Metabolic Risk Factors for the Development of Radiation Pneumonitis in Stage III Non-Small Cell Lung Cancer (NSCLC)

Abstract

Radiation pneumonitis (RP) is a dose-limiting toxicity associated with radiotherapy in the treatment of non-small cell lung cancer (NSCLC). Several studies have linked diabetes as a risk factor for development of radiation-induced lung injury and statin medications as a protective parameter. We aimed to further investigate these findings and explore other metabolic risk factors for the development of RP. After IRB approval, we retrospectively evaluated 162 patients with stage III NSCLC treated with definitive chemoradiation at our institution between 2001 and 2014. Chemotherapy was delivered as couplet therapy with the majority of patients receiving carboplatin-paclitaxel (141/162, 87.2%). Patient information including age, gender, comorbidities, medications, serum glucose values, toxicities, disease, and treatment parameters were collected from patients’ medical records. Serum glucose values spanning from 90 days prior to 90 days post-radiation therapy (n=2870) were collected. We then categorized values into pre-, during, and post-radiotherapy; defined as within 90 days prior to, during, and within 90 days post-radiotherapy. Patients were then stratified into groups achieving serum glucose values ≥150, ≥175, and ≥200 mg/dL at each of the preceding time points. Chi-Squared analyses were used to analyze the effect categorical variables had on the development of RP in our study population. Parameters achieving significance were selected for analysis against pulmonary toxicity grade. We further used a binary logistic regression model to assess how continuous variables may affect the development of RP. The median prescribed dose was 72 Gy (range= 54-84). The overall rate of RP was 17.9% (n=29). In our dataset, 3/46 (6.5%) diabetic patients developed RP vs. 22/101 (21.8%) patients without diabetes (p=0.022). The rate of radiation-induced lung injury in insulin users was 3/42 (7.1%) vs. 22/105 (21.0%) in non-users (p=0.044). Both diabetes (2.2% vs. 14.9%, p=0.022) and insulin (2.4% vs. 14.3%, p=0.036) were associated with lower incidence of RP requiring medical intervention (grade ³2). Diabetes was further correlated with reduced incidence of grade ³3 RP (0% vs. 9.1%, p=0.037). Serum glucose values at the aforementioned time points were not associated with RP. Neither statins nor other diabetic medications correlated with RP risk. Gross tumor volume (GTV), planning target volume (PTV), radiation technique, radiation dose, dose per fraction, and treatment duration were not significant predictors of our primary endpoints. Finally, patients who developed RP had similar rates of mortality, loco-regional recurrence, and distant metastasis to those without. In contrast to prior data, we identify both diabetes and insulin therapy as protective factors against the development of RP in the treatment of NSCLC. These results should be confirmed by subsequent high-quality prospective studies.