ESTRO 2022 - Abstract Book

S565

Abstract book

ESTRO 2022

and prescription to GTV/CTV volumes and OARs over repeated imaging in order to gain insight between delivered dose, local control and toxicity to optimize treatment protocols.

OC-0634 Prediction models for rectal toxicity after brachytherapy in patients with pelvic cancer

F. Tohidinezhad 1 , Y. Willems 1 , M. Berbee 1 , E. Van Limbergen 1 , F. Verhaegen 1 , A. Dekker 1 , A. Traverso 1

1 Maastricht University Medical Center, Department of Radiation Oncology (Maastro Clinic), Maastricht, The Netherlands

Purpose or Objective Over the last decades, Brachytherapy (BT) with or without supplemental External Beam Radiotherapy (EBRT) has demonstrated excellent effectiveness in localized advanced pelvic cancers, especially in patients with prostate and cervical malignancies. In recent years, quality of life after BT has become a concerning issue among physicians and patients for selecting the optimal treatment modality. Although brachytherapy allows for rapid dose falloff, the rectal wall still receives high doses of radiation due to the proximity to the tumor. Literatures show 5-7% of the patients complain of grade II or higher rectal toxicities. Better knowledge of the dose-toxicity relationship is essential for safe dose escalation to minimize rectal toxicity without impairing therapeutic benefit. This study was aimed to abstract and evaluate the studies which have developed a prediction model for rectal toxicity after BT in patients with pelvic cancer. Materials and Methods To identify relevant studies since 1995, MEDLINE was searched on August 31, 2021 using the terms related to “pelvic cancer sites”, “rectal toxicity”, “prediction model”, and “brachytherapy”. Two independent reviewers screened the citations. The papers were excluded if only a subset of patients received BT, entering/removing predictors were not reported, or mixed outcomes were used as the primary endpoint. Risk of bias associated with methodological conduct was assessed using the Prediction model Risk Of Bias ASsessment Tool (PROBAST). Results Forty-three prediction models featuring 117 distinct predictors (Figure 1) were published between 1998 and 2019. A total of 20 (47%) and 21 (49%) models were developed for prostate and cervical cancers, respectively. One study included elderly rectal cancer patients and one study developed a model for patients with different gynecological cancers. Rectal toxicity varied significantly between studies (3%-39% for prostate and 7%-47% for cervical cancers). Thirty-nine (91%) of the studies performed combined EBRT and BT and 11 (52%) studies included the gynecology cancer patients with the history of chemotherapy. Radiation Therapy Oncology Group (RTOG) and Common Terminology Criteria for Adverse Events (CTCAE) were used in 61% and 19% of the studies as the outcome measuring standard. Regression, support vector machine, and neural network were used in 40 (93%), 2 (5%), and 1 (2%) studies, respectively. Only five studies (12%) reported the area under the receiver operating characteristic curve (0.58-0.91). All studies were judged to be at high risk of bias mainly due to methods of analysis (Figure 2).