Search

Simple Summary: Studies comparing performances of machine learning (ML) methods in building predictive models of toxicity in RT are rare. Thanks to the availability of a large cohort (n = 1314) of breast cancer patients homogeneously treated with tangential fields, different ML approaches could be compared. This work shows how more complex models typically achieve higher performances. At the same time, for this test case, the importance is given mainly by a few variables, and toxicity can be predicted by simpler models with similar performances. The availability of more individually characterizing features (here partially missing) is expected to have a likely much higher impact than the choice of the best-performing ML/DL approach. Purpose. Different ML models were compared to predict toxicity in RT on a large cohort (n = 1314). Methods. The endpoint was RTOG G2/G3 acute toxicity, resulting in 204/1314 patients with the event. The dataset, including 25 clinical, anatomical, and dosimetric features, was split into 984 for training and 330 for internal tests. The dataset was standardized; features with a high p-value at univariate LR and with Spearman ρ > 0.8 were excluded; synthesized data of the minority were generated to compensate for class imbalance. Twelve ML methods were considered. Model optimization and sequential backward selection were run to choose the best models with a parsimonious feature number. Finally, feature importance was derived for every model. Results. The model's performance was compared on a training–test dataset over different metrics: the best performance model was LightGBM. Logistic regression with three variables (LR3) selected via bootstrapping showed performances similar to the best-performing models. The AUC of test data is slightly above 0.65 for the best models (highest value: 0.662 with LightGBM). Conclusions. No model performed the best for all metrics: more complex ML models had better performances; however, models with just three features showed performances comparable to the best models using many (n = 13–19) features. [ABSTRACT FROM AUTHOR]

Simple Summary: Several phase III randomized trials of moderate hypofractionation, including a higher proportion of high-risk prostate cancer patients treated only to the prostate, failed to demonstrate the superiority of hypofractionated regimens. There is only one randomized phase III trial, of moderately hypofractionated high-dose radiotherapy to the prostate-only versus pelvic irradiation and prostate boost, with a sufficiently long follow-up. It demonstrated better biochemical and disease-free survival when lymph nodal radiotherapy was added. Here we present the 10-year results of our experience based on an Institutional protocol adopted after a phase I–II study, on patients with unfavorable intermediate- (UIR), high- (HR), and very high-risk (VHR) prostate cancer (PCa) treated with pelvic lymph nodal irradiation (WPRT) and moderately hypofractionated high-dose (HD) simultaneous integrated boost (SIB) to the prostate. Prognostic factors for relapse, as well as acute and late gastro-intestinal (GI) and genito-urinary (GU) toxicity were also analyzed. Aims: To report 10-year outcomes of WPRT and HD moderately hypofractionated SIB to the prostate in UIR, HR, and VHR PCa. Methods: From 11/2005 to 12/2015, 224 UIR, HR, and VHR PCa patients underwent WPRT at 51.8 Gy/28 fractions and SIB at 74.2 Gy (EQD2 88 Gy) to the prostate. Androgen deprivation therapy (ADT) was prescribed in up to 86.2% of patients. Results: Median follow-up was 96.3 months (IQR: 71–124.7). Median age was 75 years (IQR: 71.3–78.1). At last follow up, G3 GI–GU toxicity was 3.1% and 8%, respectively. Ten-year biochemical relapse-free survival (bRFS) was 79.8% (95% CI: 72.3–88.1%), disease-free survival (DFS) 87.8% (95% CI: 81.7–94.3%), overall survival (OS) 65.7% (95% CI: 58.2–74.1%), and prostate cancer-specific survival (PCSS) 94.9% (95% CI: 91.0–99.0%). Only two patients presented local relapse. At univariate analysis, VHR vs. UIR was found to be a significant risk factor for biochemical relapse (HR: 2.8, 95% CI: 1.17–6.67, p = 0.021). After model selection, only Gleason Score ≥ 8 emerged as a significant factor for biochemical relapse (HR = 2.3, 95% CI: 1.12–4.9, p = 0.023). Previous TURP (HR = 3.5, 95% CI: 1.62–7.54, p = 0.001) and acute toxicity ≥ G2 (HR = 3.1, 95% CI = 1.45–6.52, p = 0.003) were significant risk factors for GU toxicity ≥ G3. Hypertension was a significant factor for GI toxicity ≥ G3 (HR = 3.63, 95% CI: 1.06–12.46, p = 0.041). ADT (HR = 0.31, 95% CI: 0.12–0.8, p = 0.015) and iPsa (HR = 0.37, 95% CI: 0.16–0.83, p = 0.0164) played a protective role. Conclusions: WPRT and HD SIB to the prostate combined with long-term ADT, in HR PCa, determine good outcomes with acceptable toxicity. [ABSTRACT FROM AUTHOR]