Semere, Luwam G. MD; Ko, Emily MD; Johnson, Natasha R. MD; Vitonis, Allison F. SM; Phang, Laura J. MD; Cramer, Daniel W. MD; Mutter, George L. MD
Endometrial cancer is the most common gynecologic malignancy in the United States, with an estimated 40,100 new cases diagnosed annually and 7,470 deaths occurring each year.1 The histopathologic diagnosis of premalignant lesions of endometrioid endometrial cancer has been a source of dispute among pathologists.2 The four classes of World Health Organization endometrial hyperplasia do not correspond to separate biologic entities, fail to incorporate diagnostic advances achieved in the past 15 years,3 and are poorly reproducible. The alternative endometrial intraepithelial neoplasia (EIN) schema is a two-class, functionally defined (benign endometrial hyperplasia [a hormonal effect] and EIN [a premalignant lesion]) diagnostic system that introduced several newly discovered histologic criteria associated with heightened cancer risk, such as minimum lesion size, precise extent of gland crowding, and an internal comparison standard (background compared with lesion) for interpretation of significant cytologic change.4 This has proven to be a better predictor of disease progression and, equally important, is better able to determine which lesions are likely to remain benign.5 There is no fixed concordance between World Health Organization hyperplasia and EIN schema diagnoses because of differing diagnostic criteria.6
Endometrial intraepithelial neoplasia diagnostic practices only recently have been deployed in routine clinical environments (2001, at our institution) and, as a result, little has been published regarding care of patients with EIN. The objective of this study was to estimate endometrial cancer outcomes among a series of sequentially diagnosed cases of EIN in women in a tertiary care multigroup practice. In addition, we sought to estimate if there were patient demographic, clinical, or treatment modality characteristics that influenced the outcomes of EIN involution to a benign histology compared with persistence, or progression to endometrial cancer.
MATERIALS AND METHODS
This study received approval from the Partners Human Research Committee, the Institutional Review Board for Brigham and Women's Hospital. From June 2002 until July 2006, patients receiving a diagnosis of EIN on endometrial sampling at Brigham and Women's Hospital and its community affiliates were identified through pathology record review. This sample included patients from the gynecology resident continuity practice, faculty clinics, gynecologic oncology clinics, and community health centers and satellite practices. Medical records were obtained from the Partners Health Care electronic medical record system and from paper charts. Demographic data, obstetric history, gynecologic history, medical history, surgical history, operative information, and pathology results were collected from this medical record review.
Inclusion in the study was based on a first pathologic report diagnosis of EIN within an endometrial sample (biopsy or curetting, designated here as an “index biopsy”). Women with previous EIN or endometrial carcinoma diagnoses were excluded, as were those in whom the index biopsy histologic sections were unavailable for review. The routine pathology interpretations were reported by 11 pathologists at Brigham and Women's Hospital who randomly encountered the specimens as part of rotating service coverage. Diagnosis of EIN required the following: 1) glandular volume exceeding that of stroma; 2) cytologic demarcation from surrounding normal glands; 3) lesions with largest diameter greater than 1 mm; 4) exclusion of confounding benign processes such as degenerative or hormonal effects; and 5) exclusion of carcinoma.8
Glass histologic sections of the index biopsy samples were re-reviewed by a single pathologist (G.L.M.) for presence or absence of EIN and presence or absence of concurrent carcinoma in the same specimen. Because management was based on the pathology report diagnosis, we did not exclude those patients in whom the diagnosis of EIN was not upheld on re-evaluation of available slides. Follow-up endometrial pathology specimens (obtained after the index EIN biopsy date) varied in number and format (biopsy, curetting, hysterectomy) between patients. Follow-up endometrial specimen pathology reports were reviewed and the outcome was censored at first occurrence of carcinoma or last pathology specimen diagnosis. Some women received progestin treatment after initial EIN diagnosis, and these were recorded for analysis.
A retrospective cohort was constructed from those women with cancer-free EIN at initial diagnosis to estimate endometrial cancer outcomes over time. Kaplan-Meier survivor analysis was used to determine the proportion of cancer-free patients during the available follow-up period. Cancer outcomes were then compared for those women who were, or were not, treated after diagnosis with progestins. Outcomes were categorized as cancer if any single follow-up specimen had this diagnosis. For those who did not progress to carcinoma, the outcome was based on pathology seen in the last available endometrial pathology specimen.
A case-control analysis was then conducted comparing clinical characteristics of women whose EIN regressed (control group) to two separate case groups: women with persistence of EIN on subsequent sampling or hysterectomy and women with a diagnosis of cancer either on the index biopsy or during follow-up. All women in the control group had either definitive hysterectomy or more than 18 months of follow-up to accurately reflect clinical practice outcomes. Categorical variables were compared using χ2 and Fisher exact test, and continuous variables were compared using t tests. Logistic regression analysis was used to calculate age-adjusted odds ratios (ORs). Statistical significance was defined as P<.05. SAS 9.1 statistical analysis software was used to perform all analyses.
One hundred seventy-seven new cases of EIN were identified by endometrial sampling from January 2002 through July 2006. All hemotoxylin and eosin slides were reviewed by a single pathologist (G.L.M.), with a diagnosis of EIN confirmed in 94% of the cases. Patients had an average age of 53 years, with a median of 53 years, and range from 26 to 94 years. Self-reported ethnicity was available for 103 patients. Of these, 81 (78.6%, range 69.4–86.1%) were white, eight (7.8%, 95% CI 3.4–14.7%) were Hispanic, 10 (9.7%, 95% CI 4.7–17.1%) were black, and four (3.9%, 95% CI 1.1–9.6%) were Asian. Figure 1 demonstrates the outcomes of all patients in the study. In 11.9% (21 of 177, 95% CI 7.5−17.6%) of patients, we had no information regarding follow-up endometrial sampling or hysterectomy, one of whom had a concurrent carcinoma on the index Biopsy. 82.7% (129 of 156, 95% CI 75.8−88.3%) of patients with follow-up eventually had a hysterectomy, of which the majority (91 of 129 [70.5%], 95% CI 61.9−78.2%) were within 3 months of the index biopsy. 17.3% (27 of 156, 95% CI 11.7−24.2%) of patients with available follow-up only had biopsies performed.
Overall cancer incidence was 35.7% (56 of 157, 95% CI 28.2–43.7%) diagnosed at various stages of patient management. First diagnosis of carcinoma was most common within the initial EIN-bearing biopsy sample (26 patients), but others occurred during follow-up biopsy sampling (8 patients) or at definitive hysterectomy (22 patients). Patients with EIN who remained cancer-free during follow-up (102 patients) demonstrated equal frequencies of EIN persistence 50% (51 of 102, 95% CI 40.0–60.1%) and “regression” to a benign histology (51 of 102).
All (56 of 56) of the endometrial cancers were of the endometrioid (type I) type. Most (46 of 54 [85%] 95% CI 72.9–93.4%) were well-differentiated, whereas 11% (6 of 54, 95% CI 4.2–22.6%) were moderately differentiated and 4% (2 of 54, 95% CI 0.4–12.7%) were poorly differentiated. Grading of two cases was unavailable from the pathology report. 28% (15 of 54, 95% CI 16.4–41.6%) of cancers invaded the myometrium. Of these, 87% (13 of 15, 95% CI 59.5%–98.3%) were confined to the inner half of the myometrial thickness. All cases of lymph node sampling (8 of 8, 95% CI 63.1–100%) had cancer-free nodes, and only one (1 of 54, 95% CI 0.05–9.9%) had myometrial lymphovascular invasion. There were two cases of disease spread beyond the uterine corpus, one metastasis to the ovarian surface and one with local extension to the endocervical stroma.
Twenty-one percent (32 of 151, 95% CI 15.0–28.6%) of the women in our study were treated medically with progestins (excluding 26 with unknown treatment status); all were cancer-free during the index biopsy. Cancer outcomes were not significantly different in women treated (7 of 32 [22%], 95% CI 9.3–40.0%) compared with those not treated (47 of 119 [39%], 95% CI 30.6–48.9%) with progesterone (P=.095). Women receiving medical therapy by progesterone were significantly younger than those not receiving progesterone (mean age 48.4 compared with 54.0 years, P=.012) and were significantly less likely to be parous than women not receiving progesterone therapy (13 of 30 [43.3%], 95% CI 25.5–62.6%) compared with 69.6% (80 of 115, 95% CI 60.3–77.8%, P=.01). Among women who remained cancer-free, the rate of EIN involution as seen by reversion to a benign histology was significantly greater in women treated with progesterone compared with those untreated (13 of 16 [81%], 95% CI 54.4–96.0% compared with 20 of 63 [32%], 95% CI 20.6–44.7%, P=.005).
The retrospective cohort analysis was based on 131 patients with a diagnosis of EIN only on initial sampling and available follow-up. These were variably managed by immediate hysterectomy (79 of 131 [60%], 95% CI 51.4–68.7%), biopsy surveillance only (26 of 131 [20%], 95% CI 13.4–27.7%), or biopsy surveillance followed by hysterectomy (26 of 131 [20%], 95% CI 13.4–27.7%). The pathology outcomes of all 131 patients were 23% (30 of 131, 95% CI 16.0–31.0%) endometrial cancer, 39% (51 of 131, 95% CI 30.5–47.8%) persistent EIN, and 38% (50 of 131, 95% CI 29.8–47.1%) benign pathology.
Figure 2 demonstrates the proportion of cancer-free patients during the follow-up period (median 74 days, mean 269 days, and standard deviation 462 days). Interval from index biopsy with EIN to diagnosis of cancer was a median of 56 days (mean 152 days, standard deviation 231 days). Of the 30 women who had progression to cancer, 83% (25 of 30, 95% CI 65.3–94.4%) received diagnoses within 1 year and 17% (5 of 30, 95% CI 5.6–34.7%) received diagnoses after 1 year.
For the case-control analysis, women who did not undergo definitive hysterectomy or who had less than 18 months of follow-up (n=35) were excluded, leaving a total of 142 in the case and control groups. Demographic variables including age (P=.09), ethnicity (P=.19), marital status (P=.51), insurance status (P=.10), and indications for initial endometrial sampling (P=.15) appeared to be similar (Fisher exact) between those with follow-up and those without (Table 5, available online at http://links.lww.com/AOG/A245). For those women with adequate follow-up, demographic, clinical, and reproductive characteristics of women with involuted EIN (control group n=36 [36 of 142], 25%, 95% CI 18.4–33.3%) were compared with those women with EIN persistence (case group n=50 [50 of 142], 35%, 95% CI 27.4–43.7%) or EIN progression (concurrent or future cancer, case group n=56 [56 of 142], 39%, 95% CI 31.3–48.0%).
Table 1 demonstrates the demographic characteristics of the patients in relation to their outcome diagnosis. Women who had benign pathology did not differ significantly regarding age, marital status, or insurance status from persistent EIN or cancer cases. Significantly fewer women who had cancer diagnosed were of nonwhite ethnicity (OR 0.16, 95% confidence interval [CI] 0.03–0.84). “Other” indications for initial biopsy include endometrial cells on Pap test (8 patients), atypical cells on Pap test (1 patient), thickened endometrial lining on ultrasonography (5 patients), endometrial lesion on imaging (2 patients), evaluation for infertility (2 patients), polyps seen on hysterosalpingogram during evaluation for infertility (1 patient), Lynch syndrome (1 patient), tamoxifen use (1 patient), and history of hyperplasia on hormone replacement therapy (1 patient).
Clinical variables are shown in Table 2 for each outcome group. Diabetes, hypertension, smoking status, alcohol use, family history of any type of malignancy, and personal history of any previous nonendometrial malignancy did not differ significantly. Body mass index 25 or higher was significantly associated with increased cancer outcomes (OR 3.05, 95% CI 1.10–8.45).
Table 3 compares reproductive history between those in the case and control groups with no significant differences found between the three groups for gravidity, parity, previous tubal ligation, postmenopausal status, postmenopausal bleeding, irregular (premenopausal) menses, polycystic ovarian syndrome diagnosis, ovarian cysts requiring surgery, previous intrauterine device use, and EIN presentation during infertility work-up. Among four women pursuing infertility treatment at the time of EIN diagnosis, one had EIN involution and three had cancer outcomes (OR 2.81, 95% CI 0.25–31.5).
Exogenous hormone use is illustrated in Table 4. Use of oral contraceptives, hormone replacement therapy, tamoxifen, and prior progestin for any indication did not differ significantly between patients with benign outcomes when compared with those with EIN persistence or progression to carcinoma. Progestin use specifically for the treatment of EIN did result in a decreased risk of EIN persistence (OR 0.11, 95% CI 0.03–0.42) and progression (OR 0.24, 95% CI 0.08–0.70).
Diagnostic classification of premalignant endometrial lesions is now in a state of transition from the legacy four-class World Health Organization 1994 hyperplasia schema (hyperplasia with or without atypia, complex or simple architecture) to a two-class EIN schema (benign endometrial hyperplasia or EIN).8 Advantages of the EIN system are better-defined histopathologic diagnostic criteria, superior clinical outcome prediction, and lucid communication for each diagnosis of the disease process (hormonal or premalignant).4,7 By entering information of all sequential patients with a first diagnosis of EIN within a defined practice environment, we have minimized case selection bias and are able to generate a practitioner's view of current management practices, clinical outcomes, and demographics of affected patients using the EIN diagnostic approach. Pathologic diagnosis of EIN was highly consistent among pathologists, with the initial diagnoses (made by a pool of 11 surgical pathologists) confirmed on central review in 94% of cases.
Endometrial intraepithelial neoplasia is a high risk factor for malignancy, with 35.7% (56 of 157, 95% CI 28.2–43.7%) overall having carcinoma at initial diagnosis or during followup. Among women diagnosed with EIN 15% (26 of 177, 95% CI 9.8–20.8%) had concurrent cancer in the presenting biopsy, 19% (25 of 131, 95% CI 12.7–26.9%) had cancer develop within 1 year, and an additional 4% (5 of 131, 95% CI 1.2–8.7%) had cancer develop after 1 year. This compares with a 37.8% (56 of 148) synchronous cancer rate in women with EIN undergoing immediate hysterectomy9 and a 41% cancer rate during 1 year of clinical follow-up.7 Previous estimates of concurrent endometrial cancer at the time of diagnosis of atypical endometrial hyperplasia are similar, at 43% for immediate hysterectomy.10 Short follow-up interval in our series precluded estimates of additional longer-term progression events to carcinoma, but other studies have shown that cancer occurrences beyond 1 year of EIN diagnosis are 45-times more likely than EIN-free women, occurring in 18% of patients at a median interval of 4 years (average 4.1 years).7 Although we have used the term cancer “progression” throughout for cancers diagnosed after the initial EIN-bearing biopsy, some unknown proportion of these represent occult carcinomas present in the patient, but missed by biopsy, at the time of initial EIN diagnosis.
Our case-control analysis showed that overweight and obese women with EIN (body mass index 25 or higher) were at 3.05-fold increased risk for development of endometrial cancer compared with nonobese patients with EIN. This is consistent with previous epidemiologic data in which obesity is a risk factor for endometrial carcinoma, perhaps mediated by changes in endogenous steroid hormone metabolism.11,12 We did not have sufficient numbers of morbidly obese (BMI more than 38) patients to determine whether the risk effect is proportionately scaled across the full range of body weights or if it is a discrete threshold effect.
Progestin treatment of EIN, most common in younger nulliparous patients desiring to maintain fertility, was utilized in 21% (32 of 151, 95% CI 15.0–28.6%) of our study population and associated with decreased risk for EIN persistence (OR 0.11) or progression to carcinoma (OR 0.24). This coincides with a developing experience showing that progestin therapy may, in some cases, be effective in nonsurgical treatment of endometrial precancers13,14 or even carcinoma.15 In general, however, accurate estimates of efficacy are severely limited by lack of standardization of therapeutic regimen (agent, dosage, and administration schedule) in addition to uncertainty of the best outcome measures (number of biopsies, surveillance duration, and sampling during or after treatment). Long-term failures require more protracted clinical follow-up than generally available, and there are limitations to pathologic interpretation of residual disease in an endometrial sample obtained under the influence of active progestational therapy that dramatically alters lesion cytology and architecture.15,16 Thus, some of the “benign” outcomes may be occult EIN persistence missed by the pathologist, and others may be a transient shrinkage of EIN lesions rather than long-term cure. Our own study is constrained by these factors and is best interpreted as evidence of a short-term therapeutic response rather than a permanent “cure.” More studies, preferably with multiple surveillance biopsies after completion of progestin therapy, need to be performed to determine the long-term natural history of these patients.
Endometrioid carcinoma and EIN have a common pathogenesis and thus share many overlapping risk factors that require large numbers of patients to resolve separately. Our study sample size may have limited our power to detect risk factors specific to EIN that predict cancer outcomes. Furthermore, human studies constraints in this retrospective context prohibited contacting patients and, thus, we were unable to utilize standardized exposure questionnaires.
All cancer outcomes seen in our study were of an endometrioid histologic type, and were usually well-differentiated with superficial or no myometrial invasion. However, there were two patients with more than 50% myometrial invasion, and another two patients had evidence of disease beyond the uterine corpus. Careful clinical evaluation of the possibility of concurrent carcinoma is advised in all new diagnoses of EIN. In those cases in which surgical management is the treatment chosen, examination of the hysterectomy specimen will perform this function.
Our experience with EIN is that it is a reproducible and specific diagnosis that allows us to identify those premalignant endometrial lesions that place a patient at risk for carcinoma. Management of EIN lesions should consider the heightened cancer risk that diagnosis confers and should generally follow those guidelines previously established for atypical endometrial hyperplasia. Given the high likelihood of concurrent malignancy, women identified as having EIN lesions warrant close follow-up and consideration of treatment with hysterectomy. Younger women wishing to retain fertility are, in some cases, being managed with progestin therapy, and we did find this may promote involution of EIN to a benign histology and reduce the progression likelihood to carcinoma. Despite these promising data, lack of therapeutic standardization, defined end points, and prospective blinded measures of treatment efficacy are limitations to assessment of the risks and benefits of this approach.
1. Ries LAG, Melbert D, Krapcho M, Stinchcomb DG, Howlader N, Horner MJ, et al (editors). SEER cancer statistics review, 1975–2005. Bethesda (MD): National Cancer Institute. Available at: http://seer.cancer.gov/csr/1975_2008/
. Retrieved May 11, 2011.
2. Mutter GL, Matias-Guiu X, Lax SF. Endometrial Adenocarcinoma. In: Robbboy SJ, Mutter GL, Prat J, Bentley RC, Russell P, Anderson MC, editors. Robboy's pathology of the female reproductive tract. 2nd ed. New York (NY): Elsevier; 2009. p. 393–426.
3. Bokhman JV. Two pathogenetic types of endometrial carcinoma. Gynecol Oncol 1983;15:10–7.
4. Hecht JL, Ince TA, Baak JP, Baker HE, Ogden MW, Mutter GL. Prediction of endometrial carcinoma by subjective endometrial intraepithelial neoplasia diagnosis. Mod Pathol 2005;18:324–30.
5. Mutter GL. Histopathology of genetically defined endometrial precancers. Int J Gynecol Pathol 2000;19:301–9.
6. Mutter GL. Endometrial intraepithelial neoplasia (EIN): will it bring order to chaos? The Endometrial Collaborative Group. Gynecol Oncol 2000;76:287–90.
7. Baak JP, Mutter GL, Robboy S, van Diest PJ, Uyterlinde AM, Orbo A, et al. The molecular genetics and morphometry-based endometrial intraepithelial neoplasia classification system predicts disease progression in endometrial hyperplasia more accurately than the 1994 World Health Organization classification system. Cancer 2005;103:2304–12.
8. Mutter GL, Zaino RJ, Baak JP, Bentley RC, Robboy SJ. The benign endometrial hyperplasia sequence and endometrial intraepithelial neoplasia. Int J Gynecol Pathol 2007;26:103–14.
9. Mutter GL, Kauderer J, Baak JP, Alberts D, Gynecologic Oncology Group. Biopsy histomorphometry predicts uterine myoinvasion by endometrial carcinoma: a Gynecologic Oncology Group study. Hum Pathol 2008;39:866–74.
10. Trimble CL, Kauderer J, Zaino R, Silverberg S, Lim PC, Burke JJ, et al. Concurrent endometrial carcinoma in women with a biopsy diagnosis of atypical endometrial hyperplasia: a Gynecologic Oncology Group study. Cancer 2006;106:812–9.
11. Weiderpass E, Adami HO, Baron JA, Magnusson C, Bergstrom R, Lindgren A, et al. Risk of endometrial cancer following estrogen replacement with and without progestins. J Natl Cancer Inst 1999;91:1131–7.
12. Parazzini F, La Vecchia C, Bocciolone L, Franceschi S. The epidemiology of endometrial cancer. Gynecol Oncol 1991;41:1–16.
13. Signorelli M, Caspani G, Bonazzi C, Chiappa V, Perego P, Mangioni C. Fertility-sparing treatment in young women with endometrial cancer or atypical complex hyperplasia: a prospective single-institution experience of 21 cases. BJOG 2009;116:114–8.
14. Varma R, Soneja H, Bhatia K, Ganesan R, Rollason T, Clark TJ, et al. The effectiveness of a levonorgestrel-releasing intrauterine system (LNG-IUS) in the treatment of endometrial hyperplasia–a long-term follow-up study. Eur J Obstet Gynecol Reprod Biol 2008;139:169–75.
15. Wheeler DT, Bristow RE, Kurman RJ. Histologic alterations in endometrial hyperplasia and well-differentiated carcinoma treated with progestins. Am J Surg Pathol 2007;31:988–98.
16. Lin MC, Lomo L, Baak JPA, Eng C, Ince TA, Crum CP, et al. Squamous morules are functionally inert elements of premalignant endometrial neoplasia. Mod Path 2009;22: 167–74.
Figure. No caption available.