Origin, Frequency and Clinical Significance of Glandular Cells in Liquid-Based Pap Tests from Patients Posthysterectomy
Risha B. Ramdall, M.D., Robert C. Wallach, M.D., Joan Cangiarella, M.D., Guoping Cai, M.D., Paul Elgert, C.T.(ASCP), C.M.I.A.C., Aylin Simsir, M.D., and Pascale Levine, M.D.
To quantify the frequency and significance of glandular cells in posthysterectomy liquid-based (SurePath, TriPath Imaging, Burlington, North Carolina, U.S.A.) vaginal Pap tests.
The presence of benign glandular cells in vaginal Pap tests from posthysterectomy patients represents a diagnostic challenge and may pose management issues. We investigated the presence, frequency and significance of glandular cells in 52 liquid-based (SurePath) vaginal Pap tests from posthysterectomy patients by combining cytomorphologic findings with adjunctive immunohistochemistry and mucin stains performed on cell block preparations and correlated the findings with clinical data.
After performing these special studies, the frequency of reporting glandular cells in posthysterectomy Pap tests decreased from 3.5% to 1.2% of all vaginal Pap tests performed in a 6-month period.
A strong association of the presence of benign appearing glandular cells and a previous history of gynecological malignancy (71%) and chemotherapy/radiation (59%) were noted, likely representing a regenerative process in response to injury or therapy. (Acta Cytol 2009;53:1–9)
Keywords: cytokeratin 7, glandular cells; hysterectomy, Papanicolaou smear; liquid-based cytology; vaginal Pap test.
Utilization of CK7, alcian blue–PAS
and mucicarmine enabled us
to more precisely establish the
true frequency of benign
Glandular cells are not usually expected to be present in vaginal cuff samples because only squamous epithelium is native to vaginal mucosa. However, it is known that glandular cells can be seen in the vaginal Pap tests of posthysterectomy patients.1 Various sources of these glandular cells have been reported in the literature, including vaginal endometriosis,2 vaginal adenosis,3 rectovaginal fistula, tubal prolapse,4 mesonephric duct remnants5 and adenocarcinoma. The exact source of these cells often remains unknown. Differentiating squamous metaplastic-like cells from columnar/cuboidal glandular-like cells can be difficult or sometimes impossible based on morphology alone. In some cases, a history of gynecologic malignancy may raise the question of recurrent malignancy.1 The presence of a glandular component may raise the possibility that a portion of the cervix is retained after hysterectomy. With the advent of liquid-based technology in the preparation of cervicovaginal samples, the impact on the frequency of reporting glandular cells is not yet established. It is an important goal to establish the true presence, frequency and significance of these glandular cells in posthysterectomy specimens. We prospectively collected such cases and prepared cell blocks from the specimen remaining after a routine SurePath (TriPath Imaging, Burlington, North Carolina, U.S.A.) slide was prepared for diagnosis. The cell block sections were prepared and stained with cytokeratin (CK) 7, mucicarmine, and alcian blue–periodic acid–Schiff (PAS) to better characterize these cells. CK7, a low-molecular-weight cytokeratin is an antibody to intermediate filaments. Biochemically and immunohistochemically, these glandular cells have features of simple epithelia characterized by the presence of only low-molecular-weight cytokeratins, including CK7, 8, 18 and 19.6 Special histochemical stains, mucicarmine and alcian blue are used specifically to stain acid mucins, reflecting the sulfated sialic acid mucopolysaccharide content present in glandular epithelial cells.7 There have been no prior studies that used immunohistochemistry and only 3 studies that have employed special histochemical stains8-10 to study the nature of these cells. There is no report to date that has studied the impact of liquid-based technology on the frequency of reporting glandular cells. By utilizing CK7 in conjunction with mucicarmine and alcian blue, our aim was to quantify more precisely the frequency and significance of glandular cells in posthysterectomy liquid-based (SurePath) vaginal Pap tests.
Materials and Methods
During a 6-month period, from July 2005 to January 2006, 1,478 liquid-based, posthysterectomy vaginal samples were submitted to the Division of Cytopathology at Bellevue Hospital. They represented 6.6% of all gynecologic samples received during that period. Cases with cells suggestive of glandular cells posthysterectomy were forwarded to the attending pathologists by the primary screening cytotechnologist. The attending cytopathologists identified 52 cases from 52 patients containing benign glandular cells based on their cytomorphologic resemblance to columnar, squamous metaplastic or hybrid parabasal/endocervical-type cells.8 None of the cases selected contained glandular cells with an associated atrophic background, defined as: (1) Pap test samples composed of >90% parabasal cells with preserved nuclear polarity, (2) presence of degenerated parabasal cells (blue blobs) and multinucleated histiocytic giant cells and (3) clumped granular background debris adherent to cell clusters.1 Two control groups, each consisting of 10 consecutive cervicovaginal samples, were analyzed from premenopausal women with samples containing endocervical cells and from postmenopausal women with atrophic patterns and no glandular component. The cervicovaginal samples from the premenopausal group also contained squamous metaplastic cells in 6 of 10 cases. The pertinent clinical history, prior slides and biopsies in the departmental records were obtained and reviewed. Patients were followed for 1 year. Institutional review board approval was obtained.
The preparation of Pap tests used SurePath liquid-based technology. Specimens for SurePath preparations were obtained using an endocervical broom and/or brush that was submitted in the vial of preservation fluid. The samples were processed with the AutoPrepStain Processor (TriPath Imaging). In all cases, slides were stained on the processor with the manufacturers’ modified Pap staining method. For the 52 cases and the 2 control groups, a paraffin-embedded compact cell block11 was prepared from the remaining material present in the vial. Thin sections of the cell block preparations were stained with conventional hematoxylin-eosin (H-E), an additional immunohistochemical stain for CK7 (Ventana Medical Systems, Tucson, Arizona, U.S.A.) and the special histochemical stains mucicarmine (Ventana Medical Systems) and alcian blue–PAS (Ventana Medical Systems). Immunohistochemistry was performed on sections prepared from cell blocks using an automated immunostainer (NexES, Ventana Medical Systems) by a modified avidin-biotin-peroxidase technique.
Glandular cells were identified by their resemblance to columnar, squamous metaplastic or hybrid parabasal-endocervical type cells. The cytomorphologic criteria used to assess for these particular cell types were as follows: columnar cells contained basally located nuclei with uniform, finely granular chromatin and foamy cytoplasm filled with mucin droplets (Figure 1); squamous metaplastic cells contained dense cytoplasm and were irregular or polygonal in shape with cytoplasmic processes (Figure 2); and hybrid cells, described by Tambouret et al,8 contained scant to moderate cytoplasm, distinct cytoplasmic borders, with centrally placed nuclei, 1 or 2 times the size of an intermediate cell (Figure 3).
Cell blocks were examined for adequate representation of the original vaginal slide (Figure 4). Of the 52 cases analyzed, 17 had CK7–, alcian blue–PAS– and mucicarmine+ cells; 35 cases were negative for all 3. Cells demonstrating CK7 positivity (CK7+) were determined to have diffuse, brown cytoplasmic staining (Figure 5). Mucicarmine positivity was determined as diffuse, red cytoplasmic staining (Figure 6) and alcian blue–PAS positivity as blue, diffuse cytoplasmic staining (Figure 7). Positive stains were found in specimens from patients from 45 to 86 years of age (average, 67). Twelve (71%) patients had a history of a gynecologic malignancy. Ten (59%) patients had undergone radiation or chemotherapy (Table I). Specimens containing CK7+ glandular cells compared to those without glandular cells present revealed a higher percentage of patients having a history of malignancy (71%) and radiation/chemotherapy (59%) as opposed to those without glandular cells present (31% and 23%, respectively) (Table I).
The patients’ ages ranged from 36 to 86 years (average, 63.5). Twenty-nine (56%) patients underwent a hysterectomy for benign disease and 23 (44%) for malignancy (Table I). Follow-up clinical examination was negative in 46 patients; 6 patients had no follow-up. Follow-up Pap tests were performed in 22 patients within 3–12 months and were all negative for intraepithelial lesion or malignancy. In 3 of these 22 patients, benign glandular cells were identified again. One patient had a follow-up vaginal biopsy of the right and left vaginal fornices showing squamous atrophy.
The first control group, consisting of cervical/endocervical Pap tests with an endocervical component from 10 premenopausal women, showed all of the glandular cells to be CK7+, alcian blue–PAS+ and mucicarmine+. The second control group, consisting of cervical smears devoid of endocervical cells from 10 postmenopausal women, showed a lack of expression for all 3 stains. The squamous metaplastic cells identified in the 2 control groups were negative for CK7, mucicarmine and alcian blue–PAS. The hybrid cells showed a variable pattern of staining within each case; some stained for CK7, mucicarmine and alcian blue–PAS and some did not.
Of the 17 cases containing CK7+ glandular cells, the predominant cell type seen was squamous metaplastic cells, regardless of prior history. Squamous metaplastic cells were observed in 11 (92%) patients with a history of malignancy and 4 (80%) patients without a history of malignancy (Table II). The least common cell type detected was columnar cells. No difference was encountered between patients with and without a history of malignancy. Columnar cells were observed in 9 (75%) patients with a history of malignancy and 3 (60%) patients without a history of malignancy (Table II). These findings are reflected in the patterns. The most common pattern seen in 7 of the CK7+ cases (41%) had all 3 cell types present: squamous metaplastic cells, hybrid cells and columnar cells. The second most common pattern observed contained squamous metaplastic cells and hybrid cells, detected in 5 (29%) patients (Table III). This contrasted with the most common pattern detected in CK7 negative (CK7–) cells, which were squamous metaplastic cells and hybrid cells, seen in 12 (34.2%) of the smears (Table IV). CK7+ staining was highly specific in columnar cells but was the least frequent cell type detected in the vaginal tests.
The associated background findings included inflammation, atrophy, repair and radiation effect. Inflammation was defined as a predominant population of neutrophils with or without the presence of bacterial vaginosis (1 case), Candida (1 case) and herpes (1 case). This was the predominant background seen and was observed in 7 (41%) of CK7+ vaginal Pap tests. Radiation effect was detected in 1 (6%) case. This contrasted with an atrophic background being the most common pattern observed in CK7– vaginal smears (Table V).
A history of malignancy was identified in 12 (71%) cases containing CK7+ cells, significantly higher than the cases containing CK7– cells (Table I). The types of malignancies identified in the former group were endometrial adenocarcinoma (6 cases; 5 endometrioid type, 1 mucinous type), squamous cell carcinoma of the cervix (4 cases), clear cell carcinoma of the cervix (1 case) and papillary serous carcinoma of the ovary (1 case) (Table VI).
The frequency of glandular cells in the vaginal samples of posthysterectomy patients varies tremendously in the literature, with some reported percentages as low as 1.6%8 and others as high as 19%10 (Table VII). This wide variation can be attributed to many factors. One factor contributing to this discrepancy is the lack of established morphologic criteria for the definition of glandular cells. Some reports have utilized strict criteria,4,10,12 defining true glandular cells as goblet-shaped columnar cells with foamy/granular cytoplasm, often containing apical mucin with eccentrically placed nuclei. Other reports have employed broader definitions, identifying glandular cells based on their resemblance to columnar squamous metaplastic cells, or a hybrid of both parabasal/endocervical cells.7-9,13 Another variable is the inclusion of patients with supracervical hysterectomies and/or intact uteri.10 In addition, we have found no study addressing the effect of liquid-based preparation of Pap tests on this frequency.
In our study, we utilized the broader cytomorphologic definition for identification of glandular cells. The rationale for using this broader definition was based on the extreme difficulty of identifying the true nature (parabasal vs. columnar), of the hybrid-type cell encountered in vaginal samples. Using stricter criteria excludes a portion of suspected glandular cells. On review of the 52 vaginal Pap tests, we encountered similar cell types (squamous metaplastic cells, hybrid cells and columnar cells) as described by Tambouret et al.8 However, we found a different frequency of patterns observed. The most common pattern we observed in CK7+ Pap tests was a combination of all 3 cell types: squamous metaplastic, hybrid and columnar cells (41%, Table III) compared to Tambouret et al,8 who noted a predominance of hybrid cells (59%). Based on cytomorphologic criteria alone, benign glandular cells were found in 3.5% of posthysterectomy vaginal Pap smears examined at our institution over a 6-month period.
No study to date used immunohistochemical stains to establish the true frequency of glandular cells in posthysterectomy vaginal samples. Only 3 prior studies employed special histochemical stains. Koike et al10 described goblet-like cells in atrophic vaginal smears from elderly women. In that study, alcian blue–PAS and mucicarmine stains demonstrate the presence of mucin by the goblet-like cells, with a frequency of 19% for the detection of glandular cells in posthysterectomy patients, a rate much higher than other studies.4,8,9,12,13 Because that study failed to exclude patients with intact uteri, it is impossible to rule out the origin of some cells from the cervix or upper gynecologic tract. Sodhani et al9 identified 14 smears with columnar cells and squamous metaplastic cells in women who underwent hysterectomy for benign disease. Alcian blue–PAS stain was used to confirm the presence of mucin in these cells, establishing a frequency in their population of 5.6%. Tambouret et al8 used mucicarmine and PAS-D (PAS with diastase digestion) in only 6 of 34 follow-up vaginal biopsy tissue sections. It was unclear how the 6 cases were selected for staining. No special stains were performed on the actual cytology preparations.
When immunohistochemical and special stains were applied to our cytology specimens, the percentage of true glandular cells that we reported decreased from 3.5% to 1.2% (Table VII). CK7 positivity along with alcian blue–PAS and mucicarmine staining were observed in all of the columnar cells and a portion of hybrid parabasal/endocervical type cells. No staining was observed in squamous metaplastic-type cells and in some hybrid-type cells. An explanation for the positivity seen in only some hybrid-type cells is that these cells may represent a nonspecific response to injury and are likely to evolve from reserve cells progressing to immature and mature squamous metaplastic cells. It is known that biochemically and immunohistochemically immature squamous metaplasia shares features of both mature squamous epithelium and mucinous columnar epithelium.5 Focal mucin production can be demonstrated along with less complex keratin intermediate filaments seen ultrastructurally in the cytoplasm.5 These hybrid-type cells are progressing along a continuum with some cells differentiated toward mature squamous cells, demonstrating CK7 negativity, whereas other cells differentiated toward columnar epithelium, demonstrating CK7 positivity. The predominant slide background of acute and chronic inflammatory cells (41%) would support this theory (Table V).
The CK7– cells consisted primarily of patterns composed of squamous metaplastic-type cells alone or accompanied by hybrid-type cells with a predominant background of atrophy (31%, Table V). We interpreted these cells as being representative of the goblet metaplasia first described by Koike et al10 in the vagina of postmenopausal patients. These goblet-type cells derived from atrophic vaginal epithelium can be a source of pseudoglandular-type cells.
There have been numerous etiologies suggested for the origin of these glandular cells. The most plausible hypothesis is a metaplasia related to therapy. A majority of studies4,7,8,12,13 found an association with prior malignancy (Table VII). This association may purely be incidental because posthysterectomy patients with a history of malignancy undergo more rigorous examination and follow-up. Also, vaginal Pap tests are primarily performed in patients with a history of gynecologic neoplasia. Tambouret et al8 showed this association to be statistically significant and speculated that radiotherapy and/or chemotherapy may induce a benign metaplastic process. Zaino et al5 state that, “glandular cells are more commonly found in vaginal cytology specimens after hysterectomy for malignancy.” A large percentage, 71% of our patients, had a history of a gynecologic malignancy and ~60% had received some form of chemotherapy, radiation or hormonal therapy (Table I). In 53% of slides, the predominant background pattern was either inflammation or repair (Table V). This seems to support the idea that these glandular cells may be a result of metaplasia due to a reparative or regenerative process.4,10 There are very few published reports in the literature on the effect of treatment on the vaginal mucosa. Most have described the development of acquired vaginal adenosis (adenosis arising in women with a previously normal vagina) as a result of topical 5-fluorouracil therapy or carbon dioxide laser treatments for vaginal condylomatosis.14-17 This form of adenosis can be interpreted as either a reparative response to injury or the unveiling of an occult vaginal adenosis by destruction of the overlying squamous epithelium. We found no studies that focused on the effects of chemotherapy, radiation or hormonal agents used on the vaginal mucosa.
Another possible source of glandular cells is occult vaginal adenosis, described as a partial or complete conversion of the vaginal mucosa from squamous to mucinous endocervical-type glandular epithelium. There are 3 morphologic types of vaginal adenosis recognized: occult, cystic and superficial.18 In the occult and cystic forms, glandular epithelium located in the lamina propria is completely covered by squamous epithelium. In the superficial form, the glandular epithelium replaces the surface squamous epithelium. The most common glandular cell type is the mucinous columnar cell, but tuboendometrial cells resembling endometrial and fallopian tube epithelium can also be seen. Vaginal adenosis has been commonly observed in young women who had prenatal diethylstilbestrol (DES) exposure. The prevalence of vaginal adenosis in young women exposed to DES in utero has been reported to vary from 35% to 90%.19-21 None of our patients reported DES exposure in utero. A spontaneous form of vaginal adenosis does exist, and it is reported to be surprisingly more frequent than expected. Sandberg22 found occult vaginal adenosis in 9 (41%) of 22 vaginas from postpubertal girls at autopsy and none in 13 prepubertal patients. Similar findings were obtained by Kurman and Scully,23 suggesting that vaginal adenosis can arise congenitally but that steroid hormones play a stimulatory role in their development. Clinically, patients may be asymptomatic or have complaints of excess mucous discharge. Vaginal adenosis is also known to undergo regression, the glandular epithelium gradually replaced by the process of squamous metaplasia.24 Various types of cells, including squamous metaplastic cells and mucinous columnar cells, can be seen in differing proportions in vaginal cytology specimens depending on the number of foci and the degree of involution taking place. In hysterectomies performed for malignancies, it is reported that 35% of the vaginal cuff is excised, primarily the upper third, where 75% of adenosis is found.25 Subsequent vaginal biopsies performed in studies by Bewtra4 and Tambouret et al8 failed to demonstrate well-developed glands in the vagina. Based on these findings, it is highly unlikely for the origin of the cells to be derived from vaginal adenosis.
Some lesser possible causes for the presence of glandular cells can include vaginal endometriosis,2 rectovaginal fistula,4 tubal prolapse4 and mesonephric duct remnants.5 These causes were easily ruled out in our patient population as possible sources of glandular cells based on clinical history, colposcopic examination or biopsy. It is important to note that without the pertinent clinical information, these etiologies would be extremely difficult to distinguish on cytomorphology alone from the glandular cells seen in our study.
The most important entity to consider and to successfully rule out is adenocarcinoma. A high percentage, 71%, of our patients had a history of malignancy; therefore it was crucial to rule out recurrence (Table I). The cytomorphology for adenocarcinoma, even when well differentiated, will demonstrate malignant cytologic features that include irregular nuclear contours, increased nuclear to cytoplasmic ratios, hyperchromasia, anisonucleosis, and extensive nuclear pseudostratification and overlap. None of the features of malignancy were present in our study cases.
In summary, the utilization of CK7, alcian blue–PAS and mucicarmine enabled us to more precisely establish the true frequency of benign glandular cells, with a reported percentage of 1.2% (Table VII). Our findings are similar to results reported in the literature, with benign glandular cells being present in ~2% of posthysterectomy conventional vaginal smears.4,8,12,13 CK7 in conjunction with alcian blue–PAS and mucicarmine enabled us to evaluate the hybrid-type cells and identify the true glandular cells among them. When in doubt, it is not necessary to use these 3 stains to establish a diagnosis of glandular cells. Although the most recent Bethesda 2001 reporting criteria assigned an optional category for the presence of these benign cells, there is no strong evidence that presence of these cells have any impact on patient management.1 As this study illustrates, the glandular nature of these benign cells is of little clinical import and calls into question the necessity of reporting these cells at all.
We found a strong association of a history of gynecologic malignancy (71%) and chemotherapy/radiation (59%) with the presence of inflammatory cells and reparative changes (59%) as a common background pattern (Tables I and V). We believe this to be a true association, as established by Tambouret et al.8 The strong association makes it extremely important to evaluate these cells thoroughly to rule out the possibility of recurrent cancer. Unfortunately, despite numerous reports in the literature,4,7-10,12,13 studying the true origin of these glandular cells, their origin and their clinical significance in vaginal cytology is still uncertain. We postulate that when a clear, identifiable source cannot be discerned after complete clinical investigation and there is a strong history of malignancy and treatment, coupled with a background smear pattern of inflammation/repair, the glandular cells observed are a regenerative response of reserve cells to injury or therapy. As Fluhman26 stressed, there is a bipotential nature of the basal/reserve cells of the squamous epithelium in the cervix to evolve into either columnar or squamous cells after local injury; the same phenomenon can be observed in the vagina.
From the Department of Pathology, New York University Medical Center–Bellevue Hospital Center, New York, New York, U.S.A.
Drs. Ramdall and Levine are Assistant Professors.
Dr. Wallach is Professor of Obstetrics and Gynecology.
Drs. Cangiarella and Simsir are Associate Professors.
Dr. Cai was Cytopathology Fellow, and currently is Attending Pathologist and Assistant Professor, University of Pittsburgh Medical Center, Presbyterian Shadyside Hospital, Pittsburgh, Pennsylvania.
Mr. Elgert is Cytotechnologist.
Address correspondence to: Risha B. Ramdall, M.D., Department of Pathology, New York University Medical Center, 560 First Avenue, TH461, New York, New York 10016, U.S.A. (email@example.com).
Financial Disclosure: The authors have no connection to any companies or products mentioned in this article.
Received for publication August 17, 2007.
Accepted for publication October 9, 2007.
0001-5547/09/5301-0001/$21.00© The International Academy of Cytology
Send website comments to Webmaster@acta-cytol.com