Kaplan-Meier survival functions for patients with and without positive nodal disease in major salivary gland malignancy.
Bhattacharyya N, Fried MP. Nodal Metastasis in Major Salivary Gland CancerPredictive Factors and Effects on Survival. Arch Otolaryngol Head Neck Surg. 2002;128(8):904-908. doi:10.1001/archotol.128.8.904
To determine how regional nodal metastasis affects survival in patients with major salivary gland malignancy and to identify clinical predictors for nodal disease.
Major salivary gland cancer cases with nodal sampling were identified from the Surveillance, Epidemiology, and End Results cancer database for 1988 through 1998. Kaplan-Meier survival analysis was conducted to compare patients with and without histopathologic evidence of nodal disease. Multivariate logistic regression analysis was used to determine the influence of clinical predictors on the presence of regional nodal disease.
A total of 1268 patients with major salivary gland malignancy and regional node sampling were identified. Mean age at diagnosis was 58.3 years, with a male-female ratio of 1:4. Mean tumor size was 3.0 cm. Overall mean survival time was 83 months (95% confidence interval, 80-87 months). Patients with no evidence of nodal cancer had significantly improved survival over patients with any pathologically positive nodes (mean survival time, 100 months vs 59 months, respectively; P<.001). Patient age, tumor histopathologic type, facial nerve involvement, extraglandular involvement, tumor grade, and tumor size were significant clinical predictors of nodal disease. Facial nerve involvement, tumor grade, and squamous cell carcinoma subtype exhibited the highest increased odds ratios for nodal metastasis.
Nodal disease significantly decreases survival in patients with major salivary gland malignancy. Tumor histopathologic type, facial nerve involvement, extraglandular tumor extension, and tumor grade are the most important predictors of nodal disease.
MALIGNANT TUMORS of the salivary glands constitute approximately 1% to 3% of all head and neck malignancies and only 0.3% of all malignant neoplasms.1 Because of their relative rarity, it is often hard to quantify survival and determine prognostic factors for these tumors. Most published studies reflect institutional experiences over the course of 2 or 3 decades, with relatively small overall sample sizes.
It is well known that survival in head and neck cancer, especially for mucosal squamous cell carcinoma, is strongly dependent on the disease status of the neck. In several series, the finding of positive neck nodes was associated with a decrease in survival ranging from 30% to 50% compared with patients with no evidence of metastatic neck disease.2- 4 However, for major salivary gland malignancy, management of the neck has received considerably less attention. Some authors recommend routine neck dissection for known salivary gland malignancy, whereas others recommend neck dissection only for obvious nodal disease or for certain tumor histopathologic types.
We sought to determine the effect of nodal metastasis on survival in patients with major salivary gland malignancy. Furthermore, we sought to determine clinical factors that would predict the presence or absence of nodal metastasis from major salivary gland cancer. Identification of such predictive clinical factors could assist in selecting appropriate patients for planned neck dissection in the setting of major salivary gland cancer.
For our analysis, we used the Surveillance, Epidemiology, and End Results (SEER) Program Public-Use CD-ROM (1973-1998), National Cancer Institute, Bethesda, Md, Division of Cancer Control and Population Sciences, Cancer Surveillance Research Program, Cancer Statistics Branch, released April 2001, based on the August 2000 submission. We searched the 1988-1998 period to identify all patients with malignancies of the major salivary glands (parotid gland, submandibular gland, and/or sublingual gland). From this data set, patients who had evidence of at least 1 regional node sampled as part of their evaluation and management were identified. The data were then imported into the Statistical Package for the Social Sciences version 10.0 (SPSS Inc, Chicago, Ill) for subsequent processing. Kaplan-Meier survival analysis was used to compare patients in whom no positive nodes were identified among the sampled nodes and patients in whom at least 1 positive node was identified. Survival curves were compared using the log-rank statistic with P<.05 considered significant.
The tumors were categorized into 1 of 8 histologic categories based on the International Classification of Diseases for Oncology5 as summarized in Table 1. Frequency and descriptive summary data were computed for each of the potential clinical variables. The following were chosen as predictor variables for the outcome of at least 1 positive regional node: patient age, patient sex, tumor grade, tumor histopathologic type, tumor size, facial nerve involvement, and extraglandular tumor involvement. These variables were selected as predictor variables because they can often be preoperatively determined from the physical examination, preoperative radiologic studies, and cytopathologic analysis and can therefore be used to predict the extent of surgery. To determine the effects of these predictor variables on regional nodal disease, we conducted multivariate logistic regression analysis using a backward stepwise likelihood ratio method with threshold P values of .10 for variable entry and .05 for variable exclusion. Odds ratios and associated confidence intervals (CIs) were computed for the statistically significant predictor variables.
From the SEER database for the 10-year period 1988 through 1998, 1268 patients were identified with malignancy of the major salivary glands and at least 1 regional node sampled. There were 535 women (42.2%) and 733 men (57.8%), with a mean age at diagnosis of 58.3 years (SD, 17.9 years). The mean tumor size at the time of diagnosis was 3.0 cm (SD, 1.9 cm). Patients had a mean of 10.7 nodes sampled, with a mean of 2.1 nodes pathologically positive for tumor metastasis. Overall, 748 patients (59.0%) had nodes sampled, but no positive nodes were found on subsequent pathologic analysis. The mean Kaplan-Meier survival for the overall group was 83 months (95% CI, 80-87 months).
Among patients with malignancy of the major salivary glands, the presence or absence of pathologically positive nodal disease was a significant predictor of survival. Patients with no pathologic evidence of nodal disease exhibited a mean survival of 100 months (95% CI, 96-104 months), whereas patients with nodal disease exhibited a mean survival of 59 months (95% CI, 54-64 months; P<.001). These data are graphically depicted in Figure 1. Summaries of the distributions of histopathologic tumor type and histopathologic grade are given in Table 1 and Table 2. Among squamous cell carcinomas, 12% were well differentiated, 29% were moderately differentiated, 45% were poorly differentiated or undifferentiated, and the remainder were not graded.
Logistic regression analysis was conducted to determine predictive factors associated with positive nodal disease in this patient population. The results of the logistic regression analysis are given in Table 3. Age at diagnosis, histopathologic tumor type, facial nerve involvement, extraglandular involvement, tumor grade, and tumor size were each found to be significant independent predictors for positive nodal disease. Specifically, adenocarcinoma and squamous cell carcinoma had significantly increased odds ratios for metastatic nodal disease (increase in odds of 2.0 and 2.2, respectively). Increasing tumor grade, facial nerve involvement, and extraglandular extension also strongly predicted nodal metastasis. Increasing age at diagnosis and tumor size were statistically significant but numerically weak predictors based on their odds ratios. Patient sex showed no effect on the presence of nodal disease.
Malignant tumors of the major salivary glands are fairly rare, even among head and neck lesions. Given their relative infrequency, it is difficult to determine predictors of survival for patients with major salivary gland malignancy in single-institutional patient series. The SEER database is a national tumor registry maintained by the National Cancer Institute that samples several US geographic areas and represents an estimated 10% of the US population. Through active and passive follow-up mechanisms, the database provides high accuracy and rigorous follow-up on several million cancers. Therefore, the SEER registry serves as an excellent source for the study of relatively rare tumors such as major salivary gland malignancies. This registry has been used to examine incidences, trends, and survival for several head and neck neoplastic conditions.6- 8
Treatment for major salivary gland malignancy in most instances includes surgery with or without postoperative radiation therapy, depending on the histopathologic type and grade of the original tumor, status of the surgical margins, and the presence or absence of regional disease. Based on the Kaplan-Meier analysis of the survival difference between patients with no pathologically positive regional adenopathy and patients with positive regional adenopathy, it is clear that the presence of positive nodal disease has a significant influence on overall survival in major salivary gland malignancy, regardless of subsequent additional therapy. The presence of positive nodal disease decreased mean survival in patients with major salivary gland malignancy by more than 50%. Given this effect, the ability to predict nodal involvement and appropriately treat patients with positive regional nodal disease could have a significant effect on survival. Such regional treatment could include either surgery, usually in the form of a neck dissection, or regional radiotherapy. Ideally, for most salivary gland malignancy, complete surgical extirpation of local and regional disease prior to radiotherapy is desirable.
Treatment of the neck in major salivary gland malignancy has been somewhat controversial. Some investigators recommend neck dissection only for patients with clinically evident regional disease, whereas others reserve neck dissection for tumors based on various prognostic factors. Still others recommend routine elective neck dissection. Several authors have attempted to determine predictive factors for cervical metastasis in salivary gland malignancy. In a study of 145 patients with parotid gland carcinoma, Regis de Brito Santos and colleagues9 found that histologic tumor type, size of the primary lesion, and desmoplasia were significant predictors of neck node metastasis. Similarly, Kelley and Spiro10 found that the presence of pathologically positive lymph nodes in parotid gland carcinoma had a significant negative effect on overall survival. Based on their analysis of neck node metastases in 121 patients with parotid gland carcinoma, they recommended neck dissection for patients with obvious clinical nodal involvement or those patients with a large primary tumor size, histologic indications, or high tumor grade. In a multivariate analysis of patients who had elective lymph node dissection for parotid gland malignancy, Frankenthaler and associates11 identified facial nerve paralysis, tumor grade, older patient age, lymphatic invasion, and extraparotid tumor extension as predictors of occult cervical metastasis. Other authors have confirmed neck node involvement and tumor size to be major prognostic factors.12,13
For our logistic regression analysis, with the goal of predicting regional nodal metastatic disease, we selected clinical variables that are often available to the clinician prior to surgical extirpation of the primary tumor. If the likelihood of regional nodal disease could be predicted, concurrent neck dissection might be planned at the time of resection of the primary tumor, which could provide additional prognostic information and afford improved survival.
Although we identified increasing patient age and size of the primary tumor to be statistically significant multivariate predictors of positive nodal disease, their odds ratios indicate that the influence of these variables is quite small. Only when tumor sizes approach 5.0 cm do the odds for positive nodal metastasis increase 2-fold. We did not find male sex to be independent predictor for nodal disease, although other researchers have identified male sex as a negative overall survival predictor.10
Very often, the histologic type of the salivary gland malignancy can be determined by preoperative fine-needle aspiration cytologic analysis.14 Our data highlight a significant potential advantage of using fine-needle aspiration to identify malignant tumors prior to the initial surgical extirpation. If the histopathologic type can be determined preoperatively, this information may be used to determine if a neck dissection is warranted. Our data suggest that adenocarcinomas and squamous cell carcinomas of the major salivary glands should be considered for neck dissection based on significantly increased odds ratios for nodal involvement. Although mucoepidermoid carcinoma alone was not associated with nodal metastasis, high-grade mucoepidermoid carcinomas should be considered for treatment of the neck because of the independent influence of tumor grade on nodal metastasis. Increased odds for nodal involvement in high-grade mucoepidermoid carcinoma have been confirmed by others.15,16 Similarly, the association between squamous cell carcinoma of the parotid gland and probable nodal metastasis has also been demonstrated by others, with positive nodal disease rates approximating 40%.17 Our data also indicate that sarcomas, adenoid cystic carcinoma, and other histologic types are unlikely to involve nodal disease, and neck dissection may be avoided in these cases in the absence of obvious clinical disease.18
Tumor grade was found to be a significant independent predictor of nodal metastasis on multivariate analysis. Each stepwise increase in tumor grade from grade 1 (well-differentiated) through grade 4 (undifferentiated) conferred increased odds of positive nodal metastatic disease of almost 2-fold. For example, a grade 3 tumor has a 4-fold increase in odds of involving nodal disease over a grade 1 tumor. Our data strongly suggest that tumors with higher histopathologic grade be considered for treatment of the neck, reinforcing the recommendations of others.19- 21 The likely exceptions to this are adenoid cystic carcinoma and acinic cell carcinoma, in which the grade of the tumor seems to have limited effect.22 Facial nerve involvement (odds ratio, 2.3) and extraglandular tumor extent (odds ratio, 1.7) were also found to be very strong predictors of positive nodal disease, and patients with these clinical factors should also be strongly considered for treatment of the neck.
This study has several limitations. First, the patient population in this study is likely to be biased toward more severe overall disease because one of the selection criteria for cases was operative sampling of locoregional nodes. Given that most clinicians perform neck dissection or sample the local nodes when they suspect more severe disease, based on either the tumor histologic findings or the extent of the primary tumor, these patients are likely to represent a more diseased population. Therefore, the data likely represent a "worst-case scenario" for patients with major salivary gland malignancy.
Second, because the SEER database is a national tumor registry, there is probably variability in the extent of treatment assessment it records. For example, the nodal sampling may have arisen from a selective upper jugular neck dissection (eg, regions I-III) in one patient and a comprehensive neck dissection (eg, regions I-V) in another. Other authors have shown that the extent of the neck dissection is directly related to the number of overall nodes and positive nodes recovered, and that supraomohyoid neck dissections recover on average approximately 10 nodes.23 Given that the overall mean number of nodes sampled was approximately 11, we believe that most of the patients in this series had relatively thorough regional lymphadenectomies rather than simple isolated nodal sampling.
Because cervical metastases from major salivary gland malignancy significantly influence survival, treatment of the neck deserves consideration in each newly diagnosed case. The neck may be treated with postoperative radiation therapy, or it may be addressed with neck dissection at the time of primary site surgery. The decision to subject patients to the potential (but likely limited) morbidity of neck dissection in treatment of major salivary gland cancer depends on the ability to predict which patients are likely to have pathologically positive nodal disease. Our data may be helpful in determining whether to perform a neck dissection as part of the surgical treatment of major salivary gland malignancy. For example, based on our data, patients who are deemed to have extraglandular extension or high tumor grade should be considered for neck dissection at the time of surgery for the primary site. Similarly, patients with adenocarcinoma or squamous cell carcinoma, especially with poor grade, should undergo neck dissection because they are significantly more likely than patients with other types of cancer to have nodal disease in the neck. Patients who have no pathologic evidence of nodal metastasis after neck dissection may then be spared subsequent radiation therapy to the neck and its attendant morbidities.
In conclusion, the presence of positive nodal disease confers a greater than 50% decrease in mean survival in patients with major salivary gland malignancy. Therefore, patients in whom facial nerve involvement, extraglandular tumor extension, and high tumor grade are identified should be considered for neck dissection. Similarly, patients with histologically diagnosed squamous cell carcinoma or adenocarcinoma of the major salivary glands should also be considered for surgical treatment of the neck, whereas patients with adenoid cystic carcinoma or sarcomas may be able to avoid neck dissection and their conditions managed by close observation.
Accepted for publication February 13, 2002.
Corresponding author: Neil Bhattacharyya, MD, Division of Otolaryngology, 333 Longwood Ave, Boston, MA 02115.