Composite treatment plans using BED distributions were successfully generated with appropriate anatomic accuracy (Fig. 1). This, in turn, enabled composite BED3 evaluation of several OARs, such as the spinal cord, proximal tracheobronchial tree, great vessels, brachial plexus, chest wall, heart, esophagus, and total lung. Two representative cases of the physics Gy-to-BED3 conversion using the re-RT planning CT are presented in Fig. 2.
In total, 38 patients received initial SABR followed by re-SABR (n = 14) or fractionated RT with chemotherapy (n = 24). Characteristics of the population are shown in Table 1. Notably, all patients had T1-2N0 NSCLC at initial diagnosis, 71% of re-SABR cases and 67% of chemoradiation (CRT) subjects had adenocarcinoma, and recurrence was confirmed by biopsy in 50% of re-SABR patients and 100% of CRT patients. The remaining recurrences were confirmed by dedicated diagnostic imaging. The median time to recurrence from initial SABR was 20 months for re-SABR patients and 16 months for CRT after SABR. The median follow-up time from the date of completion of re-RT was 36 months for re-SABR patients and 18 months for CRT after SABR patients. The estimated 3-year OS from the time of recurrence was 63% for re-SABR patients and 35% for CRT after SABR patients.
Overall, there were no instances of grade 4–5 events in any patient who underwent re-RT. Of the re-SABR population, the overall rate of higher-grade toxicities was low; there was one case each of grade 3 chest wall pain (7%, Fig. 2a), pneumonitis (7%, Fig. 2a), and dyspnea (7%). Other grade 2 events included brachial plexopathy (n = 1, 7%), rib fracture (n = 3, 21%), chest wall pain (n = 1, 7%), pneumonitis (n = 1, 7%), and dyspnea (n = 3, 21%). It was not possible to determine whether dyspnea was attributable to natural progression of underlying lung disease (e.g., chronic obstructive pulmonary disease) or re-treatment. Table 2 lists BED3 dosimetric parameters for OARs in all patients who experienced these toxicities.
CRT patients also demonstrated an acceptably low rate of adverse effects with no grade 4–5 toxicities reported. The only grade 3 events were two cases of pneumonitis (8%, Fig. 2b) and dyspnea (8%). The grade 2 toxicities were rib fracture (n = 2, 8%), chest wall pain (n = 3, 13%), pneumonitis (n = 1, 4%), and dyspnea (n = 2, 8%). Again, ascertainment of the cause of dyspnea was not possible and may have been related to treatment, natural progression of underlying lung disease, or other factors. Dosimetric parameters in BED3 for patients who experienced these toxicities are listed in Table 2.
Cumulative BED3 doses for patients who experienced grade 2–3 events are reported in Table 2. Cumulative BED3 doses for the entire patient cohort are presented in Table 3, including details of cumulative thoracic OAR BED3 parameters associated with toxicities not exceeding grade 3.
Lastly, for each case of grade 3 toxicity (n = 5), we evaluated the predicted doses to the corresponding OARs based on simple summation of the nominal doses versus the BED3 dose sum. It was noted that, in two (40%) instances, the former reported lower doses to OARs as compared to the latter (Fig. 2), suggesting that BED3-based dose sum planning may better predict higher-grade adverse events. These data suggest that BED3 dose sums may have higher utility in establishing dose-volume constraints as compared to nominal dose sums and may allow clinicians to anticipate long-term re-RT toxicities in a more accurate fashion.