Machine learning highlights the deficiency of conventional dosimetric constraints for prevention of high-grade radiation esophagitis in non-small cell lung cancer treated with chemoradiation

José Marcio Luna; Hann-Hsiang Chao; Russel T Shinohara; Lyle H Ungar; Keith A Cengel; Daniel A Pryma; Chidambaram Chinniah; Abigail T Berman; Sharyn I Katz; Despina Kontos; Charles B Simone 2nd; Eric S Diffenderfer

doi:10.1016/j.ctro.2020.03.007

Machine learning highlights the deficiency of conventional dosimetric constraints for prevention of high-grade radiation esophagitis in non-small cell lung cancer treated with chemoradiation

Clin Transl Radiat Oncol. 2020 Mar 24:22:69-75. doi: 10.1016/j.ctro.2020.03.007. eCollection 2020 May.

Affiliations

¹ Department of Radiation Oncology, University of Pennsylvania, Perelman Center for Advanced Medicine, 3400 Civic Center Blvd, Philadelphia, PA 19104, United States.
² Department of Radiation Oncology, Hunter Holmes McGuire Veterans Affairs Medical Center, 1201 Broad Rock Blvd, Richmond, VA 23249, United States.
³ Department of Biostatistics and Epidemiology, University of Pennsylvania, 423 Guardian Dr, Philadelphia, PA 19104, United States.
⁴ Department of Computer and Information Science, University of Pennsylvania, 3330 Walnut St, Philadelphia, PA 19104, United States.
⁵ Department of Radiology, University of Pennsylvania, 3400 Spruce St, Philadelphia, PA 19104, United States.
⁶ Albany Medical College, 43 New Scotland Ave, Albany, NY 12208, United States.
⁷ Department of Radiation Oncology, New York Proton Center, 225 East 126 St, New York, NY 10035, United States.

Abstract

Background and purpose: Radiation esophagitis is a clinically important toxicity seen with treatment for locally-advanced non-small cell lung cancer. There is considerable disagreement among prior studies in identifying predictors of radiation esophagitis. We apply machine learning algorithms to identify factors contributing to the development of radiation esophagitis to uncover previously unidentified criteria and more robust dosimetric factors.

Materials and methods: We used machine learning approaches to identify predictors of grade ≥ 3 radiation esophagitis in a cohort of 202 consecutive locally-advanced non-small cell lung cancer patients treated with definitive chemoradiation from 2008 to 2016. We evaluated 35 clinical features per patient grouped into risk factors, comorbidities, imaging, stage, histology, radiotherapy, chemotherapy and dosimetry. Univariate and multivariate analyses were performed using a panel of 11 machine learning algorithms combined with predictive power assessments.

Results: All patients were treated to a median dose of 66.6 Gy at 1.8 Gy per fraction using photon (89.6%) and proton (10.4%) beam therapy, most often with concurrent chemotherapy (86.6%). 11.4% of patients developed grade ≥ 3 radiation esophagitis. On univariate analysis, no individual feature was found to predict radiation esophagitis (AUC range 0.45-0.55, p ≥ 0.07). In multivariate analysis, all machine learning algorithms exhibited poor predictive performance (AUC range 0.46-0.56, p ≥ 0.07).

Conclusions: Contemporary machine learning algorithms applied to our modern, relatively large institutional cohort could not identify any reliable predictors of grade ≥ 3 radiation esophagitis. Additional patients are needed, and novel patient-specific and treatment characteristics should be investigated to develop clinically meaningful methods to mitigate this survival altering toxicity.

Keywords: Chemoradiation; Intensity-modulated radiation therapy; Machine learning; Non-small cell lung cancer; Proton beam therapy; Radiation esophagitis; Radiation-induced toxicity.