Personalized Prediction of Uterine Fibroid Shrinkage During
GnRH
Antagonist Treatment via Pretreatment
T2
‐Weighted
MRI Shigenobu Motoshima, Hiroyuki Nomiyama, Kenyu Hashimoto, Hideyuki Miyahara, Kenta Murotani, Kosuke Kawakami, Hiroki Ishibashi, Naofumi Okura, Toshiyuki Yoshizato
ABSTRACT
Aim
To develop and internally validate a pretreatment quantitative magnetic resonance imaging‐based predictive model for uterine fibroid volume reduction following gonadotropin‐releasing hormone antagonist therapy.
Methods
In this single‐center retrospective study, 106 women with uterine fibroids treated with relugolix underwent pre‐ and intratreatment magnetic resonance imaging. The dominant fibroid was assessed for T2‐weighted imaging fibroid‐to‐iliopsoas signal intensity ratio, apparent diffusion coefficient, flow voids, and fibroid location. The primary outcome was a ≥ 30% reduction in dominant fibroid volume between pre and intratreatment magnetic resonance imaging. Eight baseline variables were entered into a Least Absolute Shrinkage and Selection Operator‐penalized logistic regression model with bootstrap internal validation.
Results
Sixty patients (56.6%) achieved a ≥ 30% volume reduction. In the univariate analyses, a higher T2‐weighted imaging signal intensity ratio and location were significantly associated with response. A purely data‐driven Least Absolute Shrinkage and Selection Operator model retained two predictors: dichotomized T2‐weighted imaging signal intensity ratio and location. A T2‐weighted imaging ratio ≥ 2.0 was the strongest positive predictor, whereas subserosal and cervical locations showed negative associations and submucosal location showed a slight positive association relative to intramural fibroids. The model showed moderate discrimination (apparent area under the receiver operating characteristic curve, 0.746; optimism‐corrected area, 0.671) with acceptable calibration.
Conclusions
Quantitative pretreatment T2‐weighted imaging (ratio ≥ 2.0) is strongly associated with fibroid volume reduction under gonadotropin‐releasing hormone antagonist therapy. An internally validated Least Absolute Shrinkage and Selection Operator‐based model incorporating T2‐weighted imaging signal intensity and location may aid individualized treatment selection and preoperative planning.