Author: Xinxin Deng, Issam M. El Naqa, Jimm Grimm, Lijun Ma, Vitali Moiseenko, Timothy E. Schultheiss, Gopal Subedi, Wolfgang A. Tomé, Ellen D. Yorke, Albert van der Kogel 👨🔬
Affiliation: Montefiore Medical Center, Wellstar Kennestone Hospital Cancer Center, H. Lee Moffitt Cancer Center, Department of Human Oncology, University of Wisconsin–Madison, Department of Radiation Oncology, Wellstar Health System, Radiation Oncology, Keck School of Medicine of USC, Georgia Institute of Technology, University of California San Diego, City Of Hope National Medical Centre, Memorial Sloan Kettering Cancer Center 🌍
Purpose: Reirradiation is increasingly utilized in clinical practice but dose-time recovery factors for human subjects remain uncertain. We constructed a reirradiation recovery model for spinal cord entirely from published human data and compared it to animal findings.
Methods: PubMed was searched for (myelopathy OR spinal cord) AND (reirradiation OR re-irradiation) AND (SBRT). All included studies met the following criteria: prior spinal cord Dmax, reirradiation spinal cord Dmax, time between courses, and known myelopathy outcomes per patient. Only data with SBRT spinal irradiation in at least one of the courses were used. All spinal cord Dmax were converted to biological effective dose (BED) using linear quadratic model and alpha/beta = 2Gy. A proof-of-concept recovery factor was applied to each prior dose; it is a raised exponential function with two parameters, 1) RecoveryHalftime and 2) asymptotic RecoveryMaximum at which the recovery saturates at a constant level. Maximum Likelihood (ML) fitting of the logistic model to clinical data was performed to determine, TD50, g50, the RecoveryHalftime, and RecoveryMaximum. Comparisons were made to rodent and primate models.
Results: We found data for 282 lesions meeting the inclusion criteria with 6 myelopathy cases (Grade 3 or higher), in 13 papers, allowing the construction of a preliminary model as a proposal for a Reirradiation Tissue Effects in the Clinic (ReTEC) project. Similar modeling techniques were applied to the rodent (Wong,1997) and to the primate data (Ang,2001). All three models (humans, monkeys, rats) estimated ≥50% spinal cord recovery in about 1 year or less. Note that the validity of the ML results from humans is subject to sparse data and cohort challenges.
Conclusion: This concept model predicts more recovery than is used in clinical applications. For safety, conservative clinical practice should not be changed until complete human data sets with composite dose distributions are modeled robustly.