A rigorous assessment of who, when, and how to screen may lead to regular testing in a broader population.
Major organizations currently recommend screening for endometrial cancer (EC) only for women with an inherited mismatch repair defect, commonly known as HNPCC (hereditary non-polyposis colorectal cancer) and Lynch syndrome. For women with HNPCC-which typically is identified after diagnosis of EC-endometrial biopsy is recommended annually or less frequently, starting at age 35. The precise definition and diagnostic criteria for Lynch and HNPCC syndromes vary. Genetic proof of these inherited disorders is not always available. In these circumstances, clinical criteria such as Amsterdam Criteria or a high index of suspicion should be sufficient to consider EC screening at some arbitrary interval and starting age.1
The homogeneity of opinions about EC screening, however, may be ending. The Centers for Disease Control and Prevention (CDC) publicly announced that it is looking into EC screening, an initiative spurred by an unrelated, highly publicized case of morcellation.2 An incidentally discovered sarcoma is being used by at least one legislator to justify a request to the CDC to examine mandating endometrial biopsy before all hysterectomies. Many surgeons already perform biopsies before all hysterectomies and ovarian surgeries in at-risk women. In essence, these practitioners are screening for EC in an at-risk population: women about to undergo pelvic surgery.
Routine screening for EC was studied during the initial experience with selective estrogen receptor modulators (SERMS).3 The attention was due to an increase in EC in patients with breast cancer who use SERMS to prevent recurrence. The conclusion of most opinions at that time was that screening was not warranted due to a low incidence and positive predictive value of the then-favored screening modality, ultrasonography. Only ultrasonography was considered because of a perceived increased risk of office endometrial biopsy. Based on these assumptions, screening was not adopted by most practitioners because of the poor specificity of ultrasonography and the low prevalence of EC. In retrospect, that decision still seems appropriate, given the information available at that time.4
Data on complications of endometrial biopsy (perforation, bleeding, pain, infection) are derived from reports of a population with a strong indication for sampling (i.e., symptoms and presumed cancer until proven otherwise). In triage of this patient group, sampling attempts persist until successful, thus adding additional risks to the overall goal of sampling the endometrium.5,6
On the other hand, sampling asymptomatic patients, or screening, would exclude women at increased risk of complications from the procedure. In other words, patients at increased risk of complications from a screening biopsy would not be put through additional invasive procedures or attempts. For instance, a woman considered for screening who has cervical stenosis or severe comorbidities would not be subjected to dilatation and curettage in the operating room. Instead such as a patient could be triaged to serial pelvic ultrasonography or other safer office maneuvers, such as cervical “ripening,” before biopsy.7 Thus, risk associated with biopsy for EC screening could be adjusted and lower than previously reported for endometrial sampling for other indications. In general, in-office sampling for screening should have little to no risk, as there are alternatives for difficult patients and procedures.8,9
Disease prevalence as a factor
Once risk of screening with biopsy is potentially acceptable, then disease prevalence becomes the next important area to consider. Currently some populations are at higher risk than the originally proposed group of patients using SERMs for breast cancer prevention. In those women, prospective cohort data suggest an endometrial cancer incidence of approximately 1/1000 to 1/2000.3 However, more recently reported data on community-based and risk-adjusted population incidence reveal much higher rates. For instance, in communities with high non-communicable disease (NCD) rates, EC incidence has been reported to be as high as 6/1000.10,11 In other populations, high-risk cohorts can be identified on the basis of risk factors such as age and weight. Using these risk modifiers, a subpopulation at even higher risk can be identified with rates theoretically approaching 1/100 for 50-year-olds with body mass indices (BMI) of 40 or more.10,11
This prevalence approximates and compares favorably to that in patients with symptoms. In those women, who have the traditional indication for endometrial biopsy, the rate of diagnosed cancers is higher, but that should not be the sole criterion for deciding when to biopsy.1,10 Using the current standard, EC outcome is not optimal because overall survival can be as low as 80% in communities suffering from disparities. In theory, screening could eliminate all EC-related deaths in these and other populations.
To justify screening, several other conditions must be met, including existence of a premalignant condition, availability of a realistic intervention, and likelihood that patients will adhere to treatment. EC screening is conceivable based on all of these parameters, including existence of a premalignant condition that can be treated to prevent progression to cancer.
Hyperplasia in its various forms does respond to medical treatments and likely even lifestyle changes, including weight loss. Medical therapies such as a progesterone-delivering intrauterine device can be both effective for premalignant conditions and less morbid than surgical treatment for EC. The increased safety of medical treatment for a premalignant condition versus surgery for frank EC is a given. That benefit may be magnified because premalignant conditions occur years before development of EC and may have an earlier peak occurrence in women who are younger, and presumably healthier, than patients destined to be diagnosed with EC. Certainly for any patient contemplating surgery, whether for EC or a precursor, doing so at the youngest age possible may be an advantage.
Most promising of all is evidence to suggest that lifestyle changes such as exercise and weight loss and interventions such as metformin that are aimed at underlying causes may have additional health benefits beyond interrupting progression to EC.12,13 Even if screening only leads to EC diagnosis at an earlier stage, it is theoretically possible that less invasive surgery, such as simple vaginal hysterectomy, can replace more extensive procedures involving lymph node removal and staging.
Screening in EC versus other cancers
Finally, compared with other cancers for which screening is done, EC is an attractive target. Its incidence and other characteristics are similar to those of breast, prostate, colon, and lung cancer. To be clear, incidence is lower for endometrial than for breast and prostate cancer, but the latter diseases have an indolent counterpart, which EC does not. Given the potential for overtreatment of breast and prostate cancer, controversy exists regarding the risk/benefit equation for their respective screening programs. There is no way to screen for “indolent” EC, so the watchful waiting, active surveillance, and other treatment algorithms evolving for breast and prostate cancer do not exist for EC. Screening and triage for EC is relatively simple compared with the techniques used for colon and lung cancer. Incidence of the latter two diseases is slightly higher than for EC, but the complexity, expense, and risk of work-up for EC makes screening for it potentially comparable.
Other screening programs have been tailored to specific segments of the population, such as smokers and patients with chronic hepatitis, who are at risk of lung and liver cancer, respectively. EC screening could be further refined by known risk factors, such as age and BMI, until a consensus is agreed upon that the adjusted risk warrants screening. This risk adjustment could be extended nearly universally in communities that share high-risk factors. For instance, in a community with high reported NCD rates, EC was 2.5, 3.5, 8 and 10.7 times more likely to be diagnosed than breast, colon, cervical and ovarian cancers, respectively.10 All of those other benchmark cancers are either being screened for or are the subjects of intense investigations for improved screening. EC is not currently experiencing the same focus, but the relative incidence of it suggests that it should be. Further, the high mortality associated with some cancers for which effective screening is being sought-such as ovarian and cervical cancer-is the rationale for that screening. In high-risk communities, however, EC, however, may actually be a more common cause of death than those diseases.10
In-office versus hospital-based screening
Screening can take many forms and may need to be community- or hospital-based, depending on the anticipated impact on compliance. Hospital-based screening, however, might be effective, given the association between EC and comorbid conditions and NCDs that lead to frequent hospital visits. For instance, at our safety net hospital, approximately 96% of women with EC had been seen an average of 1.7 years before their diagnosis, suggesting that earlier detection at screening may have been possible.14 Many of these patients had prior Pap smears, imaging, and biopsies for other diagnoses, which suggests that a hospital-based screening program would be possible. Women at risk for EC who have intermittent, unpredictable, and uncertain access to healthcare may benefit even more from hospital-based screening for the disease because it may be effective even at long intervals between screening. This is based on the well-established relationship of EC with hyperplasia and the long latency period between EC and its premalignant precursors.
Relative incidence of EC is higher than that for many cancers for which we already screen, including cervical cancer and other cancers with active screening research programs, such as ovarian cancer. EC does not have the same controversies that exist around prostate and breast cancer screening. (Screening may do more harm than good in the subsets of patients with those cancers whose disease has a long indolent natural history.) Overall, as risks factors for EC become more prevalent, and the affected population grows older with increasing related comorbidities, the need becomes more urgent for a thorough, rigorous, and scientific approach to determining who, when and how to screen. It will also be important to consider relative costs and benefits associated with these strategies. Changes in the overall health of our society have now made it imperative for all stakeholders to reconsider screening for EC.
The authors report no potential conflicts of interest with regard to this article.