Diagnostic performance of inflammatory biomarkers and cytological analysis in salivary gland tumors

This study aimed to evaluate the diagnostic performance of serum inflammatory biomarkers in salivary gland tumors with dubious results following cytological analysis.

5][6] Sensitivity and specificity of FNAC for parotid gland lesions were reported to be 78% and 98%, respectively. 7,8he Milan system for reporting salivary gland cytopathology (MSRSGC) was conceived during the 39th European Cytology Congress held in Milan in 2015.It provides a guide for diagnosis and management according to the risk of malignancy (ROM) as divided into six categories.These are (1) nondiagnostic, (2) nonneoplastic, (3) atypia of undetermined significance, (4) neoplasm (further subdivided into benign neoplasms and salivary gland neoplasms of uncertain malignant potential), (5) suspicious for malignancy, and ( 6) malignant.The ROM of each category are 25%, 10%, 20%, 5%, 35%, 60%, and 90% respectively. 9hus, while the introduction of this system has made the stratification of lesions based on cellularity easier from a cytopathological point of view, it has not resolved diagnostic problems in a clinical setting.Use of this system still leaves a considerable margin of diagnostic uncertainty making therapeutic and management decisions potentially controversial.Diagnostic imaging investigations provide helpful support for selection of therapy, however, they are not enough to guide the decisionmaking process alone. 10n such cases, a significant role could be played by inflammatory biomarkers.2][13][14][15] The efficacy of these biomarkers as a prognostic factor in salivary gland tumors has already been widely demonstrated, but no evidence investigates a potential role for them as a preoperative diagnostic tool in SGT. 16ased on this evidence, a retrospective study was conducted to evaluate the diagnostic performance of the main inflammatory biomarkers, compared to FNAC alone, for salivary gland tumors.

| Study design
From January 2011 to June 2022 data were collected from all patients surgically treated for salivary gland tumors admitted to the Department of Maxillofacial Surgery of the University of Naples "Federico II" and a retrospective study was performed.
Ethical review and approval were waived for this study due to its observational and retrospective nature.
Informed consent was obtained from all subjects involved in the study in accordance with the Helsinki declaration.A total of 905 salivary gland tumors surgically treated in our department were collected from our clinic's digital records.Among these, 239 were eligible for this study, satisfying the following inclusion criteria.
• The presence of salivary gland tumor(s) was confirmed histologically; • Complete medical records were available; • A routine preoperative blood count was conducted; • Patient age > 18; • Preoperative ultrasound-guided FNAC examination was conducted.
The exclusion criteria were: • Patients with inflammatory diseases with potential to alter NLR, PLR, SII, and SIRI values were excluded (such as chronic or acute inflammatory disease; autoimmune hematological disorders, and anti-inflammatory/steroidal treatments).Two hundred sixty-six patients were excluded due to incomplete clinical data; 270 for chronic or acute inflammatory disease, autoimmune hematological disorders, and/or anti-inflammatory/steroid treatments; 64 were less than 18 years old; 66 had tumors at any other site, and/or had received radiation or chemotherapy.

| Data collection
The diagnostic workup for all the patients involved a complete physical examination, routine blood count, fine needle aspiration cytology of the tumor, neck ultrasound, head and neck CT scan, and/or MRI with contrast.
Relevant clinical and pathological data such as demographic information (age, sex), FNAC report, histological diagnosis, tumor site, and routine laboratory data performed before surgery, were collected from the medical records.
From 2018 to 2022, the results of the FNAC were categorized according to the Milan System.Therefore, to standardize the sample, we considered categories III, IVa, and IVb benign (ROM: 20%, <5%, and 35%) and categories V and VI malignant (ROM: 60% and 90%).
Surgical management of benign salivary gland tumors was performed according to the European Salivary Gland Society guidelines. 17uperficial parotidectomy was performed for T1 or T2 low-grade superficial tumors and a total parotidectomy was performed for high-grade or T3-T4 tumors. 10ubmandibular sialoadenectomy was performed for any tumor located in the submandibular gland.Excision including a margin of 1 cm of healthy tissue was performed for any tumor located in the minor salivary glands.Neck Dissection was performed in selected cases, based on the clinical, radiological, and cytological analysis according to current guidelines for the treatment of malignant tumors of the salivary glands. 18,19retreatment baseline NLR, PLR, SII, and SIRI were calculated using the following formulae:

| Statistical analysis
Quantitative variables are expressed as mean and standard deviation while categorical variables are expressed as frequency and percentage.The diagnostic performance of the cytological examination was summarized using raw accuracy, sensitivity, and specificity.To evaluate the diagnostic performance of the inflammatory biomarkers (SII, SIRI, PLR, and NLR), receiver operating characteristic curves (ROCs) were constructed and the corresponding areas under the curves (AUCs) were calculated.Optimal cut-offs for each marker were determined maximizing the Youden index.Additionally, the diagnostic performance of the cytological examination combined with the best-performing inflammatory marker was evaluated.The combination of the inflammatory marker at the specified threshold with the FNAC was considered dichotomous and in particular was defined as having a value of 1 if at least FNAC or the marker was positive and as 0 if neither was positive.Differences in the diagnostic performances achieved were computed with the Mc Nemar's test.Simple and multinomial logistic regression was used to investigate predictors of discordance between the FNAC and histology.For all analyses, a p-value <0.05 was considered statistically significant.Analyses were performed using the statistical software R, version 4.0.3.

| RESULTS
The  1.All 140 benign tumors were located in the parotid gland, of the 99 malignant tumors, 68 (68.5%) were located in the parotid glands, 13 (13%) in the submandibular gland, and 18 (18.5%) in the minor salivary glands.Pleomorphic adenoma was the most common in the benign tumor group (48.5%), while mucoepidermoid carcinoma and adenocarcinoma (32.3%) were most common in the malignant tumor group.The final histological diagnoses of the included tumors are summarized in Table 2.

| Diagnostic performance of fine needle aspiration cytology and inflammatory biomarkers
Cytological diagnosis did not match with the definitive histological diagnosis in 53 cases (22%).Among these, a mismatch between cytological diagnosis of benignity and histopathological diagnosis of malignancy was reported in 40 cases (75%).Furthermore, through the Youden index method, the optimal cut-offs of each individual biomarker were calculated to discriminate the tumor from benign to malignant (SII:788, SIRI:0.94,PLR: 129, and NLR:3.09).With the Youden index method, the accuracy, sensitivity, and specificity of cytological analysis were 77.8%, 59.6%, and 90.7% respectively.
Among the major biomarkers examined (SII, PLR, and NLR) SIRI performed an AUC of 0.77 (Figure 1).The best SIRI cut-off was 0.94 with an accuracy of 79.9% and a sensitivity of 66.7% (Table 3).
Logistic regression analysis of the main clinical outcome for cytological-histological discordance was statistically significant for tumor size (<0.001;Table 4).

| Diagnostic performance combining SIRI and FNAC
Since among the biomarkers SIRI showed better accuracy than the others, we decided to combine the positivity to the SIRI+ FNAC (i.e., considering a subject positive if at least one of the two indices is positive) to evaluate the diagnostic performance.
The ROC curve constructed using the combination of SIRI + FNAC had an AUC of 0.81 (Figure 1).

| DISCUSSION
The preoperative diagnosis of SGTs is still a widely debated topic in the literature. 20,21Current evidence advises against a surgical open biopsy for SGTs; this makes preoperative diagnosis of these lesions difficult.Liu et al have shown that open surgery for histological sampling may result in spillage or seeding, scarring, nerve damage, and salivary fistulas. 7Core needle biopsy (CNB) is another method of obtaining tissue for histological sampling in salivary gland neoplasms.This technique involves the collection of salivary tissue through a cutting needle with gauges ranging from 12G to 19G.
Although it is generally considered a safe technique, needle thickness may increase the risk of complications.Witt et al. highlighted that CNB is a more invasive technique that requires local anesthesia and may lead to tumor spread along the needle tract, hematoma (1.6%), and temporary facial nerve weakness (0.2%). 22,23A B L E 3 Diagnostic performance of cytological analysis and inflammatory biomarkers.In this scenario, FNAC of salivary gland tumors is considered a safer and simpler method to achieve preoperative diagnosis. 24his technique allows preservation of the integrity of surrounding tissues and parenchyma maintaining a low complication rate.

Measure
In 2018, the Milan System was introduced to classify the cytology of the salivary glands; this classification includes six categories: I, nondiagnostic; II, non-neoplastic; III, atypia of indeterminate significance (AUS); IV, neoplastic, which is further subdivided into IVa benign À IVb salivary gland neoplasm of uncertain malignant potential (SUMP); V, suspicious for malignancy; and VI, malignant. 9he Milan system has helped facilitate pathological analysis by offering a standardized criterion for classifying salivary neoplasms based on cytological characteristics.
However, as shown by Rossi et al., this system comports a certain degree of uncertainty that is expressed using the Risk of Malignancy (ROM)score ranging from <5% to >90%. 9Several authors have recently highlighted the inaccuracy of cytological sampling with FNAC.Singh et al. observed four cases of discordance between diagnoses based on cytology versus those based on histopathology in 56 examined cases.One case of carcinoma ex pleomorphic adenoma was diagnosed as pleomorphic adenoma on cytology; one case of pleomorphic adenoma was diagnosed as mucoepidermoid carcinoma on cytology; one case of adenoid cystic carcinoma was diagnosed as pleomorphic adenoma; and one case of metastatic malignant melanoma was diagnosed as chronic sialoadenitis on cytology. 25ftikhar et al. reported discordance between FNAC and histopathology in patients with mucoepidermoid carcinoma of the parotid gland.Specifically, eight cytology specimens proved falsely negative upon analysis as mucoepidermoid carcinoma could not be detected with FNAC.Three cases were reported as high-grade mucoepidermoid carcinoma on histopathology and were underdiagnosed as pleomorphic adenoma on cytology. 26In view of the above, the mucoepidermoid carcinoma (MEC) may be underdiagnosed by FNAC and, depending on the portion aspirated, may be mistaken for pleomorphic adenoma or abscess. 27,28I G U R E 2 Diagnostic performance of SIRI and cytological analysis.
Moreover, some cells of the adenoid cystic carcinoma often resemble normal acinar cells in FNAC, therefore the tumor may evade diagnosis. 29ong-Gai Huang et al. reported four misdiagnosed cases of ACC of the salivary gland.When analyzing their causes: 4 cases (12.5%) were misdiagnosed by cell morphology as benign tumors (Iva); 2 cases were misdiagnosed as pleomorphic adenomas (PA); 2 other cases were misdiagnosed as basal cell adenomas (BCA). 30,31hus, diagnostic uncertainty complicates the clinical decision-making process.Hence, the need to identify new tools capable of supporting the surgeon when choosing the best treatment.
Several authors have highlighted how the values of inflammatory biomarkers vary according to tumor histotype.In our previous study, we underlined the role of the inflammatory status in benign and malignant salivary pathology, demonstrating a statistically significant increase in NLR, PLR, and SII indices in malignant salivary gland tumors compared to benign tumors. 32,33ased on this evidence, we conducted a retrospective study to evaluate the diagnostic performance of inflammatory biomarkers in preoperative SGTs diagnosis.
SIRI has proven to be the biomarker with the highest diagnostic performance.It was accurate in 79.9% of cases, with a sensitivity and specificity of 66.7% and 83.3%, respectively.Furthermore, when the performance of SIRI is combined with cytological analysis sensitivity increases to a statistically significant value of 82.8%.
This indicates that a lesion that appears clinically benign with either a suspicious cytology or a suprathreshold SIRI carries an 82.8% risk that the lesion will be revealed as malignant upon histological examination.
Moreover, it is noteworthy that regression analysis demonstrates that tumor size has a large impact on the extent of discordance between the results of cytological and histopathological analyses.
Furthermore, we found discordance between the preoperative FNAC results and final histopathological diagnoses to be higher in patients with tumor sizes of >4 cm.
According to Yildiz et al., the reason might be that tumors with a larger volume carry a higher risk that nondiagnostic tissue will be included in the cytological aspirate. 2ur study has some limitations.This is a retrospective single-center study of 239 patients with SGTs, the cytological examination was performed by different operators with 11 years of experience; unknown inflammatory diseases that remained unreported when taking each patient's history could influence the results of inflammatory biomarkers.Moreover, the small size of the sample did not allow the differential analysis of the cutoffs values based on the tumor localization.The result turned out statistically not significant in the sample examined and therefore was not reported.Further studies with larger patient cohorts are mandatory for the validation of our results.
In conclusion, the encouraging results of our study show that the SIRI score can be routinely collected in SGTs patients as a supplement to FNAC to achieve preoperative diagnosis, especially in dubious cases.When combining SIRI with cytological analysis, sensitivity significantly increases to 82.8%.

F I G U R E 1
Receiver operating characteristic curves of the main inflammatory biomarkers.[Color figure can be viewed at wileyonlinelibrary.com]

•
Presence of previous cancers at any other sites; • Radiotherapy or chemotherapy was included in the clinical history; • Incomplete clinical data.
Sample characteristics.Histopathological types of tumor and their location.