Patient Selection for De-Escalated Radiotherapy for Early-stage HPV+ Oropharynx Cancer Can be Improved by Considering Smoking Cessation: A Multi-Institution Analysis

Abstract

De-intensified treatment strategies for early-stage HPV-positive oropharynx cancer rely on selecting patients with an excellent prognosis. Smoking status clearly affects prognosis. The criteria for enrollment on current cooperative group trials is ≤10 pack-years (pys). More nuance to the pack-year criteria may expand enrollment, improve patient outcomes and ensure trial success. Our hypothesis was that time since smoking cessation, in addition to total exposure (pys), could successfully refine selection criteria. From our IRB-approved matched two-institution database, we identified patients with HPV+ squamous carcinoma of the oropharynx treated definitively with radiation with or without chemotherapy. Only patients with a history of smoking and anatomically eligible for national de-escalation trials were included (AJCC 7th Ed. cT1-2N1-2b or T3N0-2b). Progression-free survival (PFS) was defined as time between initiation of radiation and either recurrence or death. Cessation was measured via continuous variable of time since cessation, with zero representing those smoking at diagnosis. Cox regression with penalized smoothing splines were used to evaluate non-linear effects of cessation. Recursive partitioning analysis (RPA) was used to objectively search for relationships between the two co-linear variables (pys and time since cessation). From a combined database of 1,232 oropharynx patients, we identified 200 who met the strict criteria. Of these, 141 (71%) were former smokers (cessation time >0.05 years) and 59 (29%) were smoking at the time of consultation (cessation time of 0). The median exposure to smoking was 21 pys (IQR 10-40); 26% smoked <10 pys, 24% smoked 10-20 pys and 50% >20 pys. There was a significant association between pys and smoking status (Spearman r2 0.04, p=0.004). Patients were treated to a median of 70.6 Gy (range 65-78 Gy) and 93% received chemotherapy. Median follow-up was 32 months (range 1.3-120). Non-linear Cox regression showed that a time since cessation of <5 years was associated with exponentially worse PFS with no threshold. RPA demonstrated that current smokers performed worse than former. Two-year PFS for current and former (cessation time >0.05 years) at various thresholds are reported in the table. None of the currently smoking cohorts would meet the prespecified endpoint on NRG HN002, but every former cohort would. Pack-years is partially a surrogate for active smoking, which appears to be a stronger prognostic factor. Former smokers, regardless of pys, should be considered for de-intensification studies. The predictive value of current smoking is unclear and should be explored in future studies.Abstract 2946; Table 1CohortN2-year PFS95% CICurrent ≤10 pys1089%71-100%Current ≤20 pys2471%54-94%Current ≤30 pys3372%58-91%Former ≤10 pys42100%NAFormer ≤20 pys7699%96-100%Former ≤30 pys10499%97-100%Former ≤40 pys11398%97-100%Former ≤50 pys12497%95-100% Open table in a new tab