30-Second Takeaway
- Margin-negative resection remains the primary determinant of outcome in locoregionally recurrent rectal cancer.
- Real-world data require systematic fitness assessment before oncology RCT emulation.
- Multimodal ML can predict acute dermatitis after hypofractionated breast RT but needs external validation.
Week ending June 6, 2026
Grand Rounds: Selected recent evidence for radiation oncologists
ARS guideline: margin-negative resection drives outcomes in locoregionally recurrent rectal cancer
This multidisciplinary guideline synthesizes 116 trials to frame management of locoregionally recurrent rectal cancer. The ability to achieve a margin-negative surgical resection is identified as the ultimate determinant of survival and local control. Preoperative systemic therapy, radiation, or both can downsize tumors and improve likelihood of margin-negative resection. An individualized, multidisciplinary combined-modality approach is recommended; the guideline does not mandate major practice changes.
CARE Initiative: feasibility limits for oncology RCT emulation using EHR-based RWD
The CARE Initiative evaluated six EHR-based RWD sources against 23 candidate oncology RCTs and 54 RCT-RWD combinations. Only nine combinations advanced to detailed feasibility and three reached protocol-development readiness. Fitness failures were driven by complex eligibility, biomarker capture, performance status, and outcome ascertainment. Authors recommend systematic, stagewise feasibility assessments and data linkage or custom abstraction to improve fitness.
Multimodal ML predicts acute dermatitis after IMRT hypofractionation in early-stage breast cancer
In a single-center cohort of 160 patients, a DS-LightGBM model combined clinical, radiomic, dosiomic, and deep features to predict acute dermatitis. The model achieved strong internal performance: multiclass test accuracy 0.87 and ROC-AUC 0.95, binary accuracy 0.89 and ROC-AUC ~0.88–0.90. SHAP analysis implicated high-dose dosiomic subvolumes and deep imaging features as leading predictors. External validation and prospective testing are required before clinical deployment.
References
Numbered in order of appearance. Click any reference to view details.
Additional Reads
Optional additional studies from this edition.