Performance of Candida--fungal-induced atypia and proficiency testing: observations from the College of American Pathologists proficiency testing program.
Abstract: * Context.--Candida may elicit cellular changes on otherwise negative screening Papanicolaou tests that may be misinterpreted as atypical squamous cells of undetermined significance. Although these changes have been correctly interpreted in the educational program of the College of American Pathologists, the Interlaboratory Comparison Program in Gynecologic Cytology, the performance of negative slides with Candida faltered when the same field validated slides were included in proficiency testing (PT).

Objective.--To identify the performance differences of negative for intraepithelial lesion (NILM) Candida challenges before and after PT.

Design.--We compare the performance of NILM College of American Pathologists slides with Candida as a single reference diagnosis, prior to PT (1991-2006) and after PT (2006-2007).

Results.--There were 147 186 responses for slides with NILM Candida from the College of American Pathologists programs from 1991 through 2007. After PT, 79.7% of incorrect participant responses identified Candida as Category C (low-grade squamous intraepithelial lesion), whereas prior to PT only 59.5% of the incorrect diagnoses were low-grade squamous intraepithelial lesion (P < .001) in the field-validated component of the program. Validated Candida slides performed significantly more poorly in PT (97.2%) than prior to PT (98.3%) (P < .001). Candida challenges performed better in the educational component post-PT (98.3% versus 97.2%; P < .001). Cytotechnologists (97.9%) identified Candida more frequently than pathologists (97.3%) (P < .001) and ThinPrep preparations performed the best of all preparation types.

Conclusions.--Proficiency testing adversely affects the performance of participants in the identification of NILM Candida. Slides with Candida are more likely to be identified as low-grade squamous intraepithelial lesion in a PT exercise. Misidentification is not due to lack of recognition but most likely an attempt of test takers to optimize their likelihood of passing the examination.

(Arch Pathol Lab Med. 2009;133:1272-1275)
Article Type: Report
Subject: Candida (Health aspects)
Cancer cells (Properties)
Cancer (Diagnosis)
Cancer (Methods)
Cancer (Standards)
Authors: Moriarty, Ann T.
Darragh, Teresa M.
Fatheree, Lisa A.
Souers, Rhona
Wilbur, David C.
Pub Date: 08/01/2009
Publication: Name: Archives of Pathology & Laboratory Medicine Publisher: College of American Pathologists Audience: Academic; Professional Format: Magazine/Journal Subject: Health Copyright: COPYRIGHT 2009 College of American Pathologists ISSN: 1543-2165
Issue: Date: August, 2009 Source Volume: 133 Source Issue: 8
Topic: Event Code: 350 Product standards, safety, & recalls
Geographic: Geographic Scope: United States Geographic Code: 1USA United States
Accession Number: 230247116
Full Text: Candida is a fungal organism commonly identified in cervical cytology specimens (Papanicolaou tests). The cellular reactions associated with Candida are well described and include nuclear enlargement, hyperchromasia, cytoplasmic orangeophilia, and perinuclear halos. Because of these cellular changes, negative slides with Candida are sometimes misinterpreted as atypical squamous cells of undetermined significance. (1)

The College of American Pathologists (CAP) Interlaboratory Comparison Program in Cervicovaginal Cytopathology (PAP) was originally an educational quarterly mailed glass slide quality improvement program, which included Candida challenges as a component of the negative category (negative for squamous intraepithelial lesion [NILM]) in its mailings since the inception of the program in 1989. Slides that demonstrate Candida historically performed well in the CAP PAP educational interlaboratory comparison program. Candida was one of the most reproducible (precise) categories in the CAP PAP program. (2) Slides that performed well in the educational arm of the CAP PAP became "graded" challenges after field validation. Educational slides became field validated in CAP PAP when they satisfied the following criteria: (1) at least 20 responses; (2) 90% of the responses for that slide were in the correct series, with a standard error of less than 0.05; and (3) the results reported in the category of NILM showed at least 50% match to the exact reference interpretation. Candida was always a very robust category during field validation of liquid-based cytology challenges in the CAP PAP. (3)

In 2005, national federally mandated gynecologic cytology proficiency testing (PT) began, and the CAP was approved as a PT provider for 2006 (PAP PT). Proficiency testing consists of an initial 10-slide test, which must include at least one slide from each of 4 categories: Category A, Unsatisfactory; Category B, NILM, which includes specific organisms such as Candida; Category C, low-grade squamous intraepithelial lesion (LSIL); and Category D, high-grade squamous intraepithelial lesion or carcinoma. The Center for Medicare Services also required that Category C and D slides be biopsy confirmed prior to use in the PT program.

The PAP PT program fulfills the Center for Medicare Services requirements but additionally uses field-validated slides from the original CAP PAP in its PT program. The CAP PAP PT program uses many of the same slides that originated in the CAP PAP program and were field validated in the educational program. Despite using slides with robust performance characteristics, the performance of Candida challenges has faltered since implementation of PT in 2006. Slides with Candida do not seem to perform as well as slides in the educational program, despite rigorous field validation. Some of the slides that did not perform well in PT were the same slides that easily field validated in the educational programs. To confirm the magnitude of this change, the performance of Candida PT challenges and educational challenges in 2006 and 2007 are compared with graded and educational slides with Candida in the CAP PAP from 1991 through 2005, prior to PT.

MATERIALS AND METHODS

All participant responses for slides circulated with fungal organisms consistent with Candida (reference category 111) were captured from the CAP database from 1991 through 2007. Only slides with the single diagnostic entity of Candida were used. Any slides demonstrating more than one diagnostic category were excluded from analysis. The years 1991 through 2005 were defined as those responses prior to PT (pre-PT). "Post-PT" challenges were from 2006 and 2007, after acceptance of the CAP as a national PT provider.

After PT implementation, the responses of the participants were recorded as the required responses used in PT: unsatisfactory (Category A), NILM (Category B), LSIL (Category C), and carcinoma (Category D). Prior to PT, the responses were recorded as the exact reference diagnoses used by the CAP PAP (Table 1).

In the pre-PT population, the response was considered incorrect if it was answered as any response in the 200 series (LSIL or worse in the original CAP PAP program). Post-PT the response was incorrect if it was answered as Category C or D.

A Pearson chi-square test was used to test the null hypothesis that there is no association between the reference diagnosis of Candida on field-validated and non-field-validated slides pre-PT and post-PT. Candida slides that were incorrectly identified as Category C or LSIL and Category D or high-grade squamous intraepithelial lesion were also analyzed pre-PT and post-PT implementation. Differences between slide preparation type (conventional, ThinPrep [Hologic, Marlborough, Massachusetts], or SurePath [Becton, Dickinson, and Company, Franklin Lakes, New Jersey]), and participant category (cytotechnologist, pathologist) were evaluated controlling for validation status.

All tests were run at a .05 significance level. All statistical analyses were performed using SAS v9.1 (SAS Inc, Cary, North Carolina).

RESULTS

There were 147 186 responses for the reference diagnosis of Candida from the CAP programs from 1991 through 2007. There were significant differences in all categories. The comparison of Candida challenges for field-validated slides demonstrated a significant difference pre-PT and post-PT, with post-PT challenges performing more poorly (Table 2). In the educational program there was a significantly better performance of Candida slides after PT implementation (Table 2). The incorrect responses were more often LSIL (Category C) post-PT and less often Category D(P < .001) (see Table 3).

Participants who received ThinPrep challenges had better performance than those who received conventional preparations (Table 4) and cytotechnologists more often correctly identified Candida than did pathologists (Table 5).

COMMENT

Fungal organisms morphologically consistent with Candida species are frequently seen on Papanicolaou tests. Candida may present as budding yeasts or pseudohyphae. The pseudohyphae may "spear" epithelial cells. This effect has variably been called the "lei," "puka bead," or "shish kabob" effect. The organism may incite significant inflammatory changes that may be confused with preneoplastic cellular changes. The nuclear enlargement, hyperchromasia, and perinuclear halos may be so pronounced that Candida infections may be called atypical squamous cells of undetermined significance or LSIL. (1) Figure 1 from the CAP PAP program and Figure 2 from the authors' file demonstrate the reactive cellular features that may be seen with Candida.

Despite the cellular reaction that may accompany the fungal organism, slides demonstrating Candida organisms have performed reliably in the CAP Interlaboratory Comparison Program in Gynecologic Cytology. Candida challenges have been identified as one of the most reproducible or "least difficult" reference diagnoses. Candida challenges had participant responses that exactly matched the reference diagnosis 100% of the time in 46% and 41% of the validated and educational slide responses. (2) Candida has been a robust category in the CAP PAP program, requiring very few reviews to acquire an error rate of 0.05 or less. (3) Prior to PT, slides with Candida easily validated and became graded slides rapidly in the CAP PAP program. It appears that although Candida organisms may induce changes that mimic preneoplastic changes, participants were confident in their interpretation of Candida. (4)

However, after PT implementation, there appeared to be a decline in performance of the validated Candida slides. In many instances these were the same slides that performed flawlessly as graded slides in the educational program. Candida cases performed significantly differently after PT implementation. The slides were more often called LSIL (Category C) after implementation of PT than before PT. Interestingly, Candida in the educational component of the PT program performed significantly better than the educational component pre-PT. The cellular changes associated with Candida apparently can be identified as reactive changes in an educational venue but are upgraded in the testing scenario. This confirms the observation of Hughes et al, (5) that PT itself influences the performance of participants and that participants more often upgrade borderline cases to avoid the punitive effects of "missing" a significant lesion in the testing environment. This study confirms that PT participants upgrade the reactive features seen with Candida to LSIL, probably to avoid a loss of points due to missing a possible dysplasia.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

There was a significant difference in performance due to preparation type. ThinPrep slides with fungal organisms consistent with Candida performed significantly better across the programs. Renshaw et al (4) noted this in field-validation studies in 2005, and the trend continues. It is postulated that the difference in robustness may be due to the relatively clean background of the ThinPrep cases.

Cytotechnologists have demonstrated higher specificity than pathologists in the CAP PAP program. (6) They correctly identify NILM cases as benign more often than pathologists, and this is seen for Candida challenges. Cytotechnologists have greater experience with the reactive cellular features of Candida during their screening workload. Most cases of Candida are not routinely escalated to pathologist review; hence pathologists only routinely evaluate Candida cases showing potential abnormal features. This performance continues as demonstrated in Table 5.

In summary, pathologists are more likely to overcall Candida as an intraepithelial lesion as compared with cytotechnologists. ThinPrep Candida challenges perform better than other preparations. Finally, when validated slides demonstrating Candida are used in PT, they are more likely to be called a squamous intraepithelial lesion most likely because of the test-taking event not because of lack of proficiency in identification of Candida. In educational programs, the same slides are significantly more likely to be identified correctly as demonstrating Candida and its associated cellular changes, reinforcing this hypothesis of bias associated with the test-taking environment.

References

(1.) Miguel NL, Lachowicz CM, Kline TS. Candida-related changes and ASCUS: a potential trap! Diagn Cytopathol. 1997;16:83-86.

(2.) Renshaw AA, Davey DD, Birdsong GG, et al. Precision in gynecologic cytologic interpretation: a study from the College of American Pathologists Interlaboratory Comparison Program in Cervicovaginal Cytology. Arch Pathol Lab Med. 2003;127:1413-1420.

(3.) Renshaw AA, Walsh MK, Blond B, Moriarty AT, Mody DR, Colgan TJ, for the Cytopathology Resource Committee, College of American Pathologists. Robustness of validation criteria in the College of American Pathologists Interlaboratory Comparison Program in Cervicovaginal Cytology. Arch Pathol Lab Med. 2006;130:1119-1112.

(4.) Renshaw AA, Wang E, Mody D, Wilbur DC, Davey DD, Colgan TJ. Measuring the significance of field validation in the College of American Pathologists Interlaboratory Comparison Program in Cervicovaginal Cytology: how good are the experts? Arch Pathol Lab Med. 2005;129:609-613.

(5.) Hughes JH, Bentz JS, Fatheree L, Souers R, Wilbur DC. Changes in participant performance in the "test-taking" environment: observations from the 2006 College of American Pathologists Gynecologic Cytology Proficiency Testing Program (PAP PT). Arch Pathol Lab Med. 2009;133(2):279-282.

(6.) College of American Pathologists 2004 PAP Year End Summary Report. PAPC, PAPK and PAPM; 2004. College of American Pathologists: Northfield, IL. 2009.

Ann T. Moriarty, MD; Teresa M. Darragh, MD; Lisa A. Fatheree, BS, SCT(ASCP); Rhona Souers, MS; David C. Wilbur, MD

Accepted for publication November 10, 2008.

From the Department of Pathology, AmeriPath, Indiana; Indianapolis (Dr Moriarty); the Department of Pathology and Cytology, University of California-San Francisco Mount Zion Medical Center Clinic, San Francisco (Dr Darragh); Cytology Surveys (Ms Fatheree) and Statistics (Ms Souers), College of American Pathologists, Northfield, Illinois; and Department of Pathology, Massachusetts General Hospital, Harvard Medical School, Boston (Dr Wilbur).

The authors have no relevant financial interest in the products or companies described in this article.

Reprints: Ann T. Moriarty, MD, AmeriPath Indiana, 2560 N Shadeland Ave, Suite A, Indianapolis, IN 46219-1739 (e-mail: amoriarty@ ameripath.com).
Table 1.   Series Specific Reference Diagnoses and
Series Numbers Used in the College of American
Pathologists Interlaboratory Comparison Program Prior
to Proficiency Testing

   Series 0        Series 100           Series 200

Unsatisfactory   NILM (101)       LSIL (201)
(001)            Candida (111)    HSIL (211)
                 Trichomonas      Squamous cell carcinoma
                   (113)            (221)
                 Herpes (115)     Adenocarcinoma (225)
                 Repair (120)     HSIL/carcinoma (226)
                                  Nonepithelial malignant
                                    neoplasm (227)

Abbreviations: HSIL, high-grade squamous intraepithelial lesion or
malignancy; LSIL, low-grade squamous intraepithelial lesion; NILM,
negative for intraepithelial neoplasm or malignancy without specific
organisms or repair.

Table 2. Comparison of Responses for Candida Challenges for
Field-Validated and Educational Slides Prior to and
After Implementation of National Proficiency Testing (PT)

Candida Reference Category              Pre-PAP PT, %

Field-validated correct responses   98.3 (n = 48 850) (a)
Educational correct responses       97.2 (n = 53 71 6)

Candida Reference Category              PAP-PT, %       P Value

Field-validated correct responses   97.2 (n = 271 76)    <.001
Educational correct responses       98.3 (n = 1 7375)    <.001

Abbreviations: PAP-PT, after implementation of College of American
Pathologists PT program; Pre-PAP PT, prior to the College of American
Pathologists PT program.

(a) n indicates number of total responses for Candida challenges
for each period indicated.

Table 3. Incorrect Diagnosis for Candida by Category
for Field-Validated Slides

Incorrect      Pre-PAP PT, %   PAP PT, %
Diagnosis        (N = 841)     (N = 758)   P Value

Cat C (LSIL)       59.5          79.7       <.001
Cat D (HSIL+)      40.5          20.3       <.001

Abbreviations: Cat C (LSIL), Category C (low-grade squamous
intraepithelial lesion); Cat D (HSIL+), category D (high-grade
squamous intraepithelial lesion or carcinoma); PAP PT, after
implementation of College of American Pathologists proficiency
testing program; Pre-PAP PT, prior to the College of American
Pathologists proficiency testing program.

Table 4. Candida Response by Slide Preparation Type (a)

Candida     Conventional, %   ThinPrep, %    SurePath, %
Response     (N = 64 384)     (N = 71 065)   (N = 11 737)

Correct          97.1             98.2           97.8
Incorrect         2.9              1.8            2.2

(a) P < .001.

Table 5.   Candida Response by Type of Participant (a)

Candida     Cytotechnologist, %   Pathologist, %
Response       (N = 59 641)        (N = 62 414)

Correct            97.9                97.3
Incorrect           2.1                 2.7

(a) P < .001.
Gale Copyright: Copyright 2009 Gale, Cengage Learning. All rights reserved.