The Predictive Impact of Tumor Volume Assessment Using Magnetic Resonance Imaging Before and During Chemorradiation in Patients With Locally Advanced Cervical Cancer

Abstract

The aim of this study was to evaluate the predictive impact of tumor volume (TV) measured on baseline and mid-treatment magnetic resonance imaging (MRI) on long-term outcomes in cervical cancer patients treated with chemorradiation (CRT). We analyzed 51 women with cervical cancer enrolled on an IRB-approved single-institution prospective observational trial of multiparametric MRI performed at baseline (MRI1) and after external beam radiation therapy (EBRT) prior to brachytherapy (MRI2). The initial- (GTV-Tinit) and residual-gross tumor volume after EBRT (GTV-Tres) were estimated by the product of the tumor measurements (superior-inferior; left-right; anterior-posterior) x π/6. The receiver operating characteristic (ROC) curves were built to choose the most accurate cutoff value for predicting locorregional control (LRC), progression free survival (PFS) and overall survival (OS). Uni- and Multi-variate Cox modeling were used to test associations of clinical and radiological parameters with clinical endpoints. Patients were treated with 45 Gy EBRT concomitant to platinum-based chemotherapy followed by 2D-brachytherapy (BT - most received 7Gy times 4). Mean age was 47.3 ± 1.8 years. Most had squamous cell carcinoma (86.3%) and were classified as FIGO stage I-II (58.8%). After a median follow-up time of 20.9 months six women died and ten had disease progression (1 locorregional, 3 distant, 6 combined). The median volumes were GTV-Tinit = 84.4cm3 (36.4-133.4) and GTV-Tres = 12.8cm3 (8.7-26.3). The area under ROC curve for GTV-Tinit/GTV-Tres was 0.87/0.76 for locorregional progression, 0.72/0.65 for progression and 0.70/0.77 for death, respectively. On univariate analysis the LRC was associated with GTV-Tinit (p<0.001), GTV-Tres (p=0.002), FIGO stage (p=0.04), ≥4 cycles of CT (p=0.01), adjacent structures invasion on MRI1 (p=0.018) and MRI2 (p= 0.005) and vaginal invasion at MRI2 (p=0.01); PFS was correlated with GTV-Tinit (p=0.005), GTV-Tres (p=0.01), ≥4 cycles of CT (p=0.012), endometrial invasion on MRI2 (p<0.001) and adjacent structures invasion on MRI2 (p=0.02); and OS was associated with GTV-Tinit (p=0.02), GTV-Tres (p=0.02), ≥4 cycles of CT (p=0.006) and adjacent structures invasion on MRI2 (p=0.03). On multivariate analysis LRC was related to GTV-Tinit (p=0.045), GTV-Tres (p=0.005) and FIGO staging (p=0.038); PFS was associated with GTV-Tinit (p=0.04), GTV-Tres (p=0.02) and endometrial invasion on MRI2 (p=0.003); and OS was associated with ≥4 cycles of CT (p=0.01) and invasion of adjacent structures on MRI2 (p=0.04). TV assessed on MRI was an important predictive parameter for LRC and PFS in patients with cervical cancer treated with definitive CRT. Longer follow up may be necessary to demonstrate a possible correlation with OS.