Romanian Journal of Morphology and Embryology 2009, 50(2):

Romanian Journal of Morphology and Embryology 2009, 50(2):245 250 ORIGINAL PAPER Liquid Based Preparation (LBP) cytology versus Conventional Cytology (CS) in FNA samples from breast, thyroid, salivary glands and soft tissues. Our experience in Crete (Greece) N. MYGDAKOS 1), SYLVIA NIKOLAIDOU 1), ANNA TZILIVAKI 2), D. TAMIOLAKIS 2) 1) Department of Cytopathology, Venizeleion Hospital of Heraklion, Crete, Greece 2) Department of Cytopathology, Regional Hospital of Chania, Crete, Greece Abstract Background: The improvement in quality of cytological preparations with the use of LBP methodology has been well-documented, but the cytological artifacts resulting from this technique have not been adequately described. This study describes and illustrates the cytological artifacts introduced by LBP technique when used on fine-needle aspirates (FNAs), and evaluates these artifacts as potential diagnostic pitfalls. Study Design: We reviewed a total of 96 FNAs simultaneously processed by both conventional smears and LBP. FNAs were obtained from the following sites: lymph node (38), breast (28), soft-tissue sites (nine), salivary glands (six), and thyroid gland (15). Results: The LBP smears were consistently devoid of obscuring elements, and the cells were adequately preserved and evenly dispersed. However, we noted some cytomorphological alterations that should be recognized to avoid erroneous diagnoses. The size of cell clusters was decreased, large branching sheets were fragmented, and there were more single cells, resulting in apparent discohesion. Small cells such as lymphocytes tended to aggregate. All cells were generally smaller and occasionally spindled, the chromatin detail was attenuated, and nucleoli were more prominent. Intranuclear inclusions were difficult to visualize. Background matrix was often altered in both quantity and quality. Extracellular particles, small mononuclear cells, red blood cells, and myoepithelial cells were markedly decreased in number. Conclusions: Cytopathologists should be careful in interpreting FNAs prepared using LBP technique if that is the only methodology employed. Familiarity with artifacts is essential to avoid misdiagnoses. Keywords: Liquid Based Preparation, FNA, breast, thyroid, salivary glands, soft tissues. Introduction Today, liquid based preparation (LBP) is becoming increasingly popular for evaluating non-gynecologic cytology specimens including FNA. A number of morphologic changes are inherent to LBP including altered, reduced or lost background material, smaller and more fragmented cell clusters, smaller cell size, well preserved nuclear detail, more prominent nucleoli, and more easily visualized cytoplasm. Other advantages of LBP include rapid fixation, even distribution of cells over a smaller slide area, and decreased obscuring background elements, such as blood, inflammation, and mucus. In addition, standardized LBP fixation provides advantages for centralized laboratories, especially when FNA procedures are carried out without rapid assessment. The advantages of LBP independent of diagnostic accuracy and morphology have been well-described [1 4]. Because one slide is prepared for each case, and the cellular contents are confined within a 20-mm diameter circle, the time required for evaluation by screeners and cytopathologists may be reduced. The residual material in fixative solution allows ancillary studies, such as immunohistochemistry, to be performed. Although immunohistochemistry may also be performed on cellblock material, this is not required for all FNA cytology materials, for example breast and salivary gland FNAs where morphology is usually sufficient for a diagnosis. Clinicians may prefer the simplicity of direct needle rinsing in fixative solution to preparing conventional smears (CS). In addition, the LBP sample lacks many of the CS artifacts, which may hinder interpretation, and there is a decreased need to provide on-site assessment for adequacy. In addition, greater cell dyshesion with increased single cells and smaller fragments in LBP may make interpretation challenge. Material and Methods A total of 96 cases were reviewed. In each one, FNA was performed using a 23-gauge needle, 20-mL syringe and syringe pistol. In each case, three passes were carried out. The first step produced a conventional preparation and direct smear that was immediately

246 cytofixed for Papanicolaou stain. One or more smears were also made for Diff-Quik (Harleco, Gibbstown, New Jersey, USA). For LBP smears an additional step was done. The material was collected in 30 ml of CytoLyt solution (Cytyc), to be centrifuged at 500 g for 10 minutes. The supernatant fluid was discarded and the material resuspended in a cytopreservative solution (PreservCyt, Cytyc). After 15 minutes, the material was processed in ThinPrep 2000 (Cytyc). In each case of LBP, one smear was stained by the Papanicolaou or the May-Grünwald Giemsa method and the remaining material was used to make additional smears for immunostaining if necessary. Two observers examined the CSs and LBPs on different occasions, without knowing the diagnosis. The representative CSs and LBPs were compared for cellularity, background blood and cell debris, cell architecture, informative background (such as colloid, mucus, stromal fragments), presence of cells in monolayer, and nuclear/cytoplasmic details by a semiquantitative scoring system (Table 1). Statistical analysis was made by the Wilcoxon signed rank test on the SPSS program (Chicago, Illinois, USA). Every cytological diagnosis was recorded and tabulated, and each diagnosis, with the corresponding cytologic material, was reviewed independently by each author. The study protocol was approved by the Ethics Table 2 Sites and interpretation by FNAC N. Mygdakos et al. Committee of both hospitals, and informed consent was obtained from all patients participating in the study. Table 1 Scoring system Cytologic Score features 0 1 2 3 Cellularity Zero Scanty Adequate Abundant Background Good Zero Occasional blood-debris amount Abundant Informative background Absent Present Monolayer Absent Occasional Good amount Cell Non- Moderately Well architecture recognized recognized recognized Nuclear details Poor Fair Good Excellent Cytoplasmic details Poor Fair Good Excellent Results The different sites of FNAC and diagnoses are shown in Table 2. The sites were lymph nodes (38), breast (28), soft-tissue sites (nine), salivary glands (six), and thyroid gland (15). Comparison of cytologic features between LBP and CS is seen in Table 3 (Wilcoxon signed rank test). Breast (28) Lymph nodes (38) Salivary glands (6) Thyroid (15) Soft tissues (9) Benign 2 4 Mastitis 1 Fibroadenoma 10 Fibrocystic disease 5 Ductal carcinoma 10 Granulomatous lymphadenitis 12 Reactive hyperplasia 20 Metastatic carcinoma 4 Lymphoma 2 Sialadenitis/mucocele 1 Mixed tumor 3 Warthin tumor 1 Carcinoma 1 Colloid goiter 9 Thyroiditis 4 Follicular hyperplasia 1 Papillary carcinoma 1 Inflammation 2 Endometriosis 1 Malignancy 2 Table 3 Wilcoxon signed rank test statistics LBP vs. CS Statistics (n = 96) Cellularity Blood Background Monolayers Architecture Cytoplasm Nucleus Z -1.352 (a) -6.553 (b) -1.997 (b) -6.111 (b) -0.299 (a) -3.197 (a) -3.197 (a) Asymptotic, signed test (2 tailed) p = 0.131 p<0.001 p = 0.057 p<0.001 p = 0.865 p<0.001 p<0.001 (a) Based on negative ranks; (b) Based on positive ranks; p<0.05 is statistically significant; p<0.001 is statistically highly significant. LBPs contained adequate diagnostic cells in all cases and were greatly superior to CSs regarding the presence of monolayers, absence of blood and debris, and recognition of nuclear and cytoplasmic details (p<0.001). LBPs showed cellularity, preserved cell architecture, and informative background as good as CSs, expressed by no statistically significant differences (p = 0.131, 0.865, and 0.057 respectively). It was also easier and faster to screen and interpret LBPs because the cells were in small areas with clear background and

Liquid Based Preparation (LBP) cytology versus Conventional Cytology (CS) in FNA samples from breast, thyroid excellent cell preservation. All 10 cases of breast fibroadenoma (Figures 1 and 2) were diagnosed on LBPs. Diagnostic features including ductal cell aggregates, and bipolar cells were present on LBPs, however stromal component was absent. It was apparently easy to diagnose ductal carcinoma of the breast (Figures 3 and 4) on LBPs due to rich cellularity, clear background and detailed nuclear features. However, large clusters of malignant cells were broken into smaller fragments. In the 12 cases of granulomatous lymphadenitis interpreted on CSs, epithelioid cells were noted in 10 LBPs. Although the background was clear in LBPs, occasional red blood cells and cell debris were shown in cases of necrotizing lymphadenitis. In one case of Hodgkin s disease, Reed Sternberg cells and Hodgkin cells were much easier recognized in monolayers. In this case immunocytochemistry was performed. Tumor cells were EMA+, CD30+, and CD15+. In the second case of lymphoma, lymphoid cells on LBPs were single or clustered in small aggregates mimicking oat cell carcinoma. Again, immunocytochemistry was helpful: malignant cells were LCA+ and Pan-B marker+. In the cases of salivary gland lesions, we observed the following: In the cases of mixed tumors epithelial and mesenchymal component were present only in cellular LBPs, while present in all CSs. Myxoid 247 background was poor and present in droplets. Additionally epithelial cells were clustered in small aggregates without any specific formation. Therefore, diagnosis was easier on CSs based on myxoid component. In the case of Warthin tumor, we observed small clusters of oncocytic cells on LBPs, while lymphoid cells were small in number but obviously favored oncocytic cells. Carcinoma was difficult to diagnose on LBPs due to the lack of necrotic debris and fragmentation of epithelial clusters. Cytoplasmic details were not different from those on CSs, but nuclear smudging with loss of detail was present. Occasional accentuation of nucleoli in some cases was more prominent on LBPs. In the case of thyroid lesions (Figures 5 9), LBPs allowed assessment of overall specimen cellularity but not individual passes of a FNA as in CSs. The overall amount of colloid was significantly diminished. The colloid was dense, fragmented or in droplets. Epithelial cells were crowded and tight clustered with nuclear overlapping in comparison with CSs where cells were present in flat sheets or in a honeycomb arrangement. Loss of cellular preservation was especially apparent in the large aggregates in LBPs. The peripheral edge of the preparation was commonly blurred and poorly stained, an artifact associated with the LBP technique. Figure 1 Fibroadenoma of the breast. FNAC. Conventional smear. Papanicolaou stain, 100. Figure 2 Fibroadenoma of the breast. FNAC. Liquid based preparation. Papanicolaou stain, 100. Figure 3 Breast carcinoma. FNAC. Conventional smear. Papanicolaou stain, 100. Figure 4 Breast carcinoma. FNAC. Liquid based preparation. Papanicolaou stain, 100.

248 N. Mygdakos et al. Figure 5 Thyroid nodule. FNAC. Conventional smear. Papanicolaou stain, 100. Figure 6 Thyroid nodule. FNAC. Liquid based preparation. Papanicolaou stain, 100. Figure 7 Thyroid follicular neoplasm. FNAC. Liquid based preparation. Papanicolaou stain, 100. Figure 8 Thyroid papillary carcinoma. FNAC. Conventional smear. May-Grünwald Giemsa stain, 100. Discussion Figure 9 Thyroid papillary carcinoma. FNAC. Liquid based preparation. May-Grünwald Giemsa stain, 100. LBPs appeared to have more disruption of the cytoplasm, an increased number of naked nuclei and a prominence of nucleoli. Nuclear grooves and pseudoinclusions were less apparent in papillary carcinoma. In soft tissue lesions diminished/distorted extracellular and stromal elements, cellular shrinkage, and tissue fragmentation were observed on LBPs. Although, LBP is generally favored over CS for evaluation of gynecological cytology specimens, studies comparing diagnostic accuracy and morphology of these preparations in the evaluation of non-gynecological cytology specimens have drawn variable conclusions. Initial studies found that the diagnosis of breast carcinoma and fibroadenoma were more often and more easily made with CS compared with LBP In the present study bipolar cells were present and despite the fact that stroma and an epithelial stromal relationship was lacking in LBP, a specific diagnosis (e.g. fibroadenoma) was easily made. More recent studies suggest that LBP and CS have comparable diagnostic accuracy in the evaluation of breast FNABs. Dey P et al. were able to diagnose fibroadenomas and infiltrating ductal carcinomas of the breast [2]. In their study, LBPs had adequate cellularity, preserved cell architecture and informative background, such as stromal fragments. All diagnostic elements of fibroadenoma like stromal fragments and staghorn clusters of ductal cells were observed in LBPs [2]. Also, ductal carcinoma of the breast contained abundant cells in LBPs [2]. Analogous to the present study, cell morphology was preserved, the background was clear,

Liquid Based Preparation (LBP) cytology versus Conventional Cytology (CS) in FNA samples from breast, thyroid 249 and nuclear chromatin and outline were excellent in LBP material of ductal carcinoma [2]. When Bédard YC et al. analyzed 7464 breast FNABs over a 4-year period, comparing CS with LBP, there was no significant difference in diagnostic accuracy [5]. In a large study of respiratory cytology specimens, Rana DN et al. [6] found no difference in diagnostic accuracy between LBP and CS; however, they did favor LBP over CS because of a cleaner background, more even cell distribution, and better cell preservation in the former. These results are similar to the present study, where less obscuring blood, no crush or air-drying artifacts and better nuclear detail were observed with LBP. Recently, de Luna R et al. compared LBP with CS in pancreatic FNABs; they found the diagnostic accuracy of TP to be inferior to that of CS [7]. Although they concluded that this finding was partly due to the use of a split-sample technique resulting in decreased cellularity with LBP, there were also important morphological differences, such as the lack of mucin with LBP that impeded the diagnosis of mucinous tumors. Finally, in a recent study of thyroid FNAB by Cochand-Priollet B et al., the diagnostic accuracy of CS was better than that of LBP [8]. Whereas oncocytic tumors and lymphocytic thyroiditis presented diagnostic problems in their study, the lack of background colloid with LBP was also a confounding factor [9, 10]. There are important morphological differences between LBP and CS in the evaluation of salivary gland FNABs, especially related to the quantity and appearance of stroma [11]. The diagnostic yield of CS appears to be greater than that of LBP in the diagnosis of pleomorphic adenoma, which is the most commonly rendered diagnosis in salivary gland FNABs. The alteration of extracellular material that is essential in formulating a specific diagnosis in some soft tissue lesions (e.g. endometriosis) is a potential diagnostic problem. This extracellular material is diminished and altered in quality in LBPs. It may also acquire a filamentous quality that makes it indistinguishable from fibrin. In our case of endometriosis, CSs showed a biphasic component of stromal fragments and epithelial sheets; despite stromal fragments were not seen in LBPs making the diagnosis more difficult. Conclusively, before implementing LBP methodology to FNA material, the cytology lab should obtain experience with and verify the features produced on LBPs when compared to CSs on a wide variety of lesions. Lee KR at al. outlined several differences between the two preparation techniques and emphasized the importance of experience with LBP for a correct interpretation [12]. We believe that LBP should not be the primary method for FNA, unless circumstances are absolutely prohibitive. LBP technology offers promise in other areas, and more will be found with time. Schmitt FC et al., from the University of Porto, in Portugal, have reported on the use of LBP processing for the study of breast cancer cell-lines; cells can be taken from tissue culture and prepared in an easy and uniform manner for research purposes [13]. Other investigators have used the same method to prepare controls for immunoperoxidase studies and in situ hybridization [14 16]. One of the current problems with using cytological materials for ancillary studies is that the controls employed by the immunologic or genetics labs are frequently paraffinbased. These controls are fundamentally unlike the cytological samples and may not be trustworthy. In the clinical lab, it is a dictum that the controls should be as much like the patient specimen as possible. Using histological controls for a cytological sample runs against this basic premise [17]. Conclusions While Liquid-Based Cytology (LBC) can be a valuable and relatively simple technique, careful interpretation of the slides must be applied before a secure diagnosis is reached, especially if LBC is the first and only method applied, or when the cytopathologist performing the FNA has no adequate experience with LBC. Other aspects to be considered are the amount of material yield by FNA, and the correct representation of epithelial cells versus stroma in the sample collected. ThinPrep is acceptable and safe for the diagnosis of breast, thyroid, lymph node, salivary glands, and soft tissue lesions when used in combination with conventional method. References [1] HAYAMA F. H., MOTTA A. C. F., SILVA ADE P. G., MIGLIARI D. A., Liquid-based preparations versus conventional cytology: specimen adequacy and diagnostic assessment in oral lesions, Med Oral Patol Oral Cir Bucal, 2005, 10(2):115 122. [2] DEY P., LUTHRA U. K., GEORGE J., ZUHAIRY F., GEORGE S. S., HAJI B. I., Comparison of ThinPrep and conventional preparations on fine needle aspiration cytology material, Acta Cytol, 2000, 44(1):46 50. [3] MICHAEL C. W., HUNTER B., Interpretation of fine-needle aspirates processed by the ThinPrep technique: cytologic artifacts and diagnostic pitfalls, Diagn Cytopathol, 2000, 23(1):6 13. [4] NICOL T. L., KELLY D., REYNOLDS L., ROSENTHAL D. L., Comparison of TriPath thin-layer technology with conventional methods on nongynecologic specimens, Acta Cytol, 2000, 44(4):567 575. [5] BÉDARD Y. C., POLLETT A. F., Breast fine-needle aspiration. A comparison of ThinPrep and conventional smears, Am J Clin Pathol, 1999, 111(4):523 527. [6] RANA D. N., O DONNELL M., MALKIN A., GRIFFIN M., A comparative study: conventional preparation and ThinPrep 2000 in respiratory cytology, Cytopathology, 2001, 12(6):390 398. [7] DE LUNA R., ELOUBEIDI M. A., SHEFFIELD M. V., ELTOUM I., JHALA N., JHALA D., CHEN V. K., CHHIENG D. C., Comparison of ThinPrep and conventional preparations in pancreatic fine-needle aspiration biopsy, Diagn Cytopathol, 2004, 30(2):71 76. [8] COCHAND-PRIOLLET B., PRAT J. J., POLIVKA M., THIENPONT L., DAHAN H., WASSEF M., GUILLAUSSEAU P. J., Thyroid fine needle aspiration: the morphological features on ThinPrep slide preparations. Eighty cases with histological control, Cytopathology, 2003, 14(6):343 349. [9] SALEH H. A., HAMMOUD J., ZAKARIA R., KHAN A. Z., Comparison of Thin-Prep and cell block preparation for the evaluation of thyroid epithelial lesions on fine needle aspiration biopsy, Cytojournal, 2008, 5:3.

250 [10] STAMATAKI M., ANNINOS D., BROUNTZOS E., GEORGOULAKIS J., PANAYIOTIDES J., CHRISTONI Z., PEROS G., KARAKITSOS P., The role of liquid-based cytology in the investigation of thyroid lesions, Cytopathology, 2008, 19(1):11 18. [11] PARFITT J. R., MCLACHLIN C. M., WEIR M. M., Comparison of ThinPrep and conventional smears in salivary gland fineneedle aspiration biopsies, Cancer, 2007, 111(2):123 129. [12] LEE K. R., PAPILLO J. L., ST. JOHN T., EYERER G. J. A., Evaluation of the ThinPrep processor for fine needle aspiration specimens, Acta Cytol, 1996, 40(5):895 899. [13] SCHMITT F. C., DIAS A. J., WAGNER R. A., Use of ThinPrep processing for studying breast cancer cell lines, Diagn Cytopathol, 2000, 23(1):27 28. [14] KONOFAOS P., KONTZOGLOU K., GEORGOULAKIS J., MEGALOPOULOU T., ZOUMPOULI C., CHRISTONI Z., PAPADOPOULOS O., KOURAKLIS G., KARAKITSOS P., The role of ThinPrep cytology in the evaluation of estrogen and progesterone receptor content of breast tumors, Surg Oncol, 2006, 15(4):257 266. N. Mygdakos et al. [15] VOCATURO A., NOVELLI F., BENEVOLO M., PIPERNO G., MARANDINO F., CIANCIULLI A. M., MEROLA R., PERRONE DONNORSO R., SPERDUTI I., BUGLIONI S., MOTTOLESE M., Chromogenic in situ hybridization to detect HER-2/neu gene amplification in histological and ThinPrep -processed breast cancer fine-needle aspirates: a sensitive and practical method in the trastuzumab era, The Oncologist, 2006, 11(8):878 886. [16] BÉDARD Y. C., POLLETT A. F., LEUNG S. W., O'MALLEY F. P., Assessment of thin-layer breast aspirates for immunocytochemical evaluation of HER2 status, Acta Cytol, 2003, 47(6):979 984. [17] KURTYCZ D. F., HOERL H. D., Thin-Layer technology: tempered enthusiasm, Diagn Cytopathol, 2000, 23(1):1 5. Corresponding author Nikolaos Mygdakos, MD, Department of Cytopathology, Venizeleion Hospital of Heraklion, Knossou Avenue, Heraklion, Crete, Greece; Phone +302 8103 68120, Fax +302 8210 22217, +302 8102 14447, e-mail: Nikolas@doctors.net.uk Received: July 3 rd, 2008 Accepted: May 5 th, 2009