Periprocedural outcomes comparing fibroid embolization and focused ultrasound: a randomized controlled trial and comprehensive cohort analysis.

Uterine fibroids (myomas or leiomyomas) are common and debilitating in reproductive-aged women, yet little high-quality evidence exists to guide treatment decisions ( 1 ). Most randomized controlled trials (RCTs) of fibroid therapies have been performed outside the United States ( 2 – 10 ), and those performed in the United States for the related problem of heavy menstrual bleeding have faced recruitment challenges ( 11 – 14 ). The Fibroid Interventions: Reducing Symptoms Today and Tomorrow (FIRSTT) study is a National Institutes of Health–funded RCT to evaluate the efficacy of 2 minimally invasive fibroid treatments: uterine artery embolization (UAE) and magnetic resonance imaging–guided focused ultrasound surgery (MRgFUS) ( NCT00995878 , clinicaltrials.gov ) ( 15 ). The aim of the FIRSTT study is to examine the safety, efficacy, and economics of these therapies and the ovarian reserve after treatment. In addition to the RCT participants, women who met identical enrollment criteria but declined randomization were recruited into a parallel cohort (PC1). Analysis of FIRSTT trial baseline data showed that using a comprehensive cohort design (CCD), combining the RCT and PC1 participants, yields valid results and provides additional power ( 16 , 17 ). The current report summarizes the treatment parameters, recovery trajectory, and adverse events (AEs) of patients during the first 6 weeks after treatment, using both RCT and CCD analyses.

Of 83 women who underwent treatment in the CCD cohort (43 MRgFUS, 40 UAE), 75 completed the postprocedure diaries ( Figure ). Baseline parameters were similar between women undergoing UAE and MRgFUS in both the RCT and CCD analysis ( Table 1 ) ( 16 ). The remainder of the data presented refer to the CCD cohort, unless otherwise specified. Women in the study were predominantly white and overweight. Calculated median (IQR) uterine volumes were similar between treatment arms (UAE, 540 [382–837] vs MRgFUS, 586 [395–707] cm 3 ). However, women undergoing UAE had a larger mean (SD) fibroid load than women in the MRgFUS group (362.5 [292.3] vs 249.2 [159.9] cm 3 ; P =.03), and the number of fibroids larger than 3 cm was higher in women undergoing MRgFUS. Participants in both arms had similar fibroid symptoms and substantially impaired fibroid-specific quality of life based on the UFS-QOL. Baseline pain scores tended to be higher in women undergoing MRgFUS, and the difference in visual analog scale for pain (VAS) scores reached significance in the RCT (MRgFUS, 48.0 [27.0–71.0] vs UAE, 27.0 [5.0–52.0]; P =.03) ( Table 1 ). In the UAE group, 36 of the 40 procedures (90%) were done with a unilateral puncture; all were performed in 1 day, with a mean (SD) fluoroscopy time of 42.0 (19.2) minutes (Supplemental Table 1). Total blood loss was minimal, and 100% of the treatments were deemed complete. The majority of the procedures used fentanyl (85%) and midazolam (93%). Among the 43 women undergoing MRgFUS, 23 (53%) underwent a second day of treatment (Supplemental Table 2). The number of fibroids sonicated ranged from 1 to 9, with 19 patients (44%) having 1 fibroid treated. Mean (SD) NPV ratio was 46.1% (24.8%) (Supplemental Table 2). The treating physician judged that 39 women (91%) had complete MRgFUS treatment. Similar to UAE, most procedures were performed using intravenous opioids (93%) and midazolam (98%). Women undergoing MRgFUS had significantly lower self-rated postprocedural pain than women undergoing UAE (1.0 [1.0–4.0] vs 4.0 [2.0–6.0]; P <.001) ( Table 1 ). All UAE patients remained overnight as inpatients, and all MRgFUS patients were discharged on the day of the procedure. Women undergoing UAE were more likely than the MRgFUS group to use outpatient opioid medication (75% vs 21%; P <.001) ( Table 2 ). However, among opioid users, the last day of opioid use occurred later in the MRgFUS group than the UAE group (day 27 [4–41] vs day 3 [2–5]; P =.03). In particular, in the RCT group, 2 women who underwent MRgFUS took opioid pain medication for the entire 6-week follow-up period. Nonsteroidal anti-inflammatory drug use was also higher in UAE patients than MRgFUS patients (97% vs 67%; P <.001), although the total days of use did not significantly differ between treatment arms (7 [2–19] vs 9 [6–13] days; P =.58). Antiemetic and stool softener use was also significantly more frequent in the UAE patients (both P <.001) ( Table 2 ). Women undergoing the UAE procedure took longer to return to work than did women in the MRgFUS group (8 [6–14] vs 4 [2–7] days; P <.001). The day of the week the treatment was scheduled did not affect this parameter (data not shown). The first day on which participants felt they were “totally back to normal” also occurred later in the UAE group (15 [10–29] vs 10 [10–15] days; P =.02). No significant differences were observed in hot flashes and passage of fibroid tissue from the vagina. A total of 36 patients (20 MRgFUS, 16 UAE) experienced 49 AEs; rates overall and rates of severe AEs (SIR class C–E) did not differ between treatment arms ( P =.55 and P =.71, respectively) ( Table 1 ). The AEs are described in detail in Table 3 . Most AEs (42; 86%) required only observation or nominal treatment, and none led to permanent adverse sequelae (class E) or death (class F). Two patients in the MRgFUS arm underwent second procedures within the first 6 weeks, 1 electing UAE and 1 choosing hysterectomy; both were deemed to have had incomplete MRgFUS treatment. Three patients in the UAE arm required readmission to the hospital for severe pain, which was associated with postembolization syndrome in 2. Only 8 AEs qualified for reporting using the American College of Obstetricians and Gynecologists classification. UAE treatment was significantly associated with a higher likelihood of opioid use and longer time to return to work and normal activities (all P <.001), even after adjusting for baseline pain levels and fibroid load using multivariable analysis ( Table 4 ). After adjusting for treatment group and fibroid load, women with higher baseline VAS scores were more likely to use opioid medication ( P =.001), experience an AE ( P =.01), and take longer to return to work and normal activities (both P ≤.001) ( Table 4 ). Results were similar when restricting analysis to the RCT group, except that the association between VAS score and AEs was attenuated.

In this study, women undergoing either UAE or MRgFUS were typically able to return to work within 1 to 2 weeks. However, women undergoing MRgFUS had significantly longer treatment times, with about half undergoing 2 sequential days of treatment, and 9% had incomplete treatment. Substantial differences in recovery trajectory were also observed between treatment groups, with women undergoing MRgFUS reporting lower levels of immediate postprocedure pain, using fewer outpatient pain medications, and fully returning to work sooner than patients undergoing UAE. These differences in recovery parameters persisted even with adjustments for 2 key confounders—baseline pain and fibroid load—both of which were different between treatments arms despite similar uterine volumes, the parameter on which randomization strata was predicated. Although these differences may be due to chance, it is notable that more women in our RCT declined to proceed with treatment after random assignment to UAE. This is particularly relevant when women declining randomization but consenting to be in an observational cohort chose the 2 procedures in more equal numbers. We suspect that this occurred because women were specifically seeking MRgFUS, which was less likely to be covered by insurance and was supported within the confines of the RCT. This possible subversion of the allocation process on the part of participants also supports the use of a CCD analysis. These results on postprocedural recovery are consistent with previous case series and with differences in the mechanism of tissue destruction between UAE and MRgFUS ( 25 – 27 ). UAE is a uterine-directed therapy, in which specific fibroids are not targeted for treatment; in contrast, MRgFUS, like myomectomy, is a fibroid-specific therapy ( 27 ). Thus, UAE may result in increased volume of devascularized tissue. In addition, UAE works via ischemic necrosis, whereas MRgFUS induces coagulative necrosis, which may account for differences in pain. It will be important to ascertain whether these differences correlate with long-term outcomes of these therapies. It is worth noting that both UAE and MRgFUS have substantially shorter recovery times than hysterectomy ( 28 ). Although minimally invasive hysterectomies and enhanced recovery protocols have decreased inpatient stays considerably, most women undergoing hysterectomy still stay longer than 1 inpatient day ( 29 , 30 ). Differences are magnified when comparing time to return to work. In our study, MRgFUS and UAE patients returned to work a median of 4 and 8 days after treatment, respectively. In contrast, recent studies of hysterectomy report average time to return to work of 3.8 weeks for laparoscopic-assisted vaginal hysterectomy and 5.9 weeks for total abdominal hysterectomy ( 31 ). The loss of productivity while convalescing from traditional surgical approaches for uterine fibroids can be a major economic burden for women and their employers; thus, minimally invasive alternatives to hysterectomy could have substantial benefits. Our findings also highlight the importance of assessing baseline pain levels, since greater pain was found to be an independent risk factor for prolonged recovery and AEs. Chronic pain after hysterectomy has been reported in up to 32% of women, and studies have shown that preoperative pain predicts postoperative pain ( 32 ). Although some studies have linked chronic pelvic pain to higher cumulative rates of hysterectomy, our study is the first to link preoperative pain to the incidence of AEs ( 33 ). Further research is needed comparing hysterectomy with minimally invasive options in women with chronic pelvic pain to assess the preferred treatment for this population. Strengths of our study include implementing a standardized treatment protocol and providing patients in both treatment groups with the same postprocedure instructions and prescriptions. Comparing the recovery trajectory for UAE and MRgFUS treatments is an important and understudied research area. An RCT offers the highest level of evidence, and our study offers the added benefit of increased power and generalizability gained by including similar nonrandomized patients in a parallel cohort via a CCD analysis. Our study has several important limitations. We could not recruit as many participants as we wanted; thus, some analyses may be underpowered, such as assessment of safety parameters. Some group differences were also present at baseline; although we controlled for these differences using statistical analysis, a larger sample size may have made that unnecessary. The only blinded study aspect was the analysis of postprocedural imaging. In addition, the system used for MRgFUS in this study (ExAblate 2000) is older than that currently used (ExAblate 2100), and thus the completeness of the treatment and the NPV seen in this study may be lower than those currently achieved. Finally, a key limitation of our study is the small number of black women who were enrolled, despite having a study site devoted specifically to black women and key outreach measures at other sites. Understanding how to optimize recruitment of black women to fibroid RCTs is an important research goal. Other than this limitation, we recruited a representative cohort with significant symptomatic burden based on validated measures and consistent with other studies. The critical question for these therapies, however, is whether there are differences in long-term outcomes. Long-term studies evaluating comparative effectiveness for symptom relief, economic utilization, and ovarian reserve after treatment will be forthcoming from the FIRSTT study. This will be beneficial information to empower women and their physicians to choose the correct treatment for uterine fibroids and improve overall quality of life on an individual basis.

The design and baseline data from the FIRSTT study have been previously reported ( 15 , 16 ). The Institutional Review Boards at Mayo Clinic, Rochester, Minnesota, Duke University, Durham, North Carolina, and the University of California, San Francisco, approved the same study protocol. Briefly, UAE and MRgFUS were performed according to the clinical standard of care, with follow-up for up to 36 months. All participants were premenopausal women with symptomatic uterine fibroids who were not actively seeking pregnancy and had uteri smaller than 20 gestational weeks. Full enrollment criteria have been reported previously ( 15 , 16 ). Enrollment began April 29, 2010, for the RCT, and March 24, 2011, for the PC1 group. All study procedures were performed by August 1, 2014. A study gynecologist screened all participants, and treating physicians at each site were experts in these therapies. Multiple general and disease-specific quality-of-life measures, including the Uterine Fibroid Symptom and Quality of Life (UFS-QOL) instrument, were recorded at baseline ( 15 , 18 ). Randomization was stratified by site and by calculated uterine volume (≥700 vs <700 cm 3 ) and performed using a Web-based, dynamic allocation application ( 19 ). Neither participants nor investigators were blinded to study assignments. Both UAE and MRgFUS were performed using standardized protocols. The treating physician captured key treatment variables on the day of treatment, including whether a complete treatment was achieved. This result was not disclosed to the patient. Identical standardized instructions and postprocedural prescriptions were used for both procedures. For UAE, a standardized protocol that allowed for some variation among sites was used, which included moderate sedation and anti-inflammatory and antiemetic agents ( 20 ). Prophylactic antibiotics were used at all sites; at 1 site, oral antibiotics were continued for another 5 days. UAE was performed, along with arteriography to evaluate for collateral ovarian blood supply to the fibroids. Tris-acryl gelatin microspheres (500–700 µm) were used; if necessary, 700–900–µm spheres were also used until near-stasis was achieved. Post procedure, patients were admitted overnight to a hospital-based observation unit. Treatments were performed with a clinical MRgFUS system (ExAblate 2000; InSightec, Haifa, Israel) with moderate conscious sedation ( 21 ). T2-weighted magnetic resonance images were acquired, and the sonication plan was developed. Sonication pulse duration was generally 12 to 24 seconds, with an interpulse interval of 45 to 90 seconds to allow for tissue cooling. At the conclusion, gadolinium contrast was administered, and T1-weighted images were acquired for visualization of nonperfused volume (NPV). After MRgFUS treatment, women were typically observed for 1 hour and discharged with an escort. Two sites allowed 2 sequential treatment days. For both treatment groups, baseline image analysis was performed using Vitrea 3 Software (Vital Images, Inc; Minnetonka, Minnesota). For the MRgFUS group, the NPV ratio (ie, the ratio [%] of NPV to total fibroid load) was analyzed similarly. The Data Safety Monitoring Board comprised 2 fibroid experts, 1 gynecologist, and 1 radiologist, in addition to the study statistician (A.L.W.). The National Institutes of Health project officer also reviewed safety-related issues. For both procedures, women received identical prescriptions and instructions for medication usage; slight variations among sites and modifications were allowed in the case of allergies or known prior medication use. Women were typically provided with a prescription for oxycodone (5 mg, 40 tablets), ibuprofen (600 mg, 30 tablets, 2 refills), ondansetron (4 mg, 6 tablets, 1 refill), and docusate sodium (100 mg, 10 tablets, 2 refills). Women also received written instructions that acetaminophen could be taken with the study medications. Medications taken during the study period were to be recorded in the study diaries. The postoperative instructions outlined the goal that pain should be as close as possible to 0 on a 10-point pain scale. Contact information for the site investigator and the study coordinator were provided. Women received 3 single-page color-coded diaries with stamped return envelopes at the time of hospital discharge. Each diary covered 2 weeks of recovery and included a pictorial representation of a 10-point pain scale. Information recorded in the diaries included medication use, functional status after surgery, bleeding, pain, and common symptoms. The study coordinator called women on the weekday nearest to the date of expected diary completion and reminded them to return their current diary and start the next one. Data on AEs were obtained via clinic or hospital notes, review of the diaries, and telephone calls with study staff. Characterization included severity (eg, mild, moderate, or severe) based on the Common Terminology Criteria for Adverse Events. Relatedness to the procedure (eg, definitely, probably, possibly, unlikely, or unrelated) was determined by the site investigator, and the patient’s outcome was recorded. We further classified AEs using both the Society of Interventional Radiology (SIR) classification system, which grades events on a scale from A (no therapy, no consequences) to F (death) ( 22 ), and the American College of Obstetricians and Gynecologists system, which characterizes morbidity indicators such as fever, hemorrhage, unintended procedures, life-threatening events, and readmission ( 23 ). Data were analyzed using both a CCD and an RCT design ( 16 ). Demographic and baseline characteristics, day-of-treatment parameters, and postprocedure recovery measures were summarized and reported using standard descriptive statistics: frequency (percentage) for categorical variables and mean (SD) or median (interquartile range [IQR]) for continuous variables. Comparisons between treatment arms (MRgFUS vs UAE) were evaluated using the χ 2 test or Fisher exact test for categorical variables and the 2-sample t test or Wilcoxon rank sum test for continuous variables. Because pain and total fibroid load can substantially affect recovery parameters and there was a differential loss of participants in the UAE arm after randomization ( 16 ), multivariable regression analysis was used to assess the independent effect of treatment group on posttreatment recovery measures after adjustment for pain levels and total fibroid load at baseline. Adjusted odds ratios were obtained using logistic regression for the rare (<10%) binary outcome ( AE SIR class C–E ), adjusted risk ratios were obtained using Poisson regression with a robust error variance for the common binary outcomes ( opioid use and any AE ), and linear regression was used to estimate the average difference in days for the continuous outcomes (recovery time in days) ( 24 ). In these models, logarithmic transformation was applied to the recovery time continuous outcome measures (natural log) and total fibroid load (log base 2) because of the skewed data. Model results were back-transformed to obtain adjusted estimates and confidence intervals in the original scale of the outcome variable. All calculated P values were 2-sided, and P <.05 was considered statistically significant. Analyses were performed using SAS version 9.3 software (SAS Institute Inc, Cary, North Carolina).