Mismatch repair protein deficiency (MMRD) or microsatellite instability-high (MSI-H) occurs in about 5 percent of metastatic colorectal cancers while higher frequencies are described in patients with earlier stages of disease (J Gastrointest Oncol 2018;9(4):610-617, Science 2017;357(6349):409-413). The drop in frequency in MSI-H cancers across increasing stages of colorectal cancer reflects the inherent good prognosis of earlier stage MSI-H, which is attributed to a robust tumor immune response.
MSI-H is characterized by increased tumor mutation, increased frameshift mutations, and insertion-deletion (indel) alterations. This leads to an increased incidence of neoantigen formation, which leads to a robust anti-tumor immune response. The recognition of an immune-overdrive status in MSI-H tumors, characterized by increased intratumor cytotoxic T-cell infiltration (CD8+) and countered by increased intratumor checkpoint expression (PD-L1, CTLA-4, etc.) (Cancer Discov 2015;5(1):43-51) led to the investigation of multiple checkpoint inhibitors in this group of patients.
A pivotal study evaluated pembrolizumab monotherapy in three different patient cohorts: MSI-H metastatic colorectal cancer, microsatellite-stable (MSS) colorectal cancer, and MSI-H non-colorectal cancers (N Engl J Med 2015;372(26):2509-2520). Responses were limited to the MSI-H cohorts, with 4/10 and 5/10 of the evaluable MSI-H colorectal cancer patients experiencing a partial response or stable, while only two of 11 MSS patients experienced stable disease (SD) as best response. An updated outcome on the KEYNOTE-016 trial confirmed a 52 percent response rate (RR) and 30 percent SD in MSI-H colorectal cancer compared to 54 percent RR and 18 percent SD in MSI-H non-colorectal cancers (Science 2017;357(6349):409-413).
KEYNOTE-164/158 investigated a similar treatment design and reported 28 percent RR and 23 percent SD in MSI-H colorectal cancers (KEYNOTE-162) and 37 percent RR and 21 percent SD in MSI-H non-colorectal cancers (KEYNOTE-158) (Ann Oncol 2017;28(suppl_5):v122-v141). Nivolumab monotherapy was examined on CheckMate-142 clinical trial and was associated as monotherapy with a RR of 33 percent and a SD rate of 31 percent (Lancet Oncol 2017;18(9):1182-1191).
Collectively, these studies suggest that PD-1 blockage is associated with response rates of 28-52 percent and disease control rates of 51-82 percent. Similarly, significant differences in outcome have been reported in terms of progression-free survival across these studies with 43 percent of patients being progression-free at 9 months on KEYNOTE-164 versus 59 percent at 24 months on KEYNOTE-016 versus 50 percent at 12 months on CheckMate-142. Of note, most patients achieving disease control at 1 year maintained protracted subsequent disease control.
While the above results confirm the considerable activity with PD-1 inhibitors in MSI-H colorectal cancer, they also point to the heterogeneity of the MSI-H population in terms of response and the durability of disease control. Indeed, approximately 30 percent of patients with MSI-H tumors will have progressive disease as a best response to PD-1 inhibitors.
Therefore, the identification of additional biomarkers of response may help further refine which MSI-H patients should receive anti-PD1 therapy, especially when considering the integration of checkpoint inhibitors early on in the course of disease treatment. While PD-L1 expression has been shown to predict for response to PD-1 inhibitors in a variety of solid tumors, it has failed to predict for response to PD-1 inhibitors in MSI-H colorectal cancer (Lancet Oncol 2017;18(9):1182-1191; J Clin Oncol 2018;36(8):773-779).
Tumor Mutation Burden
Tumor mutation burden (TMB) has been recently validated as a predictive biomarker of response to checkpoint inhibitors in melanoma, NSCLC, bladder cancer, and a combination of solid tumors treated with PD-1/PD-L1 inhibitors (Ann Oncol 2019;30(1):44-56). However, the relevance of TMB as a biomarker of outcome in patients with MSI-H colorectal cancer has not been assessed, especially the fact that most of these patients exhibit a high TMB that exceeds 20.
To test the impact of TMB on checkpoint inhibitor response in MSI-H colorectal cancers, we elected to focus our study on patients' MSI and TMB testing by a CLIA-certified NGS panel (Foundation One) (Ann Oncol 2019; doi:10.1093/annonc/mdz134). Enrolled patients received a PD-1 or PD-L1 inhibitor, either alone or in combination with a CTLA-4 inhibitor. Patients were identified from five cancer centers (City of Hope, The Angeles Clinic, Comprehensive Cancer Centers of Nevada, University of California-Davis, and University of Chicago). A total of 22 patients fit the above eligibility.
Twenty patients were treated with a PD-1 inhibitor (19 pembrolizumab and one nivolumab) and two received combination immunotherapy (nivolumab + ipilimumab and durvalumab + tremelimumab). Similar to prior published data, we did not see a correlation between RAS/BRAF status, primary tumor site, site of metastatic disease, and outcome. However, a significant difference in TMB was noted between responders and non-responders.
Patients with an objective response had a median TMB of 54 mutations/Mb compared with a TMB of 29 mutations/Mb in non-responders (p < 0.001). Similarly, a higher TMB was predictive for a better likelihood of disease control, with an optimal cut-point between 37 and 41 mutations/Mb. The median PFS for patients with a TMB > 41 was not reached at a median follow-up of 18 months. In contrast, the median PFS of patients with a TMB < 37 was only 2 months. Patients with higher TMB experienced a superior overall survival in comparison to the lower TMB group.
Through this study, we showed that TMB is predictive for a benefit from checkpoint inhibitors in patients with MSI-H colorectal (Ann Oncol 2019; doi:10.1093/annonc/mdz134). The cut-point identified in this study lies between 37 and 41 mutations/Mb. It is interesting that 35 percent of patients with MSI-H cancer lie below this cut-point based on a large population database analysis. This is, coincidentally, the same percentage of patients who derive little to no benefit from PD-1 targeting.
We recognize that our study population is limited in number; therefore, there is a need to validate these findings in another, larger data set. However, our results suggest that we should pause on the integration of checkpoint inhibitors in the first- or second-line settings in patients with MSI-H tumors with lower TMB (< 37 mutations/Mb).
It is important to note that, while our findings suggest that lower TMB patients are less likely to respond to PD-1 inhibitors, they do not exclude the possibility of response in this population (2/9 patients who responded had a TMB of < 37 mutations/Mb). Our results add additional confidence to the integration of checkpoint inhibitors in earlier lines of treatment in the patient population with higher TMB (>41 mutations/Mb).
In addition, we note that more than 90 percent of our patients were treated with anti-PD1 monotherapy and, therefore, our results may have limited applications on the combination of nivolumab plus ipilumumab, a combination associated with a disease control rate that exceeds 80 percent and with a 2-year progression-free survival rate of 50 percent (J Clin Oncol 2018;36(8):773-779). Whether a different TMB cut-point would apply in such settings remains to be determined.
MARWAN FAKIH, MD, is Professor in the Department of Medical Oncology and Therapeutics Research, Medical Director in the Judy and Bernard Briskin Center for Clinical Research, and Section Head of GI Medical Oncology at City of Hope, Duarte, Calif. He also holds the Judy and Bernard Briskin Distinguished Directorship in Clinical Research at the cancer center.