Abstract
Background
Brachial plexopathy is a potentially serious complication from stereotactic body radiation therapy (SBRT) that has not been widely studied. Therefore, we compared datasets from two different institutions and generated a brachial plexus dose–response model, to quantify what dose constraints would be needed to minimize the effect on normal tissue while still enabling potent therapy for the tumor.
Methods
Two published SBRT datasets were pooled and modeled from patients at Indiana University and the Richard L. Roudebush Veterans Administration Medical Center from 1998 to 2007, as well as the Karolinska Institute from 2008 to 2013. All patients in both studies were treated with SBRT for apically located lung tumors localized superior to the aortic arch. Toxicities were graded according to Common Terminology Criteria for Adverse Events, and a probit dose response model was created with maximum likelihood parameter fitting.
Results
This analysis includes a total of 89 brachial plexus maximum point dose (Dmax) values from both institutions. Among the 14 patients who developed brachial plexopathy, the most common complications were grade 2, comprising 7 patients. The median follow-up was 30 months (range 6.1–72.2) in the Karolinska dataset, and the Indiana dataset had a median of 13 months (range 1–71). Both studies had a median range of 3 fractions, but in the Indiana dataset, 9 patients were treated in 4 fractions, and the paper did not differentiate between the two, so our analysis is considered to be in 3–4 fractions, one of the main limitations. The probit model showed that the risk of brachial plexopathy with Dmax of 26 Gy in 3–4 fractions is 10%, and 50% with Dmax of 70 Gy in 3–4 fractions.
Conclusions
This analysis is only a preliminary result because more details are needed as well as additional comprehensive datasets from a much broader cross-section of clinical practices. When more institutions join the QUANTEC and HyTEC methodology of reporting sufficient details to enable data pooling, our field will finally reach an improved understanding of human dose tolerance.
Asymptotic Inference for Optimal Rerandomization Designs