Bridging Access Gaps in Radiotherapy: Knowledge-Based Planning in a Resource-Constrained Clinical Setting.

BACKGROUND/PURPOSE: Radiotherapy access in low- and middle-income countries (LMICs) is severely limited by shortages of trained personnel, high patient volumes, and large waiting times for manual treatment planning. Knowledge-based planning (KBP) has shown promise in high-income settings, yet its feasibility in resource-constrained environments remains underexplored. This study aimed to evaluate the feasibility of adapting a KBP model developed at a high-income institution for clinical use at Liga Nacional Contra el Cáncer (LNCC) in Guatemala for gynecological radiotherapy planning.

MATERIALS AND

METHODS

A RapidPlan KBP model (Varian-Siemens Healthineers, Palo Alto, CA, USA) originally developed at Washington University in St. Louis using data from over 150 patients was adapted and augmented with 118 gynecological cases from the Instituto Nacional de Cancerología (INCAN), reflecting local contouring variability and clinical heterogeneity. Cases were categorized into three planning groups based on planning target volume (PTV) geometry: pelvic only, pelvic with inguinal nodes, and pelvic with inguinal and paraaortic nodes. A validation cohort of 25 gynecological cancer patients was used to compare KBP-generated plans against manual plans produced by four dosimetrists (two junior, two senior). Plans were evaluated using a dosimetric scorecard tool adapted from Varian Medical Affairs, assessing PTV coverage, organ at risk (OAR) sparing (bladder, rectum, and femoral heads), total planning time, plan quality efficiency score (total score/planning time), gamma passing rates (3%/2 mm), and monitor units.

RESULTS

Overall plan quality was comparable between KBP-generated and manually generated plans across all planners, with no systematic differences in target coverage or OAR sparing. The KBP model produced the highest scores for rectal dose metrics compared to manual plans. All plans achieved similar gamma passing rates (mean 99.6%). The mean monitor unit was 264 ± 57 MU across all planners. Importantly, the KBP model significantly reduced plan creation time, resulting in higher plan quality efficiency scores, demonstrating that equivalent clinical quality can be achieved more rapidly.

CONCLUSION

KBP can be successfully adapted from a high-income institution for use in an LMIC setting, maintaining clinical plan quality while substantially reducing planning time. This approach supports increased throughput and more equitable access to high-quality radiotherapy in resource-constrained environments, even in the presence of contouring variability.