STARS and ROSEL pooled analysis

STARS and ROSEL both randomised patients with operable clinical stage I non-small-cell lung cancer to stereotactic ablative radiotherapy or lobectomy, and both closed early because patients and surgeons would not accept randomisation. Pooling them gave 58 patients, 31 assigned to radiotherapy and 27 to surgery.

At a median 40.2 months (radiotherapy) and 35.4 months (surgery), estimated three-year overall survival was 95 percent with radiotherapy against 79 percent with surgery (hazard ratio 0.14, 95 percent confidence interval 0.017 to 1.190, log-rank p=0.037), with six deaths in the surgical arm and one in the radiotherapy arm. Three-year recurrence-free survival was 86 against 80 percent (hazard ratio 0.69, p=0.54): the survival difference came from deaths not caused by the cancer. Grade 3 adverse events occurred in 3 of 31 radiotherapy patients and 12 of 27 surgical patients, with one surgical death.

The authors were explicit that 58 patients and three years cannot change practice, and the result is best read as a reason to run the larger trials (STABLE-MATES, VALOR and SABRTooth) rather than as evidence that radiotherapy is better. For patients who are medically inoperable, stereotactic radiotherapy is already the standard, which CHISEL established against conventional radiotherapy.