Predictors of Pneumonitis after Lung Cancer Radiotherapy
McFarlane M, Hochstedler K, Laucis AM, Sun Y, Chowdhury A, Matuszak MM, Hayman JA, Bergsma DP, Boike TP, Kestin LL, Movsas B, Grills IS, Dominello MM, Dess RT, Schonewolf CA, Spratt DE, Pierce LJ, Paximadis P, Jolly S, and Schipper M. Predictors of Pneumonitis after Lung Cancer Radiotherapy. International Journal of Radiation Oncology Biology Physics 2020; 108(3):S139.
International Journal of Radiation Oncology Biology Physics
Purpose/Objective(s): Multiple factors influence the risk of developing pneumonitis after radiotherapy for lung cancer, but few resources exist to guide clinicians in predicting risk in an individual patient using modern techniques. We analyzed toxicity data from a state-wide consortium to develop an integrated pneumonitis risk model.
Materials/Methods: All patients received radiation therapy for stage II-III NSCLC between April 2012 and July 2019. Data were prospectively collected from 24 academic and community clinics participating in a state-wide quality consortium. Pneumonitis within 6 months of treatment was graded by local practitioners (no centralized review). Weighted, univariate regression models were used to describe the risk of pneumonitis toxicity as a function of dosimetric and clinical covariates. Pneumonitis was modeled as either G ≥ 2 versus G ≤ 1 or as G ≥ 3 versus G ≤ 2. We used a stepwise modeling procedure to build a multivariable model. AUC values were calculated for each model to quantify the ability of each covariate to discriminate between patients who did or did not experience toxicity.
Results: Our analysis included 1302 patients, equally divided between male and female, with a median age of 67 years. Median comorbidity count was 1 and over 48% of patients had ≥ 2 comorbidities. 68% of patients had an ECOG performance status of 0-1. The overall rate of pneumonitis in the 6 months following RT was 16% (215 cases). 6% (92 cases) were G2+ and 1% (13 cases) were G3+. Adjusting for incomplete follow-up, estimated rates for G2+ and G3+ were 32% and 2%, respectively. In univariate analyses, V5, V10, V20, V30, and Mean Lung Dose (MLD) were positively associated with G2+ pneumonitis risk (OR 1.34, 1.34, 1.79, 1.48 per 10% increase, and 1.11 per 1 Gy increase respectively), while current smoking status was associated with lower odds of pneumonitis (OR 0.311). G2+ pneumonitis risk of > = 22% was independently predicted by MLD of > = 20 Gy, V20 of > = 35%, and V5 of > = 75% which correspond to commonly-used planning constraints. In multivariate analyses, the lung V5 metric remained a significant predictor of G2+ pneumonitis even when controlling for MLD, despite being closely correlated (coefficient 0.743). For G3+ pneumonitis, MLD (OR 1.25 per Gy) and V20 (OR 2.59 per 10% increase) were statistically significant predictors. Number of comorbidities was an independent predictor of G3+, but not G2+ pneumonitis (OR 1.61 per comorbidity).
Conclusion: We present a large, prospectively-collected dataset evaluating pneumonitis risk after definitive radiation therapy for lung cancer. We incorporate comorbidity burden, smoking status, and dosimetric parameters in an integrated risk model. Low-grade pneumonitis is associated with MLD, V5, and V20, and negatively associated with smoking. High-grade pneumonitis is positively associated with MLD, V20, and comorbidity burden. These data may guide clinicians in assessing pneumonitis risk in individual patients.