As of 2015, ovarian and endometrial cancers together accounted for over 24,000 deaths of American women yearly—approximately half as many deaths as breast cancer (Table 36-1). However, unlike breast cancer, there are no effective screening programs currently available and universally recommended to mitigate the morbidity and mortality associated with these cancers. In this chapter, we explore the possibilities and challenges for screening and early detection and discuss the role of sonographic techniques in providing early diagnosis of both ovarian and endometrial cancer. This will be followed by a discussion of each of these gynecologic cancers as related to their early detection with sonographic techniques alone and combined with other laboratory tests.

This chapter discusses and illustrates factors that influence the efficacy of screening and early detection schemes for ovarian and endometrial cancers that integrate transvaginal sonography (TVS) and related techniques such as color Doppler sonography (CDS), three-dimensional sonography (3DS), and contrast-enhanced TVS (CE-TVS) (Figure 36-1). The role of sonography in the evaluation of these two gynecologic cancers is considered together in this chapter because TVS can potentially detect both of these neoplasms in their earliest stages. TVS is probably best used in a multimodal scheme including certain lab tests such as CA-125 for the detection and evaluation of these gynecologic cancers. This chapter takes into account the recent advances in the understanding of the pathogenesis of these disorders and the use of biomarkers for their detection.

We present the potential role of TVS for the early detection of both ovarian and endometrial cancer in this chapter because they are both associated with specific, recognizable TVS findings during the early stages of disease, and so this modality can be used to investigate both of these entities in the same patient. We will also emphasize differences in the application and utility of TVS for screening and early detection in ovarian and endometrial cancers, while highlighting the relevant differences in the presentation, clinical course, and prognosis of these diseases.

Of these two cancers, TVS has a greater potential role for the detection of early ovarian cancer, which tends to have a nonspecific clinical presentation compared with the signs and symptoms associated with early stage endometrial cancer. Currently only 25% of patients with ovarian cancer are diagnosed in an early stage. Early stage disease has a significantly improved prognosis compared to late stage disease, as surgery can frequently be curative in women whose disease is confined to the pelvis. It has been postulated that if screening TVS of the ovary could increase the detection of early stage ovarian cancer to 75% of patients, the total mortality burden of the disease would decrease by 50%. One limitation of TVS for ovarian cancer screening concerns type 2 ovarian cancer, a particularly aggressive disease associated with hereditary cancer syndromes that frequently becomes widely metastatic without first producing a detectable mass.

In this chapter, we will discuss the identification of an appropriate population to benefit from early detection, consisting of women at greatest risk for ovarian cancer by clinical history, CA-125, breast/ovarian cancer genetic analysis (BRCA1 and BRCA2), and proteomics. Women with risk factors for endometrial cancer such as obesity, diabetes, anovulation, polycystic ovarian syndrome, and hypertension are also excellent candidates for early detection (Table 36-2).^1,2 Used for this purpose, TVS has a high negative predictive value but is limited by a relatively low disease prevalence, which decreases the positive predictive value and cost-effectiveness of screening methods, including sonography, for the early detection of these diseases. The accompanying high false positive rate necessitates a carefully constructed algorithm to evaluate screening results to minimize healthcare costs associated with further work-up and to avoid unnecessary surgery in women without cancer.

Additional sonographic imaging using 3D- and/or contrast-enhanced sonographic techniques may improve early detection of ovarian cancer. These will be mentioned as they relate to improving screening efficacy, taking into consideration their incremental cost and requirements for high-level technical expertise.

To best understand the applications and limitations of early detection of these two cancers, one must first consider some fundamental concepts of cancer screening. It is important to distinguish screening tests from diagnostic studies. By definition, a screening test is performed in an asymptomatic population, whereas diagnostic studies are performed on selected subjects when warranted by signs, symptoms, or clinical suspicion. The ideal screening test is one that meets the following criteria:

1. 
   

   Sensitivity of the test is high, ie, there are few false negatives.

   
   
2. 
   

   Specificity of the test is high, ie, there are few false positives.

   
   
3. 
   

   The test is easily reproducible.

   
   
4. 
   

   The test is readily available and both cost- and effort-effective.

   
   
5. 
   

   The test does not cause harm to the patient or put the patient at risk (ideally noninvasive, eg, imaging).

Other factors that influence the clinical utility of screening are intrinsic to the unique disease biology. For instance, doubling times for all histologic types of ovarian and endometrial carcinomas are influenced by many factors, such as tumor vascularity, oxygenation, and perfusion; this makes the effect of “length-time” and “lead-time” biases difficult to calculate and thus the ability of screening to decrease mortality difficult to determine. The effectiveness of a screening test therefore depends not only upon the statistical strength of the test but also upon the following disease characteristics:

1. 
   

   The disease must be serious and result in substantial morbidity or mortality.

   
   
2. 
   

   There must be a high preclinical prevalence of the disease in the screening population.

   
   
3. 
   

   Detection of disease by screening should change the natural history of the disease (eg, result in improved outcomes).

   
   
4. 
   

   There should be a low prevalence of pseudodisease (disease found by screening that would not otherwise impact the quality or length of a patient’s life).

As mentioned previously, one must consider the possibility of introducing “length-time” and “lead-time” bias when determining efficacy of a screening or early detection program. Length-time bias describes differences in outcomes between screened and unscreened populations as a result of overrepresentation of less aggressive tumors in the screened population. Length-time bias derives from the concept that a slower-growing tumor will have a longer preclinical period in which it may be detected by screening; however, outcomes for these tumors are intrinsically more favorable than outcomes for more aggressive tumors with shorter preclinical periods. Lead time refers to the period of time between development of a disease and diagnosis. Lead-time bias refers to the artifactual skewing of outcome measures such as length of time from diagnosis to death. If a disease is discovered early in the preclinical phase through screening, it may appear that time from diagnosis to death has been lengthened, although in reality the clinical course of the disease may not have changed.

These factors must all be considered when determining the efficacy of screening and/or early detection for ovarian and endometrial cancers. The sonographic techniques discussed in this chapter enjoy the advantages of posing minimal harm to the patient and being relatively reproducible, accessible, and affordable. We discuss the accuracy (sensitivity and specificity) of each technique in detail later in this chapter.

Factors intrinsic to ovarian and endometrial cancers themselves, however, pose greater challenges to effective screening. Although both of these cancers clearly merit screening on the basis of the seriousness of the diseases (the American Cancer Society predicts 14,240 deaths from ovarian cancer in 2016; 10,470 deaths from endometrial cancer), considerable barriers exist to screening. For example, the fundamental challenge facing ovarian cancer screening is detecting the tumor when still confined to the ovary (stage I) or within the pelvis (stage II). Five-year survival statistics show 85% to 90% survival for early-stage disease and much poorer survival (5%-25%) for more advanced stages. From these data, it is extrapolated that if screening could increase the number of patients whose ovarian cancers are detected at an earlier stage, this would improve long-term survival outcomes. In order to determine screening efficacy, the impact of length- and lead-time biases must be defined, yet this is difficult to calculate given the wide range of doubling times for various histologic types of ovarian and endometrial carcinomas.

The efficacy of screening is further limited by the low prevalence of these cancers in the general population. Although the annual incidence of endometrial cancer is higher than that of ovarian cancer (55,000 vs 26,500), endometrial cancer is associated with fewer mortalities (10,470 vs 14,240), presumably as a result of early symptomatic bleeding and subsequent diagnosis. It is important to note that although screening may be limited by low prevalence in the general population, it is possible to define a subset of women at higher risk for these diseases. As discussed in the next section, a selective population with certain risk factors can be identified in which the prevalence of preclinical ovarian and endometrial cancer is high enough to warrant early detection measures (see Table 36-1).

When used in a screening capacity, TVS can detect morphologic findings such as wall thickening and/or papillary excrescences within ovarian masses—findings suggestive of malignancy. Serum tests such as CA-125 are currently being studied both as an adjunct screening method to TVS and as a means to identify high-risk women who would benefit from sonography for early detection of ovarian cancer. CA-125 is an antigen that can be detected in serum and is associated with malignancies such as ovarian cancer. However, CA-125 has a relatively limited sensitivity and specificity. Only 47% of Federation of Gynecologists and Obstetricians (FIGO) stage I ovarian cancers were positive for CA-125 in one large series.¹ Furthermore, elevated levels of this glycoprotein can be seen in a variety of disorders, especially in the premenopausal age group. CA-125 is, however, an accurate means by which patients may be followed for detection of recurrent disease.² The role of TVS and CA-125 in ovarian cancer screening is being investigated in several large population screening trials.

1. 
   

   The recently completed U.S.-based study for Prostate, Colon, Lung, and Ovary (PLCO) randomly assigned 78,216 women aged 55 to 74 to undergo either annual screening with TVS and CA-125 (n = 34,523) or normal gynecologic care.³ Screening was not associated with earlier average stage of ovarian cancer at diagnosis and in fact was deemed inefficacious due to a high false positive rate and considerable morbidity associated with evaluating these false positives through unnecessary surgeries (ratio of surgeries performed to cancers detected was 19.5:1). A limitation of this study was the lack of a standard protocol to evaluate screening abnormalities. Because each patient’s personal gynecologist decided how to proceed after an abnormal screen, variability in each physician’s approach could potentially be a confounding factor, particularly when considering morbidity associated with further work-up of screening abnormalities.

   
   
2. 
   

   The United Kingdom Collaboration Trial for Ovarian Screening (UKCTOS) study enrolled 202,638 women aged 50 to 74 and randomly assigned one group to receive annual screening with TVS alone (n = 48,230), one group to receive annual CA-125 screening with TVS as a second-line test (n = 50,078), and one to receive no screening (n = 101,359).⁴ In both groups who received screening, the percentage of women with invasive cancer who were diagnosed at an early stage was increased (TVS alone: 50% at stage I or II; TVS with CA-125: 47% at stage I or II; no screening: 26% at stage I or II). The ratio of the number of surgeries performed per each diagnosis of ovarian cancer varied significantly between the two experimental groups. In the TVS screening alone group, an abnormal test resulted in the patient receiving a repeat ultrasound examination 6 to 8 weeks later; if that test was abnormal she received a clinical assessment with serum CA-125, repeat TVS, Doppler studies, and CT/MRI of abdomen and pelvis. A total of 845 women were deemed to be at risk after the clinical assessment and ultimately received surgery, with 45 cancers detected (ratio of surgeries performed to cancers detected was 18.8:1). In the multimodality treatment group, women received TVS only if they had abnormal CA-125 measurements, and this was followed by a similar clinical assessment if TVS was abnormal. Ultimately 97 of these women underwent surgery with 34 cancers detected (ratio of surgeries performed to cancers detected was 2.8:1).

   
   
3. 
   

   A prospective randomized controlled multicenter study in Japan assigned 41,688 postmenopausal women to receive annual TVS, CA-125, and annual pelvic exam and 40,799 women to receive no screening.⁵ Subsequent management of an abnormal screening result was determined by each patient’s gynecologist. Of the ovarian cancers detected in the screening group, 63% were stage 1, compared to 38% in the group that did not receive screening. Of the patients identified as high-risk on TVS, 64 underwent surgery; of these 20 were found to have primary ovarian cancer, and 10 were found to have metastatic disease to the ovary.

   
   
4. 
   

   The ongoing University of Kentucky Ovarian Cancer Screening Trial has thus far enrolled 41,413 women, including all women over 50 years of age and women over 25 years of age with a family history of ovarian cancer. Participants receive annual TVS, and data collected from this experimental group is compared to age-matched controls in the same geographic population. Evaluation after an abnormal screen is standardized; women receive a repeat screen in 4 weeks, and if abnormal, they undergo CA-125 measurement, CDS, and tumor morphology indexing. Women at high risk undergo laparotomy. Screening has so far been associated with an earlier stage of diagnosis; 68% of the 53 primary epithelial ovarian malignancies identified in the screening population were stage I or II, while 27% of the cancers detected in the population that did not receive screening were stage I or II. Furthermore, the 5-year survival of women whose ovarian cancers were detected via screening is 75%, compared to 54% for women diagnosed with ovarian cancer who did not receive screening.⁶

Endometrial cancer is more likely than ovarian cancer to be detected at an early stage because its presence is frequently associated with uterine bleeding. Although any patient who experiences postmenopausal bleeding is suspect for harboring endometrial cancer, only 10% to 15% of women with unexplained bleeding have this disease. TVS offers a complete assessment of the thickness and texture of the endometrium and provides a means to differentiate those patients who should undergo biopsy from those who may forgo this procedure. In general, endometria that measure 5 mm or less are usually atrophic, whereas those thicker than 5 mm may be proliferative, hyperplastic, or cancerous.⁷ Although the false-negative rate of TVS for the detection of clinically significant abnormal endometrial pathology is low (~2%), false-positive results can occur in up to 30% of patients.⁸ Apparent thickening of the endometrium is frequently seen in women taking tamoxifen, possibly due to cystic atrophy or stromal edema, resulting in the production of multiple interfaces depicted sonographically.⁹ A large series of women undergoing hormone replacement therapy showed overlap in the thicknesses of various endometrial abnormalities.⁸ There may be concern that cancer may be found with a thin endometrium; however, the thinnest endometrium found to be an endometrial cancer was reported to be 9 mm in one large series.⁷ There has been only one published report of an endometrial cancer whose endometrial bilayer measured 4 mm.⁷

TVS offers complete assessment of the thickness and texture of the endometrium. Occasionally, its accuracy is limited by the presence of fibroids or by the patient’s body habitus. If the endometrium is difficult to assess on the initial visit, sonohysterography should be performed for enhanced delineation of the endometrium.¹⁰ Inadvertent reflux of cancer cells is a theoretical concern, but is in practice very unlikely. Retrograde passage of cancer cells with subsequent peritoneal implantation has, to our knowledge, not been observed or reported.¹¹ In addition, positive pelvic washings are no longer considered an adverse prognostic factor unless accompanied by other adverse factors such as ovarian or lymph node metastases. Endometrial cell dissemination during diagnostic hysteroscopy has been documented in about one-quarter of patients undergoing this procedure.¹² However, one group reported a small but real risk of malignant cell dissemination in patients with endometrial carcinoma who undergo saline infusion sonohysterography (SIS).¹¹ Cytologic analysis showed the presence of malignant cells in the spilled fluid in 7% of subjects. This has led some to discourage the performance of sonohysterography in women in whom there is suspicion of endometrial carcinoma.¹³ Some have suggested that transtubal fluid leakage does not occur during hysteroscopy when intrauterine pressure is less than 40 mm Hg.¹⁴ Thus, the chance that malignant cells could reflux during SIS and survive to establish metastatic peritoneal disease is very unlikely.

In conclusion, liberal use of sonohysterography in patients with a thickened or irregular endometrium is encouraged for the clear delineation of polyps that may contain cancer.¹⁵ Inadvertent reflux and peritoneal implantation of cancer cells during the procedure is very unlikely and of unknown clinical significance.^16,17

Ovarian cancer has very few specific clinical symptoms early in its course and therefore is usually discovered late at a metastatic state. Unfortunately, screening the general population may not be efficacious with currently available tests, given the aforementioned relatively low disease prevalence and the associated morbidity associated with evaluating false positives with invasive procedures.¹⁸ Presently, the lifetime risk for developing ovarian cancer is 1 in 70 (1.4%) for the general American female population, with an average age at diagnosis of 57. However, close evaluation of women who are identified to be “at risk” will improve the efficacy of screening schemes (Table 36-2). More recent estimates indicate that the lifetime risk for developing ovarian cancer is now up to 1 in 55 (1.8%) for the general worldwide population.¹⁹ In comparison, risk factors such as nulliparity, perineal talc exposure, infertility, high-fat diet, and previous breast cancer each confer a 2.0% lifetime risk.²⁰ A family history of ovarian cancer also increases risk. Ovarian cancer in one second-degree relative increases a woman’s lifetime risk to 2.9%; one first-degree relative increases risk to 4.5%; and history of two first-degree relatives with ovarian cancer confers a lifetime risk of 39% (see Table 36-2).²⁰ Despite the greatly increased risk associated with family history of ovarian cancer, the vast majority (>90%) of ovarian cancers occur in women without a family history or known predisposing factor.²⁰

Hereditary ovarian cancer syndromes represent fewer than 10% of all ovarian cancers and can be subdivided into 3 general categories: cancer limited to the ovaries, breast–ovarian, and multiple site (Lynch type II: ovarian, proximal colon, and endometrial cancers). Women with one of these syndromes tend to be diagnosed with ovarian cancer at an earlier age (median 47 years, with 20% occurring before age 40) than the general population (median 57 years).

Recently, it has been recognized that genetics play an important role in the identification of women at risk for ovarian cancer.²¹ The most recent advancement in genetic testing to identify women at risk for developing breast and ovarian cancer is analysis of the BReast CAncer (BRCA) genes, located on the long arm of chromosome 17 (BRCA1) and the long arm of chromosome 13 (BRCA2). Mutation of BRCA1 confers a 40% risk of developing ovarian cancer by age 70, and BRCA2 confers a 20% risk. The protein products of BRCA1 and BRCA2 are part of the Fanconi’s anemia-BRCA DNA repair pathway and are involved in nonhomologous recombination repair. Other genes in this pathway that are associated with an increased risk of ovarian cancer are RAD51 and BRIP1. Hereditary cancer risk occurs when an individual inherits a mutation in one of these genes, while Knudson’s two-hit hypothesis postulates that cancer may occur when spontaneous mutation occurs at the other locus of the same gene. Loss of heterozygosity in the BRCA gene has been identified in up to 70% of sporadically occurring ovarian cancers. A high incidence of certain BRCA1 and BRCA2 gene mutations are found in Ashkenazi Jewish individuals, as well as in other specific populations, due to the founder effect.²²