Pretreatment FDG Uptake of Non-Target Lung Tissue is Significantly Associated With Radiation Pneumonitis Following Stereotactic Ablative Radiation Therapy (SABR)

Abstract

Radiation pneumonitis (RP) is the most common toxicity associated with radiation therapy to the lung. Groups have previously shown association between pretreatment, non-target lung FDG avidity, and conventionally fractionated radiation therapy. We hypothesized that baseline lung FDG avidity can similarly predict RP following lung SABR. Here, we reported our institution’s experience in using pretreatment PET-CT to calculate baseline non-target lung FDG avidity, which we found was significantly associated with symptomatic RP following stereotactic ablative radiation therapy (SABR). We retrospectively reviewed a 400-tumor lung SABR database from our institution to identify 29 patients who experienced grade 2 or higher radiation pneumonitis treated for 42 tumors and 57 control patients treated for 61 tumors. All lung tumors were T1–T3 N0. We used simulation PET-CT to calculate baseline FDG avidity in total lung minus PTV and ipsilateral lung minus PTV, and compared these values between patients who experienced symptomatic RP and those who did not. We also classified tumors as central or peripheral per the RTOG 0813 definition. All strata were compared using student’s t-test, and Cox regression for univariate and multivariate analyses adjusted for the competing risk of death, expressing hazard ratios for PET per .1 increment in SUV. Mean follow-up time was 29 months (range, 7–68). Median time-to-pneumonitis was 5.28 months (range, 1.0–25.2). Patients who experienced symptomatic RP had significantly higher baseline non-target lung FDG avidity as measured by mean SUV (P < .0001), SUV85 (P < .0001), SUV90 (P < .0001) and SUV95 (P < .0001) than those with no symptomatic pneumonitis. We obtained similar results with P < .001 for each of these SUV parameters when we analyzed non-target ipsilateral lung only. Cox regression univariate and multivariate analyses also showed significant association of symptomatic RP with baseline non-target lung FDG avidity (HR = 1.39, 95% CI = 1.21–1.59, P < .001), and with central tumor location (HR = 2.3, 95% CI = 1.11–4.87, P = .03). Receiver operating characteristic analysis revealed sensitivity of 82.8% and specificity of 57.9% when using pretreatment, non-target lung mean SUV>.56 as a cutoff for predicting symptomatic RP after SABR. As expected, specificity improved as SUV cutoff was increased. Patients with higher non-target lung FDG avidity are at significantly increased risk for experiencing symptomatic radiation pneumonitis. Pretreatment, non-target lung mean SUV>.56 may serve as a radiographically useful value for predicting clinical pneumonitis after lung SABR.