Fax: (507) 284-1599
Direct uterine sampling with the Tao brush sampler using a liquid-based preparation method for the detection of endometrial cancer and atypical hyperplasia†
A feasibility study
Article first published online: 11 JUN 2008
Copyright © 2008 American Cancer Society
Volume 114, Issue 4, pages 228–235, 25 August 2008
How to Cite
Kipp, B. R., Medeiros, F., Campion, M. B., Distad, T. J., Peterson, L. M., Keeney, G. L., Halling, K. C. and Clayton, A. C. (2008), Direct uterine sampling with the Tao brush sampler using a liquid-based preparation method for the detection of endometrial cancer and atypical hyperplasia. Cancer, 114: 228–235. doi: 10.1002/cncr.23636
See editorial on pages 219–21, this issue.
- Issue published online: 13 AUG 2008
- Article first published online: 11 JUN 2008
- Manuscript Accepted: 25 APR 2008
- Manuscript Revised: 24 APR 2008
- Manuscript Received: 16 DEC 2007
- endometrioid cancer;
- cytology screening;
- collection devices
Endometrial cytology sampling devices for direct uterine sampling have been shown in previous studies to be a reliable and relatively painless method for detecting endometrial lesions. The purpose of the current study was to determine the performance characteristics of endometrial cytology for the detection of malignancy and atypical hyperplasia using liquid-based cytology specimens collected with the Tao brush sampler.
Brushings of the endometrial cavity were obtained from 139 hysterectomy specimens before routine histopathologic evaluation. Cytology specimens were fixed in PreservCyt and processed using ThinPrep technology. Cytology diagnoses were classified as nondiagnostic, negative, atypical, or positive for malignancy. Histopathologic findings were used as the gold standard for determining the performance characteristics of cytology.
Histopathologic results from the 139 patients included 81 (58%) endometrial cancers, 7 (5%) complex hyperplasias with atypia, 2 (1%) complex hyperplasias without atypia, and 49 (35%) patients with benign histology. The number of specimens diagnosed cytologically as positive, atypical, negative, or nondiagnostic was 60 (43%), 40 (29%), 37 (27%), and 2 (1%) specimens, respectively. The overall sensitivity and specificity of cytology for detecting endometrial cancer and atypical hyperplasia were 95% and 66% when atypical cytology specimens were considered positive.
The results of the current study indicate that direct endometrial sampling by liquid-based endometrial cytology collected with the Tao brush sampler produces specimens that contain cellular material that may be identified as endometrial cancer or atypical hyperplasia. Both atypical and positive cytology diagnoses are indicators for triage to more specific methods of diagnosis. Cancer (Cancer Cytopathol) 2008. © 2008 American Cancer Society
Endometrial carcinoma is the most common cancer of the female genital tract in the US.1 Although cervical cancer has historically been more common than endometrial cancer, Papanicolaou (Pap) screening has steadily decreased the incidence of cervical cancer, whereas the number of patients with endometrial cancer remains high. In 2007, it was estimated that 7400 women would die of endometrial cancer, which is nearly double the 3670 women predicted to die from cervical cancer.1 Unfortunately, although screening algorithms for cervical cancer continue to improve (eg, combined human papillomavirus–Pap testing), screening for endometrial cancer with cytology techniques remains widely unaccepted by the medical community.
Direct endometrial sampling is currently obtained through endometrial biopsy or curetting only in symptomatic patients. Endometrial curetting is a painful and costly procedure that requires dilatation and anesthesia, and therefore is usually performed in a hospital setting. Endometrial biopsies that use the Pipelle apparatus are commonly used in an outpatient office setting, but are less efficacious as a screening tool, because significant portions of the endometrial cavity may not be sampled.2, 3
The Tao brush is a direct endometrial sampling device that has been reported in previous studies to be a reliable and relatively painless method for detecting endometrial lesions.3-10 One strength of the Tao brush sampler is that it permits direct intrauterine collection of endometrial cells without cellular contamination from the lower genital tract, in particular, endocervical cells.3 The Tao brush has not gained widespread acceptance for endometrial cancer sampling, possibly because of the perceived challenges associated with the cytologic diagnosis of directly sampled endometrial tissue. However, many architectural features associated with hyperplastic and malignant endometria are maintained in the cytologic preparation, raising the potential for use as a screening test.7 Directly sampled endometrial cytology specimens can be difficult to evaluate because of blood, overlapping cells, and hormone-associated morphologic changes occurring during the menstrual cycle or resulting from hormonal therapy.11 Recent studies have shown that using liquid-based preparation techniques (eg, ThinPrep methodology) results in an improvement in the diagnostic accuracy of endometrial cytology.5, 11 The goal of this study was to determine the feasibility of endometrial cytology by direct endometrial sampling for the detection of malignancy using specimens collected with the Tao brush sampler and processed with liquid-based ThinPrep technology.
MATERIALS AND METHODS
As approved by the Mayo Clinic Institutional Review Board, 139 cytology samples of the endometrial cavity were obtained from women with a mean age of 60 years (median, 60 years; range, 31-87 years) using the Tao brush sampler (Cook Ob/Gyn, Spencer, Ind). Women in this study included a spectrum of patients undergoing vaginal or abdominal hysterectomy for suspected or confirmed endometrial neoplasia or hyperplasia (n = 84), abnormal uterine bleeding (23 patients), uterine prolapse (13 patients), pelvic mass (11 patients), uterine leiomyosarcoma or leiomyoma (5 patients), or malignant/premalignant cervical disease (3 patients). Many of these patients had biopsy diagnosis before hysterectomy, but the Tao brush cytology result was only compared with the hysterectomy histologic result (used as the gold standard).
All cytology specimens were collected with the Tao brush sampler before uterine dissection of the hysterectomy specimen for routine histopathologic evaluation. To begin the cell collection process, the 3.5-cm sheath of the Tao brush was inserted to the level of the fundus through the cervical canal, and the sheath was pulled back. The brush was rotated 360° 3 to 5 times, the outer sheath was pushed backed to the tip, and the device was removed from the uterine cavity. After the device was removed, the 3.5-cm brush was cut off and immersed into a 20-mL PreservCyt solution vial (Cytyc, Marlborough, Mass). The vials were then sent to the Mayo Clinic Molecular Cytology and Imaging Laboratory, processed for cytology assessment using the ThinPrep 2000 (Cytyc) automated slide preparation system, and stained with Papanicolaou stain.
ThinPrep Pap–stained cytology slides were interpreted as nondiagnostic, negative for malignancy (Fig. 1a), atypical (Fig. 1b), or positive for malignancy (Fig. 1c) as previously described7, 12 using a consensus diagnosis from 2 gynecologic pathologists who were blinded to patients' clinical and histopathologic findings. Because this was a feasibility study, and the institution's cytotechnologists are not currently trained to screen directly sampled endometrial cytology specimens, no prior screening by a cytotechnologist was used for the study. In patients with a nondiagnostic result, specimen vials were reprocessed with glacial acetic acid, and an additional slide was prepared on the ThinPrep 2000 as previously described.13
Low-power examination, assessing architectural features and background, was important in categorizing the lesions. Benign endometrium most often showed an admixture of epithelial and stromal cells and abundant cellularity. Proliferative endometrium was characterized by abundant stromal cells in which discrete, elongated, straight tubular epithelial structures intermingled. The discrete and characteristic nature of the epithelial groups allowed recognition on low-power examination. On high power, the epithelial cells were round to oval with slight nuclear overlap, but with uniform chromatin distribution and even placement within the cell groups. Secretory endometrium showed more abundant cytoplasm in both epithelial and stromal components, and a tendency for the stromal cells to be discohesive. The epithelial architecture showed more flat sheet-like arrangements, rather than tight tubular structures. The nuclear features, however, supported a benign etiology with uniform nuclei, small regular nucleoli, and a honeycomb arrangement of the epithelial cells. The epithelial cell groups sometimes demonstrated cytoplasmic secretions. Endometrial atrophy was much less cellular, characterized by scant groups of glandular and stromal cells. The stromal cells occasionally appeared more like fibroblasts and had small pyknotic nuclei. The epithelial cells were arranged in small flat clusters with loss of the tubular architecture. The epithelial nuclei were most often small and uniform, although some degenerative atypia (nuclear enlargement, hyperchromasia, and nucleoli) was seen focally. Hyperplasia and low-grade adenocarcinoma were similar to each other in regard to cytologic presentation. They were characterized by variable cellularity, with irregularly shaped glandular groups that lost the tight tubular arrangements of proliferative epithelium. The epithelium often demonstrated enlarged dilated tubular structures or branched glands. Occasionally the tubular arrangement was lost altogether. There often was an increased gland to stromal ratio. There were nuclear enlargement and moderate nuclear disarray within the cell groups. Nuclear atypia, with nuclear size variability and nucleoli, was also observed to some degree. In high-grade adenocarcinomas, the epithelial clusters were small and irregular, with marked cytologic atypia. The background was usually scantly cellular, and it occasionally contained blood or tumor diathesis.
Histopathologic findings from the hysterectomy specimen were used as the gold standard for determining the sensitivity and specificity of cytology, which were calculated in 2 ways: 1) using only positive cytologic diagnoses as evidence of malignancy, and 2) using both positive and atypical cytology diagnoses as an indicator of malignancy. Hysterectomy diagnoses of cancer and complex hyperplasia with atypia were considered positive for statistical purposes. Statistical analysis was performed with SPSS for Windows software (version 11.5; SPSS Inc, Chicago, Ill) using the Student t test for paired data and McNemar test, as appropriate. All tests were 2-tailed, and a P value <.05 was considered statistically significant.
The study comprised a total of 139 specimens including 81 patients with hysterectomy diagnoses of endometrial cancer, 9 patients with hyperplasia, and 49 patients with benign endometrium, which represented 58%, 6%, and 35% of the population, respectively. The 81 malignant tumors were classified as endometrioid adenocarcinoma (63 patients; 78%) (Fig. 2a), serous adenocarcinoma (7 patients, 9%) (Fig. 2b), mixed endometrioid and serous cancer (3 patients; 4%), mixed serous and clear cell cancer (2 patients ; 2%), carcinosarcoma (2 patients; 2%) (Fig. 2c), clear cell carcinoma (3 patients; 4%) (Fig. 2d), and leiomyosarcoma (1 patient; 1%) (Fig. 2e). The 63 endometrioid tumors consisted of 33 grade 1 tumors, 17 grade 2 tumors, and 13 grade 3 tumors (determined according to the International Federation of Gynecology and Obstetrics grading system). The 9 patients with hyperplasia consisted of 7 (78%) patients with a hysterectomy diagnosis of complex hyperplasia with atypia and 2 (22%) patients with a diagnosis of complex hyperplasia without atypia. The 49 patients with benign endometrium were represented by 26 (53%) patients with proliferative endometrium, 20 (41%) with inactive endometrium, and 3 (6%) with secretory endometrium. Eleven (8%) of the patients with benign endometrium also had an endometrial polyp.
The cytology and corresponding pathology results from the 139 patients are shown in Table 1. Four patients with a nondiagnostic cytology interpretation on the first processed ThinPrep slide were reprocessed by using the glacial acetic methodology. Two of the 4 patients continued with a nondiagnostic cytology result because of inadequate cellularity, excess proteinaceous debris, and blood, whereas the remaining 2 reprocessed specimens had enough cells on the second slide to render a diagnosis. Twenty-nine percent of the 139 specimens were diagnosed as atypical by cytology, whereas 43% and 27% of cytology specimens were diagnosed as positive and negative for malignancy, respectively.
|Cytology||No. (%)||Tissue Diagnosis, No. (%)|
|Benign||Complex Hyperplasia Without Atypia||Complex Hyperplasia With Atypia||Malignant|
|Nondiagnostic||2 (1)||1 (50)||0 (0)||1 (50)||0 (0)|
|Negative||37 (27)||33 (89)||0 (0)||1 (3)||3 (8)|
|Atypical||40 (29)||13 (33)||2 (5)||5 (13)||20 (50)|
|Positive||60 (43)||2 (3)||0 (0)||0 (0)||58 (97)|
|Total||139||49 (35)||2 (1)||7 (5)||81 (58)|
Several common characteristics were observed among the specimens that were atypical by cytology and neoplastic on the hysterectomy specimen. The first was a preponderance of benign-appearing epithelium and stroma, but scattered rare clusters of epithelial groups that demonstrated mild to moderate atypia. This may have been related to architectural disorder (ie, departure from a flat-sheet–like arrangement of nuclei within the group or nuclear atypia). This cytologic presentation suggested that the patient may have a focal hyperplastic or neoplastic lesion arising in the background of benign endometrium. The second characteristic was the presence of relatively abundant cell groups with only mildly atypical features. Considerations at the time of evaluation included disordered proliferative endometrium, simple or complex hyperplasia, or possibly well-differentiated adenocarcinoma. The final feature was a paucicellular specimen without many stromal cells in the background and rare epithelial groups with cytology atypia. The atypia ranged from moderate to marked. The differential on these specimens was believed to represent either surface atypia in the setting of polyps or neoplastic change.
The overall performance of cytology is shown in Table 2. As this table illustrates, the overall sensitivity and specificity of routine cytology for the detection of endometrial cancer and atypical hyperplasia depended on how atypical endometrial cytology specimens were classified. The sensitivity of endometrial cytology was 67% when only a positive cytology diagnosis was considered positive, which was significantly lower (P<.001) than the 95% sensitivity of endometrial cytology when both positive and atypical cytology diagnoses were considered as evidence of premalignant/malignant changes. There was also a significant difference (P<.001) in the specificity of endometrial cytology, with 96% of patients without cancer (ie, benign and hyperplastic endometrium without atypia) being termed negative by cytology when atypical cytology diagnoses were considered negative, compared with a 66% specificity when equivocal cytology diagnoses were considered positive for premalignant/malignant changes.
|Tao Brush Cytology||Sensitivity†||Specificity†||PPV†||NPV†|
|Endometrial cytology‡, No. (%)||58/87 (67)||48/50 (96)||58/60 (97)||48/77 (62)|
|Endometrial cytology§, No. (%)||83/87 (95)||33/50 (66)||83/100 (83)||33/37 (89)|
The positive predictive value (PPV) and negative predictive value (NPV) of cytology are also shown in Table 2. The PPV of positive (n = 60), atypical (n = 40), and combined (positive and equivocal; n = 100) cytology diagnoses for the detection of cancer or complex hyperplasia with atypia were 97%, 63%, and 83%, respectively. The NPV of negative (n = 37), atypical (n = 40), and combined (n = 77) endometrial cytology diagnoses was 89%, 38%, and 62%, respectively.
The results of the current study suggest that directly sampled endometrial specimens processed by liquid-based cytology can detect endometrial carcinoma with reasonable sensitivity. According to the American Cancer Society guidelines,14 there is not sufficient evidence to recommend that women who are at average risk or a slightly increased risk for endometrial cancer because of a history of estrogen therapy, tamoxifin therapy, late menopause, nulliparity, infertility or failure to ovulate, obesity, diabetes, or hypertension undergo screening. However, patients with a higher risk for harboring endometrial cancer (eg, patients with abnormal uterine bleeding or hereditary nonpolyposis colorectal cancer) might benefit from an effective and low-cost screening test for endometrial cancer.3, 14, 15 Earlier tumor detection in patients at risk for endometrial cancer offers the opportunity to improve patient survival. Endometrial carcinoma deaths currently exceed cervical carcinoma deaths in the US by nearly 2-fold.1 A method that is inexpensive, is easily performed in the office setting, and causes minimal discomfort to the patient may offer potential for screening of patients with defined risk factors for developing endometrial carcinoma, in an effort to detect those patients with lower-stage disease.
Endometrial sampling in symptomatic or high-risk patients is currently obtained through dilatation and curettage (D&C) or endometrial biopsy. Unfortunately, endometrial curetting is a painful and costly procedure, and less than half of the uterine cavity is evaluated in approximately 60% of D&C procedures, which results in false-negative diagnoses.16 With regard to endometrial biopsy, a recent meta-analysis of 39 studies demonstrated that endometrial biopsy with the Pipelle is superior to other available devices for the detection of endometrial cancer.16 This study found that the Pipelle had an endometrial cancer detection rate of 91% and >99% in premenopausal and postmenopausal women, respectively, and could detect 81% of atypical hyperplasias. Unfortunately, the Pipelle biopsy yields insufficient tissue for pathologic assessment in a large percentage of women. In a study assessing 801 asymptomatic perimenopausal and postmenopausal women, 195 (25%) women had a biopsy that yielded insufficient tissue.17 This was similar to another larger study in which 36% of women were found to have insufficient tissue for pathologic assessment.18 Endometrial biopsy and curettage are also associated with significant discomfort for the patient.10 The Tao brush, however, is reported to have minimal patient discomfort, and the office collection procedure is similar to obtaining an endocervical brush specimen for routine cervical cytology.10 Other studies are needed to corroborate that investigator's findings. Therefore, although other endometrial sampling techniques exist for detecting endometrial cancer and atypical hyperplasia, each of these modalities has significant limitations, impeding implementation as a routine screening test.
The results of the current study demonstrate that the Tao brush direct endometrial sampler, when used with a liquid-based cytology preparatory method, has potential for detecting endometrial cancer. Only 2 specimens (∼1% of all specimens) were diagnosed as nondiagnostic using the Tao brush sampler (Table 1). The overall sensitivity of detecting endometrial cancer and complex hyperplasia with atypia was 95% (4 false-negative diagnoses) when atypical cytology results were considered positive. Histopathologic results from the 4 specimens with apparent false-negative results included 3 patients with grade 1 endometrioid tumors and 1 patient with complex hyperplasia with atypia. Re-review of the cytology from these 4 patients demonstrated that there were rare groups of atypical cells in a background of benign endometrium that most likely represented cancer/hyperplasia not detected on the initial scan. One drawback of the current study was that an initial scan of the slide was not performed by a cytotechnologist followed by pathologist interpretation (our usual clinical practice for most cytology specimens). Cytotechnologist screening and additional experience among pathologists assessing endometrial cytology specimens might increase the overall sensitivity to above 95%.
Another interesting finding of the current study was that 29% (40 patients) of patients had an atypical cytology diagnosis. Histopathologic results from these patients demonstrated that a large percentage had premalignant or malignant (25 patients; 63%) histopathologic results. This suggests that both atypical and positive cytology diagnoses should be used as indicators for triage to more specific methods of diagnosis, such as endometrial biopsy.
The overall specificity of the Tao brush was 66% when atypical cytology specimens were considered positive, and 96% when atypical diagnoses were considered negative. There were only 2 cases that were termed positive by cytology that did not have either cancer or complex hyperplasia with atypia within the hysterectomy specimen. Both of these patients had endometrial polyps that on re-review of the cytology still contained worrisome cells. A review of the histology found that 1 patient's polyps had surface atypia. Previous data suggest that it is not uncommon for a focal area of endometrial carcinoma, particularly of the serous type, to be present within a polyp and to be diagnosed as benign.19 Additional studies have also demonstrated that endometrial cytology is able to detect cancers that are diagnosed as benign pathologically, especially in patients with smaller (<2 cm) tumors or cancer localized to the uterine fundus.2, 20
Although the findings of the current study are informative, there were some study limitations. Specimens analyzed by cytology were collected only in women undergoing hysterectomy for routine histopathologic evaluation. This resulted in a study consisting of a very high-risk population, of which over half the patients were found to have endometrial cancer. Thus, the results obtained from this study (especially PPVs and NPVs for cytology) do not necessarily reflect what would be found in a lower-risk population that might be the target for endometrial cancer screening. Further studies will be required to determine the PPVs and NPVs of Tao brush cytology in a lower-risk screening population. Although this study does not represent a screening population, there were benefits for collecting specimens using this methodology. All specimens had a histopathologic diagnosis on the same day the cytology specimens were collected and optimal gold standard comparison (complete surgical pathology evaluation of the hysterectomy specimen). In addition, by collecting specimens from high-risk patients, we were able to determine the ability of endometrial cytology to detect rarer nonendometrioid cancers. Eighteen patients in the current study had a diagnosis of nonendometrioid cancers, including serous adenocarcinoma, mixed endometrioid and serous cancer, mixed serous and clear cell cancer, clear cell carcinoma, carcinosarcoma, and leiomyosarcoma. We found that 16 of these 18 specimens were interpreted as positive by endometrial cytology, whereas the remaining 2 patients were diagnosed as atypical. The ability to accurately flag all these specimens as abnormal is important, because nonendometrioid cancers have a higher propensity for early metastasis and demonstrate aggressive clinical behavior.21
Another limitation to the current study is that the Tao brush cytology specimens were obtained after hysterectomy rather than before hysterectomy. It is possible that the endometrial sampling obtained from a hysterectomy specimen has advantages over what might be obtained in the physician's office. Additional prospective studies that assess the sensitivity and specificity of the Tao brush on samples obtained in the office setting are needed to better understand the role of endometrial cytology in routine clinical practice.
In conclusion, the results of the current study indicate that endometrial cytology is potentially a sensitive method for detecting cancer if positive and atypical cytology diagnoses are used as indicators for potential triage to more specific methods of diagnosis. The potential for using this method as a screening test appears promising, given the reported ease of sampling and minimal discomfort to the patient. Further studies are needed to determine whether endometrial cytology alone, or cytology in conjunction with other testing modalities, can be used as a screening test for the detection of endometrial carcinoma.
- 3Evaluation of abnormal uterine bleeding: comparison of three outpatient procedures within cohorts defined by age and menopausal status. Health Technol Assess. 2004; 8:iii-iv, 1–139., , , , , .
- 7Performance characteristics of the Indiana University Medical Center endometrial sampler (Tao Brush) in an outpatient office setting, first year's outcomes: recognizing histological patterns in cytology preparations of endometrial brushings. Diagn Cytopathol. 2000; 22: 186–195..
- 12Cytopathology of the endometrium. Chicago, IL: American Society of Clinical Pathologists; 1993..