Introduction: The Milan System for reporting salivary gland cytopathology helps standardize reporting systems across institutions, improve communication between clinicians and pathologists and guide the clinical management of patients. Aims: This study was undertaken to evaluate the utility of the Milan system classification in cytology reporting. Settings and Design: The present study is a retrospective study conducted over a period of five years in tertiary care centre. Methods and Materials: All the cases of salivary gland aspirates were reviewed and reclassified into six diagnostic categories according to the Milan system of reporting salivary gland cytology (MSRSGC). Cytological diagnosis was correlated with the histopathological diagnosis wherever available. Results: A total of 258 cases were classified using the Milan system as non-diagnostic (20.9%), non-neoplastic (26.3%), atypia of undetermined significance (4.7%), neoplasm benign (37.5%), neoplasm of uncertain malignant potential (3.5%), suspicious for malignancy (0.4%), and malignancy (6.6%). Cytohistological discordance was noted among 8/76 cases (10.5%). The sensitivity and specificity of FNAC were 75% and 98.5%, respectively. The risk of malignancy was 14.2% for Category I, 9% for II, 50% for III, zero for IVA and IVB, and 83.3% for category VI. Conclusions: The new classification system helps pathologists to standardize reporting leading to better clinical and surgical management.
Keywords: Milan system, risk of malignancy, salivary gland
How to cite this article:Salivary gland neoplasms are rare, with a global annual incidence of 0.4-13.5 cases per 100,000 population. They account for 3-10% of all head and neck tumors.[1],[2]
Fine needle aspiration cytology (FNAC) has proven to be of significant value in evaluating and diagnosing salivary gland lesions. It helps differentiate between benign and malignant lesions, benign and low-grade from high-grade malignant tumors, and primary from metastatic lesions. Although FNAC has high diagnostic accuracy, the diagnosis is still challenging due to limited cellularity, sampling errors, diversity of tumors, intratumoral heterogeneity, morphological overlap, especially between benign and normal salivary gland, between certain low grade and high-grade carcinomas, lymphoid rich lesions, and cystic change/cystic lesions.[3] The sensitivity and specificity of FNAC is generally high and ranges from 54% to 98% and 88% to 99%, respectively.[4],[6],[7],[8],[9] However, regarding the subtyping of tumors, the accuracy is low and ranges from 62 to 80%.[3]
Cytology is an accepted pre-surgery modality for the evaluation of salivary gland tumors. However, the spectrum of lesions varies, with a lack of uniform consensus on the reporting classifications. Due to the difficulties faced by clinicians and pathologists, an international six-tier classification scheme, “Milan System for Reporting Salivary Gland Cytopathology (MSRSGC)” was introduced in 2015. The aim was to provide 1) an internationally accepted and user-friendly practical classification; 2) an evidence-based system for clinicians; 3) risk of malignancy (ROM) for each diagnostic category.[10],[11]
In the current study, salivary gland aspirates were retrospectively reviewed and classified using the Milan system and correlated with histopathology, wherever available. The ROM for each diagnostic category was also calculated.
Subjects and MethodsA retrospective study was conducted over a period of five years at our tertiary care hospital from January 2013 to December 2017. Aspirations done by free hand technique as well as USG-guided aspiration were included. Written consent was obtained from all the patients before the procedure. All the aspirates were carried out using a 22–24-gauge needle. Smears were routinely fixed, processed, and stained using Hematoxylin and Eosin, Papanicolaou, May Grunwald-Giemsa (H&E, PAP, and MGG) stains.
All the cases were reviewed and classified into six diagnostic categories according to the Milan system of reporting salivary gland cytology (MSRSGC) as non-diagnostic, non-neoplastic, atypia of undetermined significance (AUS), neoplasm into benign or salivary gland neoplasm of uncertain malignant potential (NUMP), suspicious for malignancy, and malignancy. Cytological diagnosis was correlated with the histopathological diagnosis wherever available. Biopsy diagnosis was considered the gold standard. Based on the presence or absence of malignancy in histopathology, cases were categorized into true positive, true negatives, false positives, and false negatives. Sensitivity, specificity, positive predictive value, and negative predictive value for differentiating benign from malignant tumors were calculated. The ROM was also calculated for each diagnostic category. IEC approval was obtained. The date of approval is 13th February 2020.
ResultsA total of 258 cases were retrieved over a period of five years. The age of the patients ranged from 3 to 83 years, with a mean of 45 years. Of the total cases, 151 were males, and 107 were females, with a male-female ratio of 1.4: 1. The most common site of aspiration was major salivary glands in 248 (96.1%) cases, followed by minor salivary glands in 10 (3.9%) cases. Among the major salivary glands, parotid involvement was seen in 71.3% cases, followed by the submandibular gland in 24.4% cases and the sublingual gland in 0.4% cases. The sites of aspiration of minor salivary glands were buccal mucosa, hard palate, and soft palate.
Non-diagnostic FNAs were seen in 54 (20.9%) cases. There was a total of 78 non-neoplastic lesions (30.2%) and 126 (48.8%) neoplastic lesions [Table 1]. The most common non-neoplastic lesion in this study was chronic sialadenitis (12.4%). Among neoplastic lesions, benign neoplasms were diagnosed in 90 (71.4%) cases and malignant neoplasms in 18 (14.3%) cases. In addition, 18 cases were diagnosed as neoplasm alone (14.3%). Pleomorphic adenoma was the most common benign neoplasm (67.5%). The most common neoplasm in the malignant category was mucoepidermoid carcinoma (58.8%).
Histological follow-up was available for 76 cases, of which 8 showed discordance on correlation with cytology (10.5%) [Table 2]. The sensitivity, specificity, positive predictive value, and negative predictive value of FNAC in this study were 75%, 98.5%, 85.7%, and 97.1%, respectively.
Table 2: Comparison of cytological diagnosis with histopathological diagnosisThe cases were categorized according to the Milan system, and the distribution was as follows: Non-diagnostic (category I) 20.9%, Non-neoplastic (category II) 26.3%, AUS (category III) 4.7%, Benign neoplasm (category IVA) 37.5%, NUMP (category IVB) 3.5%, Suspicious for malignancy (category V) 0.4%, and malignancy (category VI) 6.6%. The ROM was calculated for each category [Table 3].
Category I included 54 non-diagnostic cases either due to hemorrhage, lack of cellularity, non-mucinous cystic lesion, and normal salivary gland elements. Histological follow-up was available for 7 (12.5%) cases. Out of these, one was discordant and turned out to be mucoepidermoid carcinoma [Figure 1]a and [Figure 1]b. Overall ROM for category I in this study was 14.2%.
Figure 1: (a) Fine needle aspiration smear showing predominantly inflammatory cells with lack of epithelial component (Category I) (Papanicolaou 200×) (b) Follow-up histological examination showed low grade mucoepidermoid carcinoma (H and E 200×) (c) Aspirate showing benign salivary gland acini and variable mononuclear inflammatory cell infiltrate in the background. (Category II) (H and E 200×) (d) Follow-up histology revealed carcinoma ex pleomorphic adenoma with squamous cell carcinoma as the malignantSixty-eight cases were classified under Category II. These included acute sialadenitis, chronic sialadenitis, Granulomatous inflammation, Benign lymphoepithelial cyst, and reactive hyperplasia (intraparotid lymph node). Eleven (16.4%) cases had follow-up. One case of chronic sialadenitis was reclassified as carcinoma ex pleomorphic adenoma [Figure 1]c and [Figure 1]d. The ROM was 9%.
Category III included 12 cases with limited atypical features indefinite of a neoplasm. This category included nine cases of mucinous cystic lesion with scant or no epithelial cells, one case with a differential of retention cyst versus low-grade mucoepidermoid carcinoma, five cases of paucicellular neoplasm, including one case of paucicellular acinar cells. Follow-up was available for two (16.6%) paucicellular neoplasms with a diagnosis of Warthin tumor and acinic cell carcinoma, respectively [Figure 2]. The ROM was 50%.
Figure 2: (a) Paucicellular fine needle aspiration smear showing acinar like-cluster of cells with bland nuclei and finely vacuolated cytoplasm (Category III) (MGG 200×) (b and c) Follow-up histological examination revealed acinic cell carcinoma comprising of acinar cells admixed with intercalated duct cells and microcsytic spaces (H and E 200 × and 100×)In category IVA, there were 97 cases with histological follow-up available in 48 (50%) cases [Figure 3]a and [Figure 3]b. All were concordant, and hence the ROM was 0%. Category IVB, that is, neoplasm of undetermined malignant potential, included cases that could not be categorized as benign or malignant [Figure 3]c and [Figure 3]d. There were nine cases, including three with a diagnosis of basaloid neoplasm, three with oncocytic neoplasm, one with cellular myoepithelial cells, one case with paucicellular atypical cells, and one case with scant eosinophilic stroma with a differential of pleomorphic adenoma, basal cell adenoma, or adenoid cystic carcinoma. Two (22.2%) of them had histological follow-up and were reported as benign. Overall, ROM was 0%.
Figure 3: (a) Fine needle aspiration smear of pleomorphic adenoma with characteristic fibrillary chondromyxoid stroma (Category IVA) (MGG 200×) (b and c) Histology showing classical pleomorphic adenoma (H and E 200×) (c) Aspirate of an oncocytic neoplasm showing cellular neoplasm with cohesive clusters of cells with central nuclei and oncocytic cytoplasm. (Category IVB) (Papanicolaou 200×) (d) Cluster of basaloid epithelial cells with crowding, hyperchromasia and scant cytoplasm, reported as basaloid neCategory V included one case with paucicellular atypical lymphoid cells highly suspicious of a non-Hodgkin lymphoma. However, follow-up was not available; hence ROM could not be estimated.
Category VI included 17 cases of malignancy diagnosed on cytology. These included mucoepidermoid carcinoma (58.8%), squamous cell carcinoma (11.8%), adenocarcinoma (17.6%), small round cell neoplasm (5.8%), and papillary neoplasm (5.8%). [Figure 4] Histological follow-up was available in six (35.3%) cases. One case of Warthin tumor was misdiagnosed as squamous cell carcinoma in cytology [Figure 5]. The ROM for category VI was reported as 83.3%.
Figure 4: (a) Aspirate showing highly atypical cell cluster with moderate pleomorphism and nuclear hyperchromasia, reported as poorly differentiated carcinoma (Category VI) (H and E 400×) (b) Histology showing poorly differentiated adenocarcinoma (H and E 200×) (c) Aspirate of low grade mucoepidermoid carcinoma showing clusters of mucin secreting cells against a mucoid background (Category VI) (H and E 200×) (d) Follow-up histology of low grade mucoepidermoid carcinoma (H and E 200×)Figure 5: (a) Aspirate reported as squamous cell carcinoma with necrosis (category VI) (Papanicolaou 200×) (b) Follow-up histology revealed Warthin's tumor (H and E 200×). In this case, the dirty background of cyst contents and oncocytic cells in Warthin's tumor was misinterpreted as atypical squamous cells in a background of necrosis. (c) Aspirate showing sheets of monomorphic population of uniform round cells (Category VI) (MGG 200×) (d) Histology reported as small round cell neoplasm (H and E 200×) DiscussionThough FNA has proven its efficacy in salivary gland neoplasms, the diagnosis remains challenging due to various factors. The sensitivity and specificity of salivary gland aspirates reported in the Indian literature vary between 64 to 85% and 89 to 100%.[9],[12],[13],[14],[15],[16],[17],[18],[19],[20],[21],[22],[23] Overall, our present study had a specificity of 98.5% and a sensitivity of 75% which falls in line with the previous studies.
To address the challenges, MSRSGC provides a systematic way of reporting with terminology that provides effective communication between pathologist and clinician, improving patient care, facilitating cytohistological correlation, promoting research, and helps to share data across different laboratories.[4] Risk-based stratification has already been implemented in other organs like the thyroid, pancreas, cervix, and urinary system. MSRSGC also helps in estimating the ROM which helps in guiding the management. After reclassification, the most common category observed in this study was category IVA (37.2%).
Category I includes aspirates with insufficient quantitative/qualitative material, aspirates with benign cellular elements or non-mucinous cyst contents devoid of epithelial cell component. Though a definitive adequacy criterion is not defined for salivary gland lesions, a minimum of 60 lesional cells is suggested by MSRSGC.[24] Normal salivary gland elements in the presence of a radiological lesion are classified under category I. In this study, 54 (20.9%) cases were under Category I, 32% of which had normal salivary gland elements. However, radiological findings were not available in most of the cases. Seven (12.9%) cases had non-mucinous cyst contents. The inadequacy rate reported in the Indian literature varies between 2 and 10% [Table 4].[9],[12],[13],[14],[15],[16],[17],[18],[19],[20],[21],[22],[23] The higher rate of inadequacy in this study was mainly due to lack of cellularity, partly due to the nature of the lesion, and partly due to improper positioning of the needle. Rapid onsite evaluation for adequacy, performing aspiration under radiological guidance, and good clinical radiological correlation might help in reducing the number of non-diagnostic cases.
In category II, on follow-up, two cases of carcinoma ex-pleomorphic adenoma and pleomorphic adenoma with infarction were misdiagnosed as chronic sialadenitis. The possible explanation can be inappropriate positioning of the needle leading to a sampling of adjacent salivary gland tissue or the infarcted tissue. In the resected specimen of carcinoma ex-pleomorphic adenoma, the adjacent salivary gland tissue showed dense inflammation in response to the tumor.
Table 4: Comparison of risk of malignancy of various Indian studies with the present studyThe main objective of applying Category III is to reduce the number of false negatives in the non-neoplastic category and false positives in the neoplastic category. As per MSRSGC, the number of cases placed in AUS should preferably be less than 10%.[10] In this study, the number of cases was 12%. One case of paucicellular acinar cells turned out to be acinic cell carcinoma on resection. Implied ROM was 50% in the current study, which was higher than the ROM (20%) suggested by MSRSGC. This is due to histological follow-up being available in only two cases. As suggested by Jha et al.[14] in addition to histological follow-up, the inclusion of clinical and radiological follow-up in the denominator for calculating risk would prevent overestimation of ROM.
Cystic lesions account for 8% of all salivary gland lesions, with a diagnostic accuracy of 40%. Hence, they constitute a major pitfall in the diagnosis.[25] In this study, there was a total of 24 cystic lesions. As per the MSRSGC, these were subclassified into Categories I, II, or III based on the presence of non-mucinous cystic fluid only or non-mucinous cystic fluid with inflammatory infiltrate/crystals or mucinous cystic fluid with or without epithelial cell contents, respectively. Histological follow-up was available for three cases, of which one case reported as an inflammatory cyst with scant oncocytic cells turned out to be Warthin's tumor. WT is the most common neoplasm to yield a lymphocytic aspirate and debris; a differential should always be considered in case of inflammatory cystic aspirate.[26] Two cases with a final diagnosis of lymphoepithelial cyst on histology showed the presence of anucleate squames and debris in a proteinaceous background on aspirate. Diagnosis of acellular cystic lesions is still challenging due to the pauci-cellularity and overlapping features; implementation of MSRSGC may help reduce the number of patients requiring surgical intervention. Repeat FNAC after aspiration of cystic content is advised to reduce sampling error in such cases.
The most common benign neoplasm (category IVA) reported was pleomorphic adenoma comprising 67.5% of all cases in that category. Rossi et al.[11] and Jha et al.[14] also found a similar frequency, with pleomorphic adenoma being the most common, 70.7% and 86.54%, respectively. Histological follow-up was available for 50% of cases, all of which were concordant; hence the ROM was 0%. Gaikwad et al.[16] and Amita et al.[18] also reported 0% ROM. Most of the Indian studies have also reported the least ROM in this category, indicating high accuracy of diagnosing such lesions by FNAC.
Category IVB (NUMP) includes nine neoplastic cases where the distinction between benign and malignant is difficult to diagnose. NUMP is an uncommon category, and if reported, the patient is always considered for surgical approach.[27] There were two cases with follow-up in category IVB with predominant myoepithelial and basal cell clusters which turned out to be myoepithelioma and basal cell adenoma, respectively.
Category V includes cases lacking all the definitive features of malignancy. In this study, one of the cases had a few atypical cells highly suspicious for a lymphoproliferative disorder; however, follow-up was not available.
In category VI, the most common malignancy reported was mucoepidermoid carcinoma. Cytohistological discordance was seen in one case of Warthin tumor, which was reported as squamous cell carcinoma in cytology. Oncocytic cells with dirty background were misinterpreted as atypical squamous cells with necrosis. This is one of the pitfalls often encountered in cytology. The overall ROM for category VI in our study was 83.3%.
In this study, there were two false negatives and one false positive. The false-negative cases were necrotic debris and inflammatory lesion, which turned out to be mucoepidermoid carcinoma and carcinoma ex pleomorphic adenoma. Based on the literature, low-grade Mucoepidermoid carcinoma is the most common malignancy diagnosed as false negative on cytology.[11] The false-positive case was a misdiagnosis of Warthin tumor as squamous cell carcinoma. The present study had limitations such as the retrospective nature of the study, a relatively higher number of non-diagnostic lesions, and a lesser number of cases with histological follow-up.
ConclusionImplementation of MSRSGC in routine practice helps in uniformity of reporting and helps the pathologists minimize the number of false-positive and false-negative cases. This new classification needs to be followed across the institutions to avoid ambiguity and provide necessary information to the clinicians.
Financial support and sponsorship
Nil.
Conflicts of interest
There are no conflicts of interest.
References
Correspondence Address:
Dr. V Gnanapriya
Department of Pathology, St John's Medical College, Sarjapur Road, Bengaluru - 560 034, Karnataka
India
Source of Support: None, Conflict of Interest: None
CheckDOI: 10.4103/joc.joc_106_21
留言 (0)