Accuracy and false-positive rate of the cytologic diagnosis of follicular cervicitis: observations from the College of American Pathologists Pap educational program.
The cytologic features of follicular cervicitis, which were first described in 1965,5 are well documented in the literature (1,2,5-7); it is therefore generally assumed that it is a relatively straightforward diagnosis to make in most instances. There have been, however, sporadic mentions of follicular cervicitis being a cause of false-positive diagnoses in gynecologic cytology. (2,3,8) However, detailed and large-scale studies are not found in the literature.
The College of American Pathologists (CAP) Pap Program (CAP-PAP) is an educational program consisting of 5 glass slides (case challenges) with accompanying clinical history. Of note, all the cases circulated are reviewed by 3 anatomic pathologists (members of the CAP Cytopathology Resource Committee) who must agree with the reference diagnosis for the case to be entered into the program. The slide sets are circulated 4 times a year to participating laboratories.
Participants choose a general diagnostic category and a specific reference diagnosis based upon a menu. The data for each challenge are compiled by the CAP and specific feedback is sent to each participant and participating laboratory, allowing comparison of individual performance as well as comparison to that of the other laboratories. The CAP-PAP includes conventional smears and ThinPrep (Hologic, Marlborough, Massachusetts) cases whose reference diagnosis is follicular cervicitis." There are insufficient SurePath (BD, Franklin Lakes, New Jersey) slides circulating with that specific reference diagnosis for an accurate analysis of performance, so SurePath cases were not included in the study.
MATERIALS AND METHODS
We retrospectively reviewed the cumulative data of the CAPPAP (obtained through the CAP SCORES computer system) from 2000 to 2010 from those cases with the reference diagnosis follicular cervicitis" in order to determine the accuracy of the diagnosis and to determine the most common misinterpretation patterns leading to false-positive diagnoses according to participant and preparation types.
There were a total of 4914 participant responses available from 132 cases with a reference diagnosis of follicular cervicitis." The distribution of responses by participant type was as follows: 2012 were from pathologists, 2081 from cytotechnologists, which represented 821 unique laboratories. False-positive diagnoses were defined as any diagnosis beyond the negative for intraepithelial lesion" interpretation and included the following diagnostic categories: low-grade squamous intraepithelial lesion, high-grade intraepithelial lesion (HSIL), high-grade intraepithelial lesion or carcinoma (HSIL/carcinoma) or carcinoma not otherwise specified, adenocarcinoma in situ, adenocarcinoma not otherwise specified, nonepithelial malignant neoplasm, and squamous cell carcinoma. False-positive rates by participant type (pathologist versus cyto-technologist versus laboratory), the reference diagnosis, and preparation type (conventional smear versus ThinPrep) were analyzed.
A nonlinear mixed model was fit with participant type and preparation type. The interaction term between these factors was also included in the model. The model included a repeated measures component to model the slide factor correlation structure. A significance level of .05 was used for this analysis. All statistical analyses were performed with SAS version 9.2 (SAS Institute Inc, Cary, North Carolina).
Of the 4914 responses from the cumulative 11-year review of the CAP-PAP for challenge cases with a reference diagnosis of follicular cervicitis," 88.9% (4368 of 4914) were interpreted as negative for intraepithelial lesion or malignancy (NILM) (true negative) while 11.1% (546 of 4914) were called abnormal (false positive). An exact diagnostic match with the specific reference diagnosis of follicular cervicitis was reached in only 41.2% of the challenges, accounting for 46.4% (2026 of 4368) of the NILM answers. Figure 1, A and B, illustrates a follicular cervicitis case challenge that performed well in the CAP-PAP. The distribution of the diagnostic interpretations from the participants who opted for the NILM and for the "abnormal" categories is listed in Tables 1 and 2, respectively. Of note, adenocarcinoma and HSIL accounted for most of the false-positive answers: 42.3% (231 of 546) and 20.1% (110 of 546), respectively. A variety of other abnormal categories accounted for the remainder of the answers (Table 2). Figure 2, A and B, and Figure 3, A and B, illustrate follicular cervicitis case challenges that did not perform well in the CAP-PAP, while Figure 4 contrasts a typical case of follicular cervicitis (Figure 4, A) with 5 entities entering its differential diagnosis (Figure 4, B through F).
There was a statistically significant difference in the false-positive response rate between conventional smears (3.6%) and ThinPrep samples (12.2%) (P < .001) (Table 3). There was also significant difference in the performance between participant types (pathologists versus cytotechnologists versus laboratory) (P < .001) (Table 3).
The diagnostic features of follicular (lymphocytic) cervicitis, which have been well described in the literature, consist of a polymorphous population of lymphocytes with or without tingible-body macrophages seen in clusters or streaming out in mucus. (1-3,5-7) In conventional smears, the lymphoid cells tend to lie singly, and only rarely as cell clusters; in contrast, the cells tend to be more often aggregated in clumps in liquid-based preparations. (3)
Interestingly, there are no available data in the literature on the accuracy of the cytologic diagnosis of follicular cervicitis. In our study, which relates to an educational program, an exact match to the reference diagnosis of follicular cervicitis was reached in only about 41% of the cases. It is difficult to know if these results can also be extrapolated to daily practice situations. Indeed, it is important to remember that the CAP-PAP is not equivalent to clinical practice; participants in the CAP-PAP might not interpret challenge cases with the same frame of mind as they might in real life. Reassuringly, there are no direct consequences of not diagnosing this entity specifically, as it is benign and does not necessitate a specific follow-up or treatment. In addition, according to The Bethesda System for Reporting Cervical Cytology, (1) follicular cervicitis features among those nonneoplastic findings for which reporting is simply optional.
In our current study, a false-positive rate of 11.1% was observed. Of note, there was a statistically significant difference in the performance between participant types: the highest false-positive rate (19.2%) was seen in the "laboratory" participant responses in contrast to 7.9% and 11.1% for individual pathologists and cytotechnologists responses, respectively, when all preparatory types were considered; it even reached 22.2% when ThinPrep samples alone were analyzed. This high false-positive rate for the laboratory responses has to be taken in the context that the laboratory responses are not reported in a standardized fashion. Indeed, while some laboratories may submit a laboratory response that is most representative of all the participants in a given laboratory, others may use the cytotechnologists' responses for benign entities or use either the interpretation of the technical supervisor or of the medical director. Even if making abstraction of these less-standardized laboratory responses, the data from the current study show that, based on the individual responses of the cytotechnologists and the pathologists, follicular cervicitis is a significant cause of false-positive diagnoses, at least in this educational program.
The fact that follicular cervicitis can be misinterpreted and be a source of false-positive diagnoses has been occasionally mentioned in the literature. (2,3,7) Although most reported cases of follicular cervicitis misinterpreted as false positive have been in conventional smears, rare cases have been in ThinPrep samples. (3) The most common false-positive diagnoses for follicular cervicitis listed in the literature include HSIL, adenocarcinoma, and lymphoma; in our study, it was both adenocarcinoma and HSIL that accounted for most false-positive diagnoses. Interestingly, there was statistical difference in the current study in the false-positive response rate between conventional smears (3.6%) and ThinPrep samples (12.2%) (P < .001). This difference is most likely due to a combination of factors: (1) it is probably more challenging to notice the lymphocytes in liquid-based preparations, as the density" of lymphocytes is lower in liquid-based preparations because they are spread evenly throughout the slide instead of being concentrated focally as in conventional slides, and (2) the lymphocytes tend to aggregate more in clusters in liquid-based preparations than in conventional smears, as previously reported in the literature. (2,3) The latter tendency to aggregate may impart a glandular appearance to the lymphocytes; in fact, such aggregated lymphocytes in the context of chronic cervicitis have also been mentioned in a previous study as a source of misinterpretation as normal endometrial cells. (9) These aggregated lymphocytes can also lead to potential misinterpretation as atypical glandular cells or as adenocarcinoma. Since the response menu of the CAP-PAP does not include the category atypical glandular cells," it leaves little choice to the participants but to choose adenocarcinoma" for their response when they have concerns regarding certain glandular" groupings in a case challenge; this restricted choice of answers probably led to the increase in adenocarcinoma diagnoses in this setting, a situation which would likely not occur in real life."
An accurate cytologic diagnosis of follicular cervicitis depends upon the recognition of the polymorphous lymphoid population consisting of small round mature lymphocytes admixed with larger, more immature lymphocytes. Tingible-body macrophages may also be present, but are not required for the diagnosis. Even when in clumps, the polymorphous nature of the cell population should be recognizable, allowing one to differentiate follicular cervicitis from the usually monotonous population of syncytial HSIL or of adenocarcinoma, along with other morphologic features specific to the latter two. While syncytial HSIL is characterized by hyperchromatic crowded groups, typically composed of epithelial cells exhibiting nuclear hyperchromasia associated with variations in nuclear size and shape as well as irregular nuclear membranes, invasive adenocarcinoma is typically characterized by syncytial/3-dimensional groups, as well as single atypical glandular cells with vacuolated cytoplasm and enlarged pleomorphic nuclei with macronucleoli. (1) Finally, another differential diagnosis includes lymphoma; however, lymphoma of the cervix is very rare, and in contrast to follicular cervicitis, which shows a spectrum of mature and immature-looking lymphocytes, lymphoma of the cervix tends to be composed of a monotonous population of lymphocytes, usually of large B-cell type.
In conclusion, cytotechnologists and pathologists should be aware that follicular cervicitis can lead to false-positive responses, especially in liquid-based preparations, in an educational interlaboratory comparison program, and by extrapolation may lead to false-positive diagnoses in clinical practice.
Caption: Figure 1. Representative photomicrographs from a case in the College of American Pathologists Pap Educational Program illustrating the characteristic cytologic features of follicular cervicitis. This is a case that performed well: of the 57 participants, 49% called it follicular cervicitis, 47% called it negative for intraepithelial lesion or malignancy not otherwise specified, and 4% called it repair. A, A tingible-body macrophage is seen admixed with a polymorphous lymphoid population. B, In this field, only the polymorphous lymphoid population is present (ThinPrep Pap test; Papanicolaou stain, original magnifications 3400 [A] and 3200 [B]).
Caption: Figure 2. Representative photomicrographs from a challenge case in the College of American Pathologists Pap Educational Program. This is a case that performed poorly: of the 88 participants in this case, 28% called it abnormal (18% adenocarcinoma, 5% adenocarcinoma in situ, and 5% highgrade intraepithelial lesion/carcinoma). A, A tingible-body macrophage is seen adjacent to mature and immature lymphocytes. B, A 3-dimensional large cluster of the lymphocytic population is shown; clusters such as this one are probably the source of misinterpretations as adenocarcinoma (ThinPrep slide; Papanicolaou stain, original magnifications 3400 [A and B]).
Caption: Figure 3. Representative photomicrographs from a challenge case of follicular cervicitis in the College of American Pathologists Pap Educational Program that performed poorly. A and B, Large clustering and crowded nuclei of lymphoid cells are shown; such clusters were likely the cause of most false-positive interpretations. The presence of tingible-body macrophages (circles) and the spectrum of small lymphocytes (mature) and lymphocytes with varying degree of differentiation (immature) are helpful features to reach a correct diagnosis of follicular cervicitis (ThinPrep; Papanicolaou stain, original magnifications 3400 [A and B]).
Caption: Figure 4. Representative photomicrographs from 6 cases of Papanicolaou test showing follicular cervicitis (A) and its mimics/differential diagnoses (B through F). (Note that the cases illustrated in B through F are not from the College of American Pathologists Pap Educational Program [CAP-PAP]). A, A challenge case of follicular cervicitis in the CAP-PAP showing large clustering and crowded nuclei of lymphoid cells. B, A case of high-grade intraepithelial lesion. C, A case of adenocarcinoma in situ. D, A case of endocervical adenocarcinoma. E, A case of metastatic colonic adenocarcinoma. F, A case of diffuse large B cell lymphoma (ThinPrep; Papanicolaou stain, original magnifications 3400 [A through F]).
(1.) Solomon D, Nayar R, eds. The Bethesda System for Reporting Cervical Cytology: Definitions, Criteria, and Explanatory Notes. 2nd ed. New York, NY: Springer-Verlag;2004.
(2.) DeMay RM. The Art and Science of Cytopathology: Exfoliative Cytology. 2nd ed., Chapter 1: The Pap Test. Chicago, IL: ASCP Press;2012:31-32.
(3.) Halford JA. Cytologic features of chronic follicular cervicitis in liquid-based specimens: a potential diagnostic pitfall. Cytopathology. 2002;13(6):364-370.
(4.) Hare MJ, Toone E, Taylor-Robinson D, et al. Follicular cervicitis-colposcopic appearances and association with Chlamydia trachomatis. Br J Obstet Gynaecol. 1981;88(2):174-180.
(5.) Eisenstein R, Battifora H. Lymph follicles in cervical smears. Acta Cytol. 1965;9(5):344-346.
(6.) Roberts TH, Ng ABP. Chronic lymphocytic cervicitis: cytologic and histopathologic manifestations. Acta Cytol. 1975;19(3):235-243.
(7.) Young RH, Harris NL, Scully RE. Lymphoma-like lesions of the lower female genital tract: a report of 16 cases. Int J Gynecol Pathol. 1985;4(4):289-299.
(8.) Boon M, Gray W. Normal vulva, vagina and cervix: hormonal and inflammatory conditions. In: Gray W, ed. Diagnostic Cytopathology. Edinburgh, Scotland: Churchill Livingstone;1995:656-657.
(9.) Kapali M, Agaram NP, Dabbs D, Kanbour A, White S, Austin MR. Routine endometrial sampling of asymptomatic premenopausal women shedding normal endometrial cells in Papanicolaou tests is not cost effective. Cancer. 2007;111(1): 26-33.
Manon Auger, MD; Walid Khalbuss, MD; Ritu Nayar, MD; Chengquan Zhao, MD; Patricia Wasserman, MD; Rhona Souers, MS; Nicole Thomas, CT; Ann T. Moriarty, MD
Accepted for publication August 20, 2012.
From the Department of Pathology, McGill University and McGill University Health Center, Montreal, Quebec, Canada (Dr Auger);the Department of Pathology, University of Pittsburgh Medical Center and Shady Side Hospital, Pittsburgh, Pennsylvania (Dr Khalbuss);the Department of Pathology, Northwestern University and Northwestern Memorial Hospital, Chicago, Illinois (Dr Nayar);the Department of Pathology, University of Pittsburgh Medical Center and Magee Women's Hospital, Pittsburgh, Pennsylvania (Dr Zhao); the Department of Pathology, Hofstra School of Medicine, Long Island Jewish Medical Center, New Hyde Park, New York (Dr Wasserman); Biostatistics (Ms Souers) and Surveys Department (Ms Thomas), College of American Pathologists, Northfield, Illinois; and Cytology, AmeriPath Indiana, Indianapolis (Dr Moriarty).
The authors have no relevant financial interest in the products or companies described in this article.
Reprints: Manon Auger, MD, Department of Pathology, McGill University, 3775 University St, Room 105, Montreal, QC H3A2B4, Canada (e-mail: email@example.com).
Table 1. Negative (Benign) Participant Interpretations (N = 4368) Interpretation Frequency Percentage Follicular cervicitis 2026 46.4 NILM, not otherwise specified 1608 36.8 Atrophic vaginitis 353 8.1 Reparative changes 229 5.2 Unsatisfactory for evaluation 73 1.7 Fungal organisms consistent with Candida 44 1.0 Trichomonas vaginalis 30 0.7 Cellular changes consistent with herpes simplex virus 5 0.1 Abbreviation: NILM, negative for intraepithelial lesion or malignancy. Table 2. Positive (Abnormal) Participant 1 Interpretations (N = 546) Interpretation Frequency Percentage Adenocarcinoma, NOS 231 42.3 HSIL 110 20.1 LSIL 84 15.4 HSIL/carcinoma and/or carcinoma, NOS 39 7.1 AIS 36 6.6 Nonepithelial malignant neoplasm 24 4.4 Squamous cell carcinoma 22 4.0 Abbreviations: AIS, adenocarcinoma in situ; HSIL, high-grade squamous intraepithelial lesion; LSIL, low-grade squamous intraepithelial lesion; NOS, not otherwise specified. Table 3. Nonlinear Mixed Model Results No. of False Positive Responses Rate, % P Value Preparation type Conventional 612 3.6 <.001 ThinPrep 4302 12.2 Participant type <.001 Pathologist 2012 7.9 Cytotechnologist 2081 11.1 Laboratory 821 19.2 Preparation type(a) .23 Participant Conventional--Path 256 2.7 Conventional--CT 225 4.4 Conventional--Lab 131 3.8 ThinPrep--Path 1756 8.6 ThinPrep--CT 1856 11.9 ThinPrep--Lab 690 22.2 Abbreviations: CT, cytotechnologist;Lab, laboratory;Path, pathologist. a Two types of preparation broken down by participant types. Bolded P values indicate statistical significance.
Please note: Illustration(s) are not available due to copyright restrictions.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||CAP Laboratory Improvement Programs|
|Author:||Auger, Manon; Khalbuss, Walid; Nayar, Ritu; Zhao, Chengquan; Wasserman, Patricia; Souers, Rhona; Tho|
|Publication:||Archives of Pathology & Laboratory Medicine|
|Date:||Jul 1, 2013|
|Previous Article:||Laboratory performance in albumin and total protein measurement using a commutable specimen: results of a college of American pathologists study.|
|Next Article:||Clinical implications of current developments in genitourinary pathology.|