A radiomic approach to predicting nodal relapse and disease-specific survival in patients treated with stereotactic body radiation therapy for early-stage non-small cell lung cancer.

Purpose To describe the possibility of building a classifier for patients at risk of lymph node relapse and a predictive model for disease-specific survival in patients with early stage non-small cell lung cancer. Methods A cohort of 102 patients who received stereotactic body radiation treatment was retrospectively investigated. A set of 45 textural features was computed for the tumor volumes on the treatment planning CT images. Patients were split into two independent cohorts (70 patients, 68.9%, for training; and 32 patients, 31.4%, for validation). Three different models were built in the study. A stepwise backward linear discriminant analysis was applied to identify patients at risk of lymph node progression. The performance of the model was assessed by means of standard metrics derived from the confusion matrix. Furthermore, all textural features were correlated to survival data to build two separate predictive models for progression-free survival (PFS) and disease-specific survival (DS-OS). These models were built from the features/predictors found significant in univariate analysis and elastic net regularization by means of a multivarate Cox regression with backward selection. Low- and high-risk groups were identified by maximizing the separation by means of the Youden method. Results In the total cohort (77, 75.5%, males; and 25, 24.5%, females; median age 76.6 years), 15 patients presented nodal progression at the time of analysis; 19 patients (18.6%) died because of disease-specific causes, 25 (24.5%) died from other reasons, 28 (27.5%) were alive without disease, and 30 (29.4%) with either local or distant progression. The specificity, sensitivity, and accuracy of the classifier resulted 83.1 & x202f;+/- 24.5, 87.4 & x202f;+/- 1.2, and 85.4 & x202f;+/- 12.5 in the validation group (coherent with the findings in the training). The area under the curve for the classifier resulted in 0.84 & x202f;+/- 0.04 and 0.73 & x202f;+/- 0.05 for training and validation, respectively. The mean time for DS-OS and PFS for the low- and high-risk subgroups of patients (in the validation groups) were 88.2 & x202f;month & x202f;+/- 9.0 & x202f;month vs. 84.1 & x202f;month & x202f;+/- 7.8 & x202f;month (low risk) and 52.7 & x202f;month & x202f;+/- 5.9 & x202f;month vs. 44.6 & x202f;month & x202f;+/- 9.2 & x202f;month (high risk), respectively. Conclusion Radiomics analysis based on planning CT images allowed a classifier and predictive models capable of identifying patients at risk of nodal relapse and high-risk of bad prognosis to be built. The radiomics signatures identified were mostly related to tumor heterogeneity.