Ulipristal acetate vs gonadotropin‐releasing hormone agonists prior to laparoscopic myomectomy (MYOMEX trial): Short‐term results of a double‐blind randomized controlled trial

Abstract Introduction Laparoscopic myomectomy can be difficult when fibroids are large and numerous. This may result in extensive intraoperative bleeding and the need for a conversion to a laparotomy. Medical pretreatment prior to surgery might reduce these risks by decreasing fibroid size and vascularization of the fibroid. We compared pretreatment with ulipristal acetate (UPA) vs gonadotropin‐releasing hormone agonists (GnRHa) prior to laparoscopic myomectomy on several intra‐ and postoperative outcomes. Material and methods We performed a non‐inferiority double‐blind randomized controlled trial in nine hospitals in the Netherlands. Women were randomized between daily oral UPA for 12 weeks and single placebo injection or single intramuscular injection with leuprolide acetate and daily placebo tablets for 12 weeks. The primary outcome was intraoperative blood loss. Secondary outcomes were reduction of fibroid volume, suturing time, total surgery time and surgical ease. Results Thirty women received UPA and 25 women leuprolide acetate. Non‐inferiority of UPA regarding intraoperative blood loss was not demonstrated. When pretreated with UPA, median intraoperative blood loss was statistically significantly higher (525 mL [348‐1025] vs 280 mL[100‐500]; P = 0.011) and suturing time of the first fibroid was statistically significantly longer (40 minutes [28‐48] vs 22 minutes [14‐33]; P = 0.003) compared with GnRHa. Pretreatment with UPA showed smaller reduction in fibroid volume preoperatively compared with GnRHa (−7.2% [−35.5 to 54.1] vs −38.4% [−71.5 to −19.3]; P = 0.001). Laparoscopic myomectomies in women pretreated with UPA were subjectively judged more difficult than in women pretreated with GnRHa. Conclusions Non‐inferiority of UPA in terms of intraoperative blood loss could not be established, possibly due to the preliminary termination of the study. Pretreatment with GnRHa was more favorable than UPA in terms of fibroid volume reduction, intraoperative blood loss, hemoglobin drop directly postoperatively, suturing time of the first fibroid and several subjective surgical ease parameters.


| INTRODUC TI ON
Laparoscopic myomectomy seems to have several advantages over the laparotomic approach. Smaller incisions result in less postoperative pain, shorter hospital stay and faster recovery. [1][2][3] However, laparoscopic myomectomy can be difficult when fibroids are large and numerous. This may result in extensive intraoperative bleeding and the need for a conversion to a laparotomy. Medical pretreatment prior to surgery might reduce these risks by decreasing fibroid size and vascularization of the fibroid.
Only two medications are registered for the pretreatment of fibroids.
Gonadotropin-releasing hormone agonists (GnRHa) are considered the gold standard. Pretreatment with GnRHa improves pre-and postoperative hemoglobin level and reduces uterine and fibroid volume. 4 Ulipristal acetate (UPA), a selective progesterone receptor modulator, has recently been approved for preoperative treatment of uterine fibroids. UPA has pro-apoptotic and anti-proliferative effects on the fibroid and normal myometrial tissue remains unaffected.
No randomized trials are available reporting on surgical outcomes comparing pretreatment with GnRHa or UPA. In this double-blind randomized controlled trial, we evaluate whether pretreatment with UPA was non-inferior to pretreatment with GnRHa (11.25 mg) on intraoperative and postoperative outcomes of laparoscopic myomectomy.

| Study design
We performed a double-blind randomized controlled trial in nine hospitals in the Netherlands comparing UPA and GnRHa prior to laparoscopic myomectomy. Participating hospitals were selected on extensive experience (>150 per year) with level 3 and 4 gynecological laparoscopic surgery as defined by Royal College of Obstetricians and Gynaecologists.

| Study population
Premenopausal women for a laparoscopic resection of a maximum of two FIGO (PALM-COEIN classification) type 3, 4, 5, 6, or two to five uterine fibroids with a diameter of 5-12 cm were eligible for participation in this study. Exclusion criteria were age below 18 years, pregnancy, suspicion of malignancy, use of hormonal agents, chronic use of anticoagulants, coagulopathy, contraindication to laparoscopic procedures or allergy to leuprolide acetate or UPA.

| Randomization and treatment
Eligible women visiting the outpatient clinic of one of the participating centers were informed about the study by their gynecologist.
After written informed consent was given, participating women were randomly allocated in a one-to-one ratio to receive either GnRHa or UPA. Randomization was performed using a computer-generated randomization system and stratified by center. Women in the GnRHa group received a single intramuscular injection of leuprolide acetate (11.25 mg in 1 mL) and daily placebo tablets for 12 consecutive weeks.
Participants in the UPA group received daily oral UPA 5 mg for 12 consecutive weeks and a one-time placebo injection containing 1 mL saline. Study materials and medication packaging were identical for both groups. Treatment was preferably started in the first week of the menstruation period. Surgery was performed within a month after the last tablet. Participants and gynecologists were blinded to treatment allocation during the entire study period. Statistics were performed by an independent statistician blinded to the allocated study groups.

| Outcome measures
The primary outcome was intraoperative blood loss. Secondary outcomes were time of surgery, time of enucleation, time of suturing, surgical ease and reduction in fibroid volume. For a careful explanation of all outcome measures, see Appendix S1.

| Laparoscopic myomectomy
Surgery was performed by experienced surgeons. The procedure was performed under general anesthesia after administration with GnRHa was more favorable than UPA in terms of fibroid volume reduction, intraoperative blood loss, hemoglobin drop directly postoperatively, suturing time of the first fibroid and several subjective surgical ease parameters.

K E Y W O R D S
gonadotropin-releasing hormone agonist, intraoperative blood loss, laparoscopic myomectomy, pretreatment, surgical ease, ulipristal acetate

Key message
Non-inferiority of ulipristal acetate in terms of intraoperative blood loss could not be established. Pretreatment with gonadotropin-releasing hormone agonist seems to be more favorable than ulipristal acetate for several operative outcomes and subjective surgical parameters.
of prophylactic broad-spectrum antibiotics. An expert meeting of participants was held on 2 June 2014 to reach consensus on the surgical technique. Relevant surgical characteristics were divided in: "standard use", "never use" or "optional use", defined as: Standard use-use of barbed sutures, use of (any) uterine manipulator, use of blue dye in uterine cavity in order to diagnose whether the cavity was opened or not. For fibroids >8 cm it was allowed to apply bulldogs on the uterine artery and infundibulopelvic ligament. Never use-vasoconstrictive medication such as glypressin or use of bulldogs for fibroids <8 cm. Optional use-single administration of 1000 mg of tranexamic acid or use of hemostatic or anti-adhesive products on uterine incision were allowed only after operative blood loss has been calculated.

| Statistical analyses
The trial was a non-inferiority trial with the following null hypothesis: UPA is non-inferior to GnRHa in terms of blood loss during surgery with a maximum difference of 150 mL considered acceptable based on previous studies on this subject. 5 The assumed standard deviation was 250 mL for intraoperative blood loss based on a survey in three hospitals in the Netherlands. Based on a two-group t test of equivalence in means, using an one-sided significance level of 2.5% (one-sided), and a Type II error of 20% (80% power) this yields a sample size of 90 women (45 in each study arm).
The analyses of the primary outcome intraoperative blood loss were performed both according to the per protocol principle and intention-to-treat principle. All other analyses were performed according to the intention-to-treat principle. Normality of the data was assessed visually by means of QQ plots. Because the primary outcome itself did not appear to be normally distributed, but became normally distributed after a log-transformation, we used the following procedure for testing non-inferiority. To take into account the non-inferiority margin of 150 mL defined on the original scale, we added 150 mL to the observed blood losses for GnRHa but left observed blood losses for UPA unchanged. This was done to create a setting in which the difference of the log-transformation of the adapted blood loss was 0 exactly if the blood loss in UPA pretreated women is 150 mL higher than pretreatment with GnRHa (the null hypothesis of the non-inferiority test). Non-inferiority was concluded if the confidence interval for the differences in means for the logtransformation of these adapted outcomes lay entirely below 0 mL.
Normally distributed data were summarized as mean ± standard deviation and were compared with the independent t test. For continuous outcomes that were not normally distributed we present median and interquartile range. Depending on the exact distribution, we used an independent sample t test on the log-transformed outcome or the Mann-Whitney U test when comparing the outcomes between the groups. The Mann-Whitney test was also used for comparison of ordinal variables between the groups. Dichotomous and categorical outcomes were summarized by frequencies and percentage. Chi-square test and Fisher's exact test were used to compare the distribution of these outcomes between groups.
To assess differences between baseline and after 3 months within a group, we used the paired t test for normally distributed variables, the Wilcoxon signed rank test for non-normally distributed data and the McNemar test for dichotomous variables.
Linear regression analyses were performed to correct the analyses for comparison of mean blood loss and total surgery time between pretreatment groups for potential confounders. A variable was considered a confounder when the regression coefficient for groups changed by >10% when the confounder was added to the model. Potential confounders with a skewed distribution were logtransformed to decrease of the impact of outliers.
All analyses were performed using SPSS version 22.0 (IBM Corp., Armonk, NY, USA). Non-inferiority of blood loss was tested at a onesided significance level of 2.5%. A two-sided significance level of 5% was used for all other analyses.

| Women
Women were enrolled between May 2015 and July 2017. Due to disappointing inclusion rates in most participating centers and expiration of study medication, the intended number of inclusions was not met.
Six of nine participating hospitals included women (Table S1). Of 68 eligible women, 55 were randomized: 30 allocated to UPA and 25 to leuprolide acetate ( Figure 1). One woman randomized to UPA in retrospect did not meet the inclusion criteria (fibroid >12 cm) and was excluded from analysis. In the UPA group, two women dropped out before the end of pretreatment; one woman withdrew informed consent directly after allocation, so no follow up occurred and one woman underwent abdominal hysterectomy 6 weeks after start of medication due to persistent severe abdominal pains and fibroid growth. In the GnRHa group, all women completed pretreatment. A total of three women did not undergo a laparoscopic myomectomy. Two women (one in each group) refused surgery because they preferred homeopathic therapy. For one woman allocated GnRHa, surgery was cancelled due to major decrease in fibroid volume from 80 cm 3 to 1 cm 3 .
The two treatment groups were similar in demographic characteristics and hemoglobin level before pretreatment (Table 1).
However, treatment groups differed in fibroid characteristics, due to lack of stratification for fibroid size at baseline. The mean diameter of the largest fibroid was significantly higher in women allocated UPA than women allocated GnRHa (8.5 ± 1.9 vs 7.4 ± 1.6 cm; P = 0.035; 95% CI .1-2.0). Other fibroid characteristics at baseline (ie, type, uterine volume and total fibroid volume planned for resection) did not show any significant difference.
Compliance to study medication was high in both groups. Only two women (one in each group) reported that they forgot to take their oral study medication daily (UPA group 50 tablets remaining; GnRHa group 8 tablets remaining). For the per protocol analyses of the primary outcome, the woman who did not comply with UPA was excluded from analyses.

| Intraoperative blood loss
As intraoperative blood loss was not normally distributed, noninferiority could not be assessed in the standard manner using the 95% CI for the differences in mean blood loss. The alternative approach using a log-transformation of the adapted outcomes as described in the statistical analysis yielded a 95% CI for which the upper bound exceeded 0 (95% CI for difference −0.06 to 0.30 for both per protocol and intention to treat analyses) and hence this trial does not support the alternative hypothesis that UPA is non-inferior to GnRHa. After correction for confounder mean diameter of the largest fibroid, the 95% confidence interval for the mean difference of the logs still contained 0 (per protocol analyses: 95% CI for difference −0.20 to 0.13; intention-to-treat analyses: 95% CI for difference −0.20 to 0.14), which did not alter the conclusion.  Table 3 shows changes in fibroid characteristics and hemoglobin levels between baseline and after 3 months of pretreatment. Change

| Other intraoperative outcomes
In 81% of women pretreated with UPA, only one fibroid was removed, compared with 83% of women pretreated with GnRHa

| Surgical ease
Most items of the surgical assessment tool show a significant difference between both treatment groups (

| Side-effects, postoperative complications and serious adverse events
To assess the prevalence of the most common side-effects, head-  volume after 3 months of pretreatment between the two groups. inclusion rates. These can be explained by physician preferences for one of the treatments, an overestimation of most participating centers of the number of laparoscopic myomectomies they perform on a yearly basis, and the fact that many eligible women were already (pre)treated with UPA or GnRHa in another hospital before they were referred to one of the participating centers.
Despite this, several outcomes reached statistical significance.
It is not possible to conclude whether some of these significant differences were caused by chance (type I error). Another limitation is the fact that we did not stratify for fibroid size at baseline, resulting in an unbalanced distribution of the fibroid size in both groups. A regression analysis was performed to correct for this confounder; however, it cannot be excluded that this difference at baseline had a subsequent effect on many of the other endpoints such as blood loss and weight of fibroids. Additionally, we present the stratified results of fibroids ≤8 cm or >8 cm (Table S2). The direction and trend of the differences between UPA and GnRHa remain the same.
The majority of women were included in one center. Sensitivity analyses did not show differences in intraoperative blood loss for this center compared with other centers (Table S1). An additional limitation of the study may be that the questionnaire to assess surgical ease is non-validated due to very limited studies on this subject, so a total score could not be given. Surgeons were blinded to the pretreatment received in both treatment arms and each question should be interpreted as an individual item without calculation of a total score. Also, since the majority of women were included in one center, surgical ease was determined by a limited number of surgeons and should be interpreted as such.

| CON CLUS ION
Our study did not demonstrate non-inferiority of UPA as a pretreatment compared with GnRHa. We had an underpowered study with a relatively small number of women. Confirmation of our findings is needed to make any final conclusions and based on our data we advise that larger studies, potentially of a superior study design, are carried out. Furthermore, fibroids in our study were large and these large fibroids in particular may benefit from volume reduction to facilitate a successful laparoscopic approach.
From that perspective, volume reduction is important, since volume seems to be related to surgical ease and surgery time. Future studies should aim to confirm this.