Expansion cohort validation of a clinical predictive model for head and neck cancer survival in patients treated with immune checkpoint inhibitors.

Abstract

6030 Background: Immune checkpoint inhibitors (ICI) therapy is approved for patients (pts) with recurrent-metastatic (R/M) head and neck squamous cell carcinoma (HNSCC). The majority of pts will die within two years of diagnosis. We have shown that pretreatment clinical characteristics may predict overall survival (OS). Here, we expand our analysis to a total of 201 pts. Methods: Between January 15,2016 and April 9, 2020, 201 pts with R/M HNSCC were treated with ICI as first, second line and beyond. Data on p16 status, hemoglobin (Hb), albumin, lactate dehydrogenase (LDH), neutrophil, platelet and lymphocyte count was recorded initially. OS was defined from the start of ICI to death. Progression Free Survival (PFS) was defined from the start of ICI to disease progression (PD) or death. A nomogram was created using the rms package to generate individualized survival prediction. Results: 201 pts were analyzed, sex: 154 male (77%), 47 female (23%), median age 61 (IQR: 55-68). ICI drug: pembrolizumab 100 (50%), nivolumab 91 (45%), ipilimumab+nivolumab 10 (5%). Line of therapy: First: 98 (49%), second and beyond: 103 (51%). Tumor site: oropharynx 84 (42%), oral cavity 45 (22%), others 72 (36%). p16 status: negative 132 (66%), positive 69 (34%). Laboratory values: Median neutrophil count: 4.58 (IQR: 3.43-6.47), Median lymphocyte count: 0.69 (IQR: 0.47-1.08), Median Platelet count: 229 (IQR: 187-300), hemoglobin (Hb) normal/low 101/100 (50%/50%), albumin: normal/low 156/45 (78%/22%), LDH: normal/high 124/77 (62%/38%). Overall response rate: 36 (18%). Median OS: 12 months (CI: 9.4-14.8), median PFS: 4 months (CI: 3.5-5.7). The variables associated with OS were neutrophil count (high) [HR 1.28 (1.08 – 1.51), p=0.004], lymphocyte count (high) [HR 0.75 (0.60 – 0.95), p=0.015], albumin (low) [HR 2.06 (1.37 – 3.10), p<0.001], hemoglobin (low) [HR 1.64 (1.14 – 2.35), p=0.007], LDH (high) [HR 1.78 (1.23 – 2.56), p=0.002] and p16 status (positive) [HR 0.58 (0.39-0.87), p=0.009]. Using the prognostic index of the chosen model, we stratified patients into three risk groups at the 33rd and 66th percentile. Median OS in the good risk group was 24 months (CI: 18.5-NR), average risk group 13.8 months (CI: 11-20), poor risk group 2.3 months (CI: 1.7-4.4). The discrimination of the model after internal validation was c-index of 0.72. Conclusions: A small percentage of R/M HNSCC pts treated with ICI have good long-term survival outcomes. In a larger cohort, we internally validated the utilization of a simple, inexpensive and widely accessible nomogram based on clinical and laboratory variables which can predict OS in this patient population.