SU-E-J-226: Efficient Use of Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) for Cervical-Cancer Brachytherapy

Authors


Abstract

Purpose:

To investigate image modality selection in an environment with limited access to interventional MRI for image-guided high-dose-rate cervical-cancer brachytherapy.

Methods:

Records of all cervical-cancer patients treated with brachytherapy between 1/2013 and 8/2014 were analyzed. Insertions were performed under CT guidance (CT group) or with >1 fraction under 3T MR guidance (MRI group; subMRI includes only patients who also had a CT-guided insertion). Differences between groups in clinical target volume (CTV), disease stage (I/II or III/IV), number of patients with or without interstitial needles, and CTV D90 were investigated. Statistical significance was evaluated with the Student T test and Fisher test (p <0.05).

Results:

46 cervical-cancer patients were included (16 MRI [3 subMRI], 30 CT). CTV: overall, 55±53 cm3; MRI, 81±61 cm3; CT, 42±44 cm3 (p = 0.017). Stage: overall, 24 I/II and 22 III/IV; MRI, 3 I/II and 13 III/IV; CT, 21 I/II and 9 III/IV (p = 0.002). Use of needles: overall, 26 without and 20 with; MRI, 5 without and 11 with; CT, 21 without and 9 with (p = 0.015). CTV D90: overall, 82±5 Gy; MRI, 81±6 Gy; CT, 82±5 Gy (p = 0.78). SubMRI: CTV and D90 (as % of nominal fraction dose) were 23±6 cm3 and 124±3% for MRI-guided insertions and 21±5 cm3 (p = 0.83) and 106±12% (p = 0.15) for CT-guided insertions.

Conclusion:

Statistically significant differences in patient population indicate preferential use of MRI for patients with high-stage disease and large residual CTVs requiring the use of interstitial needles. CTV D90 was similar between groups, despite the difference in patient selection. For patients who underwent both CT and MRI insertions, a larger MR CTV D90 and similar CTVs between insertions were observed. While MRI is generally preferable to CT, MRI selection can be optimized in environments without a dedicated MRI brachytherapy suite.

This work was partially funded by the NIH R21 CA167800 (PI: Viswanathan; aviswanathan@partners.org)

Ancillary