Régis de Brito Santos I, Kowalski LP, Cavalcante de Araujo V, Flávia Logullo A, Magrin J. Multivariate Analysis of Risk Factors for Neck Metastases in Surgically Treated Parotid Carcinomas. Arch Otolaryngol Head Neck Surg. 2001;127(1):56-60. doi:10.1001/archotol.127.1.56
Copyright 2001 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2001
To analyze risk factors for neck metastases in patients with parotid carcinomas.
Cohort of patients followed up from 1 to 366.2 months at a single institution.
Referral center, private or institutional practice, hospitalized care.
A total of 145 patients with parotid carcinomas with complete clinical and pathological information. The histological diagnosis was reviewed according to the World Health Organization classification for salivary gland tumors.
Patients were treated by surgery alone (62 cases) or with postoperative radiotherapy (83 cases). A neck dissection was performed in 80 patients.
Main Outcome Measure
Rates of neck lymph node metastasis. Univariate and multivariate analyses were carried out using logistic regression evaluating the significance of demographic, clinical, and pathological data.
The following variables were significantly associated to the risk of lymph node metastasis by univariate analysis: histological type (P<.001), T stage (P<.001), desmoplasia (P = .001), facial palsy (P = .02), perineural invasion (P = .01), extraparotid tumor extension (P = .02), and necrosis (P = .003). By multivariate analysis, histological type (P<.001), T stage (P = .03), and desmoplasia (P = .006) had the highest correlation with lymph node metastasis.
The significant risk factors for neck metastasis in parotid carcinoma were histological type (ie, adenocarcinoma, undifferentiated carcinoma, high-grade mucoepidermoid carcinoma, squamous cell carcinoma, and salivary duct carcinoma), T stage (T3 and T4), and desmoplasia (severe).
SALIVARY GLAND tumors are rare and represent 0.3% of all malignant neoplasias. The vast majority have epithelial origin.1 Seventy percent of all salivary gland neoplasias are at the parotid gland with a reported percentage of malignancy varying from 17% to 34%. The heterogeneity and great diversity in histological types along with the rarity of malignant subtypes have lead to the individualized and controversial treatment of these patients.2,3 Tumor extent, neck metastases, pain at presentation, aged older than 55 to 60 years, and being male have all been considered poor prognostic indicators,4- 6 as well as the microscopic features of cellular atypia, desmoplasia, and a high mitotic index.7 Patients with acinic cell carcinoma and mucoepidermoid carcinoma have a better prognosis when compared with patients with adenocarcinoma, malignant mixed tumor, adenoid cystic carcinoma, squamous cell carcinoma, and undifferentiated carcinoma.7
Surgery is the main treatment modality and the type of resection, whether partial or total parotidectomy, with preservation or sacrifice of the facial nerve, depends on tumor location, size, and extent.8,9 The incidence of lymph node metastases in parotid carcinomas at the time of initial presentation varies from 12.4% to 24%.1,6,7,10 There is no doubt regarding the indications of neck dissection in clinically positive lymph nodes of the neck. However, the question does remain for patients with stage N0 neck. Indications for elective neck treatment (neck dissection or radiotherapy) are unclear.11 Patients with neck recurrences have a poor prognosis, since treatment in this group is inefficient and relapse is unlikely to be salvaged.
The specific point of interest in this article is to analyze demographic, clinical, and histopathological data to determine which were the N0-staged patients at a high risk for occult metastasis who could potentially benefit from elective neck treatment.
All cases of malignant parotid neoplasias treated at the Department of Head and Neck Surgery and Otorhinolaryngology, Centro de Tratamento e Pesquisa Hospital do Câncer A. C. Camargo, São Paulo, Brazil, from November 30, 1965, to March 21, 1992, were reviewed. One hundred forty-five patients with the diagnosis of primary malignant epithelial tumors of the parotid gland were eligible for this study. Melanomas, lymphomas, sarcomas, and metastatic carcinomas were excluded. We also included previously operated on patients whenever enough demographic, pathological, therapeutic, and clinical information were recovered. The patients were retrospectively staged according to American Joint Committee on Cancer classification (1997 version)12 (Table 1). Hematoxylin-eosin–stained slides were analyzed and histopathological diagnoses reviewed by 2 pathologists (V.C.A. and A.F.L.) adopting the World Health Organization classification for salivary gland tumors.13 Whenever there was disagreement between diagnoses, the slides were analyzed by a third pathologist.
There were 80 male (55.2%) and 65 female patients (44.8%), ranging in age from 2 to 82 years (median age, 53 years). The presence of a mass in the parotid region, the most frequent complaint, was reported for 141 patients (97.2%). Duration of complaint ranged from 1 to 480 months. Regarding tumor staging, 9 (6.3%) were classified as T1; 28 (19.3%), T2; 47 (32.4%), T3; 36 (24.8%), T4; and 25 (17.2%). TX. Clinical lymph node involvement was present in 38 (23.5%) of the 145 patients as follows: 18 (12.4%) N1, 15 (10.4%) N2, 1 (0.7%) N3, and 4 (2.7%)NX.
All patients underwent surgical treatment. Total parotidectomy was the most common resection performed (91 cases [62.8%]), followed by partial parotidectomy in 37 patients (25.5%) and wider resections (total parotidectomy encompassing adjacent structures as bone, skin, and muscle), whenever the tumor extended beyond the parotid confines, in 17 patients (11.7%). Sixty-four patients (44.1%) submitted to radical (classic or modified) unilateral neck dissection. Supraomohyoid neck dissection was performed in 15 patients (10.3%) and bilateral radical neck dissection in 1 (0.7%). Surgical treatment was followed by external beam radiation therapy in 83 cases (57.2%), with a total dosage ranging from 30 Gy to 70 Gy. The lymphatic drainage was included in the radiation portals in 50 cases (34.5%) and only the parotid region in 33 cases (22.8%).
The follow-up period ranged from 1 to 366.2 months (median follow-up, 66.2 months) with 77 (53.1%) of the 145 patients being followed up for 60 months and 45 (31.0%) for at least 120 months. Local recurrence alone was observed in 20 patients (13.8%). Homolateral neck lymph node metastases occurred in 9 patients (6.2%), within a median period of 37.8 months (range, 34-153.3 months), while contralateral neck nodes were the site of recurrence in only 3 patients (2.0%). Distant metastases were found in 17 patients (11.7%) and were associated with local recurrence in 2 patients (1.4%) and with neck metastases in another 2 patients (1.4%).
The following variables were evaluated as candidate risk factors for lymph node metastases in parotid carcinomas: sex, age, race, T stage, facial paralysis, extraglandular tumor extent, histological type, vascular and lymphatic invasion, necrosis, peritumoral lymphocytes, and desmoplasia (inflammatory reaction). Regarding desmoplasia, we designated the following 3 different grades based on the proportion of such histopathological features in the tumor slides reviewed: grade 1, light (14 cases); grade 2, moderate (32 cases); and grade 3, severe (27 cases) when the proportion of tumor composed of desmoplastic stromal reaction was less than 25%, between 25% and 50%, and more than 50%, respectively. Surgical margins were not evaluated in view of the impossibility to get a precise assessment in a retrospective study like this one. Logistic regression was used to perform univariate and mulivariate analysis to build models representing the most parsimonious subset of variables with an independent predictive value for neck metastases.
Mucoepidermoid carcinoma was the most common histological type (20.0%) followed by undifferentiated carcinoma (12.4%), malignant mixed tumor (11.?%), and squamous cell carcinoma (9.7%). Occult metastases were detected in 17 (37.0%) of 46 N0-staged patients who underwent elective neck dissection. Thirteen patients (76.5%) among the 17 with occult neck metastases had T3 or T4 tumors.12 The histological types observed in such cases were mucoepidermoid carcinoma, salivary duct carcinoma, squamous cell carcinoma, undifferentiated carcinoma, malignant mixed tumor, and adenoid cystic carcinoma. Nine patients had ipsilateral neck recurrences within a median period of 39.6 months. Seven of these patients were initially staged as N0 and elective neck dissection was performed in 4 of them, followed by radiotherapy in 2 cases.
The histological types were categorized according to the risk of neck metastases into 3 groups—low, moderate, and high risk, when the percentage of lymph node involvement by histological type reached, respectively, 0% to 20%, 21% to 50%, and more than 50%. Histological types with a total number of cases equal to or smaller than 6 were classified as "others" (Table 2).
The risk for neck metastases was significantly associated with the following variables by univariate analysis: histological type (P<.001), T stage (P<.001), desmoplasia (P = .001), necrosis (P = .003), facial paralysis (P = .02), perineural infiltration (P = .01), and extraparotid tumor extent (P = .02) (Table 3). There was a trend toward lymph node metastases in male patients that, however, failed to achieve statistical significance. Multivariate analysis identified the following as independent predictors for neck involvement: the histopathological type (P<.001), T stage (P = .033), and desmoplasia (P = .006) (Table 4).
The treatment of the neck in patients with parotid carcinoma is indicated whenever there is clinical evidence of lymph node metastases. This is the only situation in which there is no discussion on the therapeutic approach—usually a radical neck dissection followed by postoperative external beam radiation.9 The approach of the N0 neck, however, remains controversial; there is no consensus regarding the need for elective neck dissection or elective radiotherapy.11 The addition of neck dissection, to the surgical treatment increases the time required for the surgical procedure, the risk of postoperative complications, treatment costs, and aesthetic and functional sequelae. Therefore, it should be avoided in N0-staged patients, whenever possible. In this series, 25 (31.3%) of 80 patients who underwent neck dissection had pN0-staged stage. However, clinically N0-staged patients, with occult metastases who did not undergo treatment of the neck, may evolve with detrimental outcome, and salvage treatment is frequently inefficient in such situations.14
The overall incidence of lymph node metastases in our experience was 23.5% and the main histological type was the mucoepidermoid carcinoma (20%), in accord with the data published in the literature.1,6,7,10,15,16 The reported percentage of neck metastases at the time of patients' admission for undifferentiated carcinoma, squamous cell carcinoma, adenocarcinoma, mucoepidermoid carcinoma, and malignant mixed tumor is more than 20%.9,17,18 In this series, we observed that most of these histological types belong to the high-risk group for neck spread, ie, undifferentiated carcinoma, high-grade mucoepidemoid carcinoma, salivary duct carcinoma, adenocarcinoma, and squamous cell carcinoma. Metastasis in patients with those histological types were seen in 68%.
The question about N0-staged patients is: what is the exact percentage of such patients with occult metastases? Or how many patients will have recurrences in the neck? Are there indications for elective neck dissection in parotid carcinoma? Several authors have different answers to these questions, nevertheless, the problem still remains. Fu et al19 observed the preponderance of undifferentiated carcinomas (3 of 4 patients) in cases of neck recurrence. All patients had T3 tumors. Frankenthaler et al10 by multivariate analysis showed that extraglandular tumor extent, being older than 54 years, and lymphatic invasion were significantly associated with the risk for neck metastasis. In their series, occult metastases were observed in cases of adenocarcinoma, salivary duct carcinoma, squamous cell carcinoma, undifferentiated carcinoma, and mucoepidermoid carcinoma. Kelley and Spiro2 consider clinical stage and tumor grading as the best predictive factors. Armstrong et al9 have also shown that patients with high-grade tumors, advanced T stage, and undifferentiated carcinoma (6 of 7 patients), squamous cell carcinoma (6 of 28 patients), adenocarcinoma (11 of 49 patients), and mucoepidermoid carcinoma (30 of 209 patients) present high-risk for neck metastases and should be considered candidates to an elective neck dissection. Spiro et al6 reported metastases in 40% and 16% of squamous cell carcinoma and high-grade mucoepidermoid carcinoma, respectively. The authors suggest elective neck dissection to be performed in patients with high-grade tumors. Spiro et al6 stated that at least 58% of the patients with undifferentiated or squamous cell carcinomas will have neck metastases and could benefit from elective neck dissection. In their series, about 50% of the neck metastases were detected by this procedure. Califano et al20 found occult lymph node metastases in 17% of the patients with mucoepidermoid carcinoma and suggested benefit effects of elective neck dissection in such cases, as well as in undifferentiated and squamous cell carcinomas. In a follow-up period of 4 years, Poulsen et al16 registered neck metastases in 24.4% of 237 parotid carcinomas, showing that patients older than 60 years, with clinically positive lymph nodes at the time of first presentation and with positive surgical margins, squamous cell carcinoma, and poorly differentiated tumors, have a higher percentage of neck metastases.
High-risk histological types were observed in 13 (76.5%) of 17 patients with metastases detected by elective neck dissection in this series, and 13 (76.5%) of those cases had T3 or T4 tumors. The association of high-risk histological type and T3 or T4 was registered for 9 patients. These figures probably underestimate the actual percentage of T3 or T4 tumors in this subset of patients because T stage was undetermined (TX) in 4 of the 13 patients who had high-risk histological types with occult metastases. Neck recurrences were diagnosed in 7 patients (4.8%). The advanced T stage (T3 or T4) and the high-risk histological types were observed in 5 cases (71.4%) and 4 of them were dead of disease at the end of this study, demonstrating the remarkable and detrimental prognostic influence of neck recurrence.
Considering that the identification of risk factors for lymph node metastases, including preoperative and pathological, could be significant to indicate elective neck dissection or elective postoperative radiotherapy, with possible prognostic significance, we decided to undertake this study. Frankenthaler et al,10 analyzing 11 preoperative and postoperative variables, found patient's age, extraparotid tumor extent, and perilymphatic invasion to be the most significant predictive factors for occult lymph node metastases by multivariate analysis. In our series, histological type, T stage, facial palsy, extraglandular tumor extent, perineural invasion, necrosis, and desmoplasia reached statistical significance in univariate analysis. Of the several candidate prognostic factors, multivariate analysis revealed histological type, T stage, and severe desmoplasia as the most significant independent predictors of the risk of neck involvement.
Lewis et al3 studied a group of 90 patients with acinic cell carcinoma of the parotid gland to determine clinical and histological features that consistently predicted disease progression. The presence of desmoplastic reaction (desmoplasia) was 1 of 3 microscopic features significantly correlated with poor outcome (P<.01). In different tumors, parenchymal cells stimulate the formation of an abundant collagenous stroma, characterized by the growth of fibrous tissue, refered to as desmoplasia,21 which is a fairly common finding in parotid carcinoma. We tried, however, to estabilish not only a quantitative but also a qualitative relationship between this finding and the risk for lymph node metastases, estimating 3 different grades based on the proportion of the histopathological feature in the tumor slides that were reviewed. As we have shown, only severe desmoplasia was significantly associated with the risk for lymph node metastases (P = .001).
Unfortunately, some of this information may be unavailable at the time of surgical treatment, restricting its applicability to decide on elective neck dissection or irradiation following primary tumor ressection. Tumor stage is easily known preoperatively, and histological type is generally known by intrasurgical evaluation of a tumor specimen (frozen section). Both types of information can be incorporated in the surgical treatment plan. This is not the case for desmoplasia. However, all of this information is valuable to decide whether postoperative radiotherapy should include the lymphatic drainage area because there is no proven benefit of elective neck dissection over radiotherapyin treating parotid carcinoma.11
Elective neck dissection is considered in the management of superior aerodigestive squamous cell carcinomas whenever the percentage of occult lymph node metastases is between 15% and 20%.22 These figures are not yet established for parotid carcinomas. Our findings showed occult metastases in 22.2% of the subset of patients with T3 or T4 and a high-risk histological type.
Gallo et al,23 studying the expression of p53 in parotid tumors, observed the association of neck metastases and decreased survival in patients with high expression of this suppressor gene. However, as stated by Kelly and Spiro: "the search for factors at the subcellular level holds much promise for the future, but as of this writing, the selection of patients for elective lymphadenectomy involves a judgment based on the relevant clinical findings."2(p697) Our experience suggests that in the management of parotid carcinomas, patients with high-risk histological types (ie, adenocarcinoma, undifferentiated carcinoma, salivary duct carcinoma, squamous cell carcinoma, and high-grade mucoepidermoid carcinoma) and advanced T stage (T3 or T4) could benefit from elective neck treatment. In this series, there was neck metastases in 100% of the patients with tumors at advanced T stage, severe desmoplasia, and high-risk histological types.
Results of retrospective published series regarding the risk factors for cervical metastases in parotid carcinoma varies widely. Prospective randomized trials, however, are virtually infeasible as this is a rare tumor and long-term follow-up is required.
Accepted for publication June 29, 2000.
Reprints: Luiz P. Kowalski, MD, PhD, Department of Head and Neck Surgery and Otorhinolaryngology, Centro de Tratamento e Pesquisa Hospital do Câncer A. C. Camargo, R. Professor Antonio Prudente, 211, 01509-010–São Paulo-SP, Brazil (e-mail: firstname.lastname@example.org).