Evaluation of Preoperative Fine Needle Aspiration Accuracy in Diagnosis of Malignant Parotid Tumors

Objective: To evaluate the diagnostic accuracy of fine needle aspiration (FNA) in detecting malignant parotid tumors. Design: A retrospective study of cases series. Setting: Salmaniya Medical Complex, Otorhinolaryngology Department, Kingdom of Bahrain. Methods: A retrospective study was carried out, between January 2009 and December 2018, for all patients diagnosed with parotid tumors at the Department of Otorhinolaryngology, Head and Neck Surgery at Salmaniya Medical Complex, Manama, Kingdom of Bahrain. Data collection was made from medical records, I-seha and laboratory information system (LIS). 87 patients underwent parotidectomy, of which 30 were excluded due to unsatisfactory diagnoses or missing data. The results of fine needle aspirations were analyzed and compared with the corresponding histopathologic diagnosis. Estimation of sensitivity, specificity, negative predictive value, positive predictive value and accuracy of FNA were calculated. Results: Among the 57 evaluated cases, four cases were further excluded from the final analysis due to unsatisfactory FNA results, therefore only 53 cases remained. The mean age of patients was 48.6 ± 17.2 years. A concordance with histopathology results was observed. 47 cases were diagnosed as benign and five cases were malignant, clinical assessment radiological findings.


Introduction
The parotid glands are the largest of the three main salivary glands [1]. Their function is the secretion of saliva. Saliva is important for lubrication, swallowing, teeth protection and digestion. Salivary gland masses comprising a wide range of non-neoplastic lesions along with variety of benign and malignant neoplasms [1] [2]. Different types of tumors can arise from the parotid glands due to their histological diversity [3] [4]. Benign tumors are more common than malignant tumors [5] [6]. The two most common benign tumors are pleomorphic adenoma and Warthin's tumor [5]. Salivary gland tumors constitute 2% -6.5% of head and neck tumors [6] [7]. Of these 80% -85% are located in the parotid glands [1] [6]. The malignancy rate for parotid tumors has been reported to be usually between 14% -27% [2]. Mucoepidermoid carcinomas are the most common malignant lesions of the parotid glands [5].
The prevalence of these tumors is inconstant among different ethnicities and populations in different geographical locations [6]. Differentiation between benign and malignant tumors is difficult and cannot be carried out by physical examination and radiological studies alone [3] [4].
FNA cytology plays an important role in the initial evaluation of salivary gland lesions. FNA was first familiarized in 1920 for the assessment of parotid lesions and gained popularity 50 years later [6]. It is safe, simple, cost-effective and minimally invasive but its accuracy is doubtful, when compared to the gold standard histopathology [7]. FNA can provide helpful information for surgery planning and counselling patients regarding expectations from the surgery and after care prospects [8]. However, others emphasize that the management does not change regardless of the FNA result [9] [10].
Open biopsies are contraindicated as they can result in a potential risk for recurrence, secondary to tumor cell seeding [5]. Furthermore, a subsequent curative parotid surgery can be complicated by a previous open biopsy.
Variability in the diagnostic accuracy varies with clinical experience and geographical location owing to differences in referral patterns as well as in the prevalence of benign or malignant disease [10]. In our center, the standard methods for pre-operative assessment include radiological imaging (CT-SCAN) and FNA.
The aim of this study is to determine the diagnostic accuracy of FNA for detecting malignant parotid tumors preoperatively in Salmaniya Medical Complex (SMC), Manama, Kingdom of Bahrain, in which FNA accuracy had not been assessed previously.

Methods
A retrospective study was conducted for all patients diagnosed with benign or malignant parotid tumors who underwent parotidectomy in the Department of Otolaryngology-Head and Neck Surgery at SMC in Manama, between January 2009 and December 2018. It included all patients diagnosed with parotid tumors who underwent parotidectomy with adequate medical record and we excluded all patients with missing FNA cytology results or histopathology results or both. The patients with missing data or undetermined results were eliminated and were not involved in analysis.
In our institution, the standard procedure for FNA is performed in the outpatient setting, either the Swedish technique using a needle and needle holder in order to use suction to obtain the specimen, or the French technique (capillary method), by using only a needle [11].
The aspiration is done either by an experienced pathologist in the FNAC clinic using 23 -25 G needles or under ultrasound guidance in the radiology department by an interventional radiologist. The aspirated material is expelled on the slides and is smeared. Half of the slides are kept for air drying; which are stained by DiffQuik stain and the other half is fixed in 95% alcohol and is stained by Papanicolaou stain. Bedside adequacy is checked by the pathologists.
The remaining material usually kept for a cell block, in which some ancillary testing is done later in order to reach to a final diagnosis in controversy cases.
If a lymphoma is suspected at the time of checking the adequacy, a sample is sent for flowcytometry.
The slides are then studied carefully by the cytopathologists to reach to a final interpretation.
Patients data were obtained from medical records, I-seha and laboratory information system (LIS). The collected data were: patients demographics including age and gender; lesion characteristics including extension and location of the tumor (superficial or deep lobe or both), lesion size, patient's previous cancer history and the FNA method used (blind or ultrasound guided).
Patients data and lesions characteristics were compared and analyzed using SPSS version 23. Tumor size was categorized into 2 groups using 50 th percentile (4 cm) as cut-off point.
The cases were categorized into 3 groups: benign, malignant, or insufficient for diagnosis. Cases which were reported to have unsatisfactory FNA, were not included in the analysis and results. The proportion of each disease was calculated. After-that, the results of FNA were then compared with the histology results. The cases were classified as true positive (TP) when patients were diagnosed with malignancy by both FNA and histopathology, as true negative (TN) when patients did not have malignancy on both FNA and histopathology, as false positive (FP) when malignancy were diagnosed by FNA but not with histopathology and as false negative (FN) when malignancy was diagnosed by histopathology but not FNA.
The non-parametric statistical analysis was used to determine the sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV) and the overall accuracy with their 95% confidence interval (CI) was estimated for FNA based on histopathology results using MEDCALC.org and the following below formulas [12]: Total accuracy = TP + TN/total number of cases Our study was ethically approved by Salmaniya Medical Complex research's committee and verbal consent was obtained from patients.

Results
Of the 87 parotidectomy investigated cases, 30 of them were excluded due to missing FNA findings or for whom FNA was not performed, thus a total of 57 cases left to be examined. Results of demographic data and lesions characteristics are illustrated in than or equal 4 cm in 47 cases (82.5%) and more than 4 cm in ten patients (17.5%). In 43.9% of cases, the superficial lobe was affected, followed by 3.5% in deep lobe and 52.6% for both lobes. Fine needle aspiration diagnoses were obtained using the direct FNA blind method in 75.4% of the cases, where as 17.5% were by using ultrasound guidance. The FNA method used for the remaining cases (7%) was unknown (Table 1).
Of the 57 cases, four cases were excluded due to insufficient or inconclusive specimens. Of the satisfactory sample (N = 53), 47 cases were benign and six cases were malignant based of FNA results (Table 2)  ( Table 3). The true positive (TP) cases for malignancy were 5, while the true negative cases for malignancy were 48. There were no false negative cases for malignancy, but there was one false positive case which was diagnosed as Acinic cell carcinoma in FNA cytology; however, it was diagnosed as pleomorphic adenoma in histopathology (Table 3).
A statistical analysis of the 53 cases was carried out to assess the diagnostic accuracy to detect malignant lesions, specificity, sensitivity, PPV, NPV, and the prevalence of parotid malignant tumors (Table 4). A sensitivity of 100% was recorded. The specificity was 97.9%, and the diagnostic accuracy was 98.1%. The positive and negative predictive values were 83.7 and 100%, respectively. The prevalence of malignancy in our study was 9.4%.

Discussion
Our experience with FNA procedure for detecting malignant parotid tumors revealed that FNA cytology is safe, simple and efficient in the diagnosis of patients  with parotid gland malignancy [6]. When used preoperatively, FNA offers the clinicians with valuable information and helps making a decision regarding patient management and surgical intervention [6] [9]. The primary treatment of malignant parotid gland tumors is mostly surgery, but the extent of surgery depends on the histopathologic type [5]. However, Zbären et al. (2018), believes that for distinguishing malignant versus benign neoplasms, the sensitivity is satisfactory and the accuracy of tumor typing and grading is low [2]. Possibly, histologic type, frequency of uncommon lesions and experience of the cytologist influence these results. He also stated that FNA can be inappropriate to guide the extent of surgery for malignant neoplasms of the parotid gland [2]. In our study, in most of cases, an agreement was seen between FNA and histopathology results with 100% and 87.5% for malignant and benign cases, respectively. However, the analysis of histopathological biopsy is still considered the gold standard for assessing tumors for malignancy [13]. Alternatively, surgical biopsies can lead to significant complications such as fistula formation, facial nerve injury or tumor seeding [1]. They are also prone to sampling errors [5].
This study reports the results of FNAs of parotid masses over a 10-year period at a tertiary care center in SMC. The diagnostic accuracy, sensitivity, specificity, PPV and NPV of preoperative FNA specimens of parotid gland malignant tumors were found to be 98.1%, 100%, 97.9%, 83.7% and 100%, respectively (Table 4), indicating satisfactory results as compared with those previously reported from other institutions ( Table 5). In addition, 100% of the malignant tumors and 87.5% of the benign tumors were classified accurately with regard to tumor type on cytologic evaluation.
Similar studies reported the accuracy of FNA ranging from (80% to 92%), and a sensitivity ranging (46% -92%) as shown in  lower than other reports which probably were due to our small sample size. Limited sample size and inadequate clinical data were reasons of limitations to this study. We therefore recommend further measures to be taken in SMC, to ensure sufficient and accurate data present in future for better patient management.

Conclusion
In conclusion, based on our study, FNA is a highly specific and sensitive tool in diagnosing malignant parotid tumors. Unsatisfactory specimen was the reason behind invalid interpretation. We support its use as an initial diagnostic tool because it's a safe, rapid and minimally invasive approach. However, future studies are required for comparison and to assess factors influencing accuracy of FNA.