Nominal plan robustness may predict plan degradation in proton therapy for oropharyngeal head‐and‐neck cancer

BACKGROUND: Proton therapy for head-and-neck (HN) cancer offers superior organ-at-risk sparing compared to photon therapy, but is challenged by frequent anatomical changes during treatment. These changes need to be monitored with routine verification CTs (vCTs), which are used to trigger adaptive replans when deemed necessary by the clinical team. PURPOSE: To investigate whether nominal plan robustness evaluation (RE) data-specifically the magnitude and spatial characteristics of high-dose regions (hotspots)-can predict the development of clinically significant hotspots on verification CTs (vCTs) and guide planning strategies that minimize the need for adaptive replanning. METHODS: This retrospective study analyzed 46 patients with p16-positive oropharyngeal cancer treated with proton therapy. Clinical treatment plans were robustly evaluated using 12 uncertainty scenarios combining 3 mm setup and ± 3.5% range errors. Each plan was recalculated on periodic vCTs throughout the treatment course to assess plan degradation. The maximum RE hotspot magnitude and location were compared with vCT hotspot characteristics. A subset of five cases underwent proof-of-concept replanning to reduce RE hotspots and assess downstream vCT dose effects. RESULTS: Patients requiring adaptive replanning due to vCT hotspots had significantly higher RE hotspot magnitudes of the nominal plan compared to those who did not (p = 0.008). For replanned cases, higher RE hotspots were moderately correlated with closer proximity of RE and vCT hotspots (r = -0.59, p = 0.009). Across all patients, a modest correlation (r = 0.58, p < 0.001) was observed between RE and vCT hotspot magnitudes. Further, the rate of plan degradation over the course of treatment via hotspot formation was found to increase with increasing RE hotspot magnitude. Replanning to reduce RE hotspots led to an average 5.6% reduction in vCT hotspot dose for the five patients studied, suggesting that reducing RE hotspots may reduce the frequency of replans. CONCLUSIONS: Nominal plan robustness evaluation is predictive of both the magnitude and location of hotspots observed on vCTs, and plans with higher RE hotspots tend to degrade faster over the treatment course. Minimizing RE hotspots during treatment planning may reduce the need for adaptive replanning and enhance clinical workflow efficiency.