Multivariate survival analysis among 173 patients with mucoepidermoid carcinoma. Overall survival curves for: A, patient age; B, tumor mobility; C, T stage; D, N stage; and E, histological grade. Each plus sign represents 1 patient.
Pires FR, Paes de Almeida O, Cavalcanti de Araújo V, Kowalski LP. Prognostic Factors in Head and Neck Mucoepidermoid Carcinoma. Arch Otolaryngol Head Neck Surg. 2004;130(2):174-180. doi:10.1001/archotol.130.2.174
To analyze clinical, histological, and immunohistochemical prognostic factors in a large series of patients with mucoepidermoid carcinoma (MEC) treated in a single institution, using univariate and multivariate survival analyses.
All patients diagnosed with head and neck MEC from a single cancer referral center from January 19, 1957, to July 12, 1997.
Main Outcome Measures
Rates of local recurrence, regional and distant metastasis, and overall actuarial survival.
Men represented 53.8% of the cohort, and the parotid gland and palate were affected by MEC in 35.2% and 23.7%, respectively. TNM stage I or II lesions comprised 50.3% of the tumors, and low-grade tumors comprised 45.2%, and the 5-year overall survival was 70.2%. Univariate survival analysis revealed that age older than 40 years (P<.001), male sex (P = .005), fixed tumors (P = .002), invasion of adjacent structures (P = .004), T stage (P<.001), N stage (P<.001), clinical stage (P<.001), histological grade (P<.001), and expression of proliferating cell nuclear antigen (P<.001), Ki-67 (P<.001), and p53 (P<.001) correlated with a poor prognosis. Expression of carcinoembryonic antigen (P = .01) and bcl-2 (P<.001) correlated with a better prognosis.
Age older than 40 years, fixed tumors, T and N stage, and histological grade are independent significant prognostic factors in patients with MEC.
Mucoepidermoid carcinoma (MEC) is the most common malignant salivary gland tumor, accounting for about 3% to 15% of all salivary gland tumors and 12% to 40% of malignancies.1- 5 Histologically, MEC has been classified as low, intermediate, or high grade, according to its cytological features, pattern of invasion, and cellular type.6,7 Several clinical and histological prognostic factors for MEC have been considered, but the most consistent ones are clinical stage and histological grade of the tumor.1- 10 Expression of some immunohistochemical markers is also a prognostic factor in patients with MEC10- 17; however, few studies10,11,16,17 have considered the effect of clinical, histological, and immunohistochemical variables in a large series of MECs from different sites.
The aim of this study was to analyze clinical, histological, and immunohistochemical prognostic factors in a large series of patients with MEC treated in a single institution during 40 years.
The medical charts of A. C. Camargo Cancer Hospital were reviewed, and all medical records of patients diagnosed as having MEC from January 19, 1957, to July 12, 1997, were retrieved for this study. The project was approved by the appropriate institutional review board of the hospital. All cases were histologically reviewed using hematoxylin and eosin, periodic acid–Schiff, and mucicarmine stainings to confirm the diagnosis. Clinical, epidemiological, treatment, and follow-up data were obtained from the medical records, and the tumors were graded, according to the histological criteria of Ellis and Auclair,7 as low-, intermediate-, or high-grade lesions.
Immunohistochemical reactions against proliferating cell nuclear antigen (PCNA) (clone PC10; dilution, 1:16 000), Ki-67 antigen (clone Ki-S5; dilution, 1:200), p53 protein (clone DO-7; dilution, 1:200), c-erbB-2 protein (dilution, 1:200), carcinoembryonic antigen (CEA) (clone II-7; dilution, 1:500), and bcl-2 protein (clone 124; dilution, 1:50) (all DakoCytomation Norden A/S, Glostrup, Denmark) were performed in 3-µm histological sections. Microwave antigen retrieval was performed using citrate buffer, overnight incubation with the primary antibodies, and secondary antibodies conjugated to a streptavidin-biotin-peroxidase system (strept ABComplex/HRP duet, mouse/rabbit; DakoCytomation Norden A/S), followed by application of diaminobenzidine as the chromogen. Slides were counterstained with Carazzi hematoxylin, mounted, and analyzed by one of us (F.R.P.). Positive and negative control slides were included in all reactions. The percentage of cells expressing each of the markers, after analyzing 10 different high-power fields in each case, was classified using the following values: PCNA (negative, <1% to 5% positive cells; weak, 6%-20% positive cells; or strong, >21% positive cells) and Ki-67, p53, c-erbB-2, CEA, and bcl-2 (negative, <1% to 5% positive cells; or positive, >5% positive cells).
The Kaplan-Meier method and log-rank test were used for univariate survival analysis. Overall survival was calculated from the date of the first treatment (surgery or radiotherapy) or first consultation until the date of the last follow-up information. The data were considered uncensored when death was the outcome, independent of its cause. For multivariate survival analysis, all variables were analyzed using the Cox proportional hazards regression model.
One hundred seventy-three patients were identified, with a male-female ratio of 1.2:1 (93 men [53.8%]). The mean age of the patients was 44 years (range, 6-96 years), which was older among men (mean, 47.3 years [range, 6-79 years]) than women (mean, 40.0 years [range, 6-96 years]). Most patients (55.5%) were in their fourth to sixth decades of life. The mean duration of complaints was 38 months (range, 1-480 months), and the most commonly reported symptoms were swelling in the tumor area (93.6%), pain (39.9%), and paresthesia (6.4%).
Eighty patients (46.2%) had tumors affecting the major salivary glands: 61 parotid, 17 submandibular, and 2 sublingual. Among the 93 patients (53.8%) with tumors affecting the minor salivary glands, 75 were affected at intraoral minor salivary gland sites and 18 at extraoral sites (Table 1). The tumor size varied according to the site of the lesions and was largest in the extraoral minor salivary glands (mean, 5.5 cm [range, 2-9 cm]), followed by the major salivary glands (mean, 4.6 cm [range, 1-16 cm]) and the intraoral minor salivary glands (mean, 3.6 cm [range, 1-8 cm]). Mobility data revealed 74.3% fixed or partially fixed tumors, and invasion of adjacent structures was present in 70.5%. Eighty-seven patients (50.3%) had stage T1 or T2 disease; 31 patients (17.9%) had regional metastasis, and 3 patients (1.7%) had distant metastasis at the first evaluation. Final TNM staging revealed that 50.3% of the patients had stage I or stage II disease (Table 2).
Histologically, 73 patients (42.2%) had an intracystic component at a depth greater than 20% of the tumor, 15 (8.7%) had perineural invasion, 38 (22.0%) had necrosis, 31 (17.9%) had 4 or more mitoses in 10 high-power fields, and 87 (50.3%) had anaplasia. Sixty-one patients (35.3%) had tumors classified as low grade, 25 (14.5%) were intermediate grade, and 49 (28.3%) were high grade. Thirty-eight tumors (22.0%) were not submitted for histological grading because of the limited amount of available tissue. Low-, intermediate-, and high-grade tumors represented 45.2%, 18.5%, and 36.3%, respectively, of the 135 graded tumors.
Immunohistochemical reactions revealed that PCNA was expressed by 92.9% of the tumors (weak in 45.7% and strong in 47.2%). Expression of Ki-67, p53, c-erbB-2, CEA, and bcl-2 was found in 34.1%, 16.4%, 80.0%, 68.6%, and 63.2%, respectively.
Previous oncologic treatment was reported in 45 patients (26.0%), and surgical procedures were performed in the A. C. Camargo Hospital in 151 (87.3%) patients, including local resection in 139 (92.1%) and neck dissection in 52 (34.4%). Radiotherapy and chemotherapy were used in 72 (47.7%) patients and 10 (6.6%) patients, respectively, and 22 (12.7%) of 173 patients did not undergo any treatment in the hospital. The most common treatments were local resection alone (36.4%); local resection and postoperative radiotherapy (20.5%); local resection, neck dissection, and postoperative radiotherapy (17.9%); and local resection and neck dissection (13.9%). Among the 52 patients who underwent neck dissection, 22 (42.3%) had metastatic lymph nodes on histological evaluation. Surgical removal of the tumors in 69 (49.6%) of 139 patients was associated with resection of adjacent structures, such as maxillary bones, skin, mandible, and nerve trunks.
Twenty-two patients (12.7%) had local recurrence at a mean interval of 41.7 months (range, 2-207 months) after treatment. Metastases were regional in 17 (9.8%) of 173 patients (mean interval, 29.3 months [range, 3-77 months]), homolateral in 10 (5.8%), contralateral in 5 (2.9%), and bilateral in 2 (1.2%). Distant metastases affected 16 patients (9.2%) (mean interval, 66.1 months [range, 3-370 months]), most frequently in the lungs (12 patients [6.9%]) and bones (5 patients [2.9%]).
The mean follow-up was 98 months (range, 1-389 months) and revealed 83 (48.0%) patients alive with no evidence of disease, 6 (3.5%) alive with disease, 45 (26.0%) who had died of the disease, 16 (9.2%) who had died during treatment or of other causes, and 23 (13.3%) unavailable for follow-up. Overall survival rates were 70.2% and 60.1% for 5 and 10 years, respectively.
Univariate survival analysis of clinical and epidemiological variables revealed that age older than 40 years (P<.001), being male (P = .005), decreased tumor mobility (P = .002), invasion of adjacent structures (P = .004), T stage (P<.001), N stage (P<.001), and clinical stage (P<.001) statistically correlated with a poor prognosis (Table 3). Patients with parotid and submandibular tumors had the best and worst survival rates, respectively. Histological grade (P<.001) also correlated with a poor prognosis (Table 4). An analysis of treatment variables showed that previous oncologic treatment (P = .99) and radiotherapy (P = .19) were not prognostic factors, but patients who underwent surgery (P = .004) and neck dissection specifically (P = .007) had a better prognosis. Increased expression of PCNA (P<.001), Ki-67 (P<.001), and p53 (P<.001) also correlated with a poor prognosis; in contrast, higher expression of CEA (P = .01) and bcl-2 (P<.001) correlated with a better prognosis (Table 5). Multivariate survival analysis revealed that age older than 40 years, fixed tumors, T stage, N stage, and histological grade were independent significant prognostic factors (Table 6 and Figure 1).
Mucoepidermoid carcinoma has a predilection for adults in their fourth to sixth decades of life, without sex predilection.3,8- 11,17- 20 Our MEC patients generally presented with long-standing swelling; about 40% of them were symptomatic, which is similar to reports in the literature.1- 4,8,9,20 Mucoepidermoid carcinoma usually affects the major salivary and intraoral minor salivary glands, but can arise in other head and neck glands in about 10% of patients, including the maxillary sinus, nasopharynx, nasal cavity, oropharynx, vocal cords, larynx, and trachea.1,4,5
The parotid gland was most often affected in our patients, followed by the palate and submandibular gland; these results are generally in accord with the literature, although the intraoral minor salivary glands were the most common site in most reports.1- 5,10,11,17,20 In our patients, MECs affecting the extraoral minor salivary glands were larger than those affecting the major salivary and intraoral minor salivary glands. This was probably related to the presence of maxillary sinus MEC in the former group (38.8% of patients), which can reach considerable size before being noticed.
Despite 50.3% of our patients being diagnosed as having stage T1 and T2 MEC, 74.3% of the tumors showed at least partial fixation to adjacent tissues, and 70.5% had invaded 1 or more adjacent structures, explaining the poor prognosis of some patients. Minor salivary gland tumors were of more advanced TNM stage than major salivary gland tumors, and the frequencies of T1 and T2 tumors and N and M percentages among our patients were similar to those described in the literature.1,3,10
Mucoepidermoid carcinoma is usually classified as low-(15%-62% of patients), intermediate-(9%-48%), or high-grade (22%-38%) malignancy.1,2,4- 7,9,10,14,17,18,20- 22 Histological grade is based on a few easily observable variables present in most patients and may be a useful prognostic indicator.7,9,20,22 However, as with other grading systems, there are limitations regarding reproducibility of some subjective features, such as anaplasia,20 and Brandwein et al4 noted that these classifications tend to underestimate the histological grade of some MECs. Typical low- and high-grade MECs are usually easily classified, but intermediate-grade tumors can have low- and high-grade features. Although intermediate-grade MEC shares more histological features with low-grade tumors, it may frequently manifest positive margins and local recurrence, similar to high-grade lesions.9,10,18 This intermediate group comprises patients with good and poor prognoses and needs to be better characterized.18,20
A wide surgical approach, complemented by neck dissection and radiotherapy, is the treatment of choice for MEC.1,3,4,8,9,14,19 Resection of adjacent structures is indicated when there is previous or intraoperative evidence of invasion; this was performed in about half of our patients managed by surgery. Neck dissection is indicated when there is clinical evidence of regional metastasis, high TNM stage, high histological grade, or proximity of tumor to regional lymph nodes.1,4,18 Half of our patients who underwent neck dissection had metastatic tumoral tissue in the lymph nodes on histological analysis. Postsurgical radiotherapy is indicated in patients with clinical or histologically proven positive margins and in those with high-grade tumors.4,18
Local recurrence, regional metastasis, and distant metastasis are found in 7% to 26%, 3% to 16%, and 6% to 15%, respectively, of MEC patients.1- 4,20 Our 5- and 10-year overall survival rates of 70.2% and 60.1%, respectively, are in accord with those in the literature, which range from 57% to 92% for 5 years and from 28% to 90% for 10 years,2,10,14,19 reinforcing the good prognosis of MEC. Malignant salivary gland tumors are characterized by late recurrence and metastasis, so patients should be followed up for long periods after treatment.18
Univariate and multivariate prognostic studies of patients with MEC have been reported in the literature, and most have included clinical and histological variables. Age is an important prognostic factor, but most tumors in younger patients (in their first and second decades of life) are low-grade lesions.2,16,19,21 Among our patients, age older than 40 years was a significant prognostic factor in univariate and multivariate survival analyses. Similarly, male sex was an important prognostic factor in univariate analysis, but its prognostic significance is limited by the fact that a greater percentage of men than women have high-grade tumors.2,21
The mean tumor size was smaller in women and younger patients, contributing to their better overall survival.1 Size (T stage) and mobility of the tumor, invasion of adjacent structures, and regional metastasis are important prognostic factors in MEC. Among the major salivary glands, submandibular tumors appear to have the worst prognosis,1,8,20 although this is not universally accepted.4 Our data showed that patients with submandibular gland tumors had the worst prognosis, and some authors suggest that submandibular tumors should receive radical treatment, independent of their histological grade.9 Although the presence of symptoms, such as pain and paresthesia, was not related to a poor prognosis among our patients, it has been reported to be an important prognostic factor.1,8,9,20
Histological grade is one of the most important prognostic factors in MEC, and overall 5-year survival rates vary from 92% to 100% in low-grade tumors, 62% to 92% in intermediate-grade tumors, and 0% to 43% in high-grade tumors.1,2,8,17,19,21,22 Our study confirmed that the variables suggested by Goode et al9 and Auclair et al20 are important prognostic factors in MEC, supporting their use as components of grading systems. Our results are in agreement with other reports that most MECs are low- and intermediate-grade lesions, contributing to their overall favorable prognosis.1,4,10,14,18,21 In addition to T stage and histological grade, our multivariate survival analysis confirmed that age older than 40 years, N stage, and fixed tumors were also important independent clinical prognostic factors in MEC.8,10
Proliferating cell nuclear antigen and Ki-67 expression are associated with aggressive malignant behavior, decreasing survival rates, and poor prognosis in patients with MEC,10,11,15,16 and their expression in intermediate-grade lesions may be predictive of a poor prognosis.10,11,15- 17 Proliferating cell nuclear antigen expressions and Ki-67 expression were significant prognostic factors in patients with MEC by univariate analysis. Expression of p53 in MEC occurs in 8% to 80% of tumors, and some authors associate it with higher histological grade, increased tumor size, presence of regional metastasis, and, consequently, a poor prognosis.13,17 Expression of p53 was found in 16.4% of our patients and correlated with a poor prognosis. Expression of c-erbB-2 varies in MEC, ranging from 0% to 24% of tumors.13 Increased c-erbB-2 expression in MEC may be an important prognostic factor,14 but this was not confirmed by our results. Conversely, expression of bcl-2 by MEC cells is more common in low-grade tumors and correlates with less tumor aggression, higher survival rates, and a better prognosis,17 in accord with our results. Although expression of bcl-2 can assist MEC cells in avoiding the apoptotic pathway, other mechanisms probably control cell proliferation in these tumors. An association also seems to exist between CEA expression and high proliferative activity in pleomorphic adenoma and its malignant transformation,12 although its expression in our MEC patients correlated with low-grade tumors and, consequently, a better prognosis.
Clinical stage and histological grade are the main prognostic factors in MEC.1- 5,7,8,10 Nevertheless, our univariate and multivariate analyses showed that other clinical, histological, and immunohistochemical prognostic factors are independent significant indicators. They should be further evaluated by other comparative studies in large series of patients with MEC.
Corresponding author: Luiz Paulo Kowalski, PhD, Department of Head and Neck Surgery and Otorhinolaryngology, A. C. Camargo Cancer Hospital, Rua Prof Antônio Prudente, 211 Liberdade, CEP 01509-900 São Paulo, Brazil (e-mail: firstname.lastname@example.org).
Submitted for publication March 11, 2003; accepted April 25, 2003.
This study was supported by Fundação de Amparo à Pesquisa do Estado de São Paulo, São Paulo, Brazil.