Planning Target Volume and FDG-PET Maximum Standardized Uptake Value, But Not Dosimetric Parameters of Target Coverage, Predict for Local Recurrence After Stereotactic Body Radiation Therapy for Non-Small Cell Lung Cancer

Abstract

Prescription biological effective dose and pretreatment FDG-PET maximum standardized uptake value (SUVmax) have been associated with clinical outcomes after stereotactic body radiation therapy (SBRT) for early stage non-small cell lung cancer (NSCLC). Dosimetric parameters of target coverage were not considered in our previous analyses. This study was performed to determine the features that best model incidence of local recurrence (LR) after SBRT for NSCLC when clinical/radiographic characteristics and dosimetric parameters of target coverage are taken into account. Records of all patients with NSCLC treated with SBRT at our institution from 2006 to 2013 were reviewed. Patients with prior in-field radiation or SBRT dose less than 9 Gy x 5 fractions were excluded to yield 256 patients with 260 treated lesions. SBRT consisted of 45 to 60 Gy given in 3 to 5 fractions. Dosimetric parameters of target coverage were calculated based on treatment planning using the analytical anisotropic algorithm. The correlation between LR and dosimetric parameters of target coverage as well as established clinical and radiographic prognostic factors was analyzed using univariate Cox regression analysis. A multivariate proportional hazards model for LR was constructed. Median follow-up was 26.2 months (interquartile range [IQR], 14.8-46.1). Median age at SBRT was 77 (range, 41-95), 47% were male, 75% had KPS ≥80, and 89% were current or former smokers. Histologic subtypes included 73% adenocarcinoma, 25% squamous cell carcinoma, and 2% NSCLC not otherwise specified. Median pretreatment SUVmax was 4.9 (IQR, 2.4-9.1), median prescription BED (using an alpha/beta ratio of 10) was 48 Gy (IQR, 48-54), and median PTV was 49.9 cc (IQR, 36.2-68.6). LR occurred in 36/260 lesions (13.8%); 11/36 (31%) were biopsy proven, while 25/36 (69%) were identified on PET or CT imaging. Median time to LR was 29.7 months (95% CI, 25.4-34.0). Multivariate analysis identified larger planning target volume (PTV) in cc (HR=1.006, 95% CI 1.003-1.008; P=0.016) and higher SUVmax (HR=1.095, 95% CI 1.061-1.130; P=0.004) as independent predictors for LR. Biological effective dosimetric target coverage parameters for PTV and gross tumor volume (GTV), including prescription dose and dose received by 0-100% of the target volume (D_v) as well as target volume receiving 0-100% of the prescribed dose (V_d), did not show a significant correlation with LR (P>>0.05). Multivariate modeling of LR based on clinical and dosimetric parameters demonstrates larger PTV and higher SUVmax to independently predict for LR after SBRT for NSCLC. Dosimetric parameters of PTV and GTV coverage did not significantly correlate with LR.