The cellient automated cell block system is useful in the differential diagnosis of atypical glandular cells in Papanicolaou tests




Atypical glandular cells (AGC) is a very important diagnosis in gynecological cytology. In the current study, the authors investigated the usefulness of Cellient cell blocks (CB) for characterizing AGC on Papanicolaou (Pap) tests.


A total of 148 patients with an AGC diagnosis based on Pap tests by cytotechnologists and referred to cytopathologists were studied. Among these patients, there were 68 patients with CB preparations and 80 patients with Pap tests only (TP-AGC group). Follow-up results by Pap tests or biopsies were obtained in 117 of 148 patients. The median follow-up was 13 months (range, 1 month-36 months).


Of the 68 patients with CBs, 31 (46%) were reclassified as negative for dysplasia or low-grade intraepithelial lesion; 30 patients (44%) retained a diagnosis of AGC (CB-AGC group); and 7 patients (10%) were given specific diagnoses of high-grade intraepithelial lesion (3 patients), endocervical adenocarcinoma in situ (1 patient), and invasive adenocarcinoma (3 patients). On follow-up, the CB-AGC group was found to have a significantly lower rate of negative/low-grade squamous intraepithelial lesion diagnoses compared with the TP-AGC group (55% vs 85%; P = .006). The CB-AGC group had a significantly higher rate of endocervical or endometrial adenocarcinoma compared with the TP-AGC group (36% vs 8%; P = .003) at the time of follow-up. The rates of high-grade squamous intraepithelial lesion were not found to be statistically different between these 2 groups (9% vs 7%; P = .66).


The Cellient CB is a useful technique to further categorize a diagnosis of AGC on Pap tests. Using the Cellient CB system, the pathologist has the ability to improve the diagnostic accuracy of AGC so that unnecessary colposcopic evaluation or biopsies can be avoided. Cancer (Cancer Cytopathol) 2014;122:8–14. © 2013 American Cancer Society.


In the Bethesda System for reporting Papanicolaou (Pap) tests,[1, 2] the term “atypical glandular cells” (AGC) is used for atypical glandular cells not fulfilling the diagnostic criteria for in situ or invasive adenocarcinoma. AGC is not a well-defined entity and can reflect a variety of conditions ranging from benign to malignant. Although only 0.2% of patients are diagnosed as having AGC on their Pap tests,[3] approximately 10% to 39% of patients with AGC will be found to have significant lesions such as high-grade squamous intraepithelial lesions (HSILs) or cancers on follow-up.[4-7] Because of the substantial risk of high-grade lesions associated with AGC cytology, the 2001 Bethesda System[2] revised the former name of the finding from atypical glandular cells of undetermined significance to AGC, removing the phrase “of undetermined significance” to avoid possible confusion with the less-risky finding of ASCUS (atypical squamous cells of undetermined significance). The misleading subclassification of AGC “favor reactive” was also eliminated at that time. Accordingly, current practice guidelines[8, 9] recommend an extensive initial evaluation of women with AGC based on age and AGC subclass.[9, 10]

The accurate interpretation of AGC is an important challenge to cytopathologists. Cell block preparation with an inverted filter technique has been shown to be useful for characterizing AGC Pap tests.[11] Cell blocks also are useful for nongynecologic samples by providing information regarding tissue architectural features and facilitating immunohistochemical staining.[12-14] Liquid-based gynecologic cytology permits cell block preparation on residual material. However, conventional cell block techniques are mostly manual, difficult to standardize, and appear unsuitable for low-cellularity samples.[15] The Cellient Automated Cell Block System (CB)[15-17] allows for the efficient and rapid processing of micro-sized cytology samples without manual handling, with the collection of the cytologic sample taking place at 1 defined plane in paraffin for histologic sectioning. This system was introduced to the study institution in 2005 as a complementary diagnostic tool to the conventional cell block method. In the current study, we investigated whether Cellient CB could help resolve the diagnosis of AGC in liquid-based Pap tests.



A total of 148 patients with Pap tests were consecutively selected at the Department of Pathology at the University of Massachusetts Medical Center between January 2010 and March 2012. Of these patients, 68 were referred by cytotechnologists with a diagnosis of AGC for which pathologists ordered a Cellient CB preparation (CB group) for the final diagnosis; 80 patients were diagnosed with AGC based only on a ThinPrep slide without cell block examination (TP-AGC group). The mean age of the patients in the CB group was 45 years (range, 21 years-83 years), with a median of 42 years. The mean age of the patients in the TP-AGC group was 46 years (range, 24 years-81 years), with a median of 46 years. There was no significant difference noted between the groups with regard to age (P = .74). Pathologists used individual criteria (eg, ruling out benign endocervical and endometrial cells, squamous cell lesions, or adenocarcinomas, etc) to decide whether to obtain a cell block. Cases with a prior AGC diagnosis, prior total hysterectomy, and an AGC diagnosis with concurrent endocervical/endometrial biopsies were excluded. The study protocol was originally reviewed and approved by the University of Massachusetts Institutional Review Board.

Evaluation of Cytology and Histopathology

The Pap test cytologic slides were prepared by ThinPrep (Hologic Corporation, Marlborough, Mass). All specimens were evaluated and reported using 2001 Bethesda System terminology: AGC, atypical squamous cells cannot exclude HSIL (ASC-H), atypical squamous cells of undetermined significance (ASCUS), low-grade squamous intraepithelial lesion (LSIL), and HSIL. AGC was subclassified as favor endocervical, endometrial, or not otherwise specified (NOS).

The histopathologic diagnoses for follow-up biopsies included negative for neoplasm, low-grade dysplasia (squamous cervical intraepithelial neoplasia of grade 1 [CIN1]), high-grade dysplasia (CIN2 and CIN3), adenocarcinoma in situ (AIS), and invasive endocervical or endometrial adenocarcinoma.

Clinical Follow-Up

Follow-up data were obtained by reviewing the clinical records of patients from the date of their initial Pap tests to March 15, 2013. The median follow-up was 13 months (range, 1 month-36 months) with 1 to 5 follow-up Pap tests or biopsies performed. The worst diagnoses of squamous or glandular lesions were designated as the final results of follow-up. Of the 148 patients, 117 had clinical follow-up data available, including 44 patients in the CB group and 73 patients in the TP-AGC group.

Cell Block Preparation

The Cellient Automated Cell Block System (Hologic Corporation)[15] was developed based on a patented technique (US Patent 6,913,921) that involves trapping tissue fragments and cells on a smooth filter that does not stick to wax positioned below a modified cassette at exactly the plane at which a microtome will eventually capture a tissue section. The filter is separated from the cassette by a gasket formed of a material that can be separated easily from paraffin that forms a watertight seal between the cassette and the filter. The sample passes through an opening in the modified cassette and is loaded onto the filter. Absolute alcohol, xylene, and molten paraffin (at approximately 5 mL each) flow sequentially through the sample for approximately 3 minutes each. A column of wax is then added, cooled to harden, and the filter and gasket are removed leaving a disk of cells protruding from the undersurface of the cassette. The cells do not stick to the filter. The Cellient process takes approximately 25 minutes. The cassette is then dropped into a standard biopsy mold and remelted for approximately 15 minutes to allow a rim of wax to surround the disk of cells as an aid in aligning the microtome blade. A ribbon of sections is cut and stained with hematoxylin and eosin (H & E). The disk of cells is several hundred microns in thickness, allowing for multiple sections to be cut if necessary for immunohistochemistry. A cell block section was evaluated by a pathologist in conjunction with the Pap test of the same patient. For the CB group, the final interpretation of the cases was determined after examination of the cell block sections.

Statistical Analysis

A 2-tailed Student t test and 2-tailed Fisher exact test were conducted for statistical analysis. A P value < .05 was considered to be statistically significant.


Of the 68 cases in the CB group, 31 (46%) were signed out by pathologists after reviewing the CB slides as being negative for AGC. Diagnoses for the 31 cases were negative for squamous intraepithelial lesions or malignancy (20 patients), ASCUS (5 patients), LSIL (3 patients), and ASC-H (3 patients) (Figs. 1 and 2). Thirty of these 68 cases (44%) were signed out as AGC (endocervical origin in 12 cases, endometrial origin in 13 cases, and NOS in 5 cases). It is important to note that 7 of the 68 cases (10%) referred to the pathologist as AGC were changed to other significant diagnoses, including endocervical AIS (1 case), invasive adenocarcinomas (3 cases), and HSIL (3 cases), after pathologists reviewed CB slides (Table 1) (Figs. 3 and 4).

Figure 1.

Endometrial cells in cell block preparations are shown. (A) A ThinPrep slide demonstrating atypical glandular cells is shown (Papanicolaou stain, × 600). (B) A cell block preparation demonstrates proliferative endometrium (H & E, × 400).

Figure 2.

Reactive endocervical cells are shown in cell block preparations. (A) A ThinPrep slide showing atypical glandular cells obscured by inflammation is shown (Papanicolaou stain, × 600). (B) A cell block preparation showing reactive endocervical cells with heavy inflammation noted in the background is shown (H & E, × 400).

Figure 3.

A high-grade squamous intraepithelial lesion is shown in cell block preparations. (A) A ThinPrep slide is shown demonstrating atypical glandular cells (Papanicolaou stain, × 600). (B) A cell block preparation demonstrating a high-grade squamous intraepithelial lesion and cells with round contour and vacuolated cytoplasm, mimicking the atypical endocervical cells in the Papanicolaou test, is shown (H & E, × 400).

Figure 4.

Endocervical adenocarcinoma in situ was confirmed on cell block preparations. (A) A ThinPrep slide is shown demonstrating atypical glandular cells (Papanicolaou stain, × 600). (B) A cell block preparation demonstrating cytologic features of adenocarcinoma in situ, including pseudostratification, nuclear enlargement and elongation, coarse chromatin, mitotic figures, and apoptosis is shown (H & E, × 400).

Table 1. Frequencies of Final Cytological Diagnoses of 68 Cases Screened as AGC in the Cell Block Group
  1. Abbreviations: AGC, atypical glandular cells; AIS, adenocarcinoma in situ; ASC-H, atypical squamous cells cannot exclude high-grade squamous intraepithelial lesion; ASCUS, atypical squamous cells of undetermined significance; HSIL, high-grade squamous intraepithelial lesion; IA, invasive adenocarcinoma; LSIL, low-grade squamous intraepithelial lesion; NEG, negative for neoplasm.


Of the 30 cases diagnosed as AGC (CB-AGC group), follow-up results were available for 22 cases (73%). The diagnoses at the time of follow-up included negative for squamous or glandular neoplasm (10 cases), LSIL (2 cases), CIN3 and HSIL (2 cases), endocervical AIS (3 cases), and invasive endometrial adenocarcinoma (5 cases). Of the 3 AIS cases, 1 case was found to have coexisting CIN3.

Of the 80 cases in the TP-AGC group, the AGC subtype included favor endometrial (25 cases), favor endocervical (24 cases), and NOS (31 cases). Follow-up results were available for 73 of the 80 cases (91%). The diagnoses included negative for squamous or glandular neoplasm (49 cases), low-grade follow-up (for which we included ASCUS, LSIL, and ASC-H on repeat Pap testing or CIN1 on biopsy) (13 cases), HSIL (CIN2 and CIN3) (5 cases), endocervical AIS (2 cases), and endometrial adenocarcinoma (4 cases). Of the 2 AIS cases, both had coexisting CIN3.

Table 2 shows a comparison of follow-up results between the CB-AGC (22 cases) and TP-AGC (73 cases) groups. The CB-AGC group had a significantly lower rate of a negative or low-grade follow-up than the TP-AGC group (55% [12 of 22 cases] vs 85% [62 of 73 cases]; P < .01) (Table 2). Conversely, the CB-AGC group was found to have a significantly higher rate of endocervical/endometrial adenocarcinoma compared with the TP-AGC group (36% [8 of 22 cases] vs 8% [6 of 73 cases]; P < .01) (Table 2). There were no significant differences noted between the 2 groups for follow-up diagnosis of HSIL (9% [2 of 22 cases] vs 7% [5 of 73 cases]; P = .66).

Table 2. Comparison of Follow-Up Results of AGC With or Without Cell Blocks
ResultCB-AGC (n = 22)TP-AGC (n = 73)P
  1. Abbreviations: AGC, atypical glandular cells; CB-AGC, AGC diagnosed with ThinPrep test plus cell block; TP-AGC, AGC diagnosed by ThinPrep test alone; HSIL, high-grade squamous intraepithelial lesion; LSIL, low-grade squamous intraepithelial lesion.

Negative/LSIL12/22 (55%)62/73 (85%)<0.01
HSIL2/22 (9%)5/73 (7%).66
Endocervical/endometrial adenocarcinoma8/22 (36%)6/73 (8%)<0.01

In the CB group, 22 patients diagnosed as “negative” (13 patients) or others (LSIL, ASCUS, and ASC-H; 9 patients) had clinical follow-up data available. All 13 cases diagnosed as “negative” in the CB group (65%) were found to be negative for any squamous or glandular lesions on follow-up by Pap tests (1 time-4 times in 8 months-28 months). Nine of 11 cases (9/11, 82%) diagnosed as ASCUS (5 cases), LSIL (3 cases) and ASC-H (3 cases) in CB group had available data regarding clinical follow-up, including negative for squamous or glandular neoplasms (4 cases; prior diagnoses of ASCUS in 2 cases and LSIL in 2 cases), LSIL (3 cases; prior diagnoses of LSIL in 1 case and ASC-H in 2 cases), squamous CIN3 (1 case with a prior diagnosis of ASC-H), and endocervical AIS (1 case with a prior diagnosis of ASC-H).


The incidence of adenocarcinoma in the endocervix or endometrium has been increasing as the overall incidence of squamous carcinoma has decreased.[18] The basic criteria for a diagnosis of AGC rely on identifying hyperchromatic crowded groups of cells with features exceeding those for reactive reparative changes and falling short of the full criteria for glandular neoplasia. Although several studies have been instrumental in characterizing the cytologic features and establishing the diagnostic criteria,[19-23] the interobserver variability is still significant,[19, 24, 25] and sensitivity and specificity remain low even in the hands of experienced pathologists.[23] However, AGC is a very significant diagnosis clinically. Most women with a diagnosis of AGC should undergo a colposcopy procedure with at least endocervical curettage.[9] If there is no glandular neoplasm identified in follow-up tissue biopsies, a large excisional biopsy may be necessary if clinically indicated. Therefore, a diagnosis of AGC has a profound impact on clinical management and a patient's quality of life.

In the current study, we analyzed the data regarding the use of the Cellient CB as a complimentary tool for the diagnosis of AGC. Analysis of the Cellient CB with ThinPrep demonstrates several features that make it an attractive tool with which to improve both the sensitivity and specificity for the diagnosis of AGC.

First, the cell block analysis was able to help pathologists to classify AGC further into other diagnoses. We found that after pathologists examined the cell block slides, approximately 30% of cases with the diagnosis of AGC made by cytotechnologists were finally diagnosed as negative for dysplasia or neoplasms after the pathologists reviewed cell block slides. In these cases, the most common misclassified benign mimics of AGC included reactive/reparative changes of endocervical cells, endometrial cells, and tubal metaplasia.

Second, analysis of the Cellient CB with ThinPrep was helpful in significantly increasing the positive predictive value of AGC for the detection of endocervical/endometrial adenocarcinomas. In a comparison of follow-up results of AGC cases examined with ThinPrep plus cell block and ThinPrep alone, the CB-AGC group was found to have a significantly higher rate of endocervical/endometrial adenocarcinoma (36% vs 8%) in the follow-up tissue biopsies. One of the reasons for the increase in the specificity for the detection of adenocarcinoma in the CB-AGC group was that 56% of cases (38 of 68 cases) with cell block analysis were able to be classified further other than AGC. One case of endocervical AIS on follow-up was diagnosed as ASC-H in the CB group.

Finally, the Cellient CB system is a suitable technique to combine with ThinPrep gynecological cytology. The usefulness of cell block in gynecological cytology has been hampered largely by the inherent limitations of the sampling method and technical issues of efficiently collecting cells. In most cases, after ThinPrep Pap tests are made, there remains residual material that may contain valuable diagnostic material for supplementary diagnostic studies such as additional ThinPrep slides[26] and human papillomavirus typing.[27, 28] The Cellient CB system provides a new platform in this field. The flow-through processing maximizes the efficiency of cell recovery and of extractions during embedding, thereby decreasing the amount of cellular sample required and minimizing the amount of reagents needed for embedding. The method and apparatus also automatically place cells at the plane in the cell block at which they need to be sectioned without diluting them with carrier substances. In addition, the method and apparatus minimize or eliminate the potential for cross-contamination between 2 different cell samples. Rapid turnaround time is another advantage of the Cellient system, allowing for cases to be signed out on the same day. The section is of excellent quality and is comparable to routine paraffin sections. Moreover, this technique provides archival material that is available for ancillary studies such as immunohistochemical or molecular tests. Recently, the diagnostic value of the Cellient CB system has been studied in nongynecological specimens.[15, 29]

Although analysis of the Cellient CB block with ThinPrep demonstrates features that make it an attractive tool with which to improve the diagnosis of AGC, caution should be exercised. In the current study, the cell block was more likely to be ordered in cases in which the pathologist favored either a negative or malignant process. Therefore, there might be sampling bias in the CB group. Moreover, there were no certain numbers of glandular groups required to determine whether there was enough material on the cell block. Further studies are needed to create quantitative criteria to determine cell block adequacy.

In summary, by using the Cellient cell block system, the pathologist has the ability to improve the diagnostic accuracy of AGC on the Pap test. The Cellient CB system can be used in conjunction with cervical smears to provide a further differential diagnosis of AGC.


No specific funding was disclosed.


Dr. Fischer is a named inventor of Cellient cell block technology licensed to Hologic Corporation and receives royalties for sales of Cellient products.