Reaffirming Metastatic Risk Profiles in Human Papillomavirus-related Oropharyngeal Cancer Following Definitive Radiation Therapy With or Without Chemotherapy

Abstract

A prior publication from our group identified profiles of patients (pts) with differing risk of distant metastasis (DM), the main form of failure for HPV-related (HPV[+]) oropharyngeal cancer (OPC). These profiles are currently influencing the design of trials in HPV(+) OPC. The aim of this study is to reaffirm DM risk groups following definitive radiation therapy (RT) +/- chemotherapy (CTx) in a substantially expanded HPV(+) OPC cohort. All p16-confirmed newly diagnosed HPV(+) OPC patients treated with RT +/- CTx between 2000 and 2012 were included. Overall survival (OS), locoregional control (LRC), distant control (DC), and grade 3 to 4 late toxicity (LT) were estimated. Multivariable analysis (MVA) identified predictors for DC and OS. Recursive partitioning analysis (RPA) derived low and high DM risk groups. Within the low and high DM risk subgroup identified by RPA, we compared DC between RT with or without CTx to identify potential candidates for omission of CTx. A total of 757 HPV(+) (stage I, II, III, and IV: 8, 34, 93, and 622, respectively) pts were identified, including 605 (80%) men and 389 (51%) >10 pack-per-year (PY) smokers. Median age was 58 years. Tonsil or tongue base primary tumors were detected in 718 (95%) pts. Concurrent CTx (cisplatin 100 mg/m2 every 3 weeks) with RT was given in 382 (50%) cases. Median follow-up was 5.1 years. A total of 40 local, 35 regional, and 98 DM were identified. Five-year OS, LRC, DC, and LT were 76%, 92%, 87%, and 20%, respectively. MVA identified T4 (hazard ratio [HR] = 1.95, P < .01) or N2c-3 (HR = 3.5, P < .01), and absence of CTx (1.7, P = .03) as DM predictors; neither smoking PYs (HR = 0.99, P = .41) nor age (HR = 1.01, P = .43) were predictive for DM, but they were predictive for OS (age: HR = 1.02; smoking: 1.01, both P < .01). RPA divided the entire cohort into low-risk (T1–3N0–2b, n = 441) and high-risk (T4 or N2c–3, n = 316) subgroups with 5-year DC of 94% versus 76% and OS of 84% versus 66%, respectively (both P < .01). In the low-risk subgroup, DC was similar between RT (N = 247) versus RT + CTx (n = 194) (95% vs 93%, P = .67). Five-year DC rates by smoking ≤10 (n = 47) versus >10 (n = 52) PYs were also similar in the T1–3N2b (93% vs 90%, P = .53) subset. In the high-risk subgroup, DC was lower in the RT (n = 128) versus CRT (n = 188) (65% vs 83%, P < .01). This expanded cohort study (the sample doubled from 382 in the previous publication to 757 in the current cohort) confirms that DM is the main form of treatment failure for HPV(+) OPC pts. The T1–T3N0–N2b subgroup has low DM risk, which is not influenced by smoking PY, although smoking adversely affects survival for all groups. While these findings require prospective confirmation, the low DM risk subgroup represents a candidate for deintensification approaches, and high-risk patients merit consideration of new strategies.