Context.—Liquid-based preparations (LBPs) and human papillomavirus testing have led to changes in cervical cytology practices. The College of American Pathologists attempts to track practice patterns using a supplemental questionnaire, which allows laboratories to report diagnostic practices.
Objective.—To analyze the 2006 reporting practices and to compare the results with the 2003 survey data.
Design.—Questionnaire was mailed to 1621 laboratories. Participants included laboratories enrolled in the 2006 College of American Pathologists Gynecologic Proficiency Testing Program or the educational Interlaboratory Comparison Program in Gynecologic Cytology.
Results.—Of the 679 responding laboratories (response rate, 42%), most (97.8%; n = 664) had implemented the Bethesda 2001 terminology. The median rate for all preparations with low-grade squamous intraepithelial lesions was 2.5% (2.9% for LBPs) compared with a 2003 median rate of 2.1%; the increase was confined to LBPs. Rates for high-grade squamous intraepithelial lesions (median, 0.5%) and atypical squamous cells have changed little. High-grade squamous intraepithelial lesions and unsatisfactory rates varied at statistically significant levels between types of LBPs. Most atypical squamous cell cases were subclassified as undetermined significance (median, 4.3%). The median ratio of atypical squamous cells to squamous intraepithelial lesions and carcinomas for all specimen types combined was 1.5, similar to the 2003 median ratio of 1.4. The median rates for findings of squamous cell abnormalities for 2006 were significantly higher for LBPs than for conventional smears.
Conclusions.—Most responding laboratories have implemented the Bethesda 2001 terminology. There is an increase in LBP low-grade squamous intraepithelial lesion rates when compared with 2003 data. Liquid-based preparations have higher median squamous intraepithelial lesion and atypical squamous cell rates.
The College of American Pathologists (CAP) has conducted regular questionnaire surveys assessing cervical cytology laboratory practices since the inception of the Interlaboratory Comparison Program in Cervicovaginal Cytology.1–5 These surveys have become the basis for the benchmarks in the CAP Laboratory Accreditation Program). The first benchmarks were established using questionnaire surveys,6 conducted in 1996 and 1997, and were subsequently updated in 2003 and in 2008 using supplemental questionnaire data.
There have been 4 shifts in the practice of cytopathology since the 2003 supplemental questionnaire. First, widespread implementation of the 2001 Bethesda System7 (TBS 2001) has occurred since its publication in 2002. Although TBS 2001 has been in use for more than 5 years, the present questionnaire investigates the extent of TBS 2001 permeation in enrolled laboratories as compared with the data obtained in 2003. Second, widespread use of liquid-based preparations (LBPs) for cervical cytology may affect reporting rates for various descriptive diagnostic categories. Third, image-based and automatic-screening technologies have been introduced, which may affect reporting practices in cytology. Finally, human papillomavirus (HPV) testing in conjunction with a Papanicolaou test in women older than 30 years and as a triage test for atypical squamous cells of undetermined significance (ASC-US) in women older than 20 years are well established.8,9 The goal of the Interlaboratory Comparison Program in Cervicovaginal Cytology 2006 supplementary questionnaire was to evaluate and quantify the changes in cytology laboratory practices, to identify the extent of the implementation of new technologies, and to quantify the reporting rates of LBPs.
The 2006 Interlaboratory Comparison Program in Cervicovaginal Cytology supplementary questionnaire was mailed to 1621 laboratories enrolled in the program in mid 2006; 679 laboratories returned the questionnaire (response rate, 42%). Not every laboratory responded to every question; the number of laboratories responding is indicated for each question in the tables.
Laboratories were asked to report 2006 data reflecting TBS 2001 categories for conventional smears and the 2 types of commercially available US Food and Drug Administration–approved LBPs (SurePath, BD/TriPath, Franklin Lakes, New Jersey; and ThinPrep, Hologic, Inc/Cytyc Corporation, Marlborough, Massachusetts). Participants were to leave blank those methods or categories that their laboratory did not use. In addition to the volume of Papanicolaou tests reported in TBS 2001 categories, laboratories were also queried as to their use of the non-TBS 2001 category low-grade squamous intraepithelial lesion cannot exclude higher grade lesion. Carcinoma and adenocarcinoma in situ cases were to be separated from the atypical glandular cell (AGC) category when laboratories reported data.
Laboratories were asked to report their practices regarding HPV test methodology: whether they provided low-risk HPV testing, testing location, the method of testing, testing volume per month, and percentage of high-risk HPV+ cases for the interpretive categories of ASC-US, atypical squamous cells, cannot exclude high-grade squamous intraepithelial lesion (ASC-H), AGC, and when used as a screening test in conjunction with a Papanicolaou test in women 30 years and older (screening positive). In addition, a question was asked regarding the circumstances in which HPV testing was offered. All answers were based on actual clinical practice from 2006 only.
Reporting-rate distribution tables were constructed for each interpretive category. Reporting rate distributions were not Gaussian. Therefore, results are shown as percentile reporting rates instead of standard deviations. The 50th percentile reporting rate represents the median. The benchmark data are based only on laboratories that reported volumes of greater than 500 gynecologic cytology specimens for that type per year and included laboratories using FocalPoint (BD/TriPath Imaging) and Thin Prep Imager (Hologic/Cytyc). Reporting rates of zero were considered acceptable for uncommon categories, as long as some valid rates were reported by the laboratory. Some laboratories provided only data for all combined methods.
The ratio of positive ASCs to squamous intraepithelial lesions was calculated from participant-reported data. The calculation is the percentage of ASC-US plus ASC-H divided by the sum of the low-grade squamous intraepithelial lesion (LSIL), high-grade squamous intraepithelial lesion, and carcinoma cases.
Differences between reporting rates for conventional and LBS types were evaluated using the Wilcoxon signed-rank sum test, a nonparametric analysis of paired data. The Kruskal-Wallis rank sum test was used to compare median results when appropriate.10 Results of the survey were compared with results of the 2003 questionnaire, when available.5
Most of the laboratories that responded (397 of 656; 60.5%) offered both conventional cervical smears and LBPs. Laboratories that perform only conventional smears dropped from 181 (24.4%) to 90 (13.7%). Those performing only LBPs increased from 69 (9.3%) to 167 (25.5%). Based upon the responses of the 656 laboratories, the average rate of test use for each test method was 22.7% for conventional Papanicolaous, 57.2% for ThinPrep, and 19.6% for SurePath. Cervical cytology testing volumes for the 2006 calendar year are shown in Table 1. The largest numbers of laboratories reporting conventional slide use were those reporting fewer than 5000 tests annually; 45 laboratories performed at least 20 000 conventional smears. For ThinPrep, 80 laboratories performed at least 20 000 specimens per year. There was an increase in the number of laboratories using the SurePath method in 2006, with 52 laboratories reporting at least 20 000 annual specimens. Eight percent of laboratories used Focal Point for conventional and SurePath slides, and 21% of laboratories evaluating ThinPrep specimens used the ThinPrep Imager for ThinPrep specimens.
Most laboratories (662 of 679; 97.5%) used TBS 2001 terminology. Approximately one-half of laboratories have modified the TBS 2001 terminology to include LSIL cannot exclude a higher-grade lesion (330 of 673; 49%). The TBS 2001 minimum squamous cellularity criteria had been adopted by 94.5% of the laboratories (637 of 674); 527 of the 667 responding laboratories (79%) answered “yes” to the question, “Does your laboratory use the minimum squamous cell adequacy criteria for atrophic and/or postirradiation specimen?” Two-thirds of the laboratories (444 of 666) routinely include an educational note in cervical cytology reports.
Reporting rates (benchmark data) for various TBS 2001 categories and preparation methods are shown in Table 2. Not all laboratories provided reporting rates, but the numbers of laboratories reporting this data are provided. Reporting rates are given as percentages of total case volume for the responding laboratories. The percentiles (5th, 10th, and so forth) refer to the distribution of individual laboratory responses ranked from lowest to highest, with the 50th percentile reporting rate corresponding to the median. These benchmark data exclude data from laboratories reporting fewer than 500 gynecologic cytology specimens/year for a given method because of small-sample bias. In other words, if a laboratory performed 5000 ThinPrep preparations and 50 conventional smears, only the ThinPrep data were included for benchmarking. Because of small numbers, the carcinoma and adenocarcinoma in situ categories are excluded from the benchmark data.
Squamous epithelial abnormality rates are generally higher for LBPs compared with conventional smears. Median ASC-US rates were higher for ThinPrep (4.9%) and SurePath (4.1%) than for conventional smears (2.4%; P < .001). The median ratio of positive ASCs to squamous intraepithelial lesions was the same for manual screening and automated or image-assisted methods (1.5%). The median LSIL rates for ThinPrep (3.0%) and SurePath (2.5%) preparations were greater than for conventional smears (1.3%; P < .001). Low-grade squamous intraepithelial lesion rates for SurePath preparations were consistent, regardless of laboratory test volumes (2.3%). ThinPrep median LSIL rates showed greater variability among laboratories with different volumes (Table 3). In Table 2, median high-grade squamous intraepithelial lesion rates were higher for ThinPrep (0.6%) than either SurePath or conventional smears (both 0.3%; P = .05, Kruskal-Wallis rank sum test). Atypical squamous cells, cannot exclude high-grade squamous intraepithelial lesion, rates were 0.3% for ThinPrep and SurePath preparations and 0.2% for conventional smears. The use of LSIL cannot exclude higher-grade lesions, and adenocarcinoma in situ was not of sufficient magnitude to evaluate with statistical confidence. The median rates of the ratio of positive ASCs to squamous intraepithelial lesions were slightly lower for LBPs (ThinPrep 1.5% and SurePath 1.6%) than for conventional smears (1.7%), but these differences may not be statistically significant. The mean unsatisfactory rates were 0.3% for SurePath compared with 1.1% for ThinPrep preparations (P = .05) and 1.0% for conventional smears. The AGC mean rates were 0.2% for LBPs and 0.1% for conventional smears. There was no statistically significant difference at the .05 level between the 2 LBPs for the other 5 categories (ASC-US, P = .44; ASC-H, P = .27; LSIL, P = .08; ASC/SIL, P = .48; and AGC, P = .53).
Table 4 compares median reporting rates from the 2003 and 2006 surveys5 for all methods combined. There is an increase in median LSIL detection from 2.1% in 2003 to 2.5% in 2006; there is a similar increase in the 90th percentile reporting rates from 4.6% to 5.1% over time. In contrast, median high-grade intraepithelial lesion detection rates showed no change in distribution (0.5%); the 90th percentile reporting rate for high-grade intraepithelial lesion was similar between 2003 (1.4%) and 2006 (1.5%). The 2006 median reporting rates for ASC-US and ASC-H were 4.3% and 0.3%, compared with 2003 of 3.9% and 0.2%. Reporting rates for AGCs were low and unchanged, with a median of 0.2% for both years. The median ratio of positive ASCs to squamous intraepithelial lesions was 1.5 in 2006, similar to the 1.4 median ratio from 2003. The median unsatisfactory cervical cytology rate increased to 0.9% in 2006, compared with the 2003 rate of 0.5% for all methods combined.
Adjunctive HPV testing was offered by 90.1% (445) of laboratories (Table 5). Four hundred and forty-five laboratories (72.7%) referred HPV testing to a reference laboratory. Digene Hybrid Capture is the leading test method, according to the 558 laboratories that provided specific methods. The median volume of HPV testing was 55 tests per month. One hundred and fifty-seven laboratories know their high-risk HPV+ rate for ASC-US cases, and the median response was 36.6%.
The CAP 2006 supplemental questionnaire documents widespread acceptance and implementation of TBS 2001.7 More than 97% of laboratories had implemented TBS 2001, and a similar number had adopted the new adequacy criteria. Twenty-eight percent modified TBS 2001 terminology to include the low-grade squamous intraepithelial lesion cannot exclude higher-grade lesion category. Another modification of TBS 2001 that 79% of laboratories used was extending the minimum squamous cellularity criteria for adequacy to preparations with atrophy or postirradiation changes. The Bethesda System 2001 explicitly states that minimum cellularity criteria should be used for cervical cytology specimens, and there may be clinical instances (such as atrophy) when fewer cells are adequate.11
Two-thirds of reporting laboratories incorporate educational notes routinely in their reports. The relatively high rate may reflect the CAP Laboratory Accreditation Program criterion (CYP.07582)6 that laboratories educate providers that cervical cytology is a screening test with inherent false-negative results. Although use of educational notes is widespread, there are other acceptable ways of satisfying this educational criterion, such as conferences, directed mailing, or newsletters. In addition to educating providers about the significance of results, educational notes may also be used for management recommendations based on published clinical guidelines.9
The Bethesda System 2001 terminology subdivides ASC into 2 subcategories and states that most cases should be ASC-US, with approximately 5% to 10% categorized as ASC-H cases.7 The survey confirms that laboratories are using these subcategories with the recommended frequency; approximately 93% of ASC cases were ASC-US and 7% were ASC-H, according to results of the median reporting categories. Cases reported as ASC-US can be followed by either HPV triage testing, repeat cytology, or colposcopy, whereas ASC-H cases are most appropriately managed by colposcopy.9,12 The percentage of cases with AGCs has changed relatively little during the past 6 years, indicating that the exclusion of adenocarcinoma in situ cases from the AGC category has had only minor effect on overall reporting rates. Adenocarcinoma in situ is rare and difficult to identify definitively,13 which most likely explains the lack of significant difference between years.
There was an increased rate of LSIL detection, with increased rates confined to LBPs, with a slight decrease in the LSIL reporting rate for conventional smears. Other reporting rates showed little change in the 4-year interval. There were differences between LBP and conventional smears in several other categories in 2006; LSIL and ASC-US rates were significantly higher in LBPs compared with conventional smears. Low-grade squamous intraepithelial lesion rates varied with laboratory volumes for ThinPrep but did not show variation for SurePath.
The survey demonstrated a ratio of positive ASCs to squamous intraepithelial lesions of 1.5 in 2006. There is no single correct ASCs to squamous intraepithelial lesion ratio, and this quality improvement monitor varies according to age and relative cancer risk in the patient population and among laboratories.14 Laboratories with well-screened, stable populations may have fewer squamous intraepithelial lesions and more borderline cases and, therefore, a higher ratio of positive ASCs to squamous intraepithelial lesions. In contrast, laboratories with high-risk patients may have more definitive squamous intraepithelial lesion cases and a lower ratio of positive ASCs to squamous intraepithelial lesions. There may be a behavioral upward shift in ASC-US rates because of the option of reflex HPV testing for patients older than 20 years with ASC-US. These data are valuable as a quality assurance monitor; rates of epithelial cell abnormalities found by cytotechnologists and pathologists are often compared with the laboratory average. Comparing HPV positive rates for ASC-US is also an effective quality assurance monitor.15
The CAP Laboratory Accreditation Program checklist incorporates these new benchmarking data. A phase 1 indicator in the checklist is, if the laboratory's annual ratio of positive ASCs to squamous intraepithelial lesions for gynecologic cases falls outside of the 5th or 95th percentiles, has the laboratory determined and documented the reasons? (CYP.07650).6 The new corresponding 5th and 95th percentile ratio of positive ASCs to squamous intraepithelial lesions reporting rates are 0.4 and 4.5 for conventional smears, 0.7 and 3.4 for ThinPrep preparations, and 0.7 and 2.8 for SurePath preparations.
Unsatisfactory rates have been predicted to increase with implementation of more objective TBS 2001 criteria, and rare studies have reported higher unsatisfactory rates.16 In the current survey, there was an increase with ThinPrep and conventional preparations, but not with SurePath specimens. The increase in reported unsatisfactory rates may reflect a more generalized application of squamous cellularity criteria, whereas the SurePath unsatisfactory rates may reflect differences in technology and collection processes between SurePath and other preparations.
A comprehensive discussion of this portion of the survey has been previously published describing HPV laboratory practices.17 Reference laboratory testing is the most common form of HPV testing, although laboratories performing in-house testing reported significantly higher monthly HPV volumes than in 2003. The median volume for HPV testing is 55 tests/mo. Low-risk HPV testing is widely available, despite its lack of clinical utility. Digene Hybrid Capture II is the methodology most often used for testing, and the most common reflex pattern is for the triage of ASC-US interpretations. The high-risk HPV median, as reported by responding laboratories, is 36.6% in ASC-US cases and is lower than both the recent Q Probe data18 and data from the ALTS (ASC-US/LSIL Triage Study for Cervical Cancer) trial.8
Although technologies, in particular, adoption of automated or image-guided screening, may lead to rapid changes in reporting rates and practices, the current survey demonstrates the median ratio of positive ASCs to squamous intraepithelial lesions is the same for manual screening and automated methods (1.5).
In summary, most cytology laboratories continue to use TBS 2001 terminology, LBP methods, and HPV testing. These median reporting rates reflect most notably an increase in ASC-US and LSIL detection but only a slight increase in the average ratio of positive ASCs to squamous intraepithelial lesions for all specimen types. These reporting rates replace previous benchmarking data included in the CAP Laboratory Accreditation Program checklists. Laboratories can compare their laboratory practices with these questionnaire data and find areas for possible performance improvement.
The authors have no relevant financial interest in the products or companies described in this article.
The College of American Pathologists (Northfield, Illinois) provided support for this study.
From the Department of Pathology, Quest Diagnostics Inc, Las Vegas, Nevada (Dr Eversole); the Department of Pathology, AmeriPath Indiana, Indianapolis (Dr Moriarty); the Department of Pathology, The Methodist Hospital, Houston, Texas (Dr Schwartz); the Department of Anatomic Pathology, Mayo Clinic, Rochester, Minnesota (Drs Clayton and Henry); the Departments of Biostatistics (Ms Souers) and Cytology Surveys (Mss Fatheree and Chmara), College of American Pathologists, Northfield, Illinois; the Department of Laboratory Services, Palomar Medical Center, Escondido, California (Dr Tench); the Department of Pathology, Massachusetts General Hospital, Boston (Dr Wilbur).