Comparison of Computed Tomography- and Magnetic Resonance Imaging-based Clinical Target Volume Contours at Brachytherapy for Cervical Cancer.

Abstract

We prospectively compared computed tomography (CT)- and magnetic resonance imaging (MRI)-based high-risk clinical target volume (HR-CTV) contours at the time of brachytherapy for cervical cancer in an effort to identify patients who might benefit most from MRI-based planning. Thirty-seven patients who had undergone a pretreatment diagnostic MRI scan were included in the analysis. We delineated the HR-CTV on the brachytherapy CT and brachytherapy MRI scans independently for each patient. We then calculated the absolute volumes for each HR-CTV and the Dice coefficient of similarity (DC, a measure of spatial agreement) for the HR-CTV contours. We identified the clinical and tumor factors associated with (1) a discrepancy in volume between the CT HR-CTV and MRI HR-CTV contours; and (2) DC. The mean values were compared using 1-way analysis of variance or paired or unpaired t tests, as appropriate. Simple and multivariable linear regression analyses were used to model the effects of covariates on the outcomes. Patients with International Federation of Gynecology and Obstetrics stage IB to IVA cervical cancer were treated with intracavitary brachytherapy using tandem and ovoid (n=33) or tandem and cylinder (n=4) applicators. The mean CT HR-CTV volume (44.1 cm³) was larger than the mean MRI HR-CTV volume (35.1 cm³; P<.0001, paired t test). On multivariable analysis, a higher body mass index (BMI) and tumor size ≥5 cm with parametrial invasion on the MRI scan at diagnosis were associated with an increased discrepancy in volume between the HR-CTV contours (P<.02 for both). In addition, the spatial agreement (as measured by DC) between the HR-CTV contours decreased with an increasing BMI (P=.013). We recommend MRI-based brachytherapy planning for patients with tumors >5 cm and parametrial invasion on MRI at diagnosis and for those with a high BMI.