Atypical Epithelial Cells and Specimen Adequacy

Transcription

1 Atypical Epithelial Cells and Specimen Adequacy Current Laboratory Practices of Participants in the College of American Pathologists Interlaboratory Comparison Program in Cervicovaginal Cytology Diane D. Davey, MD; Sherry Woodhouse, MD; Patricia Styer, PhD; Janet Stastny, DO; Dina Mody, MD Context. The Bethesda System for reporting cervical/ vaginal cytologic diagnoses introduced terminology for atypical squamous and glandular cells and categories for specimen adequacy. Objectives. To analyze current laboratory reporting practices and compare trends to previous surveys. Design. Questionnaire surveys were mailed to 2 laboratories in 1996 and Participants. Laboratories enrolled in the College of American Pathologists Interlaboratory Comparison Program in Cervicovaginal Cytology. Main Outcome Measures. Laboratory policies, criteria, and reporting rates for Bethesda System categories. Results. The 1996 specimen adequacy survey had 1166 respondents, and 768 laboratories returned the 1997 questionnaire focusing on atypical squamous cells of undetermined significance (ASCUS) and glandular cells of undetermined significance (AGUS). Nearly all laboratories (92%) routinely reported specimen adequacy, an increase from the 66% rate in The median rate for unsatisfactory specimens was % (mean.95%), and the median rate for the satisfactory but limited category was 5.8% (mean 9.3%). The Bethesda criteria for designating a specimen unsatisfactory were used by more than 9% of laboratories. Nearly all laboratories (97%) used the term ASCUS in 1997, and more than 8% of laboratories used the Bethesda criteria for this category. Median reporting rates for epithelial abnormalities were as follows: ASCUS, 4.5%; AGUS,.3%; low-grade squamous intraepithelial lesion (SIL), 1.6%; and high-grade SIL, %. The median ASCUS/SIL ratio was, with 8% of laboratories reporting ratios between 4 and The median ASCUS rate and ASCUS/SIL ratio were higher than 1993 survey results. Nearly all laboratories attempted follow-up studies on patients with abnormal cytology results, and midsized laboratories achieved the highest rates of follow-up. Median rates of abnormalities following an ASCUS or AGUS diagnosis were 2% and 15%, respectively. Laboratory respondents commonly used written recommendations in ASCUS/AGUS reports. Conclusions. Most laboratories that responded to the surveys had adopted Bethesda terminology and criteria for specimen adequacy and ASCUS/AGUS. Reporting rates for SIL and adequacy categories have remained stable, but median ASCUS rates and ASCUS/SIL ratios are higher than in The AGUS category is reported infrequently, but can be associated with significant pathology. (Arch Pathol Lab Med. 2;124:23 211) The Bethesda System (TBS) for reporting cervical/vaginal cytologic diagnoses introduced terminology for atypical squamous and glandular cells and categories for specimen adequacy. 1 Previous surveys conducted by the College of American Pathologists (CAP) have indicated widespread acceptance of TBS. 2 5 The proper use of the diagnostic categories atypical squamous cells of undetermined significance (ASCUS) and atypical glandular cells of undetermined significance (AGUS) and patient followup measures remain controversial. 6 Increasing liability concerns may be putting pressure on laboratories to place more specimens into borderline categories. 7 Accepted for publication July 19, From the Department of Pathology and Laboratory Medicine, University of Kentucky, Lexington, Ky (Dr Davey); Division of Pathology and Laboratory Medicine, Cleveland Clinic Florida, Ft Lauderdale, Fla (Dr Woodhouse); College of American Pathologists, Northfield, Ill (Dr Styer); Outpatient Cytopathology Center, Johnson City, Tenn (Dr Stastny); and Department of Pathology, Baylor College of Medicine, Houston, Tex (Dr Mody). Dr Styer served as statistical consultant. Reprints: Diane D. Davey, MD, Department of Pathology and Laboratory Medicine, University of Kentucky Medical Center, 8 Rose St, Room MS117, Lexington, KY The College of American Pathologists Interlaboratory Comparison Program in Cervicovaginal Cytology (PAP) has conducted periodic questionnaire surveys of its participants concerning laboratory practices and reporting rates for TBS categories. Previous published reports have specifically addressed ASCUS and specimen adequacy. 2 5 The purposes of this study were to obtain current information from participants and identify changing trends in cytology practices. METHODS The PAP program consists of quarterly shipments of 5 glass slides of cervicovaginal material to each participating laboratory. Mailed supplementary questionnaires on specific areas of laboratory practice are also conducted, generally on an annual basis. The 1996 questionnaire addressed specimen adequacy issues, while the 1997 questionnaire examined epithelial abnormalities, concentrating on ASCUS and AGUS. Each questionnaire was mailed midyear to the approximately 2 laboratories participating in PAP. The response rate was 56% (1166 laboratories) for the 1996 survey and 38% (768 laboratories) for the 1997 survey. Responses were tabulated by the CAP staff in Northfield, Ill. Not all laboratories returning questionnaires responded to every question, so the total number of respondents for each question is indicated in Arch Pathol Lab Med Vol 124, February 2 Atypical Epithelial Cells and Specimen Adequacy Davey et al 23

2 All laboratories Unsatisfactory Limited* Table 1. Percentage of Unsatisfactory and Satisfactory but Limited Cervicovaginal Cases No. of Laboratories Responding Mean Limited rates according to cervical cytology volume th * Refers to cases labeled satisfactory but limited by. Cytology volume was not available for all responding laboratories; unknown volumes are not included th the results. Since the response rate was below 5% for the 1997 questionnaire, respondents were compared to nonrespondents to ascertain whether the respondents were representative of all laboratories enrolled in PAP. Nearly all (96%) of the questionnaires returned in 1997 were from laboratories of known institution type and case volume. There was no difference between respondents and nonrespondents according to institution (laboratory) type, but larger laboratories (annual gynecologic case volume 2, 2 or more cytotechnologists and pathologists) were more likely to return the questionnaire. For example, 36% of the responding laboratories had or 1 cytotechnologists, while this was true of 66% of the nonresponding laboratories. These findings may indicate a bias favoring practices of larger PAP laboratories. Respondents were instructed to report known or estimated rates for TBS categories to the nearest tenth of a percent. Only known rather than estimated rates were used for analysis of AS- CUS/AGUS and squamous intraepithelial lesions (SIL), and the total numbers of respondents are lower for these questions. Reporting rate distributions did not show a Gaussian distribution, but were skewed; therefore, the results illustrate median and percentile reporting rates instead of standard deviations. Participants were asked to provide the frequency of various laboratory practices using the following categories: always, 1% of the time; usually, 5% to 99% of the time; often, 2% to 49% of the time; sometimes, 5% to 19% of the time; rarely, less than 5% of the time; and never, % of the time. Results for several questions were analyzed according to laboratory or institution type and annual cervical cytology case volume, and both significant and negative analyses were reported. Items selected included those with apparent differences by visual inspection, and those of most interest to cytology practice. Demographic data were not available for every responding laboratory. Laboratories with unknown demographics were included in overall tabulations, but were excluded in the analysis of responses by institution type and volume. The Kruskal-Wallis test, a nonparametric analysis of variance model, and the Pearson 2 test were used to evaluate the institution type and case volume effects on the dependent variable. 8 In some cases, responses by case volume were further analyzed using a Spearman rank correlation coefficient to assess if there was evidence of a monotonic relationship between case volume and the dependent variable. 9 To adjust for multiplicity of tests, P.1 was used as evidence of a statistically significant result. Results of previous published questionnaire data were compared with the current surveys when applicable. Specimen adequacy practices were surveyed in 199 and 1991, and ASCUS questionnaires were conducted in 1993 and ,5 Responders to surveys in different years were not perfectly matched nor statistically independent. Therefore, formal statistical tests concerning trends in practices are not presented. Descriptive information is provided for some questions repeated from previous years. Some responses in the current survey were compared with responses to the 1993 survey by assessing previous results relative to confidence intervals for the medians of the current results. 8 RESULTS More than half of the laboratories surveyed were hospital-based facilities; independent laboratories were next in frequency. The median cervicovaginal cytology case volume of responding laboratories was approximately 1. Specimen Adequacy In 1996, nearly all laboratories (176/1166 [92.3%]) provided a statement of specimen adequacy in every cervicovaginal cytology report. This represents an increase from the 35% and 66% rates in 199 and 1991, respectively. 5 Reporting rate distributions for the unsatisfactory and the satisfactory for evaluation but limited by... (SBL) categories are provided in Table 1. The median ( percentile) unsatisfactory rate was %, and the mean was.95%. The and percentile reporting rates were.1% and %, meaning that 8% of laboratories reported unsatisfactory rates within this range. No significant differences were seen according to laboratory type or case volume. Adequacy reporting rates in 1991 were queried by a choice of bracketed ranges, and the median unsatisfactory rate was % to.9%. 5 The median SBL rate was 5.8% in 1996, and the mean was 9.3%. In contrast, the median less than optimal rate in the 1991 survey was between 3.% and 4.9%. 5 Eighty percent of laboratories had SBL rates between % and 21.4% in Laboratories with larger cytology case volumes reported higher SBL rates, for which we found a statistically significant positive association (Table 1; P.1, Kruskal-Wallis and rank correlation tests). Laboratory type was not significantly associated with SBL rates. Laboratories were asked to provide the 3 most common reasons for designating a specimen either unsatisfactory or SBL. The percentage of laboratories and number listing each reason in their top 3 are shown in Table 2. For the unsatisfactory category, the most common reasons were obscuring inflammation, scanty squamous cellularity, and obscuring blood. Partially obscuring inflammation, no/ scanty endocervical component, and obscuring blood were the leading reasons for designating a specimen SBL. Cell types or other constituents accepted as evidence of an endocervical or transformation zone component are provided in Table 3. Most laboratories accepted either en- 24 Arch Pathol Lab Med Vol 124, February 2 Atypical Epithelial Cells and Specimen Adequacy Davey et al

3 Table 2. Most Common Reasons for Designating a Specimen Unsatisfactory or Satisfactory but Limited* Unsatisfactory, % (N) Limited, % (N) Obscuring inflammation 84.4 (969) 88.8 (999) Scanty squamous cellularity 81.8 (939) 29.8 (335) Obscuring blood 66.5 (763) 56.5 (636) No/scanty endocervical component 1 (122) 73.9 (831) Obscuring by insufficient cell preservation 32.4 (372) 27.8 (313) Incomplete clinical information (15) 1. (113) Obscuring cytolysis or autolysis 3.9 (45) 5.8 (65) Obscuring contaminant (12) (11) No name or incorrect name (identification) 5.5 (63) (6) Broken/damaged slide 4.7 (54).1 (1) * Laboratories were asked to name their 3 top reasons for each specimen adequacy category. Columns indicate the percentage and number listing each reason; N indicates number of laboratories responding. docervical columnar or squamous metaplastic cells (79.8%), and no associations were apparent according to laboratory type or case volume. In comparison, only 43% of respondents in the 1991 questionnaire used this criterion. 5 Endocervical columnar cells were required by 27% of laboratories in This criterion was used less by large laboratories, a statistically significant association (Table 3; P.9, 2 and rank correlation tests). The absence of an endocervical/transformation zone component did not generally result in a specimen being categorized as unsatisfactory, according to 87.6% (111/1154) of laboratories. However, 81% of laboratories agreed that absence of an endocervical/transformation zone component would render a specimen SBL under certain circumstances. Nearly all laboratories used the following TBS criteria for unsatisfactory specimens in the 1996 questionnaire: smear without well-visualized squamous cells over at least 1% of the slide (153/1161 [9%]), and at least % of epithelial cells obscured by inflammation, blood, or other (185/1162 [93.4%]). A written recommendation to repeat the smear was frequently made in unsatisfactory reports. The following answers were provided: 47.6% always, 16% usually, 4.5% often, 7.1% sometimes, 14.4% rarely, and % never. By contrast, half of the responding laboratories (49.7%) rarely or never gave a written recommendation for a repeat smear in less than 1 year in a report of an SBL specimen. When queried whether epithelial abnormalities were ever reported when specimens were categorized as unsatisfactory for evaluation, 61.6% answered No, which is the correct response according to TBS. The remainder stated that cytology reports could list both an unsatisfactory adequacy designation and an epithelial abnormality. About half of laboratories deemed specimens SBL when pertinent clinical information was absent on the requisition (564/1139 [49.5%]). Atypical Epithelial Cells In 1997, 97.1% (74/725) of laboratories reported using the term ASCUS, compared to 86% in Cellular changes included in the ASCUS category in 1997 were as follows: (1) cells with nuclear enlargement at least 2.5 times the size of normal intermediate cell nucleus, 88.2% (677); (2) cells suggestive of koilocytes but demonstrating fewer than the 3 required diagnostic features (abnormal nucleus, large perinuclear clear zone, peripheral cytoplasmic condensation, 86.3% (663); (3) atypical squamous metaplasia, below the diagnostic threshold for high-grade SIL, 85% (653); (4) hyperchromatic crowded groups, 55.1% (423); (5) parakeratosis with spindled cells or increased nuclear-cytoplasmic ratio, 72.9% (56); (6) parakeratosis alone, 7.3% (56); and (7) hyperkeratosis alone, 1.8% (14). The first 3 criteria, which are similar to those recommended by TBS, were used by 7% to 74% of laboratories in Parakeratosis alone qualified as ASCUS according to 19% of respondents in 1993, and hyperkeratosis was a criterion of 1%. Criteria 4 and 5 were not queried in the 1993 survey. Laboratory reporting rates for SIL, ASCUS, and AGUS are illustrated in Table 4. The median reporting rate for low-grade SIL (LSIL) was 1.6% in 1996, which was identical to the 1993 median. 3 Eighty percent of laboratories had LSIL rates between % and 4.3% in For highgrade SIL (HSIL), the,, and percentile reporting rates in 1996 were %, %, and %, respectively. The median HSIL reporting rate in 1993 was %. Reporting rates for LSIL and HSIL in 1996 varied significantly according to laboratory type (P.1 for both; Kruskal-Wallis tests). University laboratories had the highest median LSIL and HSIL rates (3.8% and %, respectively; Table 5). Laboratory volume had a marginally significant effect, and there was no evidence of a simple Table 3. Cell Types Acceptable as Evidence of Endocervical or Transformation Zone Component* No. of Laboratories Responding ECC or Sq Met, % ECC Only, % ECC or Sq Met or Mucus, % ECC and Sq Met, % Sq Met Only, % No Criteria, % All laboratories Cervical cytology volume * ECC indicates endocervical columnar cells; Sq Met, squamous metaplastic cells. Statistical analysis was only performed for the ECC or Sq Met and ECC Only results (see text). There were insufficient responses to analyze other columns. Cytology volume was not available for all responding laboratories; unknown volumes are not included. Arch Pathol Lab Med Vol 124, February 2 Atypical Epithelial Cells and Specimen Adequacy Davey et al

4 Table 4. Percentages of Cases Reported in 1996 as Epithelial Abnormalities and Atypical Squamous Cells of Undetermined Significance (ASCUS)/Squamous Intraepithelial Lesion (SIL) Ratio. Category No. of Laboratories Responding Mean 25th (Median) th LSIL HSIL ASCUS AGUS ASCUS/SIL * LSIL indicates low-grade SIL; HSIL, high-grade SIL; and AGUS, atypical glandular cells of undetermined significance Table 5. Percentages of Cases Reported as Low-Grade (LSIL) and High-Grade (HSIL) Squamous Intraepithelial Lesion in 1996 According to Laboratory Type LSIL (N)* Hospital (351) Independent (1) Federal (31) Government (37) University (17) Other, clinic, office (17) HSIL (N) Hospital (352) Independent (98) Federal (27) Government (37) University (17) Other, clinic, office (16) N indicates number of laboratories responding. Mean Table 6. Atypical Squamous Cells of Undetermined Significance/Squamous Intraepithelial Lesion Ratio in 1996 According to Laboratory Type and Volume Mean Laboratory type (N)* Hospital (31) Independent (84) Federal (26) Government (33) University (16) Other, clinic, office (16) Cervical cytology volume (N) 5 (136) (18) (91) (92) 5 (53) monotonic relationship between reported rates and case volume. The median 1996 ASCUS rate was 4.5%, and 8% of laboratories had ASCUS rates between 1.5% and 9.9%. The 1993 median ASCUS rate was only 2.8% 3 ; this number does not fall within the 1996 median confidence intervals of 4.% to 4.8%, suggesting that the 1996 figures probably represent an increase in reporting rates. Rates for ASCUS did not vary significantly according to either laboratory type or case volume. Laboratories were asked how often an ASCUS response was qualified, when possible, to indicate whether a reactive or premalignant response was favored. Of the 6 respondents, 23.3% answered always, 44.6% usually, 14.8% often, 8.9% sometimes, 5.2% rarely, and 3.3% never. For laboratories reporting both SIL and ASCUS rates, an ASCUS/SIL ratio was calculated (Tables 4 and 6). The median ASCUS/SIL ratio was (95% confidence interval, ), and 8% of laboratories had ASCUS/SIL ratios between 4 and In contrast, the 1993 median ASCUS/SIL ratio was, outside the 95% confidence intervals for the 1996 survey. 3 Both laboratory case volume and type were significantly associated with the ASCUS/ SIL ratio (P.1 for both; Kruskal-Wallis tests). The 26 Arch Pathol Lab Med Vol 124, February 2 Atypical Epithelial Cells and Specimen Adequacy Davey et al

5 Table 7. Proportion of Cases Receiving Cytohistologic Follow-Up Studies Within 1 Year Percentage of Laboratories Responding Yes in Each Category % 5% to 74% 5% All laboratories (N 717)* Laboratory type (N) Hospital (427) Independent (133) Federal (42) Government (47) University (21) Other, clinic, office (2) Cervical cytology volume (N) 5 (26) (144) (133) (122) 5 (83) Table 8. Percentage of Women With Diagnoses of Atypical Squamous Cells of Undetermined Significance (ASCUS) or Atypical Glandular Cells of Undetermined Significance (AGUS) That Have Follow- Up Cytology or Histology Within 1 Year ASCUS AGUS N* Mean lowest median ASCUS/SIL ratio was observed in university laboratories (), and the highest was calculated in federal laboratories (2.49). Case volume was positively associated with the ASCUS/SIL ratio (P.1; rank correlation test); laboratories with case volumes of greater than 5 had a median ratio of Slightly fewer laboratories provided rates of abnormalities for the first 6 months of 1997, and these were generally similar to the 1996 entire-year figures. Median LSIL and HSIL rates were identical, the median ASCUS rate for 484 laboratories was 4.4%, and the median ASCUS/SIL ratio was 1.9 (N 468). Nearly all laboratories performed follow-up studies on women with abnormal cervicovaginal cytology results (731 [96.8%]). Cancer and HSIL lesions were followed up by at least 94% (719) of laboratories, while 69% (528) of laboratories performed follow-up on all or a portion of LSIL cytology cases. Methods to obtain follow-up information included review of laboratories files (667 [86.8%]), written requests for follow-up sent to health care providers (53 [69%]), and requests for follow-up included with original cytology report (173 [22.5%]). The proportion of cases (based on cases selected for follow-up) for which follow-up cytologic diagnoses, histologic diagnoses, or both were received within 1 year is illustrated in Table 7. The type of laboratory had a marginally significant influence (P.5), while a significant association was observed according to laboratory case volume (P.1; 2 test of homogeneity). Note that both the largest and smallest laboratories tended to have lower follow-up rates. About half (36/743) of laboratories included women with ASCUS in follow-up programs; 33% attempted follow-up on all women with ASCUS, and 16% attempted follow-up on a portion. Laboratories were queried about the percentage of women with ASCUS that had follow-up cytology or histology within 1 year. The mean rate for ASCUS follow-up was 65% (median %; Table 8). No significant differences were noted according to laboratory type or case volume. Average rates of significant abnormality on follow-up of ASCUS cases were queried. Laboratories were instructed to use the following calculation for ASCUS cases: (Number With SIL/Cancer) (Number With Follow-Up). The mean and median rates of abnormalities following ASCUS were 25.4% and 2%, respectively (Table 9). Laboratory case volume, but not type, affected the rates of abnormality following both ASCUS and AGUS (P.1; Kruskal- Table 9. Percentage of Women With Diagnoses of Atypical Squamous Cells of Undetermined Significance (ASCUS) or Atypical Glandular Cells of Undetermined Significance (AGUS) That Have Squamous Intraepithelial Lesion/Carcinoma/ Glandular Neoplasia Within 1 Year Mean ASCUS (N)* All laboratories (256) Cervical cytology volume (N) 5 (86) (55) (44) (39) 5 (17) AGUS (N) All laboratories (232) Cervical cytology volume (N) 5 (86) (55) (44) (39) 5 (17) Arch Pathol Lab Med Vol 124, February 2 Atypical Epithelial Cells and Specimen Adequacy Davey et al

6 Wallis test). Small laboratories were particularly likely to have very low rates of abnormalities on follow-up, and this raises the possibility of incorrect calculation methods. The 1994 survey queried only rates of SIL following ASCUS, and the mean was 28% (median 15%). 3 The and percentile AGUS reporting rates were.3% and %, respectively; these rates have not been queried in previous PAP questionnaires. Fifty-four percent of laboratories sought follow-up information for some or all patients diagnosed with AGUS. In those laboratories seeking data, a mean of 69% of women with AGUS diagnoses had follow-up cytology or histology information within 1 year (median %; Table 8). No significant differences were noted according to laboratory type or case volume. A similar calculation for abnormalities following an AGUS cytology was requested; diagnoses of glandular neoplasia/dysplasia, SIL, and cancer on follow-up were included in the numerator. For AGUS, the mean rate of abnormality was 26.7%, and the median was 15% (Table 9). Laboratory case volume but not type affected the rates of abnormality following an AGUS diagnosis (P.1; Kruskal-Wallis test). Since AGUS data had not been queried previously, several other published AGUS studies were reviewed. Reporting rates for AGUS based on 8 large studies from 1992 to 1998 were between.17% and 1.8% (Table 1) The mean AGUS incidence in these studies based on cases was 3%. 17 If the outlying 1.8% rate in the study by Nasu et al 11 is eliminated, then the mean AGUS incidence was.35%. Follow-up results for AGUS diagnoses from 1 published and unpublished studies from 1992 to 1998 are also shown in Table 1. Approximately 29% had SIL on follow-up, nearly 4% had adenocarcinoma in situ, about 6% had carcinoma, and variable numbers had endometrial hyperplasia The use of written recommendations in reports of AS- CUS/AGUS was queried; 3% (233/768) always made recommendations, 21.8% usually, 9.5% often, 13.9% sometimes, 2% rarely, and 3.3% never. Specific types of recommendations used by laboratories in 1997 were as follows: (1) repeat cytology at short interval (4 6 months), 81.4% (619); (2) treatment with estrogen followed by a repeat smear for further evaluation of ASCUS associated with atrophy, 87% (661); (3) antibiotic/anti-inflammatory treatment followed by a repeat smear for further evaluation of ASCUS, 67.4% (512); (4) human papillomavirus (HPV) DNA testing for follow-up of ASCUS/AGUS, 3.3% (25); (5) suggest colposcopy and biopsy or endocervical curettage for ASCUS, 49.6% (3); and (6) suggest colposcopy and biopsy with endocervical curettage, endometrial curettage, or both for AGUS, 64.7% (485). The recommendation for estrogen therapy was positively associated with laboratory size (P.1, 2 and rank correlation test), indicating that larger laboratories were more likely to use this recommendation (Table 11). The antibiotic recommendation was negatively associated with laboratory size (P.8; 2 and rank correlation test). There was no statistically significant association according to either laboratory type or volume for the other recommendations. COMMENT The PAP questionnaire surveys represent the largest collection of data concerning current cytology laboratory practices in the United States. These surveys have provided valuable information on a variety of topics during the last 1 years and have focused particularly on implementation of TBS. Most laboratories participating in PAP have spent considerable time and effort to complete the often detailed surveys. The lowest response rate was for the 1997 questionnaire (38%); comparison of the responding and nonresponding groups demonstrated that small laboratories were less likely to return the surveys. Since more gynecologic cytology specimens are analyzed in medium to large laboratories, the results reported are likely representative of current US cytology practices. The questionnaire results demonstrate widespread implementation of TBS terminology and criteria among US laboratories. Nearly all laboratories (92%) now report specimen adequacy on each gynecologic cytology specimen, which represents an increase from the 66% rate in More than 9% of laboratories used TBS criteria for designating a specimen unsatisfactory in Unsatisfactory reporting rates in 1996 were similar to those of the 1991 survey. 5 The median unsatisfactory rate was %, with 9% of laboratories reporting rates of % or less. The reporting rate distribution for SBL specimens was much broader, with a median of 5.8% and a percentile rate of 21%. This broad range is likely related to variations in sampling methods, patient risk factors, and laboratory application of criteria. One of the leading reasons for designating a specimen SBL was lack of transformation zone component. The Bethesda System recommends accepting either endocervical columnar or squamous metaplastic cells as evidence of a transformation zone component, and 8% of laboratories agreed, compared to only 43% in Despite the observations that most laboratories have adopted TBS criteria and that unsatisfactory rates appear relatively similar among laboratories, areas of controversy have become apparent since the last Bethesda conference. The 1% cellularity rule is often difficult to use in practice and may be interpreted in a variety of ways. Some demand only that 1% of fields contain numerous squamous cells, while others attempt to look at overall coverage by nuclei and cytoplasm. Renshaw et al 19 demonstrated that a visual estimate of 1% coverage (by nuclei and cytoplasm) actually corresponds to 3% when measured by image analysis. In an interobserver reproducibility study by Spires et al, 2 overall reproducibility rates were good to excellent for the satisfactory and unsatisfactory categories, and fair for the limited category. However, scant squamous cellularity cases were among those with poor reproducibility. The increasing use of liquid-based gynecologic cytology specimens has made reproducible and scientifically valid criteria for minimum cellularity even more critical. The authors recommend that the next Bethesda conference in 2 carefully consider criteria for minimum cellularity. An area of confusion noted was how to report cases with abnormal cells and perceived adequacy limitations. Thirty-eight percent of laboratories reported epithelial abnormalities in cases designated unsatisfactory for evaluation. This is a contradiction in terms, according to TBS. 1 Any specimen with carcinoma is by definition satisfactory, and a specimen with any type of abnormality cannot be considered unsatisfactory for evaluation of epithelial abnormalities. Another controversial question was whether to designate those specimens SBL when there was absence of clinical information on the requisition; only half of laboratories currently do so. 28 Arch Pathol Lab Med Vol 124, February 2 Atypical Epithelial Cells and Specimen Adequacy Davey et al

7 Table 1. Author, y Goff et al, Nasu et al, Taylor et al, Jones and Novis, Kennedy et al, Zweizig et al, Eddy et al, Duska et al, Veljovich et al, Baylor College of Medicine Other Studies Reporting Atypical Glandular Cells of Undetermined Significance and Follow-Up Results* No. of Pap Smears AGUS Rate, % SIL, % AIS, % EMH, % CA, % * AGUS indicates atypical glandular cells of undetermined significance; SIL, squamous intraepithelial lesion; AIS, adenocarcinoma in situ; EMH, endometrial hyperplasia; CA, carcinoma; and, not available or not applicable. Includes glandular dysplasia. Unpublished data Table 11. Recommendations for Follow-Up of Atypical Squamous Cells of Undetermined Significance Cervical Cytology Volume (N)* 5 (223) (152) (142) (127) 5 (84) Percentage of Laboratories Responding Yes in Each Category Estrogen Plus Repeat Antibiotic Plus Repeat Recommendations for early repeat were frequently used by laboratories when issuing unsatisfactory reports, given the unreliability of such specimens for detecting abnormalities. Retrospective studies have demonstrated that false-negative specimens in patients with carcinoma were frequently unsatisfactory on retrospective review. Ransdell et al 21 recently published a follow-up study demonstrating that patients with unsatisfactory index smears were at higher risk for abnormalities in subsequent specimens. In contrast, the 1996 questionnaire demonstrated no consensus for recommending early repeat of the satisfactory but limited specimen. Cross-sectional studies have demonstrated higher rates of abnormalities in those patients with transformation zone cells, but longitudinal studies have failed to demonstrate that patients whose entry smears lacked such cells are at higher risk in subsequent screening events As a result, interim guidelines from a National Cancer Institute (NCI)-sponsored workshop stated that lack of endocervical component alone should not be the sole reason for early repeat of a smear. 26 Additional research on recommended frequency of sampling in patients with SBL specimens would be useful. Despite controversy regarding the ASCUS category, 1997 questionnaire results show widespread acceptance and use of this term. Most laboratories also attempted to use the cytologic criteria recommended by TBS for the ASCUS category. These criteria include nuclear enlargement at least 2.5 times normal, cells with some but not all HPV/koilocyte criteria, and atypical squamous metaplasia. Other borderline cytologic processes, such as atypical parakeratosis and hyperchromatic crowded groups, were also included in the ASCUS category by the majority of laboratories. Hyperchromatic crowded groups represent a diagnostic challenge in which the differential diagnosis includes benign endocervical or endometrial groups, tubal metaplasia, benign metaplastic groups, and squamous or glandular epithelial abnormalities. Such groups are also classified as AGUS or atypical epithelial groups by some laboratories. 6,27 According to TBS, atypical parakeratotic cells should be categorized as ASCUS or SIL, depending on the degree of cellular abnormalities. 1 Atypical parakeratotic cells have been discussed in detail by Voytek et al. 28 Many such cases are associated with more serious lesions, such as HSIL or even carcinoma. Parakeratosis and hyperkeratosis alone without nuclear abnormalities are generally considered benign cellular changes according to TBS, and very few PAP laboratories placed such changes in the ASCUS category in The median reporting rate for ASCUS of 4.5% in 1996 is higher than the 2.8% figure in the previous survey performed in In contrast, LSIL and HSIL rates have remained relatively constant, with the highest rates in university laboratory settings. Possible reasons for the increasing ASCUS rate are the widespread acceptance of this category by laboratories and increasing liability concerns for false-negative cytology reports. 7 Rates for both ASCUS and SIL are dependent on the risk factors of the laboratory patient population, and the AS- CUS/SIL ratio provides some degree of correction. While the median ASCUS/SIL ratio has also increased to since 1993, survey results show a tighter reporting range for the ASCUS/SIL ratio than for ASCUS rates alone. The ASCUS/ SIL ratio is a useful intralaboratory and interlaboratory comparison tool, especially when used in conjunction with follow-up studies and other quality improvement tools. Cytologists should keep in mind that the ASCUS/SIL ratio will vary not only with the ASCUS criteria in use, but also with SIL criteria. Of note, the highest median ratios were reported in large laboratories, and institution type also affected ASCUS/SIL. The reason for this observation is uncertain, but one could speculate that higher ASCUS/SIL ratios are common in laboratories that primarily serve a screening rather than diagnostic function. Most laboratories are attempting follow-up studies on patients with significantly abnormal cytology results, in compliance with federal regulations and CAP laboratory accreditation checklists. While more than 9% attempt fol- Arch Pathol Lab Med Vol 124, February 2 Atypical Epithelial Cells and Specimen Adequacy Davey et al 29

8 low-up on HSIL and cancer cases, fewer do so following LSIL results. The highest rates of follow-up were recorded in laboratories with medium gynecologic cytology volumes (5 5 annually). These laboratories most commonly represent community-based practices, both hospital-based and independent, with well-established quality assurance programs. Cytologic-histologic correlation is important in promoting optimal patient care. Maintaining midvolume community-based cytology practices is one way of ensuring quality patient care and optimal communication between laboratories and clinicians. The majority of laboratories also attempted follow-up studies on at least a portion of ASCUS cases. The median rate of SIL or cancer following an ASCUS diagnosis was 2%. This result is similar to that found in the 1993 questionnaire; other recent studies in the literature have cited dysplasia/cancer rates of about 2% to 5%. 3,29 33 Most laboratories also use recommendations in some or all reports of ASCUS. The most widely used recommendation was repeat cytology at short interval. About half of the responding laboratories recommended colposcopy with biopsy in some ASCUS reports. Interim guidelines resulting from the 1992 NCI workshop suggest differing management guidelines for AS- CUS, depending on clinical circumstances and whether the diagnosis is qualified. 26 Repeat cytology is generally appropriate when the ASCUS diagnosis is not qualified or when a reactive process is favored. Colposcopy should be considered when a preneoplastic process is favored or if the patient is at high risk. According to an international expert conference sponsored by the International Academy of Cytology in 1997, this is particularly important for those cases in which there is concern about a possible high-grade lesion. Solomon et al stated, In contrast to favoring a reactive process versus SIL, a more rational approach to qualification of ASCUS may be to separate cases equivocal for LSIL from those suspicious for HSIL. 6 Atypical metaplastic cells, hyperchromatic crowded groups, and atypical parakeratosis are all examples of the types of ASCUS in which the differential includes a highgrade lesion or cancer. Colposcopy with biopsy may be desirable as the next step, and communication of the level of concern is helpful. In addition to phrases that qualify the ASCUS diagnosis, the same results can be achieved by the use of recommendations in the report or by including laboratory statistics listing the likelihood of various clinical outcomes. 6 A few recent studies analyzing the value of qualifying ASCUS diagnoses have found that the yield of SIL diagnoses, particularly high-grade lesions, is increased when the report indicates that a neoplastic process is favored. 3,31,34 Interim guidelines from the NCI workshop recommend reevaluation after 2 to 3 months when ASCUS is associated with severe inflammation. 26 These guidelines discuss the use of topical estrogen therapy to aid in clarification of AS- CUS associated with atrophy. Antibiotic therapy is recommended when a specific infection is present. According to the NCI guidelines, HPV testing is advised only for clinicians experienced with the limitations. In contrast, guidelines issued by the American Society for Colposcopy and Cervical Pathology (ASCCP) include algorithms for intermediary adjunctive HPV testing, cervicography, or both. 35 These adjunctive testing methods are currently being evaluated by a multicenter NCI trial. Questionnaire results show that most PAP laboratories frequently issue recommendations for use of estrogen or anti-inflammatory therapy, but HPV testing is only mentioned by a minority. This survey represents the largest database on AGUS rates and follow-up practices. The median rate was.3% in 1996, with 9% of the laboratories reporting rates of % or less. The average AGUS incidence in 8 previously reported studies was 3%. 17 If the study reporting a 1.8% incidence is eliminated, then the mean AGUS incidence was.35%, similar to the median AGUS reporting rate from the 381 PAP laboratories. The AGUS category includes a large variety of entities, ranging from reactive processes to adenocarcinoma in situ, and TBS recommends subcategorization as to endocervical or endometrial origin when possible. 1 The endocervical AGUS cases are to be further subclassified into favor reactive and favor neoplastic categories, based on criteria published in the monograph by Kurman and Solomon. 1 More recently, experts have advocated that cases that meet all of the criteria for adenocarcinoma in situ be separated from the AGUS category and placed in a free-standing category of adenocarcinoma in situ. 6 However, the CAP study was performed prior to this recommendation. Both the NCI guidelines and those published by the ASCCP advocate careful follow-up of women with AGUS reports. 26,36 The type of follow-up varies depending on the subclassification by the cytology laboratory, as well as clinical and physical findings. Endometrial biopsy, fractional dilatation and curettage, or hysteroscopy are recommended for endometrial AGUS cases. According to NCI guidelines, repeat cytology utilizing a cytobrush in 3 to 6 months may be appropriate in reliable patients in whom a reactive endocervical process is favored. 26 If the repeat is still interpreted as AGUS or higher, then colposcopy and biopsy are recommended. Colposcopy and endocervical curettage are generally used for the unqualified endocervical AGUS case. The ASCCP guidelines recommend colposcopy and endocervical curettage for all AGUS patients, even those favored as reactive. 36 Both guidelines recommend immediate colposcopy and histologic biopsies when a neoplastic process is favored. Similarly, if the findings are suspicious for adenocarcinoma or if there is persistent unexplained AGUS following initial negative colposcopy, cone biopsy is generally indicated. Based on the survey responses, 54% of the laboratories sought follow-up information on some or all women with AGUS reports. In those laboratories seeking such data, a median of % of women actually had follow-up cytology or histology results available. Of those receiving followup, a mean of 26.7% (median 15%) were found to have carcinoma, glandular neoplasia, or SIL within 1 year of AGUS diagnosis. In the 1 published and unpublished AGUS studies cited in Table 1, about 4% of the patients had SIL, carcinoma, or significant glandular lesions when investigated by biopsy. The average abnormality rates reported from the PAP laboratories are lower than those reported in published studies, including the 1994 CAP Q- Probes study by Jones et al. 18 Possible reasons include differences in types of practice audited as well as study methodology. The lowest rates of abnormality were noted in low-volume laboratories in the PAP questionnaire. In contrast, the single institution studies utilized biopsy follow-up, and most have been reported from university or teaching hospitals, which tend to attract high-risk patients. The Q-Probes study, performed in 348 institutions in 1994, analyzed cytologic-histologic correlation in patients with 21 Arch Pathol Lab Med Vol 124, February 2 Atypical Epithelial Cells and Specimen Adequacy Davey et al

9 all types of epithelial abnormalities, and AGUS reports accounted for only a small fraction of the cases. Because the aim was to specifically examine cervical biopsy correlation, curettage specimens and follow-up cytology cases were not considered. These exclusions likely skewed the analysis in favor of more squamous abnormalities. 18 Regardless of the study source or design, a significant percentage of what is interpreted as AGUS actually turns out to be an SIL or carcinoma on subsequent follow-up. Glandular intraepithelial lesions and endometrial hyperplasia generally represent the minority of results. Other studies have documented poor interobserver reproducibility of the AGUS category. 37 Hence, better definition of AGUS criteria, additional education, and diligent application of currently published criteria are of paramount importance. 1,38,39 This subject was alluded to by Selvaggi in her thoughtful editorial following a review of AGUS abstracts at a national cytopathology meeting. 4 In summary, most PAP laboratories have adopted TBS terminology and criteria for specimen adequacy and AS- CUS/AGUS classifications. Unsatisfactory, LSIL, and HSIL reporting rates have remained relatively stable when recent questionnaire results are compared with those collected in the early 199 s. In contrast, median ASCUS reporting rates and ASCUS/SIL ratios are higher than those reported in the 1993 survey. Such trends may reflect both widespread acceptance of TBS and the increasing liability risks associated with false-negative gynecologic cytology reports. These surveys were performed prior to widespread implementation of liquid-based gynecologic cytology sampling methods. Liquid-based methods could affect reporting rates for adequacy categories and various epithelial abnormalities, and future surveys will attempt to separate reporting rates according to type of preparation. This is the first large questionnaire survey that has queried AGUS reporting rates. While AGUS appears to be an uncommon entity in most laboratories, the risk of serious lesions in such patients has prompted organizations such as NCI and ASCCP to aggressively recommend follow-up studies in these cases. References 1. Kurman RJ, Solomon D. The Bethesda System for Reporting Cervical/Vaginal Cytologic Diagnoses. New York, NY: Springer-Verlag; 1994: Davey DD, Naryshkin S, Nielsen ML, Kline TS. Atypical squamous cells of undetermined significance: interlaboratory comparison and quality assurance monitors. Diagn Cytopathol. 1994;11: Davey DD, Nielsen ML, Naryshkin S, Robb JA, Cohen T, Kline TS. Atypical squamous cells of undetermined significance: current laboratory practices of participants in the College of American Pathologists Interlaboratory Comparison Program in Cervicovaginal Cytology. Arch Pathol Lab Med. 1996;12: Davey DD, Nielsen ML, Rosenstock W, Kline TS. Terminology and specimen adequacy in cervicovaginal cytology: the College of American Pathologists Interlaboratory Comparison Program experience. Arch Pathol Lab Med. 1992;116: Nielsen ML, Davey DD, Kline TS. Specimen adequacy evaluation in gynecologic cytopathology: current laboratory practice in the College of American Pathologists Interlaboratory Comparison Program and tentative guidelines for future practice. Diagn Cytopathol. 1993;9: Solomon D, Frable WJ, Vooijs GP, et al. ASCUS and AGUS criteria: IAC task force summary. Acta Cytol. 1998;42: Frable WJ. Litigation cells: definition and observations on a cell type in cervical vaginal smears not addressed by the Bethesda System. Diagn Cytopathol. 1994;11: Snedecor GW, Cochran WG. Statistical Methods. Ames, Iowa: The Iowa State University Press; Agresti A. Analysis of Ordinal Categorical Data. New York, NY: John Wiley & Sons Inc; Goff BA, Atanasoff P, Brown E, Muntz HG, Bell DA, Rice LW. Endocervical glandular atypia in Papanicolaou smears. Obstet Gynecol. 1992;79: Nasu I, Meurer W, Fu YS. Endocervical glandular atypia and adenocarcinoma: a correlation of cytology and histology. Int J Gynecol Pathol. 1993;12: Taylor RR, Guerrieri JP, Nash JD, Henry MR, O Connor DM. Atypical cervical cytology: colposcopic follow-up using the Bethesda system. J Reprod Med. 1993;38: 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. 1996;63: Zweizig S, Noller K, Reale F, Collis S, Resseguie L. Neoplasia associated with atypical glandular cells of undetermined significance on cervical cytology. Gynecol Oncol. 1997;65: Eddy GL, Strumpf KB, Wojtowycz MA, Piraino PS, Mazur MT. Biopsy findings in 531 patients with atypical glandular cells of uncertain significance as defined by the Bethesda System. Am J Obstet Gynecol. 1997;177: 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. 1998;91: Veljovich DS, Stoler MH, Anderson WA, Covell JL, Rice LW. Atypical glandular cells of undetermined significance: a five year retrospective histopathologic study. Am J Obstet Gynecol. 1998;179: Jones BA, Novis DA. Cervical biopsy cytology correlation: a College of American Pathologists Q-Probes study of correlations in 348 laboratories. Arch Pathol Lab Med. 1996;12: Renshaw AA, Friedman MM, Rahemtulla A, et al. Accuracy and reproducibility of estimating the adequacy of the squamous component of cervicovaginal smears. Am J Clin Pathol. 1999;111: Spires SE, Banks ER, Weeks JA, Banks HW, Davey DD. Assessment of cervicovaginal smear adequacy: the Bethesda System guidelines and reproducibility. Am J Clin Pathol. 1994;12: Ransdell JS, Davey DD, Zaleski S. Clinicopathologic correlation of the unsatisfactory Papanicolaou smear. Cancer Cytopathol. 1997;81: Elias A, Linthorst G, Bekker B, Vooijs PG. The significance of endocervical cells in the diagnosis of cervical epithelial changes. Acta Cytol. 1983;27: Kivlahan C, Ingram E. Papanicolaou smears without endocervical cells: are they adequate? Acta Cytol. 1986;3: Mitchell H, Medley G. Longitudinal study of women with negative cervical smears according to endocervical status. Lancet. 1991;337: Vooijs PG, Elias A, van der Graaf Y, Veling S. Relationship between the diagnosis of epithelial abnormalities and the composition of cervical smears. Acta Cytol. 1985;29: Kurman RJ, Henson DE, Herbst AL, Noller KL, Schiffman MH. Interim guidelines for management of abnormal cervical cytology. JAMA. 1994;271: Bose S, Kannan V, Kline TS. Abnormal endocervical cells: Really abnormal? Really endocervical? Am J Clin Pathol. 1994;11: Voytek TM, Kannan V, Kline TS. Atypical parakeratosis: a marker of dysplasia? Diagn Cytopathol. 1996;15: Alanen KW, Elit LM, Molinaro PA, McLachlin CM. Assessment of cytologic follow-up as the recommended management for patients with atypical squamous cells of undetermined significance or low grade squamous intraepithelial lesions. Cancer Cytopathol. 1998;84: Genest DR, Dean B, Lee KR, Sheets E, Crum CP, Cibas ES. Qualifying the cytologic diagnosis of atypical squamous cells of undetermined significance affects the predictive value of a squamous intraepithelial lesion on subsequent biopsy. Arch Pathol Lab Med. 1998;122: Kline MJ, Davey DD. Atypical squamous cells of undetermined significance qualified: a follow-up study. Diagn Cytopathol. 1996;14: Sheils LA, Wilbur DC. Atypical squamous cells of undetermined significance: stratification of the risk of association with, or progression to, squamous intraepithelial lesions based on morphologic subcategorization. Acta Cytol. 1997; 41: Williams ML, Rimm DL, Pedigo MA, Frable WJ. Atypical squamous cells of undetermined significance: correlative histologic and follow-up studies from an academic medical center. Diagn Cytopathol. 1997;16: Collins LC, Wang HH, Abu-Jawdeh GM. Qualifiers of atypical squamous cells of undetermined significance help in patient management. Mod Pathol. 1996;9: ASCCP Practice Guideline. Management guidelines for follow-up of atypical squamous cells of undetermined significance (ASCUS). Colposcopist. Winter 1996;XXVII(1): Cox JT, ASCCP Practice Committee. ASCCP practice guidelines: management of glandular abnormalities in the cervical smear. J Lower Genital Tract Dis. 1997;1: Raab SS, Geisinger KR, Silverman JF, Thomas PA, Stanley MW. Interobserver variability of a Papanicolaou smear diagnosis of atypical glandular cells of undetermined significance. Am J Clin Pathol. 1998;11: DiTomasso JP, Ramzy I, Mody DR. Glandular lesions of the cervix: validity of cytologic criteria used to differentiate reactive changes, glandular intraepithelial lesions and adenocarcinoma. Acta Cytol. 1996;4: 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. 1995;14: Selvaggi S. Please! No AGUS swamp. Diagn Cytopathol. 1999;21: Arch Pathol Lab Med Vol 124, February 2 Atypical Epithelial Cells and Specimen Adequacy Davey et al 211