Principal findings of randomised trial
This randomised controlled trial compared the LNG-IUS and UPA in a population of women with self-reported HMB, with either small or no fibroids. At 12 months after randomisation, the impact of HMB, measured with the MMAS, was improved substantially from baseline in both groups. There was no evidence of a statistically significant difference in MMAS scores between treatments, though those allocated UPA exhibited a more rapid rate of improvement in MMAS scores. Those allocated to UPA had a greater magnitude of improvement in overall menstrual bleeding scores (as measured by PBAC) and were more likely to achieve amenorrhoea at 12 months (64% vs. 25%), yet this was not reflected in greater improvement in quality of life or sexual functioning. In those with uterine fibroids, there were no changes in fibroid or uterine volume in either treatment group.
Principal findings of mechanism of action study
Effects of UPA administration on the uterus
UPA produced a reduction in cell proliferation in the endometrium, as well as alteration of other local endometrial cellular markers (steroid receptor and steroid metabolising enzyme expression) consistent with a local endometrial oestrogenic environment. The effects on endometrial cellular markers were reversed upon withdrawal of UPA treatment. Stereological analysis in 19 patients showed that UPA did not produce a reduction in the volume of the uterus, irrespective of coexisting fibroids or adenomyosis. DCE-MRI in 15 patients showed that UPA appears not to have an effect on uterine blood flow. If adenomyosis was present in the uterus there was a significant increase in plasma volume in the endometrium. However, one of the five women with adenomyosis also had fibroids.
Effects of UPA administration on uterine fibroids
DCE-MRI studies showed that UPA produced an average reduction in plasma volume in 11 fibroids, which may be interpreted as due to a reduction in extracellular matrix components. This finding was not supported by stereological analysis, which failed to show a reduction in the total volume of fibroids in eight patients. However, it should be noted that the number of subjects studied is small.
Safety findings of UPA
There were no malignant changes in the endometrium of either patient group although one participant allocated to UPA administration developed endometrial hyperplasia with atypia. The MoA studies further support the absence of pathological changes despite a local oestrogenic microenvironment following treatment with UPA. Rates of PAEC following treatment were low and all had resolved within six months of treatment cessation. Two participants developed liver transaminase levels over three times the upper limit of normal, rising to three in the post-treatment period; none required hospital admission for management thereof. SAEs were infrequent and typically unrelated to treatment. There was one SUSAR, development of an acute hepatitis during the final course of UPA. This occurred prior to the initial USM. At initial presentation there was concern that this could represent a DILI, given the previous hepatoxic effects of other SPRMs such as onapristone109 and was thus reported as a SUSAR. However, the strong family history of autoimmune hepatitis led her hepatology team to conclude that this was the likely aetiology, and liver biopsy demonstrated widespread lymphoplastic hepatitis. The diagnosis of autoimmune hepatitis was not adjusted following the USM. The participant continues on long-term azathioprine with normal liver function. Following the USM, LFT eligibility and monitoring criteria were instigated in line with MHRA recommendations and it is likely that this participant would not have been eligible to participate.
Generalisability, equality, diversity and inclusion
In contrast to previous studies of SPRMs, typically limited to those with large fibroids only, the UCON trial has reflected a more ‘real-world’ participant population affected by the symptom of HMB, particularly given two-thirds of study participants had structurally normal uteri and those with small fibroids were evenly distributed between the two treatment regimens. Among the individuals recruited to the UCON trial, the mechanism of heavy bleeding is likely different to those with large uterine fibroids, and fewer therapeutic options are available. The average age of the participant population was 42.5 years and the duration of their symptom of HMB was considerable, with a median duration of symptoms of 3 years. This participant population reflects one with a high burden of symptoms, further reflected in a median MMAS score at entry of 37 in the UPA arm and 33 in the LNG-IUS arm out of a possible 100 (0 worst possible health, 100 best possible health state). This underscores the burden of HMB, reflecting a large and underserved patient population with unmet therapeutic needs.
The study population was predominantly white, with relative underrepresentation of other ethnicities, particularly black women, who represented only 2% of study participants. This may contribute to the relatively low rates of uterine fibroids. This lack of ethnic diversity is likely as a result of the majority study participants being recruited from Scotland, and that those with large fibroids were excluded. The potential impact of this lack of diversity is unclear. We did not collect data regarding social deprivation and level of education and so are unable to comment on diversity within these areas.
The UCON clinical trial recruited participants both from primary care and secondary care; however, the relative proportions of these differing populations were not recorded. Those presenting to secondary care may reflect a more refractory subset of women with HMB, having likely failed initial medical management in the community. As such they may have held differing views of acceptability of treatment strategy, which may in part explain why 40% of those eligible for, but declining participation to UCON, was due to a strong preference either for or against LNG-IUS. Furthermore, those who were relatively treatment naive may have had differencing expectations of what would constitute treatment success.
Adenomyosis may cause HMB,7 and although its presence was not an exclusion criterion, the incidence of adenomyosis within the UCON trial may have been underestimated. Rates of adenomyosis vary in the literature, reflecting both the impact of coexisting pathology and the challenge of diagnosis using non-invasive techniques, but a background rate of 20–30% is frequently reported,110 even in younger women.111 The sensitivity of ultrasound for detecting adenomyosis is less than MRI, but remains the first-line image tool of choice.112 In the MoA study, described in Chapter 4, participants taking UPA underwent MRI. In the 19 participants of the MoA study, 6 (32%) had adenomyosis on MRI imaging. However, an overall incidence of adenomyosis in those recruited to the main clinical trial, as determined by ultrasound scan, was 8.5%, and was evenly distributed between the two treatment groups. Given the findings of the MoA study, and background rates of adenomyosis it is possible that there may have been unrecognised cases of abnormal bleeding due to adenomyosis. As such, it is difficult to ascertain the impact of this coexisting pathology may have had upon treatment efficacy.
Limitations of the randomised trial
Impact of USMs on UCON
Undoubtedly, the impact of two USMs remains the greatest limitation of the UCON clinical trial. The impacts of these USM events were profound. Firstly, the halt in recruitment after the first USM resulted in early cessation of UPA treatment for many participants, impacting both on primary outcome data, including failure to recruit the necessary sample size, and introduced a significant level of statistical complexity, as outlined in Chapter 2. Consequently, the early termination of UPA treatment for many participants resulted in a significantly increased sample size requirement, following resumption of UPA use. The challenge of recruitment was further complicated by increased monitoring requirements (i.e. regular monthly LFTs for those participants allocated to UPA). This requirement may also have impacted retention and participant satisfaction. The perceived risk of DILI likely impacted upon study acceptability, both to potential participants and their managing clinicians resulting in reduced equipoise.
That the UCON clinical trial did still successfully recruit participants, despite these potential barriers following resumption after the first USM, reflects the unmet need of this participant population. However, following the second USM, it was deemed unacceptable to continue with study recruitment, resulting in failure to achieve the required sample size to ensure 90% power to address the study hypothesis. A smaller sample size than intended has hampered our ability to detect a conclusive difference in MMAS scores, so we would recommend caution in not interpreting our finding as equivalence for this outcome as the estimates of uncertainty were noticeably wide. In addition, MMAS scores were heavily skewed (to maximum score) and we were forced to rethink our original planned analysis which was designed to maximise our ability to find a difference between groups. Although we felt our re-engineered analysis treated the data in an appropriate manner, our approach will have had a negative impact on power to an unknown extent.
The second USM coincided with the rapid rise of COVID-19 cases and came into force on the 13 March 2020, just prior to the national lockdown in late March 2020. This placed additional burden on LFT monitoring for UPA-allocated participants, as well as delay in some end-of-study procedures such as ultrasound and endometrial biopsy. The impact of the pandemic on quality of life and satisfaction with treatment, particularly within the context of reduced access to alternative treatments strategies is unknown, but likely affected all participants in the final year of the study.
Absence of qualitative study
The absence of an embedded qualitative study reduced the ability to further assess the impact of prior treatment experience and symptoms duration on satisfaction and acceptability of the differing treatment arms. Similarly, the absence of a qualitative component precluded exploration of the disconnect between high rates of amenorrhoea and MMAS score while taking UPA, yet similar scores as those allocated LNG-IUS, who were significantly less likely to achieve amenorrhoea. Thus, the relative merits of reduction in menstrual bleeding volume versus predictable menstrual bleeding pattern can only be speculated. On first inspection, the inconsistency between the MMAS and PBAC results may be disconcerting, with reduced quantity of bleeding with UPA (as evidence by a higher amenorrhoea rate) not necessarily translating to higher MMAS scores compared with LNG-IUS. One hypothesis on why this may have been the case is that participants allocated UPA were considering the ‘off-treatment period’ of their UPA treatment schedule when completing the MMAS questionnaire, even though responses were solicited in the ‘on-treatment’ period. This may be understandable as five of the six MMAS questions (see additional editorial documentation) ask the responder to consider their previous menstrual cycle when completing the questions. Some women allocated UPA may have reflected on when they last bled, which may have been in the ‘off-treatment period’. While there was considerable variation in responses, there was some evidence of negative correlation between the two measures prior to randomisation (see Appendix, Figure 25). However, it was noticeable at follow-up assessment that participants allocated to UPA were less likely than the LNG-IUS group to report a high MMAS score even if they reported amenorrhoea (PBAC score = 0). PBAC scores were also solicited in the ‘off-treatment period’ in the UPA group (see Appendix, Figure 26). As can be observed, at these times many women returned to bleeding, with a number experiencing heavy bleeding (score > 100).
Absence of longer-term follow-up
Surgical intervention rates were low in the UCON clinical trial; however, follow-up was only for one year after commencing treatment. Previous studies of long-term follow-up of LNG-IUS usage for the symptom of HMB suggests a recourse to surgical intervention of 21% by 5 years and 29% by 10 years after medical treatment with LNG-IUS or other recommended oral treatments.8,10 However, the longer-term acceptability of the LNG-IUS needs to be interpreted in the context of changing fertility requirements across the life course, as well as reflecting possible treatment failure. Exploration of the rates of longer-term surgical intervention through data-linkage within the UCON clinical trial population would be of limited value, given the impact of the USM and the ongoing impact of COVID-19 on surgical waiting times.
Interpretation in context of other literature
Improvement on quality of life
The existing literature on impact of UPA use in women with HMB and resultant improvement in quality of life has been predominantly addressed in women with uterine fibroids, and as a result the condition specific quality of life outcome the UFS-QoL has been typically used, rather than MMAS, thereby precluding direct comparison. In the subset of those participants with uterine fibroids participating in the UCON clinical trial improvement in whom the UFS-QoL was additionally completed, the overall UFS-QoL was similar to that observed in the PEARL II study,24 despite participants in UCON having smaller fibroids (and less associated change in volume following treatment with UPA). However, some of the studies of UPA in women with fibroids have also assessed ED-5D-5L, suggesting an improvement in VAS of 15.6 points after three courses of UPA,54 slightly more than that observed in our UCON trial, although the UCON study reflects a smaller and clinically slightly different population, and thus direct comparisons should be interpreted with this limitation. Improvement in MMAS in LNG-IUS observed in the UCON trial were broadly similar to that observed in the EcLiPSE study.10 Similarly, improvement in the ED-5D-5L descriptive system in the LNG-IUS arm was similar to that reported in the EcLiPSE trial, but improvement in the health thermometer component was more marked in the LNG-IUS UCON trial arm (8.6-point improvement) than those receiving LNG-IUS in the EcLiPSE trial (1.2-point improvement).10
With an average age of participants of 42.5 years, it might be inferred that many of the participant population may no longer have been seeking fertility. As such surgical intervention for HMB, such as endometrial ablation or hysterectomy, may have been attractive alternatives. Hysterectomy is definitive treatment for the symptom of HMB. The recent HEALTH study17 reported that 69% of participants had a MMAS score of 100 after undergoing laparoscopic supracervical hysterectomy (LASH) for HMB, in contrast to 40% of those allocated UPA and 46% of those LNG-IUS within the UCON study. Furthermore only 12% had a score of 75 or less following LASH, compared with 38% receiving UPA and 30% receiving LNG-IUS. However, endometrial ablation and hysterectomy are fertility ending, and an older average age does not equate with completion of family, since 2013 in the UK over 50% of children are born to women in their fourth decade and above.113 Moreover, there may be a desire for uterine preservation irrespective of future fertility desire. Furthermore, hysterectomy is associated with both immediate surgical risk, and long-term consequences, including risk of cardiovascular disease and dementia,114,115 particularly if combined with concurrent oophorectomy.116 Therefore, despite improved quality of life compared to medical treatments, surgical intervention remains inappropriate for many, thus effective medical management options for HMB are likely to remain important to women117 and thus clinicians and researchers must continue to reflect this in management options offered. This was further reflected by very low rates of surgical intervention during study participation. Two participants were recommended surgical treatment during UCON participation for other indications, one hysterectomy following excision of a benign but mitotically active fibroid, a second participant underwent risk reducing bilateral salpingo-oophorectomy in view of a family history of breast cancer. While the indication for surgery was not HMB, in both cases definitive end of menstruation would have been achieved, and it is unclear whether their menstrual bleeding symptoms contributed to their decision for definitive surgery during their participation in the trial.
Adherence
Good adherence to the UPA treatment schedule in those unaffected by the USM was observed, with only 13/118 (11%) discontinuing treatment due to perceived lack of efficacy or side effects, although we accept that those women lost to follow-up may have been less likely to adhere with reasons unknown. Headache was the most frequently reported side effect and was responsible for nearly half of treatment discontinuations. This finding was similar to that observed in the longer-term UPA in the PEARL IV trial, where 75% completed four 12-week courses of treatment.118 Slightly more of those allocated LNG-IUS (17/118, 14%) discontinued treatment by 12 months [with a further 7/118 (6%) declining fitting], similar to rates observed in EcLiPSE.10 As expected, the predominant reason for discontinuation was impact on bleeding pattern.
Control of bleeding and amenorrhoea
Those allocated to UPA treatment had a greater control of menstrual bleeding at 12 months (PBAC < 100) compared with the group using the LNG-IUS. The control of menstrual bleeding observed with those participants receiving UPA treatment was 82%, slightly lower than that observed in the first PEARL study, reflecting a single course of UPA, but in PEARL IV, where participants received four courses of UPA, control of bleeding was achieved in 73% of those receiving 5 mg UPA. 64% of UCON participants achieved amenorrhoea at the end of the third treatment course, similar to PEARL IV, in which 67.1% of those receiving 5 mg UPA continued to achieve amenorrhoea at the end of treatment course 4. Control of menstrual bleeding at 12 months was achieved in 70% of those allocated the LNG-IUS, yet only 25% achieved amenorrhoea.
Other studies have reported improvements in haemoglobin levels following treatment with UPA,23 while there was no significant treatment difference observed in our study, mean haemoglobin at entry was 128 in the UPA arm, and 131 in the LNG-IUS, and thus potential for improvement was minimal.
Impact on uterine fibroid size
The majority of participants in our UCON clinical trial had structurally normal uteri, and only 24% of participants had fibroids. Of those with fibroids, these were typically small as inclusion was limited to those with 14-week or smaller-size uterus. In contrast, the PEARL23,24,54,118 and VENUS119,120 studies only included those with fibroids larger than 3 cm. In our UCON trial, the median volume of the largest fibroid was 13.4 ml in the UPA arm of UCON, compared with 100.7 cm3 (total fibroid volume), 79.6 cm3 (total of largest three fibroids), 25.1 cm3 (total fibroid volume) and 10.7 cc (total fibroid volume) in the corresponding 5 mg UPA arms of PEARL I,23 II,24 and VENUS I,120 and II,119 respectively. While the PEARL and VENUS studies used different metrics, the percentage reduction in size of uterine fibroids in those participants receiving 5 mg of UPA ranged from 21% to 36% after a single course of UPA23,24 to 71.8% after four courses118 in the PEARL studies. In the VENUS studies, where participants had a lower median fibroid volume at entry, reduction in size of fibroid was more modest, with 9.6% reduction in fibroid volume after one course120 and 13.7% after two 12-week courses of 5 mg UPA.119
In the UCON trial here reported, the size of the largest uterine fibroid, as assessed by ultrasound, was relatively unchanged after three 12-week courses, with mean volume increasing by 1.3 ml. There is clearly complexity in contrasting results between reported studies, given differing modalities and techniques of assessing fibroid volume, and difference in outcome measures used with regard to number of fibroids per participant included for analysis. However, the findings of the main UCON study were supported by the embedded MoA study (see Chapter 4), where no significant change in the three largest or total volume of fibroids, as assessed by MRI, was observed. Other studies have also demonstrated limited impact of UPA on fibroid size,69,70 particularly in smaller fibroids, similar to those participating in the UCON study.
On DCE-MRI, UPA treatment increased plasma volume, but not plasma flow within the uterine fibroids, as described in Chapter 4. This change in plasma volume may reflect change in the proportion of extracellular matrix, changes in which have previously been identified following treatment with UPA.102,103,121 Equally shrinkage of fibroid size following UPA treatment has been associated with reduction in Versican proteins, which could lead to expulsion of water, thereby decreasing the hydrostatic swelling potential.121 Absence of change in fibroid plasma volume may reflect absence of change in Versican proteins. This may be a factor contributing to the lack of expected reduction in fibroid volume seen in the UCON trial. Other factors in the local uterine microenvironment of uterine fibroids that vary between small and large fibroids may contribute to an altered response to UPA administration observed in our study. As UCON participants were not undergoing surgical intervention we are unable to further explore the lack of expected shrinkage in size observed in our population.
Adenomyosis
The mechanism of HMB in those women with adenomyosis may be different to that of an endometrial or fibroid aetiology;15,122 however, the LNG-IUS is an established treatment for AUB due to adenomyosis,123 and SPRMs have shown promising utility in the management of this condition.124 None of the 12 participants allocated LNG-IUS with coexisting adenomyosis remained within the study at 12 months, whereas 5 of the 8 participants with adenomyosis allocated UPA completed three cycles of treatment. Due to the small number of those identified with adenomyosis in our UCON study, it is not possible to comment further on the efficacy of UPA versus LNG-IUS within our trial participant population. A small placebo-controlled trial of women with adenomyosis and symptom of HMB had a significant effect on PBAC score with amenorrhoea rates of 95.2% following 12 weeks of 10 mg UPA, with concurrent reduction in pain score.125 However, this was not associated with improvement in quality of life, as assessed by the UFS-QoL. Future studies of HMB in those without large uterine fibroids would need to encompass the potential impact of pain, as well as control of menstrual bleeding, as both contribute to quality of life.
Influence of ethnicity
Following treatment of women with the symptom of HMB and uterine fibroids > 3 cm with 10 mg UPA for 12 weeks, a study of Japanese women, observed slightly higher amenorrhoea rates and greater reduction in fibroid volume than that in the previously reported (predominantly white) PEARL I study (87% vs. 82% and −23.7% vs. −12%, respectively).23,126 Furthermore, previous studies have observed higher rates of amenorrhoea and patient satisfaction in white populations compared with black participants following treatment with UPA, despite similar improvement in fibroid symptomatology.127 However, further studies would be required to determine whether a similar effect would have been observed following treatment with UPA in participants with non-white ethnicity who are experiencing symptoms of HMB, but in whom no fibroids are present.
Progesterone receptor modulator-associated endometrial changes
The UCON trial used a differing treatment schedule to previous studies, with a four-week treatment-free interval between UPA treatment courses, irrespective of whether a withdrawal menstrual bleed had occurred. A previous study with prolonged treatment the SPRM asoprisnil without withdrawal bleeds resulted in a case of complex hyperplasia without atypia and another case of low-grade endometrial adenosarcoma.38 While the latter may have been pre-existing, there remains concern about unknown long-term endometrial effects. In our treatment schedule we did not observe endometrial malignancy despite the potentially shortened ‘off-treatment’ interval. The endometrial thickness in the UPA arm was not increased relative to baseline thickness. Furthermore, rates of PAEC at the end of study were similar to those observed in reported studies of longer off-treatment intervals between treatment cycle schedules118 (and considerably lower than reported in the VENUS I study).120 There was no case of residual PAEC six months following cessation of treatment. The MoA study observed reduction in cell proliferation, in keeping with previous published literature,28 and also demonstrates that the alteration of the endometrial microenvironment reverses on cessation of treatment. This may provide added reassurance to those contemplating pregnancy following treatment given limited outcome data of pregnancies following UPA treatment to date.128,129
Implications for decision makers
On the basis of recruited participants, our study has shown that UPA treatment achieves amenorrhoea, improves quality of life and has a high rate of patient satisfaction. However, the risk of DILI has meant that UPA is no longer recommended as a therapeutic option for those with small (< 3 cm) fibroids in the UK, and UPA unlikely to proceed to be licensed as a treatment for those patients with structurally normal uterus with the symptom of HMB. While surgical intervention for HMB, particularly hysterectomy, is associated with greater improved quality of life, it is typically fertility ending, has a risk of serious complications, and higher associated mortality risk compared with that of fatal liver injury following UPA treatment (> 1 : 100 vs. 0.1 : 100,000).32 At present, given very low rates of DILI, and that multiple studies have demonstrated improvement in quality of life, there remains a role for UPA treatment as a therapeutic option for symptomatic uterine fibroids as outlined by NICE,9 particularly given the medical alternative with GnRH analogues is associated with more deleterious impact on bone density and other unwanted side effects such as vasomotor symptoms24. However, in those with small uteri, the risk–benefit may be less, given the equivalent impact on quality of life compared with existing alternative option of LNG-IUS, a treatment strategy typically less successful or inappropriate in those with large fibroids. While LNG-IUS remains a popular choice for many women with HMB, and in some countries use has increased exponentially over the last two decades,130 data from previous studies are consistent with our finding that the LNG-IUS is not effective for all women, and for many is unacceptable for an initial trial of treatment. There remains a pressing need for effective oral medical treatments for HMB with an acceptable side effect profile.
Recommendations for future research
SPRMs remain an attractive class of compounds, given their oral route of administration, high rates of amenorrhoea, preservation of bone density and lack of long-term endometrial effects. While UPA is now unlikely to be licensed for use in the patient population reflected in the UCON study, in the future, other SPRMs may be developed that do not have the reactive metabolite formation131 and thereby should be considered as new therapeutic options, given the demonstrable efficacy and acceptability to patients. Until these are available a promising class of compounds are the oral GnRH receptor antagonists, explored as a therapy for endometriosis associated pain,132 which are well tolerated, and have minimal impact on long-term risk of fracture.133 When used for HMB in association with fibroids, control of bleeding was achieved in up to 84.1% of participants compared with placebo,90 and is again associated with preserved bone mineral density when used with HRT combination therapy.134
Our view is that a study in a similar population as that in our UCON clinical trial, namely those with HMB in conjunction with small or no uterine fibroids, is a comparison between an oral GnRH receptor antagonist compared with a progestin. Ideally this would be large enough to permit differential analysis between those with HMB resultant from AUB due to adenomyosis, leiomyoma and endometrial dysfunction,7 informing both precision-based care but also reflecting a population where the primary complaint is HMB. Whether the underlying mechanism between these three pathologies is significant in the context of medical treatment strategy is unknown.
The optimal condition-specific patient-reported outcome measures remain uncertain, with a multitude of measures previously reported.135 Development of a core outcome set for HMB by Cooper et al.136, under the CROWN/COMET initiative,137 will aid potential to draw insights from pooled data. Additional learning points from the UCON trial include the need for inclusion of an embedded qualitative study, to explore the disconnect between achieving amenorrhoea and improvement in quality of life. Finally, our experience with not one but two urgent safety measures, and in particular the resultant complexity for statistical analysis, we hope will aid other clinical trialists in the future.
Given our findings from the UCON study, our recommendations for future research include:
- Further studies of medical treatments for HMB
- a. Developing and utilising other SPRMs, not associated with DILI
- b. Other hormonal/non-hormonal medical treatments for HMB
- Patient populations that encompass both the symptoms of HMB, and underlying aetiologies, including structurally normal uteri, adenomyosis and small fibroids
- Study design with outcome measures including impact on bleeding pattern, pain and impact on haemoglobin and iron-deficiency, as well as quality of life
- Qualitative studies to determine what are the most important outcomes to women who suffer HMB.
Conclusion
Both UPA and LNG-IUS improve bleeding symptoms and alleviate the adverse impact of heavy menstrual bleeding on quality of life. UPA is now only available for intermittent treatment of moderate to severe uterine fibroid symptoms before menopause and when surgical procedures (including uterine fibroid embolisation) are not suitable or have failed. New, effective and acceptable oral medical treatment options are needed to address an important unmet clinical need.