Role of TMB and PD-L1 as Prognostic Factors and Predictors of Response to Immunotherapy in Head and Neck Squamous Cell Cancer

Purpose/Objective(s)

Immunotherapy with checkpoint inhibitors (ICI) improved survival in HNSCC. Response rates to ICIs remain low, in part due to lack of effective predictors of response to guide patients' selection. PD-L1, as the only biomarker used in prospective research of ICI in HNSCC has generated inconsistent results. Developments in Precision Oncology in HNSCC brings the prospect for new biomarkers. Tumor mutational burden (TMB) is under investigation in many solid tumors and this study assessed the significance and potential role of TMB and PD-L1 in the immunotherapy of HNSCC.

Materials/Methods

This single institution retrospective review includes HNSCC patients (pts) with tumor tissue evaluated for PD-L1 (IHC 22C3 pharmDx kit) and TMB by FoundationOne. PD-L1 was reported as TPS score (prior to 2018) or CPS score (thereafter). TMB and PD-L1 were analyzed for individual association with demographics, disease characteristics and survival, and with progression free survival (PFS) and response to ICI. Pts tested for TMB (mut/Mb) were grouped as TMB(1-5) and TMB(6+) and pts tested for PD-L1 were grouped as PD-L1(0), PD-L1(1-19) and PD-L1(20+).

Results

139 pts were included, 128 pts had TMB, 95 pts had PD-L1 and 92 pts had both results. TMB(6+) was associated with smoking (p=0.03) and with tumor location, with laryngeal cancer pts having higher TMB (p<0.01) and oropharyngeal cancer pts having lower TMB (p p<0.01). PD-L1 was higher in African Americans (p=0.04), in never smokers (p=0.04) and never drinkers (p=0.01), and in pts with earlier cancer stage (p=0.03) and lower tumor stage (p p<0.01). There was no correlation of TMB or PD-L1 with HPV status. TMB(6+) was associated with improved survival in univariate analysis (p=0.02) and in Cox proportional hazards regression model adjusted for age, tobacco, tumor site, nodal stage, previous chemo/radiotherapy and PD-L1 level (p<0.01). No correlation was found between PD-L1 level and prognosis. There was no correlation between TMB and PD-L1 level. Treatment efficacy with ICI was evaluated in 51 pts, of whom 40 pts had TMB and 36 pts had PD-L1 results. There was significant association of continuous TMB score with PFS (p=0.013) and with response to ICI (mean TMB=11.2 in responders and 4.9 in non-responders; p=0.01). Similarly, in categorical analysis, PFS was 538 days in TMB(6+) and 261 days in TMB(1-5) (p=0.04). There was no association between response to ICI or PFS and PD-L1 analyzed as categorical (p=0.89 and p=0.62) or continuous variable (p=0.99 and p=0.66).

Conclusion

Inadequate PD-L1 results in this study could be explained by inconsistence in methods. The association of TMB with prognosis and response to ICI in this study warrants further investigations, and prompts a methodology harmonization effort and advancement of TMB in future prospective studies. The recent availability of circulating TMB makes it a further attractive biomarker.