Biomarkers for Immunotherapy: Results from ASCO 2016
| Jul 8, 2016
Co-authored by Gena Kanas, M.P.H., Ph.D.
At ASCO 2016, new data was presented on potential biomarkers that could predict response to immunotherapy in the treatment of a variety of cancers. Current U.S. Food and Drug Administration (FDA)-approved PD-1/-L1 checkpoint inhibitor immunotherapy approvals include the use of Keytruda® (pembrolizumab, Merck & Co.) or Opdivo® (nivolumab, Bristol Myers-Squibb/Ono Pharmaceuticals) in advanced melanoma and advanced non-small cell lung cancer (NSCLC); Opdivo for use in advanced renal cell carcinoma and Hodgkin’s lymphoma; and most recently approved Tecentriq™ (atezolizumab, Genentech/Roche) in advanced urothelial carcinoma.
One goal for the choice of biomarkers is their utility to select only those patients who will respond to a given treatment. Current immunotherapies target the interaction between PD-1 and PD-L1, and the inhibition of this interaction leads to antitumor activity. Unfortunately, over half of patients treated with immunotherapy do not show a clinical response, which has led to further investigation into possible immunotherapy combination strategies and to identify potential biomarkers to better predict those patients who best respond to immunotherapy.
The use of tumor mutation burden, mismatch repair deficiency and PD-L1 expression as possible biomarkers were discussed at this year’s ASCO conference.
Two abstracts from the KEYNOTE-010 and KEYNOTE-021 studies focused on PD-L1 expression as a potential predictive biomarker for response in NSCLC patients. KEYNOTE-010 reported a general correlation between improved efficacy with increasing PD-L1 expression, as defined by a tumor proportion score (TPS), in Keytruda-treated NSCLC patients compared to docetaxel.1 However, since PD-L1-negative patients were not included in the study, KEYNOTE-010 was unable to evaluate PD-L1 status as a biomarker for Keytruda. KEYNOTE-021 reported improved efficacy of Keytruda combined with different chemotherapy regimens regardless of PD-L1 status in three cohorts of advanced NSCLC patients.2 Across all three cohorts, the overall response rate (ORR) did not significantly differ between patients with TPS of 1% or greater and those with TPS less than 1%. These studies suggest that PD-L1 expression is weakly correlated with clinical response to immunotherapy.
Tumor Mutation Burden
Three abstracts reported on tumor mutation burden (TMB) as a possible predictor of immunotherapy response in NSCLC, melanoma, and urothelial carcinoma patients. Each study used the same set of genes for the analysis of TMB (calculated as the number of synonymous and nonsynonymous variants from a series of 236-315 genes).
The first study reported a correlation between TMB and longer immunotherapy (Keytruda, Opdivo, or avelumab (Pfizer)) treatment duration in 64 NSCLC cases.3 Patients with TMB of 15 mutations/Mb or more had a longer immunotherapy treatment duration compared to those with TMB of less than 15 mutations/Mb (p=0.010). However, the authors did not report data on efficacy of immunotherapy treatment by level of TMB.
The authors of the second analysis reported a correlation between increasing TMB and higher progression free survival (p<0.001) and overall survival (p<0.001) in melanoma patients treated with immunotherapy (Opdivo, Keytruda or Tecentriq).4 A high TMB in this study was defined as more than 23.1 mutations/Mb. TMB was not predictive for survival for patients not treated with immunotherapy.
The third study reported the results of an analysis of predictors of metastatic urothelial carcinoma patients’ response to Tecentriq in the IMvigor 210 Phase II trial.5 The patients within the highest quartile of median mutational load (more than 16 mutations/Mb for platinum treated and more than 13.5 mutations/Mb for first line cisplatin-ineligible) had higher overall survival (p=0.0012 and p=0.0079, respectively) than patients in the lower quartiles. The authors also reported an improved efficacy among patients with higher PD-L1 expression. When included in a multivariable model, both PD-L1 and TMB were independent and significant predictors of response to Tecentriq (p=0.0109 and p<0.0001, respectively).
The results of these three studies suggest that a high TMB is correlated with clinical response to immunotherapy; however, each study provided a different definition of “high” TMB. While assessing TMB as a potential biomarker, the threshold to be used to identify those patients who would benefit most from immunotherapy treatment will need to be determined via consensus.
Mismatch Repair Deficiency
Two abstracts reported updated results that included more patients from the same study6 that focused on mismatch repair (MMR) deficiency or microsatellite instability (MSI) as markers of immunotherapy response.7, 8
In the first analysis,7 30 patients with MMR-deficient metastatic or locally advanced non-colorectal cancers who had received at least one prior therapy were treated with Keytruda. After a 10-month median follow-up, 53% of the patients had an objective response, and responses were observed in all of the cancer types evaluated with durable disease control. Unfortunately, patients with MMR-proficient cancers were not included in this analysis, thus this study is unable to determine the utility of testing for MMR as a biomarker for Keytruda. The second analysis8 focused on metastatic colorectal cancer patients treated with Keytruda and included both MMR-deficient (n=28) and MMR-proficient (n=25) patients. In contrast to the previous results for PD-L1 and TMB, there were no responses to Keytruda in MMR-proficient colorectal tumors, highlighting the selective nature for this biomarker.
Future studies will be necessary to confirm the utility of MMR deficiency as a biomarker as well as the relationship for MMR deficiency with PD-L1 expression and/or TMB. Several Phase III and Phase II trials are currently recruiting in patients with MSI-high or MMR-deficient advanced or metastatic colorectal cancer.
Among the FDA-approved indications for the three approved PD-1/-L1 inhibitors, only Keytruda has a label currently limiting use to a biomarker-defined patient subtype (Keytruda in PD-L1-positive NSCLC). Nevertheless, as the development of these agents is rapidly expanding across tumor types, there remains strong interest to identify predictive biomarkers of response.
Of the three possible biomarkers discussed above, MMR deficiency has the strongest evidence supporting its predictive behavior to PD-1 inhibitors. PD-L1 status and TMB may ultimately be most effective as predictive biomarkers when combined with each other and potentially other tumor-specific factors. Thus, the current oncology model of using a single biomarker to select patients for treatment may not be the best approach in the era of immunotherapy.
- Bass P, et al. J Clin Oncol. 2016; 34(suppl); abstr 9015.
- Gadgeel SM, et al. J Clin Oncol. 2016; 34(suppl); abstr 9016.
- Spigel DR, et al. J Clin Oncol. 2016; 34(suppl); abstr 9017.
- Johnson DB, et al. J Clin Oncol. 2016; 34(suppl); abstr 105.
- Rosenberg JE, et al. J Clin Oncol. 2016; 34(suppl); abstr 104.
- Le DT, et al. N Engl J Med. 2015; 372(26):2509-20.
- Diaz LA, et al. J Clin Oncol. 2016; 34(suppl); abstr 3003.
- Le DT, et al. J Clin Oncol. 2016; 34(suppl); abstr 103.