- Email: [email protected]
Significance of Atypical Glandular Cells of Undetermined Significance on ThinPrep Papanicolaou Smears
Gynecologic Oncology 78, 245–250 (2000) doi:10.1006/gyno.2000.5884, available online at http://www.idealibrary.com on
Significance of Atypical Glandular Cells of Undetermined Significance on ThinPrep Papanicolaou Smears Gamal H. Eltabbakh, M.D.,* ,1 Jennifer N. Lipman,* Sharon L. Mount, M.D.,† and Ann Morgan, M.S.‡ *Department of Obstetrics and Gynecology, †Department of Pathology, and ‡Department of Biostatistics, University of Vermont College of Medicine, Burlington, Vermont 05401 Received February 22, 2000
epithelial cell abnormality termed atypical glandular cells of undetermined significance (AGCUS). Although recent reports [1–7] have described a high incidence of dysplasia and cancer among women with this cytology, our understanding of this group of Papanicolaou smear reporting and the most costeffective method of management of these women is still evolving. To address the problem of a high false-negative rate of Papanicolaou smear secondary to sampling or slide preparation, the ThinPrep (Cytec Corp., Boxborough, MA) slide preparation technique has been used recently [8]. This method has been shown to improve specimen adequacy and diagnostic yield, lower the percentage of atypical squamous or glandular cells of undetermined significance diagnoses, and increase the percentage of more accurate Papanicolaou smear diagnoses such as squamous intraepithelial lesions (SIL) [8, 9]. However, the accuracy of ThinPrep Papanicolaou smears in diagnosing glandular abnormalities has not been extensively studied. Conventional Papanicolaou smears have been reported to have a poor accuracy in diagnosing glandular lesions of the cervix [10]. Most of the reports investigating the significance of AGCUS cytology included patients whose smears were obtained by the conventional method [1– 6]. The significance of this cytologic diagnosis on Papanicolaou smears obtained by the ThinPrep method is not clear. The aim of the current study is to investigate the clinical significance of AGCUS on ThinPrep Papanicolaou smears and to study the factors that could predict significant histopathologic findings among these patients.
Objective. The aim of this study was to assess the incidence and risk factors predictive of significant histopathologic findings among women with atypical glandular cells of undetermined significance (AGCUS) on ThinPrep Papanicolaou smears. Methods. ThinPrep smears with AGCUS obtained between 1997 and 1999 were reviewed. Patients’ charts were reviewed and patients’ characteristics, follow-up information, and colposcopy and biopsy results were recorded. Pathologic slides were reviewed. The demographic features of women with favor reactive smears were compared with those with favor neoplasia and risk factors predictive of significant histopathologic findings (high-grade squamous intraepithelial lesion, endometrial hyperplasia, and cervical or endometrial cancers) were calculated. Results. The rate of diagnosis of AGCUS was 0.65%. Eightyfour patients with follow-up information were identified. The demographic features of women with smears favor neoplasia (n ⴝ 43) were similar to those with smears favor reactive (n ⴝ 41). The rates of incidence of any dysplasia or cancer and significant histopathologic findings were 32.1 and 22.6%, respectively, and were higher among women with smears favor neoplasia than among women with smears favor reactive (41.9% versus 22.0%, P ⴝ 0.051, and 34.9% versus 9.8%, P ⴝ 0.006, respectively). The subtype of cytology was the only factor that predicted significant histopathologic findings (odds ratio ⴝ 5.0, 95% confidence interval 1.6, 15.6, P < 0.010). Conclusions. In women with AGCUS on ThinPrep smears, significant histopathologic findings were found in 34.9% versus 9.8%, depending on the subtype of the smear (favor neoplasia vs reactive). Further studies are needed to validate the cytologic criteria for subtyping AGCUS smears and base management of women with AGCUS cytology on the subtype of the smear. © 2000 Academic Press
MATERIAL AND METHODS INTRODUCTION In our institution, the ThinPrep technique has been available for cervical smears since 1997. Through a computer search, we identified all patients with ThinPrep Papanicolaou smears demonstrating AGCUS seen at the Department of Obstetrics and Gynecology, University of Vermont/ Fletcher Allen Health Care, and the Womens Choice Clinic in Burlington, Vermont, between 1997 and August 1999. We excluded patients with
Implementation of the Bethesda system in Papanicolaou smear reporting has resulted in the creation of a new class of 1
To whom correspondence and reprint requests should be addressed at Division of Gynecologic Oncology, University of Vermont, MCHV Campus, 111 Colchester Avenue, Shepardson 330, Burlington, VT 05401. Fax: (802) 847-2360. E-mail: [email protected]. 245
0090-8258/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
246
ELTABBAKH ET AL.
similar cytology on conventional Papanicolaou smears. The ThinPrep Papanicolaou smears were originally diagnosed by a group of five American Board of Pathology certified pathologists using the Bethesda system criteria. These smears were then formally reviewed by one of the coauthors (SLM) to confirm the diagnosis of AGCUS. Since the distinction between AGCUS reactive and AGCUS neoplastic can show interobserver variability, the smears were evaluated by SLM to determine whether the original diagnosis was acceptable. Patients’ charts were reviewed and the following information was abstracted: age, gravidity, parity, menopausal status, use of hormonal replacement therapy, smoking, history of abnormal Papanicolaou smear and its treatment, history of human papillomavirus (HPV) infection by patient recall or previous Papanicolaou smears or biopsies, history of adenocarcinoma in situ, history of pelvic radiation, colposcopic findings, biopsy results, and follow-up Papanicolaou smears. In addition, histologic slides were reviewed by the same pathologist to confirm the histologic diagnosis. Patients were followed up with particular emphasis on the development of pelvic or extrapelvic cancers. Smoking was defined as having ever smoked for a period equal to or more than 1 year. Cervical smears were collected using the Cervex brush (Unimar, Inc., Wilton, CT). The brush was then rinsed into a vial containing 20 mL of preservative (PreservCyt) and the vial was capped and sent to the cytopathology laboratory where slides were processed using the ThinPrep processor. Cytologic diagnosis of AGCUS was based on established criteria [11]. The cells are of glandular type and display nuclear atypia that exceeds obvious reactive or reparative changes, but lack the unequivocal features of invasive carcinoma. The smears were classified as favor reactive when they exhibited inflammatory changes with nuclear enlargement associated with variability or irregularity of the shape of the nuclei and as favor neoplasia when the cells had more abnormal nuclei with hyperchromasia and overlapping. During the study period, management of patients with AGCUS has been rather consistent. Most patients with this cytology underwent colposcopy. Depending on the colposcopic findings, directed biopsies and endocervical curettage (ECC) were obtained. Patients with normal colposcopic findings, those with abnormal bleeding, and those whose cytology suggested endometrial origin underwent endometrial biopsy. Patients with persistent AGCUS cytology despite negative colposcopy and endometrial biopsy were offered pelvic ultrasound, hysteroscopy, and cone biopsy. We did not use HPV DNA testing as an adjunct to abnormal Papanicolaou smear in our study population. We compared patients with AGCUS favor neoplasia and those with smears favor reactive in demographic features and the incidence of any degree of dysplasia or cancer and the incidence of significant histopathologic findings. Significant histopathologic findings were defined as high-grade SIL, carcinoma in situ, cervical or endometrial cancers, and endome-
trial hyperplasia. Statistical analysis was performed using the two-sample t test, 2 test, and Fisher’s exact test. Assessment of the risk factors was performed by calculating the Mantel– Haentzel odds ratio and the 95% confidence interval. The statistical analysis was done using the SAS statistical package (SAS Institute, Cary, NC). Two-tailed P values ⬍0.05 were considered statistically significant. RESULTS We identified 86 patients with the confirmed cytologic diagnosis of AGCUS from 13,134 ThinPrep Papanicolaou smears obtained from patients seen between 1997 and August 1999. Thus, the proportion of women with AGCUS cytology was 0.65%. Review of the ThinPrep smears by SLM found no significant discrepancies from the original diagnoses. Although in individual cases the difference between an AGCUS favor reactive and an AGCUS favor neoplastic can be subjective, the original diagnoses were judged to be within reason. Forty-two (48.8%) patients had favor reactive and 44 (51.2%) had favor neoplasia smears. Of the 42 patients with smears favor reactive, 1 had no follow-up information, 5 were followed up with repeat Papanicolaou smears, and 36 had colposcopy. Two of 5 patients who had repeat Papanicolaou smears had abnormal smears and underwent colposcopy and biopsy revealing high-grade SIL in both. Of the 38 patients who underwent colposcopy, 35 underwent directed biopsy and ECC and 4 underwent endometrial biopsy. Of the 35 patients who had histologic diagnoses, 1 had high-grade SIL, 5 had low-grade SIL, and 1 had complex endometrial hyperplasia with atypia. Of the 44 patients with smears favor neoplasia, 1 had no follow-up information, 5 were followed up with repeat Papanicolaou smears, and 38 underwent colposcopy. Of the patients who underwent colposcopy, directed biopsies and ECC were obtained in 37 and endometrial biopsies were obtained in 10 patients. The histopathologic diagnoses among these patients were normal (n ⫽ 16), high-grade SIL (n ⫽ 3), low-grade SIL (n ⫽ 3), adenocarcinoma in situ (n ⫽ 1), small cell cervical carcinoma (n ⫽ 1), endometrial adenocarcinoma (n ⫽ 5), and mixed mu¨llerian tumor of the uterus (n ⫽ 1). Of the 5 patients who were followed up with repeat cytology, 3 had abnormal Papanicolaou smears. Among these 3 patients, colposcopy, cervical, and endometrial biopsies revealed highgrade SIL (n ⫽ 1), endometrial adenocarcinoma (n ⫽ 1), and cervical polyp (n ⫽ 1). The 5 patients who had no biopsies (3 with AGCUS favor reactive and 2 with AGCUS favor neoplasia) had normal repeat Papanicolaou smears. None of these patients developed cervical or vaginal dysplasia or gynecologic cancers. We excluded the 2 patients who had no follow-up information from further analysis. The demographic features of patients with AGCUS cytology are illustrated in Table 1. The mean (range) age of our patients was 46.4 (20 –79) years. There were no significant differences
247
AGCUS ON THINPREP PAP
TABLE 1 Patient Characteristics Atypical glandular cells of undetermined significance
Age in years (mean ⫾ SD) Gravidity (mean ⫾ SD) Parity (mean ⫾ SD) History of STD History of abnormal Papanicolaou smear History of treatment for abnormal smear History of dysplasia History of hysterectomy Postmenopausal Hormonal replacement therapy History of smoking History of HPV History of adenocarcinoma in situ History of radiation therapy
Favor reactive (%)
Favor neoplasia (%)
P value
All patients (%)
44.8 ⫾ 10.3 2.0 ⫾ 1.6 1.7 ⫾ 1.4 12.5 60.0 20.0 25.0 5.0 33.3 18.0 27.5 10.0 2.5 5.0
47.9 ⫾ 13.9 2.5 ⫾ 2.3 2.0 ⫾ 2.1 11.9 59.5 26.2 36.8 9.5 38.5 20.5 30.8 12.8 7.7 2.6
0.246 0.191 0.462 1.0 1.0 0.507 0.257 0.676 0.637 0.774 0.749 0.693 0.293 0.571
46.4 ⫾ 12.3 2.2 ⫾ 2.0 1.8 ⫾ 1.6 12.2 59.7 23.2 30.8 7.3 35.9 19.2 29.1 11.4 5.1 3.8
Note. STD, sexually transmitted diseases; HPV, human papillomavirus.
in the characteristics between patients with smears favor reactive and those with favor neoplasia. Approximately 60% of the study population had a history of abnormal Papanicolaou smears and 30% had a history of smoking. The overall incidence of cervical or vaginal dysplasia among our study patients was 17.9%. The rates of incidence of adenocarcinoma in situ and endometrial cancers were 3.6 and 8.3%, respectively. One (1.2%) patient had complex endometrial hyperplasia with atypia and one (1.2%) had small cell cervical carcinoma. The histology of the patients with endometrial cancers was endometrioid adenocarcinoma (n ⫽ 5), papillary serous adenocarcinoma (n ⫽ 1), and mixed mu¨llerian tumor (n ⫽ 1). All women who had endometrial hyperplasia or cancer were more than 40 years of age. These women were significantly older than those with cervical dysplasia or cancer (mean age ⫾ standard deviation: 55.1 ⫾ 11.96 versus 43.5 ⫾ 13.8, respectively, P ⫽ 0.043). As demonstrated in Table 2, the incidence of significant histopathologic findings was significantly higher among patients with smears favor neoplasia compared with those with smears favor reactive (34.9% versus 9.8%, respectively, P ⫽ 0.006). The incidence of any degree of dysplasia or cancer was higher among women with smears favor neoplasia compared to those with smears favor reactive (41.9% versus 22.0%, respectively, P ⫽ 0.051). Review of the smears identified a probable endometrial origin in 31 (36.9%), a probable endocervical origin in 33 (39.3%), both in 4 (4.8%), and neither in 16 (19.0%) patients. There was no significant difference in the incidence of significant histopathologic findings based on the origin of the smear (19.4, 21.2, 50.0, and 25.0%, respectively, P ⫽ 0.537). Review of the smears of the patients who had endometrial hyperplasia or cancer identified a probable endometrial origin in 5, a
probable endocervical origin in 1, a probable endometrial and endocervical origin in 1, and neither in 1 patient. The incidence of endometrial hyperplasia or cancer was higher if the atypical cells were of endometrial than if they were of endocervical origin (5/31 [16.1%] versus 1/33 [3.0%], respectively, P ⫽ 0.075). Assessment of risk factors predictive of significant histopathologic findings demonstrated that only the subtype of AGCUS cytology (reactive versus neoplasia) was significantly predictive of significant histopathologic findings. As demonstrated in Table 3, age, menopausal status, smoking, history of TABLE 2 Histopathologic Findings Atypical glandular cells of undetermined significance
Low-grade SIL High-grade SIL Any dysplasia Adenocarcinoma in situ Cervical cancer Endometrial hyperplasia Endometrial cancer Total with significant histopathology
Favor reactive (n ⫽ 41) (%)
Favor neoplasia (n ⫽ 43) (%)
All patients (n ⫽ 84) (%)
5 (12.2) 3 (7.3) 8 (19.5) 0 0 1 (2.4) 0
3 (7.0) 4 (9.3) 7 (16.3) 3 (7.0) 1 (2.3) 0 7 (16.3)
8 (9.5) 7 (8.3) 15 (17.9) 3 (3.6) 1 (1.2) 1 (1.2) 7 (8.3)
4 (9.8)
15 (34.9)*
19 (22.6)
Note. SIL, squamous intraepithelial lesion. Significant histopathology is defined as high-grade SIL, adenocarcinoma in situ, cervical cancer, endometrial hyperplasia, and endometrial cancer. * P ⫽ 0.006.
248
ELTABBAKH ET AL.
TABLE 3 Risk Factors for Significant Histopathologic Findings No. with dysplasia or cancer/total number (%)
Factor Age ⱕ46 ⬎46 Menopausal status Postmenopausal Premenopausal Smoking Yes No History of HPV Yes No History of abnormal Papanicolaou Yes No Subtype Neoplasia Reactive
Odds ratio
95% Confidence interval
P value
8/34 (23.5) 11/50 (22.0)
1.8
0.6, 5.2
0.30
9/31 (29.0) 10/51 (19.6)
1.7
0.6, 4.7
0.333
6/23 (26.1) 13/41 (31.7)
0.8
0.2, 2.4
0.641
2/8 (25.0) 18/76 (23.7)
1.1
0.2, 5.9
1.0
11/49 (22.4) 8/35 (22.9)
1.0
0.3, 2.8
0.721
15/43 (34.9) 4/41 (9.8)
5.0
1.6, 15.6
⬍0.010
Note. HPV, human papillomavirus. Data for menopausal status and smoking were not known for some patients.
HPV, and history of abnormal Papanicolaou smears were not predictive of significant histopathologic findings. Patients with smears favor neoplasia were five times more likely to have significant histopathologic findings compared with those with smears favor reactive (odds ratio ⫽ 5.0, 95% confidence interval 1.6, 15.6, P ⬍ 0.010). The power of the current study to detect a 20% difference between the two groups assuming a 10% incidence in the lower group was 66.2%. None of the patients enrolled in the study developed extrauterine cancers during the period of the study. DISCUSSION AGCUS is an uncommon cytologic diagnosis with a reported incidence between 0.08 and 0.74% among women who had conventional Papanicolaou smears. The incidence of this cytology on ThinPrep Papanicolaou smears among our patients (0.65%) was within the reported range using the conventional Papanicolaou smear. Ashfag et al. [7] reported that among women attending Parkland Memorial Hospital the incidence of AGCUS cytology on ThinPrep Papanicolaou smears was similar to that on conventional smears (0.14 and 0.11%, respectively). Although several studies [1– 6] have reviewed the clinical significance of AGCUS cytology on conventional Papanicolaou smears, our study focuses specifically on the clinical significance of the AGCUS diagnosis on ThinPrep smears. As demonstrated in Table 4, these studies found that the rates of incidence of high-grade SIL, adenocarcinoma in situ, cervical cancer, and endometrial hyperplasia or cancer were 5.2–21.9,
0 –3.9, 1.4 – 4.3, and 1.3–11.0%, respectively. To our knowledge, the current study is the first to report the histopathologic findings among women with AGCUS on ThinPrep Papanicolaou smears as a separate entity. The incidence of high-grade SIL, adenocarcinoma in situ, cervical cancer, and endometrial cancer or hyperplasia reported in our study (8.3, 3.6, 1.2, and 9.5%, respectively) falls within the reported ranges in studies using conventional Papanicolaou smears [1– 6]. Comparing our results with the histopathologic findings among women with AGCUS cytology on ThinPrep smears to those with historical controls of women with the same cytology on conventional smears [1– 6] reveals similar findings. Compared with conventional smears, we did not find AGCUS diagnosis on ThinPrep smears to be associated with a higher incidence of significant histopathologic findings. Ashfag et al. [7] found that ThinPrep Papanicolaou smears provided more accurate diagnoses of glandular disease with an increase in both sensitivity and specificity for glandular lesions compared to conventional smears. These authors described the histopathologic findings among patients with ThinPrep Papanicolaou smears demonstrating AGCUS (n ⫽ 27), adenocarcinoma in situ (n ⫽ 2), and adenocarcinoma (n ⫽ 7). They found adenocarcinoma, adenocarcinoma in situ, carcinoma in situ, and miscellaneous squamous abnormalities in 22.2, 5.6, 8.3, and 11.1%, respectively [7]. Unfortunately, Ashfag et al. [7] grouped patients with different cytologic diagnoses together and did not analyze the histopathologic findings according to the individual type of cervical cytology. Recently, Papillo et al. [12] reported that, compared with conventional smears, the incidence of cytologic diagnosis of glandular cell
249
AGCUS ON THINPREP PAP
TABLE 4 Review of Reports on Atypical Glandular Cells of Undetermined Significance
Reference
Number of patients
Raab et al. [1] Kennedy et al. [2] Zweizig et al. [3] Duska et al. [4] Kim et al. [5] Cheng et al. [6]
116 77 85 73 268 75
Total/range (conventional Papanicolaou smear) Current study (ThinPrep Papanicolaou smear)
694 84
High-grade SIL (%) 12.1 5.2 7.1 21.9 9.0 19.0
Adenocarcinoma in situ (%) 0 3.9 1.2 0 1.9 3.0
5.2–21.9 8.3
0–3.9 3.6
Cervical cancer (%) 4.3 2.6 3.5 1.4 2.6 2.0 1.4–4.3 1.2
Endometrial cancer (%) 9.5 1.3 5.9 4.1 7.1 11.0 1.3–11.0 9.5
Note. SIL, squamous intraepithelial lesion. Endometrial cancer cases include some cases of endometrial hyperplasia or mixed mu¨llerian tumors.
abnormalities was reduced with ThinPrep but that histologically confirmed glandular disease was equivalent with the two techniques. The mean age of our study population (46.4) was similar to that reported in other studies [1–7]. In our study all women with AGCUS cytology and endometrial hyperplasia or cancer were older than 40 years and in the majority of them the origin of the atypical cells was identified as the endometrium. We found that these women were significantly older than those with cervical dysplasia or cancer (mean age 55.1 versus 43.5, respectively, P ⫽ 0.043). Similar findings were reported by Zweizig et al. [3]. One of our patients had a benign cervical polyp. Benign lesions reported in patients with AGCUS cytology include cervical and endometrial polyps, microglandular hyperplasia of the cervix, atypical squamous or tubal metaplasia of the cervix, endometritis, endocervicitis, cervical endometriosis, cervical or uterine myoma, Arias-Stella reaction, adenomyosis, ovarian endometrioma, and pelvic endometriosis [1, 4, 5]. AGCUS cytology can also occur as a result of repair in women who had cervical conization [13]. We found a higher incidence of significant histopathologic findings among women with smears favor neoplasia compared with those with smears favor reactive (34.9% versus 9.8%, respectively, P ⫽ 0.006). This finding is in agreement with other reports using conventional smears [4, 6]. Cheng et al. [6] found that patients with smears favor neoplasia had a significantly higher incidence of significant histopathologic findings compared with those with smears favor reactive and unspecified smears (72, 20, and 26%, respectively, P ⫽ 0.003). Duska et al. [4] reported that the subtype suggestive of reactive was a significant negative predictor of significant lesions. Raab et al. [1] identified the cytologic criteria of irregular nuclear membranes, atypical single cells, and decreased cytoplasm as useful in separating clinically significant from benign lesions. We found no extrauterine cancers among our patient population. Ovarian cancer, metastatic pancreatic, gastric, colon, and breast cancers have been described among women with
AGCUS cytology [2– 4, 6]. Patients with repeated AGCUS and normal colposcopic and endometrial histology should be offered pelvic ultrasound and followed up closely. Areas of strength in the current study include the formal review of the cytology and histologic slides by a single pathologist, subtyping the smears into favor reactive and favor neoplasia, comparison of the demographic features of women with smears favor reactive versus those with smears favor neoplasia, and calculation of the risk factors for significant histopathologic findings. Areas of weakness include its retrospective nature and its relatively small patient population. It is possible that with a larger study population different findings might have been observed. The most appropriate and cost-effective method for clinical evaluation of women with AGCUS has not been established. Some authors [2, 5] have recommended a full workup including colposcopy, biopsy, ECC, and endometrial biopsy for all women with this cytology. Others [3] have recommended the same workup for all patients with endometrial biopsy being offered only to women older than 35 years. Raab et al. [1] and Duska et al. [4] suggested that management could be based on subtyping the smears. Further prospective studies are needed to evaluate the appropriate diagnostic tests. Since preventing invasive cervical cancer by detecting highgrade dysplasia is the ultimate goal of Papanicolaou smears and considering the mounting costs of investigating and treating abnormal Papanicolaou smears, a triage system for women with AGCUS cytology is highly desirable. Understanding the limitations of retrospective studies in providing management algorithms, we believe that our study may provide some guidelines to help in the management of patients with this cytology on ThinPrep Papanicolaou smears. Among our patients with smears favor reactive only three (7.3%) and one (2.4%) patients had high-grade SIL and endometrial hyperplasia, respectively, and none had adenocarcinoma in situ, or cervical or endometrial cancers. In agreement with our findings, Cheng et al. [6] reported no cases of adenocarcinoma in situ or cervical or endometrial cancer among patients with smears favor reac-
250
ELTABBAKH ET AL.
tive (n ⫽ 27). Based on our findings and those of others [1, 4, 6] we recommend that AGCUS on ThinPrep smears be subtyped into favor reactive and favor neoplasia. Although it is tempting to suggest that women with favor reactive smears may be managed less aggressively than those with smears favor neoplasia, it is important to remember that 9.8% of patients with favor reactive smears were found to have significant histopathologic findings. We suggest that the guidelines for management of women with smears showing AGCUS favor reactive be established based on prospective studies. Based on our experience we suggest that all women with AGCUS favor neoplasia smears undergo colposcopy and that patients older than 40 years or with smears suggesting atypical cells of endometrial origin undergo endometrial biopsy. However, it is important to note that the criteria for subtyping AGCUS have not been thoroughly tested, particularly with monolayer technology (ThinPrep), and that management algorithms based on the subtyping of these smears are at least partially dependent on the cytopathologist’s experience [14]. Further prospective studies are needed to validate the cytologic criteria for subtyping smears with AGCUS cytology which might then allow management of women with AGCUS cytology to be based on the smear subtype. ACKNOWLEDGMENTS The authors thank Cheryl Gibson, M.D., Elizabeth Wegner, M.D., and Susan Smith, M.D., for allowing us to include their patients in the study.
REFERENCES 1. Raab SS, Isacson C, Layfield LJ, Lenel JC, Slagel DD, Thomas PA: Atypical glandular cells of undetermined significance. Cytologic criteria to separate clinically significant from benign lesions. Am J Clin Pathol 104:574 –582, 1995 2. Kennedy AW, Salmieri SS, Wirth SL, Biscotti CV, Tuason LJ, Travarca MJ: Results of the clinical evaluation of atypical glandular cells of
undetermined significance (AGCUS) detected on cervical cytology screening. Gynecol Oncol 63:14 –18, 1996 3. Zweizig S, Noller K, Reale F, Collis S, Resseguie L: Neoplasia associated with atypical glandular cells of undetermined significance on cervical cytology. Gynecol Oncol 65:314 –318, 1997 4. Duska LR, Flynn CF, Chen A, Whall-Strojwas D, Goodman A: Clinical evaluation of atypical glandular cells of undetermined significance on cervical cytology. Obstet Gynecol 91:278 –282, 1998 5. Kim TJ, Kim HS, Park CT, Park IS, Hong SR, Park JS, Shim JU: Clinical evaluation of follow-up methods and results of atypical glandular cells of undetermined significance (AGUS) detected on cervicovaginal Pap smears. Gynecol Oncol 73:292–298, 1999 6. Cheng RJ, Hernandez E, Anderson LL, Heller PB, Shanjk R: Clinical significance of a cytologic diagnosis of atypical glandular cells of undetermined significance. J Reprod Med 44:922–928, 1999 7. Ashfag R, Gibbons D, Vels C, Saboorian MH, Iliya F: ThinPrep Pap test. Accuracy for glandular disease. Acta Cytol 43:81– 85, 1999 8. Lee KR, Ashfaq R, Birdsong GG, Corkill ME, McIntosh KM, Inhorn SL: Comparison of conventional Papanicolaou smears and a fluid-based, thinlayer system for cervical cancer screening. Obstet Gynecol 90:278 –284, 1997 9. Papillo JL, Zarka MA, St. John TL: Evaluation of the ThinPrep Pap test in clinical practice. A seven-month, 16,314-case experience in northern Vermont. Acta Cytol 42:203–208, 1998 10. Lee KR, Minter LJ, Granter SR: Papanicolaou smear sensitivity for adenocarcinoma in situ of the cervix: a study of 34 cases. Am J Clin Pathol 107:30 –35, 1997 11. Kurman RJ, Henson DE, Herbst AL, Noller KL, Schiffman MH: Interim guidelines for management of abnormal cervical cytology. J Am Med Assoc 271:1866 –1869, 1994 12. Papillo JL, Warren SB, Zarka MA: Increased specificity in the detection of glandular lesions: decreased false-positive AGUS with ThinPrep Pap tests. Acta Cytol 43:902, 1999 (abstract) 13. Lee KR: Atypical glandular cells in cervical smears from women who have undergone cone biopsy: a potential diagnostic pitfall. Acta Cytol 37:705–709, 1993 14. Raab SS, Snider TE, Potts SA, McDaniel HL, Robinson RA, Nelson DL, Sigman JD, Thomas PA: Atypical glandular cells of undetermined significance. Diagnostic accuracy and interobserver variability using select cytologic criteria. Am J Clin Pathol 107:299 –307, 1997
Significance of Atypical Glandular Cells of Undetermined Significance on ThinPrep Papanicolaou Smears Gamal H. Eltabbakh, M.D.,* ,1 Jennifer N. Lipman,* Sharon L. Mount, M.D.,† and Ann Morgan, M.S.‡ *Department of Obstetrics and Gynecology, †Department of Pathology, and ‡Department of Biostatistics, University of Vermont College of Medicine, Burlington, Vermont 05401 Received February 22, 2000
epithelial cell abnormality termed atypical glandular cells of undetermined significance (AGCUS). Although recent reports [1–7] have described a high incidence of dysplasia and cancer among women with this cytology, our understanding of this group of Papanicolaou smear reporting and the most costeffective method of management of these women is still evolving. To address the problem of a high false-negative rate of Papanicolaou smear secondary to sampling or slide preparation, the ThinPrep (Cytec Corp., Boxborough, MA) slide preparation technique has been used recently [8]. This method has been shown to improve specimen adequacy and diagnostic yield, lower the percentage of atypical squamous or glandular cells of undetermined significance diagnoses, and increase the percentage of more accurate Papanicolaou smear diagnoses such as squamous intraepithelial lesions (SIL) [8, 9]. However, the accuracy of ThinPrep Papanicolaou smears in diagnosing glandular abnormalities has not been extensively studied. Conventional Papanicolaou smears have been reported to have a poor accuracy in diagnosing glandular lesions of the cervix [10]. Most of the reports investigating the significance of AGCUS cytology included patients whose smears were obtained by the conventional method [1– 6]. The significance of this cytologic diagnosis on Papanicolaou smears obtained by the ThinPrep method is not clear. The aim of the current study is to investigate the clinical significance of AGCUS on ThinPrep Papanicolaou smears and to study the factors that could predict significant histopathologic findings among these patients.
Objective. The aim of this study was to assess the incidence and risk factors predictive of significant histopathologic findings among women with atypical glandular cells of undetermined significance (AGCUS) on ThinPrep Papanicolaou smears. Methods. ThinPrep smears with AGCUS obtained between 1997 and 1999 were reviewed. Patients’ charts were reviewed and patients’ characteristics, follow-up information, and colposcopy and biopsy results were recorded. Pathologic slides were reviewed. The demographic features of women with favor reactive smears were compared with those with favor neoplasia and risk factors predictive of significant histopathologic findings (high-grade squamous intraepithelial lesion, endometrial hyperplasia, and cervical or endometrial cancers) were calculated. Results. The rate of diagnosis of AGCUS was 0.65%. Eightyfour patients with follow-up information were identified. The demographic features of women with smears favor neoplasia (n ⴝ 43) were similar to those with smears favor reactive (n ⴝ 41). The rates of incidence of any dysplasia or cancer and significant histopathologic findings were 32.1 and 22.6%, respectively, and were higher among women with smears favor neoplasia than among women with smears favor reactive (41.9% versus 22.0%, P ⴝ 0.051, and 34.9% versus 9.8%, P ⴝ 0.006, respectively). The subtype of cytology was the only factor that predicted significant histopathologic findings (odds ratio ⴝ 5.0, 95% confidence interval 1.6, 15.6, P < 0.010). Conclusions. In women with AGCUS on ThinPrep smears, significant histopathologic findings were found in 34.9% versus 9.8%, depending on the subtype of the smear (favor neoplasia vs reactive). Further studies are needed to validate the cytologic criteria for subtyping AGCUS smears and base management of women with AGCUS cytology on the subtype of the smear. © 2000 Academic Press
MATERIAL AND METHODS INTRODUCTION In our institution, the ThinPrep technique has been available for cervical smears since 1997. Through a computer search, we identified all patients with ThinPrep Papanicolaou smears demonstrating AGCUS seen at the Department of Obstetrics and Gynecology, University of Vermont/ Fletcher Allen Health Care, and the Womens Choice Clinic in Burlington, Vermont, between 1997 and August 1999. We excluded patients with
Implementation of the Bethesda system in Papanicolaou smear reporting has resulted in the creation of a new class of 1
To whom correspondence and reprint requests should be addressed at Division of Gynecologic Oncology, University of Vermont, MCHV Campus, 111 Colchester Avenue, Shepardson 330, Burlington, VT 05401. Fax: (802) 847-2360. E-mail: [email protected]. 245
0090-8258/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
246
ELTABBAKH ET AL.
similar cytology on conventional Papanicolaou smears. The ThinPrep Papanicolaou smears were originally diagnosed by a group of five American Board of Pathology certified pathologists using the Bethesda system criteria. These smears were then formally reviewed by one of the coauthors (SLM) to confirm the diagnosis of AGCUS. Since the distinction between AGCUS reactive and AGCUS neoplastic can show interobserver variability, the smears were evaluated by SLM to determine whether the original diagnosis was acceptable. Patients’ charts were reviewed and the following information was abstracted: age, gravidity, parity, menopausal status, use of hormonal replacement therapy, smoking, history of abnormal Papanicolaou smear and its treatment, history of human papillomavirus (HPV) infection by patient recall or previous Papanicolaou smears or biopsies, history of adenocarcinoma in situ, history of pelvic radiation, colposcopic findings, biopsy results, and follow-up Papanicolaou smears. In addition, histologic slides were reviewed by the same pathologist to confirm the histologic diagnosis. Patients were followed up with particular emphasis on the development of pelvic or extrapelvic cancers. Smoking was defined as having ever smoked for a period equal to or more than 1 year. Cervical smears were collected using the Cervex brush (Unimar, Inc., Wilton, CT). The brush was then rinsed into a vial containing 20 mL of preservative (PreservCyt) and the vial was capped and sent to the cytopathology laboratory where slides were processed using the ThinPrep processor. Cytologic diagnosis of AGCUS was based on established criteria [11]. The cells are of glandular type and display nuclear atypia that exceeds obvious reactive or reparative changes, but lack the unequivocal features of invasive carcinoma. The smears were classified as favor reactive when they exhibited inflammatory changes with nuclear enlargement associated with variability or irregularity of the shape of the nuclei and as favor neoplasia when the cells had more abnormal nuclei with hyperchromasia and overlapping. During the study period, management of patients with AGCUS has been rather consistent. Most patients with this cytology underwent colposcopy. Depending on the colposcopic findings, directed biopsies and endocervical curettage (ECC) were obtained. Patients with normal colposcopic findings, those with abnormal bleeding, and those whose cytology suggested endometrial origin underwent endometrial biopsy. Patients with persistent AGCUS cytology despite negative colposcopy and endometrial biopsy were offered pelvic ultrasound, hysteroscopy, and cone biopsy. We did not use HPV DNA testing as an adjunct to abnormal Papanicolaou smear in our study population. We compared patients with AGCUS favor neoplasia and those with smears favor reactive in demographic features and the incidence of any degree of dysplasia or cancer and the incidence of significant histopathologic findings. Significant histopathologic findings were defined as high-grade SIL, carcinoma in situ, cervical or endometrial cancers, and endome-
trial hyperplasia. Statistical analysis was performed using the two-sample t test, 2 test, and Fisher’s exact test. Assessment of the risk factors was performed by calculating the Mantel– Haentzel odds ratio and the 95% confidence interval. The statistical analysis was done using the SAS statistical package (SAS Institute, Cary, NC). Two-tailed P values ⬍0.05 were considered statistically significant. RESULTS We identified 86 patients with the confirmed cytologic diagnosis of AGCUS from 13,134 ThinPrep Papanicolaou smears obtained from patients seen between 1997 and August 1999. Thus, the proportion of women with AGCUS cytology was 0.65%. Review of the ThinPrep smears by SLM found no significant discrepancies from the original diagnoses. Although in individual cases the difference between an AGCUS favor reactive and an AGCUS favor neoplastic can be subjective, the original diagnoses were judged to be within reason. Forty-two (48.8%) patients had favor reactive and 44 (51.2%) had favor neoplasia smears. Of the 42 patients with smears favor reactive, 1 had no follow-up information, 5 were followed up with repeat Papanicolaou smears, and 36 had colposcopy. Two of 5 patients who had repeat Papanicolaou smears had abnormal smears and underwent colposcopy and biopsy revealing high-grade SIL in both. Of the 38 patients who underwent colposcopy, 35 underwent directed biopsy and ECC and 4 underwent endometrial biopsy. Of the 35 patients who had histologic diagnoses, 1 had high-grade SIL, 5 had low-grade SIL, and 1 had complex endometrial hyperplasia with atypia. Of the 44 patients with smears favor neoplasia, 1 had no follow-up information, 5 were followed up with repeat Papanicolaou smears, and 38 underwent colposcopy. Of the patients who underwent colposcopy, directed biopsies and ECC were obtained in 37 and endometrial biopsies were obtained in 10 patients. The histopathologic diagnoses among these patients were normal (n ⫽ 16), high-grade SIL (n ⫽ 3), low-grade SIL (n ⫽ 3), adenocarcinoma in situ (n ⫽ 1), small cell cervical carcinoma (n ⫽ 1), endometrial adenocarcinoma (n ⫽ 5), and mixed mu¨llerian tumor of the uterus (n ⫽ 1). Of the 5 patients who were followed up with repeat cytology, 3 had abnormal Papanicolaou smears. Among these 3 patients, colposcopy, cervical, and endometrial biopsies revealed highgrade SIL (n ⫽ 1), endometrial adenocarcinoma (n ⫽ 1), and cervical polyp (n ⫽ 1). The 5 patients who had no biopsies (3 with AGCUS favor reactive and 2 with AGCUS favor neoplasia) had normal repeat Papanicolaou smears. None of these patients developed cervical or vaginal dysplasia or gynecologic cancers. We excluded the 2 patients who had no follow-up information from further analysis. The demographic features of patients with AGCUS cytology are illustrated in Table 1. The mean (range) age of our patients was 46.4 (20 –79) years. There were no significant differences
247
AGCUS ON THINPREP PAP
TABLE 1 Patient Characteristics Atypical glandular cells of undetermined significance
Age in years (mean ⫾ SD) Gravidity (mean ⫾ SD) Parity (mean ⫾ SD) History of STD History of abnormal Papanicolaou smear History of treatment for abnormal smear History of dysplasia History of hysterectomy Postmenopausal Hormonal replacement therapy History of smoking History of HPV History of adenocarcinoma in situ History of radiation therapy
Favor reactive (%)
Favor neoplasia (%)
P value
All patients (%)
44.8 ⫾ 10.3 2.0 ⫾ 1.6 1.7 ⫾ 1.4 12.5 60.0 20.0 25.0 5.0 33.3 18.0 27.5 10.0 2.5 5.0
47.9 ⫾ 13.9 2.5 ⫾ 2.3 2.0 ⫾ 2.1 11.9 59.5 26.2 36.8 9.5 38.5 20.5 30.8 12.8 7.7 2.6
0.246 0.191 0.462 1.0 1.0 0.507 0.257 0.676 0.637 0.774 0.749 0.693 0.293 0.571
46.4 ⫾ 12.3 2.2 ⫾ 2.0 1.8 ⫾ 1.6 12.2 59.7 23.2 30.8 7.3 35.9 19.2 29.1 11.4 5.1 3.8
Note. STD, sexually transmitted diseases; HPV, human papillomavirus.
in the characteristics between patients with smears favor reactive and those with favor neoplasia. Approximately 60% of the study population had a history of abnormal Papanicolaou smears and 30% had a history of smoking. The overall incidence of cervical or vaginal dysplasia among our study patients was 17.9%. The rates of incidence of adenocarcinoma in situ and endometrial cancers were 3.6 and 8.3%, respectively. One (1.2%) patient had complex endometrial hyperplasia with atypia and one (1.2%) had small cell cervical carcinoma. The histology of the patients with endometrial cancers was endometrioid adenocarcinoma (n ⫽ 5), papillary serous adenocarcinoma (n ⫽ 1), and mixed mu¨llerian tumor (n ⫽ 1). All women who had endometrial hyperplasia or cancer were more than 40 years of age. These women were significantly older than those with cervical dysplasia or cancer (mean age ⫾ standard deviation: 55.1 ⫾ 11.96 versus 43.5 ⫾ 13.8, respectively, P ⫽ 0.043). As demonstrated in Table 2, the incidence of significant histopathologic findings was significantly higher among patients with smears favor neoplasia compared with those with smears favor reactive (34.9% versus 9.8%, respectively, P ⫽ 0.006). The incidence of any degree of dysplasia or cancer was higher among women with smears favor neoplasia compared to those with smears favor reactive (41.9% versus 22.0%, respectively, P ⫽ 0.051). Review of the smears identified a probable endometrial origin in 31 (36.9%), a probable endocervical origin in 33 (39.3%), both in 4 (4.8%), and neither in 16 (19.0%) patients. There was no significant difference in the incidence of significant histopathologic findings based on the origin of the smear (19.4, 21.2, 50.0, and 25.0%, respectively, P ⫽ 0.537). Review of the smears of the patients who had endometrial hyperplasia or cancer identified a probable endometrial origin in 5, a
probable endocervical origin in 1, a probable endometrial and endocervical origin in 1, and neither in 1 patient. The incidence of endometrial hyperplasia or cancer was higher if the atypical cells were of endometrial than if they were of endocervical origin (5/31 [16.1%] versus 1/33 [3.0%], respectively, P ⫽ 0.075). Assessment of risk factors predictive of significant histopathologic findings demonstrated that only the subtype of AGCUS cytology (reactive versus neoplasia) was significantly predictive of significant histopathologic findings. As demonstrated in Table 3, age, menopausal status, smoking, history of TABLE 2 Histopathologic Findings Atypical glandular cells of undetermined significance
Low-grade SIL High-grade SIL Any dysplasia Adenocarcinoma in situ Cervical cancer Endometrial hyperplasia Endometrial cancer Total with significant histopathology
Favor reactive (n ⫽ 41) (%)
Favor neoplasia (n ⫽ 43) (%)
All patients (n ⫽ 84) (%)
5 (12.2) 3 (7.3) 8 (19.5) 0 0 1 (2.4) 0
3 (7.0) 4 (9.3) 7 (16.3) 3 (7.0) 1 (2.3) 0 7 (16.3)
8 (9.5) 7 (8.3) 15 (17.9) 3 (3.6) 1 (1.2) 1 (1.2) 7 (8.3)
4 (9.8)
15 (34.9)*
19 (22.6)
Note. SIL, squamous intraepithelial lesion. Significant histopathology is defined as high-grade SIL, adenocarcinoma in situ, cervical cancer, endometrial hyperplasia, and endometrial cancer. * P ⫽ 0.006.
248
ELTABBAKH ET AL.
TABLE 3 Risk Factors for Significant Histopathologic Findings No. with dysplasia or cancer/total number (%)
Factor Age ⱕ46 ⬎46 Menopausal status Postmenopausal Premenopausal Smoking Yes No History of HPV Yes No History of abnormal Papanicolaou Yes No Subtype Neoplasia Reactive
Odds ratio
95% Confidence interval
P value
8/34 (23.5) 11/50 (22.0)
1.8
0.6, 5.2
0.30
9/31 (29.0) 10/51 (19.6)
1.7
0.6, 4.7
0.333
6/23 (26.1) 13/41 (31.7)
0.8
0.2, 2.4
0.641
2/8 (25.0) 18/76 (23.7)
1.1
0.2, 5.9
1.0
11/49 (22.4) 8/35 (22.9)
1.0
0.3, 2.8
0.721
15/43 (34.9) 4/41 (9.8)
5.0
1.6, 15.6
⬍0.010
Note. HPV, human papillomavirus. Data for menopausal status and smoking were not known for some patients.
HPV, and history of abnormal Papanicolaou smears were not predictive of significant histopathologic findings. Patients with smears favor neoplasia were five times more likely to have significant histopathologic findings compared with those with smears favor reactive (odds ratio ⫽ 5.0, 95% confidence interval 1.6, 15.6, P ⬍ 0.010). The power of the current study to detect a 20% difference between the two groups assuming a 10% incidence in the lower group was 66.2%. None of the patients enrolled in the study developed extrauterine cancers during the period of the study. DISCUSSION AGCUS is an uncommon cytologic diagnosis with a reported incidence between 0.08 and 0.74% among women who had conventional Papanicolaou smears. The incidence of this cytology on ThinPrep Papanicolaou smears among our patients (0.65%) was within the reported range using the conventional Papanicolaou smear. Ashfag et al. [7] reported that among women attending Parkland Memorial Hospital the incidence of AGCUS cytology on ThinPrep Papanicolaou smears was similar to that on conventional smears (0.14 and 0.11%, respectively). Although several studies [1– 6] have reviewed the clinical significance of AGCUS cytology on conventional Papanicolaou smears, our study focuses specifically on the clinical significance of the AGCUS diagnosis on ThinPrep smears. As demonstrated in Table 4, these studies found that the rates of incidence of high-grade SIL, adenocarcinoma in situ, cervical cancer, and endometrial hyperplasia or cancer were 5.2–21.9,
0 –3.9, 1.4 – 4.3, and 1.3–11.0%, respectively. To our knowledge, the current study is the first to report the histopathologic findings among women with AGCUS on ThinPrep Papanicolaou smears as a separate entity. The incidence of high-grade SIL, adenocarcinoma in situ, cervical cancer, and endometrial cancer or hyperplasia reported in our study (8.3, 3.6, 1.2, and 9.5%, respectively) falls within the reported ranges in studies using conventional Papanicolaou smears [1– 6]. Comparing our results with the histopathologic findings among women with AGCUS cytology on ThinPrep smears to those with historical controls of women with the same cytology on conventional smears [1– 6] reveals similar findings. Compared with conventional smears, we did not find AGCUS diagnosis on ThinPrep smears to be associated with a higher incidence of significant histopathologic findings. Ashfag et al. [7] found that ThinPrep Papanicolaou smears provided more accurate diagnoses of glandular disease with an increase in both sensitivity and specificity for glandular lesions compared to conventional smears. These authors described the histopathologic findings among patients with ThinPrep Papanicolaou smears demonstrating AGCUS (n ⫽ 27), adenocarcinoma in situ (n ⫽ 2), and adenocarcinoma (n ⫽ 7). They found adenocarcinoma, adenocarcinoma in situ, carcinoma in situ, and miscellaneous squamous abnormalities in 22.2, 5.6, 8.3, and 11.1%, respectively [7]. Unfortunately, Ashfag et al. [7] grouped patients with different cytologic diagnoses together and did not analyze the histopathologic findings according to the individual type of cervical cytology. Recently, Papillo et al. [12] reported that, compared with conventional smears, the incidence of cytologic diagnosis of glandular cell
249
AGCUS ON THINPREP PAP
TABLE 4 Review of Reports on Atypical Glandular Cells of Undetermined Significance
Reference
Number of patients
Raab et al. [1] Kennedy et al. [2] Zweizig et al. [3] Duska et al. [4] Kim et al. [5] Cheng et al. [6]
116 77 85 73 268 75
Total/range (conventional Papanicolaou smear) Current study (ThinPrep Papanicolaou smear)
694 84
High-grade SIL (%) 12.1 5.2 7.1 21.9 9.0 19.0
Adenocarcinoma in situ (%) 0 3.9 1.2 0 1.9 3.0
5.2–21.9 8.3
0–3.9 3.6
Cervical cancer (%) 4.3 2.6 3.5 1.4 2.6 2.0 1.4–4.3 1.2
Endometrial cancer (%) 9.5 1.3 5.9 4.1 7.1 11.0 1.3–11.0 9.5
Note. SIL, squamous intraepithelial lesion. Endometrial cancer cases include some cases of endometrial hyperplasia or mixed mu¨llerian tumors.
abnormalities was reduced with ThinPrep but that histologically confirmed glandular disease was equivalent with the two techniques. The mean age of our study population (46.4) was similar to that reported in other studies [1–7]. In our study all women with AGCUS cytology and endometrial hyperplasia or cancer were older than 40 years and in the majority of them the origin of the atypical cells was identified as the endometrium. We found that these women were significantly older than those with cervical dysplasia or cancer (mean age 55.1 versus 43.5, respectively, P ⫽ 0.043). Similar findings were reported by Zweizig et al. [3]. One of our patients had a benign cervical polyp. Benign lesions reported in patients with AGCUS cytology include cervical and endometrial polyps, microglandular hyperplasia of the cervix, atypical squamous or tubal metaplasia of the cervix, endometritis, endocervicitis, cervical endometriosis, cervical or uterine myoma, Arias-Stella reaction, adenomyosis, ovarian endometrioma, and pelvic endometriosis [1, 4, 5]. AGCUS cytology can also occur as a result of repair in women who had cervical conization [13]. We found a higher incidence of significant histopathologic findings among women with smears favor neoplasia compared with those with smears favor reactive (34.9% versus 9.8%, respectively, P ⫽ 0.006). This finding is in agreement with other reports using conventional smears [4, 6]. Cheng et al. [6] found that patients with smears favor neoplasia had a significantly higher incidence of significant histopathologic findings compared with those with smears favor reactive and unspecified smears (72, 20, and 26%, respectively, P ⫽ 0.003). Duska et al. [4] reported that the subtype suggestive of reactive was a significant negative predictor of significant lesions. Raab et al. [1] identified the cytologic criteria of irregular nuclear membranes, atypical single cells, and decreased cytoplasm as useful in separating clinically significant from benign lesions. We found no extrauterine cancers among our patient population. Ovarian cancer, metastatic pancreatic, gastric, colon, and breast cancers have been described among women with
AGCUS cytology [2– 4, 6]. Patients with repeated AGCUS and normal colposcopic and endometrial histology should be offered pelvic ultrasound and followed up closely. Areas of strength in the current study include the formal review of the cytology and histologic slides by a single pathologist, subtyping the smears into favor reactive and favor neoplasia, comparison of the demographic features of women with smears favor reactive versus those with smears favor neoplasia, and calculation of the risk factors for significant histopathologic findings. Areas of weakness include its retrospective nature and its relatively small patient population. It is possible that with a larger study population different findings might have been observed. The most appropriate and cost-effective method for clinical evaluation of women with AGCUS has not been established. Some authors [2, 5] have recommended a full workup including colposcopy, biopsy, ECC, and endometrial biopsy for all women with this cytology. Others [3] have recommended the same workup for all patients with endometrial biopsy being offered only to women older than 35 years. Raab et al. [1] and Duska et al. [4] suggested that management could be based on subtyping the smears. Further prospective studies are needed to evaluate the appropriate diagnostic tests. Since preventing invasive cervical cancer by detecting highgrade dysplasia is the ultimate goal of Papanicolaou smears and considering the mounting costs of investigating and treating abnormal Papanicolaou smears, a triage system for women with AGCUS cytology is highly desirable. Understanding the limitations of retrospective studies in providing management algorithms, we believe that our study may provide some guidelines to help in the management of patients with this cytology on ThinPrep Papanicolaou smears. Among our patients with smears favor reactive only three (7.3%) and one (2.4%) patients had high-grade SIL and endometrial hyperplasia, respectively, and none had adenocarcinoma in situ, or cervical or endometrial cancers. In agreement with our findings, Cheng et al. [6] reported no cases of adenocarcinoma in situ or cervical or endometrial cancer among patients with smears favor reac-
250
ELTABBAKH ET AL.
tive (n ⫽ 27). Based on our findings and those of others [1, 4, 6] we recommend that AGCUS on ThinPrep smears be subtyped into favor reactive and favor neoplasia. Although it is tempting to suggest that women with favor reactive smears may be managed less aggressively than those with smears favor neoplasia, it is important to remember that 9.8% of patients with favor reactive smears were found to have significant histopathologic findings. We suggest that the guidelines for management of women with smears showing AGCUS favor reactive be established based on prospective studies. Based on our experience we suggest that all women with AGCUS favor neoplasia smears undergo colposcopy and that patients older than 40 years or with smears suggesting atypical cells of endometrial origin undergo endometrial biopsy. However, it is important to note that the criteria for subtyping AGCUS have not been thoroughly tested, particularly with monolayer technology (ThinPrep), and that management algorithms based on the subtyping of these smears are at least partially dependent on the cytopathologist’s experience [14]. Further prospective studies are needed to validate the cytologic criteria for subtyping smears with AGCUS cytology which might then allow management of women with AGCUS cytology to be based on the smear subtype. ACKNOWLEDGMENTS The authors thank Cheryl Gibson, M.D., Elizabeth Wegner, M.D., and Susan Smith, M.D., for allowing us to include their patients in the study.
REFERENCES 1. Raab SS, Isacson C, Layfield LJ, Lenel JC, Slagel DD, Thomas PA: Atypical glandular cells of undetermined significance. Cytologic criteria to separate clinically significant from benign lesions. Am J Clin Pathol 104:574 –582, 1995 2. Kennedy AW, Salmieri SS, Wirth SL, Biscotti CV, Tuason LJ, Travarca MJ: Results of the clinical evaluation of atypical glandular cells of
undetermined significance (AGCUS) detected on cervical cytology screening. Gynecol Oncol 63:14 –18, 1996 3. Zweizig S, Noller K, Reale F, Collis S, Resseguie L: Neoplasia associated with atypical glandular cells of undetermined significance on cervical cytology. Gynecol Oncol 65:314 –318, 1997 4. Duska LR, Flynn CF, Chen A, Whall-Strojwas D, Goodman A: Clinical evaluation of atypical glandular cells of undetermined significance on cervical cytology. Obstet Gynecol 91:278 –282, 1998 5. Kim TJ, Kim HS, Park CT, Park IS, Hong SR, Park JS, Shim JU: Clinical evaluation of follow-up methods and results of atypical glandular cells of undetermined significance (AGUS) detected on cervicovaginal Pap smears. Gynecol Oncol 73:292–298, 1999 6. Cheng RJ, Hernandez E, Anderson LL, Heller PB, Shanjk R: Clinical significance of a cytologic diagnosis of atypical glandular cells of undetermined significance. J Reprod Med 44:922–928, 1999 7. Ashfag R, Gibbons D, Vels C, Saboorian MH, Iliya F: ThinPrep Pap test. Accuracy for glandular disease. Acta Cytol 43:81– 85, 1999 8. Lee KR, Ashfaq R, Birdsong GG, Corkill ME, McIntosh KM, Inhorn SL: Comparison of conventional Papanicolaou smears and a fluid-based, thinlayer system for cervical cancer screening. Obstet Gynecol 90:278 –284, 1997 9. Papillo JL, Zarka MA, St. John TL: Evaluation of the ThinPrep Pap test in clinical practice. A seven-month, 16,314-case experience in northern Vermont. Acta Cytol 42:203–208, 1998 10. Lee KR, Minter LJ, Granter SR: Papanicolaou smear sensitivity for adenocarcinoma in situ of the cervix: a study of 34 cases. Am J Clin Pathol 107:30 –35, 1997 11. Kurman RJ, Henson DE, Herbst AL, Noller KL, Schiffman MH: Interim guidelines for management of abnormal cervical cytology. J Am Med Assoc 271:1866 –1869, 1994 12. Papillo JL, Warren SB, Zarka MA: Increased specificity in the detection of glandular lesions: decreased false-positive AGUS with ThinPrep Pap tests. Acta Cytol 43:902, 1999 (abstract) 13. Lee KR: Atypical glandular cells in cervical smears from women who have undergone cone biopsy: a potential diagnostic pitfall. Acta Cytol 37:705–709, 1993 14. Raab SS, Snider TE, Potts SA, McDaniel HL, Robinson RA, Nelson DL, Sigman JD, Thomas PA: Atypical glandular cells of undetermined significance. Diagnostic accuracy and interobserver variability using select cytologic criteria. Am J Clin Pathol 107:299 –307, 1997