PROgesterone Therapy for Endometrial Cancer Prevention in Obese Women (PROTEC) Trial: A Feasibility Study Free

Endometrial cancer is the sixth most common cancer in women, with more than 382,000 new diagnoses and 89,900 deaths recorded globally in 2018 (1). The incidence of endometrial cancer is rising sharply in parallel with escalating obesity rates (2). Obesity is the strongest risk factor for the most common histologic subtype, endometrioid (type I) endometrial cancer, and its precursor lesion, atypical hyperplasia (3). Such is the strength of the association that approximately 40% of endometrial cancers are thought to be directly attributable to obesity (4), and a marked dose–response relationship bestows higher risk as body mass index (BMI) rises (5). It has been estimated that women with obesity class III (BMI > 40 kg/m²) have a 7-fold increased risk of endometrial cancer compared with normal weight women (BMI 18.5–25 kg/m²; ref. 3). The biological mechanism responsible for this association relates to the endometrial stimulatory effect of adipose-derived estrogen, which is unopposed by progesterone in an ovulatory and postmenopausal women, and augmented by the negative consequences of insulin resistance and chronic inflammation (6). Weight loss achieved and sustained through bariatric surgery reduces endometrial cancer risk (7, 8) with measurable impact on circulating biomarkers of adiposity, reproductive hormones, and insulin status, accompanied by downregulation of pro-oncogenic signaling pathways in the endometrium (9). Bariatric surgery is neither available, appropriate, nor acceptable to everyone with an elevated BMI, however, and cannot be recommended solely for the purpose of primary prevention of endometrial cancer (10). Dietary caloric restriction can facilitate weight loss, particularly if accompanied by increased levels of physical activity, but the amount of weight lost and duration of benefit is considerably lower than following bariatric surgery (11). Alternative strategies are therefore urgently needed to provide protection to women at greatest risk of endometrial cancer in order to thwart the explosion in incidence rates predicted by modeling studies (12, 13).

The levonorgestrel intrauterine system (LNG-IUS) delivers progestin directly to the endometrium, counteracting the stimulatory effect of estrogen through stromal decidualization, downregulation of proliferative signaling pathways, and glandular atrophy (14). Epidemiologic studies have shown ever-users of the LNG-IUS have a reduced risk of endometrial cancer (15, 16), and several meta-analyses have demonstrated its effectiveness as a therapeutic agent for women with atypical hyperplasia and low-grade cancers confined to the endometrium (17–19). Despite strong evidence for its anticancer activity, no previous studies have investigated use of the LNG-IUS as a chemopreventive agent for the primary prevention of obesity-driven endometrial cancer. It is not known whether women with a raised BMI are aware of their increased risk of endometrial cancer or whether they would be prepared to engage in risk reduction with a LNG-IUS. In preparation for a clinical efficacy study, we measured feasibility, participation rate, and compliance with the LNG-IUS for endometrial protection in women with class-III obesity. We studied its short-term effects on endometrial morphology, proliferation, and hormone receptor status and on circulating biomarkers of endometrial cancer risk. Further, we explored the impact of the LNG-IUS on menstrual bleeding patterns, as well as mental wellbeing, through validated questionnaires.

The study was sponsored by Manchester University NHS Foundation Trust (MFT) and approved by the Cambridge East Research Ethics Committee—(15/EE/0063), Medicine and Healthcare Products Regulatory Authority (MHRA, reference 21387/0234/001–0001), and local Research and Development departments. The trial was prospectively registered on the European (EudraCT number 2014–005610–37) and UK (ISRCTN40940943) clinical trial databases and conducted in accordance with Good Clinical Practice guidelines and the Declaration of Helsinki.

This was a single-arm feasibility study of the LNG-IUS for endometrial protection in women with class III obesity. Women attended clinic at baseline (screening visit, T0), 2 ±1 months (LNG-IUS inserted in clinic, T1), and 8 ±3 months (final assessment, T2; Fig. 1). Serial assessment of anthropometric measures (weight, BMI, and waist:hip ratio), serum biomarkers (hormone status, insulin resistance, and adiposity), endometrial biomarkers (endometrial morphology, hormone receptor status, Ki-67 proliferation index, and pro-proliferation signaling molecules), menstrual bleeding patterns, and mental wellbeing was performed at all time points.

Willingness to receive the LNG-IUS for endometrial protection was determined as the proportion of eligible women who agreed to its insertion. The feasibility of using the LNG-IUS in women with class-III obesity was calculated as the proportion of successful LNG-IUS insertions. Complications of insertion, side effects, and adverse events were recorded. At T2 final visit, women chose whether to keep their LNG-IUS for ongoing endometrial protection or have it removed. Adherence with repeated endometrial sampling and other trial procedures was recorded. Compliance with the LNG-IUS was calculated as the proportion of women who chose to keep their LNG-IUS.

Eligible women were ≥18 years of age with a BMI ≥40 kg/m² and histologically normal endometrium at baseline. All participants gave written, informed consent. We advertised the study on the University of Manchester and MFT websites, Cancer Research UK and UK ISRCTN clinical trials databases, on social media platforms, and by word of mouth. We recruited women who approached the research team for participation directly and those attending gynecology and sleep apnea outpatient clinics at MFT and Salford Royal Hospitals NHS Foundation Trust, respectively. Exclusion criteria included previous hysterectomy; LNG-IUS or other intrauterine device within the past 6 months; planning pregnancy, pregnant, or breastfeeding; previous endometrial ablation; congenital or acquired uterine anomaly; history of pelvic inflammatory disease or genital actinomyces; breast cancer; overdue cervical screening or last screen abnormal; immunodeficiency; actively trying to lose weight; contraindications to LNG-IUS, including coagulopathy, liver disease, migraine, raised blood pressure, arterial disease, postpartum endometritis, infected abortion during the past 3 months or recent trophoblastic disease with persistently elevated hCG levels; and inability to tolerate endometrial sampling/ LNG-IUS insertion as an outpatient.

At baseline, we recorded last menstrual period (LMP), menstrual bleeding pattern, and contraceptive use. Postmenopausal status was defined as LMP occurring >1 year before if follicle-stimulating hormone (FSH), luteinizing hormone (LH), and estradiol levels were confirmatory; the remaining participants were considered premenopausal. A urinary pregnancy test was performed if indicated. Cervical screening was offered in accordance with the National Health Service Cervical Screening program. High vaginal and endocervical swabs were taken to exclude active lower genital tract infection. Medical history was documented. Screening bloods, including full blood count, urea, and electrolytes and liver function tests were taken to confirm medical fitness for participation in the trial.

The Mirena LNG-IUS (Bayer plc) was inserted in clinic at T1; women were advised to take paracetamol and NSAIDs 1 hour before insertion, if not contraindicated. The procedure was carried out on a colposcopy couch using a Winterton speculum under aseptic conditions, according to the manufacturer's instructions. Safety monitoring was by telephone call at 6, 12, and 18 weeks following LNG-IUS insertion. Side effects, adverse events, and complications were recorded. Participants were advised to attend their General Practitioner for a coil thread check 4 weeks after LNG-IUS insertion.

Height was measured using a stadiometer with shoes removed. Weight was measured using electronic scales following removal of bulky clothing and BMI derived using the formula kg/m². Waist to hip ratio was calculated from waist (midpoint between lower margin of last palpable rib and top of the iliac crest measured with a tape measure) and hip circumference (widest portion of the buttocks).

Serum obtained by venepuncture following a 6-hour fast was used to measure (i) reproductive function (LH; FSH; sex hormone binding globulin, SHBG; testosterone; free androgen index, FAI; estradiol; progesterone), (ii) insulin resistance [glucose and insulin to derive Homeostasis Model Assessment: Insulin Resistance, HOMA-IR (20); glycosylated hemoglobin A1c, HbA1c), (iii) adiposity (adiponectin, leptin), and (iv) inflammation (C-reactive protein, CRP). With the exception of adiponectin and leptin, all analytes were measured using automated routine clinical service protocols in the MFT Clinical Biochemistry Laboratory. Adiponectin and leptin were measured with a DuoSet ELISA development kit (R&D Systems).

Endometrial sampling was performed using a Pipelle (Carefusion) or MedGyn Endosampler (MedGyn). Premenopausal participants were sampled on day 12 ±2 of the menstrual cycle, where possible. Endometrial tissue was formalin-fixed, paraffin-embedded, sectioned, and stained with hematoxylin and eosin. Endometrial morphology was assessed by a consultant gynecological pathologist. In premenopausal participants, endometrial morphology and reproductive hormone profile was used alongside LMP to determine menstrual cycle phase. Abnormalities were confirmed by a second consultant gynecological pathologist and classified according to World Health Organization guidelines (21, 22).

Tissue sections (4 μm) were baked for 30 minutes at 70°C. The automated Ventana BenchMark Ultra IHC Staining Module (Ventana Co.) was used with the Ultraview 3, 3′ diaminobenzidine (DAB) v3 detection system (Ventana Co.). Tissue sections were deparaffinized and incubated in EZPrep Volume Adjust (Ventana Co.). A heat-induced antigen retrieval protocol was carried out using a TRIS–ethylenediamine tetracetic acid–boric acid pH 8 buffer, Cell Conditioner 1 (CC1). The sections were incubated with ultraviolet inhibitor blocking solution for 4 minutes, followed by an optimized concentration of antibody (Supplementary Table S1). Sections were then incubated with horseradish peroxidase–linked secondary antibody, DAB chromogen, and copper. Counterstain (Hematoxylin II) was applied for 12 minutes before a 4-minute incubation with bluing reagent. Slides were dehydrated through three steps of 99% industrial methylated spirits (IMS) and two changes of Xylene. Sections were coverslipped using ClearVue Mount XYL (Thermo Scientific). Negative (isotype control) and positive tissue controls were used for quality assurance.

The Ki-67 score was the proportion of glandular cells with positive nuclear staining. The Ki-67 score was determined from >1,000 nuclei scored in 3 representative high-powered fields (x20), chosen by the study pathologists; scanty samples were scored in their entirety (23). Estrogen (ER) and progesterone receptor (PR) staining was assessed by modified H-score (0–18), the product of area score (proportion of positively stained tissue, scored 0–6), and intensity of staining score (0 = none, 1 = mild, 2 = moderate, and 3 = strong). Phosphorylated (p)AKT staining was scored using the percentage of positively stained tissue [H = (3× % strong staining) + (2× % moderate staining) + (% weak staining)] to account for within tissue heterogeneity (0–300). PTEN status was scored “PTEN null” if there were endometrial glands negative for PTEN adjacent to positive stroma. Slides were scored as “PTEN positive” if all endometrial glands expressed PTEN (24). Scoring was performed manually by two independent scorers who were blinded to time point. Discrepant scores (>10% or disagreement as to PTEN status) were reviewed and resolved by consensus agreement.

Two validated questionnaires, the Hospital Anxiety and Depression Scale (HADS; refs. 25, 26) and Warwick–Edinburgh Mental Wellbeing Scale (WEMWBS; refs. 27, 28), were completed at baseline and follow-up to determine whether the LNG-IUS had an impact on mental wellbeing. For the HADS, different cutoffs are indicative of a mental health disorder, depending on clinical context (26), but lower scores indicate absent or lower severity of symptoms. For the WEMWBS, the mean score in the general population is 51, with higher scores reflecting improved mental wellbeing (27). Premenopausal participants completed the Menstrual Bleeding Questionnaire (29) at baseline and follow-up.

This was a preliminary study designed to inform recruitment rates, feasibility of and likely adherence to a clinical efficacy trial of the LNG-IUS for endometrial protection in women with class-III obesity. We considered that a clinical efficacy trial could be successfully conducted if >50% of eligible women agreed to participation, >50% of those eligible had a LNG-IUS successfully fitted, and >75% of women kept their LNG-IUS for >6 months. We also measured LNG-IUS–induced change in circulating and tissue biomarkers to inform intermediary biomarker endpoints for our definitive study. We did not perform a formal sample size calculation and planned the pragmatic recruitment of 30 to 40 women over a 6- to 12-month recruitment period.

Statistical analyses were performed using GraphPad Prism 5.0b for Mac (GraphPad Software) and SPSS 23.0 for Mac (IBM Corp.). Descriptive statistics included mean and SD for normally distributed, and median and interquartile range (IQR) for nonnormally distributed, data. Within-individual changes over time were compared using paired t test and Wilcoxon signed-rank test for normally distributed and nonnormally distributed data, respectively. To assess the short-term impact of the LNG-IUS on endometrial proliferation, a mixed effects regression model was fitted, with Ki-67 score set as the dependent variable, time point (baseline set as reference category) as the predictor of interest and the covariates baseline Ki-67 score, age, menopausal status (pre/post), smoking (never, ever, current), type II diabetes mellitus (yes/no) baseline BMI, and baseline waist:hip ratio. A further analysis was performed that included weight at follow-up, to determine if change in weight was responsible for change in Ki-67 at outcome. To account for repeated measures within participants, a random effect intercept was included to account for the within subject versus between subject variation. To account for possible departures in normality, a cluster bootstrapping procedure was employed with 1,000 replications. In an effort to emphasize clinically over statistically important effects, data are reported in terms of mean difference effect estimates and 95% confidence intervals (CI).

Between October 2015 and September 2016, 103 women were approached, 54 were offered a participant information sheet, 35 agreed to participate, and 25 received a LNG-IUS (Fig. 1). Forty-nine women (48%) were ineligible to receive the participant information sheet for the following reasons: LNG-IUS in situ (n = 13); previous hysterectomy (n = 12); pending bariatric surgery (n = 10); social/capacity reasons (n = 7); LNG-IUS contraindicated (n = 4), or BMI < 40 kg/m² (n = 3). Nineteen women (35%) declined participation for the following reasons: none given (n = 8); declined device (n = 3); declined “hassle” (n = 3); declined procedures (n = 3), or other reasons (n = 2). Thus 35 of 54 (65%) eligible women consented to the baseline screening assessment, 10 of 35 (29%) either failed screening or withdrew their consent, and 25 of 35 (71%) proceeded to LNG-IUS insertion. Three of the six screen failures (3/35, 9%) had occult atypical hyperplasia or endometrial cancer on their baseline biopsy. All four withdrawals (4/35, 11%) found baseline endometrial sampling too painful to consider further procedures. Our final study population comprised 25 women with a median age and BMI of 54 years (IQR 52–57) and 47 kg/m² (IQR 44–51), respectively (Table 1). There were no significant differences between women who consented to participate and women who declined in terms of age (54 vs. 52 years), BMI (47 vs. 46 kg/m²), or ethnicity/race (all except three were White British). Nine (36%) were premenopausal but just 4 had regular menstrual cycles; most were either amenorrheic (2/12) or experienced irregular menstrual bleeding (3/12). Four (16%) were using hormone replacement therapy (3/25, 12%) or oral contraceptives (1/25, 4%) at baseline, which they continued throughout the trial. All had at least one comorbidity, most commonly type II diabetes (10/25, 40%), hypertension (15/25, 60%), or asthma (8/25, 32%), and 48% had more than three comorbidities.

All 25 women received the LNG-IUS in clinic without complication. There were no insertion failures, expulsions, uterine perforations, or lost devices. One woman (4%) developed mild symptoms of endometritis following LNG-IUS insertion, which was treated with oral antibiotics. One patient complained of pelvic discomfort/mild pain following LNG-IUS insertion that settled with oral analgesia. Other adverse and serious adverse events, specifically urinary tract infection (1/25, 4%), vasculitis (1/25, 4%), sciatica (1/25, 4%), and attempted suicide (1/25, 4%), were not thought to be related to the LNG-IUS. All women kept their LNG-IUS until their final assessment when one woman (4%) chose to have it removed (“easier now than later”); the remaining 24 women (96%) kept their LNG-IUS for ongoing endometrial protection. All 25 women were compliant with study procedures, including sequential endometrial biopsies (all 3 biopsies taken, 25/25).

Three of 35 participants (9%) had an incidental finding of atypical hyperplasia or endometrial cancer on a pre–LNG-IUS biopsy and were excluded from the study. All other women had histologically normal endometrium at T0 (baseline) and T1 (time of LNG-IUS insertion). Many of the samples were scanty. Morphology was consistent with menopausal status and/or reported phase of menstrual cycle, as appropriate. At follow-up (T2), all endometrial biopsies showed stromal decidualization and glandular atrophy, consistent with the progesterone effect associated with LNG-IUS treatment.

The LNG-IUS was associated with a significant decrease in endometrial proliferation as assessed by Ki-67 score. The mean Ki-67 score was 27.1% (SD 23.4) at baseline, 21.8% (SD 14.8) at the time of LNG-IUS insertion, and 12.7% (SD 10.9) at follow-up. A mixed effects regression model adjusting for within participant clustering, potential confounders, and weight change between time points is shown in Table 2. Between baseline (T0) and time of LNG-IUS insertion (T1) and between baseline (T0) and follow-up (T2), the change in Ki-67 score was −5.4% (95% CI, −17.1%, 6.3%) and −14.6% (−25.3%, −3.9%), respectively. These results were consistent across all three models, indicating that potential sources of confounding, including change in weight during follow-up, had little effect on Ki-67 score. PR expression decreased with LNG-IUS treatment (Table 3). There was no significant change in expression of the other endometrial biomarkers, estrogen receptor, PTEN, or pAKT. Interestingly, all three women excluded because of occult endometrial abnormalities had PTEN-null glands, and a further 2 of 25 participants in the study had PTEN-null glands before but not after LNG-IUS insertion.

Overall, women lost weight during the trial, although this was not clinically significant [median weight 124.4 kg (IQR 111–143), 123.9 kg (IQR 111–142), and 123 kg (IQR 111–144) at T0, T1, and T2, respectively]. There were no clinically significant changes in circulating biomarkers of reproductive function, insulin resistance, adiposity, or inflammation across the three time points (Table 3), with the notable exception of altered serum FSH, LH, and progesterone levels over time, which likely reflect natural reproductive aging in our perimenopausal cohort.

Of the 9 premenopausal participants, 2 (22%) were amenorrheic, 4 (44%) had regular, and 3 (33%) irregular menstrual bleeding at baseline. As expected, the 7 women who experienced menstrual bleeding reported a significant reduction in blood loss with the LNG-IUS; all but two became amenorrheic according to the Menstrual Bleeding Questionnaire. Mental wellbeing improved with the LNG-IUS according to both the WEMWBS and HADS scales (Table 4). A change in score of 2–3 points is clinically significant, but did not reach statistical significance, most likely because of small numbers.

The PROTEC trial was undertaken to assess the feasibility of a future clinical efficacy trial of the LNG-IUS for endometrial protection in women with class-III obesity. In a 12-month recruitment period, we approached 103 women, 54 (52%) of whom met the inclusion criteria, 35 (65%) agreed to participate, and 25 (71%) proceeded to LNG-IUS insertion. There were no insertion failures, and all women were fully compliant with all study procedures, including an endometrial biopsy at the final visit. There were no related serious adverse events but one case each of endometritis and postinsertion pain, both recognized complications of LNG-IUS treatment. We observed no detrimental impact of the LNG-IUS on mental wellbeing and self-reported menstrual bleeding profiles improved for our premenopausal participants. These data suggest that women at greatest risk of obesity-driven endometrial cancer are willing to engage in risk reduction with a LNG-IUS and that a clinical efficacy trial could be feasible. Roughly a quarter of women approached for the trial had a LNG-IUS inserted, indicating a relatively high proportion of screen failures and noncontinuation rate. These findings must be factored in to the design of a clinical efficacy trial.

To determine the short-term impact of the LNG-IUS on biomarkers of endometrial cancer risk, we measured change in anthropometric variables, reproductive hormones, insulin resistance, endometrial morphology, and glandular proliferation status between baseline, 2 months, and 8 months. As predicted, we observed stability in these biomarkers prior to LNG-IUS insertion. Short-term treatment with the LNG-IUS was associated with changes in endometrial morphology, reduced proliferation, and PR expression; there were no associated changes to circulating hormone levels, measures of insulin resistance, or adiposity. An unexpected finding was that women lost an average 2.5 kg in weight during the study; however, this was neither clinically nor statistically significant. Given our conviction that women with class-III obesity are at sufficiently high risk of endometrial cancer that they would benefit from risk-reducing measures, it is striking that 3 of 35 (9%) of our participants had an incidental finding of atypical endometrial hyperplasia or endometrioid endometrial cancer at baseline, requiring hysterectomy. This is consistent with our previous study that found 10 of 72 (14%) women with class-III obesity referred for weight loss management had occult underlying endometrial neoplasia (9).

Endometrial glands are clonal cell populations that frequently harbor driver mutations in cancer genes (30). PTEN-null glands confer a proliferative advantage, predisposing to endometrial carcinogenesis, and have been shown to persist between menstrual cycles, but only a small proportion progresses to endometrial cancer (31). We found PTEN-null glands in all three women with occult endometrial abnormalities and in two participants with histologically normal endometrium before, but not after, LNG-IUS insertion. This is consistent with the hypothesis that PTEN-null glands in morphologically normal endometrium represent latent endometrial cancer precursors that regress with LNG-IUS treatment (24).

Although the endometrial impact of the LNG-IUS is well studied (32, 33), this is the first trial to offer the LNG-IUS to women with class-III obesity for the primary prevention of endometrial cancer. Confirmation that the expected endometrial effects of the LNG-IUS are observed in this population is important given their defining characteristics, specifically their class-III obesity and amenorrhea or irregular menstrual bleeding, which distinguish them from regular users of the device. It is known that Ki-67 score is higher in endometrium harvested from women with obesity compared with that collected from normal weight women (34). We considered that the expected change in endometrial morphology, a reduction in glandular proliferation as assessed by Ki-67 score, and downregulation of endometrial progesterone receptors would confirm the utility of the LNG-IUS at standard doses in this population. We also quantified the Ki-67 drop at 6 months post–LNG-IUS insertion for the purposes of developing an intermediary molecular endpoint for a definitive trial of the LNG-IUS for endometrial protection upon which a sample size calculation could be based. It is interesting that the 15% Ki-67 drop observed after 6-month treatment with the LNG-IUS was similar to that observed after an average bariatric surgery–induced weight loss of 22 kg at 2 months in women with class-III obesity (9). Bariatric surgery is known to reduce endometrial cancer risk (7, 8, 35), and although the mechanisms underlying risk reduction are not fully understood, it is thought that downregulation of endometrial pro-proliferative signaling pathways could be important (9, 36). Ki-67 is only expressed by proliferating cells, a hallmark of cancer; indeed, Ki-67 is known to differentiate benign from malignant endometrium, with higher Ki-67 scores observed in high-grade, advanced stage cancer and correlating with poor survival outcomes in this group (37). We considered that a reduction in glandular proliferation in benign peri- and postmenopausal endometrium could reduce the risk of mutational events that trigger malignant transformation (10).

We have demonstrated proof of principle that some women at high risk of obesity-driven endometrial cancer are prepared to engage in risk reduction with a LNG-IUS, paving the way for a clinical efficacy trial in this population. Despite concerns that LNG-IUS insertion would be challenging in the outpatient setting in postmenopausal women with class-III obesity, we had no insertion failures, consistent with previous studies (38). Concerns that uterine instrumentation would be unacceptable to women who did not have a gynecological complaint were also unsubstantiated, with 32 of 35 (91%) participants consenting to and undergoing three sequential biopsies, without complication. Further, we found the LNG-IUS was not associated with a detrimental impact on mental wellbeing using two validated questionnaires, with even some suggestion that mental wellbeing improved during the trial, possibly due to improved menstrual bleeding profiles and peace of mind regarding endometrial health; indeed 24 of 25 (96%) participants chose to keep their LNG-IUS at the end of the trial for ongoing endometrial protection. A short-term study of this kind cannot confirm that women will be compliant with the LNG-IUS in the medium to long-term, however. Nor can it help define the optimal duration of a clinical efficacy trial. We did not deliberately target perimenopausal women for trial participation, although long-term use of the LNG-IUS in a clinical efficacy trial would ideally avoid women whose compliance could be compromised by future pregnancy plans. The single-center nature of this research is a limitation of the study, because we cannot necessarily extrapolate feasibility of our approach to other centers, countries, or healthcare settings. The lack of racial and ethnic diversity in our study population precludes any insight into the acceptability and uptake of the LNG-IUS for uterine protection in non-White British women. We do not know whether women would consent to randomization to a no intervention arm, which would be the ideal clinical efficacy trial design, and would certainly affect feasibility of the definitive study. Furthermore, our biomarker findings should be interpreted with caution given the small sample size and marked heterogeneity of participating women with respect to age, menopausal status, and use of exogenous hormones at baseline.

Although invasive, the advantage of the LNG-IUS is that it releases a continuous supply of levonorgestrel directly to the endometrium, avoiding the peaks and troughs observed with oral administration and eliminating compliance issues (39). Apart from insertion problems, there are few contraindications to its use, at least partly because systemic concentrations of the drug are much lower than those achieved with oral administration (40). Serum levonorgestrel levels are 20-fold lower in LNG-IUS users than levonorgestrel-containing combined oral contraceptive pill users, for example (41). There is an inverse correlation between serum levonorgestrel concentrations and BMI (42), suggesting even lower systemic levels in our population. A further advantage of the LNG-IUS for this indication is that it would be expected to eradicate or treat latent endometrial cancer precursors, atypical hyperplasia, and occult obesity-driven endometrial cancer, as previously demonstrated (24, 17–19). Regression of established endometrial abnormalities takes 6 to 12 months or longer and is more likely in the case of atypical hyperplasia (approximately 90% complete response rate) than early stage endometrial cancer (67% complete response rate; ref. 43). There are currently no validated biomarkers that predict LNG-IUS response to established disease (44), although some show promise (45, 46), mandating careful assessment of any new bleeding that develops following device-induced amenorrhea (47) in an endometrial cancer prevention trial.

Overall, we found the LNG-IUS to be safe and well-tolerated, with no unacceptable side effects in our study population. This is particularly important if the LNG-IUS is being used for endometrial protection rather than an established clinical indication, and should be a focus of future work. There is no evidence that the LNG-IUS increases the risk of cardiometabolic disorders in obese women (48), but a recent systematic review found LNG-IUS users have a modestly increased breast cancer risk [OR = 1.16 (95% CI, 1.06–1.28), I2 = 78%, P < 0.01], although most of the included studies failed to adjust for BMI (49). Clearly this is an important area for further research. The LNG-IUS may prevent the endometrial consequences of excess adiposity, but it does not address the cardiovascular or metabolic sequelae of obesity and competing risks for death.

In summary, identifying women at greatest risk of endometrial cancer and developing evidence-based prevention strategies are important given the escalating rates of obesity and associated emerging epidemic of endometrial cancer across the world (50). The LNG-IUS may be an effective deterrent against obesity-driven endometrial cancer and offer endometrial protection for those at highest risk. It reduces risk in the general population by 50% during and for at least 5 years after discontinuation of its use (16), and as such is likely to be a cost-effective prevention strategy for women with class-III obesity (51). Here, we demonstrate that a LNG-IUS is acceptable to some women with class-III obesity and that a clinical efficacy trial would be feasible. The specifics of trial design require careful consideration because a large cohort with sufficient follow-up will be challenging and expensive to achieve. Minimizing trial size, duration of follow-up, and cost is an important goal for women, researchers, and funders of such a trial. Although endometrial cancer risk is high in women with class-III obesity, absolute risk is modulated by reproductive, metabolic, and genetic factors (13, 52), as well as competing risks for death. More sophisticated risk prediction models, calibrated for clinical use, must now be developed to establish the optimal prevention trial target population, maximize the benefits of participation, and reduce unnecessary harms (53).

R.J. Edmondson reports personal fees from Astra Zeneca, personal fees from Arquer Diagnostics, and grants from Tesaro Inc. outside the submitted work. H.C. Kitchener reports grants from William Walter Will Trust and grants from Central Manchester Foundation Trust during the conduct of the study. No disclosures were reported by the other authors.

A.E. Derbyshire: Data curation, formal analysis, investigation, writing-original draft, writing-review and editing. J.L. Allen: Data curation, writing-review and editing. M. Gittins: Formal analysis, supervision, writing-review and editing. B. Lakhiani: Data curation, writing-review and editing. J. Bolton: Data curation, formal analysis, writing-review and editing. J. Shaw: Formal analysis, writing-review and editing. P.W. Pemberton: Data curation, writing-review and editing. M. Needham: Data curation, supervision, writing-review and editing. M.L. MacKintosh: Supervision, writing-review and editing. R.J. Edmondson: Supervision, writing-review and editing. H.C. Kitchener: Conceptualization, resources, supervision, funding acquisition, writing-original draft, writing-review and editing. E.J. Crosbie: Conceptualization, resources, supervision, funding acquisition, methodology, writing-original draft, writing-review and editing.

We would like to thank the women who participated in this study. We are grateful to all the clinical staff involved in their care who helped facilitate recruitment, especially Samantha Johnson and Bryan Wilson. We would particularly like to thank Linsey Nelson, who contributed to study set-up, and Tina Pritchard, who supported patient recruitment, provided nursing care, and helped with administrative tasks. We are grateful to the independent members of the Trial Steering Committee, Professor Sudha Sundar, Professor Martin Rutter, Professor Steve Roberts, and Anne Lowry for providing study oversight.

A.E. Derbyshire was a Manchester University NHS Foundation Trust Clinical Research Fellow and E.J. Crosbie an NIHR Clinician Scientist (NIHR-CS-012–009), and their work was supported through the NIHR Manchester Biomedical Research Centre (IS-BRC-1215–20007) and the William Walter Will Trust. This article presents independent research funded by the NIHR. The views expressed are those of the authors and not necessarily those of the NHS, NIHR, or the Department of Health.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.