Abstract
Objective
Development of a multivariate prediction model based on MRI and clinical parameters for histological adenomyosis diagnosis.
Materials and methods
This single centre retrospective cohort study took place in the gynaecological department of a referral hospital. In all, 296 women undergoing hysterectomy with preoperative pelvic MRI between 2007–2022 were included. MRI scans were retrospectively assessed for adenomyosis markers (junctional zone [JZ] parameters, high signal intensity [HSI] foci in a blinded fashion. A multivariate regression model for histopathological adenomyosis diagnosis was developed based on MRI and clinical variables from univariate analysis with p < 0.1 and factors deemed clinically relevant.
Results
131/296 women (44.3%) had histopathological adenomyosis. Patients had comparable age at hysterectomy, BMI and clinical symptoms, p > 0.05. Adenomyosis patients more often had: undergone a curettage (22.1% vs. 8.9%, p = 0.002), a higher mean JZ thickness (9.40 vs. 8.35 mm, p < .001), maximal JZ thickness (16.00 vs. 13.40 mm, p < .001), mean JZ/myometrium ratio (0.56 vs. 0.49, p = .040), and JZ differential (8.60 vs. 8.15 mm, p = .003). Presence of HSI foci was the strongest predictor for adenomyosis (39.7% vs. 8.9%, p < .001). Based on the parameters age and BMI, history of curettage, dysmenorrhoea, abnormal uterine bleeding (AUB), mean JZ, JZ differential ≥ 5 mm, JZ/myometrium ratio > 40, and presence of HSI foci, a predictive model was created with a good area under the curve (AUC) of .776.
Conclusions
This is the first study to create a diagnostic tool based on MRI and clinical parameters for adenomyosis diagnosis. After sufficient external validation, this model could function as a useful clinical decision-making tool in women with suspected adenomyosis.
Development of a multivariate prediction model based on MRI and clinical parameters for histological adenomyosis diagnosis.
Materials and methods
This single centre retrospective cohort study took place in the gynaecological department of a referral hospital. In all, 296 women undergoing hysterectomy with preoperative pelvic MRI between 2007–2022 were included. MRI scans were retrospectively assessed for adenomyosis markers (junctional zone [JZ] parameters, high signal intensity [HSI] foci in a blinded fashion. A multivariate regression model for histopathological adenomyosis diagnosis was developed based on MRI and clinical variables from univariate analysis with p < 0.1 and factors deemed clinically relevant.
Results
131/296 women (44.3%) had histopathological adenomyosis. Patients had comparable age at hysterectomy, BMI and clinical symptoms, p > 0.05. Adenomyosis patients more often had: undergone a curettage (22.1% vs. 8.9%, p = 0.002), a higher mean JZ thickness (9.40 vs. 8.35 mm, p < .001), maximal JZ thickness (16.00 vs. 13.40 mm, p < .001), mean JZ/myometrium ratio (0.56 vs. 0.49, p = .040), and JZ differential (8.60 vs. 8.15 mm, p = .003). Presence of HSI foci was the strongest predictor for adenomyosis (39.7% vs. 8.9%, p < .001). Based on the parameters age and BMI, history of curettage, dysmenorrhoea, abnormal uterine bleeding (AUB), mean JZ, JZ differential ≥ 5 mm, JZ/myometrium ratio > 40, and presence of HSI foci, a predictive model was created with a good area under the curve (AUC) of .776.
Conclusions
This is the first study to create a diagnostic tool based on MRI and clinical parameters for adenomyosis diagnosis. After sufficient external validation, this model could function as a useful clinical decision-making tool in women with suspected adenomyosis.
| Original language | English |
|---|---|
| Article number | 100028 |
| Number of pages | 8 |
| Journal | Journal of Endometriosis and Uterine Disorders |
| Volume | 2 |
| DOIs | |
| Publication status | Published - Jun 2023 |