Uterine artery embolisation versus myomectomy for premenopausal women with uterine fibroids wishing to avoid hysterectomy: the FEMME RCT

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 08/53/22. The contractual start date was in June 2011. The draft report began editorial review in June 2020 and was accepted for publication in August 2021. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Copyright statement

Copyright © 2022 Daniels et al. This work was produced by Daniels et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This is an Open Access publication distributed under the terms of the Creative Commons Attribution CC BY 4.0 licence, which permits unrestricted use, distribution, reproduction and adaption in any medium and for any purpose provided that it is properly attributed. See: https://creativecommons.org/licenses/by/4.0/. For attribution the title, original author(s), the publication source – NIHR Journals Library, and the DOI of the publication must be cited.

2022 Daniels et al.

Symptomology of uterine fibroids

Uterine fibroids are the most common tumour in women of reproductive age and increase in prevalence as women get older. Around 80% of women will have developed a fibroid by the time that they are 50 years old. Approximately half of women with fibroids experience significant symptoms, which can include heavy menstrual bleeding (HMB), pain on intercourse, abdominal pain and a feeling of pressure,2 all of which can have a significant impact on a woman’s quality of life. 3,4

The symptoms experienced by women with fibroids may vary depending on the position, size and number of fibroids. Intramural fibroids are the most common form of fibroid, but are frequently asymptomatic. Subserosal fibroids, located on the outer surface of the uterus, can become very large and create feelings of bulkiness. Submucosal fibroids project into the uterine cavity and are associated with HMB. As submucosal fibroids may distort the uterus and change the local morphology of the uterine tissue, some clinicians believe that the presence of fibroids may have a negative impact on fertility,5,6 although data are contradictory. 7

Previous research comparing uterine artery embolisation with surgery

Previous research on the cost and cost-effectiveness of treatments for uterine fibroids

The cost-effectiveness of different interventions for the management of uterine fibroids has been evaluated in nine studies. 30,38–45 These studies utilised data from different sources and one study was based solely on a RCT. 30 Economic studies reporting only cost data were not considered here. The interventions used in these studies consisted of MRgHIFU, hysterectomy, myomectomy, UAE and medical management. Cost-effectiveness was assessed from the perspectives of the payer and the society. Only three studies30,39,40 were specific to the UK setting, the remainder being from Canada, the USA and Hong Kong.

The clinical studies underpinning the economic evaluations sometimes gave inconsistent and conflicting results on the clinical effectiveness and safety. Some of the variation can be explained by small sample size and, particularly in the case of the non-randomised studies, considerable differences in baseline characteristics. For example, those undergoing UAE tended to be older and of higher parity and to have poorer quality of life and larger fibroids. This has a direct impact on the results of the associated economic evaluations.

All studies included UAE as a comparator. Six30,38,39,41–43 of the nine studies also included myomectomy as a comparator. Often, myomectomy was considered to be a second-/third-line treatment43,44 or a treatment option when less-invasive methods failed to improve symptoms. 40 The RCT-based study30 categorised UAE as ‘surgical treatment’, along with hysterectomy.

The results showed UAE to be cost-effective when compared with hysterectomy41,44,45 and myomectomy. 41 However, it was not cost-effective when compared with MRgHIFU. 46 In the study conducted by Wu et al. ,40 UAE was cost-effective compared with hysterectomy over the first year after treatment, but this was no longer the case when adopting a longer time horizon. The RCT-based study supports this result,30 as UAE was associated with a greater resource use and higher costs than the surgical treatments (i.e. hysterectomy and myomectomy) during the 1- to 5-year follow-up. However, within the first year, it was associated with lower costs than the surgical treatments. In another two studies,41,43 myomectomy was found to incur higher costs and provide less health benefit than MRgHIFU and UAE (i.e. myomectomy was dominated by these two procedures). 41,43 By contrast, one study42 found that myomectomy was cost-effective when compared with MRgHIFU if productivity costs were not included. However, even when productivity costs were included, the differences in cost and health benefits between myomectomy and MRgHIFU were small. Similarly, in the same study by Cain-Nielsen et al. ,42 UAE was found to be dominated by myomectomy, but, again, the differences in cost and health benefits were small. A study by You et al. ,38 which directly compared UAE with myomectomy from a Hong Kong society perspective over 5 years, reported similar findings (i.e. myomectomy compared with UAE was cost-effective, but with minimal difference).

National and international guidelines on treatment for fibroids

The Royal College of Obstetricians and Gynaecologists (London, UK) and the Royal College of Radiologists (London, UK) jointly issued clinical practice guidelines during recruitment to FEMME (randomised trial of treating Fibroids with either Embolisation or Myomectomy to Measure the Effect on quality of life). 47 These guidelines47 stated that UAE should be considered alongside surgical treatments, endometrial ablation and medical management for women with symptomatic fibroids. These guidelines47 also highlighted that evidence for fertility and pregnancy outcomes after UAE and myomectomy is limited and that there is no robust comparative evidence. Therefore, the guidelines47 deferred making a recommendation for women with fibroids who might want to become pregnant.

In 2012, the American Association of Gynecologic Laparoscopists (Cypress, CA, USA) published a practice guideline48 solely about submucosal fibroids, stating that UAE and MRgHIFU are not appropriate for women who want to get pregnant, while acknowledging that this recommendation was based on expert opinion alone. The American College of Radiology (Reston, VA, USA) presented a variety of clinical vignettes of women with fibroids and derived appropriateness criteria for each potential treatment strategies for each scenario. 49 The American College of Radiology considers UAE appropriate in all situations and myomectomy in most situations, except where there are multiple submucosal fibroids. The Royal Australian and New Zealand College of Obstetricians and Gynaecologists (Melbourne, VIC, Australia) recommendations advise caution and recommends that the routine use of UAE in young women wishing to conceive should be avoided as the effects are uncertain50 again based on expert opinion. We are unaware of the production of any other subsequent guidelines on fibroid management from professional societies in English-speaking countries.

Rationale

Both myomectomy and UAE appear to improve the quality of life of women with symptomatic uterine fibroids, but data derived from high-quality RCTs that would support these very different options are sparse. 51 For symptom control, the choice is currently very uncertain, and indications and clinical preferences for either modality are varied. There is a perception that fertility and pregnancy outcomes are poorer following UAE, but there are few RCT data to support this. 52,53 Given that the options are so different, with very different impact on clinical, patient-reported and economic outcomes, a fair and comprehensive comparison is both overdue and necessary.

Specific objectives

Design

The FEMME trial was a randomised, open, parallel, multicentre trial of UAE and myomectomy in women with symptomatic uterine fibroids wishing to avoid a hysterectomy. The trial had a favourable ethics opinion from National Research Ethics Service Committee Coventry and Warwickshire (reference 11/WM/0149).

Trial oversight

Trial oversight and monitoring were provided by a Trial Steering Committee (TSC) and by an independent Data Monitoring Committee.

The TSC provided independent supervision for the trial, providing advice to the chief investigator, co-investigators and the sponsor on all aspects of the trial throughout the trial. The Data Monitoring Committee adopted the DAMOCLES charter54 to define its terms of reference and operation in relation to oversight of the FEMME trial. Both the TSC and Data Monitoring Committee met on an approximately annual basis during the period of recruitment and follow-up.

Participants

The participants in the FEMME trial were recruited in gynaecology and interventional radiology clinics in a diverse range of NHS hospitals located across the UK. Hospitals could participate if their patient pathway allowed eligible women to undergo either myomectomy or UAE, whether at the same hospital, in another hospital within the same trust/board or under an arrangement with another NHS trust/board.

Screening of potential participants

A full gynaecological and general history was taken by the gynaecologist alongside a general and pelvic examination. The initial diagnosis of fibroids was informed by the woman’s medical history and symptoms, and was confirmed by transabdominal or transvaginal ultrasound scan (TVUS) or hysteroscopy.

If fibroids were visible and considered to be the likely cause of symptoms, and the felt that the women could respond equally well to UAE or myomectomy, then the clinician made the initial approach to the woman to discuss treatment options and participation in the FEMME trial. Women who were referred to a recruiting gynaecologist with a diagnosis then received written information about the FEMME trial before their appointment. If a woman expressed an interest in taking part in the FEMME trial, an approved member of the local research team discussed the trial with them and provided written information. Women who consented to participate in the FEMME trial were provided with a baseline questionnaire and a diary to estimate their menstrual blood loss and were asked to use this to record when they started their next menstrual period (prior to randomisation). Prior to treatment, most women underwent magnetic resonance imaging (MRI), CEMRI ideally, to enable the gynaecologist to accurately visualise the uterus and locate any fibroids.

Eligibility criteria

Women were eligible for randomisation if they met all inclusion criteria and no exclusion criteria.

Recruitment

In the majority of cases, CEMRI was performed to confirm the presence of fibroids, to determine whether or not the fibroids were amenable to treatment by either myomectomy or UAE and to rule out any other pathologies. Consent was obtained in writing from confirmed eligible women who agreed to participate in the trial. The recruitment pathway is summarised in Figure 1.

FIGURE 1.

Eligibility pathway to recruitment and randomisation. BCTU, Birmingham Clinical Trials Unit; GnRH, gonadotropin-releasing hormone; PIS, participant information sheet; TAUS, transabdominal ultrasound.

Participating sites were asked to record the number of women who were initially eligible but later excluded or women who were eligible but who declined consent, alongside the reason for non-inclusion.

Randomisation

Interventions

Following randomisation, the allocated treatment was scheduled for the woman in accordance with local hospital practice. A negative pregnancy test was required immediately prior to the intervention being performed. Pre-myomectomy treatment of fibroids using gonadotropin-releasing hormone analogues and, later, pre-treatment with UPA could be initiated at the discretion of the gynaecologist, but was not recommended. Immediate pre-procedure use of any medical treatment or hormonal contraceptive to reduce HMB was recorded.

Blinding

Owing to the very different natures of the interventions, it was not possible to blind participants, investigators, research nurses and other attending clinicians to the trial treatment allocation.

Outcomes and assessment

Sample size

The original sample size was 650 participants. This would have provided 90% power (at a two-sided alpha level of 0.05) to detect a small to moderate difference of 8 points [i.e. 0.29 of a standard deviation (SD)] in the primary outcome, allowing for approximately 20% loss of primary outcome data. There is no validated minimally important difference for the UFS-QOL scale58 and, therefore, this target difference to detect was considered meaningful and plausible based on the results of a similar published study. 29

Following slower than anticipated recruitment to the trial, and with access to individual participant UFS-QOL data from a previous study,12,29 the sample size target was revised to 250 participants in October 2013, when 114 women had been randomised. A re-analysis of these data, using more appropriate regression methods accounting for baseline imbalances, suggested that a larger difference of 12 points was attainable and that the pooled-group SD of UFS-QOL scores was slightly lower than that originally estimated. The revised sample size had 90% power to detect a moderate-sized difference between groups (i.e. 0.55 of a SD). The Data Monitoring Committee and TSC, which remained blind to contemporaneous primary outcome data, approved the changes on the grounds that this revised effect difference was plausible.

Statistical methods

A comprehensive statistical analysis plan was drawn up prior to the 2-year analysis and was provided to the independent TSC for review. A summary of the analytical approaches is described here.

The analysis of the primary outcome was performed in accordance with the intention-to-treat (ITT) principle. Analyses were performed on (1) complete observed data and (2) all randomised participants at all assessment times through imputation of missing responses. Repeated-measures linear regression models, including data at all time points, were used to estimate least-square mean differences (with 95% two-sided CIs) in the primary outcome at 2 years. 57,62 Parameters allowing for participant, treatment group, baseline score, time, time-by-treatment interaction and the minimisation variables were included. For analysis 1, participants were included in the model provided that they had at least one response at any of the three assessment times. For analysis 2, multiple imputation using a Markov chain Monte Carlo method, assuming a joint multivariate normal distribution, was used. The imputation model was consistent with the analysis model in terms of the variables used. All response times were included. 63

For the primary outcome, a p-value was generated through the aforementioned linear regression model. Observed data from secondary continuous outcomes were analysed in a similar fashion to the primary outcome. Reproductive hormone levels were log-transformed and, therefore, for ease of interpretation, are presented as geometric mean ratios. Log-binomial regression was used to estimate relative rates and 95% CIs for binary outcomes, making similar adjustments to the other analyses. The widths of the CIs were not adjusted for multiplicity and, therefore, the intervals should not be used to infer definitive treatment effects.

Several sensitivity analyses for the primary outcome were also performed, including inclusion of time as a continuous linear predictor, assuming no interaction with treatment; addition of a parameter for treating hospital; and a per-protocol analysis, including only those who received the allocated treatment. Some questionnaires were incomplete and, therefore, an additional sensitivity analysis used available subscale scores to generate an overall score.

For 4-year data, continuous outcomes were analysed by adding responses at this time point to the aforementioned regression models. For time to first pregnancy and time to first further procedure for treatment of fibroids, a Cox proportional hazard model was carried out, adjusting for the minimisation variables. Kaplan–Meier plots were produced in which women were censored if they had withdrawn, were lost to follow-up or had undergone a hysterectomy.

We analysed the treatment effect on the primary outcome in prespecified subgroups that matched the minimisation variables. These analyses involved adding the subgroup-by-treatment group interaction parameters to the linear regression model. All analyses were performed using SAS (SAS Institute Inc., Cary, NC, USA) software, version 9.4.

The Data Monitoring Committee reviewed accruing safety data during the period of recruitment.

Recruitment of participants

The randomisation of participants commenced on 6 February 2012 and the last woman was randomised on 21 May 2015. The monthly rate of recruitment of participants into the FEMME trial is shown in Table 2. Participants were recruited from 29 hospitals across England, Scotland and Wales, as shown in Figure 2.

FIGURE 2.

Recruitment of participants to the FEMME trial over time.

Information was sought for reasons why women declined their invitation to participate in the FEMME trial up to March 2014, at which point 170 women had been randomised. Not every participating hospital provided data on the number of women declining participation and, even among hospitals that did, many did not do so consistently and, therefore, the ratio of women randomised to women declining participation cannot be reliably calculated. Data on 335 women who were assumed to be eligible for randomisation and who declined participation are available. Of these 335 women, 84 (25.1%) requested UAE, 79 (23.6%) requested a myomectomy and 56 (16.7%) requested a hysterectomy. The remaining 116 (34.6%) women gave either another reason or no reason. No accurate information is available on the number of women who were initially approached for participation but were ultimately found to be ineligible for randomisation.

Participant follow-up within 2 years

A total of 127 women were assigned to myomectomy and 127 women were assigned to UAE (Figure 3). The follow-up rate for the primary outcome was 206 out of 254 (81%) women at 2 years (see Figure 3). A total of 227 (89%) women provided scores at least at one assessment time.

FIGURE 3.

Flow of participants through the FEMME trial up to 2 years of follow-up. LTFU, lost to follow-up. Manyonda et al. 64 Copyright © 2020 Massachusetts Medical Society. Reprinted with permission from Massachusetts Medical Society.

Manyonda et al. 64 Copyright © 2020 Massachusetts Medical Society. Reprinted with permission from Massachusetts Medical Society.

Compliance to treatment allocation

Of the 123 women who were randomised to myomectomy and who did not withdraw from the trial prior to the procedure, 105 (85%) had a myomectomy as their initial operation. Similarly, 98 of the 122 (80%) women in the UAE group underwent UAE. Four women in the myomectomy group and nine women in the UAE group chose not have either procedure initially. The initial procedures following randomisation for each allocation group are shown in Figure 4.

FIGURE 4.

Compliance with treatment allocation, by group. LTFU, lost to follow-up. Manyonda et al. 64 Copyright © 2020 Massachusetts Medical Society. Reprinted with permission from Massachusetts Medical Society.

Manyonda et al. 64 Copyright © 2020 Massachusetts Medical Society. Reprinted with permission from Massachusetts Medical Society.

Of those participants who crossed over groups on the basis of a clinical recommendation after randomisation (as opposed to a participant-driven decision), one participant in the myomectomy group was found to have adenomyosis (alongside fibroids) and was offered UAE and two women in the UAE group had a myomectomy (one woman because of heavy bleeding and anaemia and one for an unknown reason).

Four women in the myomectomy group underwent hysterectomy: one because of suspected pelvic infection, one because MRI showed no fibroids but, instead, a large area of adenomyosis, one because she was admitted with acute abdominal pain and one because the surgeon was unable to perform myomectomy. There was one emergency conversion to hysterectomy during a laparoscopic myomectomy following a massive haemorrhage. In the UAE group, one woman underwent a transcervical resection of her fibroids instead.

Baseline characteristics of trial participants

The baseline characteristics of the trial participants are shown in Table 3. Women were, on average, 41 years old and classed as overweight by their body mass index. Although 48% of participants in both groups responded positively to the question ‘at this time, are you seeking to get pregnant?’, about 58% of women were taking treatments, including contraceptives and hormones, to control their HMB or shrink their fibroids.

Procedural details

The details of the initial procedures undertaken are shown in Table 4. Combinations of procedure specifics were possible, according to the clinical situation: one participant from the UAE group had a hysteroscopic myomectomy and then went on to have an open myomectomy; one participant from the myomectomy group had a laparoscopic myomectomy that was converted to an open myomectomy; and six participants from the myomectomy group had a hysteroscopic myomectomy and then went on to have an open myomectomy.

The vast majority of the UAE procedures were successfully completed in the opinion of the radiologist at the time of procedure. However, among those women who had MRI after 6 months, around one-quarter had fibroids that were considered to be < 90% infarcted.

Primary outcome results

The average HRQoL score at 2 years was substantially improved in both groups, approximately doubling in each group, but these improvements were greater at 2 years in those assigned to the myomectomy group (mean difference from observed data 8.0 points, 95% CI 1.8 to 14.1 points, p = 0.01; mean adjusted difference with missing responses imputed 6.5 points, 95% CI 1.1 to 11.9 points). Significant results were seen at earlier time points, as shown in Table 5 and Figure 5.

FIGURE 5.

Evolution over mean responses over time of UFS-QOL HRQoL scores by group. Manyonda et al. 64 Copyright © 2020 Massachusetts Medical Society. Reprinted with permission from Massachusetts Medical Society.

Manyonda et al. 64 Copyright © 2020 Massachusetts Medical Society. Reprinted with permission from Massachusetts Medical Society.

Sensitivity and subgroup analyses of the primary outcome

Secondary outcomes within 2 years

Procedural complications and adverse events

Participant flow within 4 years

Compared with return rates at 2 years, a further 25 women in the myomectomy group and 33 women in the UAE group did not return questionnaires and/or were uncontactable, with 67 (53%) and 81 (64%) women, respectively, returning complete UFS-QOL quality-of-life scores. Interpretation should take this level of missing data into account because it may limit the generalisability of the results.

Participant-reported outcomes within 4 years

Pregnancy and associated outcomes within 4 years

At 4 years, pregnancies, and their outcomes, continued to be reported by the participants or members of the local study team. The number of women getting pregnant, reported as cumulative rates, is shown in Table 22. Appropriate denominators cannot be presented here because of the high levels of drop-out between 2 and 4 years. Data are also reported for the per-protocol (i.e. only those who went on to receive the randomised intervention) and treatment-received populations. Figure 6 presents the ITT data as Kaplan–Meier estimates and takes into account the lack of full follow-up for all women.

TABLE 22 - Pregnancy outcomes within 4 years

UAE group: one participant had two pregnancies that both ended in miscarriage and two participants had two pregnancies that both ended in live birth. Myomectomy group: one participant had two pregnancies that both ended in live birth. These events have been primarily included once in this table, with repeat events in the same woman shown in brackets. All other events occurred in separate women. Percentages of the total population cannot be calculated, as women withdrew from the trial or were lost to follow-up at different intervals up to 4 years.

UAE group: one participant had two pregnancies that both ended in live birth. Myomectomy group: one participant had two pregnancies that both ended in live birth. These events have been primarily included once in this table, with repeat events in the same woman shown in brackets. All other events occurred in separate women. Percentages of the total population cannot be calculated, as women withdrew from the trial or were lost to follow-up at different intervals up to 4 years.

UAE group: one participant had two pregnancies that both ended in live birth. Myomectomy group: one participant had two pregnancies that both ended in live birth and one participant had two pregnancies that both ended in miscarriage. These events have been primarily included once in this table, with repeat events in the same woman shown in brackets. All other events occurred in separate women. Percentages of the total population cannot be calculated, as women withdrew from the trial or were lost to follow-up at different intervals up to 4 years.

Outcome	Number of women (number of events)
	UAE group	Myomectomy group
Pregnancy by ITT
Women reporting pregnancya	12 (15)	6 (7)
Pregnancy (in population desiring pregnancy at time of randomisation)	12 (15)	6 (7)
Live birth	7 (9)	5 (6)
Miscarriage	4 (5)	0
Termination	1	1
Pregnancy by per protocol
Women reporting pregnancyb	7 (8)	6 (7)
Live birth	4 (5)	5 (6)
Miscarriage	2	0
Termination	1	1
Pregnancy by treatment received
Women reporting pregnancyc	7 (8)	8 (10)
Live birth	4 (5)	6 (7)
Miscarriage	2	1 (2)
Termination	1	1

FIGURE 6.

Time to pregnancy.

The cumulative pregnancy rate was 15% in the UAE group and 6% in the myomectomy group (hazard ratio from ITT data 0.48, 95% CI 0.18 to 1.28) (see Figure 6).

Further procedures for fibroids within 4 years

The cumulative number of further procedures for treatment of fibroids was 22 in the UAE group and 13 in the myomectomy group (Table 24), which were hysterectomies for 11 and 8 women, respectively. Again, appropriate denominators are not presented here because of the high level of drop out between 2 and 4 years. Figure 7 presents these data as Kaplan–Meier estimates and takes into account the lack of full follow-up for all women.

TABLE 24 - Further procedures for fibroids within 4 years

Procedure	Number of participants, (n)
	UAE group	Myomectomy group
Any further procedure	22	13
Method
Hysterectomy	11	8
Myomectomy	6	2
Transcervical resection	5	3

The cumulative repeat procedure rate was 24% in the UAE group and 13% in the myomectomy group (hazard ratio 0.53, 95% CI 0.27 to 1.05).

FIGURE 7.

Time to further procedure for treatment of fibroids. Operation dates were estimated to be half-way between follow-up times if exact dates were not available.

Resource use and costs

Direct health-care resource use data collected during the treatment and follow-up periods of the trial include items related to diagnostic procedures, interventions, medication, GP visits, outpatient attendance and inpatient admissions (Table 25). The resource use items were categorised into two parts: (1) treatment-related resource use items and (2) post-treatment resource use items. Treatment-related resource use items referred to items that were recorded from the time of pre-procedure fibroid assessment to the time that participants were discharged following the initial treatment. This period included any imaging performed to assess fibroids, actual interventions received, immediate repeat procedures, excess bed-days and medications that were prescribed on discharge. Post-treatment resource use items were recorded during the period from post discharge from initial treatment to the first follow-up at 6 months and, subsequently, to follow-ups at 1, 2 and 4 years.

All unit costs were collected from routine sources, including the NHS reference costs,68 the Personal Social Services Research Unit69 and the British National Formulary (BNF). 70 All costs were expressed in GBP (£) for the price year 2018/19 (Table 26). Costs associated with all resource use were summed as the total cost for each participant.

A number of assumptions related to resource use and unit costs were made. In the case of treatment-related resource use, all interventions were assigned unit costs according to the Healthcare Resource Group (HRG). All participants were assumed to be admitted as elective inpatients. Repeat interventions were assumed to incur the same resource use and cost as the initial intervention. Length of stay (LOS) was calculated as the difference between admission date and discharge date. In the case of LOS exceeding the ‘trim points’ (i.e. the expected LOS for each HRG), an additional per-diem cost was assigned to estimate costs associated with these excess bed-days. 72 Medications used during the procedure or during the time in a ward were assumed to be included in the HRG episode cost. Any additional medications prescribed on discharge were costed separately using unit costs from the BNF.

In the case of post-treatment resource use, all re-admissions from post discharge to first follow-up were assigned an average cost of non-elective short stay and long stay. Stays of ≤ 2 days were considered to be short stays and stays of > 2 days were considered to be long stays. All participants who did not report hospital re-admissions but had complications, infections or medications during the follow-up period were assumed to attend outpatient clinics. It was assumed that all reinterventions for fibroid removal during the follow-up period were performed on an elective inpatient basis. Resource use data captured in hospital records from post discharge to follow-up and data separately recorded by the participants at their first follow-up (i.e. 6 months) were cross-checked to avoid double-counting.

Health outcomes

Patient-reported HRQoL was measured using the EQ-5D-3L. The EQ-5D-3L was completed by the participants at baseline, 6 months, 1 year, 2 years and 4 years. The EQ-5D-3L describes health status of a patient in five dimensions (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) and there are three levels of response for each dimension. 73 Responses to the five health domains were used to calculate health utilities using a standard UK value set. The area under the curve method was used to estimate quality-adjusted life-years (QALYs) for each participant by adjusting the life-years gained by the quality of life experienced over the study period. 74 The method considers the utilities obtained at different time points because it assumes a linear relationship between these utilities. Subsequently, QALYs from baseline to 2 years and from baseline to 4 years were estimated.

Handling missing data

For the purpose of analysis, the following assumptions were made:

• With regard to medication, the median duration of treatment was assumed where data were missing and standard BNF doses were assumed where data on dosages were missing.

• In the case of participants with initial fibroid assessment scans who did not undergo any procedure, but remained in the trial and had cost/utility follow-up data (nine participants in the UAE group and four in the myomectomy group).

  • We did not make assumptions on additional resource use for participants with no resource use data throughout the study period (six in total); participants who had no other resource use data apart from those collected during the follow-up (three in the UAE group); nor participants who had additional fibroid imaging but no fibroid removal reintervention during the follow-up (two in the myomectomy group).

  • We assumed that all participants who had reintervention, but no other imaging related resource use, received fibroid imaging (two in UAE group).

The association between missingness, baseline variables and observed outcomes was also explored. A binomial logistic regression was carried out to investigate which variables and outcomes were associated with probability of missingness. We assumed that the data were missing at random. 75 Multiple imputation was performed using multiple imputation by chained equations. 76 Ten imputation data sets were generated with predictive mean matching (the mean of five nearest values). Imputation was performed for total costs at a subaggregate level of treatment and non-treatment costs, whereas QALYs were imputed at aggregate level of the total QALYs level and not on a health utilities level.

Cost-effectiveness analysis

The base-case analysis followed an ITT strategy and was conducted post multiple imputation. Generalised linear models were used to analyse cost and QALY data because of their non-normal distribution, which would otherwise render the calculation of a simple mean as inappropriate. 77 Generalised linear models accommodate this non-normal distributions of cost and QALY data by allowing specification of a more appropriate distributional family and link function through the modified Park test and other tests.

The cost model was analysed using a gamma family and log-link. It was adjusted for minimisation variables used in the clinical analysis (i.e. the participant’s desire to be pregnant at the time of randomisation, the longest dimension of largest fibroid and the number of fibroids). The QALY model was analysed with Gaussian family and identity link. It was adjusted for minimisation variables, as well as the statistically significant predictors of QALYs (i.e. baseline utilities and body mass index). Baseline utilities are usually highly correlated with the patient’s QALY gain, and any imbalance of baseline utility level between treatment groups may lead to misleading cost-effectiveness estimates. 74 The models were then used to predict the marginal mean costs and mean QALYs, which were then used to calculate their cost and QALY differences.

Cost-effectiveness was expressed as incremental cost per QALY gained [incremental cost-effectiveness ratio (ICER) = ΔC/ΔQ)], where appropriate. Incremental mean cost (C_UAE – C_Myomectomy = ΔC) is the difference in mean costs between the two treatment groups, and incremental QALY (Q_UAE – Q_Myomectomy = ΔQ) is the difference in QALYs gained between the two treatment groups. The ICER provides the additional cost of achieving 1 perfect year of health and can be compared against the NICE willingness-to-pay (WTP) threshold of £20,000. 67 In addition, incremental net monetary benefit (NMB), a measure of the health benefit expressed in monetary terms, was also estimated. The NMB was calculated using the following formula:

where ΔQ is the incremental QALY, λ is the WTP threshold (i.e. £20,000 in the UK) and ΔC is the incremental cost. 77 An intervention is generally considered cost-effective compared with the alternative if the incremental NMB is positive.

To quantify the uncertainty around incremental costs and QALYs, a 1000-iteration bootstrap was performed. The results were presented on cost-effectiveness planes. Cost-effectiveness acceptability curves were used to present uncertainty in the decision regarding cost-effectiveness over a range of WTP thresholds. 78 All analysis was carried out using Stata (StataCorp LP, College Station, TX, USA) version 16.0.

Sensitivity analyses

The second scenario involved varying the unit cost applied to the procedures during initial intervention and any reinterventions for fibroid removal. In the base-case analysis, costs were estimated using a top-down approach, which emphasises national average costs. The English NHS reference costs incorporate costing on the basis of HRG, which is a measure of case mix, demonstrating groups of clinically similar treatments that utilise a common set of health-care resources. 72 Therefore, the HRG tariffs are a reflection of NHS average costs and may not capture the difference in practice across different FEMME trial sites. Therefore, we performed a sensitivity analysis to test the robustness of our results according to variations in unit costs that may have resulted from difference in practices. A 20% cost increment and decrement were applied to cost of initial intervention, as well as to the cost of reinterventions for fibroid removal that happened during the study period.

Health-care resource use and costs

Overall, 127 women were randomised to each treatment group (Table 27). The majority of women underwent pre-procedure imaging prior to UAE and myomectomy. Across both groups, the majority of women received MRI (UAE group, 71%; myomectomy group, 79%). In the UAE group, 98 women received UAE, 14 received myomectomy and one received endometrial ablation (in addition, 14 women did not receive treatment or withdrew from the study). In the myomectomy group, 105 women received myomectomy, six received UAE and eight received hysterectomy (in addition, eight women did not receive treatment or withdrew from the study). UAE was associated with a median LOS of 2 (IQR 2–3) days, compared with 4 (IQR 3–5) days for myomectomy. Across both groups women were prescribed analgesics on discharge (UAE group, 87%; myomectomy group, 97%).

Post-treatment-related resource use is shown in Table 28. During the period from discharge to first follow-up (at 6 months), UAE was associated with a slightly higher proportion (16%) of women with re-admissions than myomectomy (11%). Similar patterns were observed for the mean number of re-admissions from discharge to 2 and 4 years. Resource use associated with outpatient appointments and medications prescribed was similar between the UAE and the myomectomy treatment groups and this pattern was consistent at 2 and 4 years. A larger number of women in UAE group (n = 18) than in the myomectomy group (n = 8) received reintervention within the first 2 years. At the end of 4 years, 22 women from the UAE group had reintervention compared with 13 women in the myomectomy group.

Table 29 provides a breakdown of average costs according to resource use categories. In the case of most resource use categories, the costs incurred are very similar in both treatment arms. The minimal difference in pre-procedure fibroid assessment can be attributed to the type of scans that these women received at the beginning. Most women in the myomectomy group received MRI rather than ultrasound, and MRI is slightly more expensive. The same explanation is applicable for the cost of imaging follow-up for fibroid reassessment. The average cost of treatment is higher in the myomectomy group, given that myomectomy is a more expensive procedure and requires a longer stay in the hospital. By contrast, costs incurred during the period from post treatment to discharge were higher in the UAE group. This reflects the unit cost applied for UAE excess bed-days (see Table 26), which was almost double the excess bed-days unit cost for myomectomy. Mean costs during post discharge to 6-month follow-up period were higher in the UAE group than in the myomectomy group. Women in the UAE group, again, incurred higher mean costs during the follow-up, which included the cost of GP visits, outpatient appointments and inpatient admissions.

Health outcomes

The proportion of responding participants and their corresponding responses to five health domains at baseline, 6 months, 1 year, 2 years and 4 years are presented in Table 30. Across the two treatment groups, at baseline, only a minority of women reported any problems with mobility (UAE, 16%; myomectomy, 17%) and self-care (UAE, 4%; myomectomy, 2%). Twenty-nine per cent of women in the UAE group and 33% of women in the myomectomy group reported some problems with usual activity. However, in both groups, 79% of women reported any problems with pain/discomfort, and 50% of women in the UAE group and 55% in the myomectomy group reported any problems with anxiety/depression.

In the UAE group, the proportions of women reporting any problems with mobility and self-care were consistent over time. There were substantial improvements in the usual activities and pain/discomfort domains, and a modest improvement in the anxiety/depression domain. In the myomectomy group, there were modest improvements in the mobility and self-care domains over time, and substantial improvements in the remaining three domains. In particular, the improvement in the pain/discomfort and anxiety/depression domains was greater than that observed in the UAE group.

Cost-effectiveness analysis

Participants who withdrew, were lost to follow-up, had missing resource use at the main time points or had any missing health utilities data were considered to be missing. EQ-5D-3L data were missing for 32% and 45% of participants at the 2- and 4-year follow-up, respectively. The proportion of participants with missing resource use data was low at both time points (4% and 8%, respectively).

The cost-effectiveness results for the FEMME trial are presented in Table 31. Predicted total mean costs were estimated separately based on resource use from baseline to 2 years and from baseline to 4 years. Each of these costs constituted cost components (i.e. treatment cost, as well as post-treatment costs up to 2 and 4 years).

The mean treatment costs were lower in the UAE group than in the myomectomy group (£3064 vs. £3862). Conversely, the UAE group was associated with a higher post-treatment cost than the myomectomy group over 2 years’ follow-up (£4918 vs. £3431) and over 4 years’ follow-up (£5288 vs. £4151). The total mean cost incurred over 2 years was £7958 in the UAE group compared with £7314 in the myomectomy group. The total mean cost over 4 years was £8362 in the UAE group compared with £8010 in the myomectomy group.

Over the period of 2 years, the QALY gain was 0.74 (95% CI 0.70 to 0.78) in the UAE group compared with 0.83 (95% CI 0.79 to 0.87) in the myomectomy group. UAE was associated with higher costs (i.e. a £645 difference in cost, but this was not statistically significant) and lower QALYs (i.e. a –0.09 difference in QALYs) than myomectomy. UAE is dominated by myomectomy.

Similar results were observed over the 4-year time horizon. The QALY gain was 0.73 in the UAE group (95% CI 0.69 to 0.76) compared with 0.82 (95% CI 0.79 to 0.87) in the myomectomy group. UAE was associated with higher costs (i.e. a £352 difference in cost, but this was not statistically significant) and lower QALYs (i.e. a –0.09 difference in QALYs). Again, UAE is dominated by myomectomy. Compared with the 2-year follow-up, the cost difference between the UAE and the myomectomy group at 4 years’ follow-up was slightly lower, while the QALY gain remained the same.

The cost-effectiveness planes for 2 and 4 years are presented in Figure 9. The lines represent the WTP thresholds (i.e. £20,000 per additional QALY gained). The bootstrapped replications were displayed in the north- and south-west quadrants. Most replications were concentrated in the north-west quadrant. Regardless of the time horizons, there is little difference in costs between UAE and myomectomy. The deterministic estimate is close to zero. There is little uncertainty that UAE has lower health benefits than myomectomy. There is some uncertainty relating to the magnitude of the difference in costs between the two groups at 4 years.

FIGURE 9.

Cost-effectiveness plane for (a) 2 years and (b) 4 years. Reproduced with permission from Rana et al. 66 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from Rana et al. 66 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Both of the cost-effectiveness acceptability curve figures (Figure 10) showed that myomectomy had a higher probability of being cost-effective than UAE at WTP thresholds of ≥ £20,000. At a £20,000 WTP threshold, the probability of myomectomy being cost-effective is 98% at 2 years and 96% at 4 years.

FIGURE 10.

Cost-effectiveness acceptability curves for (a) 2 years and (b) 4 years. Reproduced with permission from Rana et al. 66 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from Rana et al. 66 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Sensitivity analysis

The results from the scenario in which we performed a complete-case analysis are presented in Table 32. Similar to the base-case analysis, analysis of cost components showed that women in the UAE group had lower treatment costs than women in the myomectomy group (£3073 vs. £3870, respectively). Women in the UAE group also incurred higher post-treatment costs over 2 years than women in the myomectomy group (£4663 vs. £3384, respectively) and 4 years (£5057 vs. £4127, respectively). UAE was associated with higher costs (i.e. a £456 difference in cost, but this was not statistically significant) and a lower QALY gain (difference of –0.06 ) over a time horizon of 2 years. Similar results were observed over a 4-year time horizon. The cost difference was slightly lower than at the 2-year follow-up, but the QALY difference remained the same. A 6% decrease in ICER was observed when we compared the 2-year follow-up ICERs obtained from the base-case analysis with those obtained from the sensitivity analysis. Relatedly, an increase in ICER of 20% was observed for the 4-year follow-up. It is imperative to note that ICERs are not normally calculated when an intervention is dominated by the comparator.

The results of the sensitivity analysis on the impact of varying unit cost of procedures on mean total costs show that costs were higher in UAE group than in the myomectomy group (Table 33). Overall, varying cost of interventions had an overall effect on the magnitude of the ICERs (Figure 11). However, the overall results remained consistent with the base-case analysis. UAE was associated with higher costs and lower QALYs gained when compared with myomectomy across the two time horizons. Therefore, myomectomy dominated UAE.

FIGURE 11.

Sensitivity analysis on cost of interventions. (a) ICERs at 2 years and (b) ICERs at 4 years. ICERs are normally not presented when an intervention is dominated by its comparator. However, we present them for completeness. The mean incremental QALY for the ICER was –0.09, which is consistent with the base case, as we only varied the unit cost of procedures in sensitivity analysis. Reproduced with permission from Rana et al. 66 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Reproduced with permission from Rana et al. 66 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure includes minor additions and formatting changes to the original figure.

Principal findings

Although both myomectomy and UAE improved participant-reported HRQoL scores, women assigned to the myomectomy group reported higher scores than those assigned to the UAE group. Menstrual bleeding scores appeared similar in both groups. Overall complication rates from all initial procedures occurred in a similar proportion of women in both groups. The need for additional treatments was higher and hospital stay was shorter in the UAE group than in the myomectomy group. There were no consistent differences between groups in biomarkers of ovarian reserve. In total, there were 15 pregnancies in the UAE group and seven pregnancies in the myomectomy group, but these numbers were too small draw a conclusion on the effect of the procedures on fertility. The economic evaluation showed that UAE was associated with higher costs and lower QALYs than myomectomy. At a £20,000 WTP threshold, the probability of myomectomy being cost-effective is 98% at 2 years and 96% at 4 years.

Interpretation

Doubling of the UFS-QOL quality-of-life score from baseline to each time point shows that, on average, both treatments are effective, with an additional 8-point benefit accrued from myomectomy, equating to a small to moderate standardised treatment benefit at 2 years. 79 In a review of the use of EQ-5D in 11 varied populations, a mean minimally important difference of 0.07 was derived. 80 This is consistent with the difference in EQ-5D score observed in this study, corroborating the between-group difference observed for UFS-QOL.

There were substantially more surgical reinterventions in the UAE group in the first 2 years of follow-up, possibly reflecting the higher residual impact on quality of life observed in the UAE group and the marginal patient-reported preference for myomectomy. However, the number of hysterectomies performed as the initial procedure was higher in the myomectomy group because of either patient preference or clinical decision.

Our results show that UAE is associated with higher costs and lower QALYs than myomectomy (£645 and £352 over the 2- and 4-year time horizons, respectively). This incremental cost should be interpreted with caution, as the difference was small and not statistically significant. The difference in QALYs over both the 2- and the 4-year time horizons was 0.09, and this is equivalent to 33 days of perfect health. The cost drivers were GP visits, outpatient appointments and inpatient admissions during the follow-ups and the associated reinterventions for fibroid removal. The key driver for this QALY difference was the substantial improvements associated with pain/discomfort and anxiety/depression observed with the myomectomy group.

Findings in the context of the existing effectiveness literature

The only comparable randomised trial using UFS-QOL29 suggested a between-group difference of 12.1 (95% CI 4.0 to 20.2) points in the HRQoL domain at 1 year, following our reanalysis of their individual patient data, which is consistent with our observation of a 10.8-point difference (95% CI 4.2 to 17.5 points) at the same time point. The changes in score from baseline to 1 year are substantial: an improvement of at least 33 points (maximum score for domain is 100 points) in both the FUME29 and the FEMME trial and following either treatment. Similar large improvements in UFS-QOL scores have been reported following various fibroid treatments. 81,82

In a study from Czechia,27,28 the rate of reintervention within 2 years was 33% in the UAE group and 3% in the myomectomy group, but these rates are not directly comparable with the FEMME results, as the protocol of the former study recommended myomectomy if improvements were not seen 6 months after UAE. Only one participant had a subsequent hysterectomy following myomectomy in the other RCT,29 although the reintervention rate was 15% (i.e. nine participants) more than 1 year after an initial UAE.

There were substantially more pregnancies in the study from Czechia,27,28 with 13 out of 26 women who attempted conception becoming pregnant in the UAE group and 30 out of 40 women in the myomectomy group achieving pregnancy within a mean follow-up period of around 2 years. Half of those women who became pregnant in the UAE group had a subsequent myomectomy. Comparing their pregnancy and pregnancy outcome data with data from the FEMME trial is difficult because of the different proportions of women intending to get pregnant and differences in the age profile of participants.

Previous randomised trials that measured FSH, and used varying thresholds for ovarian failure, also found no evidence of harm from UAE over short and longer time frames. 35 To the best of our knowledge, no other randomised studies have assessed AMH.

Findings in the context of the existing economic literature

To identify the main driver of total mean costs, the costs were categorised according to treatment-related cost and post-treatment cost. The treatment cost of UAE was lower than that of myomectomy. This can be attributed to the fact that the observed LOS was longer for myomectomy than for UAE (i.e. a median of 4 days vs. a median of 2 days), and this is consistent with the literature. 81,83 Women in the UAE group incurred greater post-treatment costs than women in the myomectomy group. The main driver of post-treatment costs was utilisation of health-care resources (e.g. GP visits, outpatient appointments and inpatient admissions), as recorded during the follow-ups associated with reinterventions for fibroid removal. Over the 4 years’ follow-up, the rate of reintervention was higher in the UAE group than in the myomectomy group (hazard ratio 0.53, 95% CI 0.27 to 1.05). Our findings are in line with existing studies that reported on UAE continuously accruing costs in the long term40 and higher reintervention cost than myomectomy. 38 In our study, the majority of the post-treatment costs were accrued within 2 years and only a small amount of additional post-treatment costs incurred between the 2-year and the 4-year follow-up.

Strengths and limitations of the randomised trial

To the best of our knowledge, this study is the largest ever randomised clinical trial to report on treatment of symptomatic fibroids by UAE and myomectomy. The robust study design ensures internal validity, enabling the results to be interpreted with confidence. Randomisation via computer-generated allocation sequence was effective in achieving balanced groups with respect to important prognostic factors.

In the absence of a minimally important difference for between-group comparisons on the UFS-QOL scale,58 the size of the study was driven by a small to moderate standardised effect. This was derived from the published results of the FUME study. 29 Later, we reanalysed the individual participant UFS-QOL data of the FUME study using more appropriate regression methods, accounting for baseline imbalances. This reanalysis generated a larger potential difference and a smaller pooled-group SD. This enabled us to reduce the FEMME trial’s sample size from the original projection, which we were then able to achieve. However, this cannot be considered data driven, as the recalculation was performed before any interim analysis and ratified by the TSC and Data Monitoring Committee, which were blind to any contemporary outcome data.

Our trial design offered a number of other strengths with respect to data collection and analysis. The primary outcome measure and many secondary outcome measures focused on the participants’ report of their symptoms and quality of life. In addition, the primary outcome was fibroid specific and co-developed with women with fibroid symptoms. This ensured that the outcomes reflect the aspects of the condition of most importance to women.

The techniques used for both myomectomy and UAE were determined by the fibroid presentation and the preference of the operating clinician. This enabled a diverse range of fibroid diagnoses to be included in the study, but limits the interpretation of the results to a class comparison of myomectomy with UAE (i.e. the trial did not analyse operative subgroups, e.g. laparoscopic myomectomy vs. UAE, and we did not balance the randomisation by intended myomectomy approach).

We consider the trial design and conduct to have been methodologically robust, but there are some limitations of our study that should be considered. Nineteen per cent of participants failed to return the UFS-QOL questionnaire at 2 years, despite our attempts to retrieve this information. Our analytical approach involved imputation of missing responses using a recognised method, but assumed that data were missing at random and that the reason for non-response was not related to these participants’ quality of life. Any deviation from this assumption could give rise to inconclusive results, given that the lower end of our CIs around effect estimates were close to zero.

A substantial number of women were not recruited because of their preference for a particular treatment option and expectations of treatment benefit were not captured pre randomisation. As with all surgical or interventional procedure trials, not every participant ultimately undergoes the allocated procedure, either for clinical reasons or because of patient preference The proportion of women who did not undergo their allocated procedure differed between groups (UAE group, 20%; myomectomy group, 15%). The numbers undergoing different types of alternative treatment also varied; for example, the number of women undergoing no immediate procedure was nine in the UAE group and four in the myomectomy group. All participants were analysed in the groups to which they were allocated and the per-protocol analysis gave a treatment effect very similar to ITT population analysis, suggesting little impact of non-adherence to the allocation.

The two procedures have considerably different recovery periods, which may be reflected in the first outcome measures reported by participants at 6 months. The duration from randomisation to the procedure averaged around 13 weeks in both groups, and the primary outcome was at 2 years post procedure, as this is a realistic time point at which to compare quality of life.

Some blood samples were unable to be analysed for FSH and LH, as they were not obtained within 5 days of the start of the last menstrual period, reducing the data available to corroborate the AMH results. Despite randomisation, there was a small difference in participant age between the two groups, prompting a post hoc-adjusted analysis of the ovarian reserve markers.

Our statistical analysis plan did not include correction for multiplicity of statistical testing. We have not carried out multiplicity correction for the secondary outcomes in terms of size of CI (p-values were not reported).

Strengths and limitations of the economic evaluation

The economic evaluation is based on robust data collected alongside, to the best of our knowledge, the largest RCT that evaluated this comparison and adhered to good practice guidelines set out by NICE. 67 In accordance with the guidance,67 we took into account relevant resource use and estimated costs, and we estimated health benefits based on the EQ-5D. Robust methods were used in the analysis to explore uncertainties.

A limitation of the analyses is the number of missing health utility data. However, incomplete participant data is an expected feature of trials with long follow-up periods and is attributable to participants being withdrawn or lost to follow-up or non-compliance. We performed sensitivity analysis to examine the impact of missing data. Another limitation is that the health outcomes were assessed using the EQ-5D-3L, rather than the EuroQol-5 Dimensions, five-level version, which is considered to be a more sensitive and precise measure of health status. The former has a tendency to underestimate health utilities by overestimating health problems. 84 Moreover, one inherent limitation of this approach is that it does not take into account patient preference, despite that the literature that suggests that there is no statistically significant difference between UAE and surgery (hysterectomy or myomectomy) in terms of health-related quality of life. 81 Furthermore, this suggestion is supported by the substantial number of women who were not recruited into the trial because of their preference for a particular treatment option. A larger number of participants with similar resource use and outcomes would have decreased uncertainty in our results. Based on the cost–utility framework, the potential trade-off between the additional QALYs gained associated with myomectomy and the potential benefits of avoiding a surgical procedure associated with UAE is not known.

Generalisability

Unlike previous comparisons27–29 of myomectomy and UAE, the FEMME trial was a multicentre clinical trial and did not include or exclude women based on their pregnancy intentions. The generalisability of the findings is increased by the inclusion of multiple centres, gynaecological surgeons and interventional radiologists, allowing both interventions’ impact to be evaluated without confounding by individual variance in clinical practice and skill, although nearly two-thirds of participants were recruited from just three hospitals. A substantial number of participants were of African-Caribbean ethnicity and presented with a wide range of fibroid diagnoses.

Patient and public involvement

We have been supported throughout the project by the Fibroid Network, in particular, and also FEmISA and the British Fibroid Trust. Public and patient involvement was crucial in improving the acceptability of the FEMME trial, in providing authenticity for the trial among women and in promoting recruitment. We engaged with the Fibroid Network founder throughout, developing an appreciation of the lack of choice often presented to women around the treatment of fibroids and uncertainty surrounding UAE and myomectomy among women, as well as the opinions of clinicians and barriers to accessing UAE that the chairperson had encountered. This prompted us to work harder to engage with clinicians to reiterate the uncertainty around both treatments, in particular for women desiring pregnancy.

We will engage with the Fibroid Network and British Fibroid Trust regarding the dissemination of our findings (FEmISA has ceased activity), providing a plain English summary of the findings and the uncertainties around the evidence we have discussed here. This will be distributed via their respective websites and via Twitter (Twitter, Inc., San Francisco, CA, USA; www.twitter.com). Any future research groups taking forward the research recommendations from this project would benefit from engaging with these groups and women directly via social media.

Implications for health care

Myomectomy and UAE are both established procedures within the repertoire of many gynaecologists and interventional radiologists, respectively, and training of junior doctors should continue to include these procedures. Services should continue to offer both procedures to women where both are potential options. Women, including those desiring a future pregnancy, should be provided with the evidence generated by the FEMME trial to enable them to make a fully informed decision regarding their fibroid treatment.

Research recommendations

The FEMME trial provides high-quality evidence on the clinical effectiveness and cost-effectiveness of UAE compared with myomectomy and, in our opinion, further clinical trials addressing this pragmatic question around quality of life are not required. An individual patient data meta-analysis is under way (PROSPERO CRD42018098676), which aims to identify which patients have the highest risk of unchanged or worsened quality of life and reintervention by 1 year.

Recommendations for further research encompass fertility concerns, long-term outcomes and technical aspects of the interventions. These may be prioritised as follows:

• The impact of myomectomy, compared with UAE, on live birth rates is the most urgent unanswered question. A comparable trial36 to the FEMME trial, specifically for women with fibroids and no other infertility factors, commenced in France (NCT02577055); however, this trial struggled to recruit in the face of the same concerns regarding the impact of UAE on ovarian function and was terminated after 15 participants were enrolled. 36 The lack of compelling evidence for adverse effects of myomectomy and UAE from the FEMME trial and other sources37 should reduce the barriers to a new randomised trial in women seeking to get pregnant naturally or undergoing assisted reproduction treatment. The effectiveness of hysteroscopic myomectomy of submucosal fibroids in women seeking treatment for infertility or who have had recurrent miscarriages, compared with deferred surgery, is being addressed in a UK randomised trial (URL: https://fundingawards.nihr.ac.uk/award/NIHR128969). The extent of uterine cavity distortion from submusocal fibroids and its impact on the success of fertility following myomectomy may also be revealed from this trial.

• Long-term follow-up of women who have undergone myomectomy or UAE using routine data to capture further reinterventions in the FEMME cohort would be of benefit.

• If and when the emerging drug classes of progesterone receptor modulators and oral gonadotropin-releasing hormone receptor antagonists gain marketing authorisation for use by women with uterine fibroids, then these drugs should be evaluated both against standard medical treatments, such the levonorgestrel-releasing intrauterine-releasing system, and against myomectomy or UAE.

• Adenomyosis is another common menstrual disorder, with symptoms that overlap with those of fibroids, that can often co-exist with fibroids, but is relatively overlooked. A randomised trial of UAE compared with hysterectomy is under way in the Netherlands (URL: www.trialregister.nl/trial/5471; accessed 15 December 2021). Many of the drugs that potentially provide benefit for fibroids or endometriosis could also help with the symptoms of adenomyosis and should also be evaluated in RCTs.

• Comparison of the surgical approaches to myomectomy may be considered, but the number, size and location of the fibroids are key determinants to the decision to operate hysteroscopically, laparoscopically or by open laparotomy. Outcome data, such as those collected in the FEMME trial and fibroid registries, may help to define and revise the criteria for deciding which types of fibroids require open myomectomy and which can be removed laparoscopically.

• There are opportunities for further research into some of the technical aspects of myomectomy (e.g. the minimisation of blood loss through the use of tranexamic acid or vasopressin85) and in pain control. Morcellation of fibroid tissue remains a controversial issue and, although modifications to the technique, such as in-bag morcellation, have been compared with uncontained power morcellation,86 the theoretical risk of dissemination of leiomyosarcoma will remain a barrier to further investigation of the value of morcellation.

• The choice of embolic agent for UAE has generally been at the discretion of the interventional radiologists, although randomised comparisons have been undertaken. 87,88 Further trials could help to refine the endovascular techniques and determine the impact of embolic agent not only on radiological features, such as the extent of infarction, but also on symptom and patient-reported outcomes.

• Post-embolisation syndrome is a common AE following UAE, particularly associated with larger fibroids, and, although self-limiting, can delay discharge and may mask infection. Identification of risk factors and optimisation of antipyretic and antiemetic therapy would improve the patients’ experience.

• There are few data on how widely thermal and microwave ablation, and ultrasound-guided high-intensity transcutaneous-focused ultrasound and MRgHIFU techniques, are being used in the UK, as routine data sources do not distinguish between modalities. Anecdotally, these techniques do not appear to be gaining traction in the UK, but if any become more mainstream then their clinical effectiveness and cost-effectiveness should be established against both myomectomy and UAE.

There is also scope for methodological research around the evaluation of fibroid treatments. A core outcome set of the most important outcomes to be measured and reported in clinical trials of fibroid treatment is under development. 89 This initiative will reduce research waste, but will define only the main domains not the actual outcome measures. The UFS-QOL is validated, but was developed from responses from women who had completed their family and, therefore, does not refer to the fertility and pregnancy concerns that women may have. Evaluation of the validity of generic fertility quality-of-life or anxiety questionnaires specifically among women with fibroids desiring pregnancy would be valuable and, if necessary, a new outcome measure for this population could be developed.

Cost–utility analyses using health utilities valued using a generic measure of HRQoL do not take into account patient preference. It is conceivable that some women may place additional value on a non-surgical procedure over a surgical alternative. Further research on how such fully informed patient preference may be quantified and incorporated into subsequent economic analyses of medical, surgical and non-surgical interventions for fibroids would be worthwhile. For example, patient preferences may be elicited through discrete choice experiments as a supplement to QALY measurement.

FEMME Collaborative Group

Contributions of authors

Jane Daniels (https://orcid.org/0000-0003-3324-6771) (Professor of Clinical Trials) was involved in the conceptualisation, writing (original draft, except Chapters 5 and 6, review and editing), project administration, investigation and funding acquisition.

Lee J Middleton (https://orcid.org/0000-0003-4621-1922) (Lead Medical Statistician) was involved in the statistical analysis plan, data curation, formal analysis and writing (review and editing).

Versha Cheed (https://orcid.org/0000-0002-6713-0913) (Medical Statistician) was involved in data curation, formal analysis and writing (review and editing).

William McKinnon (https://orcid.org/0000-0001-8734-2932) (Trial Management Team Leader) was involved in writing (review and editing), investigation and project administration.

Dikshyanta Rana (https://orcid.org/0000-0001-9133-3094) (Research Assistant, Health Economics) was involved in data curation, formal analysis (economic evaluation) and writing (original draft, including Chapters 5 and 6, review and editing).

Fusun Sirkeci (https://orcid.org/0000-0002-8221-5763) (Clinical Research Fellow, Gynaecology) was involved in investigation and writing (review and editing).

Isaac Manyonda (https://orcid.org/0000-0003-3080-2357) (Professor of Obstetrics and Gynaecology) was involved in the conceptualisation, writing (review and editing), investigation, funding acquisition and was a site principal investigator.

Anna-Maria Belli (https://orcid.org/0000-0002-3703-9596) (Professor of Interventional Radiology) was involved in the conceptualisation, writing (review and editing), investigation, funding acquisition and was a site principal investigator.

Mary Ann Lumsden (https://orcid.org/0000-0003-2998-9185) (Professor of Obstetrics and Gynaecology) was involved in the conceptualisation, writing (review and editing), investigation, funding acquisition and was a site principal investigator.

Jonathan Moss (https://orcid.org/0000-0002-4589-9174) (Professor of Interventional Radiology) was involved in the conceptualisation, writing (review and editing), investigation, funding acquisition and was a site principal investigator.

Olivia Wu (https://orcid.org/0000-0002-0570-6016) (Professor of Health Economics) was involved in the conceptualisation, formal analysis (economic evaluation), writing (review and editing) and funding acquisition.

Klim McPherson (https://orcid.org/0000-0002-5461-2569) (Emeritus Professor of Public Health) was involved in the conceptualisation, statistical analysis plan, writing (review and editing) and funding acquisition.

Publications

McPherson K, Manyonda I, Lumsden MA, Belli AM, Moss J, Wu O, et al. A randomised trial of treating fibroids with either embolisation or myomectomy to measure the effect on quality of life among women wishing to avoid hysterectomy (the FEMME study): study protocol for a randomised controlled trial. Trials 2014;15:468.

Manyonda I, Belli A, Lumsden MA, Moss J, McKinnon W, Middleton L, et al. Uterine artery embolisation or myomectomy for uterine fibroids. N Engl J Med 2020;383:440–51.

Rana D, Wu O, Cheed V, Middleton LJ, Moss J, Lumsden M-A, et al. Uterine artery embolisation or myomectomy for women with uterine fibroids wishing to avoid hysterectomy: a cost–utility analysis of the FEMME trial. BJOG 2021;128:1793–802.

Daniels J, Middleton LJ, Cheed V, McKinnon W, Sirkeci F, Manyonda I, et al. Uterine artery embolization or myomectomy for women with uterine fibroids: four-year follow-up of a randomised controlled trial. Eur J Obstet Gynecol Reprod Biol 2021;13:100139.

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to available anonymised data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health and Care Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, the HTA programme or the Department of Health and Social Care.