A clinicopathological study of parotid carcinoma: 18-year review of 171 patients at a single institution

Parotid carcinoma is a rare tumor that accounts for approximately 0.5% of all carcinomas and represents less than 5% of head and neck cancers [1, 2]. While most oral, pharyngeal, and laryngeal cancers are squamous cell carcinomas, parotid carcinoma varies widely and is classified into 23 histopathological types according to the 2017 classification of the World Health Organization [3]. In addition, a single histopathological type often has several pathological grades. Therefore, treatment planning cannot be based on tumor staging alone, and preoperative diagnosis of both the histopathological type and pathological grade is also important. Although there have been various reports of methods for preoperative histopathological diagnosis and determination of the pathological grade, including fine-needle aspiration cytology (FNA), diagnosis depends largely on the pathologist’s experience and results can be inconsistent among different institutions [4, 5]. Surgery is the first-line therapy for parotid carcinoma, but the incision plan and the criteria for facial nerve preservation have not been standardized. In addition, the indications for postoperative radiation therapy are still under discussion [6, 7].

It is not easy to assess the prognostic factors for parotid carcinoma or the survival rate for several reasons, including the low incidence of this disease with various histopathological types and pathological grades [3], the high prevalence of low-grade carcinoma that needs long-term follow-up [8], and the difficulty in following a particular diagnosis and treatment plan for a long period. In particular, since treatment should be determined according to the pathological grade and histopathological type, accurate results can only be obtained from accumulated data on patients managed according to a consistent plan for diagnosis and treatment. Therefore, it is difficult to find solid evidence based on patient data obtained from multiple institutions.

During the past 18 years, our department has followed an essentially consistent diagnostic and treatment approach for the management of parotid carcinoma. Accordingly, we conducted a retrospective clinicopathological investigation of a total of 171 patients accumulated during that 18-year period.

It has been reported that parotid carcinoma is infrequent and has an incidence of 1.3 per 100,000 persons in the general population [13]. In addition, parotid carcinoma has various pathological grades and histopathological types, thus making it difficult to accumulate and investigate a large series of patients with the disease.

Pathological grade is one of the most important prognostic factors for parotid carcinoma. The correct preoperative diagnosis is necessary for adequate surgical management, so pretreatment diagnosis of the pathological grade is important for deciding the treatment plan. Although the pathological grade can be diagnosed by FNA and FS, the results are not always satisfactory and diagnosis largely depends on the experience of the pathologist. Thus, diagnostic expertise for parotid carcinoma differs among institutions and results in different treatment plans. During the past 18 years, our department has followed a consistent diagnostic and treatment policy. For the reasons mentioned above, we consider that it is significant to investigate a large case series of parotid carcinoma managed at a single center.

Pain is the most important feature that cannot be overlooked among the symptoms of parotid carcinoma. In this series, pain was noted in 52% of all parotid carcinoma patients. Pohar et al. [14] reported that pain was a symptom in 34% of their patients, while Godballe et al. [15] reported pain in 31% of parotid carcinoma patients and stated that pain was a poor prognostic factor. While Stodulski et al. [16] reported that pain was present in 40% of parotid carcinoma patients, they also concluded that it was not a prognostic factor. In this study, classification by pathological grade showed that pain was present in 67% of patients with high-grade disease and even 41% of those with low/intermediate disease. Particularly in patients with low-/intermediate-grade carcinoma, pain should not be overlooked because these tumors cause few other symptoms and often follow a course similar to benign lesions. It has generally been reported that benign tumors do not cause pain, and pain was only noted by 33 out of 668 patients (5%) with benign parotid tumors at our department. Since the frequency of pain was approximately 10 times higher in patients with parotid carcinoma than in those with benign tumors, the symptom of pain is considered to be the first indicator of possible malignancy. Although adhesion to the surrounding tissues is a slightly vague sign due to difficulty in performing quantitative assessment, it is an important feature of malignancy. It was reported that adhesion to the skin was observed in 9% of patients and adhesion to deep tissues in 13–18% of patients [15, 17, 18]. We consider that these findings correspond to “fixation” in our study, and the frequency was similar. It has been reported that the frequency of facial palsy is 9–25%, and a similar frequency of 20% was obtained in our study [17‐19]. Facial palsy is known to be associated with recurrence and a poor prognosis [20, 21], and there was a significant difference of this sign between low-grade and high-grade tumors in the present study.

FNA is the only method for preoperative determination of the histological type and pathological grade, and it has been reported that the diagnostic accuracy of FNA for parotid carcinoma is unfavorable [22‐24]. Although it is desirable to determine both the histological type and pathological grade, we consider that the latter is more important for treatment planning. Out of 163 patients, both the histological type and pathological grade were diagnosed correctly in only 19%, while the pathological grade was diagnosed correctly (histological type unknown) in 15%. Thus, the pathological grade was determined in 34% of the patients, i.e., only 1 out of 3. The diagnosis rate was comparatively favorable in patients with high-grade carcinoma, whereas it was unsatisfactory in patients with low-/intermediate-grade carcinoma [25]. The pathological grade was diagnosed by FS in 72% of the patients and, although this diagnosis rate is considerably higher compared with FNA, the problems to be noted here are that the diagnosis rate was worse for low-grade carcinoma as with FNA [25] and that surgery cannot be explained to patients preoperatively if the pathological grade is only diagnosed correctly by FS after the operation has already started. In histopathological types, non-diagnostic cases were predominant in basal cell carcinoma and epithelial–myoepithelial carcinoma. The correct diagnosis in FS, as well as FNA, was often difficult because these histopathological types have poor cellular atypia. We tried to study the relationship between prognosis and preservation/sacrifice of the facial nerve. However, it was impossible to analyze it because the number of cases was extremely imbalanced. In low-/intermediate-grade cases, only five cases were dead/recurrent among the total 75 cases. On the other hand, in high-grade cases, only four cases were preservation of the facial nerve among the 42 cases. These results suggest that it was impossible to analyze the relationship between prognosis and the facial nerve preservation/sacrifice, even though it was a very important and interesting point. As a result of multivariate analysis, we found that histological grade was a major prognostic factor and that correct preoperative diagnosis of pathological grade contributed to improvement of the surgical planning.

With regard to handling of lymph node metastasis in patients with parotid carcinoma, although most surgeons have no objection to neck dissection for N+ patients, consensus has not been reached on the indications for END in N0 patients [26]. Preoperative diagnosis of lymph node metastasis has limitations, and it is not rare for lymph node metastasis to be found by END (occult metastasis). Armstrong et al. [27] reported that occult metastasis was observed in 38% of patients. It was also found in 20% of patients by Zbaren et al. [28] and in 18% of the patients by Klussmann et al. [29]. In the present study, END was performed in 67 out of 127 patients clinically diagnosed as N0 (cN0), and pathological N+ (pN+) was observed in 8 patients. Among the 58 patients in whom END was planned, secondary lymph node metastasis was found in 2 cases. Accordingly, occult metastasis was detected in a total of 10 patients (8%). We consider that the occult metastasis rate is decreasing with the progress of imaging methods, but occult metastasis may exist in approximately 10% of patients. There have been various reports on the indications for END in patients with parotid carcinoma, and Sinha et al. [30] reported that they did not perform END at all. In contrast, Eneroth et al. [31] conducted END in all patients except for those with low-grade tumors. In addition, Armstrong et al. [27] reported that END should be conducted in patients with T3 or higher disease and patients with high-grade tumors. In the present study, lymph node metastasis was observed in 49 out of 171 patients (29%), and the lymph node metastasis rates were 59% for high-grade disease and 6% for low-/intermediate-grade disease. Based on these results, it may not be necessary to perform END in patients with low-/intermediate-grade tumors. However, it is still a problem that preoperative diagnosis of pathological grade by FNA and FS is not adequate. Even though not conducting END for low-/intermediate-grade tumors seems reasonable, this treatment policy should be based on preoperative diagnosis of the pathological grade. We define the range of END as levels I–III and the upper part of level V. We also preserve the sternocleidomastoid muscle, accessory nerves, and internal jugular vein. Considering that preoperative diagnosis of the pathological grade is difficult and there are limitations on the diagnosis of lymph node metastasis, we conclude that it is appropriate to perform END for all patients with parotid carcinoma since END causes less surgical stress. However, in the present study, among the 41 pN+ patients undergoing neck dissection, positive lymph nodes were found at level II in 74% and at the periparotid region in 62%. Since level II and the periparotid region can be approached by an S-shaped skin incision, another option is for FS of the lymph nodes to be performed first and neck dissection omitted if metastasis proves to be negative.

Salivary gland cancers are histologically similar to certain types of breast cancer [32]. Use of anti-HER2 therapy is currently approved for HER2 overexpressing breast cancer [33], which has led to interest in studying anti-HER2 treatment for salivary gland cancers. According to previous reports, the HER2-positive rate of salivary gland carcinoma is not always high. Glisson et al. [34] screened 137 patients and reported that the HER2-positive rate was 17%. In this study, there were 15 HER2-positive patients out of 107 patients (14%), which was a similar result. Among salivary gland carcinomas, the HER2-positive rate is reported to be higher for SDC [34, 35]. In our study, 5 out of 13 patients with SDC were HER2 positive. The HER2-positive rate is also higher in patients who have CXPA [36], and the positive rate is reported to be higher for invasive tumors. In this study, 6 out of 25 patients with CXPA were HER2 positive, and all 6 patients had invasive tumors with histological predominance of the SDC component. Since the number of HER2-positive patients is higher among those with invasive SDC and CXPA, it was reported that HER2-positive status is associated with a poor prognosis due to rapid progression. In this study, the DSS of HER2-positive patients was significantly worse. It has frequently been reported that the AR-positive rate is higher compared with that of HER2, and Locati et al. [37] found an AR-positive rate of 43%. In this study, the AR-positive rate was only 14% and was lower than in other reports. Similar to HER2, there were more AR-positive patients in the high-grade group, and 12 out of 15 AR-positive patients had high-grade tumors. With regard to the histopathological diagnosis, 5 patients were classified as SDC, while 6 patients had CXPA (3 high-grade tumors) and 3 patients had mucoepidermoid carcinoma (all high-grade tumors). The prognosis of AR-positive patients was poor, similar to that of HER2-positive patients. Since recurrence is frequent in patients with HER2- or AR-positive tumors, it is possible that molecular-targeting therapy is indicated for such disease. Recently, Keller et al. [38] summarized new concepts for personalized therapy in salivary gland carcinoma. According to this paper, there has not been any large-scale study on anti-HER2 and anti-AR therapies. However, we could find some studies stating that anti-HER2 and anti-AR therapies were effective in salivary gland carcinoma in small-scale studies. There have been a few reports showing that trastuzumab (an anti-HER2 antibody) is not effective as a monotherapy, but has some efficacy in combination with other agents [39, 40]. As antiandrogen therapy, Jasper et al. [41] reported that administration of bicalutamide was effective for AR-positive SDC. Since distant metastasis is frequently seen in patients with high-grade tumors, development of effective molecular-targeted therapy is needed.

This is the largest published series of parotid cancer patients from a single center in Japan. Pathological grade was the most important prognostic factor according to multivariate analysis, suggesting that treatment planning for parotid cancer should depend on the pathological grade. However, preoperative diagnosis of the pathological grade by FNA was often incorrect or impossible. The disease-specific 5-year survival rate of patients in stages I to IV was 100, 95.2, 70.4, and 45.1%, respectively. Overexpression of HER2 and AR was observed in high-grade tumors, especially SDC, and was associated with a poor prognosis.