Fig. 2

Distribution curve of the cumulative incidence of GARD from 231 STS samples. GARD pathologic response prediction based on historic (8%) and institutional (18%) FPR rates to standard neoadjuvant radiotherapy (50 Gy in 25 fractions), with GARD thresholds of 31.37 and 22.96, respectively. Note the effect that dose escalation with 60 Gy (red) and 70 Gy (green) in 25 fractions has on the percent of patients that are predicted beyond the thresholds. Since both GARD targets are after the linear region in the curve, a uniform dose increase quickly impacts the number of patients that achieve each of the GARD target values. Since habitats are dichotomized based on the gross tumor volumes median value, then 50% of the volume will receive standard dose and the other half will receive dose escalation of either 60 or 70 Gy. Therefore, the probability of achieving a FPR is 8 to 18% for the standard half and an average of 15.2–60.2% to the escalated half of the GTV, would predict for a > 24.3% estimated FPR for the cohort. A tripled FPR rate (8–24%) is a modest estimate that assumes complete dose conformality, with neighboring habitats adjacent to one another, this would be a conservative minimum FPR increase we expect to see clinically. The estimated FPR is assuming the same probability of response for each habitat, but knowing that radioresistant hypoxic regions often require > 30% higher dose for response [40], the radiomic habitat directed approach may better identify the regions that would most benefit from dose escalation, therefore improving overall response