Correlation of Salivary Gland FNA Reporting Using the Modified Milan System for Reporting Salivary Gland Cytopathology (MSRSGC) and Intraoperative Frozen Section Evaluation with Final Histopathology Report for Major Salivary Gland Neoplasms

Abstract

Fine needle aspiration (FNA) is a commonly used investigation for salivary gland lesions. The modified MSRSGC was introduced to standardise the FNA reporting of major salivary gland lesions. A frozen section is also used intraoperatively for these lesions. In this retrospective study, we included all patients who were treated at our institute between January 2012 to December 2019. The FNA reports of all the patients were reclassified based on the modified MSRSGC, and the sensitivity, specificity, and positive and negative predictive values were calculated. We also assessed the indication for utilising the frozen section and correlated it with the preoperative FNA and the final histopathology report. A total of 325 patients satisfied the eligibility criteria and were included in this study. The sensitivity and specificity, positive predictive value, and negative predictive value of the modified MSRSGC were 64.18% and 91.94% and 92.27% and 63.11%, respectively. The frozen section (FS) was done in 131 patients, the commonest reason was for obtaining a primary diagnosis (n = 104,79.3%). When the FNA was type V and above the primary diagnosis was that of malignancy in the vast majority (p < 0.001, kappa 0.563). The values of the modified MSRSGC were comparable with available literature in all categories except categories I and II which were higher than what is reported in the literature. When the preoperative FNA was modified to MSRSGC V and above, the possibility of malignancy was high, and the use of FS for primary diagnosis may not add much value intraoperatively.

Introduction

Fine needle aspiration cytology (FNAC) plays a very important role in the workup of (major) salivary gland lesions. The FNAC results help the surgeon counsel regarding the treatment decisions (extent of surgery), possible complications, and prognosis [1]. However, the reporting of (major) salivary gland lesions FNAC lacks a universally accepted uniform reporting system such as the Bethesda system for reporting thyroid gland FNAC [2]. This is compounded by the diverse spectrum of histopathological diagnoses available for salivary gland neoplasm [3]. The need for a uniform reporting system was long felt for better communication between the surgeons and the pathologists across institutions worldwide and more importantly with the patients. To address this, the Milan System for Reporting Salivary Gland Cytopathology (MSRSGC) was developed in 2015 [4]. Subsequently, a modified MSRSGC was introduced to improve relevance in terms of uniform surgical decision-making [5]. One of the criticisms of MSRSGC is its suboptimal diagnostic sensitivity in differentiating benign and malignant lesions. [6] Also, a frozen section is utilised during surgery of the salivary gland for various indications such as for primary diagnosis, identification of metastatic nodes, and margins among others.

In the present study, we attempted to understand the utility of the modified MSRSGC system for diagnosing salivary gland cytopathology in routine clinical practice and to understand its correlation with intraoperative frozen section in obtaining a primary diagnosis for major salivary gland neoplasms.

Methodology

We took our institutional ethics committee’s approval for this retrospective study. All patients with major salivary gland neoplasms, aged 18 years and above, having undergone surgery at our institute between January 1st, 2012, and December 31st, 2019, with the availability of preoperative FNAC diagnosis and final histopathology report were included in this study. Patients who underwent preoperative FNAC outside the institute and when these slides were unavailable for review by in-house pathologists were excluded from the study. Minor salivary gland neoplasms, non-salivary gland primaries i.e. metastasis to the salivary gland from known primaries, and infective and inflammatory lesions were also excluded from the study.

Our in-house head and neck pathologist went through all the reports of the preoperative FNAC and classified them based on the description of the FNAC findings in the report into one of the categories of the modified MSRSGC [5] (Table 1). These FNAs were reported by experienced oncopathologists with 13 of them having experience greater than 10 years in the field and another 13 of them with 5–10 years’ experience in the field.

Table 1.

The modified Milan system for reporting salivary gland cytopathology (MSRSGC) categories

Diagnostic category	Benign (in final HPR) (n = 124))	Malignant (in final HPR) (n = 201)	Risk of malignancy (ROM)	Risk of malignancy (ROM) reported in literature [5]
Non-diagnostic (I) (n = 18)	11	7	38.8%	19%
Non-neoplastic (II) (n = 21)	8	13	61.9%	11.8%
Benign (III) (n = 65)	56	9	13.8%	5.6%
Salivary lesion of uncertain malignant potential (SLUMP) (IV) (n = 82)	39	43	52.4%	45%
Suspicious for malignancy (V) (n = 18)	4	14	77.7%	71.4%
Malignant (VI) (n = 120)
Low-grade salivary gland neoplasm (VIa)	4	69	94.5%	86.4%
High-grade salivary gland neoplasm or other malignant neoplasms (VIb)	2	45	95.7%	97%
Hematologic malignant neoplasm (VII) (n = 1)	0	1	100%	100%

Sensitivity = 64.18%, specificity = 91.94%, PPV = 92.27%, NPV = 63.11%, diagnostic accuracy = 75.28%

The various demographic and clinical details are age, gender, major salivary gland location of neoplasm, history of previous treatment if any, preoperative FNAC report, surgery done for the primary and the neck, frozen section details, and the final histopathology report, including the grade. Based on the FNAC report, each patient was assigned a category as per the three systems and documented.

For each system, the risk of malignancy (ROM) for every category was calculated. An agreement analysis between the preoperative FNAC diagnosis and the final histopathology report was also done. Subsequently, the sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic accuracy were calculated for each system.

Statistical Analysis

Analysis was done using SPSS version 24 (IBM, New York). An agreement analysis was performed to assess the preoperative FNAC diagnosis of the primary tumour and the final histopathology reporting. A p-value of < 0.05 was considered significant, and a kappa value was used to interpret the agreement between the preoperative FNAC and final histopathology report for each of the three systems. Also, the sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic accuracy were calculated for each system.

Results

Three hundred and twenty-five patients satisfied the eligibility criteria and were included in this study. The median age of the cohort was 49 years (range: 18–85 years), and the majority were males (n = 192, 59.1%). Most of the patients were treatment naïve (n = 231, 71.1%), whereas 94 patients (28.9%) had received some form of prior treatment, mostly surgery with or without adjuvant therapy. The parotid gland was the most common site (n = 296, 91.1%), followed by the submandibular gland (n = 28,8.6%) and the sublingual gland (n = 1, 0.3%). Superficial parotidectomy (n = 128, 39.4%) was the most common surgery performed for the primary followed by total conservative parotidectomy (n = 69, 21.2%), radical/extended radical parotidectomy (n = 74, 22.7%). Neck dissection was performed in 100 (30.7%) patients. Most of them were malignant lesions (n = 207, 63.7%). Pleomorphic adenoma (n = 83, 25.5%) was the most common benign neoplasm followed by Warthin’s tumour (n = 18, 5.5%) and others (n = 17, 5.2%). Mucoepidermoid carcinoma (n = 68,20.9%) was the most common malignant major salivary gland carcinoma followed by adenoid cystic carcinoma (n = 28, 8.6%), carcinoma ex-pleomorphic adenoma (n = 23, 7.1%), salivary duct carcinoma (n = 18,5.5%), acinic cell carcinoma (n = 14,4.3%), and others (n = 56, 17.2%).

FNAC Reporting as per Modified MSRSGC (Table 1)

The number of non-diagnostic (I) and non-neoplastic (II) FNAC reported was 18 (5.5%) and 21 (6.4%), respectively. There were 65 (20%) FNACs reported as benign (III) and 82 (25.2%) SUMP reported. FNAC was reported as suspicious for malignancy (V) in 18 (5.5%). Seventy-three (22.4%) FNACs were reported as low-grade neoplasms (VIa) and 47 (14.4%) as high-grade neoplasms (VIb). In one patient, the FNAC was reported as a haematolymphoid malignancy. The risk of malignancy for each of the categories was 38.8% for I, 61.9% for II, 13.8% for III, 52.4% for IV, 77.7% for V, 94.5% for VIa, and 95.7% for VIb. The sensitivity, specificity, PPV, NPV, and diagnostic accuracy were 64.18%, 91.94%, 92.27%, 63.11%, and 75.28%, respectively. There was moderate agreement (p < 0.001, kappa = 0.512) between the FNAC categories assigned and the diagnosis of malignancies in the final histopathology report.

The sensitivity of the modified MSRSGC was low compared to the specificity which was higher than 90%. The diagnostic accuracy too was around 75% for modified MSRSGC. The lower sensitivity could be attributed to the fact that the FNA reporting systems were being used recently and may get better with experience. This may also be true for the lower diagnostic accuracy.

Correlation Between Preoperative FNA Based on Modified MSRSGC and Intraoperative Frozen Section (FS) (Table 2)

Table 2.

Correlation of preoperative FNA (modified MSRSGC) and intraoperative frozen section findings for the diagnosis of the primary lesion

Variables	Final HPR		p-value	Kappa
	Benign	Malignant
Milan I–IV			< 0.001	0.649
Frozen report
Benign	43	1
Malignant	4	27
Inconclusive	3	3
No primary diagnosis	1	5
Milan V-VII			0.004	0.165
Frozen report
Benign	2	3
Malignant	0	25
Inconclusive	1	3
No primary diagnosis	0	10

Overall, kappa score was 0.563, p < 0.001

The intraoperative frozen section was used in 131 (40.3%) patients out of 325 patients for various reasons such as for primary diagnosis, nodal status (metastatic/non-metastatic), margins alone, or in various combinations (Fig. 1). FS was more useful in patients with MSRSGC category I-IV FNA reporting (p < 0.001, kappa 0.649). In our previous publication, we reported on the role of dedicated head and neck pathologists in our disease management group (DMG) and the functioning of head and neck pathologists and non-head and neck pathologists [7]. The frozen section was reported predominantly by non-head and neck pathologists. There was a good correlation between the frozen section report and the final HPR with both. Still, it was better with head and neck pathologists (p < 0.001, kappa 0.698) compared to non-head and neck pathologists (p < 0.001, kappa 0.519) in terms of being able to give a possible primary diagnosis of frozen section. The grade was reported in 17 patients (13%), and there was a reasonable agreement with the grades given in the final histopathology report; it was slightly better when reported by head and neck pathologists (p = 0.010, kappa 0.401) compared to when reported by non-head and neck pathologist (p < 0.001, kappa 0.349).

Fig. 1.

Indications for a frozen section in our cohort

The sensitivity, specificity, PPV, NPV, and diagnostic accuracy for the FS (compared to the final histopathology report (gold standard) are 91.84%, 92.86%, 91.84%, 92.86%, and 92.38%, respectively.

Discussion

Identifying malignancies before surgery or during surgery helps in the appropriate management of the major salivary gland tumours. A standardised system of reporting FNACs with a defined risk of malignancy (ROM) is preferred by clinicians [8]. Ever since the introduction of the MSRSGC [4], multiple publications have appeared wherein these reporting systems were used in their respective institutional cohorts with or without the final histopathology report following surgery [7–13]. There are fewer reports on the modified MSRSGC in similar lines. At our institute, we have also been using these reporting systems for salivary gland FNACs.

In the present study, we report the various aspects such as the sensitivity (64.18%), specificity (91.94%), PPV (92.27%), NPV (63.11%), and diagnostic accuracy (75.28%) of the modified MSRSGC. The ROM in our cohort is higher for MSRSGC categories I–IV compared to what is available in the literature [5], but similar for MSRSGC categories V–VII. When the preoperative FNA was Milan’s type V and above in modified MSRSGC, the possibility of malignancy was high, and the use of FS for primary diagnosis may not add much value. For Milan’s type, I–IV, FS was useful in differentiating between benign and malignant lesions. The grade was reported on FS in only 17 cases.

The ROM given in the literature for the various categories of modified MSRSGC is given in Table 1. In our series, the ROM was 38.8% for the non-diagnostic category, 61.9% for the non-neoplastic category, 13.8% for the benign category, 52.4% for SUMP, 77.7% for suspicious for the malignancy category, 86.4% for low-grade malignancy, 97% for high-grade malignancy, and 100% for haematologic malignant lesion category in the modified MSRSGC reporting system. The ROM was within the reported range in all categories except for the non-diagnostic, non-neoplastic, and benign categories. This could probably be explained by the referral bias of the cases to a tertiary cancer referral centre. Also, note that the number of patients in categories I (7/18, 38.8%) and II (13/21, 61.9%) are less overall. The other reason could be due to the inherent limitations of the MSRSGC itself [3].

Though FNA is the initial investigation of choice in the evaluation of major salivary gland lesions, it has various pitfalls [14]. Frozen section (FS) is a useful tool in the management of major salivary gland lesions along with the FNA. Apart from helping to differentiate between benign and malignant lesions of the major salivary gland, it is also useful to assess margins, refine the diagnosis of the lesion, and assess the nodal status of malignant major salivary gland lesions [15, 16]. FNA and FS can be complementary to each other in the management of major salivary gland lesions. In our study, we have seen that the sensitivity, specificity, PPV, NPV, and diagnostic accuracy for the FS are almost similar to what is reported in the literature. However, FS was more common in patients with MSRSGC category I–IV lesions and major salivary gland lesions; it does not appear to add in the management for category V–VII lesions. The value of intraoperative FS has been shown to vary widely in literature (40–100%) [15]. FS has various intraoperative utilities: one is to differentiate between neoplastic and non-neoplastic lesions; second, to differentiate between benign and malignant neoplastic lesions; and third, to assess the grade of the malignant lesion if possible (which may help intensify the surgery in the form of adding a neck dissection in high-grade lesions, particularly) and the neck node status (metastatic/non-metastatic), to assess margins, and at times, to take decisions regarding the possibility of saving or sacrificing the facial nerve. In our study, we used the FS for all of these reasons in our cases (Fig. 1).

The strengths of the study are the reasonably good sample size with which we have provided the various parameters, such as the sensitivity specificity, PPV, NPV, and diagnostic accuracy, for the modified MSRSGC and how it may be used effectively with FS. The limitation, however, is the retrospective nature of the study.

Conclusions

The values of the modified MSRSGC in our cohort were comparable with those available in the literature except for categories I and II which were higher than what is reported in the literature. When the preoperative FNA was type V and above, as per the modified MSRSGC, the possibility of malignancy was high, and the use of FS for primary diagnosis may not add much value in that regard. FS may be used for various other reasons other than the primary diagnosis in these instances.

Data Availability

Data can be considered to be made available only on reasonable request (for reasons of confidentiality and the necessity of obtaining further clearances to do the same).

Declarations

Ethics Approval

Institutional ethics committee approval was taken for this study.

Conflict of Interest

The authors declare no competing interests.