The myometrial‐cervical ratio (MCR): Assessing the diagnostic accuracy of a novel ultrasound measurement in the diagnosis of adenomyosis

Abstract

Background:Adenomyosis is histologically defined by the presence of endometrial glands and stroma in the myometrium. To date, there are no standardised ultrasound findings that reliably predict histological adenomyosis.Aims: This study aimed to determine the diagnostic accuracy of a novel sonographic measurement for adenomyosis, the myometrial-cervical ratio (MCR), when compared with histopathological diagnosis. Materials and Methods:A single-centre retrospective study was performed. The MCR was calculated from the pre-operative ultrasound, and histopathology reviewed for each case. Accuracy data were analysed in the form of 2 × 2 tables. The discriminative value of the MCR was summarised with a receiver operator characteristic (ROC) curve. Sub-analysis examined the impact of fibroids, hormonal suppression, menopausal status, parity and indication for surgery.Results: Between 1 January 2016 and 31 December 2018, 982 patients underwent hysterectomy for benign non-obstetric indication and adequate pre-operative ultrasound was available for 260. The MCR demonstrated limited diagnostic ability for adenomyosis (area under the receiver operating characteristic curve (AUROC) 0.58, 95% CI 0.51-0.65). However, when applied to cases with no uterine fibroid included in the MCR calculation (n = 133) there was a strong association between MCR and diagnosis of adenomyosis (odds ratio: 5.79, 95% CI: 2.15, 15.62, P = 0.001) with AUROC for this model 0.68 (95% CI: 0.59, 0.77). At an MCR cut-point of 1.74, sensitivity is 67.16% and specificity is 66.15%, with 66.67% of samples correctly classified. Conclusions:While diagnostic accuracy was suboptimal, the MCR outperforms traditional ultrasound diagnostic features of adenomyosis. The MCR may offer a simple imaging measurement for adenomyosis.

The use of the myometrial‐cervical ratio in the ultrasound diagnosis of adenomyosis – A validation study

“…The optimal MCR cut‐point in this subgroup was 1.79, which achieved 55.6% sensitivity and 62.8% specificity, with 58.5% of samples correctly classified (Table 2). Using the previously described cut‐point of 1.74 13 59.0% sensitivity, 59.0% specificity, and 59.0% correct classification were achieved. There was no statistical evidence of a difference in the AUROC using either cut‐point ( P = 0.88).…”

“…Mooney et al showed an association between MCR and histological diagnosis of adenomyosis when samples with fibroids in the MCR measurement were excluded 13 . Using an MCR cut‐point of 1.74 achieved a sensitivity of 67.2% and specificity of 66.2%.…”

“…Medical records were reviewed by a single examiner. Data points collected were identical to the study by Mooney et al 13 Demographics, including age, time between ultrasound and surgery, menopausal status, presenting complaint, gravidity and parity, previous twin pregnancy, previous cervical or myometrial surgery and hormonal treatment at the time of ultrasound. Histopathological diagnosis was extracted from the pathologist report and recorded as adenomyosis present or absent.…”

“…The MCR was calculated by author 5, using the same technique as Mooney et al to ensure uniform image review between the two studies. As previously described, 13 a single standard sagittal view was used where the uterine body and cervix filled 75% of the screen. Two perpendicular measurements were performed, one at the mid‐point in the cervical canal and a second at the greatest anterior posterior diameter of the myometrium at the fundus.…”

“…The optimal MCR cut‐point was identified to maximise model sensitivity and specificity (with 95% CIs determined from 10 000 bootstrapped samples) using the ‘cutpt’ package 15 . Performance (as area under ROC) was compared to a previously described out‐of‐sample MCR cut‐point of 1.74 13 using the algorithm proposed by DeLong, DeLong, and Clarke‐Pearson (1988) 16 as implemented in Stata. Logistic regression was used to calculate the performance of models including MCR alone and traditional ultrasound measurements.…”