Applied Radiology

Some of the material contained in this article was presented in part at the Society of Radiologists in Ultrasound Meeting, October 2007, Chicago,IL,and the Society of Uroradiology Meeting, February 2008, Rancho Mirage, CA.

Postmenopausal bleeding (PMB) refers to any bleeding other than the expected bleeding that occurs with sequential hormones in postmenopausal women. It is a common problem that affects 1 in 10 postmenopausal women older than 55 years. ^1,2 Approximately 70% of all gynecological consultations in perimenopausal and postmenopausal women are for PMB. ³ Abnormal perimenopausal or postmenopausal bleeding is associated with endometrial carcinoma in approximately 10% of cases, and PMB is considered to be caused by endometrial neoplasia until proven otherwise. ⁴ Approximately 90% of women with endometrial carcinoma present with vaginal bleeding. ⁵ In 18% to 40% of women, benign focal lesions, such as endometrial polyps and fibroids, are common. Although PMB is most commonly due to other conditions, it is essential to exclude endometrial carcinoma. Thus, accepted practice in the United States includes further evaluation to exclude endometrial carcinoma in women with PMB. ⁶

Formerly, dilatation and curettage (D&C) was the principal means of investigation for PMB. But D&C has now been replaced by newer methods of investigation, incuding outpatient endometrial biopsy, transvaginal ultrasound (TVUS), and hysteroscopy. However, there is still uncertainty about the most accurate, acceptable, and efficient diagnostic approach.

Definition of postmenopausal bleeding

Menopause is defined by the World Health Organization as the permanent cessation of menstruation resulting from the loss of ovarian follicular activity. From a symptomatic perspective, PMB is vaginal bleeding following a woman's last menstrual cycle, which is considered an episode of bleeding ≥12 months after her last period. ⁷

It may be difficult to assess abnormal bleeding in women who are using hormone replacement therapy (HRT). Unscheduled bleeding is the term used to describe breakthrough bleeding occurring in women on HRT. Therefore, assessment depends on the type of regimen. For sequential HRT regimens, abnormal bleeding may be heavy or prolonged at the end of or after the progestogen phase or may occur at any time (breakthrough bleeding). Continuous combined regimens induce amenorrhea and, therefore, avoid cyclical bleeding. According to Ettinger et al, ⁸ however, nearly half of women on continuous combined HRT regimens will experience bleeding at some time, usually within the first 6 months of treatment. Bleeding on continuous combined HRT regimens should be evaluated if it occurs after the first 6 months of treatment or if it occurs after amenorrhea has been established.

Etiology of postmenopausal bleeding

Various causes of postmenopausal bleeding are listed in Table 1 and include endometrial atrophy (approximately 75% of cases), submucosal fibroids, endometrial polyps, endometrial hyperplasia, endometrial carcinoma (approximately 10%), and estrogen withdrawal. ⁹ Patients with atrophic endometritis usually have been postmenopausal for >10 years. Treatment of atrophic endometritis is topical or systemic estrogens. Curettage has also been known to be curative for an atrophic endometrium. ¹⁰ Endocervical polyps are more common than ectocervical polyps. Endometrial hyperplasia is the result of unopposed estrogens on the endometrium and is commonly encountered in perimenopausal women. It is an abnormal proliferation of endometrial stroma and glands and represents a spectrum of endometrial changes that range from glandular atypia to frank neoplasia. As suggested by Kurman et al, ¹¹ up to one third of endometrial carcinoma is thought to be preceded by hyperplasia. Endometrial carcinoma is the most common gynecologic malignancy and the fourth most frequent site of malignancy in females in North America and Europe. Risk factors include obesity, unopposed estrogens, polycystic ovarian syndrome, and nulliparity. The common presenting symptom is postmenopausal bleeding or abnormal uterine bleeding in the premenopausal patient. Though a majority of the patients will be postmenopausal, approximately 20% of the patients will be premenopausal and will present with irregular bleeding. ¹²

Diagnostic tools available and choice of study

Since endometrial carcinoma is an important differential diagnosis for postmenopausal bleeding, the standard of care is further evaluation of any PMB. Various investigations available for evaluation of the endometrium are TVUS, saline infusion sonohysterography (SIS), D&C, hysteroscopy, and endometrial biopsy. So the question is, which test is to be done first? That depends on the accuracy, safety, comfort, efficiency, and availability of each modality. The sensitivity, specificity, and adverse events associated with TVUS, hysteroscopy, and endometrial biopsy are shown in Table 2.

A definitive diagnosis in PMB is made by histology. Historically, endometrial samples have been obtained by D&C. This involves curetting the walls of the uterine cavity in a systematic fashion. However, the technique is blind, and it covers as much as 60% of the cavity but may miss polyps as they recoil from the passing curette. ¹³ In a series by Grimes et al, ¹⁴ endometrial lesions were overlooked in up to 10% of cases in which D&C was the only procedure used. Also, D&C must be performed as an inpatient procedure, which makes it a costly procedure and increases the risks associated with the use of general anesthesia. Outpatient endometrial sampling devices, such as the Pipelle endometrial sampler (Unimar, Inc., Wilton, CT), have a narrow bore (typically with an external diameter of approximately 3 mm) and sample the endometrium by aspiration. There are a variety of different samplers available, such as the Novak (Cooper Surgical, Trumbull, CT), the Kevorkian curette (Cooper Surgical), and the Vabra aspirator (Berkeley Medevices, Richmond, CA), but no systematic comparisons have been made between them. All methods of sampling the endometrium will miss some cancers. ^15-18 Unlike directed biopsy, the sample by these devices is obtained blind and as little as 4% of the cavity is sampled. Also, polypoid lesions may not be sampled or removed by these devices. Guido et al ¹⁹ evaluated the sensitivity of the Pipelle endometrial suction curette in the detection of endometrial carcinoma in patients with known endometrial cancer. They enrolled 65 patients with known endometrial cancer who were undergoing hysterectomy as part of a formal staging procedure. Endometrial biopsies were performed prior to each surgical procedure, and the biopsy results were compared with the hysterectomy specimen. They found that Pipelle biopsy was adequate for analysis in 63 of 65 patients (97%). When the cancer was localized to <25% of the surface, however, 7 cases were missed; when cancer was localized to <50% of the surface, 11 cases were missed; and when cancer involved >50% of the surface, no cases were missed. They concluded that Pipelle endometrial suction curette is an effective office device for evaluating patients at risk of endometrial cancer but that tumors localized to a polyp or small area of endometrium may go undetected. ¹⁹ The authors also noted that D&C had a greater complication rate, but the diagnostic accuracies of the 2 techniques were comparable.

The main disadvantage of the use of the Pipelle is the relatively high proportion of inadequate samples, particularly among postmenopausal women. Dijkhuizen et al ²⁰ conducted a meta-analysis by examining 39 studies that included 8000 patients who underwent endometrial biopsy and compared results with those from D&C, hysteroscopy, or hysterectomy. The authors found that the detection rate for malignancy was higher in postmenopausal women. The Pipelle device proved to be the best sampling tool, with cancer detection rates of >99% for postmenopausal women and 81% sensitivity in detecting complex atypical hyperplasia.

Pelvic ultrasound can be used to evaluate the uterus and endometrium. In the absence of visible abnormalities (eg, fibroids), endometrial thickness (ET) and homogeneity have been used as markers for endometrial pathology. Details of ultrasound and sonohysterography in postmenopausal bleeding are mentioned below.

The guidelines for evaluation of PMB suggested by the Society of Radiologists in Ultrasound (SRU) consensus panel ⁶ are limited since there have been no full-fledged cost-effectiveness analysis and no large-scale randomized controlled trials have compared the effectiveness of ultrasound or biopsy. The SRU panelists stated that, in their opinion, either TVUS or office biopsy can be used as the initial test in the evaluation of women with PMB. The panelists agreed that SIS or hysteroscopy was appropriate when a focal abnormality is suspected on TVUS. One advantage of hysteroscopy is that it permits biopsy of a focal mass. ²¹

Imaging of normal and abnormal endometrium in postmenopausal patients

Standards for ultrasound

Transabdominal ultrasound (Figure 1) alone is not sufficient to assess the uterus due to the suboptimal resolution of the uterus and endometrium. However, a transabdominal ultrasound should be performed so that large masses or fluid collections are not missed. A transabdominal ultrasound should be followed by a TVUS (Figure 1). The uterus and adnexae are evaluated by long- and short-axis views (Figure 2). This should include assessment of the cervix. The endometrium is evaluated on both the long- and short-axis views through the uterus; however, the endometrium is measured on the long-axis view (Figure 2). The entire endometrium should be evaluated to avoid missing focal lesions. The thickest measurement of the endometrium in the midline is reported (Figure 3). Any focal abnormalities should be noted and their appearance and location described (Figures 4 and 5). Any vascularity within the focal lesion should also be noted. Polyps may be broad-based and sessile or pedunculated with a central stalk of vascularity on color Doppler. ²² Fluid within the endometrium is not included in the measurement. In cases of endometrial fluid, ET is the sum of the thickness of the 2 endometrial layers (Figure 6). Non-visualized endometrium should not be considered as a normal endometrium. The nonvisualization of the endometrium could be due to large fibroids displacing the endometrium (Figure 7) or endometrial carcinoma infiltrating the myo metrium (Figure 8). ²³ Two separate studies by Briley et al ²⁴ and Karlsson et al ²⁵ have suggested that nonvisualized endometrium or poorly seen endometrium should be biopsied.

Imaging of normal and abnormal endometrium

The endometrium in a postmenopausal patient should be thin and homogenous. There is controversy regarding the cutoff value of ET that should be considered normal. Some investigators have found that the ET decreases with age ^26,27 ; however, others believe that there is no statistically significant change during menopause. ²⁸ Warming et al ²⁹ reported the results of their study that followed 1182 asymptomatic patients between 45- and 83-years-old without HRT. They noted that when using transvaginal endometrium measurement, the normal range of ET was correlated with years since menopause (YSM). For women with <5 YSM, the mean ET was significantly greater than for women >5 YSM. Furthermore, the mean ET decreased in the first 5 YSM, whereas it remained almost unchanged from 5 to 13 YSM, only to increase slightly during the rest of the postmenopausal years. ²⁹

In general, a double-layer thickness <5 mm without focal thickening excludes significant disease. ^30,31 Homogenous endometrium measuring ≤5 mm is considered within the normal range with or without HRT. ³² Several studies have established that the ET generally does not exceed 4 or 5 mm in a normal postmenopausal woman. ^30,31 However, there is considerable confusion as to which measurement should be used as a cutoff to trigger further investigations. In a study by Gull et al, ³² the prevalence of endometrial carcinoma in patients with ET of <5 mm was 0.6%. A meta-analysis study by Smith-Bindman ² found that with the use of a cutoff of 5 mm, 96% of endometrial carcinomas would be detected in postmenopausal patients with bleeding. When this threshold was changed to 4 mm, it did not alter the sensitivity for cancer detection (sensitivity 96%); however, it increased the false-positive rate from 9% to 30%. ² They also found that TVUS was better at detecting cancer than it was at detecting polyps or hyperplasia. By using a 5-mm threshold, 96% of women with cancer had an abnormal ultrasound result, whereas 92% of women with endometrial disease had an abnormal ultrasound. ² In a prospective study by Granberg et al ³³ of 1110 women with PMB, endometrial pathology was found most frequently when the ET was >8 mm, and no endometrial cancers were detected in women with ET <4 mm. Similarly, an evaluation of 419 women with PMB by Garuti et al ¹⁷ assessed the sensitivity of 2 ET thresholds: 4 mm and 8 mm. The authors reported a diagnostic sensitivity of 95.1% and a specificity of 54.8% with the use of the 4-mm cutoff, and 83.8% sensitivity and 81.3% specificity when the 8-mm cutoff was used. Using a maximum of 5-mm ET to exclude carcinoma, a study by Briley et al ²⁴ of 182 women with PMB found no cases of carcinoma; however, 3 patients had hyperplasia. According to Levine et al, ²⁷ the variation in what is considered the upper limit of normal ET seems to be correlated with the body-mass index and type of HRT, with higher values in heavier women and in those taking combinations of estrogen and progesterone. Another study by Gull et al ³⁴ reported a 0.6% prevalence of endometrial cancer in women with PMB and ET <4 mm. This prevalence increased to 19% in women with an ET <5 mm. The authors concluded that endometrial biopsy is not required in women with an ET <4 mm. Several studies also suggest that an ET >15 mm is highly suggestive of endometrial carcinoma, although on sonograms polyps can present as an endometrium >15 mm thick. Other authors suggest that a minimum sonographically measured ET of 6 mm should be utilized to reduce the number of false-positives. ³⁵

Some other markers have been reported as suggestive of malignancy, including collection of fluid in the endometrial cavity and irregularity of the myometrial-endometrial interface. In 1995, Weigel et al ³⁶ suggested that adding assessment of the endometrium for homogeneity, presence of a central echo, and echogenicity would be most useful in diagnosing cancers in women whose ET is in the ''gray area'' of 4 to 10 mm. Another prospective trial by Fistonic et al ³⁷ found a 100% sensitivity for irregular myometrial-endometrial interface in predicting endometrial carcinoma. Doppler imaging has been reported to increase the sensitivity and specificity of sonography in detecting endometrial pathology. In a prospective study of 60 women with PMB, Amit et al ³⁸ reported a sensitivity of 86% and specificity of 89% for pulsed Doppler (pulsatility index point cutoff, 1.0). Sheth et al, ³⁹ however, evaluated pulsed Doppler in postmenopausal women with thickened endometria and found that low-impedance arterial flow did not differ significantly between benign and malignant lesions.

Errors in TVUS

Even in the most experienced hands, TVUS may provide inaccurate results. According to Fleischer, ⁴⁰ errors arise from improper measurement, single linear measurement, and distortion by fibroids. The most common error is improper measurement of the endometrium. The shape of the endometrium is not purely geometric, and measurements of its width rather than its anteroposterior 2-layer thickness can result in overestimation of its thickness. The echotexture of the endometrium should be carefully assessed. Any echogenic mass that disrupts the smooth texture with or without punctate cystic areas may signify the presence of a polyp (Figure 9). ⁹ Since the shape of the endometrium is not geometric, the thickness of the endometrium may vary in the fundus and the body. Also, focal lesions may make the endometrium thicker focally. According to Karlsson et al, ²⁵ readers of ultrasonograms also vary in their interpretations, with a difference of almost 1.5 mm seen in a study among sonographers. Submucosal (Figure 10) and intramural fibroids may distort the endometrium, making it difficult to measure thickness in the anteroposterior plane. According to Levine et al, ²⁷ the endometrium also varies with the cycle in patients on cyclic estrogen and progesterone with a difference of up to 3 mm.

Saline infusion sonohysterography

Saline infusion sonohysterography is a diagnostic technique consisting of an intrauterine infusion of saline solution by means of a catheter with an inflating balloon positioned in the cervical canal. The expansion of the uterine cavity is directly observed through TVUS. Saline infusion sonohysterography allows reliable differentiation between focal and diffuse endometrial and subendometrial lesions. ⁴¹ According to Farquhar et al, ⁴² SIS is primarily indicated in symptomatic and asymptomatic patients who have an irregular endometrial appearance on TVUS when endometrial lesions (such as polyps [Figure 11], submucosal fibroids, and focal endometrial hyperplasia) are suspected. Saline infusion sonohysterography can be performed safely and easily as an outpatient procedure, and anesthesia is not needed. The detection rates for focal abnormalities are comparable with those of hysteroscopy. ⁴³ The SRU consensus panel favored SIS when a focal endometrial abnormality is suspected on the TVUS to confirm that a focal abnormality is indeed present and to better determine the nature of the focal abnormality. ⁶ According to Goldstein et al, ⁴⁴ SIS may also be helpful when a thickened endometrium has been identified on the sonogram to allow more efficient triaging of patients. Saline infusion sonohysterography will show whether the abnormality is diffuse (in which case a biopsy or D&C would be the next step) or focal (in which case hysteroscopy would be the next step). Saline infusion sonohysterography is also helpful if there is a discrepancy between the findings on the TVUS and biopsy. Saline infusion sonohysterography is also more sensitive than TVUS or biopsy in detecting focal endometrial abnormalities. ^43,45

In a multicenter study investigating the utility of SIS, Bree et al ⁴³ reported polyps in 47% and endometrial hyperplasia in 4% of women with PMB. In a recent study, Neele et al ⁴⁵ found that nearly 35% of asymptomatic postmenopausal women had endometrial abnormalities detected with SIS.

Issues related to tamoxifen and HRT

Tamoxifen is a selective estrogen receptor modulator that is commonly used as an adjuvant therapy in the treatment of breast cancer. It binds to the estrogen receptor and acts as both estrogen antagonist and agonist, depending on the target organ. ^46,47 Tamoxifen has been shown to help prevent the original breast cancer from recurring after breast surgery while also hindering the development of new cancers in the opposite breast. ⁴⁸ It is well known that tamoxifen use increases the risk of developing endometrial carcinoma. ⁴⁹ In a study by Cuzick et al, ⁵⁰ women taking tamoxifen were 2.4 times more likely to have endometrial carcinoma than women who were not taking the drug. This risk also increased with increasing duration of use (relative risk 2.0 times for 2 to 5 years and 6.9 times for >5 years).

In a patient on tamoxifen, TVUS shows a thickened endometrium with a characteristic appearance of tiny cystic areas (Figure 12). ⁴⁹ Goldstein et al ⁵¹ suggested that these changes actually represented small subendometrial anechoic areas in the proximal myometrium; however, others maintain that these changes are actually in the endometrium. Using histopathologic analysis, McGonigle et al ⁵² concluded that endometrial cystic atrophy may be responsible for these benign changes.

Estrogens had been used as a monotherapy to relieve postmenopausal women's complaints since the 1940s. Estrogen has a proliferative effect on the endometrium glands as well as on stroma. The proliferative activity of the endometrium is preserved after menopause. ⁵³ Although shortly after the appearance of postmenopausal HRT a cause-and-effect relationship between estrogens and endometrial carcinoma was suspected, it could be proven until 1974. ⁵⁴ In 1961 hat scientists proved the capacity of progesterone to counteract the proliferative effect of estrogen on the endometrium. Since 1975, the use of combined preparations has grown; estrogen monotherapy is now used only under limited circumstances. ^55,56 Progesterone, administered to counteract the effect of estrogen, can also counteract endogenous estrogen and thus decrease the risk of endometrial carcinoma below the original level. According to Erkkola et al, ⁵⁷ during postmenopausal sequential HRT, time-dependent appearance of endometrial hyperplasia can be expected, which poses an increased risk for the development of endometrial carcinoma, since the prevention of osteoporosis requires 10 years of treatment. Therefore, sequential treatment is indicated in the frame of short-term palliative therapy to relieve complaints. ⁵⁷

Conclusion

Postmenopausal bleeding is a common problem, with endometrial carcinoma being an important differential diagnosis. At present, endometrial sampling is the gold standard for the diagnosis of endometrial carcinoma; however, endometrial sampling is an invasive procedure. Ultrasound can be used for noninvasive assessment of the endometrium. However, further evaluation of PMB depends on the expertise available at a particular institution and physician/patient preference.