328 - Cellular Immune and Genomic Biomarkers in NRG-HN003, a Phase I Study Adding Pembrolizumab to Adjuvant Cisplatin and Radiation Therapy (CRT) in Pathologically High-Risk Head and Neck Cancer (HNSCC)

J. E. Bauman¹, J. Harris², R. Uppaluri³, M. Yao⁴, J. Chen⁵, R. Jordan⁶, J. L. Geiger⁷, S. Jujjavarapu⁸, A. Chakravarti⁹, M. Phan¹⁰, F. Siddiqui¹¹, A. Kulkarni¹², P. Upadhyay¹², L. Vujanovic¹², B. Isett¹², G. L. Sica¹³, C. Reeder¹⁴, Q. T. Le¹⁵, and R. L. Ferris¹⁶; ¹George Washington University Cancer Center, Washington, DC, ²American College of Radiology, Philadelphia, PA, ³Department of Surgery/Otolaryngology, Brigham & Womens Hospital and Dana-Farber Cancer Institute, Boston, MA, ⁴University Hospitals Cleveland Medical Center, Cleveland, OH, ⁵UCSF, San Francisco, CA, ⁶University of California San Francisco, San Francisco, CA, ⁷Department of Hematology and Medical Oncology, Taussig Cancer Center, Cleveland Clinic, Cleveland, OH, ⁸OSF Healthcare, Peoria, IL, ⁹The Ohio State University, Columbus, OH, ¹⁰University of Oklahoma Health Sciences Center, Oklahoma City, OK, ¹¹Department of Radiation Oncology, Henry Ford Cancer Institute, Detroit, MI, ¹²University of Pittsburgh, Pittsburgh, PA, ¹³Department of Pathology, University of Pittsburgh, Pittsburgh, PA, ¹⁴UPMC Hillman Cancer Center, Pittsburgh, PA, ¹⁵Stanford University, Stanford, CA, ¹⁶University of Pittsburgh Medical Center, Pittsburgh, PA

Purpose/Objective(s): NRG HN003 was a phase I study that assessed the safety and recommended phase 2 schedule (RP2S) for adding concurrent and maintenance pembrolizumab to adjuvant CRT in patients with pathologically high-risk, HPV-negative HNSCC. As PD-1/L1 immune checkpoints can be upregulated by HNSCC during chemotherapy and RT, their inhibition may reverse associated immunosuppression. Here, we report the immune and genomic correlates of disease-free survival (DFS). Materials/

Methods: All 34 eligible patients treated at RP2S are included in the exploratory biomarker analyses. Only patients with available, relevant biospecimens were included in each analysis, thus n varies by biomarker category. To quantify the in situ immunophenotypic microenvironment, formalin-fixed, paraffin-embedded (FFPE) baseline primary tumor tissue was evaluated by multispectral imaging (n=34). FFPE tumor cores (n=24) were also subjected to whole exome sequencing (WES), with somatic variant calls facilitated by germline sequencing of peripheral blood mononuclear cells (PBMC; n=17). Baseline (n=20) and post-treatment (n=16) PBMC were evaluated by spectral flow cytometry characterizing T and natural killer (NK) cells, and monocytes with immune checkpoint (PD-1, LAG-3, TIGIT, TIM-3) and effector markers (CD39, CD69, GZMK, GZMB, IFN-?) if applicable. DFS rates were estimated by Kaplan-Meier method. Hazard ratios (HR) for biomarker values > vs. = median were estimated by Cox models. All analyses are exploratory.
Results: With a median follow-up of 3.1 years, the estimated 2-year DFS for the RP2S population was 59.4% (95% CI: 42.4-76.4). Post-treatment blood memory T and T regulatory (Treg) cells were associated with superior DFS: CD8+ central memory (T_CM; HR 0.10; 0.01-0.83); CD8+ effector memory (T_EM; HR 0.11; 0.01-0.89); Treg (HR 0.10; 0.01-0.83). Hypothesis-generating numerical associations with DFS included the following baseline tumor biomarkers: PD-L1+ stromal cells at the invasive tumor margin or within all stroma (HR 0.48; 0.16-1.40); PD-L1+ CD3- immune cells within the tumor bed (HR 1.77; 0.63-4.98); disruptive TP53 mutations (1-2 vs. 0 mutations per tumor; HR 2.18; 0.63-7.51). In baseline PBMC, CD8+ T_EM (HR 0.44; 0.10-1.86) and CD4+ T_CM (HR 0.38; 0.09-1.62); monocytes (HR 0.47; 0.11-1.99); and CD56dim CD16- NK cells (HR 0.48; 0.11-2.04) may be associated with DFS. Analysis of WES as well as immune checkpoint and effector markers is ongoing.
Conclusion: In NRG HN003, post-treatment circulating memory T cells and Tregs were associated with improved DFS. This cohort also generated testable hypotheses that baseline PD-L1+ stromal cells and various circulating immune subsets may be protective, while intratumoral PD-L1+ myeloid cells and disruptive TP53 mutations may be adverse in this setting.