Performance Characteristics of Adenoid Cystic Carcinoma of the Salivary Glands in Fine-Needle Aspirates

CAP Laboratory Improvement Programs Performance Characteristics of Adenoid Cystic Carcinoma of the Salivary Glands in Fine-Needle Aspirates Results From the College of American Pathologists Nongynecologic Cytology Program Z. Laura Tabatabai, MD; Manon Auger, MD; Daniel F. I. Kurtycz, MD; Alice Laser, MD; Rhona J. Souers, MS; Rodolfo Laucirica, MD; Guliz A. Barkan, MD; Barbara A. Crothers, DO; Walid E. Khalbuss, MD, PhD, FIAC Context. Although the cytomorphology of adenoid cystic carcinoma (ACC) has been well described, the accuracy of this diagnosis in fine-needle aspirates (FNAs) of the salivary glands has not been extensively evaluated. Objective. To assess participants responses in the College of American Pathologists (CAP) Nongynecologic Cytology (NGC) Program to determine the accuracy and false-negative rate of ACC cases in salivary gland FNAs. Design. A retrospective review of the CAP NGC Program s cumulative data from 2000 2010 was performed for the general and the specific reference diagnosis categories for ACC in salivary gland FNAs according to preparation and participant types. Results. Of 5156 responses, the overall concordance rates for both the general category of malignancy and the specific category of ACC were 63.6% (3279 of 5156) and 38.6% (1966 of 5088), respectively, with a false-negative rate of 36.4% (1877 of 5156). The most frequent falsenegative responses were pleomorphic (1080) and monomorphic (526) adenoma (1614 of 5088, 31.5%), while lymphoma was the most frequent malignant misinterpretation. There was a significant statistical difference in concordance to the reference interpretation between the reader types: 39.9% (1006 of 2521) concordance rate for pathologists compared to 33.8% (503 of 1488) for cytotechnologists. However, there was no significant statistical difference for concordance to the general category or reference interpretation, based on preparation type (Papanicolaou versus modified Giemsa stained). Conclusions. In this interlaboratory comparison educational program, accurate identification of ACC has shown to be problematic, with ACC representing an important cause of false-negative responses. The most common diagnostic pitfall is distinguishing this entity from pleomorphic and monomorphic adenoma in the benign category and from lymphoma and adenocarcinoma in the malignant one. (Arch Pathol Lab Med. 2015;139:1525 1530; doi: 10.5858/arpa.2013-0173-CP) Accepted for publication March 19, 2015. A From the Department of Pathology, University of California San denoid cystic carcinoma (ACC) is a relatively rare, slowgrowing, malignant salivary gland tumor with poor Francisco, San Francisco (Dr Tabatabai); the Department of Pathology, McGill University Health Sciences Center, Montreal, long-term prognosis. 1 Cytologic features of ACC have been Quebec, Canada (Dr Auger); the Department of Pathology, Wisconsin State Lab of Hygiene, Madison (Dr Kurtycz); the Department of well described 2 14 ; however, the reproducibility of this Anatomic Pathology Services, North Shore Long Island Jewish diagnosis in fine-needle aspirates (FNAs) of the salivary Hospital System, New Hyde Park, New York (Dr Laser); the gland has not been extensively evaluated. This study was Department of Biostatistics, College of American Pathologists, undertaken to analyze participants responses of ACC in Northfield, Illinois (Ms Souers); the Department of Pathology and Immunology, Baylor College of Medicine, Houston, Texas (Dr salivary gland FNAs in the College of American Pathologists Laucirica); the Department of Pathology, Loyola University, Maywood, Illinois (Dr Barkan); the Department of Pathology and Area also address the cytomorphology and some of the diagnostic (CAP) Nongynecologic Cytology (NGC) Program and to Lab Services, Walter Reed National Military Medical Center, pitfalls of ACC in salivary gland FNAs. Bethesda, Maryland (Dr Crothers); and the Department of Pathology, University of Pittsburgh Medical Center (UPMC)-Shadyside, Pittsburgh, Pennsylvania (Dr Khalbuss). MATERIALS AND METHODS The authors have no relevant financial interest in the products or The members of the CAP Cytopathology Committee (CC) companies described in this article. submit and review cases for the CAP Interlaboratory Comparison AllauthorsaremembersoftheCollegeofAmericanPathologists Cytopathology Committee. Program in Nongynecologic Cytology (NGC). After submission, Reprints: Z. Laura Tabatabai, MD, Department of Pathology, each case is reviewed by at least 3 members of the CC to assess University of California San Francisco, 4150 Clement St, Suite slide quality and whether it is representative of the reference 113B, San Francisco, CA 94121 (e-mail: laura.tabatabai@ucsf.edu). diagnosis. Only those slides that have consensus agreement Arch Pathol Lab Med Vol 139, December 2015 Performance of Adenoid Cystic Carcinoma Tabatabai et al 1525

Table 1. Participant Interpretation Summary (N ¼ 5088) Reference Interpretation No. Percentage Malignant neoplasms 3111 61.1 Adenoid cystic carcinoma 1966 38.6 Malignant lymphoma 306 6 Adenocarcinoma, NOS 274 5.4 Other malignancies (acinic cell carcinoma; undifferentiated carcinoma; small cell undifferentiated carcinoma; mucoepidermoid carcinoma, high grade/low grade; metastatic carcinoma, NOS; sarcoma, NOS/spindle cell malignancy; and squamous cell carcinoma) 565 11.1 Benign neoplasms 1717 33.7 Pleomorphic adenoma 1080 21.2 Monomorphic adenoma 526 10.3 Others (Warthin tumor, myoepithelioma, and oncocytoma) 111 2.2 Benign nonneoplastic lesions/normal/others 260 5.1 Lymphoepithelial lesions/lymphoepithelial cyst; sialoadenitis/granulomas; normal salivary gland; intraparotid lymph node; cyst, NOS Abbreviation: NOS, not otherwise specified. between the submitter and 3 reviewers are used for the educational NGC Program. The CAP NGC Program consists of 4 educational challenges per year, each challenge composed of 5 slides mailed to participants who voluntarily subscribe to the continuing education program. After review of the slides, participants choose a general category (benign/negative, suspicious, positive/malignant, or unsatisfactory) and a specific reference diagnosis. After circulation to the participating laboratories, all slides are evaluated for their performance profile such that the slide can be compared to other slides in the same reference category. Every 6 months, each slide associated with a slide set is evaluated for stain quality and diagnostic performance by a CAP cytotechnologist. Any slide with a discordant rate of 70% or higher after circulation is reviewed by a CC member to determine if it will continue circulation in the program. Any slide that has been identified as technically unacceptable by participants is reviewed by the CC. Such reviews occur twice per year. Slides are retired most frequently owing to breakage or suboptimal stain. This quality control system ensures that the slide challenges remain consistent over time. The cumulative histories of participant responses for salivary gland FNA cytology challenges with the reference diagnosis of ACC circulated from 2000 2010 were obtained through the CAP SCORES computer system. The preparation types included modified Giemsa and Papanicolaou-stained smears. Participant results were analyzed for the general categorization (benign/negative, suspicious, positive/malignant, or unsatisfactory) and for the reference diagnosis of ACC. Concordance to the general category was classified as a positive or suspicious participant response. Concordance to the reference interpretation indicated the participant reference interpretation was ACC. A correct response was a response that was concordant to either the general category (ie, malignancy) or to the exact reference diagnosis of ACC. General and exact diagnosis discordant rates between slide preparation type as well as participant type (pathologist versus cytotechnologist) were also analyzed. The results were analyzed by using a nonlinear mixed model for each of the concordance rates. Three factors preparation type, participant type, and 2-interval program year were included in the models. The interaction terms between these factors were also included. Both models included a repeated-measures component to account for the multiple responses per slide. A significance level of.05 was used for these analyses. All statistical analyses were performed with SAS version 9.2 (SAS Inc, Cary, North Carolina). RESULTS A total of 2543 pathologist responses, 1500 cytotechnologist responses, and 1113 laboratory responses submitted for ACC challenges were available for review. Two hundred sixteen responses were excluded from the reference interpretation analysis owing to invalid response. The cumulative responses were based on 97 case challenge slides. The analysis examined concordance to the general category (malignant neoplasm, benign neoplasm, and benign entities), as well as the specific reference interpretation (subtyping of the neoplasm as ACC). Concordance to the general category was classified as a positive or suspicious participant response. Concordance to the reference interpretation indicated the participant reference interpretation was ACC (Table 1). Furthermore, 2 main factors, preparation type and participant type, were analyzed (Table 2). The overall concordance rate to the general category was 63.6% (3279 of 5156) and the specific reference interpretation concordance rate was 38.6% (1966 of 5088). The most frequent incorrect responses in the benign category were pleomorphic adenoma (21.2%, 1080 of 5088) and monomorphic adenoma (10.3%, 526 of 5088), while the most frequent erroneous responses in the malignant/suspicious category were lymphoma (6.0%, 306 of 5088) and adenocarcinoma (5.4%, 274 of 5088). There was no significant statistical difference for concordance to the general category or specific reference interpretation, based on preparation type. The general category concordance rate for Papanicolaou-stained slides was 62.6% (2457 of 3926) compared to the modified Giemsa stained slide concordance rate of 66.8% (822 of 1230), and the reference interpretation concordance rate for Papanicolaoustained slides was 37.6% (1457 of 3870) versus the modified Giemsa stained slides (41.8%, 509 of 1218). The program year was significantly associated with concordance to the general category with a higher concordance between years 2000 2006 than years 2007 2010 (P ¼.007). Participant type was significantly associated with concordance to the reference interpretation (P ¼.006) with pathologists showing 39.9% (1006 of 2521) concordance rate compared to 33.8% (503 of 1488) for cytotechnologists. COMMENT Adenoid cystic carcinoma is a rare, slow-growing, salivary gland malignant tumor with poor long-term prognosis. 1 This tumor metastasizes hematogenously to lung and bone and only rarely spreads to lymph nodes. It accounts for 3% 1526 Arch Pathol Lab Med Vol 139, December 2015 Performance of Adenoid Cystic Carcinoma Tabatabai et al

Table 2. Concordance to the General and Reference Categories: Preparation Type and Participant Type Factor No. of Concordant Evaluations/ Total No. of Evaluations Concordance Rate, % P Value Concordance to the general category (N ¼ 5156) Preparation type Pap stain 2457/3926 62.6.18 Modified Giemsa stain 822/1230 66.8 Participant type Pathologist 1594/2543 62.7 Cytotechnologist 956/1500 63.7.33 Laboratory 729/1113 65.5 Concordance to the reference interpretation (N ¼ 5088) Preparation type Pap stain 1457/3870 37.6.39 Modified Giemsa stain 509/1218 41.8 Participant type Pathologist 1006/2521 39.9 Cytotechnologist 503/1488 33.8.006 Laboratory 457/1079 42.4 Abbreviation: Pap, Papanicolaou. to 5% of all salivary gland neoplasms and is the second most common salivary gland malignancy after mucoepidermoid carcinoma. 1 Most cases of ACC present with an asymptomatic mass that is usually indistinguishable from benign salivary gland tumors, such as pleomorphic or monomorphic adenomas, on clinical presentation. However, it is critical to differentiate ACC from pleomorphic or monomorphic adenoma on FNA because of the malignant biology of ACC and poor prognosis of the evolving tumor. Indeed, a diagnosis of ACC typically leads to total parotidectomy and wide dissection, often with sacrifice of the facial nerve, whereas the surgical intervention for pleomorphic or monomorphic adenomas is much more conservative. Although the cytologic features of ACC have been well described, 2 14 the accuracy of this diagnosis in FNAs of the salivary gland has not been extensively evaluated. The cytologic features of ACC in FNAs of salivary glands are summarized in Table 3. 2 13 The presence of basaloid cells (Figure 1) and homogenous matrix material (Figure 2, A and B) constitute the cornerstones for recognizing ACC in FNAs. Upon aspiration, the FNA often yields glistening, gelatinous material similar to that of a pleomorphic adenoma. Samples are usually moderately to highly cellular. Low-power examination of smears exhibits a monotonous population of small, basaloid cells with very scant cytoplasm. The nuclei of ACC are round to angulated, small to intermediate, uniform in appearance, and usually without significant atypia; however, hyperchromasia is common. Nucleoli are usually not observed or are inconspicuous. Cells typically exhibit minimal cytoplasm and at most demonstrate a small amount of basophilic cytoplasm. Sometimes, there is a pattern of naked nuclei. Mitoses or foci of necrosis are rare. The cells of ACC are typically arranged in tightly cohesive clusters and may form tubules, cribriform arrays, or rosettes. Cells may show a single, isolated arrangement as naked nuclei, or may more characteristically cluster around an amorphous stroma with a sharp epithelial-stromal transition. The stroma is characteristically intensely metachromatic on modified Giemsa stained slides; however, because the stroma appears as pale and semitranslucent on Table 3. Adenoid Cystic Carcinoma Versus Pleomorphic/Monomorphic Adenoma Adenoid Cystic Carcinoma Pleomorphic/Monomorphic Adenoma Matrix Metachromatic (most cases) Metachromatic Homogenous/glassy; with rare or no fibrillary clusters Myxofibrillary or fibrillary Well-defined hyaline spheres ( gum balls ) and/or cylindrical/fingerlike tubules Ill-defined with embedded cells, large, irregular or in tubules Content: acellular or with naked-like nuclei along Content: rare spindle cells or capillaries the border Cells Monomorphic Heterogeneous Basaloid cells Plasmacytoid and spindle cells Little cytoplasm or naked nuclei Moderate cytoplasm/plasmacytoid Angulated nuclear contour Smooth nuclear contour Round nuclei Round, oval, or spindle nuclei Fine to coarse chromatin Fine and bland chromatin Naked nuclei are common Naked nuclei are rare No or inconspicuous nucleoli No nucleoli Cell-stroma interface Stroma and the cells are separated by a sharp smooth border or cells at external border Collagenous stroma interdigitates with adjacent cells Cluster shape Small, more cohesive Large, loosely discohesive Smooth round to angulated border Irregular, fuzzy border Clear spaces within clusters (sieves) No spaces within clusters Background Necrosis may be seen Clean Naked nuclei Single plasmacytoid cells Arch Pathol Lab Med Vol 139, December 2015 Performance of Adenoid Cystic Carcinoma Tabatabai et al 1527

Figure 1. Fine-needle aspirate of adenoid cystic carcinoma showing basaloid cells with little or no cytoplasm, mild nuclear enlargement, nuclear overlapping, and mildly coarse chromatin (modified Giemsa, original magnification 3300). Papanicolaou-stained smears, it can easily be overlooked. This stromal material typically has sharp, rounded outlines and presents as hyaline globules or as cylindrical, branching tubules, giving rise to a fingerlike morphology. Despite the fact that the cytologic features of ACC of the salivary glands in FNAs are well known and have been taught for decades, our study results show that distinguishing ACC from other benign and malignant salivary gland tumors constitutes a difficult diagnostic challenge. Only 63.6% (3279 of 5156) of responses were correctly categorized as malignant, leading to a false-negative rate of 36.4% (1877 of 5156), and only 38.6% (1966 of 5088) of the responses correctly subtyped the malignancy as ACC. Falsenegative diagnoses and difficulties in diagnosis of ACC have been reported in the literature. 2,3,5,6,10 16 However, the results of our study confirm that the distinction between neoplastic and nonneoplastic salivary gland lesions in FNA material is reliable, since approximately 95% of the respondents chose the neoplastic (benign or malignant) category. This is somewhat reassuring, since a diagnosis of salivary gland neoplasm is often sufficient to prompt complete surgical resection guided by preoperative imaging studies and intraoperative frozen section evaluation. Additionally, a diagnosis of salivary gland neoplasm with basaloid features is often sufficient to lead to surgical excision with wide margins, particularly when there is clinical evidence of symptoms of facial nerve pain and tenderness owing to ACC s tendency toward perineural invasion. Previous studies on salivary gland FNAs have reported sensitivity of 70% to 98% and specificity of 86% to 100% in distinguishing benign from malignant neoplasms. 9,10 Our data show a lower sensitivity (63.6%). This may be because our study is based on a single-slide interpretation. This setting probably does not truly mimic the actual practice of cytology where more clinical and radiologic information may be available and where interpretation is usually made with several slides, including various and complementary preparation types such as Papanicolaou- and modified Giemsa stained slides, and possibly cell block and/or ancillary testing. Additionally, while our data differentiate pathologists from cytotechnologists, they do not subdivide pathologists into board-certified cytopathologists and those without that specialty certification, nor do the data differentiate the level of experience and training in cytology or practice setting among the program participants. In this study of FNAs of ACC, the most common falsenegative diagnoses (31.5% of the responses, 1606 of 5088) were pleomorphic adenoma and monomorphic adenoma. Conversely, only 1.3% (360 of 27 545) of pleomorphic adenoma slide reviews, and 5.4% (14 of 257) of monomorphic adenoma slide reviews were diagnosed as adenoid cystic carcinoma. Review of the literature and review of slides of the problematic cases in our study provide clues on the source of the pitfalls in the diagnosis of ACC in salivary gland FNAs and how to distinguish it from other entities. The most important distinguishing feature of ACC on FNA from a pleomorphic adenoma is the stroma. On modified Giemsa stained slides, the stroma of ACC appears as a metachromatic basement membrane like material that is typically amorphous and uniform. In contrast, the stroma of pleomorphic adenomas is characteristically fibrillar (Figure 2, C and D). A fibrillar type of stroma is rarely seen in ACC. In fact, stromal fragments in ACC usually do not show any internal structure and can be described as uniform and hyaline (Figure 3A). In truth, one can occasionally find the ill-defined chondromyxoid fibrillar-like material characteristically seen in pleomorphic adenoma in ACC (Figure 3A, inset). However, when present in ACC, it is lesser in amount than the nonfibrillar homogenous material. These rare fibrillary stromal fragments may be caused by a desmoplastic reaction to the infiltrating tumor and mimic the stroma of pleomorphic adenoma. 15 17 It should also be noted that the presence of the homogenous glassy (hyaline) material resembling pink gum balls is characteristic but not pathognomonic of ACC. 7 9,11 Such extracellular material has also been described focally in other salivary gland neoplasms including pleomorphic adenoma, basal cell adenoma, polymorphous low-grade adenocarcinoma, and epithelial-myoepithelial carcinoma. Stromal features are more difficult to appreciate with a Papanicolaou stain, since stroma stains lightly and appears pale green and translucent. Of diagnostic importance is the presence of sharp epithelial-stromal transitions in ACC compared to the gradual, sunburst epithelial-stromal transition seen in pleomorphic adenoma. The stromal fragments in ACC are often present as globules or tubules that are variable in size with many associated basaloid cells that are intimately arranged along the periphery (Figures 3, A and B; Figure 4). Alternatively, these stromal globules can be seen isolated and dispersed throughout the smears. In cases where acinar and cribriform-like cellular arrangements are seen, the stroma can be found centrally in lumenlike foci. Finally, the absence of stroma can make the diagnosis of ACC extremely challenging, as seen in some of the cases in our study (Figure 5, A and B). When stroma is absent or minimally present, the solid variant of ACC can be considered. 16 20 Other features to look for in support of the diagnosis of ACC include enlargement of cells; prevalence of single cells with overlapping, variably sized and shaped nuclei; identifiable nucleoli; and increased mitotic activity and necrosis. The solid variant of ACC is therefore much more difficult to diagnose on FNA than the classic type, and it is usually suggested as part of a differential diagnosis, typically in the setting of basaloid neoplasms. 16 20 1528 Arch Pathol Lab Med Vol 139, December 2015 Performance of Adenoid Cystic Carcinoma Tabatabai et al

Figure 2. Fine-needle aspirates of adenoid cystic carcinoma (A, B) and pleomorphic adenoma (C, D). The cells in adenoid cystic carcinoma are arranged around homogenous, spherical to tubular matrix fragments (A). B, Clear spaces between clusters create a sievelike effect. C, Pleomorphic adenoma shows larger cells in a background of metachromatic, myxofibrillary to fibrillary matrix with ill-defined borders and cells embedded within the matrix. D, The cells have a moderate amount of cytoplasm with plasmacytoid morphology (modified Giemsa, original magnifications 3300 [A] and 3200 [C]; Papanicolaou, original magnification 3200 [B and D]). While the stroma is the most characteristic feature of ACC, cytologic features can help make the distinction from other entities. As noted above, ACC shows little or no cytoplasm in most cases, 5 whereas cells of pleomorphic and monomorphic adenoma show cytoplasm that tends to be moderate to abundant with plasmacytoid morphology. Another helpful feature is the shape of epithelial clusters. In ACC, the clusters are dense and cohesive, often in trabecular arrangement with smooth round to angulated borders (Figure 6), whereas those in pleomorphic or monomorphic adenoma tend to exhibit larger, looser clusters with ill-defined borders. In addition to the high false-negative rate, the results of our study also highlight a relatively high rate of misclassification of malignancy; in particular, 6% (306 of 5088) of respondents interpreted the samples as lymphoma instead of ACC. This error is most likely due to several factors. Certainly, the scant cytoplasm of the basaloid cells causes a resemblance to lymphocytes. If the participant fails to notice the presence of stroma, then it is not difficult to see why such a misclassification may occur. 3,13 These carefully adjudicated slides did have the features of ACC, but as in day-to-day practice, diagnosis requires careful study, identification of the stromal component, and recognition of the absence of lymphoglandular bodies. The performance of participants in this CAP NGC Program should reflect performance in the general practice of cytology. Our observation of a failure to make a malignant diagnosis 36.4% (1877 of 5156) of the time in face of ACC is problematic. If cytology practitioners cannot make the call in general practice, then clinicians will have to use other modalities to make the diagnosis in order to institute appropriate therapy. If the methodology is not reliable, clinicians will not use it. However, the fact that nearly 95% of the participants recognized the lesion as neoplastic is reassuring, since patients will ultimately receive the appropriate surgical management, albeit sometimes in more than 1 step. In summary, our study illustrates the diagnostic difficulties associated with ACC in salivary gland FNA and highlights several diagnostic pitfalls that have been already published in the literature. Cytologists should be aware of the Arch Pathol Lab Med Vol 139, December 2015 Performance of Adenoid Cystic Carcinoma Tabatabai et al 1529

Figure 3. Matrix of adenoid cystic carcinoma in (A) tubule/fingerlike morphology and (B) round spheres or gum balls. Most matrix fragments are homogenous with only very rare clusters showing fibrillary quality (A inset) mimicking pleomorphic adenoma (modified Giemsa, original magnification 3300 [A, B, and inset A]). Figure 4. Fine-needle aspirate of adenoid cystic carcinoma showing cells with cytologic atypia. The cells are basaloid with little cytoplasm, and show mild nuclear enlargement, mild coarse chromatin, and inconspicuous nucleoli. The matrix fragments are homogenous and well defined. The cells are mostly aligned along the borders of the matrix material (Papanicolaou, original magnification 3400). Figure 5. Fine-needle aspirate of adenoid cystic carcinoma, solid type. A, There is almost no visible matrix component. B, The cells are basaloid with cytologic atypia and almost no visible cytoplasm, giving nuclei a naked appearance. There is mild nuclear enlargement, mildly coarse chromatin, and mild pleomorphism (modified Giemsa, original magnifications 3100 [A] and 3400 [B]). Figure 6. Fine-needle aspirate of adenoid cystic carcinoma. A cell cluster with a smooth, rounded to angulated border is seen (Papanicolaou, original magnification 3400). cytomorphologic characteristics of ACC and perform careful evaluation of the background, the matrix quality including shape, border, content, and cell-stromal interface, as well as the cell morphology and architectural characteristics, in order to minimize the possibility of diagnostic errors. However, in particularly difficult cases of ACC, where basaloid features are evident and predominate but the stromal component is scant to absent on FNA material, a more practical approach may be to diagnose such lesions as salivary gland neoplasm with basaloid features, which, when concordant with the clinical history, imaging findings, and clinical symptoms consistent with perineural involvement, should lead to appropriate surgical management of the patient. References 1. Bradley PJ. Adenoid cystic carcinoma of the head and neck: a review. Curr Opin Otolaryngol Head Neck Surg. 2004;12(2):127 132. 2. Klijanienko J, Vielh P. Fine-needle sampling of salivary gland lesions, III: cytologic and histologic correlation of 75 cases of adenoid cystic carcinoma: review and experience at the Institute Curie with emphasis on cytologic pitfalls. Diagn Cytopathol. 1997;17(1):36 41. 3. Nagel H, Hotze HJ, Laskawi R, Chilla R, Droese M. Cytologic diagnosis of adenoid cystic carcinoma of salivary glands. Diagn Cytopathol. 1999;20(6):358 366. 4. Oliai BR, Sheth S, Burroughs FH, Ali SZ. Parapharyngeal space tumors: a cytopathological study of 24 cases on fine-needle aspiration. Diagn Cytopathol. 2005;32(1):11 15. 5. Lee S, Cho K, Jang J, Ham E. Differential diagnosis of adenoid cystic carcinoma from pleomorphic adenoma of the salivary gland on fine needle aspiration cytology. Acta Cytol. 1996;40(6):1246 1252. 6. Stanley M, Horwitz C, Henry M, Burton L, Lowhagen T. Basal-cell adenoma of the salivary gland: a benign adenoma that cytologically mimics adenoid cystic carcinoma. Diagn Cytopathol. 1988;4(4):342 346. 7. Hara H, Oyama T, Suda K. New criteria for cytologic diagnosis of adenoid cystic carcinoma. Acta Cytol. 2005;49(1):43 50. 8. Yang GC, Waisman J. Distinguishing adenoid cystic carcinoma from cylindromatous adenomas in salivary fine-needle aspirates: the cytologic clues and their ultrastructural basis. Diagn Cytopathol. 2006;34(4):284 288. 9. Stewart CJ, MacKenzie K, McGarry GW, Mowat A. Fine-needle aspiration cytology of salivary gland: a review of 341 cases. Diagn Cytopathol. 2000;22(3): 139 146. 10. Cohen EG, Patel SG, Lin, O et al. Fine-needle aspiration biopsy of salivary gland lesions in a selected patient population. Arch Otolaryngol Head Neck Surg. 2004;130(6):773 778. 11. Mukunyadzi P. Review of fine-needle aspiration cytology of salivary gland neoplasms, with emphasis on differential diagnosis. Am J Clin Pathol. 2002; 118(suppl):S100 S115. 12. Cajulis RS, Gokaslan ST, Yu GH, Frias-Hidvegi D. Fine needle aspiration biopsy of the salivary glands: a five-year experience with emphasis on diagnostic pitfalls. Acta Cytol. 1997;41(5):1412 1420. 13. Hughes JH, Yolk EE, Wilbur DC. Pitfalls in salivary gland fine needle aspiration cytology: lessons from the College of American Pathologists Interlaboratory Comparison Program in Nongynecologic Cytology. Arch Pathol Lab Med. 2005;129(1):26 31. 14. Lü BJ, Zhu J, Gao L, Xie L, Xu JY, Lai MD. Diagnostic accuracy and pitfalls in fine needle aspiration cytology of salivary glands: a study of 113 cases [in Chinese]. Zhonghua Bing Li Xue Za Zhi. 2005;34(11):706 710. 15. Desai S, Krishnamurthy S. Stromal fragments in invasive carcinoma: source of diagnostic difficulty in aspiration cytology. Acta Cytol. 1997;41(6):1747 1750. 16. Stanley MW, Horwitz CA, Rollins SD, et al. Basal cell (monomorphic) and minimally pleomorphic adenomas of the salivary glands: distinction from the solid (anaplastic) type of adenoid cystic carcinoma in fine-needle aspiration. Am J Clin Pathol. 1996;106(1):35 41. 17. Viguer JM, Vicandi B, Jiménez-Heffernan JA, López-Ferrer P, Limeres MA. Fine needle aspiration cytology of pleomorphic adenoma; an analysis of 212 cases. Acta Cytol. 1997;41(3):786 794. 18. Chhieng DC, Paulino AF. Basaloid tumors of the salivary glands [review]. Ann Diagn Pathol. 2002;6(6):364 372. 19. Chen L, Ray N, He H, Hoschar A. Cytopathologic analysis of stroma-poor salivary gland epithelial/myoepithelial neoplasms on fine needle aspiration. Acta Cytol. 2012;56(1):25 33. 20. Spiro RH, Huvos AG, Strong EW. Adenoid cystic carcinoma of salivary origin: a clinicopathologic study of 242 cases. Am J Surg. 1974;128(4):512 520. 1530 Arch Pathol Lab Med Vol 139, December 2015 Performance of Adenoid Cystic Carcinoma Tabatabai et al