Project 2: InHANCE: Radiation Associated Dysphagia

ABSTRACT: Project 2 Radiation-associated dysphagia (RAD) is a leading driver of QOL and a potentially life-threatening adverse treatment sequelae, afflicting more than half of patients treated with curative radiotherapy (RT) for head and neck cancers (HNC). Aspirators are almost 5-times more likely to develop pneumonia than non-aspirators, and pneumonia confers a 42% increased risk of mortality among HNC survivors. RAD is a radiation dose and field dependent phenomena. Despite laudable progress to define dysphagia-aspiration related structures (DARS), their normal tissue complication probabilities (NTCP), and level 1 evidence for their dose constraints (i.e., ISRCTN25458988), critical opportunities remain to advance care by innovating the evidence base for RT field/dose reduction via personalized care pathways and clinical trials. Using the investigators’ one-of-a-kind PRO-ACTIVE trial (NCT03455608) dataset, we propose to first accomplish this using InHANCE P01 core resources to define effect modification of the dysphagia-RT dose-response relationship by pragmatic co-factors like age and multi-modality treatment – factors known to influence in uncertain magnitude the radio-sensitivity and functional reserve or the ability of an individual survivor to compensate for neuromuscular RT injury. The investigators will next exploit the integrated InHANCE P01 infrastructure to for the first time characterize the soft tissue toxicity profiles underlying RAD risk-reduction using image-based RT field reduction by Tc99- lymphoscintigraphy (SPECT-CT) in oropharyngeal cancer. Finally, the investigators will use deep learning to deliver an interpretable AI tool to help investigators more efficiently and reliably clinically implement the image based swallowing grading method we developed called DIGEST (PMC5161634). The long-term goal of this work is to reduce dysphagia burden through clinical adoption of pragmatic imaging innovations. Our central hypothesis is that pragmatic imaging algorithms and image-guided RT field reduction can improve dysphagia care through improved risk stratification, classification, primary prevention of RAD. The objective of the proposed studies is to analyze and augment imaging sets from 4 large existing data sources to: 1) develop risk-stratified normal tissue complication probability (NTCP) models of RAD (Aim 1), 2) deliver image-based swallowing secondary endpoint and lymphedema/fibrosis characterization after SPECT-CT image-guided RT field reduction (Aim 2), and 3) deliver a clinically interpretable automated DIGEST grade for clinical use (Aim 3). Using integrated Core support from the InHANCE P01 Program to process and analyze RT planning CT, MRI, and radiographic MBS imaging studies linked to clinical data, we are uniquely positioned to address these complementary aims. We expect this work to deliver practical quantitative imaging solutions to help real-world clinicians detect: 1) who will get RAD (Aim 1), 2) LEF and RAD mitigation through image-based RT field reduction (Aim 2), and 3) classification of RAD (Aim 3).