Intercepting Endometrial Cancer: Opportunities to Expand Access Using New Technology

Endometrial cancer has the highest incidence among gynecologic cancers in the United States. Endometrial cancer rates are increasing, especially among Black women (1), likely reflecting the high prevalence of obesity and related conditions, which are strong causal endometrial cancer risk factors (2). Black women have poorer survival and higher mortality rates than white women (8.7 per 100,000 for Blacks versus 4.4 per 100,000 for whites; ref. 3). Worse outcomes among Black women are ascribed to later stage at diagnosis, a higher incidence of more aggressive histologic subtypes, and suboptimal surgical treatment (3). We hypothesize that delays in clinical presentation contribute to racial disparities in endometrial cancer mortality and increase treatment-related morbidity, secondary to the requirement to administer adjuvant radio- and chemotherapy to manage late-stage disease. Accordingly, we propose a multipronged strategy that combines innovative patient education with novel early detection strategies to reduce health impacts of endometrial cancer and its precursors, especially among Black women.

Self-reported abnormal uterine bleeding (AUB) is among the most frequent gynecologic complaints for which women in the United States seek care. Menstrual disorders account for approximately 19% of all outpatient visits for gynecologic conditions (4). A central issue facing women and their providers is distinguishing variation in “normal” or “physiologic” uterine bleeding from “abnormal” uterine bleeding (AUB), which may represent an early sign of endometrial cancer. As women reach the menopausal transition, the distinction between physiologic, premenopausal AUB versus postmenopausal bleeding (PMB) is challenging.

Natural menopause is defined as permanent cessation of menstrual periods after 12 months of amenorrhea without other identifiable pathologic or physiologic causes. As such, menstrual status may be indeterminate during short or intermittent periods of amenorrhea. While PMB is abnormal, clinical determination of menopause may be uncertain due to contraception, medical conditions, and treatments affecting bleeding patterns (e.g., cancer, obesity, HIV). Endometrial cancer risks related to variation in menstrual patterns are ill-defined. Obesity, the most significant risk factor for endometrial cancer, is associated with elevated estrogen levels, which may stimulate endometrial growth that results in vaginal bleeding after cessation of menstrual cycling, which occurs on average at age 51 years.

Defining menopausal status is crucial because PMB is always considered abnormal and mandates evaluation to rule out endometrial cancer, and represents the presenting symptom among 90% of patients with endometrial cancer; however, fewer than 10% of women with PMB have endometrial cancer (5). Women having bleeding associated with the menopausal transition (i.e., physiologic variants of normal) may be misclassified as PMB or have benign anatomical or functional conditions (6).

The clinical heterogeneity in AUB and PMB profiles results in significant variability in the evaluation for endometrial cancer, which is magnified among black women. U.S. data show that Black women are less likely than non-Hispanic white women to receive guideline-concordant care, including recognition of PMB and appropriate diagnostic evaluation (70% vs. 79%), and that guideline nonconcordant care was associated with higher odds of advanced stage at endometrial cancer diagnosis (7). Therefore, improving recognition and management of PMB among Black women may diminish racial disparities in endometrial cancer outcomes.

The current approach to diagnostic evaluation of premenopausal AUB or PMB is presented in Fig. 1, which relies on patient-reported symptoms triggering an interaction with a primary healthcare provider who orders diagnostic tests to rule out endometrial cancer. Successful management requires clear communication of test results to patients, and referral to specialists who provide care. Gaps anywhere in the clinical pathway may lead to suboptimal outcomes. Prompt investigation of AUB can lead to detection and eradication of endometrial cancer or endometrial cancer precursors. Diagnostic testing to exclude endometrial cancer may include sonography [with or without saline infusion (sonohysterography)], endometrial biopsy or dilation and curettage, and/or hysteroscopy in the clinic or the operating room.

We propose that a multi-pronged approach for endometrial cancer prevention and detection that incorporates novel technologies and patient education and evaluation strategies has the potential to transcend current barriers to optimal management of AUB/PMB. Established and emerging technologies, such as smartphone apps that collect endometrial cancer risk factors, search engine presentation of information about the endometrial cancer evaluation (Table 1), novel molecular biomarkers of endometrial cancer risk and/or presence, and direct lines of electronic communication with subspecialists may overcome barriers to optimal management of AUB/PMB (Fig. 2). To succeed, these approaches must be affordable, acceptable to patients and providers, available, and implementable in diverse communities. Specifically, in communities that face social stressors or access issues, approaches that deliver information and testing outside of the specialized gynecology setting may be needed.

The utility of endometrial cancer screening in reducing endometrial cancer mortality and treatment-related morbidity remains unclear and benefits must be weighed against costs and the potential for “overdiagnosis” (detection of clinically insignificant disease). In an Israeli study, endometrial cancer diagnosed in asymptomatic women (i.e., without self-reported AUB) was not associated with better overall or recurrence-free survival than among symptomatic women (overall survival 79.7% vs. 76.8%, P = 0.37; ref. 8). A meta-analysis including international studies found similar results [RR = 1.04; 95% confidence interval (CI), 0.96–1.12, P = 0.32; ref. 9]. However, asymptomatic patients were more frequently diagnosed with stage I EC (RR = 2.33; 95% CI, 1.17–4.64) and less often required radiotherapy than symptomatic women (30.5% vs. 40.6%, P = 0.02) suggesting a decrease in treatment-related morbidity, costs, and inconvenience. This meta-analysis found that risks of high-grade endometrial cancer did not differ significantly between asymptomatic and symptomatic patients overall (RR = 0.92; 95% CI, 0.77–1.10; P = 0.36); however, when the Israeli study was omitted, the meta-analysis found a lower risk of high-grade endometrial cancer among asymptomatic women (RR = 0.83; 95% CI, 0.70–0.97). The analysis did not include U.S. studies, and did not address racial differences. Furthermore, asymptomatic patients were diagnosed with endometrial cancer following abnormal ultrasonographic findings that were performed as part of well-woman care. It is unclear whether these patients had unrecognized or unreported AUB/PMB that was not captured in the retrospective analysis. Lack of recognition of disease-associated symptoms and subsequent delays in patients seeking care have been identified in breast, cervical, and colorectal cancers (10).

Risk stratification has the potential to improve the efficiency of early detection testing (i.e., increased positive predictive value) by targeting groups at elevated risk, such as bariatric surgery candidates (11) or patients with Lynch syndrome (12). A case–control analysis nested within an ovarian cancer screening trial using transvaginal ultrasound examination detected occult endometrial cancer or endometrial hyperplasia among 0.3% of patients (13). On the basis of an optimal endometrial thickness cutoff for endometrial cancer detection of 5.15 mm, as defined in the analysis, ultrasound testing achieved a sensitivity of 80.5% (95% CI, 72.7%–86.8%) and specificity of 85.7% (85.4%–86.2%; refs. 13, 14). Epidemiologic models for prediction of endometrial cancer risk were developed in the U.S. and Europe (15, 16). However, the generalizability to non-white patient populations is unclear. It is also recognized that highly aggressive subtypes of endometrial cancer, which contribute disproportionately to endometrial cancer mortality, especially among Black women, comprise the minority of endometrial cancers and tend to occur at later ages (17).

Risk factors for endometrial cancer include obesity measures (body mass index, body fat distribution, weight change), reproductive measures (prolonged anovulation/polycystic ovarian syndrome), and tamoxifen use (2). As more women at risk for endometrial cancer are studied, the diagnostic role of novel biomarkers for endometrial cancer and its precursor lesions may increase. Although BMI is strongly and proportionately related to endometrial cancer risk, understanding which obese women will develop endometrial cancer is a critical area of research.

Data suggest that lower genital tract sampling in combination with molecular testing holds promise for distinguishing which women with AUB have endometrial cancer. We and others have demonstrated that tampons and Papanicolaou tests may represent an acceptable approach for collecting DNA for molecular testing among women with AUB (18–20). Self-collection of biospecimens such as vaginal fluid through at-home tampon collection represents a promising approach to make the evaluation of AUB and PMB more accessible to women (21). Molecular testing for endometrial cancer could be combined with high-risk human papillomavirus testing to screen for cervical neoplasia to achieve early detection and provide strong and long-lasting negative predictive value (i.e., reassurance) that these cancers are not present. Results of negative tests could be leveraged as an inflection point at which women could be influenced to pursue health promotion activities (e.g., weight loss). Large-scale studies, especially among underrepresented patient populations, to assess the feasibility of tampon-based collection of DNA as an early detection test for endometrial cancer and its precursors among women with AUB and asymptomatic women with endometrial cancer risk factors appear warranted.

Synergistic development of digital tools and artificial intelligence may offer a powerful approach for improving risk assessment and early diagnosis across healthcare settings. Smartphone apps and search engines (Table 1, example) that collect data from diverse sources including patients, academic journals, healthcare electronic medical records, population databases, and wearables are emerging as potentially powerful tools to enhance patients' engagement in their own healthcare. The ability to collect large amounts of granular serial patient data with annotated outcomes offers the possibility of applying artificial intelligence algorithms to improve accuracy of prediction of individual endometrial cancer risk and optimal triage of women for invasive diagnostic testing. Application of artificial intelligence approaches to “big data” may enable development of improved risk prediction models for endometrial cancer, and such models may improve over time through “continuous learning.” Furthermore, patient engagement may incentivize health promoting interventions, such as “gamification,” which “…harnesses a desire for competition, incorporating gaming elements such as badges, leaderboards, competitions, rewards, and avatars to engage and motivate people” (22). While a promising tool, the impact of gamification on clinical outcomes such as health promotion activities to prevent cancer have not been assessed and this represents an important area of future research (22).

Smartphone apps that incorporate over 30 variables including menstrual cycle history, pregnancy test results, age, height, weight, and lifestyle statistics have been developed to collect data that is analyzed using artificial intelligence algorithms to predict timing of menses and ovulation. These applications offer a potential patient-centric approach to family planning that can benefit individual users and also provide a “pay forward” dividend by adding to accumulating data that enhance the algorithm and increase its value for later users. However, the prediction of ovulation and menstruation is only as good as the data provided by patients. For menstrual prediction, such data tracking may allow for more accurately predicting the timing of when future periods will occur so that a woman can make informed choices about symptom management and improve the discrimination of AUB from physiologic albeit uncommon bleeding patterns.

Because these tools provide information and testing outside of the specialized gynecology setting, they have potential to address existing cancer health disparities related to health care access. However, access does not ensure patient uptake. The current lack of patient engagement in promising prevention tests was clear in a randomized clinical trial of mailed human papillomavirus test kits compared with usual care reminders. Screening uptake improved modestly: 26.3% in the mailed collection kit compared with 17.4% in the usual care group (23). The CanTest framework addresses these limitations by proposing a 5-phase process proceeding from selection of the test and measures of single test performance to implementation and effects at healthcare and population (24). The framework highlights the importance not only of diagnostic accuracy but patient-centric outcomes (e.g., acceptability) and cultural factors that may influence patient decisions about testing (25).

Important aspects of endometrial cancer–relevant data monitoring using app-based technology include providing a cue to the user requesting data and recommendations to change a routine/habit in order to obtain a reward, such as visual feedback of risk reduction that can be appreciated in real-time on the app. This habit loop was described and popularized in Charles Duhigg's The Power of Habit (26). The miPlan contraceptive app for African American and Latina women was developed and tested with young women who were potential application users, and it subsequently achieved high levels of user acceptability (25). Further large-scale testing in underrepresented communities should be prioritized for other risk prediction and education apps. We envision approaches in which women with AUB would interface with a risk-prediction dashboard that prompts the entry of additional risk factor data as needed to optimize endometrial cancer risk prediction. These tools may offer suggested risk-reducing strategies and quantify the risk-reduction expected from these strategies. Analysis of personal inputs to predict individual endometrial cancer risk prediction has the potential to drive increased patient engagement and provide patients with increased understanding of the expected risk and benefits of interventions, including medical management (progestogens), bariatric surgery, exercise, or definitive endometrial cancer risk–reducing surgery (hysterectomy). This approach rests on the assumption that patients will adopt the technology, and that the information provided to women will motivate them to action. The technology itself, once developed for market, should be affordable, and user engagement during the development process can increase acceptability and ensure the app is easy to navigate and useful. Engagement of women from diverse socioeconomic and cultural backgrounds can also increase our understanding of how individuals perceive cancer risk and make health behavior decisions. The use of behavioral theory is one strategy for identifying behavioral influences, including social factors, and linking those to specific components of health interventions (27, 28).

Patient empowerment holds potential to enable improved individualized medical care by directing healthcare resources to women at the greatest risk. Enabling healthcare providers to function in a digital healthcare environment may require training and acquisition of new skills for counseling patients and acting on results. Sorting out medicolegal issues associated with patients assuming greater responsibility for healthcare choices and reimbursement models for these activities will be needed.

Futuristic approaches using gamification, smartphone apps, artificial intelligence, and health promotion outside of the physical clinic holds promise in preventing endometrial cancer and reducing morbidity and mortality related to the disease, but it also raises questions. Will underrepresented populations utilize new technologies, and if not, will disparities in care widen? If the initial risk prediction screen indicates low risk, should a self-collected triage test still be offered? Similarly, if the triage test is negative, when and/or how often should it be repeated and what roles should providers/patients have in overriding results? Would and should healthcare payers incentivize the patient centric approach over the traditional approach of patients evaluated in a physical clinic? How should quality of life and patients with low risk of endometrial cancer that develop AUB be factored into the decision making?

Despite these remaining questions, patient-centric approaches have untapped potential to not only improve EC outcomes for patients, but to reduce disparities for women who face barriers to care by delivering information and testing outside of the physical clinic setting engaging underrepresented patient populations.

Endometrial cancer, the most common of gynecologic malignancies among U.S. women, is often diagnosed at an early curable stage. Nonetheless, the incidence and mortality from endometrial cancer is rising and minority women are particularly at risk. Improved methods for endometrial cancer risk stratification and distinguishing AUB/PMB from physiologic variation are needed. Proposed new approaches must offer favorable ratios of costs, benefits, and risks for patients and society. A future emphasis on patient-centric care, collection of “big data,” artificial intelligence, and self-collection methods could yield new models of care. If healthcare visits are reduced, these models of care would enable providers to spend more time with fewer patients who most need extended counseling, sophisticated testing and treatment. Given disparities resulting from the current approaches to endometrial cancer diagnosis and management, involvement of underrepresented populations in the development, testing, and use of these new strategies will be paramount.

No potential conflicts of interest were disclosed.

Conception and design: C.C. DeStephano, J.N. Bakkum-Gamez

Development of methodology: C.C. DeStephano, J.N. Bakkum-Gamez

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.C. DeStephano

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.C. DeStephano, J.N. Bakkum-Gamez

Writing, review, and/or revision of the manuscript: C.C. DeStephano, J.N. Bakkum-Gamez, A.M. Kaunitz, J.L. Ridgeway, M.E. Sherman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.C. DeStephano, M.E. Sherman

Study supervision: C.C. DeStephano

Other (critical review of literature): M.E. Sherman

This study was supported in part by the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery.

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.