-1 therapy, Topalian indicated that the microsatellite instability (MSI)-high genotype, which is present in 15% of colorectal cancer and 2% to 4% of all solid tumors and contributes to high mutational load, is an exception.
A recent study13
showed that 62% and 60% of mismatch repair (MMR)-deficient subsets of colorectal and noncolorectal solid tumors, respectively, had an objective response to anti-PD-1 therapy whereas none of the patients with MMR-proficient colorectal cancers responded. Mehnert also described a case study of a patient with DNA polymerase epsilon (POLE)-mutant endometrial cancer, which is associated with high mutational burden and elevated expression of several immune checkpoint genes, that had an exceptional response to pembrolizumab.14
Although Topalian indicated that MSI assays are relatively standard for most hospitals, they may not detect all cases with high mutational load. Furthermore, Mehnert stated that approximately half of POLE-mutant cancers show MMR proficiency on MSI assays.
Although whole-exome sequencing was used to determine mutational load in a recent study associating mutational load with clinical benefit and progression-free survival in pembrolizumab-treated NSCLC,15
Douglas B. Johnson, MD, of Vanderbilt-Ingram Cancer Center, stated that the method would be difficult and expensive to implement in the clinic. He presented a method that uses hybrid capture-based next generation sequencing (NGS) of a 236- or 315-gene coding sequence to profile a fraction of the genome and act as a surrogate for mutational load. He and his colleagues analyzed archived samples from patients with melanoma who received anti-PD-1/PD-L1 therapy and showed that the NGS-estimated mutation load was strongly correlated with the whole-exome sequencing results.16
Responders to pembrolizumab had a higher mutational load than did nonresponders, and further stratification showed that patients with a high mutational load had a significantly greater response rate and median OS than patients with intermediate or low mutational load, respectively. Rosenberg described the same NGS assay that was used in the IMvigor 210 trial11 to show that response to atezolizumab was associated with a significantly higher mutational load. Furthermore, the quartile with the highest mutational load had better OS than the quartiles with lower mutational load.
Although these results suggest high mutational burden as a potential biomarker for response to anti-PD-1 therapy, Topalian indicated that the cutoff of mutational load for predicting clinical response is unclear.
Johnson agreed that mutational load alone should not be used to exclude patients from anti-PD-1 therapy, although he speculated that mutational load could play a role in prioritizing therapy for certain patients. For example, he suggested that monotherapy with PD-1 blockade could be effective for patients with high mutational load and less toxic than combination therapy with ipilimumab and PD-1 blockade.
Like the neoantigens produced by hypermutated tumors, viral antigens are strong stimulants of the immune system, according to Topalian. She stated that because virus-positive tumors express PD-1 on infiltrating T cells and PD-L1 on tumor cells and macrophages, they may respond to anti-PD-1/PD-L1 therapy. She cited an analysis of archived tumor samples from Johns Hopkins that showed Merkel cell carcinoma expressing the Merkel cell polyomavirus is more likely to express PD-L1 in an immune-front pattern than virus-negative Merkel cell carcinoma.17
She also described a follow-up study18
showing that patients with advanced Merkel cell carcinoma had a 56% response rate to pembrolizumab. Response rates were similar between viruspositive and virus-negative tumors even though the virus-positive tumors had a lower mutational load.