Dosimetric Outcomes of SBRT to Ultra-central Lung Tumors: Lessons from the SUNSET trial

The BLINDED FOR REVIEW trial investigated the maximum tolerated dose for ultracentral lung tumors treated with stereotactic body radiotherapy (SBRT). Here we report a spatial and dosimetric secondary analysis of the treatment plans and assess relationships between doses to targets, organs-at-risk (OARs) and clinical outcomes. Five institutions enrolled patients with ultracentral lung cancer, cT1-3N0M0, and all received 60 Gy in 8 fractions. Maximum dose (Dmax) was limited to 120% of prescription. Planning datasets and treatment plans were imported into a central repository. Univariable logistic and Cox proportional hazards regression modelling were performed to identify significant dosimetric predictors for related grade ≥ 2 adverse events, overall survival (OS) and local control (LC). Thirty patients were included in this analysis. At median follow-up of 36.5 months, 11 patients experienced grade ≥2 toxicity. The PTV overlapped with 58 central OARs of which airway was most common (n=26). The mean Dmax was 69.3 Gy (range: 63.8-72.0 Gy); all were within the ITV. Mean ± SD PTV D98 was 56.1 ± 7.8 Gy, while D0.1cc of PBT, esophagus and pulmonary artery were 53.1 ± 12.6 Gy, 26.5 ± 10.5 Gy and 57.0 ± 8.6 Gy, respectively. On regression analysis, the combined overlap volume (cc) of the two primary overlapping OARs with PTV was associated with inferior LC (hazard ratio [HR]: 2.86, p=0.012); however, it was not associated with increase in grade ≥2 adverse events (odds ratio [OR]: 1.17, p=0.49). There was no association between OAR doses (D1cc and D0.1cc) with toxicity. PTV undercoverage (D98) was not associated with worse LC (HR per 5 Gy: 1.54, p=0.68), however lower PTV coverage was significantly associated with reduced overall survival for D98 (HR: 0.65, p=0.014) and D95 (HR per 5 Gy: 0.66, p=0.035). Within the dose constraints used in the trial, there was no relationship identified between OAR doses and toxicity. Local control decreased with increasing overlap of PTV with OARs, however, this was not associated with dosimetric under-coverage of the target.