Abstract
The risk of distant metastasis of salivary gland cancers has usually been associated with histological type, tumour size, and site. The aim of this study was to evaluate a series of patients with major salivary gland carcinomas in order to identify potential risk factors associated with distant metastasis. 255 patients treated for major salivary gland carcinoma in Brazil from 1953 to 2004 were reviewed. Clinical and treatment data were obtained from the medical records and histological features reviewed. 57 (22%) of 255 patients had distant metastasis. The lungs were the most common metastatic site (40 cases, 65%) and adenoid cystic carcinoma the most frequent histological type involved (27 cases, 47%). The percentage of tumours in the submandibular, parotid, and sublingual glands that presented distant metastasis was 42%, 20%, and 17%, respectively. These results provide evidences that clinicopathological factors (tumour site and histology) are significant predictors of distant metastasis in patients with major salivary gland carcinomas.
Salivary glands carcinomas comprise a morphologically diverse group of rare tumours of unknown cause . These malignant tumours account for 0.4–2.6 cases per 100,000 people in the USA. According to the National Cancer Data Base Report on Cancer of the Head and Neck in the USA, major salivary gland tumours accounted for 6% of all malignant tumours in the head and neck and 0.3% of all malignancies .
These tumours are interesting because of their remarkable variation in clinical presentation, histology, and biological behaviour, showing varied rates of local, regional, and distant recurrence in accordance with tumour site and therapy . The most frequent failures affecting the prognosis involve distant metastases . Data from the literature show that distant metastasis rates in patients with carcinomas of parotid, submandibular, and minor salivary glands are higher than 17%, 37% and 24%, respectively . The histological types more frequently associated with the high risk of metastasis are adenoid cystic carcinoma, salivary duct carcinoma, undifferentiated carcinomas, adenocarcinoma not otherwise specified (NOS), squamous cell carcinoma, high grade mucoepidermoid, and carcinoma ex-pleomorphic adenoma . The anatomical sites in which distant metastasis from salivary gland tumours have been observed more frequently are lungs, bones, brain, liver, and skin .
Adenoid cystic carcinoma is the main histological type characterized by delayed and distant metastasis, but it has not been associated with regional lymph node metastasis . Distant metastasis from adenoid cystic carcinoma and the other histological types can occur irrespective of locoregional control , but regional treatment failure can be a predictive factor for distant metastasis development .
Metastasis is the main cause of human cancer deaths, and its early detection and treatment continues to be a major challenge. The identification of factors associated with distant metastasis is of paramount relevance for therapeutic planning. The aim of this study was to analyse the risk factors associated with distant metastasis in a series of 255 patients with primary carcinomas of the major salivary glands undergoing treated in a single institution and followed-up for at least 5 years.
Materials and methods
A retrospective review of medical records was performed for 255 patients with major salivary gland carcinomas diagnosed and treated from 1953 to 2004 at a hospital in São Paulo, Brazil. This study was approved by the Institutional Ethics Committee (Protocol #1025/08). The eligibility criteria included previously untreated patients, submitted to treatment in the single hospital. The medical records of all patients were examined for demographic data (age, gender, and race), clinicopathological information (clinical stage, invasion of adjacent structure, lymph nodes involvement, histological type, and capsular rupture). TNM restaging was reviewed according to the International Union Against Cancer criteria (UICC, 2002) . Treatment and follow-up data were also recovered.
The statistical associations amongst variables and distant metastasis were analysed using χ 2 and Fisher’s exact test. Overall survival was defined as the interval between the beginning of treatment and the date of death or the last information for censored observations. Overall survival probability was estimated by the Kaplan–Meier method, and the log-rank test was applied to assess the significance of differences amongst actuarial survival curves with a confidence interval of 95%. All analyses were performed using the statistical software package STATA (STATA Corporation, College Station, TX, USA).
Results
The studied population consisted of 255 patients with major salivary gland carcinomas, of which 131 were male (51%) and 124 female (49%), with a mean age of 49.1 years (range 1–91 years). 214 patients (84%) were Caucasian and 41 (16.1%) were non-Caucasian. The mean time of complaints was 31.2 months (range 1–480 months), and the most common symptoms were tumour growth (213 cases, 84%) and pain (91 cases, 36%). Most tumours involved the parotid (218 cases, 86%), followed by the submandibular (31 cases, 12%), and sublingual gland (six cases, 2%). The mean tumour size was 5 cm (range 1–15 cm), with 39 patients (17%) showing invasion of adjacent structures such as skin, bone, nerve, muscle, and adipose tissue.
At diagnosis, 139 patients (57%) presented in an advanced clinical stage with T3 or T4 tumours, 62 cases (25%) had lymph node metastasis, and 14 patients (6%) distant metastasis involving the lungs (nine cases), liver (one case), brain (one case), and other sites (three cases) ( Table 1 ). The final clinical stage showed that 39 cases (16%) were classified as clinical stage I, 48 (20%) stage II, 92 (39%) stage III, and 59 (23%) stage IV ( Table 1 ). The most common histological type was mucoepidermoid followed by adenoid cystic carcinomas corresponding to 31% and 22%, respectively, in patients with major salivary gland carcinomas ( Table 2 ).
| Variable | Category | N (%) |
|---|---|---|
| Tumour size (T) | T1 | 46 (18.9) |
| T2 | 59 (24.2) | |
| T3 | 105 (43) | |
| T4 | 30 (12.3) | |
| TX | 4 (1.6) | |
| Lymph nodes (N) | N+ | 62 (25.4) |
| N0 | 182 (74.6) | |
| Distant metastasis (M) | M+ | 14 (5.7) |
| M0 | 230 (94.3) | |
| Clinical stage (TNM) | I | 39 (16.3) |
| II | 48 (20.2) | |
| III | 92 (38.7) | |
| IV | 59 (24.8) | |
| Variable | Category | N (%) |
|---|---|---|
| Histological type | Mucoepidermoid carcinoma | 80 (31.4) |
| Adenoid cystic carcinoma | 57 (22.4) | |
| Undifferentiade carcinoma | 32 (12.6) | |
| Adenocarcinoma NOS * | 25 (9.8) | |
| Squamous cell carcinoma | 21 (8.2) | |
| Acinic cell carcinoma | 19 (7.5) | |
| Carcinoma ex pleomorphic adenoma | 16 (6.3) | |
| Other | 5 (2) |
Patients were treated mainly by surgery alone (112 cases, 44%) or associated with adjuvant radiotherapy (102 cases, 40%). Other therapeutic modalities for advanced or irresectable tumours, included radiotherapy alone (22 cases, 9%), and combined treatments such as surgery and chemotherapy (five cases, 2%), radiotherapy and chemotherapy (three cases, 1%), or surgery, radiotherapy and chemotherapy (11 cases, 4%). Total or partial parotidectomy was performed in 130 cases (52%) and 53 cases (21%), respectively, and submandibulectomy in 23 cases (9%). 24 patients (9%) were treated with other resection types. Following surgical treatment a residual tumour was found in 29 patients.
107 patients (42%) underwent neck dissection. The majority of these patients (62 cases, 58%) had clinically positive lymph nodes (N+). 45 patients (42%) with neck carcinoma staged as N0 had advanced T stage or tumours of more aggressive histological types (undifferentiated carcinoma, adenocarcinoma NOS, and carcinoma ex-pleomorphic adenoma). Histological analyses confirmed microscopically lymph nodes metastasis (pN+) in 58 patients (54%), with capsular rupture in six patients (10%).
86 patients (34%) experienced tumour recurrences during the follow-up, and 48 of these (56%) had distant metastasis. Of these 48 cases, 35 patients (41%) had only distant metastasis, seven (8%) distant metastasis and local recurrences, four (5%) distant metastasis and regional recurrences, and two (2%) distant metastasis, local, and regional recurrences. The most common sites involved in distant metastasis were lungs (30 cases, 65%), bones (six cases, 13%), brain (two cases, 4%), and liver (two cases, 4%). Only one patient presented distant metastasis involving the liver (2%), bones, and brain (one case, 2%); lungs and bones (one case, 2%); bones and other site not specified (one case, 2%); brain and other site not specified (one case, 2%). Some isolated patients (four cases, 9%) had distant metastasis in infrequent anatomical sites.
Metastasis did not correlate with gender or age (data not shown). Distant metastasis was influenced by the tumour site. Carcinomas in the submandibular gland were more frequently associated with distant metastasis (13 cases, 42%), followed by parotid (43 cases, 20%), and sublingual glands (one case, 17%) ( P = 0.020) ( Table 3 ). Considering histological types, adenoid cyst carcinoma, and adenocarcinoma NOS had more distant metastasis (47% and 20%, respectively) than other histological types ( Table 3 ). Patients with tumours showing invasion of adjacent structures, advanced clinical stage (T3 + T4), and positive lymph nodes had metastasis in 31%, 26%, and 27%, respectively ( Table 4 ). Facial paralysis was observed in 22 patients and 59% of them developed distant metastasis ( P = 0.001) ( Table 4 ). Amongst the 58 patients with microscopic lymph node metastasis (pN+), 33% developed distant metastasis, the figure increasing to 57% when capsular rupture was present ( Table 4 ).
| Variable | Category | Distant metastasis, N (%) | ||
|---|---|---|---|---|
| No | Yes | P value | ||
| Tumour site | Parotid | 175 (80.3) | 43 (19.7) | |
| Submandibular | 18 (58.1) | 13 (41.9) | 0.020 | |
| Sublingual | 5 (83.3) | 1 (16.7) | ||
| Histologic type | Mucoepidermoid carcinoma | 67 (83.7) | 13 (16.3) | NA ** |
| Adenoid cystic carcinoma | 30 (52.6) | 27 (47.4) | ||
| Undifferentiated carcinoma | 26 (81.2) | 6 (18.8) | ||
| Adenocarcinoma NOS * | 20 (80) | 5 (20) | ||
| Squamous cell carcinoma | 20 (95.2) | 1 (4.8) | ||
| Acinic cell carcinoma | 17 (89.5) | 2 (10.5) | ||
| Carcinoma ex pleomorphic adenoma | 13 (81.2) | 3 (18.8) | ||
| Other | 5 (100) | 0 (0) | ||
| Variable | Category | Distant metastasis, N (%) | ||
|---|---|---|---|---|
| No | Yes | P value | ||
| Tumour size | T1 + T2 | 87 (82.9) | 18 (17.1) | 0.103 |
| T3 + T4 | 103 (74.1) | 36 (25.9) | ||
| Lymph nodes | N0 | 145 (79.7) | 37 (20.3) | 0.245 |
| N+ | 45 (72.6) | 17 (27.4) | ||
| Clinical stage | I + II | 72 (82.8) | 15 (17.2) | 0.192 |
| III + IV | 114 (75.5) | 37 (24.5) | ||
| Invasion of adjacent structures | No | 124 (81.5) | 35 (18.5) | 0.085 |
| Yes | 27 (69.2) | 12 (30.8) | ||
| Facial paralysis | No | 165 (85) | 29 (15) | 0.001 |
| Yes | 9 (40.9) | 13 (59.1) | ||
| Pathological lymph nodes | pN− | 40 (81.6) | 9 (18.4) | 0.092 |
| pN+ | 39 (67.2) | 19 (32.8) | ||
| Capsular rupture | No | 36 (70.6) | 15 (29.4) | 0.143 |
| Yes | 3 (42.9) | 4 (57.1) | ||
The mean follow-up time was 100 months, ranging from two days to 544 months. At the end of the follow-up period, 53 patients (21%) were alive without evidence of disease and four patients (2%) were alive with recurrent disease. 95 patients (37%) died of causes related to the disease (range from two days to 544 months) and 51 patients (20%) died of other causes not related to the disease, while 52 (20%) were lost during the follow-up. Overall survival rates were 59% (5 years) and 49% (10 years) ( Fig. 1 A) . Survival analysis revealed that parameters such as tumour size ( P = 0.0001), positive lymph nodes ( P = 0.0001), advanced clinical stage ( P = 0.0001), race ( P = 0.0193), facial paralysis ( P = 0.0001), and invasion of adjacent structures ( P = 0.0043) were statistically correlated with lower overall survival ( Table 5 ). Patients with tumours from the submandibular gland presented lower survival probability than those with tumours from other sites, but this was not statically significant ( Table 5 ).
