Cancer of the Endometrium

Published on 09/04/2015 by admin

Filed under Hematology, Oncology and Palliative Medicine

Last modified 09/04/2015

Print this page

This article have been viewed 1641 times

49 Cancer of the Endometrium

I-Chow Joe Hsu, MD, Kaled M. Alektiar, MD, Dattatreyudu Nori, MD, FACR, FACRO, FASTRO

Epidemiology

Endometrial cancer is the most common gynecologic cancer and the fourth most frequently diagnosed cancer in women in the United States. According to the 2009 cancer statistics, the estimated number of newly diagnosed cases is 42,160.¹ Endometrial cancer occurs most commonly in postmenopausal women, although 25% of cases occur in women before menopause and 5% in patients younger than 40 years of age.² Worldwide, the incidence of endometrial cancer is higher in North America and Western Europe than in developing countries and Japan. Immigrant populations generally assume the risk of native ones, reflecting the importance of environmental factors in this disease. In the United States, Caucasian women have a twofold higher risk of developing this disease than African-American women, and it is more common among urban than rural residents. The expected number of deaths from endometrial cancer in 2009 is 7780, making it the eighth-leading cause of death from cancer in women.¹

Anatomy

The uterus is a hollow muscular organ situated in the pelvis between the urinary bladder anteriorly and the rectum posteriorly. Structurally and functionally, it is divided into two parts—the body and the cervix. At the transition between these two, there is a slight narrowing called the isthmus, which is the location of the internal os of the cervix. It is this region that differentiates into the lower uterine segment during parturition. The body of the uterus is made up of three layers—the endometrium, consisting of a complex glandular lamina functionalis; a basal layer, which is closely applied to the underlying myometrium, consisting of smooth muscle fibers; and the outer serosa, which is essentially the visceral peritoneum. The nulliparous uterus is approximately 7.5 cm long, the upper two-thirds (5 cm) being the corpus and the lower one-third (2.5 cm), the cervix. The uterine cavity is approximately 6 cm in length and the wall is approximately 1.2 cm in thickness. After menopause, the uterus gradually atrophies, the myometrium thins out, and the cavity may be less than 5 cm in length, with a markedly attenuated cervix.

The main supports of the uterus are fibromuscular condensations of tissue in the pelvic fascia, namely the broad ligaments, the round ligaments, the uterosacral ligaments, and the transverse or Mackenrodt cardinal ligaments. The arterial supply is from the paired uterine arteries, which usually arise from the anterior division of the internal iliac arteries. The uterine artery enters the uterine wall at the isthmus, after crossing the ureter, and has free anastomoses with the ovarian and vaginal arteries. The venous drainage follows the arterial supply. There are relatively few lymphatics in the endometrium, but the myometrium and the subserosa have a rich lymphatic network. The uterine body drains primarily to the external and internal iliac lymph nodes, which then drain to the lateral aortic nodes. When the cervix is involved by disease, the pattern of spread of cancer parallels that of cervical cancer. Anastomoses between lymphatics from the body of the uterus and those in the round ligament provide a pathway for the rare metastasis to the inguinal lymph nodes. Similarly, there is a direct pathway for the lymphatics from the fundus of the uterus to the lateral aortic nodes, bypassing the iliac nodes (Fig. 49-1).

FIGURE 49-1 Lymphatic drainage pathways from the uterine body. A, Aortic nodes: all nodes anterior to the lumbar spine, which lie anteriorly, posteriorly, and laterally to the abdominal aorta, superior to its bifurcation. These drain the cornual and fundal region of the uterus via the lymphatics of the ovarian pedicle (1). B, Interiliac nodes: nodes lying within the pyramidal space bounded by the medial surface of the external iliac artery, the internal iliac (hypogastric) artery and its anterolateral branches, and the lateral pelvic wall. The middle and lower portions of the body of the uterus drain here via the broad ligament lymphatics (2). C, Deep femoral nodes: all nodes within the fatty tissue of the femoral triangle, which lie beneath the fascia lata. The most medial of these is the Cloquets (Rosenmüller) node. This constitutes an alternative pathway for drainage of the cornual region, via the lymphatics of the round (teres) ligament (3).

Risk Factors

The true cause of endometrial cancer remains unknown. However, it is safe to say that in a large proportion of patients, the cancer seems to be related to estrogen stimulation.

Hormones

Alterations in the hormonal milieu are closely associated with an increased risk of developing endometrial cancer. The interaction between estrogens and endometrial cancer has been discussed for several decades.³ The coexistence of hyperplasia and adenocarcinoma and the apparent evolution of adenomatous hyperplasia into florid adenocarcinoma have been recognized by pathologists for more than 3 decades,⁴ and cytopathologists have long reported on the increased prevalence of well-cornified vaginal epithelium, indicating high levels of estrogens, in women with endometrial cancer.⁵ The clinical counterpart of these pathologic findings is the increased risk for endometrial cancer in several conditions associated with abnormally high levels of estrogens. Thus, Stein-Leventhal syndrome, or polycystic ovarian syndrome, has traditionally been associated with a very high risk (up to 25%) of endometrial cancer. Although this number may be somewhat of an overestimate, these cancers account for a large proportion of cases in young women.⁶ Similarly, a fairly large number of patients with granulosa or theca cell tumors of the ovary, or both, also develop a concomitant endometrial carcinoma. This association was first described by Novak and Yui⁷ in 1936, and it is now recognized that the probability of developing endometrial carcinoma is related to the predominance of theca cells in the ovarian tumor, because these are the important source of feminizing estrogens.³ It is also interesting to note that the prognosis of endometrial cancer in either of these settings is more favorable than for endometrial cancer in general.⁸

Late menopause, nulliparity, and obesity are classically associated with the development of endometrial cancer. Thus, delayed menarche is associated with a lower risk of endometrial cancer, whereas menopause after 52 years of age results in a 2.5-fold increase in the risk. The effect of delayed menopause (and to a lesser extent, early age at menarche) may reflect prolonged exposure of the endometrium to unopposed estrogen stimulation in the presence of anovulatory cycles.⁹ Similarly, nulliparous women have the highest risk, with 25% to 30% of all endometrial cancers occurring in nulliparous women. Having one child reduces this risk by half, and quintipara have less than 20% the risk faced by nulliparous women.¹⁰ This adverse effect of nulliparity may also be a manifestation of prolonged periods of infertility.¹¹ In one study, nulliparous women who sought advice for infertility were at an almost eightfold excess risk compared with nulliparous women without an infertility problem.¹² A number of biologic mechanisms have been proposed for this increased risk with infertility: anovulatory menstrual cycles resulting in prolonged exposure to estrogens without sufficient progesterone, high serum levels of androstenedione, lower levels of serum sex-hormone binding globulin (with resultant high levels of free estrogen), and the absence of the monthly sloughing of the endometrial lining (allowing the residual tissue to become hyperplastic).¹³ Finally, obesity is a well-recognized risk factor for endometrial cancer, possibly accounting for up to 25% of occurrences of the disease.¹⁴^,¹⁵ Obesity affects the incidence of both premenopausal and postmenopausal endometrial cancer.¹⁴^–¹⁶ Being 21 to 50 pounds overweight increases the risk by almost threefold, and women who are more than 50 pounds overweight may have as high as a tenfold increase in risk of developing endometrial cancer.¹⁷ The increased peripheral conversion of androstenedione to estrone may be the underlying pathophysiologic mechanism for this risk. Apart from the absolute increase in weight, the duration of obesity¹⁵ and the distribution of body fat may also play an important role.^15,^18,¹⁹ Conversely, physical activity may exert a protective effect on the risk of endometrial carcinoma.²⁰^,²¹ This association is biologically quite feasible, because physical activity is well known to effect changes in hormone levels and menstrual cycles.

Exogenous estrogens, in the form of estrogen replacement therapy, are also implicated in endometrial carcinoma. Several studies have looked at this correlation, and case reports have existed in the literature for 40 years.²² These studies have been performed in at least eight countries, under a wide spectrum of medical care systems and conventions, in urban and rural settings, in high-usage and low-usage populations, using cohort and case-control designs, and with population-based and hospital-based comparison groups. Most of the studies have shown a strong association between past estrogen use and the development of endometrial cancer. The biologic plausibility of this association is further enhanced by the positive correlations between the dose of estrogen used, the duration of use, and the risk of subsequent endometrial cancer. At least 2 to 3 years’ use was required to develop an increased risk, and the highest relative risks of 10 to 20 are observed after at least 10 years of estrogen use.²³^–²⁵ The relative risk is highest among thin, nondiabetic, normotensive women.^23,^25,²⁶ This may hint at a difference in estrogen metabolism in this group of patients, or that the risk in the typical obese, diabetic, hypertensive patient is already high enough as to not be affected by additional exogenous estrogens. Like the tumors arising in the setting of high levels of endogenous estrogen, most of these tumors are also relatively nonaggressive: They are generally diagnosed at an early stage, are well differentiated, and have little myometrial penetration²⁶^,²⁷ and a relatively small effect on mortality.²⁸^,²⁹

Further evidence for the role of estrogens comes from the studies evaluating the effects of oral contraceptives. These studies have demonstrated significantly higher risks in users of the sequential oral contraceptives—which have a relatively high dose of estrogen with a weak progestin—and substantially lower risks of endometrial cancer in the women using estrogen-progestin combination pills, which have a higher dose of progestins.^11,^16,^30,³¹ The protective effect of the combination “pill” may be most marked in nulliparous women and in nonobese patients who have not been exposed to noncontraceptive estrogens.³²

Progesterone is known to produce regressive changes in endometrial hyperplasia, and recently there has been an upsurge in enthusiasm for combining estrogen therapy with progestin to counteract the carcinogenic effects.^23,^33,³⁴ The recently published Women’s Health Initiative randomized trial of healthy, postmenopausal women showed that estrogen and progestin did not increase the risk of endometrial or colorectal cancers, but it did significantly increase the risk of invasive breast cancer.³⁵

The risk of endometrial cancer from tamoxifen use has been the subject of many studies in light of its widespread use for breast cancer treatment and prevention. The mechanism of action is competition with endogenous estrogen for estrogen receptors. In premenopausal women, tamoxifen has an antiestrogenic effect, but in postmenopausal women it has a weak estrogenic effect because of the up-regulation of estrogen receptors. In an overview of the breast cancer prevention trials, Cuzick et al.³⁶ reported that the rates of endometrial cancer were increased with tamoxifen in all prevention trials (relative risk 2.4, 95% confidence interval [CI]: 1.5-4; P = 0.0002). Most of the excess risk was seen in women 50 years old or older.³⁶ Most authors³⁷^–³⁹ believe that there is no real difference in the stage, grade, or prognosis of endometrial cancers associated with tamoxifen, although there are emerging data to the contrary.⁴⁰

Diet

In contrast to cervical cancer, uterine cancer is a disease of the Western countries; even there, it is more common in the urban population belonging to the higher socioeconomic strata of society. The rates are highest in North America and Northern Europe; intermediate in Israel, Southern Europe, and Latin America; and lowest in Africa and Asia.⁴¹ This geographic difference may be a reflection of the high content of animal fat in the Western diets,⁴² because both observational and interventional studies have shown a higher level of plasma estrone, estradiol, and prolactin among women consuming a high-fat diet.⁴³^,⁴⁴ Diet probably plays a multifaceted role in the risk of developing endometrial cancer, as other studies have shown a protective effect of high levels of micronutrients⁴⁵ and diets high in fruits and vegetables.⁴⁶

Other Comorbidities

Endometrial carcinoma is also known to be associated with several other medical conditions. Diabetes and hypertension have long been related to a significant increase in the endometrial cancer risk, and recent reports confirmed that this risk persisted over time and was independent of the patient’s weight.⁴⁷^,⁴⁸ Other interesting correlations include the interrelationship with smoking. The bulk of evidence from epidemiologic studies suggests that smoking reduces the risk of developing endometrial cancer. The exact cause is not clear, but mechanisms such as the antiestrogenic effect of smoking on circulating estrogen concentrations, a reduction in relative body weight, and earlier age at menopause have been suggested.⁴⁹ Alcohol also has been associated with a reduction in the risk of endometrial cancer, perhaps by attenuating the endogenous estrogens.⁵⁰

Genetics

Genetic predisposition to endometrial carcinoma occurs primarily within the context of hereditary nonpolyposis colorectal cancer (HNPCC), which probably accounts for less than 5% of all endometrial cancer cases. In patients with HNPCC there is increased risk of developing not only colon cancer but other cancers as well. The latter (Lynch syndrome II) includes endometrial cancers; ovarian cancers; gastric adenocarcinomas; carcinomas of the small bowel, pancreas, and biliary tree; transitional cell carcinomas of the ureters and renal pelvis; and tumors of the skin (Muir-Torre syndrome).⁵¹ The lifetime risk estimates for endometrial carcinoma range from 40% to 60%, corresponding to a relative risk of 13 to 20. These estimates are not population-based, but are derived from analyses of HNPCC families and thus may be higher than expected in the population generally. The genes involved in HNPCC include MSH2, MLH1, and MSH6. Tumors from patients with HNPCC syndrome are characterized by genetic instability that leads to error-prone deoxyribonucleic acid (DNA) replication-RER+ phenotype, also referred to as microsatellite instability.⁵² Apart from the special considerations of colorectal cancer, screening for endometrial and ovarian cancers should also be carried out. Currently for endometrial cancer, this consists of assessment of the endometrial stripe at the time of transvaginal ultrasonography, as well as endometrial aspiration for cytologic and histologic examinations. With respect to risk-reducing surgery, there are no data on the efficacy of hysterectomy and bilateral salpingo-oophorectomy (BSO) for cancer prevention in women at risk for HNPCC, and no recommendation was made for or against risk-reducing surgery by the Cancer Genetics Studies Consortium that addressed this issue.⁵³ Nevertheless, it is reasonable to assume that hysterectomy would be completely preventive for endometrial carcinoma, and there are no published case reports to the contrary.

Pathologic Conditions

Hyperplasia

On gross inspection, hyperplastic endometrium has no distinctive features. There is a diffuse thickening resulting in an increase in the amount of endometrial tissue. The endometrium appears pale and knobby, and has a spongy, velvety feel. Microscopically, the diagnostic criterion for hyperplasia is an increase in the number and size of proliferating glands. The International Society of Gynecologic Pathologists has standardized the subclassification of endometrial hyperplasia.⁵⁴ In simple hyperplasia, there is only glandular proliferation and enlargement with increased stromal cellularity. This rarely progresses to carcinoma (<1%). Complex hyperplasia is characterized by back-to-back proliferation of glands with intraluminal papillae, epithelial pseudostratification, and few mitotic figures. If there is no cytologic atypia, the risk of malignant degeneration is again quite low, on the order of 3%. Any proliferation demonstrating cytologic abnormalities (in cellular or nuclear morphology) is classified as atypical hyperplasia. Atypical hyperplasia has a much higher risk of progression to an invasive carcinoma—8% for simple atypical hyperplasia, increasing to almost 30% for complex hyperplasia associated with atypia. Based on the histologic type and the patient’s age and general medical condition, treatment options for endometrial hyperplasia include periodic endometrial curettage, high-dose progestin, ovulation induction, total abdominal hysterectomy (TAH) with BSO, and even intracavitary radiotherapy.

Carcinomas

On gross inspection, most endometrial cancer involves the endometrial surface diffusely and is found frequently in the upper part of the uterine body, especially on the posterior wall. Only 40% of endometrial cancers are localized, which may take the form of a polyp or a localized, flattened, pyramidal thickening involving a part of the endometrial surface. The tumor obtained from endometrial curettage has a typical firm feel with a dry, granular, and often friable appearance. Necrotic areas may be yellow, white, or red. The International Society of Gynecologic Pathologists has proposed a classification for endometrial carcinomas (Table 49-1). Although the typical carcinoma is easy to diagnose, it may not always be easy to distinguish between a well-differentiated adenocarcinoma and an advanced atypical hyperplasia.⁵⁵^,⁵⁶ The chief feature that helps in this distinction is the extent of the proliferative changes: To diagnose a specimen as well-differentiated adenocarcinoma, the proliferative changes must be widespread and involve at least one half of a low-power microscopic field, 4.2 mm in diameter. Nuclear atypia is not required for this distinction.

Table 49-1 Classification of Endometrial Carcinoma

Clear Cell Carcinoma

Clear cell carcinoma of the endometrium resembles renal carcinoma. Previously believed to arise from the mesonephric duct remnants (hence the synonyms mesonephroma or adenocarcinoma of mesonephric origin), its origin from müllerian structures is now well established.⁶² Unlike vaginal and cervical clear cell carcinoma, it is not related to intrauterine diethylstilbestrol exposure. The microscopic structure may vary from solid patterns to glandular differentiation. In the latter pattern, the small cells resembling “hobnail” cells line spaces and glands. These are cells that have extruded their cytoplasm, leaving bare nuclei that protrude into the glandular lumens. The prognosis of this cancer is somewhat similar to that of serous cancer.⁶³

Mixed Histologic Findings

Mixed cell–type endometrial cancer composed of two or more pure types is not uncommon. By convention, to be designated mixed, the other cell type component has to compose at least 10% of the tumor. Except for mixed endometrioid and serous or clear cell carcinoma, the clinical significance of mixed cell type is questionable.

Simultaneous Tumors

Cancers of an identical type may be discovered in the ovary and endometrium simultaneously.⁶⁴ Usually, the site of the largest tumor is assigned the primary origin, but occasionally true primary endometrial and ovarian malignancies may coexist. This field effect of the müllerian system may occur in as much as 15% to 20% of the ovarian endometrioid tumors.⁶⁵ If the endometrial tumor is less than 5 cm in diameter, is well differentiated, has no vascular invasion, is limited to less than the middle third of the myometrium, and the ovarian lesions are bilateral, it is more likely that there are two concomitant primary tumors.⁶⁶

Molecular Biology

Several investigators in the past pointed out that there are two distinct types of endometrial cancer.⁶⁷^,⁶⁸ In type I endometrial cancer there is strong correlation with prior estrogen stimulation. The cancers in this category are often indolent in nature, with minimal myometrial invasion and low-grade histologic findings. They affect premenopausal and perimenopausal women. Type II endometrial cancer often affects postmenopausal women with no prior history of estrogen stimulation. The histologic characteristics of the tumors is often high grade, such as serous or clear cell cancers with deep invasion, and at a more advanced stage at the time of presentation. What is intriguing is the fact that at the molecular level, the existence of two distinct types of endometrial cancer seems to be validated.⁶⁹

Inactivation of the p53 tumor suppressor gene is seen in almost 90% of cases of serous carcinoma.⁷⁰^,⁷¹ Mutation in the p53 gene, however, is encountered in only 10% of endometrioid adenocarcinoma, with most occurring in International Federation of Gynecology and Obstetrics (FIGO) grade 3 tumors. Several studies have also demonstrated that p53 mutation is an independent predictor of poor outcome.⁷²^,⁷³ In contrast, mutation in the PTEN tumor suppressor gene is found in 30% to 50% of endometrioid adenocarcinoma.⁷⁴^,⁷⁵ PTEN mutation is also associated with early-stage, nonmetastatic disease and more favorable survival.⁷⁶ Microsatellite instability, which is found in patients with HNPCC, is also seen in approximately 20% of “sporadic” endometrial cancers.⁷⁷ Data in the literature suggest a favorable survival associated with microsatellite instability in endometrioid endometrial cancers.⁷⁸ Mutation in k-ras proto-oncogene is seen in 10% to 30% of endometrial cancer patients,⁷⁹ but it seems that there is no correlation between k-ras mutation and clinical outcome. Over expression of HER-2/NEU is also seen in 10% to 15% of endometrial cancers and is associated with more advanced disease and poor outcome.⁸⁰^,⁸¹ Other oncogenes⁸²^–⁸⁵ that have been found to be overexpressed or amplified include c-myc, bcl-2, c-fms, and B-catenin. Microarray analysis has further revealed distinct gene expression profiles among different histologic types of endometrial cancer.⁸⁶

Mode of Spread

Endometrial cancer tends to remain confined to the body of the uterus for a long time. In most cases, the initial spread is by local extension, along the endometrial surface with or without enlargement of the uterus. Subsequent growth continues in radial and longitudinal directions. Longitudinal growth results in cervical involvement, initially through the openings of the endocervical glands, and later with frank cervical stromal infiltration. The tumor can also extend along the cornua to involve the fallopian tubes (although ovarian involvement is more often embolic in nature). As the tumor grows radially, it penetrates the myometrium and makes its way toward the subserosa and serosa. In a small number of cases, following extensive penetration through the myometrium or the cervix, the tumor involves the pelvic soft tissues and continues to reach the pelvic sidewall; the tumor may also directly invade the bladder or the rectum. Once the tumor breaches the serosa, transperitoneal dissemination can occur, resulting in a pattern of dissemination much like that of ovarian cancer. Overall, of the clinical stage I tumors, 15% are limited to the endometrium, 45% penetrate the inner third of the myometrium, 20% reach the middle third of the myometrium, and only approximately 20% reach the subserosa. The extent of myometrial penetration correlates very well with the histologic grade of the tumor (Table 49-2). Peritoneal seeding can result from transmural spread of the tumor once the serosa is breached, or because of transtubal spillage of tumor cells into the general peritoneal cavity.

Table 49-2 Correlation of Tumor Grade with Extent of Myometrial Invasion

The endometrium proper has few lymphatics. However, once the tumor penetrates the myometrium, and especially when it reaches the lymphatic-rich subserosa, spread via lymphatic embolization is common. Overall, only approximately 11% of patients with clinical stage I and occult stage II endometrial carcinoma have pelvic or para-aortic lymph node involvement.⁸⁷ However, a number of adverse pathologic factors can affect this number, as shown in Table 49-3. There is a close correlation between the tumor grade and the depth of myometrial invasion, as discussed earlier. This risk of pelvic and para-aortic lymph node involvement as related to grade and depth of myometrial invasion is given in Table 49-4 and Table 49-5. Lymph node involvement also increases with increasing surgical stages. Thus, approximately 10% to 15% of all patients with stage I disease have pelvic lymph node involvement; this rises to 30% to 40% in stage II.

Table 49-3 Frequency of Nodal Metastases as a Function of Various Risk Factors

Table 49-4 Grade, Depth of Invasion, and Pelvic Node Involvement

Table 49-5 Grade, Depth of Invasion, and Aortic Node Involvement

Lymph node involvement in endometrial cancer is of immense prognostic importance. Whereas stage I endometrial carcinoma is generally associated with a 5-year disease-free survival (DFS) rate in excess of 90%, patients with pelvic lymph node involvement have a 5-year DFS rate of only 65% to 70%, and the minority with para-aortic lymph node involvement have only a 30% to 35% 5-year DFS rate. Lymphatic spread is also believed to be responsible for involvement of the vagina, without contiguous cervical involvement (relapse in the distal vagina following a prior hysterectomy, without a vault recurrence) and for the isolated adnexal involvement often seen in stage IIIA disease.

Hematogenous spread is infrequent in endometrial cancer, occurring mainly in high-grade tumors or tumors with an unfavorable histologic finding (e.g., serous adenocarcinoma). The common sites of metastases include the lungs, liver, and bones.

Clinical Presentation

Early endometrial cancer has few symptoms, and a high index of suspicion is necessary to avoid any delay in diagnosis. The perimenopausal phase has only one normal pattern: menstrual periods that get progressively lighter and further apart. Any variant of this should always sound a warning note in the treating clinician. Although postmenopausal bleeding can have several nonmalignant causes,⁸⁸ abnormal bleeding is the presenting symptom in more than 80% of cases of endometrial cancer. Another fairly frequent (10%), but rather nonspecific, symptom is that of leukorrhea. Pain or other symptoms of pelvic pressure are rarely present in early-stage disease.

Physical examination is also not helpful in the early stages, as uterine enlargement or extrauterine spread usually implies advanced disease. Any pyometra in a postmenopausal woman warrants an exclusion of an underlying endometrial cancer.

The diagnosis is usually established by endometrial sampling, usually via a fractional dilation and curettage (D&C) or an office biopsy. The first step in the D&C procedure is curettage of the endocervix. This is followed by sounding of the uterus, cervical dilation, and a systematic curettage of the entire endometrium. Cervical and endometrial specimens should be labeled and processed separately. The fractional D&C also provides an opportunity to examine the relaxed patient under anesthesia and allows for a cystoscopy or proctoscopy in the rare cases in which bladder or rectal involvement is suspected. Outpatient procedures, such as an endometrial biopsy or aspiration curettage with endocervical sampling, are diagnostic only if positive for cancer (10% are false-negative). If these are negative, a formal fractional D&C must be carried out.

Following histologic confirmation of the diagnosis, patients must undergo a thorough evaluation and staging workup. It is important to realize that these are often frail, elderly patients. They may have a number of concurrent medical problems that can affect the potential treatment-related morbidity and thus influence selection of the appropriate treatment modality. Routine blood and urine studies are obtained, as is a chest x-ray examination (to rule out metastases and evaluate the cardiopulmonary status of the patient). Other studies, including a cystoscopy, proctosigmoidoscopy, and barium enema may be required in selected patients in whom more extensive disease is suspected.

Several imaging studies are being evaluated for their ability to accurately define the disease extent preoperatively. The typical finding on ultrasonogram is a thickened endometrium. It is suggested that less-differentiated tumors may be hypoechoic ⁸⁹ or that distinct sonographic patterns allow one to distinguish among hyperplasia, polyps, and carcinoma.⁹⁰ In this respect, transvaginal ultrasonography (TVS) is believed to be superior to conventional transabdominal sonography.⁹¹ TVS is also useful in determining the depth of myometrial invasion.⁹² Its overall accuracy in determining invasion is 79% and its accuracy is in determining the degree of myometrial invasion is 60% to 76%. Good-quality pelvic computed tomography (CT) scans obtained with oral and intravenous contrast can demonstrate the extent of the endometrial tumor. The endometrial carcinoma is a hypodense mass relative to the normal myometrium⁹³ and may be seen as a diffuse, circumscribed vegetative or polypoidal mass within the uterine cavity. If myometrial invasion is seen, it usually implies involvement of greater than one third to one half of the myometrial thickness.⁹⁴ Endometrial involvement of the cervix is seen on CT as cervical enlargement greater than 3.5 cm in diameter with heterogeneous low-attenuation areas within the fibromuscular stroma.⁹⁵ Parametrial or sidewall extension is seen by the loss of periureteral fat in the former and less than 3 mm of intervening fat between the soft tissue mass and the pelvic sidewall in the latter.⁹⁶ Involvement of the fallopian tubes, ovaries, and lymph nodes is detected in the usual fashion.

Although the largest study looking at the ability of CT to detect myometrial invasion by dynamic and delayed images reported an overall accuracy of approximately 76%, the authors noted that CT did provide clinically useful information in almost 90% of cases.⁹⁴ The major limitation of CT is the failure to detect microscopic parametrial, lymph nodal, bladder, or rectal invasion.⁹⁵^,⁹⁷ Other problems include an unreliability in assessing myometrial invasion in the atrophic uteri of postmenopausal women and an occasional difficulty in differentiating a uterine from an adnexal mass.^93,^94,⁹⁷ Because of these problems, and because surgical staging is the standard approach in most patients, CT is not routinely recommended in clinical stage I or II disease. Its greatest effect may lie in clinical stage III patients, in whom it may confirm extrauterine spread, detect pelvic lymphadenopathy, or reclassify the disease as stage IV by detecting extrapelvic involvement.⁹⁵^,⁹⁷ It may also help in staging a patient with an equivocal pelvic examination or a contraindication to surgery and in confirming advanced stage III or IV disease.⁹³^–⁹⁵ ⁹⁷ CT scanning is also useful in the evaluation of recurrent disease by helping to define the extent of disease at relapse and in facilitating a directed biopsy for histologic confirmation of relapse.^95,^97,⁹⁸

Magnetic resonance imaging (MRI) is increasingly being recommended as part of the preoperative evaluation of endometrial carcinoma.⁹⁹^–¹⁰² Because unenhanced T1- and T2-weighted images lack specificity in diagnosing endometrial cancer, use of intravenous contrast is necessary.⁹¹^,¹⁰³ Following contrast injection, even small tumors contained within the endometrium can be diagnosed by their enhancement relative to the normal endometrium, and dynamic MRI with rapid administration of gadolinium diethylenetriaminepentaacetic acid using the turbo–fast, low-angle shot technique can delineate the inner myometrium from the outer layers even in postmenopausal women.

MRI scans have been correlated to the FIGO subdivisions of stage I endometrial cancer.¹⁰⁴ Stage IA tumors typically show a clear junctional zone or preservation of a sharp delineation between the tumor and the myometrium; IB disease is characterized by disruption of the junctional zone or increased signal-intensity tumor in the inner half of the myometrium with preservation of the outer myometrium, or both. Finally, in stage IC patients, there is extension of the high-signal-intensity tumor into the outer myometrium with preservation of a peripheral rim of normal, intact myometrium.¹⁰⁵ MRI also helps delineate stage II disease by showing an expansion of the cervical canal with stromal heterogeneity. Similarly, imaging of extrauterine spread is facilitated, although lymph node involvement is limited to the criterion of increased size. State-of-the-art MRI using gadolinium chelates is probably accurate in more than 90% of cases.¹⁰⁶^,¹⁰⁷ It definitely outperforms CT in determining myometrial invasion^103,^107,¹⁰⁸ and is possibly better than TVS.¹⁰⁹^,¹¹⁰

Tumor markers such as CA-125 have been shown to be elevated in 63% to 67% of patients with advanced or metastatic disease compared with 10% to 19% in those with early-stage disease. Whether CA-125 is a useful tool in screening for endometrial cancer is not determined, but most authors agree about its potential in patients with aggressive histologic findings, advanced disease, or pretreatment elevated levels.¹¹¹^,¹¹²

Staging and Prognostic Factors

The Gynecologic Oncology Group (GOG) conducted two large prospective trials ⁸⁷^,¹¹³ in 1984 and 1987. The major prognostic factors in endometrial cancer were defined from these studies, and surgical staging has gradually become the standard in this disease. This surgical approach led FIGO, in 1988, to design a surgical staging scheme for endometrial cancer, incorporating most of the known prognostic factors (Table 49-6).¹¹⁴ For the small minority of patients who are not surgical candidates and are treated with primary radiotherapy, the earlier clinical staging system is still applied (Table 49-7).¹¹⁵

Table 49-6 Corpus Cancer Surgical Staging (FIGO 1988)

Stage		Definition
IA	G123	Tumor limited to the endometrium
IB	G123	Invasion to <50% of the myometrium
IC	G123	Invasion to >50% of the myometrium
IIA	G123	Cervical involvement restricted to the endocervical glands
IIB	G123	Cervical stromal invasion
IIIA	G123	Tumor invades uterine serosa, or adnexae, or positive peritoneal cytology
IIIB	G123	Tumor involving the vagina
IIIC	G123	Pelvic or para-aortic lymph nodal involvement
IVA	G123	Invasion of bladder, bowel mucosa, or both
IVB		Distant metastases, including intra-abdominal spread or inguinal lymph nodes

FIGO, International Federation of Gynecology and Obstetrics.

Table 49-7 Clinical Staging of Endometrial Cancer (FIGO 1971)

Stage	Definition
I	Carcinoma confined to the uterus
IA	Length of uterine canal is ≤8 cm
IB	Length of uterine canal is >8 cm
Histologic Subtypes of Adenocarcinoma
G1	Highly differentiated adenomatous carcinoma
G2	Differentiated adenomatous carcinoma with partly solid areas
G3	Predominantly solid or entirely undifferentiated carcinoma
II	Carcinoma involves the corpus and the cervix
III	Carcinoma extends outside the uterus but is confined to the true pelvis
IV	Carcinoma extends outside the true pelvis or involves the bladder, rectum, or both

FIGO, International Federation of Gynecology and Obstetrics.

Several clinicopathologic factors have been identified in patients with endometrial carcinoma to help predict the prognosis and individualize the treatment plan. Based on clinical experiences, Nolan and Huen ¹¹⁶ developed a mathematical model for prognostication, wherein patient prognosis is directly proportional to host resistance, inversely related to tumor virulence, and positively affected by appropriate treatment factors.

Host Resistance Factors

Age

The importance of age to endometrial cancer has been well documented. Frick and associates,¹¹⁷ in 1973, reported a better outcome for patients younger than the age of 59 with stage I endometrial cancer, even after correcting for comorbid medical conditions. This was subsequently confirmed by other investigators.¹¹⁸^–¹²⁰ The adverse effect of advanced age on outcome persists even when elderly patients are treated as aggressively as their younger counterparts.¹²¹ The improved survival in younger patients may reflect the tendency toward earlier, better-differentiated tumors with less myometrial invasion in these patients; in older postmenopausal women with atrophic uteri, relatively little myometrial penetration brings the tumor close to the lymphatic-rich subserosa, with a correspondingly high risk of lymph node metastases.¹²² Better host immunocompetence may be an additional factor in the younger patients.

Race

Caucasian women tend to fare better than African Americans independent of other prognostic factors.¹²³ It is important to note that although the prevalence of endometrial cancer is lower in African American women, the incidence of high-risk tumors in this group is higher.¹²⁴ The influence of histologic type and differentiation on outcome was discussed earlier.

Performance Status and Anemia

A poor performance status and anemia, which may predict a poorer response to radiotherapy, are other adverse, host-related “resistance” factors.

Tumor Virulence Factors

Stage at Diagnosis

Because the very concept of stage grouping is based on creating groups with similar prognoses, it is not surprising that the outcome gets less favorable with advancing stage. Reported results of stage-based survival are remarkably consistent, with stages I, II, III, and IV disease having 5-year survival rates of 92%, 80%, 40%, and 5% respectively.¹²⁵

Size of the Uterus

Uterus size has long been recognized as a predictor of survival and was incorporated in the earlier clinical staging for endometrial cancer. Creasman and associates showed a direct correlation between uterine size and the incidence of para-aortic and pelvic nodal metastases.⁸⁷ It has been pointed out, however, that uterine size may not always reflect tumor volume, because a number of benign conditions can cause enlargement of the uterus.¹²⁶ Furthermore, because the size of the uterus correlates with tumor grade and myometrial penetration, it is very possible that size merely acts as a surrogate for these far more significant prognostic factors.

Lower Uterine Segment Involvement

Although 92% of stage I endometrial carcinomas lie in the fundus, it has been proposed that cancers involving the isthmus or lower uterine segment may have a worse prognosis.¹²⁷ The GOG study of surgical-pathologic spread patterns found a doubling of the incidence of pelvic nodal involvement from 8% to 16% and an increase in para-aortic nodal involvement from 4% to 14% when the tumor arose from or involved the isthmus.⁸⁷ Mayr and associates¹²⁸ found that lower uterine segment involvement, in the absence of any other adverse prognostic factors, resulted in an almost 50% local failure rate when postoperative radiotherapy was omitted; in patients receiving postoperative radiotherapy, the relapse rates dropped to 3%.

Cervical Involvement

It is important to separate pathologic (occult) from clinical involvement of the cervix in endometrial cancer because only occult cervical involvement is recognized in the 1988 FIGO staging system. Prognosis for patients with cervical involvement (stage II) is much poorer than for those with earlier-stage lesions. Surwit and associates ¹²² reported that cervical stromal involvement dropped the survival rate to less than 50%, as compared with almost 75% with tumor involvement limited to the endocervical glands; access to the cervical stroma probably increases the propensity for lymphatic dissemination, as positive pelvic lymph nodes are seen in approximately 35% of the patients.¹²⁹ Similarly, Rubin and associates reported on the Memorial Sloan-Kettering Cancer Center experience with stage II endometrial cancer.¹³⁰ At 60 months, the actuarial disease-free survival rate was 39% in patients with gross cervical involvement, 62% in those with clinically occult stromal involvement (detected only on a fractional D&C), and 88% in the subset having only detached tumor fragments on fractional D&C.

Histologic Subtypes

Although the most common subtype—endometrioid adenocarcinoma—has a worse prognosis with increasing grade,⁸⁷^,¹³¹ there are cell types that have a high propensity to early systemic dissemination with a uniformly bad prognosis, regardless of the pathologic grade. These include the serous, clear cell, squamous, and undifferentiated subtypes. The overall survival rate in this group is less than 35%.¹³²^,¹³³

Tumor Grade

Tumor grade is one of the most sensitive indicators of prognosis. Grade directly affects the depth of myometrial penetration and the frequency of lymph node involvement (see Table 49-3).^87,^131,^134,¹³⁵ Thus, most grade 1 tumors are limited to the endometrium or have superficial myometrial penetration, and the overall risk of pelvic and para-aortic lymph nodal metastases is 3% and 1.5%, respectively; however, 10% of grade 1 tumors have deep myometrial penetration, and the pelvic and para-aortic lymph nodal involvement in these are 12% and 6%, respectively. Similarly, more than 50% of grade 3 tumors have a greater than 50% myometrial penetration, and these have pelvic and para-aortic nodal involvement on the order of 30% and 20%, respectively.⁸⁷

Myometrial Invasion

Myometrial invasion is probably the factor most consistently identified with treatment failures. Conventionally, the depth of invasion has been reported as inner, middle, or outer thirds of the myometrium. The FIGO staging system, however, refers only to inner and outer halves, and the pathologic reporting should change to incorporate this recommendation. In patients with stage IB disease (<50% invasion), it seems that invasion to less than one-third versus greater than one-third is not a significant predictor of outcome.¹³⁶ Lutz and associates also noted that proximity to the serosa may be more important than absolute depth of penetration.¹²⁶ They found a 65% survival rate for patients whose tumors had invaded to within 5 mm of the serosa, compared with a 97% survival rate for patients with tumors more than 10 mm from the serosa. Both grade and depth of penetration are predictive for lymph nodal involvement; however, depth is more influential in determining lymphatic and extrauterine spread, treatment failure, and relapse. Thus, regardless of grade, only 1% of tumors limited to the endometrium had lymph nodal involvement, as compared with 25% pelvic and 17% para-aortic involvement with deep penetration.⁸⁷

Vascular or Capillary-Like Space Invasion

Vascular or capillary-like space (CLS) invasion is seen in approximately 15% of the cases of endometrial cancer.^87,^137,¹³⁸ The GOG study found that CLS-positive tumors were associated with a 27%, or fourfold, increase in the pelvic lymph nodal metastases, and a 19%, or sixfold, increase in para-aortic nodal metastases.⁸⁷ This translates into more frequent relapses and a poorer survival for the CLS-positive patients.¹³⁷

Adnexal Involvement

Approximately 5% of patients with stage I and occult stage II disease have adnexal involvement.⁸⁷ This is associated with a fourfold increase in lymph node metastases; thus, pelvic lymph nodal positivity rises to 32% (as compared with 8% without adnexal spread) and para-aortic nodal involvement is seen in 20% (as opposed to only 5% in patients with no adnexal spread). The depth of myometrial invasion is still an important factor: Patients with adnexal spread but less than 50% myometrial involvement have a 5-year survival rate of approximately 80%, as compared with a 45% 5-year survival rate when adnexal involvement is associated with deep myometrial invasion. The extent of adnexal involvement is also of prognostic significance. Patients with occult, microscopic invasion do far better (5-year survival rate >80%) than those with gross adnexal involvement (5-year survival rate <40%).¹³⁹

Peritoneal Cytologic Findings

Peritoneal fluid positive for malignant cells is found in 12% to 15% of all patients undergoing surgical staging. This is associated with 25% pelvic lymph nodal involvement and 19% para-aortic nodal involvement. However, only approximately 5% of patients with positive peritoneal cytologic finding have no extrauterine disease,⁸⁷ and the literature regarding the true effect of positive peritoneal cytology is mixed. Several small series indicate that a positive peritoneal cytologic finding has no independent prognostic value.¹⁴⁰^–¹⁴² On the other hand, two relatively large series show a distinct adverse effect from positive peritoneal cytology.¹⁴³^,¹⁴⁴ One confounding factor, mainly in patients with no other risk factors, is whether all endometrial cancer cells that gained access to the peritoneal cavity are capable of independent growth. The meta-analysis by Milosevic and associates concluded that, although malignant cytologic findings were strongly associated with disease recurrence (odds ratio of 4.7 with 95% CI of 3.5-6.3), this was largely due to the association of malignant cytologic findings with other adverse prognostic factors, which dominate the clinical presentation.¹⁴⁵ In the absence of any other adverse features, the true incidence of malignant cytologic findings is quite low and is unlikely to act as an independent predictor for relapse. Recent data suggest a higher rate of positive cytologic findings for patients undergoing laparoscopic-vaginal hysterectomy, in which there is manipulation of the uterine cavity, compared with TAH, in which there is no such manipulation.¹⁴⁶

Serosal Involvement

This is the least-studied feature of all stage III-A subsets. Its incidence ranges from 7% to 16% and it is often associated with other poor prognostic factors.¹⁴⁷ But even when it is isolated, the 5-year disease-free survival is only 41.5%.

Gross Intraperitoneal Spread

This again correlates very highly with lymph nodal metastases. Pelvic lymph nodal positivity increased to 51% in the presence of gross intraperitoneal spread as compared with only 7% without such dissemination; similarly, para-aortic nodal involvement rises to 23% with intraperitoneal spread, as compared with only 4% without this spread.

Pelvic and Para-Aortic Lymph Nodes

Overall, approximately 11% of patients with stage I and occult stage II endometrial cancer have pelvic nodal involvement.⁸⁷ This increases to 25%, 30%, and 50% with deep myometrial penetration, adnexal involvement, and extrauterine spread, respectively. Lymph nodal involvement is a major predictor for distant disease: the 5-year disease-free survival rates drop to 65% to 70% in patients with pelvic lymph nodal involvement as their only risk factor.¹⁴⁸ Similarly, although para-aortic nodal metastases are seen in approximately 5% of all patients with stage I and occult stage II disease, the biggest risk factor for para-aortic node involvement is the presence of pelvic nodal metastases; more than 30% of patients with pelvic nodal involvement have para-aortic disease. The pattern of lymphatic spread in endometrial cancer is different than in cervical cancer. In endometrial cancer, a simultaneous spread to both pelvic and para-aortic nodes could occur, whereas in cervical cancer the spread to para-aortic nodes is almost always secondary to pelvic lymph nodal involvement. Despite optimal debulking and radiation, 5-year DFS rates drop to approximately 30% in this subpopulation.