MATERIALS AND METHODS: A literature review was performed following the PRISMA guidelines. Systematic searches were performed in PubMed, Scopus, Cochrane and Embase databases from the earliest record up to September 2022. Related studies on deep learning models for radiotherapy toxicity prediction were selected based on predefined PICOS criteria.
RESULTS: Fourteen studies of radiotherapy-treated patients on different types of cancer [prostate (n=2), HNC (n=4), liver (n=2), lung (n=4), cervical (n=1), and oesophagus (n=1)] were eligible for inclusion in the systematic review. Information regarding patient characteristics and model development was summarized. Several approaches, such as ensemble learning, data augmentation, and transfer learning, that were utilized by selected studies were discussed.
CONCLUSION: Deep learning techniques are able to produce a consistent performance for toxicity prediction. Future research using large and diverse datasets and standardization of the study methodologies are required to improve the consistency of the research output.
MATERIALS AND METHODS: We searched PubMed and Scopus electronic databases to identify original studies reporting toxicity outcomes following PBT of primary NPC. Quality assessment was performed using NIH's Quality Assessment Tool. Reports were extracted for information on demographics, main results, and clinical and dose factors correlates. Meta-analysis was performed using the random-effects model.
RESULTS: Twelve studies were selected (six using mixed particle-photon beams, five performed comparisons to photon-based therapy). The pooled event rates for acute grade ≥2 toxicities mucositis, dermatitis, xerostomia weight loss are 46% (95% confidence interval [95% CI]-29%-64%, I2 = 87%), 47% (95% CI-28%-67%, I2 = 87%), 16% (95% CI-9%-29%, I2 = 76%), and 36% (95% CI-27%-47%, I2 = 45%), respectively. Only one late endpoint (xerostomia grade ≥2) has sufficient data for analysis with pooled event rate of 9% (95% CI-3%-29%, I2 = 77%), lower than intensity-modulated radiotherapy 27% (95% CI-10%-54%, I2 = 95%). For most endpoints with significant differences between the PBT and photon-based therapies, PBT resulted in better outcomes. In two studies where dose distribution was studied, doses to the organs at risk were independent risk factors for toxicities.
CONCLUSION: PBT may reduce the risk of acute toxicities for patients treated for primary NPC, likely due to dose reduction to critical structures. The pooled event rate for toxicities derived in this study can be a guide for patient counseling.