Objective
The objective of the study was to identify additional factors that may improve the ability to predict underlying carcinoma risk in patients with complex atypical hyperplasia (CAH) of the uterus.
Study Design
All subjects diagnosed with CAH of the uterus on endometrial sampling from March 1994 to May 2008 were identified. CAH was classified as CAH suspicious, CAH polypoid, CAH focal, or CAH not otherwise specified (NOS). Subjects were then exclusively assigned to 1 of 3 categories: CAH suspicious, CAH NOS, or CAH focal and/or polypoid.
Results
We identified 197 cases of CAH diagnosed on preoperative endometrial sampling. Carcinoma was subsequently diagnosed in the hysterectomy specimen in 67 subjects (34%). The risk of underlying carcinoma if assigning subjects as CAH suspicious, CAH NOS, or CAH polypoid and/or focal was 56%, 36%, and 20%, respectively ( P < .001).
Conclusion
The risk of underlying carcinoma in patients with CAH on preoperative endometrial sampling is associated with the method of sampling and age and can be significantly modified by the nature of pathologic assessment.
The International Society of Gynecological Pathologists classifies endometrial hyperplasias into 4 groups according to their likelihood of progressing to carcinoma. The 4 classifications (simple hyperplasia, complex hyperplasia, simple hyperplasia with atypia, and complex hyperplasia with atypia) are differentiated based on architectural complexity and the presence of cytologic atypia. It is often difficult to distinguish complex atypical hyperplasia (CAH) from well-differentiated carcinoma, but it is important to do so on final hysterectomy because their outcomes vary when myometrial invasion is present. Distinguishing the 2 is more challenging when attempting to make the distinction via endometrial biopsy alone.
CAH is generally considered a precursor lesion to endometrial carcinoma. Kurman et al retrospectively reviewed 170 patients with any degree of endometrial hyperplasia who were not treated for at least 1 year. The risk of progression to carcinoma was 1% for simple hyperplasia, 3% for complex hyperplasia, 8% for simple atypical hyperplasia, and 29% for CAH.
There is a definite risk of underlying occult carcinoma at the time of a hyperplasia diagnosis. The Gynecologic Oncology Group (GOG) conducted a large prospective trial (GOG 167) in patients with atypical hyperplasia of the endometrium. Patients diagnosed with atypical hyperplasia on a preoperative endometrial biopsy underwent an immediate hysterectomy, and the findings in the uterus were analyzed to determine the rate of underlying carcinoma. Carcinoma was diagnosed in the hysterectomy specimen in 123 of 289 cases (43%). The reproducibility of the referring institution’s pathologists’ diagnosis of atypical hyperplasia in the preoperative biopsy specimen by the study panel of pathologists was poor. The level of reproducibility among the subspecialist study panel was also poor.
These findings clearly highlight the need for improved reproducibility of the diagnosis of atypical hyperplasia in endometrial biopsies as well as an improvement in the ability to predict underlying carcinoma, which would effectively guide the clinical management of these patients.
Recent work has suggested that atypical hyperplasia of the endometrium can be further stratified into subcategories that would better distinguish atypical hyperplasias associated with underlying carcinoma. The sampling method also appears to have significance because atypical hyperplasia diagnosed in dilatation and curettage (D&C) specimens may be less strongly associated with carcinoma on follow-up compared with atypical hyperplasia diagnosed via office biopsy alone.
We reviewed our experience with patients who had undergone a hysterectomy at our institution for a preoperative diagnosis of CAH, based on endometrial sampling, and sought to determine whether we could identify additional factors that may improve our ability to predict the risk of underlying carcinoma. We also sought to determine the histologic features of the underlying carcinomas that were identified.
Materials and Methods
Following approval from our institutional review board, we identified all patients diagnosed with CAH on preoperative endometrial sampling performed between March 1994 and May 2008 using our Department of Pathology database. All specimens obtained from the endometrial sampling as well as the subsequent hysterectomy were reviewed at our institution.
Diagnoses were extracted from the clinical record without retrospective review of the slides. The diagnoses had been rendered by a mixed group of pathologists, some of whom had subspecialty expertise in gynecologic pathology.
Pathology reports for the endometrial sampling specimens were then reviewed in detail, and CAH was further classified as CAH suspicious if the pathologists deemed the specimen was suspicious for or bordering on cancer or they could not rule out cancer and no obvious invasive carcinoma was present. The specimen was subclassified as CAH polypoid if it arose in or was associated with a polyp, whether confined or not. If CAH was reported as only focally present and/or exhibited mild atypia, it was subclassified as CAH focal. Cases were deemed CAH not otherwise specified (NOS) if the CAH was not further qualified.
These classifications were not mutually exclusive initially in that each case could have multiple classifications. The rate of underlying carcinoma for each classification was then determined by review of the hysterectomy specimens. Pathologic diagnoses of the preoperative endometrial samplings from referring institutions were not abstracted. There has been an overall general consistency in the terms used among our pathologists over the study period.
We then assigned each case exclusively to 1 of 3 categories using the initial diagnostic information and based on the highest risk feature in the specimen. These 3 exclusive categories were: (1) CAH suspicious; (2) CAH NOS; and (3) CAH focal and/or polypoid. The rate of underlying carcinoma was then determined for each category. The features of each carcinoma identified were extracted from the pathology reports.
The carcinomas were dichotomized into low or high risk (LR and HR, respectively) based on final International Federation of Gynecology and Obstetrics (FIGO) grade and depth of invasion (DOI), as previously described ( Table 1 ). LR-FIGO was defined as endometrioid carcinoma limited to the endometrium of any grade (FIGO stage IA, grades 1-3) or FIGO grade 1 with DOI less than 50% (FIGO stage IB, grade 1). HR-FIGO carcinomas had serous and clear cell histology and were considered as final FIGO grade 3 for the purpose of this dichotomization.
| DOI | Grade 1 | Grade 2 | Grade 3 a |
|---|---|---|---|
| Limited to endometrium | LR-FIGO | LR-FIGO | LR-FIGO |
| <50% DOI | LR-FIGO | HR-FIGO | HR-FIGO |
| ≥50% DOI | HR-FIGO | HR-FIGO | HR-FIGO |
χ 2 and Fisher’s exact test were used, as appropriate, to compare nominal variables. Median values were compared using the Mann-Whitney U test. Data were not available for every subgroup analysis and accounts for some of the subgroup analyses not having all identified cases. Multivariate modeling was performed incorporating factors found to be significant on univariate analysis using binary logistic regression. All statistical analyses were performed using SPSS 15.0.1 (SPSS, Inc., Chicago, IL).
Results
We identified 197 cases of CAH diagnosed on preoperative endometrial sampling and reviewed at our institution. The clinical characteristics of the cohort are described in Table 2 . Carcinoma was identified in the hysterectomy specimen in 67 of the 197 cases (34%). The association of various factors with an underlying carcinoma is detailed in Table 3 . The multivariate model included 188 of the 197 cases.
| Variable | n (%) |
|---|---|
| Age, y | |
| Median (range) | 54 (32–86) |
| BMI, kg/m 2 | |
| Median (range) | 28.6 (17–74) |
| Median gravidity (range) | 2 (0–9) |
| Median parity (range) | 1 (0–7) |
| Time from biopsy to hysterectomy, d | |
| Median (range) | 47 (5–572) |
| Uterine weight, g | |
| Median (range) | 140 (8–750) |
| Menopause status | |
| Pre/perimenopausal | 74 (40) |
| Postmenopausal | 113 (60) |
| Method of sampling | |
| D&C | 123 (62) |
| EMB | 74 (38) |
| Megace use | |
| Past | 5 (2.5) |
| Current | 8 (4) |
| Variable | n a | No cancer in hyst | Cancer in hyst | Univariate P value | Multivariate HR (95% CI) | Multivariate P value |
|---|---|---|---|---|---|---|
| Method of sampling | ||||||
| D&C | 123 | 90 (73%) | 33 (27%) | |||
| EMB | 74 | 40 (54%) | 34 (46%) | .006 | 2.33 (1.12–4.85) | .02 |
| Menopause status | ||||||
| Pre/perimenopause | 74 | 60 (81%) | 14 (19%) | |||
| Postmenopause | 113 | 64 (57%) | 49 (43%) | .001 | 1.64 (0.6–4.51) | .3 |
| CAH type | ||||||
| Nonsuspicious | 152 | 110 (72%) | 42 (28%) | |||
| Suspicious | 45 | 20 (44%) | 25 (56%) | .001 | 3.29 (1.41–7.67) | .006 |
| CAH focal | ||||||
| Yes | 54 | 44 (81%) | 10 (19%) | |||
| No | 143 | 86 (60%) | 57 (40%) | .005 | 0.4 (0.17–0.95) | .04 |
| CAH polypoid b | ||||||
| Yes | 45 | 36 (80%) | 9 (20%) | |||
| No | 152 | 94 (62%) | 58 (38%) | .02 | 0.36 (0.13–0.99) | .048 |
| CAH polypoid, focal | ||||||
| Yes | 14 | 11 (79%) | 3 (21%) | |||
| No | 183 | 119 (65%) | 64 (35%) | .4 | — | — |
| Age, y | .004 | |||||
| Median (range) | 197 | 52 (32–79) | 58 (40–86) | < .001 | 0.93 (0.89–0.98) | |
| BMI, kg/m 2 | .06 | |||||
| Median (range) | 196 | 28.1 (17–54) | 32.4 (20–74) | .01 | 0.97 (0.93–1.0) | — |
| Median gravidity (range) | 186 | 2 (0–9) | 2 (0–5) | .6 | — | — |
| Median parity (range) | 186 | 1 (0–7) | 1 (0–4) | .9 | — | |
| Time from biopsy to hyst, d | — | |||||
| Median (range) | 197 | 47.5 (5–572) | 47 (6–279) | .9 | ||
| Uterine weight, g | — | — | ||||
| Median (range) | 151 | 135 (30–750) | 145 (8–370) | .5 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
