Comparison of Oncologic Outcomes and Treatment-Related Toxicity of Carbon Ion Radiotherapy and en Bloc Resection for Sacral Chordoma

Abstract

Importance: Maximal resection is the preferred management for sacral chordomas but can be associated with unacceptable morbidity. Outcomes with radiotherapy are poor. Carbon ion radiotherapy (CIRT) is being explored as an alternative when surgery is not preferred. Objective: To compare oncologic outcomes and treatment-related toxicity of CIRT and en bloc resection for sacral chordoma. Design, Setting, and Participants: Univariable logistic regression was performed to evaluate the association between treatment type and oncologic and toxicity outcomes in this retrospective cohort study. Nearest-neighbor propensity score matching was used to match the CIRT cohort with the en bloc resection cohort and 10 National Cancer Database (NCDB) cohorts separately, with the objective of obtaining more homogeneous cohorts when comparing treatments. Patient- and tumor-related characteristics from 2 institutional cohorts were collected for patients diagnosed with sacral chordomas between April 1, 1994, and July 31, 2017. The NCDB was queried for data on patients with sacral chordoma from January 1, 2004, to December 31, 2016, as a comparator in overall survival (OS) analyses. Data analysis was conducted from February 24, 2020, to January 16, 2021. Exposures: En bloc resection, incomplete resection, photon radiotherapy, proton radiotherapy, and CIRT. Main Outcomes and Measures: Overall survival was estimated using the Kaplan-Meier method and compared using the Cox proportional hazards model. Peripheral motor nerve toxic effects were scored using Common Terminology Criteria for Adverse Events, version 4.03. Results: A total of 911 patients were included in the study (NCDB: n = 669; median age, 64 [IQR, 52-74] years; 410 [61.3%] men; CIRT: n = 188; median age, 66 [IQR, 58-71] years; 128 [68.1%] men; en bloc surgical resection: n = 54; median age, 53.5 [IQR 49-64] years, 36 [66.7%] men). Comparison of the propensity score-matched institutional en bloc resection and CIRT cohorts revealed no statistically significant difference in OS (CIRT: median OS, 68.1 [95% CI, 44.0-102.6] months; en bloc resection: median OS, 58.6 [95% CI, 25.6-123.5] months; P =.57; hazard ratio, 0.71 [95% CI, 0.25-2.06]; P =.53). The CIRT cohort experienced lower rates of peripheral motor neuropathy (odds ratio, 0.13 [95% CI, 0.04-0.40]; P