Room: Exhibit Hall | Forum 9
Purpose: To quantify dose response for lung irradiation and other predictors associated with IP (interstitial pneumonitis) for patients undergoing TBI (total body irradiation).
Methods: A literature search on TBI identified 42 publications (including 3194 patients) that reported IP incidence rates along with lung doses, dose rates, dose fractionation schemes, and chemotherapy agents. Mean lung dose from these different fractionation regimens was converted to equivalent dose in 2Gy fractions (EQD2) (Î±/Î² =3Gy). Data from fractionated and single fraction regimen subsets were analyzed separately and adjusted for incomplete lung tissue repair (half-life=4 hours) for fractionation schemes with > 1 fraction per day (EQD2_repair). Cox logistic regression was performed using multivariate analysis to identify predictors of IP. Dose response functions were generated using two different models and recursive partitioning was used to identify a lung dose threshold.
Results: Cox logistic regression models found that EQD2, EQD2_repair, and chemotherapy agent cyclophosphamide (Cy) are significant predictors of IP. The models failed to deliver a reasonable dose response curve. Discrete data analysis of multi-fraction per day regimens identified a lung dose threshold of 7Gy and 7.6Gy for EQD2 and EQD2_repair, respectively, below which no IP toxicity was observed. Dose rate was not found to be an independent risk factor for IP. The most common chemotherapy agent was found to be Cy with 80% prevalence and mean dose of 120 mg/kg.
Conclusion: Mean lung dose and Cy were identified as predictors of IP but the prediction model did not accurately reflect the IP rates of published data. Restricting lung EQD2 below the thresholds of 7.0Gy and 7.6Gy identified here for single and multiple fraction per day regimens, respectively, results in negligible IP incidence irrespective of dose rate. These lung doses are easily achievable with lung blocks or intensity modulation for standard 12Gy prescription regimens.