Cytology of gastrointestinal (GI) samples has a high

Transcription

1 Comparison of ThinPrep Preparations to Other Preparation Types in Gastrointestinal Cytology Observations From the College of American Pathologists Interlaboratory Comparison Program in Nongynecologic Cytology Amy C. Clayton, MD; Joel S. Bentz, MD; Patricia G. Wasserman, MD; Mary R. Schwartz, MD; Rhona J. Souers, MS; Beth Anne Chmara, CT(ASCP); Rodolfo Laucirica, MD; Karen M. Clary, MD; Ann T. Moriarty, MD; for the College of American Pathologists Cytopathology Resource Committee N Context. Differences in participant responses for Thin- Prep (TP) and non-thinprep (NTP) preparations for gastrointestinal cytology challenges, which circulated in the College of American Pathologists Interlaboratory Comparison Program in Nongynecologic Cytology (CAP NGC), may help to identify performance variations between preparation types. Objective. To compare the performance of TP-prepared slides of gastrointestinal exfoliative cytology specimens to that of NTP preparations in the CAP NGC program. Design. Participant responses between 2000 and 2007 were evaluated for esophageal wash/brush, gastric wash/ brush, and biliary tract brush specimens with a reference diagnosis of adenocarcinoma, squamous cell carcinoma, carcinoid, or spindle cell neoplasm. ThinPrep challenges were compared with NTP preparations (conventional smears, cytospins) for discordant responses. Results. In all, 6023 pathologist responses and 3825 cytotechnologist responses were reviewed. Non-ThinPrep preparations comprised 93% (n = ) of the challenges, while 7% (n = 912) were TP material. A match for a positive/suspicious diagnosis was seen in 88.5% of NTP and 95.9% of TP preparations (P,.001). These results were statistically significant when the specific reference diagnosis was adenocarcinoma (P,.001). Overall performance of cytotechnologists was not different from that of pathologists (89.2% versus 89.0%; P =.75). Cytotechnologists had better performance for detecting squamous cell carcinoma (96.3% versus 92.6%; P,.001), while pathologists had better performance for detecting spindle cell neoplasm (79.7% versus 42.9%; P,.001). Conclusions. ThinPrep preparations performed significantly better than NTP preparations in gastrointestinal cytology specimens circulated in an interlaboratory comparison program. Performance varied by reference interpretation, with the best performance for the interpretation of adenocarcinoma. Cytotechnologists and pathologists performed at the same level overall, but with differences for the diagnosis of spindle cell neoplasm and squamous carcinoma. (Arch Pathol Lab Med. 2010;134: ) Accepted for publication January 26, From Anatomic Pathology, Mayo Clinic, Rochester, Minnesota (Dr Clayton); the Department of Pathology, University of Utah, Salt Lake City (Dr Bentz); the Department of Pathology, Long Island Jewish Medical Center, New Hyde Park, New York (Dr Wasserman); the Department of Pathology, Methodist Hospital, Houston, Texas (Dr Schwartz); Biostatistics Department (Ms Souers) and Surveys Department (Ms Chmara), College of American Pathologists, Northfield, Illinois; the Department of Pathology, Baylor College of Medicine, Houston, Texas (Dr Laucirica); the Department of Pathology, Rochester General Hospital, Rochester, New York (Dr Clary); and AmeriPath Indiana, Indianapolis (Dr Moriarty). The authors has no relevant financial interest in the products or companies described in this article. This study and article were prepared without financial support other than the resources provided by the College of American Pathologists. Reprints: Amy C. Clayton, MD, Division of Anatomic Pathology, Department of Laboratory Medicine and Pathology, Hilton 11, Mayo Clinic, Rochester, MN ( clayton.amy@mayo.edu). Cytology of gastrointestinal (GI) samples has a high specificity but often has significantly lower sensitivity for detecting malignancy. 1 3 For example, brush cytology of pancreaticobiliary strictures has a specificity approaching 100%, but a sensitivity ranging from 37% to 89%, with an average of 59%. Both sampling error and interpretation error may account for this lower sensitivity. Gastrointestinal specimens can be processed for cytologic examination by a number of different methods, including conventional smears, membrane filters, cytocentrifugation or cytospin (CS), ThinPrep (TP), and SurePath preparations. A number of studies have suggested that cytology slides prepared with some of the newer liquid-based preparations (TP or SurePath methodologies) may offer improved cellularity and cellular preservation, 4 6 as well as the potential for using these slides for molecular studies designed to enhance diagnosis, treatment evaluation, and prognosis. 7 Several studies 8 13 have reviewed the morphology and performance of TP preparations in GI cytology Arch Pathol Lab Med Vol 134, August 2010 Comparison of Preparation Methods for GI Cytology Clayton et al

2 The TP method of cytology preparation has been available for use with nongynecologic specimens since Because of the potential to improve slide quality and the ensuing familiarity with TP from gynecologic cytology, many laboratories are switching to this technology for a variety of nongynecologic specimens, including respiratory specimens, body cavity fluids, and urine. Since the introduction of TP for cytology preparation, few large comparison studies of preparation type for gastrointestinal cytology specimens have been reported in the literature. The College of American Pathologists Interlaboratory Comparison Program in Nongynecologic Cytology (CAP NGC) began in In the following years, it grew to include more than 2000 laboratories and participants. As a result of this shift in the processing paradigm of nongynecologic samples, this program contains challenges including both non-thinprep (NTP) and TP material. We reviewed the responses of pathologists and cytotechnologists for GI cytology cases distributed from 2000 to 2007 in the CAP NGC program. Our primary goal was to determine if there was a significant difference in the participants ability to diagnose a variety of neoplastic GI lesions when comparing NTP-prepared specimens to TPprepared specimens in this educational interlaboratory comparison program. METHODS The members of the CAP Cytopathology Resource Committee (CRC) submit cases used for the CAP NGC program. Cases are selected for the NGC program by consensus at CRC screening sessions. Each case is independently reviewed by 2 members of the committee, who must agree on general and exact reference interpretations for acceptance of the slide into the program. The 2 members must also agree that the slide is a good representation of the entity, and that it is technically adequate for assessment. Each slide distributed in the program has a specific reference diagnosis assigned to a body site, as well as a general category of unsatisfactory, negative/benign, positive/malignant, or suspicious. The CAP NGC consists of 4 educational challenges per year, with each challenge composed of 5 slides mailed to participants. The slides are circulated to participants who voluntarily subscribe to the continuing education program. After review of the slides, participants are able to choose a general category (benign/negative, suspicious, positive/ malignant, unsatisfactory) and a specific reference diagnosis, in which participants hone their responses to a more specific diagnosis. After circulation to many laboratories, all slides have a unique performance profile that can be compared against others in the same reference category. The quality and performance of the slides is continually monitored. Every 6 months, each slide associated with a slide set is evaluated for stain quality and diagnostic value, as assessed by a CAP cytotechnologist. Any slide with a discordant rate of 70% or higher after circulation is reviewed by a CRC member to determine if it will continue circulation in the program. Any slide that has been identified by a participant on the result form as technically unacceptable is reviewed by the CRC. Slides may be retired, most often because of breakage or fading stain. This quality control system ensures that the slide challenges remain equivalent over time, such that newer slides entering into the program perform similarly to older slides in circulation. Table 1. Participant Concordance Rate for Preparation Type and the General Category of Positive/ Malignant and Suspicious Conventional Cytospin ThinPrep No. (%) (88.7) 471 (83.9) 912 (95.9),.001 We reviewed the cumulative histories of all participant responses for GI cytology challenges with a reference diagnosis of adenocarcinoma, squamous cell carcinoma, carcinoid, or spindle cell neoplasm, which were obtained through the CAP SCORES computer system (Northfield, Illinois) and circulated in the program from 2000 to A separate comparison for GI challenges with a reference diagnosis of benign was also performed for the same period. Gastrointestinal samples included esophageal wash/brush, gastric wash/brush, and biliary tract brush specimens. The preparation type was noted for each challenge. The specimens used in this analysis included conventional smears (either direct from brushing specimens or after laboratory concentration methods from a liquid specimen the distinction between direct smears and laboratory concentrated and prepared smears was not available to the authors of this study), cytospin preparations, and TP material (Hologic Inc, Bedford, Massachusetts). Millipore filter preparations (Millipore Corporation, Billerica, Massachusetts) have not been circulated in the program. There were GI exfoliative specimens included in this analysis. Participant results were identified for misclassification for both the general categorization (eg, negative versus positive) and for the exact reference diagnosis on a given slide (eg, adenocarcinoma versus squamous cell carcinoma). Specific reference diagnoses included adenocarcinoma; squamous cell carcinoma; carcinoma, not otherwise specified; carcinoid; or spindle cell neoplasm. Breakdown by specific site (gastric brush, esophageal brush, and biliary tract brush) was also performed. A participant response was considered discordant (incorrect) if the participant general category response was unsatisfactory or negative and the reference interpretation was positive or suspicious. A correct response was a response that was concordant to either the general category or to the exact reference interpretation. General and exact diagnosis discordant rates between slide-preparation type, as well as participant type (pathologist versus cytotechnologist), were analyzed. Statistical analysis was performed with Pearson x 2 and Fisher exact tests to test the null hypothesis to ensure that no association exists between the participants performance and 2 distinct factors, namely, slide preparation and reader type. Performance by specific reference diagnoses was also tested for these 2 factors. Additionally, the Cochran-Mantel-Haenszel test was used for the analysis of preparation type controlling for the specific gastrointestinal site. All statistical tests were evaluated using P,.05 to indicate a significant result. All statistical analyses were performed with SAS version 9.1 (SAS Institute Inc, Cary, North Carolina). RESULTS A total of 6023 pathologist responses and 3825 cytotechnologist responses submitted for gastrointestinal cytology challenges were available for review within the neoplastic categories. Non-ThinPrep preparations com- Arch Pathol Lab Med Vol 134, August 2010 Comparison of Preparation Methods for GI Cytology Clayton et al 1117

3 Figure 1. Conventional smear of biliary brush specimen demonstrating adenocarcinoma. A, Papanicolaou stain demonstrates a relatively flat architectural arrangement. B, Papanicolaou stain reveals marked nuclear atypia, with associated partially obscuring necrotic debris (original magnifications 3400 [A] and 3600 [B]). Figure 2. ThinPrep of biliary brush specimen demonstrating adenocarcinoma. A, Papanicolaou stain reveals nuclear atypia similar to that of the cells depicted in Figure 1, but with enhanced preservation of the nuclear features. B, Papanicolaou stain demonstrates a greater 3-dimensional architecture, characteristic of liquid-based preparations (original magnifications 3600 [A] and 3400 [B]). prised 93% (n ) of the GI challenges, while 7% (n 5 912) were TP preparations. Of the NTP specimens, were conventional smear preparations (CSPs) and 471 were cytospin specimens (Table 1). An additional 8216 specimens from the benign category were also included in the study (CSP, 6324; cytospin, 941; and TP, 951). A statistically significant difference was noted between the diagnostic performance of TP specimen types and the conventional smear and cytospin preparation types (P,.001) (Table 1). Examples of representative images of conventional smear and TP preparations are demonstrated in Figures 1 and 2, respectively. Concordance to the general category of positive/malignant or suspicious was made in 95.9% of TP challenges as Table 2. Participant Concordance Rate for Preparation Type and the General Category of Benign Conventional, Cytospin, ThinPrep, No. (%) 6324 (96.3) 941 (96.9) 951 (98.1).01 compared to 88.7% for CSP and 83.9% for cytospin preparations. There was also a statistical difference between the 3 preparation types for the reference category of benign (Table 2), but the difference was less than that for the general category of positive/malignant or suspicious (CSP, 96.3; cytospin, 96.9; TP, 98.1). ThinPrep specimens also outperformed NTP preparations for the exact reference diagnosis of adenocarcinoma (TP, 95.7%; NTP, 87.5%; P,.001) (Table 2). Representative images of adenocarcinoma from a conventional smear preparation and from a Table 3. Participant Concordance Rate for Preparation Type and Exact Reference Diagnosis Diagnosis Non-ThinPrep ThinPrep Adenocarcinoma 8287 (87.5) 812 (95.7),.001 Squamous cell carcinoma 2976 (93.8) 43 (95.4)..99 Carcinoid tumor 126 (78.6) 0 (NA) Spindle cell neoplasm 120 (70.0) 0 (NA) Abbreviation: NA, not applicable Arch Pathol Lab Med Vol 134, August 2010 Comparison of Preparation Methods for GI Cytology Clayton et al

4 Table 4. Concordance Rate by Participant Type for Gastrointestinal Cytology Challenges Cytotechnologist Pathologist 3825 (89.2) 6023 (89.0).76 ThinPrep-processing method are shown in Figures 1 and 2. There was no significant difference between preparation types for the reference diagnosis of squamous cell carcinoma (TP, 95.4%; NTP, 93.8%; P..99) (Table 3). Carcinoid tumors and spindle cell neoplasms represented a minor portion of the NTP category but were not represented in the TP category; therefore, a specific comparison could not be made for these reference diagnoses. Pathologists and cytotechnologists had equivalent performance characteristics for most of the diagnostic categories (Tables 4 and 5). No performance difference between participant types was noted for either all categories combined (89.0% versus 89.2%; P 5.76) or adenocarcinoma (88.4% versus 88.0%; P 5.64). A difference was seen between pathologists and cytotechnologists in the category of carcinoid (83.7% versus 67.3%) but was not statistically significant (P 5.06). Cytotechnologists performed better than pathologists in diagnosing squamous carcinoma (cytotechnologist, 96.3%; pathologist, 92.6%; P,.001), but pathologists performed better than cytotechnologists for a specific reference diagnosis of spindle cell neoplasm (cytotechnologist, 42.9%; pathologist, 79.7%; P,.001). Performance characteristics by specimen site (esophageal brush, gastric brush, biliary brush) revealed significant performance improvement with TP preparations for both esophageal brush and gastric brush specimens (P,.001, Tables 6 and 7), but no significant difference for biliary brush specimens. Biliary brush specimens comprised 7.4% of the total GI challenges. Gastric and esophageal brush specimens made up 49.4% and 43.2% of the analysis group, respectively. DISCUSSION Exfoliative gastrointestinal cytology specimens are an important method for screening and diagnosing gastrointestinal malignancies. Traditionally, exfoliative GI samples for cytology have been prepared by using filters, concentrated smears, or cytocentrifugation. With the advent of newer liquid-based preparation technology in the laboratory, this instrumentation is being used with increased frequency for processing of nongynecologic samples, including GI specimens. Our results demonstrate that participants in the CAP NGC educational interlaboratory comparison program Table 5. Concordance Rates by Participant Type and Exact Reference Diagnosis for Gastrointestinal Cytology Challenges Diagnosis Cytotechnologist Pathologist Adenocarcinoma 2729 (88.0) 4404 (88.4).64 Squamous cell carcinoma 964 (96.3) 1441 (92.6),.001 Carcinoid tumor 52 (67.3) 49 (83.7).06 Spindle cell neoplasm 35 (42.9) 64 (79.7),.001 Table 6. Frequency Distribution of Participant Responses by Gastrointestinal Site (N = ) Site Frequency Percentage Gastric brushing/washing Esophageal brushing/washing Biliary tract brushing are better able to give the correct diagnosis for exfoliative gastrointestinal cytology specimens with TP specimens than with NTP preparations (cytospin or conventional smear). Malignancy was correctly diagnosed in 88.5% of the NTP preparations, as compared with 95.9% in the TP slides. This statistically higher rate of concordance for the TP specimens was also noted in the adenocarcinoma group, but not for squamous cell carcinoma. This difference in performance was statistically significant for both gastric and esophageal brush specimens. A difference was also seen in biliary brush specimens (not statistically significant, however). This performance difference was also seen in the reference category of benign, but not to the same degree, indicating that TP preparation may impact sensitivity more than specificity in GI cytologic specimens. Other studies with nongynecologic cytology preparation types have also demonstrated equivalent or superior performance with newer liquid-based specimen types compared to traditional preparations, but the study numbers were relatively small and the specimens were a mixture of nongynecologic cytology specimen types (predominantly body fluids or fine-needle aspiration specimens). In these reports, statistically significant differences for diagnostic accuracy were hard to achieve, although the preference was for liquid-based cytology because of perceived morphology enhancement. Two large studies similar to the current study design (CAP NGC Interlaboratory Comparison Program) found enhanced diagnostic performance of peritoneal and pelvic fluid TP specimens as compared to NTP preparations of the same body site, 17 although there was no difference when looking at all body fluids categorized together. Laucirica et al 18 demonstrated that a diagnosis of highgrade urothelial carcinoma was rendered correctly more often in urine TP specimens than NTP preparations. A large study (n ), 2 devoted specifically to biliary or pancreatic brush cytology, demonstrated a statistically significant difference (favorable to TP) in sensitivity and accuracy for TP specimens compared to conventional smear and cytospin preparations. Additional smaller studies on biliary or pancreatic brushings 11,13,19 failed to confirm a statistical difference for newer liquid-based cytology preparations compared to traditional preparations, but did state a preference for the former because of Table 7. Performance by Slide-Preparation Type Controlling for Gastrointestinal Site a Gastrointestinal Site Non-ThinPrep ThinPrep Gastric brushing/ washing 5925 (88.1) 248 (98.0),.001 Esophageal brushing/ washing 4428 (89.1) 419 (96.8),.001 Biliary tract brushing 612 (87.3) 208 (92.0).053 a Cochran-Mantel-Haenszel test (P,.001). Arch Pathol Lab Med Vol 134, August 2010 Comparison of Preparation Methods for GI Cytology Clayton et al 1119

5 the morphologic presentation. Minamiguchi and colleagues 9 found a lower false-negative rate for their TP biliary brushes than for direct smears in a study of 68 biliary brush specimens. To the best of our knowledge, there are no studies in the literature specifically comparing TP specimens of esophageal or gastric brushing specimens to other preparation types. Newer liquid-based preparations offer advantages over traditional preparations, which may lead to enhanced diagnostic accuracy, as described in the above studies. Hoda 20 summarized aspects that are favorable or unfavorable for newer liquid-based nongynecologic cytology specimens, after an extensive review of the current literature. Morphologic advantages include immediate fixation with minimal air-drying artifact, homogeneous cellular distribution with optimized cellular presentation, clean background, and minimal cellular overlap and clumping. In addition, more standardized collection and preparation processes are achieved with newer liquidbased preparations. The screening area is smaller and potentially may be screened in a more time-efficient manner. Residual fluid can be used for ancillary studies, which might not be possible for direct smears from brush specimens or small-volume specimens. Disadvantages enumerated in the same study relate to altered morphologic presentation of cell fragments (breakage of papillary groups and increased disaggregation of cells), smaller appearance of cells and nuclei because of rounding up in the liquid, and loss of important background material (colloid, stroma, mucus, chondromyxoid matrix). Loss of background material is not likely to negatively impact GI exfoliative cytology specimens, as the diagnosis of neoplasms from these specimens rarely depends on background findings. Improved cellular fixation with enhanced morphologic detail would certainly facilitate accurate diagnosis in these specimens, as the distinction between reactive gastrointestinal mucosal lining cells and neoplasm can be difficult and requires careful evaluation of nuclear sizes, shapes, and arrangements. In our study, participants ability to diagnose adenocarcinoma specifically improved with TP material, while their ability to diagnose squamous carcinoma did not differ between TP and NTP preparations. This suggests that the improved morphologic detail afforded by TP may be more important for recognizing adenocarcinoma than squamous carcinoma (see Figures 1 and 2). Pathologists were more successful at recognizing spindle cell neoplasms on the GI challenges. Gastrointestinal stromal neoplasms are unusual tumors, but they would be part of most surgical pathology practices over time. Cytopathologists often participate in surgical pathology practice responsibilities and would thus have more exposure to cases of spindle cell neoplasms than cytotechnologists. Cytotechnologists exposure to spindle cell neoplasms may be limited to educational programs such as the CAP NGC educational slide program. Conversely, cytotechnologists were better able to correctly diagnose squamous carcinoma than cytopathologists. The reason for this is not entirely clear, but could relate to the fact that cytotechnologists traditionally spend a large part of their practice screening gynecologic cytology specimens for squamous carcinoma and precursor lesions, thus affording them more familiarity with squamous neoplasms. In summary, TP challenges performed significantly better than NTP preparations of neoplastic exfoliative GI cytology in an educational interlaboratory comparison program. The GI challenges also performed better for the exact reference interpretation of adenocarcinoma, but no difference was observed for challenges for other malignancies. Cytotechnologists appear to perform at the same level as pathologists in interpretation of malignant GI specimens. Use of TP technique is becoming widespread for GI specimens, and the results of this study confirm that TP preparations perform better than NTP preparations in an educational interlaboratory comparison program. References 1. Stewart CJ, Mills PR, Carter R, et al. Brush cytology in the assessment of pancreatico-biliary strictures: a review of 406 cases. J Clin Pathol. 2001;54(6): Volmar KE, Vollmer RT, Routbort MJ, et al. Pancreatic and bile duct brushing cytology in 1000 cases: review of findings and comparison of preparation methods. Cancer. 2006;108(4): Kipp BR, Stadheim LM, Halling SA, et al. A comparison of routine cytology and fluorescence in situ hybridization for the detection of malignant bile duct strictures. Am J Gastroenterol. 2004;99(9): Papillo JL, Lapen D. Cell yield: ThinPrep vs cytocentrifuge. Acta Cytol. 1994;38(1): Piaton E, Faynel J, Hutin K, Ranchin MC, Cottier M. Conventional liquidbased techniques versus Cytyc Thinprep processing of urinary samples: a qualitative approach. BMC Clin Pathol. 2005;5: Piaton E, Hutin K, Faynel J, et al. Cost efficiency analysis of modern cytocentrifugation methods versus liquid based (Cytyc Thinprep) processing of urinary samples. J Clin Pathol. 2004;57(11): Tisserand P, Fouquet C, Marck V, et al. ThinPrep-processed fine-needle samples of breast are effective material for RNA- and DNA-based molecular diagnosis: application to p53 mutation analysis. Cancer. 2003;99(4): Duggan MA, Brasher P, Medlicott SA. ERCP-directed brush cytology prepared by the Thinprep method: test performance and morphology of 149 cases. Cytopathology. 2004;15(2): Minamiguchi S, McEvoy R, Fraig M, et al. Bile duct brushings on ThinPrep: experience with 68 specimens. Diagn Cytopathol. 2004;30(4): Sheehan MM, Fraser A, Ravindran R, et al. Bile duct brushings cytology improving sensitivity of diagnosis using the ThinPrep technique: a review of 113 cases. Cytopathology. 2007;18(4): Siddiqui MT, Gokaslan ST, Saboorian MH, et al. Comparison of ThinPrep and conventional smears in detecting carcinoma in bile duct brushings. Cancer. 2003;99(4): Wang HH, Sovie S, Trawinski G, et al. ThinPrep processing of endoscopic brushing specimens. Am J Clin Pathol. 1996;105(2): Ylagan LR, Liu LH, Maluf HM. Endoscopic bile duct brushing of malignant pancreatic biliary strictures: retrospective study with comparison of conventional smear and ThinPrep techniques. Diagn Cytopathol. 2003;28(4): Bishop JW, MacFarlane K, Cheubront D, et al. Cell recovery and appearance in thin-layer preparations in nongynecologic cytology. Anal Quant Cytol Histol. 1998;20(4): Elsheikh TM, Kirkpatrick JL, Wu HH. Comparison of ThinPrep and cytospin preparations in the evaluation of exfoliative cytology specimens. Cancer (Cancer Cytopathol). 2006;108(3): Leung CS, Chiu B, Bell V. Comparison of ThinPrep and conventional preparations: nongynecologic cytology evaluation. Diagn Cytopathol. 1997; 16(4): Moriarty AM, Schwartz MR, Ducatman BS, et al. A liquid concept do classic preparations of body cavity fluid perform differently than ThinPrep cases? Arch Pathol Lab Med. 2008;132(11): Laucirica R, Bentz JS, Souers RS, et al. Do liquid-based preparations of urinary cytology perform differently than classically prepared cases: observations from the College of American Pathologists Interlaboratory Comparison Program in Nongynecologic Cytology. Arch Pathol Lab Med. 2010;134(1): Ylagan LR, Liu LH, Maluf HM. Endoscopic bile duct brushing of malignant pancreatic biliary strictures: retrospective study with comparison of conventional smear and ThinPrep techniques. Diagn Cytopathol. 2004;30(4): Hoda RS. Non-gynecologic cytology on liquid-based preparations: a morphologic review of facts and artifacts. Diagn Cytopathol. 2007;35(10): Arch Pathol Lab Med Vol 134, August 2010 Comparison of Preparation Methods for GI Cytology Clayton et al