Validating the Utility of the Milan System of Reporting Salivary Gland Cytopathology for the Diagnosis of Salivary Gland Fine-Needle Aspirates in a North Indian Tertiary Care Center and Comparing the Risk of Malignancy with its Two Editions

Abstract

Background:

Salivary gland (SG) fine-needle aspiration (FNA) is a well-accepted tool for preoperative management of salivary gland lesions (SGLs). First and second editions of the Milan System for Reporting Salivary Gland Cytopathology (MSRSGC) were published in the years 2018 and 2023, respectively. The MSRSGC aims to improve patient care by providing better communication between clinicians and institutions. It has six ascending diagnostic categories having a definite risk of malignancy (ROM) and corresponding management schema. Our study assessed the incidence and corresponding ROM of each MSRSGC category within a tertiary care northern India referral center and compared each ROM with those laid down in the two editions of the MSRSGC.

Aims and Objectives:

The main aims and objectives are as follows:

1. Validating the utility of the MSRSGC for the diagnosis of SG-FNAs in a North Indian Tertiary Care Center.

2. Comparing the Risk of Malignancy with its Two Editions.

Materials and Methods:

Data on SG-FNAs performed between 01.01.2010 and 31.03.2024 were searched from the hospital information system. The cytopathology and histopathological slides were retrieved, and each case was separately reviewed and retrospectively classified into the I) non-diagnostic (ND), II) non-neoplastic (NN), III) atypia of undetermined significance (AUS), IVA) benign neoplasms (NB), IVB) SG neoplasm of uncertain malignant potential (SUMP), V) suspicious for malignancy (SFM), or VI) malignant categories according to the MSRSGC. Cyto-histopathological correlation analysis was performed, and the ROM for each category was calculated.

Results:

A total of 793 such cases were obtained. Histopathological follow-up was available in 267 cases, which included 2 (0.7%) ND, 29 (10.9%) NN, 5 (1.9%) AUS, 190 (71.2%) NB, 11 (4.1%) SUMP, 10 (3.7%) SFM, and 20 (7.5%) malignant cases. The calculated ROM for these ascending categories were 0%, 27.6%, 60%, 11.8%, 45.5%, 90%, and 100%, respectively.

Conclusion:

MSRSGC is a valuable means to standardize the reporting and preoperatively stratify SGLs as per the ROM for each ascending risk category. This study, to the best of our knowledge, is the largest Indian study to authenticate the utility and reproducibility of the MSRSGC in our tertiary care institute.

INTRODUCTION

Salivary gland (SG) fine-needle aspiration (FNA) is now considered a powerful and competent tool for the diagnosis of salivary gland lesions (SGLs). They form a cost-effective and minimally invasive mode for the diagnosis of SGLs, in addition to their early and timely management. FNA can not only delineate between neoplastic and non-neoplastic pathologies but can also differentiate between benign and malignant, as well as low-and high-grade carcinomas, in most cases. This knowledge is important to decide the type of management (e.g., surgical in neoplastic, non-surgical in non-neoplastic, and a few benign neoplasms, where clinical follow-up and imaging can help) and to decide the extent of surgery in cases requiring surgical intervention.

Two editions of the Milan System for Reporting Salivary Gland Cytopathology (MSRSGC) have been published so far. The first edition was published in February 2018, and the second edition was published recently in July 2023.[1,2] The MSRSGC comprises six diagnostic categories, including non-diagnostic (ND; category-I), benign non-neoplastic (NN; category-II), atypia of undetermined significance (AUS; category-III), benign neoplasm (NB; category-IVA), SG neoplasm of uncertain malignant potential (SUMP; category-IVB), suspicious for malignancy (SFM; category-V), and malignant (M; category-VI). The establishment of the MSRSGC has paved the way to standardized reporting of SG cytology across institutions, improvement of the overall effectiveness of SG-FNAs, better communication between the pathologist and clinicians, stratification of ascending MSRSGC categories as per the risk of malignancy (ROM), and corresponding risk-based management of the cases, all ultimately leading to improved patient care. Thus, after the introduction of the MSRSGC, studies have affirmed promise in its use to classify various SGLs. Moreover, these studies provide information to help the system modify, revise, and improve its predictability and corresponding management.

In this study, we retrospectively categorized the SG-FNAs according to the MSRSGC and assessed the incidence and ROM of the six MSRSGC categories. In addition, we compared our ROM with the older ROM in the 1st edition of MSRSGC and the newer refined ROM in the 2nd edition of MSRSGC.

MATERIALS AND METHODS

This was a retrospective study of FNAs performed from the SGLs, including metastasis and lymphomas of intra-glandular lymph nodes. The hospital information system was searched for all the retrospective cases of SG-FNAs performed at our institute from January 2010 to March 2024. The relevant clinico-radiological details, like age, gender, complaint, site, size, and side of the swelling; follow-up data; and histopathological diagnosis, were recorded.

FNAs were mostly performed with USG guidance with rapid on-site evaluation (ROSE). In each case, both air-dried and alcohol-fixed smears were made with the FNA aspirates. Air-dried smears were stained with May–Grunwald–Giemsa stain, and alcohol-fixed smears were stained with hematoxylin & eosin (H&E) and papanicolaou (Pap) stains. All the smears of each case were reviewed individually by two experienced cytopathologists (who were blinded to the histopathological diagnosis). In the case of any discrepancy, an opinion was taken from the third cytopathologist. Each case was allocated one of the six diagnostic categories of MSRSGC, namely, ND; category-I, NN; category-II, AUS; category-III, benign neoplasm NB; category-IVA, SUMP; category-IVB, SFM; category-V, and M; category-VI, as per the diagnostic criteria given by Faquin et al.[1,2]

Histopathology (HPE) was taken as the gold standard. Immunohistochemistry (IHC) was applied in cases as and when required. Histopathological diagnoses like mucoepidermoid carcinoma (MEC), squamous cell carcinoma (SCC), acinic cell carcinoma (ACC), adenoid cystic carcinoma (AdCC), salivary duct carcinoma (SDC), poorly differentiated carcinoma, non-Hodgkin lymphoma, carcinoma ex pleomorphic adenoma (CaExPA), epithelial–myoepithelial carcinoma, secretory carcinoma (SC), malignant mixed tumor, basal cell adenocarcinoma (BCA), malignant peripheral nerve sheath tumor (MPNST)/other malignant mesenchymal tumors, metastasis from infiltrating ducal carcinoma (breast), and sebaceous carcinoma on surgical follow-up were considered malignant outcomes. For the sake of statistical analysis, any histopathological specimens reported as “no evidence of granulomatous pathology or malignancy” were assumed to be of the non-neoplastic category. In cases where repeat FNAs were performed, the MGRSGC category was given according to the first FNA.

The cases were first divided into non-neoplastic, benign, and malignant categories depending upon their histopathological diagnosis. Cyto-histopathological correlation analysis was performed. The risk of neoplasia (RON) and ROM were calculated for each diagnostic category in cases for which surgical follow-up with histopathological examination was available. The RON is defined as the ratio of the number of histopathologically neoplastic cases (i.e., benign plus malignant) reported in a particular MGRSGC category with respect to the total number of FNAs performed within the same MGRSGC category for which HPE was performed. The ROM similarly is defined as the ratio of the number of malignant cases (on HPE) reported in a particular MGRSGC category with respect to the total number of FNAs performed within the same MGRSGC category for which HPE was performed.

Additionally, the cytological cases were further typed as true positives (diagnosed as malignant or suspicious of malignancy on both cytology and HPE), true negatives (diagnosed with the absence of malignancy on both cytology and HPE), false positives (diagnosed incorrectly as malignant or suspicious of malignancy on cytology), and false negatives (failure to diagnose malignancy on cytology). Subsequently, diagnostic efficacy was determined in terms of sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).

RESULTS

A total of 793 FNA procedures from SGLs were performed between January 2010 and February 2024, that is, 13 years and 2 months. Of these 793 cases, surgical follow-up with histopathological examination was available in 267 cases and was not available in 526 cases. The clinico-pathological profile of all the 793 cases and 267 cases with available surgical follow-up has been summarized in Table 1. The mean age of the 793 cases was 43.6 ± 0.6 years (2–92 years), consisting of 444 (56%) males and 349 (44%) females, with a male:female ratio of 1.27:1. Among the cases with available surgical follow-up, the mean age of the patients was 40.3 + 0.9 years (6–78 years), and the male:female ratio was 1:1.08. The parotid gland was the most common gland for FNA (617 and 228 cases), followed by the submandibular gland (171 and 38 cases) and sublingual gland (one case each) in both types of cases.

Table 1.

The clinicopathological profile of the cases

Parameters and categories		Overall cases (n = 793)	Cases with surgical follow-up (n = 267)
Mean age (in years)		43.6 ± 0.6 (2–92)	40.3 ± 0.9 (6–78)
Gender	Male	444	128
	Female	349	139
	Male: female ratio	1.27:1	1:1.08
Site (gland)	Parotid	617	228
	Submandibular	171	38
	Sublingual	01	01
	Minor salivary gland (palate)	04	00
Histopathological category	Non-neoplastic	–	17
	Benign	–	183
	Malignant	–	67

Among 267 cases in which the surgical follow-up was available, the number of non-neoplastic, benign, and malignant cases as per the histopathological examination was 20 (7.5%), 180 (67.4%), and 67 (25.1%) cases, respectively [Table 2]. Among the cases without available histopathological follow-up, the most common MGRSGC category was non-neoplastic (224 cases; 42.6%), followed by benign (156 cases; 29.7%) and malignant (80 cases; 15.2%) cases.

Table 2.

MSRSGC categories and ROM in cases with surgical follow-up

Diagnostic category	No. of FNAs done	No. of cases without surgical follow-up (%)	No. of cases with available surgical follow-up
			Total cases with surgical follow-up (%)	Non-neoplastic	Benign	Malignant	Total	RON, (%)	ROM, (%)	ROM as per MGRSGC (1st edition)	ROM as per MGRSGC (2nd edition) (%)
I. Non-diagnostic	32	30 (5.7%)	2 (0.7%)	1	1	0	2	1/2 (50%)	0/2 (0%)	25% (0%–67%)	15%
II. Non-neoplastic	253	224 (42.6%)	29 (10.9%)	16	5	8	29	13/29 (44.8%)	8/29 (27.6%)	10% (0%–20%)	11%
III. AUS	27	22 (4.2%)	5 (1.9%)	1	1	3	5	4/5 (80%)	3/5 (60%)	20% (10%–35%)	30%
IVA. Benign neoplasm (NB)	346	156 (29.7%)	190 (71.2%)	0	167	23	190	190/190 (100%)	23/190 (12.1%)	<5% (0%–13%)	<3%
IVB. SUMP	22	11 (2.1%)	11 (4.1%)	1	5	5	11	10/11 (90.9%)	5/11 (45.5%)	35% (0%-100%)	35%
V. Suspicious for malignancy	13	3 (0.5%)	10 (3.7%)	1	1	8	10	8/10 (80%)	9/10 (90%)	60% (0%-100%)	83%
VI. Malignant	100	80 (15.2%)	20 (7.5%)	0	0	20	20	20/20 (100%)	20/20 (100%)	90% (57%-100%)	>98%
	793	526 (100%)	267 (100%)	20	180	67	267

MSRSGC = Milan System for Reporting Salivary Gland Cytopathology, FNA = fine needle aspiration, ROM = risk of malignancy, AUS = atypia of undetermined significance, SUMP = salivary gland neoplasm of uncertain malignant potential

The most common MGRSGC category among the cases with available surgical follow-up was the benign/category-IVA (190 cases; 71.2%), followed by non-neoplastic/category-II (29 cases; 10.9%) and malignant/category-VI (20 cases; 7.5%). The histopathological distribution of these cases with available surgical follow-up has been displayed in Tables 3 and 4. The most common benign neoplasm was pleomorphic adenoma/PA (153 cases; 86.9%), and the most malignant neoplasm was MEC (43 cases; 59.7%). The spectrum of malignant cases has been displayed in Table 4. The ROM was calculated for each category and was found to be 0%, 27.6%, 60%, 11.8%, 45.5%, 90%, and 100% for ND, NN, AUS, NB, SUMP, SFM, and malignant categories, respectively [Table 2].

Table 3.

Distribution of cases according to the histopathological diagnosis in the cases with available surgical follow-up

HPE category	HPE diagnosis	Cat I	CatII	Cat III	Cat IVA	Cat IVB	CatV	Cat VI
Non-neoplastic
	Epidermal inclusion cyst (1)		1 (3.4%)
	Granulomatous inflammation (1)		1 (3.4%)
	Benign cyst/lymphoepithelial cyst (3)		3 (10.3%)
	Sialadenitis acute/chronic (6)		4 (13.8%)			1 (9.1%)	1 (10%)
	Sjogren’s syndrome		2 (6.9%)
	Tubercular lymphadenitis (2)		2 (6.9%)
	Unremarkable parenchyma/no evidence of malignancy/non-diagnostic (5)	1 (50%)	3 (10.3%)	1 (20%)
Benign
	Castleman disease (1)		1 (3.4%)
	Clear cell oncocytoma (1)				1 (0.5%)
	Lymphangioma (1)				1 (0.5%)
	Monomorphic adenoma (2)				1 (0.5%)		1 (10%)
	Myoepithelioma (2)				2 (1%)
	Pleomorphic adenoma (157)		2 (6.9%)		153 (80.5%)	2 (18.2%)
	Cellular and atypical pleomorphic adenoma(2)			1 (20%)	1 (0.5%)
	Neurofibroma (1)				1 (0.5%)
	Schwannoma (4)				4 (2.1%)
	Warthin tumor (9)	1 (50%)	2 (6.9%)		3 (1.6%)	3 (27.1%)
Malignant
	Acinic cell carcinoma (3)				2 (1%)			1 (5%)
	Adenoid cystic carcinoma (3)				2 (1%)			1 (5%)
	Basal cell adenocarcinoma (1)				1 (0.5%)
	Carcinoma ex pleomorphic adenoma (1)						1 (10%)
	Epithelial–myoepithelial carcinoma (1)				1 (0.5%)
	Mucoepidermoid carcinoma (40)		8 (27.2%)	3 (60%)	14 (7.4%)	4 (36.4%)	4 (40%)	7 (35%)
	Metastatic infiltrating ductal carcinoma (IDC) breast (1)							1 (5%)
	Malignant mesenchymal tumor/MPNST (2)				2 (1%)
	Malignant mixed tumor (1)							1 (5%)
	Myoepithelial carcinoma (1)				1 (0.5%)
	Non-Hodgkin’s lymphoma (2)							2 (10%)
	Poorly differentiated carcinoma (2)							2 (10%)
	Salivary duct carcinoma (2)						1 (10%)	1 (5%)
	Squamous cell carcinoma (metastatic) (4)						2 (20%)	2 (10%)
	Sebaceous carcinoma (metastatic) (2)							2 (10%)
	Secretory carcinoma (1)					1 (9.1%)
Total cases in each category		2 (7.4%)	29 (10.9%)	5 (1.9%)	190 (71.2%)	11 (4.2%)	10 (3.7%)	20 (7%)

HPE = histopathology, Cat = category, IDC = infiltrating ductal carcinoma, MPNST = malignant peripheral sheath tumor

Table 4.

Type of malignant tumor

Primary (60 cases)	No. of cases	Metastatic (7 cases)	No. of cases
Mucoepidermoid carcinoma	40	Squamous cell carcinoma	4
Acinic cell carcinoma	3	Sebaceous carcinoma (metastatic)	2
Adenoid cystic carcinoma	3	Metastatic IDC-breast	1
Salivary duct carcinoma	2
Poorly differentiated carcinoma	2
Non-Hodgkin lymphoma	2
Malignant mesenchymal tumor/MPNST	2
Carcinoma ex pleomorphic adenoma	1
Epithelial–myoepithelial carcinoma	1
Myoepithelial carcinoma	1
Basal cell adenocarcinoma	1
Malignant mixed tumor	1
Secretory carcinoma	1

MPNST = malignant peripheral sheath tumor, IDC = infiltrating ductal carcinoma

For calculation of the diagnostic efficacy of the MSRSGC in determining the malignancy, the number of cases positive/suspicious and negative for malignancy on cytopathology were compared with the cases that were positive/negative for malignancy on surgical follow-up. Metrics of diagnostic efficacy, namely, sensitivity, specificity, PPV, and NPV of MSRSGC in reporting malignancy, were calculated and found to be 49.3%, 96%, 80.5%, and 85%, respectively. These data have been tabulated in Table 5.

Table 5.

Evaluation of the diagnostic efficacy of the Milan System for Reporting Salivary Gland Cytopathology (MSRSGC) categories in our study

	Positive for malignancy on surgical follow-up with histopathological examination	Negative for malignancy on surgical follow-up with histopathological examination	Total
Positive/suspicious for malignancy on cytopathology	True-positive (TP) = 33	False-positive (FP) = 8	41
Negative for malignancy on cytopathology	False-negative (FN) = 34	True-negative (TN) = 192	226
Total	67	200	267
Diagnostic efficacy metrics of the MSRSGC categories in our study
Sensitivity [TP/(TP+FN)]		49.3%
Specificity [TN/(FP+TN)]		96%
Positive predictive value (PPV) [TP/(TP+FP)]		80.5%
Negative predictive value (NPV) [TN/(FN+TN)]		85%

DISCUSSION

The study aimed at validating the role of the MSRSGC in making the diagnosis of SG-FNAs within a tertiary care referral center in Northern India. The distribution of SG-FNA cases in our study [Table 2] shows a predominance of benign/category-IVA (71.2%) cases, followed by NN/category-II (10.9%), malignant (M)/category-VI (7.5%), and SG neoplasm of uncertain malignant potential (SUMP)/category-IVB (4.1%) and compares the number of cases and ROM of the six MSRSGC categories in our study with that of the others [Table 6]. Interestingly, our non-diagnostic category (ND)/category-I included only 0.7% of the total cases, which is the lowest among all the studies. This is probably because (1) in our institute, FNA is performed by experienced pathologists and cytotechnicians under USG guidance with ROSE; (2) our institute is a tertiary care referral center offering cancer treatment where higher-category cases are frequently referred for surgeries rather than the lower-category cases, which can be managed more conservatively. Harb et al. and Johnson et al. demonstrated that the USG guidance allows better diagnostic yield in the SG-FNAs when compared to unguided palpation guidance.[3,4,5]

Table 6.

Comparison of ROM among different studies and Milan System for Reporting Salivary Gland Cytopathology (MSRSGC)

Published report (included cases with surgical follow-up)	Milan diagnostic category, % cases, ROM
	I		II		III		IVA		IVB		V		VI

	%cases	ROM (%)	% cases	ROM (%)	% cases	ROM (%)	% cases	ROM (%)	% cases	ROM (%)	% cases	ROM (%)	% cases	ROM (%)
MSRSGC (1st edition), 2018	–	25	–	10	–	20	–	<5	–	35	–	60	–	90
MGRSGC (2nd edition), 2023	–	15	–	11	–	30	–	<3	–	35	–	83	–	>98
Lubin et al. diagnostic cytopathology, 2020 (373 cases)	14.5	37.5	20.0	31.3	5.4	52.2	24.8	3.8	13.8	41.3	2.5	92.3	18.9	99.0
Tochtermann et al. Cancer Medicine, 2023 (753 cases)	6.0	15.7	9.3	12.3	6.6	26.9	48.7	2.9	10.4	35.7	3.3	84.8	15.7	98.5
Ratzon et al. Journal of American Society of Cytopathology, 2023 (638 cases)	3.2	29	9.7	22	9.4	51	16.1	25	21.5	62	6.2	88	33.8	99
Gomez et al. Journal of American Society of Cytopathology, 2024 (212 cases)	4.75	40	17.22	25	1.78	51	27.33	25	11.29	62	3.17	88	34.46	99
Our study (267 cases)	0.7	0	10.9	27.6	1.9	60	71.2	11.8	4.1	45.5	3.7	90	7.5	100

ROM = risk of malignancy

The ROM for various categories calculated in our study has been compared with MSRSGC (both first and second editions) and other previous reports in the literature in Table 6.[1,2,5,6,7,8]

It is noteworthy that the ROM of the MSRSGC has been refined in the second edition. Modifications have been made in the ROM in all the categories, except category-IVB, which remains at 35%. While the ROM has been decreased in categories-I and IVA, it has increased in categories-II, III, V, and VI.

Category-I/non-diagnostic

The ROM of all categories in our study, except that of Category-I, is more than that of the MSRSGC. In the latest MSRSGC, the ROM of category-I is 15%, and for other studies, it varied from 15.7% to 40%. The ROM in Category-I in our study is 0%, probably because this category includes only 0.7% of the total study cases, due to the reasons mentioned above.

Category-II/non-neoplastic

The cases included in this category have been summarized in Table 3. Category-II formed the second largest category in our study (both with and without surgical follow-up). Surgical follow-up was available in only 11.5% cases: 16/29 cases (55.2%) in this category were truly non-neoplastic. This category included cases like sialadenitis, benign/lymphoepithelial cysts [Figure 1a], and tubercular lymphadenitis/granulomatous pathology [Figure 1e]. Of these, four (13.8%) cases were non-neoplastic cystic lesions like benign/lymphoepithelial and epidermal inclusion cysts. The RON in this category was 44.8% (13/29 cases): 16/29 (17.2%) cases were actually benign, belonging to entities like pleomorphic adenoma (PA) [Figure 1b], Castleman disease, Warthin tumor [Figure 1c], and schwannoma [Figure 1d], which were misdiagnosed as sialadenitis or benign cystic lesion. A large portion (8/29 cases; 27.2%) of cases were falsely labeled as negative and actually belonged to MEC, mostly low to intermediate grade, which were mistakenly reported as benign cystic lesion (4) and sialadenitis (2). Two cases were reported to have no evidence of malignancy. In contrast to this, Rossi et al. Viswanathan et al. Song et al. and Tochtermann et al. reported the highest false negative rates for lymphoma, rather than MEC, due to misinterpretation as a reactive lymph node.[7,9,10,11]

Figure 1.

(a) Lymphoepithelial cyst: Smear shows mixed infiltration of lymphocytes, histiocytes, and plasma cells on a thin proteinaceous background (May–Grunwald–Giemsa [MGG], 40× magnification). (b) Pleomorphic adenoma: smear shows both benign epithelial/myoepithelial cells on chondromyxoid stroma background (40× magnification). (c) Warthin tumor: Smear shows few oncocytic cells on lymphoid cell background (MGG, 40× magnification). (d) Schwannoma: smear shows few clusters of uniform spindle cells with elongated nuclei having tapering end (40× magnification). (e) Granulomatous pathology: Smear shows granuloma composed of epithelioid histiocytes, giant cells, necrosis, and lymphocytes (MGG, 40× magnification). (f) Carcinoma ex pleomorphic adenoma: Smear shows malignant epithelial cells arising from a primary or recurrent benign pleomorphic adenoma cell (MGG, 40× magnification). (g) Adenoid cystic carcinoma: Smear shows monomorphic round to oval hyperchromatic atypical cells with scant cytoplasm and variable hyaline globule (MGG, 40× magnification). (h, i) Mucoepidermoid carcinoma: Smear shows Malignant epithelial cells, mucous-secreting cells, and intermediate and epidermoid cells along with tumor cells display mucin globules in cytoplasm (PAD-positive, MGG, 40× magnification)

Although these cases were diagnosed as false-negative for neoplasm, they were followed-up and underwent surgical removal due to constant clinico-radiological indication of malignancy. The ROM of this category was calculated to be 27.6% in comparison to the MSRSGC, second edition (11%).[1] The ROM of other studies ranged from 10.8% to 31.3%. Higher ROM similar to ours was reported by Lubin et al., Ratzon et al., and Gomez et al. which are in cancer treatment centers like ours.[5,6,8]

Category-III/atypia of undetermined significance (AUS)

This AUS category included those for which definite cytomorphological features of neoplasm are quantitively or qualitatively lacking, but cytological atypia prohibits the impartment of the non-neoplastic category. Thus, this category, on the one hand, prevents the diagnosis of false-negative cases in category-II, like cystic MEC and lymphomas [Figure 2c]. On the other hand, it prevents diagnosis of false-positive cases like inflammatory pathology or metaplasia in the neoplastic categories. As per the MSRSGC, this category should be <10%, which is a marker of good quality.

Figure 2.

(a) Salivary duct carcinoma: Smear shows atypical epithelial cells arranged in clusters and singly scattered with round to oval nucleus, vesicular chromatin, focally distinct nucleoli, and moderate amount of basophilic cytoplasm (hematoxylin & eosin [H&E], 40× magnification). (b) Sebaceous carcinoma: Smear shows atypical epithelial cells arranged in clusters and singly scattered on a necrotic background and some cells with eccentrically placed nucleus, occasional large bizarre cells and foamy histiocytes (May–Grunwald–Giemsa [MGG], 40× magnification). (c) Non-Hodgkins lymphoma: Smear shows dispersed population of small- to medium-sized atypical lymphoid cells on hemorrhagic background (MGG, 40× magnification). (d and e) Secretory carcinoma: Smear shows nests of tumor cells with a papillary or glandular structure floating in mucinous secretions and multiple characteristic cytoplasmic vacuoles (MGG, 40× magnification). (f) Squamous cell carcinoma: Smear shows clusters of malignant polygonal epithelial cells, vesicular nucleus, inconspicuous nucleoli, and few keratinized squamous cells on necrotic background (H&E,40× magnification). (g) Metastatic infiltrating ductal carcinoma from breast: Smear shows atypical ductal epithelial cells arranged in clusters and singly scattered on a necrotic background (MGG, 40× magnification). (h) Poorly differentiated carcinoma: Smear shows mainly singly scattered and occasional clusters of atypical epithelial cells on necrotic background (PAP, 40× magnification). (i) Epithelial–myoepithelial carcinoma: Smear shows the biphasic population of tumor cells of ductal epithelial and myoepithelial cells arranged in small clusters and sheets. The myoepithelial cells are small, uniform nuclei; clear cytoplasm and distinct cell borders (May–Grunwald–Giemsa, 40× magnification)

In our study, this category included 5/267 (1.9%) cases and included cases like cellular/atypical PA (1) and MEC (3). One case showed unremarkable SG parenchyma only. We reported ROM to be 60%, which is clearly higher than 30% in MSRSGC, second edition.[2] Similar high ROMs in category-III have also been reported by Lubin et al., Ratzon et al., and Gomez et al.[5,6,8] These differences may be attributed to the unpredictable nature of these categories as there is a tendency for underdiagnosis in cases where definite features of malignancy are absent.

Tochtermann et al.,[7] in their study, performed repeat FNA in 13% of the cases in this category and obtained the ultimate diagnosis in 71.4% of cases. Thus, they concluded that one can follow recommendations by the MSRSGC to either repeat the FNA or to proceed for surgery.[7]

Category-IVA/benign neoplasm

This category allows strictly only those cases that present with typical cytological features of a particular benign neoplasm so as to keep the ROM to be <3%, which is revised in the second edition of MSRSGC. In our study, this category was the biggest, comprising 190/267 (71.2%) cases and predominately included benign entities like PA (154 cases; 80.5%), schwannoma (four cases; 2.1%), and Warthin tumor (3 cases; 1.6%). This category also contained a few false-negative cases like MEC [14, Figure 1h and i], which was misinterpreted as PA(10) and Warthin tumor (4); ACC(2), AdCC [2 Figure 1g], BCA(1), secretory carcinoma [1, Figure 2c and 2d], SCC [4, Figure 2f], epithelial–myoepithelial carcinoma [1, Figure 2i], metastatic infiltrating ductal carcinoma [Figure 2g], poorly differentiated carcinoma [Figure 1h], and myoepithelial carcinoma (1), which all were misdiagnosed as PA; and also malignant mesenchymal tumors/MPNST, which were misinterpreted to be benign (2). Thus, we reported the ROM to be 11.8%, which is again >3% of the cutoff laid down in the MSRSGC, second edition.

Category-IVB/SUMP

This category is made for those neoplastic cases where benign and malignant neoplasms cannot be differentiated cytologically. A typical example is that of a cellular variant of PA. The ROM continues to be 35%, even in the second edition of MSRSGC. In our study, this category had 11/267 (4.1%) cases and included non-neoplastic entities like sialadenitis (1); benign entities like PA (2) and Warthin tumor (3); and malignant entities like MEC (4) and SC(1). The ROM was calculated to be 45.5%, which is again more than that of MSRSGC, that is, 35%. Similar high ROMs of category-IVB have also been reported by Lubin et al., Ratzon et al., and Gomez et al., which are cancer treatment centers like ours.[5,6,8]

As per the MSRSGC, conservative surgery or a further work-up to refine the diagnosis of the tumor can be opted for. A definite diagnosis dictates the type and extent of surgery. And thus, it should be the goal of the cytopathologists to keep this category to the minimum. Ancillary techniques like immunocytochemistry, fluorescence in situ hybridization (FISH), or DNA-sequencing methods may be helpful not only in clinching the correct diagnosis in this category but in categories-III and V as well.

Category-V/suspicious for malignancy

This category had 10/267 (3.7%) cases and included non-neoplastic entities like sialadenitis (1); benign entity like monomorphic adenoma (1); and malignant entities like MEC(4), SCC [2, Figure 2f], and CaExPA [1, Figure 1f]. Smears from malignant mixed tumors of the SGs show biphasic tumors that exhibit both carcinomatous and sarcomatous elements. These tumors are thought to develop de novo in the SG and contain malignant stromal and epithelial elements on histology. The carcinomatous component was composed of adenocarcinoma, SCC, or undifferentiated carcinoma. Smears from CaExPA show the presence of variable-sized clusters of polygonal cells with relatively rich cytoplasm and large round-to-oval nuclei in a necrotic background. Interestingly, carcinoma cells mixed with the PA component arise from a preexisting benign mixed tumor. Usually, the patient will have a long-standing history of a SGL and may have multiple recurrent pleomorphic adenoma. Most of these tumors will have malignant epithelial components, but not malignant stromal components.

SDC [1, Figure 2a] and sebaceous carcinoma [Figure 2b]. The ROM provided by the second edition MSRSGC is 83%. Other studies have shown similar ROM. However, the ROM of ≥90% was seen by us (90%) and by Lubin et al. (92.3%).6

Category-VI/malignant

20/267 (7.5%) cases belonged to this category and included only true-positive cases. The spectrum of cases included in this category is shown in Table 3. MEC formed the commonest entity. The ROM is revised from 90% to >98% in the 2nd edition of MSRSGC. However, our ROM is more than both (i.e., 100%), similar to most other studies. The malignant cases included both the primary and secondary malignancies, as shown in Table 4. Notably, only 20/60 (33.3%) cases of primary malignancies and all (7/7) cases of secondary malignancies belonged to this category.

Calculation of metrics of diagnostic efficacy

The calculation of the metrics has been organized in Table 5. Our study reports the sensitivity, specificity, PPV, and NPV of MSRSGC in reporting the malignancy to be 49.3%, 96%, 80.5%, and 85%, respectively. We compared the metrics in our study with those of the others in Tables 6 and 7.[7,10,12,13,14,15,16]

Table 7.

Metrics of diagnostic efficacy compared with different studies

	No of cases in the study (with surgical follow-up)	Sensitivity (%)	Specificity (%)	Positive predictive value (%)	Negative predictive value (%)
Rohilla et al. Cancer Cytopathology, 2017	94	79.4	98.3	96.4	89.2
Vishwanathan et al. Cancer Cytopathology, 2018	373	79	98	92	94
Chen et al. Diagnostic Cytopathology, 2019	349	70.4	99.2	90.5	96.7
Wu H et al. American Journal of Clinical Pathology, 2019	694	75	98.4	88.9	95.3
Jha et al. Acta Cytologica, 2020	102	64.28	97.01	90	86.67
Tochtermann et al. Cancer Medicine, 2023	753	94.6	99.3	97.9	98.2
Nguyen et al. American Journal of Otolaryngology–Head and Neck Medicine and Surgery, 2023	215	90.9%	98.2%	90.9%	98.9%
Present study	267	49.3%	96%	80.5%	85%

The sensitivity reported in our study (49.3%) is much less than that in the other studies (64.28–94.6%). However, the specificity and PPV and NPV are also lowest among the compared studies but are still within a good range.

CONCLUSION

This study puts forward our experience with the application of the MSRSGC at a tertiary care institute located in northern India. To the best of our knowledge, this is the largest Indian study involving 793 cases overall and 267 cases with available histopathological follow-up data, consulting our institute over a duration of 13 years and 2 months. Notably, our study had the lowest number of cases in category-I (0.7%); probably, thus the ROM for category-I in our study was also quite low as compared to the report of the MSRSGC and other studies. This fact also reverberates the fact that USG guidance and ROSE performed by experienced hands improve the diagnostic yield of salivary gland FNAs, thereby decreasing the amount of category-I. The bulk of cases in our study, quite unlike most other studies, were from category-IVA (71.2%).

The calculated ROMs for categories IVB, V, and VI are largely similar to the MSRSGC, especially the 2nd edition. On the contrary, ROM is found to be increased for the categories-II, III, and IVA, which may be attributed to the false positivity of the cases due to the vast complexity and wide overlapping spectrum of diagnoses involved. However, ancillary techniques, including IHC and molecular studies, may be helpful on this front. We also report the sensitivity, specificity, PPV, and NPV of MSRSGC in reporting malignancy to be 49.3%, 96%, 80.5%, and 85%, respectively.

So, to conclude, our study strongly supports the use of MSRSGC for the categorization of the salivary gland FNAs by authenticating its validity and effectiveness. We believe that the MSRSGC provides a schema for categorizing the nature of SGLs, providing their accurate diagnosis, and suggesting an apt clinico-radiological correlation or ancillary work-up to reach the correct diagnosis, followed by relevant management of the lesion.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.