Enhancing Dosimetric Quality in Cervical Cancer HDR Brachytherapy: A Study on Empirically Derived Cost Function Templates

Summary: Delivering quality high dose rate (HDR) brachytherapy for locally advanced cervical cancer requires significant time, expertise, and clinical resources. Volumetric HDR plan optimization often involves multiple iterations from an arbitrarily defined initial optimization template. This dosimetric quality improvement study aims to develop an improved set of initial optimizer metrics for HDR boost planning for cervical cancer using tandem and ring hybrid applicators with interstitial needles. We identified 18 cervical cancer patients treated with a tandem and ring hybrid applicator and 4-6 needles. The final cost function values from the optimizer were obtained for CTV High, bladder, rectum, urethra, small bowel, and sigmoid colon. For each structure, the most common cost function metrics were averaged across plans to create a more robust starting point. Five similar patients were retrospectively re-planned using both the institution's usual template and the new empirically generated template. The clinical goal pass rate and dosimetry for each structure were compared between plans using a paired 2-tailed T-test. All doses are EQD2 with a/b of 3 and 10 for normal structures and CTV, respectively. The new empirically derived cost function template generally asked for more CTV High coverage and lower dose to organs at risk (OARs). Plans generated using the new cost function metrics were significantly more likely to meet clinical goals than plans generated using our institution’s usual template (88% vs 78%, p = 0.016). CTV High coverage generally improved with the new template with D90% averaging 88.22 Gy vs 87.92 Gy (p = 0.0025). Non-bladder OAR dose was not statistically different between templates. Interestingly the bladder D2cc trended slightly higher with the new template averaging 79.76 Gy vs 78.07 Gy (p = 0.10) and produced one new major goal violation in the validation set. This was remedied by adding a second empirically derived bladder metric to the cost function seen in several plans of the initial data set (D1cc < 77.85 Gy). Plans produced with these metrics had lower bladder doses than the usual template (p = 0.003) without compromising target coverage. An empirically derived cost function template outperformed this institution’s usual template created with clinical planning expertise alone. This approach produces higher quality starting plans, easing HDR treatment logistics. Future work could assess time savings and generalization to other disease sites. Roman Travis, MD (Presenting Author) - University of Alabama at Birmingham Department of Radiation Oncology; Samantha Simiele, Ph.D., DABR (Co-Author) - University of Alabama Department of Radiation Oncology; Eric Simiele, Ph.D. (Co-Author) - University of Alabama at Birmingham Department of Radiation Oncology; Samuel Marcrom, MD (Co-Author) - UAB Radiation Oncology