Tumor volume discrepancies between FDG-PET and MRI for cervical cancer.

Abstract

To compare the MRI-defined gross tumor volume (MR-GTV) to the metabolic tumor volume (Metabolic GTV) defined by FDG-PET in patients with cervical cancer. Forty seven patients with cervical cancer underwent FDG-PET/CT and MRI simulations. FDG-PET images were acquired with a spatial resolution of 5mm. MR imaging was performed on a 1.5 Tesla scanner. MR-GTV was contoured on the T2 axial images while Metabolic GTV was auto-contoured using a 40% SUV threshold. Tumor volumes were compared. Eighteen patients (38.3%) had tumors ≥62 cc on MRI. Seventeen patients (36.2%) had tumors ≥14 and <62 cc, while 12 patients (25.5%) had tumors <14 cc. 83% of patients in the ≥62 cc, 23.5% in the 14-62 cc, and 16.7% in the <14 cc cohort had Grade 3 images as defined by Dimopoulos et al. In the ≥62 cc cohort, MR-GTV and Metabolic GTV had a lesion coverage factor of 0.68 (mean MR-GTV(vol) 124.1, mean Metabolic GTV(vol) 119.9 cc). This fell to 0.51 (32.4, 33.1 cc) for the 14-62 cc, and 0.28 (8.4, 8.7 cc) for the <14 cc cohort. These differences were statistically significant on ANOVA testing (p<0.001). MRI provided better visualization of larger tumors than smaller tumors in reference to FDG-PET/CT. FDG-PET/CT visualized tumor volumes different from T2-weighted MRI, especially in tumors <14 cc in regard to location.