NTCP Model for Radiation Pneumonitis after Stereotactic Body Radiation Therapy in Non-Small Cell Lung Cancer Patients

Abstract

Radiation pneumonitis (RP) is a dose limiting toxicity in stereotactic body radiation therapy (SBRT) for early stage non-small cell lung cancer (NSCLC). Currently there is no consensus on which lung dosimetric parameter would best predict RP. This study aimed to determine the optimal lung dosimetric parameter for predicting RP in SBRT using a lung normal tissue complication probability (NTCP) model. A total of 339 early-stage NSCLC patients with SBRT were included in the study. Total lung minus gross tumor volume was used to calculate lung dose volume histogram. V5Gy, V10Gy, V20Gy, V30Gy, as well as generalized equivalent uniform dose (EUD) with “a”=1 to 6 were calculated as representative lung dosimetric parameters. RP1 (RP>=1), RP2 (RP>=2), RP3 (RP>=3) were used as the toxicity endpoint, respectively. Maximum log likelihood (MLL) was used to determine the optimal lung dosimetric parameter for the logistic NTCP model with D50 and γ50 as fitting parameters. The RP>=1 was reported in 91 (27%) patients, RP>=2 in 34 (10%) patients and RP>=3 in 19 (6%) patients. V20Gy was the best lung dosimetric parameter for predicting RP>=1, with V50=16.5 % ± 95% CI (8.6%, 24.3%). EUD (a=5) was the best parameter for RP>=2, with D50=46.7 (Gy) ± 95% CI (30.8 Gy, 62.5 Gy). EUD (a=3) was the best for RP>=3, with D50=30.9 (Gy) ± 95% CI (24.4 Gy, 37.4 Gy). Considering 30%, 15% and 10% of RP for RP1, RP2, and RP3 as the acceptable toxicity rate, respectively, the cutting lung dosimetric parameter should be V20Gy < 7.2%, EUD (a=5) < 28.3 Gy, and EUD (a=3) < 19.0 Gy. With the NTCP analysis of RP endpoints, V20Gy, EUD (a=5), and EUD (a=3) appear to be the best predictors for RP1, RP2 and RP3 following SBRT for NSCLC patients. The cutting off dosimetric parameters for 30% RP1, 15% RP2 and 10% RP3 are 7.2%, 28.3 Gy and 19.0 Gy, respectively.