Comparative effectiveness of proton versus photon chemoradiotherapy for patients with private insurance.

Abstract

7049 Background: Proton therapy may increase the tolerability/efficacy of concurrent chemoradiotherapy (CRT) but is controversial & generally not covered by private insurers. There is little data on the comparative effectiveness (CE) of proton vs photon CRT among private insurance pts to guide payers on proton coverage policies. Methods: We conducted a CE study of adult non-metastatic cancer pts with private insurance treated with curative-intent proton vs photon CRT from 2011-2016 at Penn. The choice of radiation modality was heavily influenced by the insurer’s proton coverage policy. Data on adverse events (AEs) & survival were gathered prospectively using standardized templates. Primary endpoint was 90-day AEs associated with unplanned hospitalizations (CTCAEv4 grade ≥3 AEs). Secondary endpoints included 90-day grade ≥2 AEs, decline in ECOG performance status (PS) during treatment, disease-free survival (DFS) & overall survival (OS). Modified Poisson regression models with inverse propensity score weighting were used for adverse event outcomes. Weighted Cox proportional hazards models were used for survival outcomes. Propensity scores were estimated using an ensemble machine-learning approach. P<0.01 was significant. Results: 920 pts were included (178 proton/742 photon), with H&N(25 proton/296 photon); CNS(44/128); lung(41/120); upper GI(34/78) & lower GI/GYN(34/120). Median age was 57. Race, comorbidity score, BMI, baseline AEs & baseline PS were similar (p>0.05 for all). 11.2% of proton pts had grade ≥3 AE’s vs 26.8% of photon pts. On propensity score weighted-analyses, proton CRT was associated with significantly lower relative risk (RR) of 90-day grade ≥3 AEs (RR 0.51, 95%CI 0.32-0.81, p<0.01). 90-day grade ≥2 AE’s (RR 0.91, 95%CI 0.83-0.99, p=0.03); decline in PS (RR 0.85, 95%CI 0.70-1.04, p=0.11); DFS (HR 0.64, 95%CI 0.27-1.52, p=0.31) & OS (HR 0.53, 95%CI 0.18-1.52, p=0.24) favored protons. Sensitivity analysis showed that a substantial imbalance in an unmeasured confounder would be needed to alter the significance of the primary outcome. Proton accepting insurance status was not associated with a difference in 90-day grade ≥3 AE’s (RR 1.02, 95%CI 0.95-1.10, p=0.54) for pts treated with photon CRT (608 with non-proton accepting insurance & 134 with proton-accepting insurance). Conclusions: In adults with private insurance, proton CRT was associated with significantly reduced acute grade ≥3 AE’s with similar DFS & OS. Proton-accepting insurance status was not associated with better health outcomes when adjusting for RT modality.