A clinical review of borderline glandular cells on cervical cytology

British Journal of Obstetrics and Gynaecology May 2000, Vol07, pp. 605-609 A clinical review of borderline glandular cells on cervical cytology D. Kusnah A. Mohammed Research Senior House Oflcer, Ofer Lavie Subspecialty Fellow, Albert0 de B. Lopes Consultant (Gynaecological Oncology), Paul Cross Consultant (Pathology), John M. Monaghan Senior Lecturer (Gynaecological Oncology) Gynaecological Oncology Centre, Queen Elizabeth Hospital, Gateshead Objective To review the diagnoses and diagnostic pathway of women presenting with borderline glandular cells on cervical cytology. To outline the basis of clinical approach of these women. Design Retrospective review. Population Forty-three women referred to the hospital department over a 32-month period. Methods Review of the casenotes for the demographic data, previous cervical cytology andor histology report, indication for the smear resulting in borderline glandular cells, colposcopic findings, diagnostic and/or treatment procedures, final diagnosis and current status. Results The average age was 36.7 years. Twenty-four women (56%) had clinically significant lesions: seven women (6%) presented with cancers, of which one was endometrial in origin, and 7 (40%) with intraepithelial neoplasia (CIN and cervical glandular intraepithelial neoplasia (CGIN)). Sixtyseven percent of all clinically significant lesions were of squamous origin. Thirty-seven had histological diagnosis, while six went on to cytological surveillance. Colposcopy was the most significant predictor for clinically significant lesions (P < 0-05). Punch biopsies and loop excisions were diagnostic when based on abnormal colposcopic findings. Brush cytology was appropriate follow up for asymptomatic, premenopausal women with no colposcopic abnormality. In addition, endometrial sampling was recommended in the peri- and postmenopausal women. Conclusion Borderline glandular cells have a high incidence of clinically significant lesions. Immediate referral for colposcopy and assessment is strongly recommended in women with two borderline glandular smears to avoid delays in potential cancer diagnosis. INTRODUCTION Abnormal glandular cytology on routine smears raises the possibility of a malignancy (of cervical and extracervical origin), cervical intraepithelial neoplasia (which may require treatment) or a benign diagnosis which usually does not require treatment. To identify those women needing treatment, a search for the origin of these cytologically abnormal cells is essential. There is no standard guideline for the clinical management of women with abnormal glandular cytology. The variable clinical approach is attributable to the lack of consistency in the interpretation and reporting of abnormal glandular cells, the spectrum of conditions associated with it and the variable experience each unit has in dealing with this cytology report. Kennedy et al. recommended immediate referral for colposcopy, endocervical curettage and endometrial Correspondence: Dr D. K. A. Mohammed, Gynaecological Oncology Centre, Queen Elizabeth Hospital, Gateshead NE9 6SX, UK. biopsy of women with atypical glandular cells of undetermined significance, with increased cytological surveillance during the subsequent two years if no cause was found. From the 29 cases with abnormal glandular cells reviewed by Jackson et al., 45% had glandular neoplasia, of which only one was invasive, and colposcopy was recommended for the assessment of coexisting squamous lesions and a diagnostic cone biopsy for glandular lesions. The term borderline glandular cells as used in this study is based on the British Society of Cervical Cytology/NHS Cervical Screening Programme (BSCC/ NHSCSP) national guidelines criteria3. These cells exhibit borderline changes in their nuclear size, shape and staining properties. They include irregular nuclear membrane, which is not due to distortion by a cytoplasmic vacuole, irregular coarse grainy chromatin and hyperchromasia with intercellular variation in the depth of nuclear staining. Borderline glandular cells as a group have a disordered three dimensional arrangement losing the normal honeycomb appearance. In reporting Q RCOG 2000 British Journal of Obstetrics and Gynaecology 605

606 D. K. A. MOHAMMED ET AL. borderline glandular cells, the cytologist, in the absence of any other cytological abnormality including squamous cell changes, has genuine doubt as to whether the cells are pre-malignant or not. Based on the NHSCSP ABC publication4, the abnormal glandular cytology of this study is of borderline category and not of glandular neoplastic category, the latter representing definite glandular dyskaryosis. Leeson et al5 found, on review of 89 women with glandular dyskaryosis, 22 endometrial cancers (24-7%), cervical cancers (%), 5 cases of cervical intraepithelial neoplasia (6.8%), one adenocarcinoma in situ, and one vaginal intraepithelial neoplasia. The average age of this group was 54.3 years and approximately one-third presented with postmenopausal bleeding. This study recommended endometrial sampling and colposcopy for all at first presentation of this cytological grade. The inter-observer and inter-laboratory variation in the recognition and reporting of borderline glandular cells is well recognised. Guidelines published by a working party in the United Kingdom3 for the recognition and management of borderline nuclear changes in both endocervical and squamous cells on cervical cytology were aimed to reduce the cytological inconsistencies and to provide guidelines on follow up and colposcopy referral for these women. We reviewed the clinical management and outcome of women with borderline glandular cytology within our department over a 32-month period. This descriptive study aims to:. Outline the pathology associated with borderline glandular cells; 2. Describe the diagnostic pathway of our study group; and 3. Outline the basis of a clinical approach to women with borderline glandular cells. METHODS The Cervical Cytology Screening Department at the Queen Elizabeth Hospital in Gateshead provides the cervical screening services for the Gateshead and South Tyneside areas. Over 3,000 cervical smears a year are screened, of which 0.07% are reported as borderline glandular cells. The recommendation for colposcopy referral is made after no more than two borderline glandular cytologies. Some women are referred after the first borderline glandular cytology if they have had previous abnormal cytology or are symptomatic. Borderline smears are repeated at three to six month intervals and referral for colposcopy is advised if they are persistent. Between January 995 and August 997, there were 48 women seen with borderline glandular smears. They were identified from the cervical cytology database. Five women were excluded due to unavailability of casenotes in three cases, coexisting severe squamous dyskaryosis reported in one case and recurrent borderline glandular cells during pregnancy in one woman, which had already been treated elsewhere. This study reviewed the remaining 43 women who presented with borderline glandular cytology during this period. The casenotes of these women were reviewed to obtain the following data: age, menopausal status, the use of exogenous hormones, the use of intrauterine devices, previous cervical treatment, indication for taking the initial smear resulting in the report of borderline glandular cells, previous abnormal histology/cytology, colposcopic findings, procedure performed at first assessment, provisional diagnosis, subsequent procedures, final diagnosis and current status. The cytology slides were not reviewed for the purpose of this study, using the initial pathologist issued cytological report. A x2 test (SPSS 8.0 for Windows) is used to correlate significant disease (defined as all grades of intraepithelial neoplasia or worse) outcome with age, menopausal status, use of exogenous hormones, history of abnormal cytology or histology and colposcopic findings. P < 0-05 is taken to be significant. RESULTS The women s average age was 36.7 years, ranging between 24 and 63 years. Of the 43 women in the study group, 37 (86%) were premenopausal, three (7%) were perimenopausal (women with irregular cycles and menopausal symptoms) and three (7%) were postmenopausal. Six were using oral contraception, three were on hormone replacement therapy, two were on depo provera, two were using intrauterine devices and 30 were using either no contraception or non hormonal methods of contraception. All the smears included in this study were reported as borderline glandular cells, in the absence of a squamous or a higher grade abnormality. Five women had had previous cervical treatments. They were in the form of laser for CIN III, cone biopsy for CIN II, cryotherapy for postcoital bleeding and large loop excision of the transformation zone (LLETZ) in two cases for CIN II. The indications for smears in this study group were identified from the referral letters, the cytology forms or previous records. They were routine screening in 7 cases (39.5%), follow up of an abnormal smear in 3 cases (30.2%), on yearly recall from previous abnormal cytology or histology in eight cases (8.6%), overdue in two cases (4.6%), irregular bleeding in one case (%), following loop excision in one case and not recorded in another. Thirty-seven women had histological diagnoses, and the remaining six went on to cytological surveillance. 0 RCOG 2000 Br J Obstet Gynaecol 07,605-609

BORDERLINE GLANDULAR CELLS 607 There were seven cancers (6%), 7 intraepithelial neoplasia (40%) and 9 insignificant abnormalities (44%). Table details the final diagnoses of these women. All patients underwent a colposcopic examination at their first visit. The colposcopic findings were normal with visible squamo-columnar junction in eight women (8.6%) and unsatisfactory in four (9.3%) in whom this junction was not seen. Acetowhite epithelium was seen in 9 women (44.2%) and in five (.6%), there was associated mosaicism or punctation. Atypical vessels were noted in five women (.6%) and in two (4.6%) there was an obvious invasive lesion seen. All six women presenting with cervical cancers had abnormal colposcopic findings, of which two had obvious visible lesions, while one early stromal invasion was diagnosed on LLETZ following CIN III on punch biopsy. The woman with endometrial cancer had a LLETZ for atypical vessels and at the same time, an endometrial sampling was diagnostic of endometrial adenocarcinoma. Of the 7 women with CIN, 5 had colposcopic abnormalities and two had unsatisfactory colposcopy with no abnormality on the ectocervix. The diagnoses were made in the latter two women as a result of persistent cytological abnormality in one and from a diagnostic LLETZ in the other. Of the remaining 9 women with human papillomavirus (HPV), inflammation or no abnormality, nine had colposcopic abnormalities, eight had normal satisfactory colposcopy and two had unsatisfactory colposcopy. Forty-two women had other investigative procedures performed at the time of the initial colposcopy (Table 2). Eleven women had two or more procedures performed. The remaining woman had a normal and satisfactory colposcopy and was followed up by cytology. Table. Final diagnosis of the study group. Diagnosis Frequency Percentage Malignant Adenocarcinoma Adenosquamous carcinoma Endometrial adenocarcinoma Squamous cell carcinoma Microinvasive squamous cell Intraepithelial neoplasia High grade CGIN CIN I CIN I Other wart virus Inflammation No abnormality TOTAL 7 3 7 3* 2 2+ 9 6 3 O* 43 6.2 39.6 27.9 4.7 44.2 3.9 23.3 00.0 *Includes one case of hysterectomy specimen diagnosis. +Includes one case that had a loop excision for a punch biopsy CIN III. *Six cases by cytology, four cases by histology. Table 2. Procedures at the first visit. LLETZ = large loop excision of the transformation zone. Procedures LLETZ Repeat smear Punch biopsy Endornetrial sampling Endocervical curettage No. 2 3 3 7 LLETZ was performed in 2 women, including three with normal and satisfactory colposcopy. There were three adenocarcinomas, one high grade CGIN, eight CIN III, one CIN I and eight with negative histology. None of the women with normal and satisfactory colposcopy showed any abnormality on histology. A repeat smear was performed in 3 women at the time of the initial assessment, eight as the only procedure and five with other procedures. Of these women, 0 had satisfactory colposcopy and three had unsatisfactory colposcopy. Of the 0 with visible squamocolumnar junction five had no abnormalities, while the other five had acetowhite epithelium. The cytology was negative in nine of the 3, abnormal in three and unsatisfactory in one. Discrepancy between cytology and histology occurred in one out of four who had a histological diagnosis at the same time. The cytology was reported as negative in the woman with early stromal invasion. Thirteen women had colposcopically -directed punch biopsies. There were two obvious cancers, six CIN, one high grade CGIN and the remaining four had no significant abnormality. The two cancer cases had lesions seen on the cervix and the remaining women had acetowhite areas, of which four also had mosaicism and atypical vessels. Endometrial sampling was performed in seven women who were between 3 and 63 years old, none of whom had symptoms recorded. Four of these women were premenopausal, one was perimenopausal and two were postmenopausal. All samples were adequate for diagnosis. The only endometrial carcinoma in our series was in the oldest woman, who was 63 years of age. There was only one woman who had endocervical curettage which was reported as normal. Fourteen (32.6%) women underwent definitive treatment following their initial assessment. Of the six cervical cancers, five had radical surgery and the woman with early stromal invasion was managed by LLETZ. The woman with endometrial cancer was treated by hysterectomy. LLETZ was performed for CIN III in six cases. Another woman had hysterectomy for high grade CGIN following recurrent borderline glandular cytology. 0 RCOG 2000 Br J Obstet Gynaecol 07,605409

608 D. K. A. MOHAMMED ET AL. The mean duration of follow up is 0.8 months, ranging between 3 to 4 months. Disease status was based on repeat cytology and colposcopy in the six women without histological diagnoses, of whom four had normal and satisfactory colposcopy and two had unsatisfactory colposcopy. All six had at least one negative brush smear and three have had two consecutive negative smears following the borderline glandular cytology. Forty-two women are currently disease free and one woman, who had wart virus changes on loop histology and a borderline glandular post loop smear, has defaulted further follow up. Using a x2 test to correlate significant disease outcome of the study group with age, menopausal status, the use of exogenous hormones and history of previous abnormal, cytology or histology did not reveal any statistically significant association. However, an abnormal colposcopic finding was found to be significantly (P < 0.05) correlated to the final diagnosis, with Yates correction applied to this small study group. DISCUSSION Most reports on atypical glandular cells in the literature describe atypical glandular cells of undetermined significance. Clinically significant lesions were found in between 7% and 50% ofcases and the majority were found to be squamous cervical lesion^'*^."^. In our study group 57% of cases had clinically significant lesions: 6% invasive disease, 33% CIN and 7% high grade CGIN. The variation in the proportion and nature of clinically significant lesions between gynaecological units again highlights the inconsistencies in the terminology and definition of abnormal glandular cells between cytologists and between laboratories. The concomitant reporting of squamous dyskaryosis by some laboratories, the influence of sampling methods and slide preparation or a true geographical variation can perpetuate this dis~repancy~~ ~. Based on the Bethesda system, atypical glandular cells of undetermined significance are comparable to borderline endocervical cells (abnormal glandular cells) used in the British Society of Cervical Cytology terminology. The Bethesda system further classifies, when possible, the atypical glandular cells of undetermined significance to suggest their origin and clinical significance. This classification appears to be useful for identifying significant lesions albeit preinvasive or invasive lesions of the cervix, endometrium or other origins. Several authors have reported on ways of improving the cytological interpretation of borderline nuclear changes to distinguish those of benign from malignant origins. Raab et al identified irregular nuclear membranes, atypical single cells and decreased cytoplasm in the atypical glandular cells to be important cytological criteria in distinguishing benign from clinically significant cases. However, these criteria were found to be most effective when combined with the experience of the cytologist 2. Bose et az.i3 also found further cytological criteria to categorise the lesions associated with abnormal endocervical cells and found on review of the slides that some abnormal endocervical cells are actually either reactive cells or metaplastic cells. Abnormal glandular cells can arise from several malignancies, including cervical carcinoma (0-2-6-8%), endometrial carcinoma (0.24%) and metastatic disease from the ovaries, fallopian tube, colon and pan- ~reas..~.~.~. In comparison, we have a higher rate of cervical malignancy at 3.9% but an average % for endometrial malignancy. The rate of intraepithelial neoplasia in our study group is comparable to the published data. We had 7% of CGIN compared with 2-8% and 33% of CIN compared with 940% from other author^'*^^^*^. Lee at al4 found that the cytological distinction of squamous cells from borderline glandular cells was not always possible and used this to explain the higher rate of CIN compared with CGIN. Benign lesions which can be associated with the presence of abnormal glandular cytology include microglandular hyperplasia, inflammation, tuboendometrial metaplasia, cervical endometriosis, sampling of lower uterine segment post cone biopsy, human papillomavirus and endometrial hyperplasia 2 * 3 * 7 5-7. only inflammation and human papillomavirus effects were noted in 7% and 4% of our series. The presence of polyps and the use of an intrauterine contraceptive device can also be associated with abnormal glandular cells3, although these were not demonstrated in our study population. Different approaches and management of these women are adopted by different units relating to the range of differential diagnoses and the experience of the individual unit when dealing with abnormal glandular cytology. Referral for colposcopy is recommended after a persistent borderline abnormality in the United Kingdom. We found colposcopic findings to be the most important predictive factor of significant disease in this small series of patients. A loop excision, when performed in the absence of colposcopic abnormality and a visible transformation zone, did not reveal any disease in our group of patients. Therefore, we would recommend that if colposcopy is normal and satisfactory, then adequate sampling of the endocervical canal with a cytobrush would be an acceptable step in an asymptomatic, premenopausal women. The use of LLETZ at the first visit is justifiable when a high grade lesion is seen colposcopically, especially if its upper limit is not defined, and in unsatisfactory colposcopy, if a repeat smear 0 RCOG 2000 Br J Obstet Gynaecol 07,605-609

BORDERLINE GLANDULAR CELLS 609 continues to be abnormal then a diagnostic loop is required. In the presence of small areas or mild abnormalities on colposcopy, a punch biopsy may be adequate for diagnosis. Cone biopsy was advocated by Jackson et al2 for assessing the endocervix, but endocervical curettage is a more popular method in North Americaq6. Dusk et al6 commented that endocervical curettage was unyielding, being positive in only one of 60 women. We routinely use the cytobrush as an alternative to adequately sample the endocervix. Chakrabarti et al8 showed that a cytobrush samples endocervical cells better than a spatula, but that it has a tendency of upgrading the cytological grade. In the peri- and postmenopausal group, especially those with symptoms, we would recommend a pipelle sampling of the endometrium in this group of women at higher risk of endometrial malignancy. Zweizig et al9, however, found vaginal bleeding not indicative of significant lesions and suggested endometrial biopsy in all women over the age of 35 years. Kennedy et al.', on the other hand, suggested endometrial biopsy in all women as part of their primary assessment, while Duska et al6 recommended it in the postmenopausal group only, in which all the endometrial carcinomas in his review were found. As a result of our findings, our protocol for management of women referred with borderline glandular cytology is as follows:. Colposcopy by an experienced accredited colposcopist. 2. If colposcopy is normal and the squamo-columnar junction is seen, repeat cytology with spatula and brush is performed. If negative, two further reviews by cytology and colposcopy is necessary prior to discharge. A diagnostic loop should be performed if the cytological abnormality persists. 3. If colposcopy is normal but the squamo-columnar junction is not seen, our experience from four women in this study would lead us to believe that cytological follow up as described under 2. (above) would be a reasonable approach. 4. Punch biopsy for small areas or mild abnormalities. 5. Loop excision for high grade lesion detected colposcopically, especially when the upper limit is not seen. 6. Pipelle endometrial sampling in the peri- and postmenopausal women. A woman referred with a borderline glandular smear in our series had a one in two chance of having clinically significant disease. This is greater than for its squamous counterpart. A stepwise rather than a blanket approach in these women is preferable. There is a danger that an over zealous approach may result in unnecessary interventions that are not cost effective and potentially harmful. References 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 Oncoll996; 63: 4-8. 2 Jackson SR, Hollingsworth TA, Anderson MC, Johnson J, Hammond RH. Glandular lesions of the cervix-cytological and histological correlation. Cytopathol 996; 7: 0-6. 3 Buckley CH, Herbert A, Johnson J et al. Borderline nuclear changes in cervical smears: guidelines on their recognition and management. J Clin Path 994; 47: 48492. 4 Working Party of the Royal College of Pathologists BSCC, NHSCSP. Achievable standards, benchmarks for reporting & criteria for evaluating cervical cytopathology. NHSCSP Publication 995: 5. 5 k son SC, Inglis TCM, Salman WD. A study to determine the underlying reason for abnormal glandular cytology and the formulation of a management protocol. Cytopathol 997; 8: 20-26. 6 Duska LR, Flynn CF, Chen A, Whall-Strojwas D, Goodman A. Clinical evaluation of atypical glandular cells of undetermined significance on cervical cytology. Obster Gynecoll998; 9: 278-282. 7 Goff BA, Atanasoff P, Brown E, Muntz HG, Bell DA, Rice LW. Endocervical glandular atypia in Papanicolaou smears. Obsfer Gynecol 992; 79 0-04. 8 Eddy GL, Strumpf KB, Wojtowycz MA, Piraino PS, Mazur MT. Biopsy findings in five hundred thirty one patients with atypical galndular cells of uncertain significance as defined by the Bethesda system. Am J Obstet Gynecoll997; 77: 88-95. 9 Zweizig S, Noller K, Reale F, Collis S, Resseguie L. Neoplasia associated with atypical glandular cells of undetermined significance on cervical cytology. Gynecol Oncoll997; 65: 34-3 8. 0 Fiorella RM, Casafrancisco D, Yojota S, Krangel PJ. Artifactual endocervical atypia induced by endocervical brush collection. Diagnostic Cytopatholl994; : 79-84. 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 Putholl995; 04: 57&582. 2 Raab SS, Snider TE, Potts SA et al. Atypical glandular cells of undetermined significance: diagnostic accuracy and interobserver variability using select cytologic criteria. Am J Clin Parholl997; 07: 299-307. 3 Bose S, Kannan V, Kline TS. Abnormal endocervical cells: really abnormal? Really endocervical?arn J Clin Parholl994; 0: 708-73. 4 Lee KR, Manna EA, StJohn T. Atypical endocervical cells: accuracy of cytologic diagnosis. Diagnostic Cytopatholl995; 3: 202-208. 5 Heaton RE3, Harris TF, Larson DM, Henry MR. Glandular cells derived from direct sampling of the lower uterine segment in patients status post-cervical cone biopsy: a diagnostic dilemma. Am J Clin Patholl996; 06: 5-56. 6 Hanau CA, Begley N, Bibbo M. Cervical endornetiosis: a potential pitfall in the evaluation of glandular cells in cervical smears. Diagnostic Cyroparholl997; 6 274-280. 7 k KR. Atypical glandular cells in cervical smears from women who have undergone cone biopsy: a potential diagnostic pitfall. Acta Cytologica 993; 37: 705-709. 8 Chakrabarti S, Fernando FB, Parakevas M. Brush vs spatula for cervical smears: histologic correlation with concurrent biopsies. Acta Cytologica 994; 38: 35-38. Accepted 7 December 999 0 RCOG 2000 Br J Obstet Gynaecol 07,605-609