Detecting endometrial cancer

Endometrial cancer (EC) is the most common gynaecological cancer in the UK. Ninety percent of women with EC present with postmenopausal bleeding (PMB), but less than 10% of women with PMB have a sinister underlying cause. National Institute for Health and Care Excellence guidance advises that symptomatic postmenopausal women undergo urgent investigation; however, guidance is unclear for premenopausal women. Current investigations for PMB, including transvaginal ultrasound scan, endometrial biopsy and/or outpatient hysteroscopy, have advantages and disadvantages. Novel detection tools are in development, which combine minimally invasive sampling with genomic, proteomic and single cell technologies.

Endometrial cancer (EC) is the most common gynaecological cancer in the UK. Ninety percent of women with EC present with postmenopausal bleeding (PMB), but less than 10% of women with PMB have a sinister underlying cause. National Institute for Health and Care Excellence guidance advises that symptomatic postmenopausal women undergo urgent investigation; however, guidance is unclear for premenopausal women. Current investigations for PMB, including transvaginal ultrasound scan, endometrial biopsy and/or outpatient hysteroscopy, have advantages and disadvantages. Novel detection tools are in development, which combine minimally invasive sampling with genomic, proteomic and single cell technologies.

Learning objectives
To understand who is at risk of EC and who should be referred for urgent investigations.
To understand the evidence underpinning the current diagnostic pathway for EC. To highlight unique and promising perspectives for EC detection and their potential to transform clinical care.

Introduction
Endometrial cancer (EC) is the fourth most common cancer in women in the UK and the most common gynaecological malignancy. In the UK, there are over 9000 new cases each year and the incidence has risen by 57% since the early 1990s. This is attributed to the ageing population, a growing prevalence of obesity and declining rates of hysterectomy for benign disease. Survival rates are dependent on stage at diagnosis, ranging from 95% for stage I cancers to 15% for stage IV, therefore early diagnosis is essential for good outcomes. 1 Early diagnosis may also enable conservative treatment for women of reproductive age or for those for whom surgery carries considerable risks, such as the elderly or morbidly obese. proliferation. Unimpeded by apoptosis, these mutations expand clonally and acquire additional mutations that drive carcinogenesis. Estrogen is produced by adipose tissue through the aromatisation of adrenal androgens. After menopause, a lack of endogenous progesterone leaves the endometrium unprotected from the effects of estrogen. 3 Thus, obesity confers a higher risk of endometrial cancer, 4 with every additional 5 kg/m 2 of body mass index (BMI) associated with a 50% (95% confidence interval [CI] 40-60%) increased risk. 5 Around 85% of EC is diagnosed in women older than 55 years of age. 6 In premenopausal women, anovulatory cycles in polycystic ovary syndrome (PCOS) and obesity are a major risk factor. 7 Lynch syndrome is an autosomal dominant inherited condition of defective DNA mismatch repair, which affects the MSH2, MLH1, MSH6 and PMS2 genes. 8 Women with Lynch syndrome have a 25-60% lifetime risk of EC and present at younger ages than women with sporadic EC. 9,10 Red flag symptoms for endometrial cancer Postmenopausal bleeding (PMB), defined as vaginal bleeding occurring more than 12 months after the cessation of menstruation at menopause, is the most common red flag symptom for EC. 11 Over 90% of women with EC present with PMB, but over 90% of women with PMB have a benign underlying cause for their symptoms (Box 1). 12,13 PMB can be confused with haematuria, triggering urgent urological investigations. 14 Vaginal discharge or pyometria is less commonly described. Premenopause, women complain of intermenstrual or persistent heavy menstrual bleeding. Late-stage disease presents with abdominal distension, pelvic pressure symptoms or pain. 15 Cytology-based cervical screening detects atypical glandular cells in up to half of women subsequently diagnosed with EC, 16 who might also be identified incidentally on ultrasound, computed tomography (CT) or magnetic resonance imaging (MRI) performed for other reasons.

Referral from primary care
The 2015 National Institute for Health and Care Excellence (NICE) suspected cancer guidance (Box 2) recommends that women with PMB or heavy, irregular bleeding who are over 45 years of age should have a full history, pelvic and speculum examinations and urinalysis performed in primary care. 17 Examination is important to exclude a pelvic mass or pathology of the lower genital tract. The probability of EC in women with PMB rises from less than 1% in women under the age of 50 to 24% in women over 80 years old. 18 Women on hormone replacement therapy (HRT) require special consideration. Those with persistent unscheduled bleeding for more than 6 months after starting HRT should be referred for investigation. 17 Those with new onset PMB should only be referred if bleeding continues 6 weeks after stopping HRT.
EC should be considered in premenopausal women with abnormal bleeding, particularly those with obesity, PCOS, a strong family history or other risk factors. A systematic review of premenopausal women with abnormal uterine bleeding found the risk of ECor its precursor lesion, atypical hyperplasiawas just 1.31%, with intermenstrual bleeding being a better predictor than heavy menstrual bleeding. 19 The risk for premenopausal women increases with BMI and is reportedly five times higher at a BMI ≥30 kg/ m 2 . 7 A retrospective review of two-week wait referrals from primary care in England between 2006 and 2010 found that women aged 35-44 years eventually diagnosed with EC were significantly less likely to be referred urgently than women aged 65-74 years (odds ratio 0.09, 95% CI 0.07-0.12, p < 0.001). 6 The lack of clear guidance from NICE on which premenopausal women require urgent review risks diagnostic delay for those at highest risk.

Diagnostic pathway for endometrial cancer
The most effective diagnostic strategy for the investigation of PMB remains controversial; there is no evidence-based upto-date guidance from NICE, the Scottish Intercollegiate Guidelines Network (SIGN), or the Royal College of Obstetricians and Gynaecologists (RCOG). Selective transvaginal sonography (TVS) for 'high-risk' women based on age, BMI and other risk factors is the most cost-effective strategy, but risks missing cases. 20 The British Gynaecological Cancer Society (BGCS) recommend TVS as first-line investigation for PMB, followed by endometrial biopsy, with or without hysteroscopy, if the endometrium is thickened ( Figure 1). 21

One-stop postmenopausal bleeding clinics
One-stop clinics, in which patients are scanned, reviewed by a clinician and offered endometrial biopsy and/or hysteroscopy in a single visit, reduce delays, improve patient experience and are cost-effective. [20][21][22] Transvaginal sonography for investigation of postmenopausal bleeding TVS provides a non-invasive assessment of double-layered endometrial thickness, which can be used to triage women for further investigations. A recent systematic review of women with PMB found that women with EC have a mean endometrial thickness of 16.4 mm (95% CI 14.8-18.1 mm), compared with 4.1 mm (95% CI 3.5-4.7 mm) for those investigated but found not to have EC. 23 Interestingly, mean endometrial thickness has increased significantly over time.
The mean endometrial thickness of women with PMB found not to have EC was 3.5 mm in studies published before 2000, but 5.7 mm in later studies; this can possibly be attributed to improved resolution of imaging, increased HRT use, or higher obesity rates. This influences the clinical utility of TVS for EC detection, since endometrial thickness cut-offs derived from historical studies might not be transferable to modern day populations.
The diagnostic accuracy of TVS for EC detection depends on the endometrial thickness cut-off used. BGCS guidelines currently recommend an endometrial thickness cut-off of ≥4 mm. This is based on a systematic review published in 2010, 23 (Table 2). Based on this systematic review, increasing the cut-off from ≥4 mm to ≥5 mm in future UK guidance would offer comparable sensitivity and negative predictive value (NPV) to a cut-off of ≥4 mm, but improved specificity, reducing the need for invasive diagnostic procedures by up to 17%. 23 The character of the endometrium can further define risk of malignancy by TVS. A heterogenous endometrium with cystic change is highly suspicious of underlying disease (Figure 2A). 25 A caveat is the benign subepithelial stromal hypertrophy associated with tamoxifen treatment, which causes a grossly abnormal endometrial signal and, consequently, a high false-positive rate. 26 This is particularly challenging because tamoxifen is associated with a three-fold increased risk of EC and therefore triggers a high level of clinical suspicion. 27 Grayscale and colour Doppler sonographic features can be useful to distinguish malignant from nonmalignant endometrial pathology ( Figure 2B). 25 Endometrial thickness is of limited utility as a triage test in premenopausal women, in whom endometrial thickness fluctuates naturally during the menstrual cycle. It is also technically challenging to accurately measure endometrial thickness where the uterine cavity is distorted by fibroids and when body habitus compromises test validity (for example, in women with a high BMI).

Incidental findings in asymptomatic women
An incidental finding of a thickened endometrium in an asymptomatic postmenopausal woman is a thorny issue because there is no consensus as to what endometrial thickness cut-off requires further investigation. A prospective study of 81 asymptomatic women referred for endometrial sampling following an incidental finding of a thickened endometrium on TVS found EC or atypical endometrial hyperplasia (AEH) in just four women (4.9%), all of whom had an endometrial thickness ≥10 mm. 28 A theoretical cohort study combining published and unpublished data from 10 000 postmenopausal women found an endometrial thickness cut-off of ≥11 mm differentiated between women whose risk of cancer was 6.7% and those whose risk was just 0.002%. 29 In contrast, a systematic review of 32 studies and 11 100 women failed to identify a discriminatory endometrial thickness in asymptomatic postmenopausal women of sufficient clinical utility to rationalise further investigations. 30 The UK Collaborative Trial for Ovarian Cancer Screening (UKCTOCS) reported asymptomatic endometrial pathology in 125 of 36 861 postmenopausal women within 12 months of TVS and found a cut-off endometrial thickness of ≥5 mm had a sensitivity of 77.1% and specificity of 85.8% for the detection of EC or AEH. 31 A caveat was the lack of a standardised protocol for endometrial investigations at the ≥5 mm cut-off; EC diagnoses were made up to 12 months later, possibly after women developed symptoms, when an up-to-date TVS may have returned a different endometrial thickness. With no clear guidance, an incidental finding of a thickened endometrium is investigated at the discretion of individual clinicians, taking patient preference and risk factors into account.

Endometrial sampling
An endometrial biopsy is indicated if a woman presenting with PMB has a thickened endometrium on TVS. While easy and quick to perform in an outpatient setting, endometrial biopsy is an invasive procedure with potential for harm, including failure (11%), inadequate sample (31%), pain,  bleeding, infection andvery rarelyperforation. 32 Numerous aspirating, brush and cannulation devices are available that show similar diagnostic accuracy to traditional dilatation and curettage, but enable outpatient sampling. 33,34 The Pipelle aspirator (Pipelle de Cornier Mk II, Eurosurgical Ltd., Guildford, UK) is the most commonly used sampling device, with a sensitivity of 90-100% for EC detection when an adequate sample is obtained. 32,34-36 It was previously thought that the 30% of women with an inadequate sample could be safely reassured; 35 however, given that 4.5% of women were diagnosed with EC after an initial inadequate sample in one study, 37 this might not be appropriate and emphasises the importance of restricting endometrial sampling to women with a thickened endometrium in the first place. While most women tolerate endometrial sampling well, pain is a significant barrier for some. 34 Failed endometrial sampling is usually associated with pain or cervical stenosis, which are more common in nulliparous women. 38 As a blind procedure, endometrial sampling has the potential to miss small, localised cancers. 39 Women with benign or inconclusive histology, but persistent symptoms or suspicious ultrasound findings, should be offered hysteroscopy.

Hysteroscopy
Hysteroscopy is direct visualisation of the uterine cavity via a fine bore scope to identify pathology, take directed biopsies and carry out therapeutic procedures, such as polypectomy. Hysteroscopy is indicated for women with a thickened, irregular endometrium, or other concerning features on ultrasound; those with recurrent or prolonged bleeding; or where random endometrial sampling has been nondiagnostic. 21 A randomised controlled trial comparing hysteroscopic resection of endometrial polyps with expectant management of PMB found endometrial (pre)malignancy in 6% of women that had been missed by random endometrial sampling. 40 This highlights the importance of hysteroscopic assessment in cases where focal pathology is suspected on TVS. A systematic review of 65 studies, including 26 346 women, reported that hysteroscopy has a sensitivity of 86.4% and specificity of 99.2% for EC detection. 41 This is the most recent systematic review at the time of writing, although modern technology may offer improved accuracy.
Hysteroscopy can be carried out as an outpatient procedure and, when offered as part of a one-stop clinic, this is the most cost-effective and efficient way of investigating unexplained PMB. 42 The risks of hysteroscopy include failure (4.2%),  pain (31%), bleeding, infection (0.25%) and perforation (less than 0.1%). 41,43,44 Although well tolerated by most women, a significant proportion report severe (13%) or moderate to severe (31%) pain. 44 Women are advised to take standard doses of nonsteroidal anti-inflammatory drugs (NSAIDS) 1 hour before their scheduled appointment; 45 however, this may be insufficient analgesia for many women. A Cochrane review of 32 trials and 3304 participants found no evidence for a clinically meaningful improvement in pain scores at outpatient hysteroscopy associated with opioid, antispasmodic, intracervical or paracervical local anaesthetic administration. 46 Vaginoscopic techniques using fine bore scopes improve procedural tolerance. 47,48 It is important that women are appropriately counselled, provide informed consent and are offered a choice of analgesia, including the option to undergo the procedure under general anaesthesia. Hysteroscopy under general anaesthesia has significantly lower postoperative pain scores, but carries greater risks and is much more expensive. 49 Magnetic resonance imaging MRI is usually reserved for the preoperative staging of EC. On rare occasions, it might be required for more detailed assessment of a thickened endometrium on TVS where hysteroscopy fails or is contraindicated.

Diagnostic models
Current approaches to the investigation of PMB do not take risk factors into account, yet certain groups of women have a much higher pre-test probability of EC than others. The integration of clinical parameters and ultrasound findings in a diagnostic model could support a more sophisticated riskbased assessment of symptomatic women, which is likely to be more cost-effective. 20 High-risk women could be fast tracked through urgent invasive investigations, while low risk women are safely reassured. To date, six diagnostic models have been developed specifically for women presenting with PMB. [50][51][52][53][54][55] Predictors used in these models are age, age of menopause, BMI, parity, recurrent PMB, hypertension, diabetes, HRT and warfarin use, endometrial thickness, detailed ultrasonographic findings and serum HE4 levels. These models are not currently used in clinical practice because none have been externally validated and their clinical efficacy has not yet been established. 56 In their systematic review, Alblas et al. 56 called for the validation of previously published models and their extension with new predictors and biomarkers to build a model for clinical use.

Screening for endometrial cancer
The aim of screening is to identify occult atypical hyperplasia or EC in asymptomatic women. Detecting cancer at its earliest possible stage is expected to improve cure rates, reduce the morbidity associated with aggressive treatment and offer uterus-sparing management options for younger women wishing to preserve their fertility. 57 There is currently no established EC screening programme in the UK, neither for average-risk nor high-risk populations. The ideal screening tool is a minimally invasive, inexpensive and easy-to-perform test that is effective at detecting pre-invasive and early invasive disease. 58 In average-risk postmenopausal women, TVS has the advantage of being tried and tested, but the endometrial thickness cut-off chosen is a trade-off between sensitivity and specificity. Ensuring cases are not missed might expose large numbers of women to unnecessary invasive diagnostic tests. 31 Cervical cytology has not been explicitly tested as a screening tool for endometrial cancer. However, according to a systematic review of 45 studies and 6599 endometrial cancer cases, 16 atypical glandular cells are found in cervical samples of 77% of women with type II, and 44% of those with type I endometrial cancers, respectively. Indeed, a disadvantage of the switch to primary human papillomavirus (HPV) cervical screening is that most samples do not now undergo cytological assessment, thus missing the opportunity to incidentally diagnose EC.
In Japan, endometrial cytology features in an established screening programme for high-risk women. Such high-risk women are defined as those attending routine cervical screening who are nulligravid or postmenopausal and report abnormal bleeding in the past 6 months. 58 Endometrial cytology requires uterine instrumentation and is therefore less acceptable as a screening tool; however, it is associated with much lower rates of inadequate/failed sampling than endometrial biopsy. 59 Screen-detected endometrial cancers are also identified at an earlier stage and boast improved survival outcomes compared with women diagnosed following acute symptomatic presentation. 58 Screening high-risk groups for endometrial cancer Current BGCS guidelines recommend that women with Lynch syndrome are offered annual TVS, hysteroscopy and endometrial biopsy after the age of 35 years; 21 however, the evidence supporting this strategy is limited. 8 Sceptics question the benefit of screening when most EC presents at an early stage and has an excellent overall 5-year survival rate. 60 Other highrisk groups include women with class III obesity referred for weight loss management, in whom a high prevalence of occult endometrial abnormalities has been described 61 and breast cancer survivors receiving tamoxifen treatment, although no screening strategy is currently recommended for either group. Risk-stratifying women from the general population based on obesity, insulin resistance, reproductive and genetic biomarkers might identify other high-risk groups that could benefit from screening. 62 Developing novel endometrial cancer detection tools Current investigations for suspected EC are unpleasant, invasive and expensive. No single test is sufficient to both 'rule in' and 'rule out' disease in women presenting with red flag symptoms, or to identify occult endometrial (pre)cancer in asymptomatic women with risk factors. Technological innovation and an improved understanding of cancer biology have paved the way for novel ways of detecting endometrial cancer early. The goal is to combine patient-friendly tools for biofluid collection with EC biomarker discovery to develop minimally invasive sampling methodologies for screening and diagnosis (Table 3, Figure 3). 63 Uterine samples, including endometrial brushings and uterine lavage fluid, are an excellent source of cancer-specific biomarkers, but their collection is invasive and poorly tolerated by some. 64,65 The anatomical continuity between the upper and lower genital tracts provides the opportunity Routinely available Low concentrations of cancer-specific biomarkers in early cancer for naturally shed tumour debris to pass through the cervix, enabling collection from the vagina using tampons, brushes and swabs. 66,67 These collection tools lend themselves to selfsampling, enabling women to collect their own biofluids at home for postal return to the laboratory. While tampons can be left in the vagina for several hours to collect a representative sample, they are uncomfortable for postmenopausal women to use. 68 Urine is the perfect biofluid for non-invasive sampling because it is easy to collect, with the potential for large volumes, repeat samples or collection at pre-specified times of the day. It depends on urinary excretion of systemic cancer biomarkers, or the reliable contamination of urinary flow with uterine-shed tumour debris. Initial studies have yielded promising results. 69 Blood biomarkers are also undergoing development. Cancer antigen 125 (CA125) is elevated in advanced stage, poor prognosis EC, but its relatively normal levels in early stage disease precludes its utility as an early detection tool. A panel of protein biomarkers may have more value than a single protein biomarker; indeed, studies combining CA125 with HE4 and other proteins have demonstrated encouraging proof of concept. 70 Circulating tumour cells (CTCs), circulating tumour DNA (ctDNA) and microRNA (miRNA) are genomic biomarkers that could facilitate diagnosis, treatment monitoring and the detection of relapse, but their low concentration in early stage cancer limits their applicability at current detection limits. 71 Conclusion Despite its reputation for good prognosis, the rapidly rising incidence and devastating outcomes from advanced disease mean that 40% more women are now dying from EC than they did at the turn of the 21st century. 1 Early diagnosis is key to improving outcomes and there is much interest in the development of minimally invasive detection tools for the rapid triage of symptomatic women. Used in combination with validated diagnostic models, these tools could enable fast-tracked invasive diagnostics for those at greatest risk, while avoiding the physical and psychological harms of investigating healthy women. Such a tool may also be useful for screening asymptomatic women who are deemed to be high risk by virtue of age, obesity or hereditary predisposition, where currently no standardised programme exists.

Disclosure of interests
There are no conflicts of interest.

Contribution to authorship
ERJ and EJC designed and wrote the article. HO'F and KN provided expert material and review. All authors provided critical comment, edited the manuscript, and approved its final version. EJC is this article's guarantor.