Clinical prognostic factors in patients with recurrent or metastatic carcinoma of the head and neck treated with immune checkpoints therapies.

Abstract

e17530 Background: Immunotherapy (IT) provides advantages in a small proportion of recurrent / metastatic head and neck squamous cell carcinoma (RM HNSCC) patients (pts). Predictive biomarkers are still an unmet need. We aimed to determine if known prognostic clinical factors in RM HNSCC as well as new variables hold prognostic value in the IT era. Methods: We conducted a retrospective analysis of a cohort of 106 RM HNSCC pts enrolled into IT trials at National Cancer Institute of Milan-Italy between Oct 2014 - Jul 2018. Using Log Rank test and Cox proportional hazard model we computed univariate and multivariate analysis (UVA; MVA) of demographics, clinical and biological data to assess their prognostic value for progression free survival (PFS) and overall survival (OS). Results: Ninety-three pts (M 75; F: 18) were included in the final analysis with a median follow up of 20.8 months. The majority (96%) were treated in platinum-refractory setting. Approximately half of pts received anti-PD-1 or anti-PDL-1 monotherapy while the remaining 52% received them in combination with other IT agents. Median OS and PFS were 6.5 (CI 95%: 4.6-10.4) and 2.1 months (CI 95%: 1.8-2.2), respectively. Statistically significant factors at UVA for PFS were smoking history - both pos vs neg and < vs ≥ 30 PY, disease subsite, weight loss > 5%, Karnosfky performance status (PSK), occurrence of any and G3-G4 drug related (DR) toxicities, neutrophil to lymphocytes ratio (NLR). Without stratification for disease subsite, MVA showed independent predictive value of prolonged PFS for positive smoking history (HR 0.5, p 0.02) and any DR toxicities (HR 0.43, p < 0.01). Several factors reached significance for OS in UVA (smoking history and PY, weight loss > 5%, ECOG PS, PSK, occurrence of any toxicities, CRP, baseline C-reactive protein, antibiotic use, steroid use in the 30 days before IT). Of these, PSK (HR 0.95, p 0.01) and any DR toxicities (HR 0.3, p < 0.01) resulted to be predictive in the MVA. Conclusions: Clinical parameters, especially pretreatment PSK, smoking history and occurrence of any DR toxicities, appear to be strong predictors of outcome in RM HNSCC pts treated with IT. These results should be validated in an independent cohort.