Cancer Progress and Priorities: Uterine Cancer

Background

Uterine corpus cancer is the most common invasive gynecologic cancer among United States (U.S.) women. Studies of endometrial cancers, which comprise approximately 90% of all uterine cancers, have identified numerous risk factors, many of which appear to reflect high levels of estrogens in the absence of sufficient progesterone. Recent advances have indicated that the disease is etiologically heterogeneous, consisting of at least two major subgroups. This heterogeneity extends to important racial differences in both incidence and survival, possibly partially attributable to genetic factors.

Descriptive Epidemiology

Uterine cancer incidence is highest in North America and Northern Europe, intermediate in Southern Europe and temperate South America, and lowest in Southern and Eastern Asia and most of Africa (Figure 1) (1). This likely reflects prevalence differences in risk factors, including obesity and reproductive patterns. In the U.S., uterine cancer is the fourth most frequently diagnosed cancer, with estimates of 63,230 diagnoses in 2018 (lifetime risk of 1 out of every 40 women) (2). The average annual age-adjusted incidence of uterine cancer from the Surveillance, Epidemiology and End Results Program (SEER) was 25.7 per 100,000 women between 2010–2014 (3). The disease is rare before the age of 45 years, but risk rises sharply among women of all races in their late 40s to middle 60s (Figure 2). Worldwide, uterine cancer ranked in 2012 as the sixth most common cancer, with 319,600 estimated cases (4).

Figure 1 –

Age-standardized incidence rates for corpus uteri cancer, GLOBOCAN, 2012 shows the age-standardized incidence rates for corpus uteri cancer using data from GLOBOCAN 2012.

Figure 2 –

Age-specific uterine cancer incidence rates by race among U.S. women, SEER-18, 2003–2014 shows age-specific uterine cancer incidence rates among non-Hispanic White, Hispanic White, Black, American Indian/Alaskan Native, and Asian/Pacific Islander U.S. women using data from the SEER Program.

Dramatic changes in the incidence of uterine cancers have occurred over time. A marked increase in U.S. incidence peaked around 1975, a trend later linked with the widespread use of menopausal estrogens in the late 1960s and early 1970s (Figure 3). After a subsequent period of steady or declining incidence rates in many countries, endometrial cancer is again on the rise, mirroring increases in obesity prevalence (4, 5).

Figure 3 –

Trends in uterine cancer incidence and mortality among U.S. women, SEER-9, 1973–2014 shows uterine cancer incidence among White and Black U.S. women using data from the SEER Program.

In the U.S., age-adjusted mortality is 4.6 per 100,00 women, while in Europe mortality ranges between 2–4 per 100,000 (3, 6) (Figure 4). Similar to recent incidence increases, endometrial cancer mortality rates are also on the rise (4, 7). Overall, five-year survival is approximately 82%, which represents a marked increase since the 1960’s when it was 60% (8, 9). The distribution of uterine cancer stage, a strong prognostic factor, has remained stable (8, 10–12). Five-year survival is 95.3% for localized, 67.5% for regional, and 16.9% for distant-stage diseases (9).

Figure 4 –

Age-standardized mortality rates for corpus uteri cancer, GLOBOCAN, 2012 shows the age-standardized mortality rates for corpus uteri cancer using data from GLOBOCAN 2012.

Disparities

Historically, endometrial cancer incidence was lower among black compared to white women; however, that gap has narrowed significantly over time (13–17). However, once hysterectomy rates are taken into account, incidence in blacks surpasses that of whites (18). Although the associations for established endometrial cancer risk factors among black and white women are similar (19), prevalence differences may partially explain the markedly higher incidence increases among blacks. Endometrial cancer mortality is twice as high among black compared to white women (8.1 vs. 4.2 per 100,000 women) and has been attributed to aggressive clinical characteristics, lower socioeconomic status, higher prevalence of comorbid conditions, poor patient-provider interactions, and inferior treatment (20). Although less frequently studied, Asian and Hispanic women have lower risks of endometrial cancer compared with white women; however, five-year survival is the same or better (17, 21).

Risk Factors (see Table 1 and Figure 5)

Table 1:

Summary of endometrial cancer risk factors, genetics, and biomarkers

Domain	Factor	Estimated Relative Riska	Heterogeneity of risk	Comments	Highest level of evidence	Refs
Metabolic factors	Obesity	2.0–5.0	Association stronger for type I than II cancers	Each 5 kg/m2 increase in body mass index (BMI) is associated with a 62% increased risk	Cohort study	(22, 83)
	Diabetes	2.0	No heterogeneity observed	Uncertain extent to which relations are confounded by obesity	Meta-analysis of cohort studies	(26, 83)
	Hypertension	1.1–1.3	Not examined	Association between hypertension and endometrial cancer was weaker, but still significant, among studies with adjustment for BMI	Meta-analysis of case-control and cohort studies	(27)
	Metabolic syndrome	1.4–2.0	No heterogeneity observed	Adjustment for overweight/obesity does not eliminate increased risks associated with metabolic syndrome factors	Meta-analysis of case-control and cohort studies	(29, 114)
	Polycystic ovary syndrome	2.8	Not examined	Uncertain extent to which relations are confounded by obesity	Meta-analysis of case-control studies	(28)
Reproductive factors	Nulliparity	3.0	Association restricted to type I cancers	Further reductions for multi-parous women	Meta-analysis of case-control and cohort studies	(31, 115)
	Infertility	1.8	No heterogeneity observed	Even after adjusting for nulliparity, infertile women had increased risk	Pooled analysis of case-control and cohort studies	(32)
	Early age at menarche	1.5–2.0	No heterogeneity observed	4% reduction in risk per 2 years delay in menarcheal age	Meta-analysis of cohort studies	(33, 86)
	Late age at natural menopause	1.5–2.2	No heterogeneity observed	Pronounced risks among nonusers of menopausal hormones	Cohort studies	(86, 115, 116)
	Breastfeeding	0.9	No heterogeneity observed	Greatest reductions for long-term breastfeeding	Pooled analysis of case-control and cohort studies	(36)
Contraceptives	Combination oral contraceptives	0.3–0.5	No heterogeneity observed	Risk reduction persists for > 30 years	Pooled analysis of case-control and cohort studies	(83, 86)
	Intrauterine device use	0.5–0.8	Association stronger for type I than II cancers	More studies needed on the effects of progestin-releasing devices	Pooled analysis of case-control and cohort studies	(37, 38)
Menopausal hormone therapy	Menopausal estrogens	10.0–20.0	Not examined	Highest risks for long-term and high dose users of unopposed estrogens	Cohort study	(39)
	Menopausal estrogen plus progestins	0.7	Association stronger for type I than II cancers	Risk reduction is greatest for obese women	Randomized trial	(39, 42, 43)
Tamoxifen use	High cumulative doses of tamoxifen	2.2	Non-endometrioid histology subtypes appear to be especially affected by tamoxifen	Endometrial cancer risks highest shortly after exposure	Randomized trial	(44, 117)
Lifestyle factors	Cigarette smoking	0.5	No heterogeneity observed	Effects of cigarette smoking are particularly strong among postmenopausal women and menopausal hormone users	Meta-analysis of case-control and cohort studies	(46, 83)
	Moderate-to-vigorous physical activity	0.8	No heterogeneity observed	Inverse relation with physical activity restricted to overweight or obese women	Meta-analysis of case-control and cohort studies	(48, 86, 118)
Family history	Family history	1.8	No heterogeneity observed	Association is independent of Lynch syndrome status	Meta-analysis of case-control and cohort studies	(73, 119)
High penetrance gene mutations	MLH1	18–54% lifetime risk	Not examined			(74–76)
	MSH2	21–49% lifetime risk	Not examined			(74–76)
	MSH6	16–61% lifetime risk	Not examined			(75, 76)
	PMS2	12% lifetime risk	Not examined			(77)
	EPCAM	12% lifetime risk	Not examined			(78)
Low and moderate penetrance genes		1.1–1.4	Some SNP associations differ according to histology			(120)
Serum biomarkers	Estradiol and other endogenous estrogens	2.0–6.2	Some support for stronger relations with type I than II cancers	Associations persist after adjustment for body mass and show slightly stronger relations for type I than II cancers		(91)
	Insulin	Significant mean difference between endometrial cancer cases and controls: 33.94	Not examined	This meta-analysis did not detect an association among studies restricted to postmenopausal women, possibly due to small numbers		(94)
	C-peptide	Significant mean difference between endometrial cancer cases and controls: 0.14	Not examined	A lack of information on fasting time since the last meal may have led to misclassification of C-peptide levels		(94)
	Androgen	Postmenopausal: 1.7 Premenopausal: 0.9	Similar associations observed when restricted to women with type I	Higher circulating levels of androgens are associated with endometrial cancer among postmenopausal women		(88–90, 92, 93)
	Inflammatory markers	SERPINE1: 2.4 VEGF-A: 2.6 Anti-inflammatory cytokines (IL13, IL21): 0.5–0.6 Pro-inflammatory cytokines (CCL3, IL1B, IL23): 0.5–0.6	No heterogeneity observed although the number of women with type II was small	Endometrial cancer risk was most pronounced among obese women with the highest inflammation score		(97)
	Adiponectin	0.5	Not examined	Inverse associations were strongest among postmenopausal women, nulliparous women, and non-hormone users		(98)
	Leptin	2.2	Not examined	Associations were strongest among non-hormone users, diabetic women, and in prospective studies		(98)

Figure 5 –

Summary of the magnitude of association for established endometrial cancer risk factors summarizes the magnitude of associations for established endometrial cancer risk factors. Risks are approximate and can vary depending on the extent of exposure.

Metabolic factors.

A strong risk factor for endometrial cancer is obesity, accounting for 40–50% of all U.S. cases (22, 23). Overall body size appears to be more important than body fat distribution (24). Women with obesity-associated diseases such as diabetes (25, 26), hypertension (27), and polycystic ovary syndrome (28) are also at elevated risk, although obesity may contribute to these relationships. Metabolic syndrome has also been associated with significant risk elevations, although to a lesser extent than obesity (29).

Reproductive factors.

Nulliparous women are at substantially higher risks than parous women (30, 31), with infertility additionally contributing to risk (32). Other established reproductive risk factors include young ages at menarche and/or late ages at menopause (30, 33), potentially reflecting increased numbers of lifetime ovulatory cycles (34). Breastfeeding has also recently emerged as a possible protective factor (35, 36).

Contraceptives.

The use of combination oral contraceptives has been linked with marked risk reductions which persist for more than 30 years after discontinuation. Intrauterine devices also appear to reduce endometrial cancer risk (37, 38).

Menopausal hormone therapy.

Menopausal hormones have been strongly linked with risk increases, particularly for extended usage of high-dose unopposed estrogens (39). Progestins cause regression of estrogen-induced endometrial hyperplasia, the presumed precursor of most endometrial cancers (40), leading to estrogens commonly being prescribed with a progestin (particularly among non-hysterectomized women). Sequential progestin use, i.e., <10 days per month, is associated with only slight risk reductions compared to unopposed estrogen use (41). However, continuous combination therapy reduces risk compared to non-hormone usage (39, 42). Associations of hormone usage are strongly modified by body mass index (BMI) (39, 43).

Tamoxifen use.

Clinical trials have demonstrated increased endometrial cancer risk among tamoxifen-treated breast cancer patients, with risks highest shortly after exposure, among those receiving high cumulative doses, and for histologies usually associated with a poor prognosis (44).

Lifestyle factors.

Cigarette smoking (45, 46) and moderate to active physical activity levels (47, 48) have been associated with reduced risks, relations that are independent of other risk factors, including obesity.

Other factors.

It remains less clear whether risk reductions associated with high levels of fruit and vegetable consumption and/or of micronutrients are independent risk factors (49–52). Higher dairy product intake (53), coffee consumption (54), and consumption of green, but not black, tea (55) may lead to risk reductions. High-fat diets (51, 52) and alcohol consumption (56) have not generally been associated with risk. Use of the anti-diabetic drug metformin (57) or aspirin (58) appear to slightly reduce risk.

Controversial risk factors.

Less accepted as potential risk factors are several occupational exposures (59, 60); talcum powder use (61–63); thyroid diseases, cholecystectomy and endometriosis (64–66); antidepressants, statins, and acetaminophen (67–69); endocrine disruptors (70); tubal ligation (71); and electromagnetic radiation (72).

Familial and genetic factors.

Elevated endometrial cancer risks have been noted among women with a first-degree family history of endometrial cancer (73). This could reflect familial obesity (genetic or environment) or inherited risk, such as Lynch syndrome, an autosomal dominant cancer predisposition syndrome attributed to germline mutations in one of several mismatch repair (MMR) genes. Specific mutations have been estimated to result in cumulative lifetime endometrial cancer risks ranging between 12–61% (74–78), with MSH6 showing the highest risks (79) (Table 1). However, the higher range estimates may reflect reliance on data from clinical cancer genetic cohorts that are biased to include patients with family histories of cancer. Although Lynch syndrome is associated with a high cumulative lifetime risk of endometrial cancer, the relative rarity of the condition translates to an attributable fraction of only 5%.

The genome-wide association study (GWAS) approach has identified 18 risk loci for endometrial cancer, which are modestly associated with risk (odds ratios 0.8–1.4) (80). Some risk loci are significant only for endometrioid cancers. Few rare variants have been identified through exome-wide association studies (81), but candidate gene studies (82) have identified a number of single nucleotide polymorphisms (SNPs) in genes that may possibly impact risk.

Etiologic Heterogeneity (see Tables 1 and 2)

Table 2:

Summary of associations of established risk factors with type and histology of endometrial cancer

Risk factor	Type I	Type II	Endometrioid	Serous	Carcinosarcoma	Clear cell
Obesity	+++	++	+++	++	++	++
Diabetes	+	+	+	+	NA	NA
Metabolic syndrome	+	+	+	+	+	+
Nulliparity	++	+	++	+	+	NA
Infertility	+	+
Early age at menarche	+	++	+	++	NA	NA
Late age at natural menopause	--	NA
Breastfeeding	-	NA
Combination oral contraceptives	--	--	---	NA	NA	--
Intrauterine device use	-	NA
Menopausal estrogen plus progestins	--	NA
High cumulative doses of tamoxifen				++	++
Cigarette smoking	--	--	--	--		NA
Moderate-to-vigorous physical activity	--	--
Family history	+	++

Red indicates that the factor is positively associated with risk of the particular subtype.

⁺⁺⁺indicates a strong association (RR/OR ≥5.0),

⁺⁺a moderate association (RR/OR, 2–5), and

⁺, a modest association (RR/OR <2).

Green indicates that the factor is negatively associated with risk of the particular subtype.

^---, a strong association (RR/OR ≤0.6);

^--, a moderate association (RR/OR, 0.6–0.8); and

^-, a modest association (RR/OR ≥0.8).

Blue (N/A) indicates the factor is not associated with risk of the particular subtype

Important heterogeneity has been noted between type I (predominantly endometrioid adenocarcinomas with a hormonally driven etiology) and type II (mainly non-endometrioid malignancies that occur frequently among older and non-white women) cancers. Several epidemiological studies have found that type II cancers are less strongly linked to classic risk factors, such as obesity, nulliparity and hormones (44, 83).

Stronger relationships of hormonal, reproductive, and anthropometric risk factors have been found for endometrioid endometrial cancers compared with serous, clear cell, mucinous, or mixed tumors (44, 83–86). Furthermore, the Cancer Genome Atlas (TCGA) study has identified four molecular subtypes of endometrial cancer: polymerase Ɛ (POLE) ultramutated, microsatellite instability hypermutated, copy-number low, and copy-number high clusters (87). A comprehensive evaluation of endometrial cancer risk factors according to TCGA subtype has not yet been conducted.

Biologic Underpinnings of Identified Risk Factors

Estrogens are strongly related to risk (Table 1) (88–90), with one study showing generalized uterotropic activity of both parent estrogens and metabolites (91). Circulating androgens, the main source of estrogens in postmenopausal women, have also been linked with increased risk (88–90, 92, 93). Consistent with an association between diabetes and endometrial cancer risk, insulin and c-peptide have been demonstrated to be elevated among women with endometrial cancer (94). Insulin-like growth factor 1 (IGF)-1 and the IGF binding proteins are less consistently linked with risk (95, 96). Risk has also been related to circulating levels of inflammatory biomarkers (97) and with several obesity-related hormones (98).

Risk Prediction Models

Two risk prediction models, one developed in U.S.-based cohorts (99) and the other in a European cohort (100), demonstrated moderate discriminatory ability for established endometrial cancer risk factors (respective discrimination assessed by the area under the curve of 0.68 and 0.77). In the latter model, the addition of pre-diagnostic serum biomarkers only modestly (1.7%) increased discrimination (101).

Future Trends

Projection models indicate that endometrial cancer incidence will continue to rise, mainly as a consequence of rising obesity prevalence (7, 102). Changes in the distribution of other endometrial cancer risk factors also contribute to the projected growth in incidence, including increases in diabetes and metabolic syndrome (103, 104), declines in use of combination hormone therapy (5), and decreases in childbearing and smoking (105, 106). Moreover, hysterectomy for benign conditions has declined in recent decades, particularly among whites, contributing to more at-risk women (18, 107). In the next decade, mortality rates are also projected to increase (108).

Prevention

Primary prevention efforts focused on weight loss or use of medications are attractive prevention strategies. For high-risk patients, bariatric surgery is associated with a 44% reduced risk of developing endometrial cancer (109). Among Lynch syndrome patients, there is some evidence that oral contraceptive use may reduce risk (110).

Screening

Endometrial cancer screening is not recommended for women in the general population (111). Studies evaluating the use of endometrial biopsy and/or transvaginal ultrasound have generally shown low detection specificity (112). Nonetheless, the American Cancer Society Cancer recommends annual screening for Lynch syndrome patients with endometrial biopsy beginning at age 35 years. Development of early detection blood-based biomarkers are being explored (113).

Future Directions

Although considered an indolent tumor, the rapid increase in both endometrial cancer incidence and mortality warrants additional etiologic and prevention research. While progress has been made in identifying risk factors for the most common endometrial cancer subtype, this has not translated into effective primary prevention strategies. Future efforts should be directed at reducing the prevalence of modifiable risk factors (e.g., obesity). Additional research is needed to identify risk factors for aggressive endometrial cancer subtypes, particularly among black women.

To favorably impact survival, research on screening modalities to identify endometrial cancer at early stages is needed. Currently, screening in the general population is not recommended, but efforts to identify high-risk women could be beneficial.

Abbreviations list:

U.S.

United States

SEER

Surveillance, Epidemiology and End Results Program

GWAS

genome-wide association study

TCGA

The Cancer Genome Atlas

POLE

polymerase Ɛ

SNPs

single nucleotide polymorphisms

IGF

insulin like growth factor