The present retrospective cohort study enrolled women who presented to the Gynecology Fibroid Clinic at Mayo Clinic, Rochester, MN, USA, for treatment of symptomatic uterine fibroids between April 1, 2013, and April 1, 2014. Participants underwent magnetic resonance imaging (MRI) for clinical care and were not enrolled in any other clinical trial. The Mayo Clinic institutional review board approved the study and patients who had authorized the use of their medical record for research were included in the study; additional informed consent was not required.
A gynecology research fellow who was not involved in staging the fibroids randomly selected one to three fibroids from each patient; they attempted to select the dominant fibroid and a second additional fibroid for women with multiple fibroids.
Static images of each fibroid were captured with sagittal, axial, and coronal orientation in the plane of their maximal diameter; fibroids were labeled, “A” representing the first fibroid, with subsequent fibroids designated “B” and “C” as applicable. These images were reviewed by radiologists (G.K.H., K.R.B.) for accuracy. Using the static images and the series/image number, the fibroids could be mapped to the original MRI examination using the QREADS online system (Mayo Clinic Ventures, Rochester, MN, USA).
The staging of each leiomyoma was determined using the FIGO anatomic sub-classification system by four independent readers; the manuscript by Munro et al. [6 ] and a figure demonstrating location of the fibroids were provided for the readers. The four physicians—two gynecologists (S.K.L-T. and M.R.H.) and two radiologists (G.K.H. and K.R.B.)—were experts in the evaluation and treatment of uterine fibroids; the study center averages 15–20 external referrals of women with symptomatic fibroids monthly. Both gynecologists perform myomectomies routinely, and both radiologists perform focused ultrasound ablation of fibroids.
The number of unique classifications assigned to each fibroid was recorded. By way of example, if three readers determined that a fibroid was type 5 and the fourth reader labeled it as type 6, this would be recorded as two unique answers; if all four readers recorded different fibroid types, there would be four unique answers. A sensitivity analysis was performed, excluding any discrepancies if the ranges of stages overlapped with other answers. For instance, for the sensitivity analysis, a range of type 2–5 was considered no different than a type 4 fibroid.
The significance of staging differences was then categorized based on the clinical relevance of the discrepancy. Staging differences were recorded based on whether they would have clinical implications (Yes/No classification). An example of a staging difference with clinical implications would be a fibroid staged as type 2 by one expert and as type 3 by others because a hysteroscopic myomectomy could be performed on a type-2 fibroid, but not a type-3 fibroid (Table 1 ). For patients who went on to choose surgical management (myomectomy or hysterectomy), attempts were made to match the description of fibroid location in the operative report to that assigned by the readers.
Finally, each fibroid was measured in three perpendicular planes with QREADS and volumes were calculated using the formula for the volume of a prolate ellipsoid. The size of the fibroid was analyzed in terms of impact on the staging results using the Shapiro–Wilk test to compare the significance of two groups and the Kruskal–Wallis test was used for analyses of the number of unique stages. Fibroid size was expressed as the mean±SD. A Cohen kappa statistic was calculated for inter-reader and intra-specialty agreement. Statistical analyses were performed using STATA version 12.1 (StataCorp, College Station, TX, USA) and P≤0.05 was considered statistically significant.
A gynecology research fellow who was not involved in staging the fibroids randomly selected one to three fibroids from each patient; they attempted to select the dominant fibroid and a second additional fibroid for women with multiple fibroids.
Static images of each fibroid were captured with sagittal, axial, and coronal orientation in the plane of their maximal diameter; fibroids were labeled, “A” representing the first fibroid, with subsequent fibroids designated “B” and “C” as applicable. These images were reviewed by radiologists (G.K.H., K.R.B.) for accuracy. Using the static images and the series/image number, the fibroids could be mapped to the original MRI examination using the QREADS online system (Mayo Clinic Ventures, Rochester, MN, USA).
The staging of each leiomyoma was determined using the FIGO anatomic sub-classification system by four independent readers; the manuscript by Munro et al. [6 ] and a figure demonstrating location of the fibroids were provided for the readers. The four physicians—two gynecologists (S.K.L-T. and M.R.H.) and two radiologists (G.K.H. and K.R.B.)—were experts in the evaluation and treatment of uterine fibroids; the study center averages 15–20 external referrals of women with symptomatic fibroids monthly. Both gynecologists perform myomectomies routinely, and both radiologists perform focused ultrasound ablation of fibroids.
The number of unique classifications assigned to each fibroid was recorded. By way of example, if three readers determined that a fibroid was type 5 and the fourth reader labeled it as type 6, this would be recorded as two unique answers; if all four readers recorded different fibroid types, there would be four unique answers. A sensitivity analysis was performed, excluding any discrepancies if the ranges of stages overlapped with other answers. For instance, for the sensitivity analysis, a range of type 2–5 was considered no different than a type 4 fibroid.
The significance of staging differences was then categorized based on the clinical relevance of the discrepancy. Staging differences were recorded based on whether they would have clinical implications (Yes/No classification). An example of a staging difference with clinical implications would be a fibroid staged as type 2 by one expert and as type 3 by others because a hysteroscopic myomectomy could be performed on a type-2 fibroid, but not a type-3 fibroid (
Finally, each fibroid was measured in three perpendicular planes with QREADS and volumes were calculated using the formula for the volume of a prolate ellipsoid. The size of the fibroid was analyzed in terms of impact on the staging results using the Shapiro–Wilk test to compare the significance of two groups and the Kruskal–Wallis test was used for analyses of the number of unique stages. Fibroid size was expressed as the mean±SD. A Cohen kappa statistic was calculated for inter-reader and intra-specialty agreement. Statistical analyses were performed using STATA version 12.1 (StataCorp, College Station, TX, USA) and P≤0.05 was considered statistically significant.