The objective of treatment in patients with Stage I/II carcinoma of the tongue1 is to control the primary tumor and regional neck metastasis.2 Several types of treatment are used to control metastasis: resection of the primary tumor with or without elective neck treatment (dissection or irradiation) and radiotherapy to the primary tumor with or without elective neck treatment. At the Department of Head and Neck Surgery of the National Cancer Center Hospital East in Chiba, Japan, resection of the primary tumor without elective neck dissection is the standard treatment in patients with Stage I/II carcinoma of the tongue. However, in our institution, cervical metastases developed in approximately 40% of such patients managed with partial glossectomy only. Naturally, the survival of these patients is poorer than that of patients without cervical metastasis.
What features can be used to distinguish tumors that metastasize to the neck after surgery from those that do not? Several studies have shown that tumor thickness3 and vascular invasion4 are risk factors for cervical metastasis in patients with squamous cell carcinoma of the head and neck. However, few studies have used multivariate analysis to identify factors associated with cervical metastasis in patients with Stage I/II squamous cell carcinoma of the tongue. In this study, we attempted to identify the clinicopathologic factors most significantly associated with cervical metastasis in patients with International Union Against Cancer TNM Stage I/II squamous cell carcinoma of the anterior two-thirds of the tongue.
MATERIALS AND METHODS
We reviewed the medical records of 253 patients who were treated for squamous cell carcinoma of the anterior two-thirds of the tongue at the Department of Head and Neck Surgery of the National Cancer Center Hospital between 1980-1991 and the National Cancer Center Hospital East between 1992-1995. This study was limited to patients with Stage I/II carcinoma of the tongue.1 All patients underwent partial glossectomy through the mouth without elective neck dissection. Moreover, no patients received either preoperative or postoperative treatment. There were no recurrences at the primary site. Finally, 44 patients were selected for the study: 21 who developed cervical metastases postoperatively and 23 who had no evidence of disease for 5 years after surgery. The median patient age was 63 years (range, 36-85 years), and the male to female ratio was 1.6:1 (27 men and 17 women).
Paraffin embeded specimens were retrieved from the files of the pathology department. Serial sections 4-μm thick were cut. Alternating sections were stained with hematoxylin and eosin.
All pathologic slides were reviewed at the maximum cross-section by two observers (T.A. and T.Y.) without knowledge of the clinical data. The histologic evaluation was based on the grading system of Jakobsson et al.5 with slight modification (Table 1). The grading scheme of Crissman et al. was used for nuclear polymorphism.6 Mitoses were counted in 10 high-power fields (HPF) (x400). Tumor cells were graded separately in the superficial and deep portions because cellular morphology in these portions differed. The border between carcinoma and normal tissue, cancerous nest formation, and the infiltrative growth ratio were examined to assess the mode of invasion. The infiltrative growth ratio was calculated as the percentage of the tumor area comprised of small nests of ten or fewer cells at the periphery of the tumor to the total area of the tumor. Depth, expressed as the deepest layer of tumor invasion, was recorded.
Table 1. Grading of Histologic Factors
|Differentiation||Well: marked keratinization||Moderate: some keratinization||Poor: minimal keratinization||Poor: no keratinization|
|Nuclear polymorphism||A few enlarged nuclei||Moderate number of enlarged nuclei||Markedly irregular, enlarged nuclei||Enlarged anaplastic immature nuclei|
|Tumor border||Smooth||Fairly even||Rough||Irregular|
|Nest formation||Thick trabecular||Thin trabecular||Loose||Scirrhous|
|Infiltrative growth ratio (%)||0-10||11-20||21-30||30+|
Tumor thickness also was measured to evaluate invasion. Regardless of whether the growth type of the carcinoma was exophytic or invasive, according to Breslow's classification for melanoma,7 tumor thickness was measured from the surface of the normal mucosa to the deepest portion of the tumor. Lymphocytic infiltration and increased connective tissue were evaluated to assess the cellular response of the host. Perineural invasion also was recorded.
All data were tabulated and statistical tests were performed with the SAS statistical software package (SAS Institute, Inc., Cary, NC). The chi-square test with Fisher's exact test was used to assess the association of cervical metastasis with clinicopathologic factors. The prognostic significance of clinicopathologic factors in cervical metastasis was assessed with univariate and multivariate logistic regression analysis. The correlations between clinicopathologic factors and the disease free survival/overall survival were analyzed with the log rank test and generalized Wilcoxon test. Survival curves were plotted using the Kaplan-Meier method. The prognostic significance of clinicopathologic factors for disease free survival and overall survival were assessed using Cox's multivariate proportional hazards regression analysis. Correlations were considered significant when their overall P values were <0.05.
Univariate Analysis of Clinicopathologic Factors in Cervical Metastasis
The associations between clinicopathologic factors and cervical metastasis are presented in Table 2. Cervical metastasis correlated with growth type (P = 0.011), tumor differentiation of the deepest layer (P < 0.001), nuclear polymorphism of the deepest layer (P = 0.001), tumor border between the carcinoma and peripheral normal tissue (P = 0.032), cancerous nest formation (P = 0.011), the infiltrative growth ratio (P = 0.007), depth (P = 0.006), and tumor thickness (P < 0.001). However, cervical metastasis did not correlate with age, gender, smoking, alcohol, location, T classification, mitosis in the deep and superficial layers, differentiation and nuclear polymorphism of the superficial layer, connective tissue, lymphocytic infiltration, or perineural invasion.
Table 2. Univariate Analysis of Clinicopathologic Factors in Cervical Recurrence
| <70||4 (24)||13 (76)|
| 70||17 (63)||10 (37)||2.7||0.7-10.5||NS|
| Men||16 (59)||11 (41)|
| Women||5 (29)||12 (71)||0.3||0.1-1.0||NS|
| Dorsum, margin||15 (43)||20 (57)|
| Inferior||6 (67)||3 (33)||2.7||0.6-12.4||NS|
| Exophytic, superficial||4 (24)||13 (76)|
| Invasive||17 (63)||10 (37)||5.5||1.4-21.7||0.011|
| T1||6 (35)||11 (65)|
| T2||15 (56)||12 (44)||2.3||0.7-8.0||NS|
|Differentiation of the deep portion|
| Well, moderate||11 (32)||23 (68)|
| Poor||10 (100)||0 (0)||NA||NA||<0.001|
|Nuclear polymorphism of the deep portion|
| Few, moderate||5 (23)||17 (77)|
| Markedly enlarged||16 (73)||6 (27)||9.1||2.4-35.7||0.001|
| Smooth, even||6 (30)||14 (70)|
| Rough, irregular||15 (63)||9 (37)||3.9||1.1-13.8||0.032|
| Thick or thin trabecular||4 (24)||13 (76)|
| Loose, scirrhous||17 (63)||10 (37)||5.5||1.4-21.7||0.011|
|Infiltrative growth ratio|
| 20%||8 (31)||18 (69)|
| >20%||13 (72)||5 (28)||5.9||1.6-22.0||0.007|
| ep, sep||0 (0)||7 (100)|
| m||21 (57)||16 (43)||NA||NA||0.006|
| <4 mm||2 (12)||15 (88)|
| ≥4 mm||19 (70)||8 (30)||17.8||3.3-96.5||<0.001|
| (-)||18 (45)||22 (55)|
| (+)||3 (75)||1 (25)||1.5||0.7-1.2||NS|
Logistic Regression Analysis of Clinicopathologic Factors in Cervical Metastasis
The factors that correlated with cervical metastasis were grouped into three categories. The first category was tumor thickness (growth type, depth, and tumor thickness). The second category was cellular morphology (differentiation and nuclear polymorphism at the deep portion). The third category was infiltrative pattern (the border between the carcinoma and normal tissue, cancerous nest formation, and the infiltrative growth ratio).
Differentiation could not be examined with logistic regression analysis because the tumor recurred in all poorly differentiated squamous cell carcinoma cases. Thus, for cellular morphology, only nuclear polymorphism was examined. For tumor thickness and the infiltrative pattern, tumor thickness and the infiltrative growth ratio, respectively, were selected. Multivariate logistic regression analysis of the three categories demonstrated that only tumor thickness had a predictive value for cervical metastasis (P = 0.016; odds ratio = 9.4) (Table 3).
Table 3. Logistic Regression Analysis in Cervical Recurrence
|Nuclear polymorphism |
| few, moderate vs. markedly, enlarged||9.1||2.4-35.7||0.001||3.2||0.6-16.7||0.154|
|Infiltrative growth ratio |
| ≤20% vs. >20%||5.9||1.6-22.0||0.007||2.5||0.5-12.4||0.250|
|Tumor thickness |
| <4 mm vs. ≥4 mm||17.8||3.3-96.5||<0.001||9.4||1.5-57.7||0.016|
Univariate Analysis of Disease Free Survival
Disease free survival correlated with the same clinicopathologic factors on univariate analysis of cervical metastasis (log rank test and generalized Wilcoxon test). Moreover, disease free survival was associated with perineural invasion.
Cox's Proportional Hazards Regression Analysis of Histopathologic Factors for Disease Free Survival
As with logistic regression analysis of histopathologic factors in cervical metastasis, the factors were divided into four broad categories: cellular morphology, infiltrative growth ratio, tumor thickness, and perineural invasion. Cox's multivariate proportional hazards regression analysis of these four categories demonstrated that only tumor thickness had a predictive value (P = 0.032; risk ratio = 5.6).
Univariate Analysis of Overall Survival
No statistical correlation was found between overall survival and any of the clinicopathologic factors.
Relation between Cellular Morphology of Cervical Metastasis and Disease Free Survival in Patients with Tumors > 4 mm
Patients with poorly differentiated squamous cell carcinoma had a higher incidence of cervical metastasis than did those with well or moderately differentiated squamous cell carcinoma when the tumor was > 4 mm (odds ratio = 3.8, logistic regression analysis). Furthermore, patients with poorly differentiated squamous cell carcinoma with tumors > 4 mm had shorter disease free survival than those with well or moderately differentiated carcinoma (risk ratio = 2.4, Cox's proportional hazards regression analysis).
The unpredictable clinical behavior of squamous cell carcinoma of the head and neck has led several investigators to search for factors that may be useful as indicators of cervical metastasis and prognosis. Jakobsson et al. proposed a semiquantitative grading scheme for laryngeal carcinoma and found a correlation between the malignant grading score and survival.5 Willen et al. proposed a modification of the grading scale of Jakobsson et al. for gingival squamous cell carcinoma and found a strong correlation between histologic malignancy and survival.8 The modified grading scale of Jakobsson et al. was used in this study. Our univariate analysis revealed the factors correlating with recurrence and disease free survival were tumor thickness, cellular morphology in the deep portion, infiltrative growth, and perineural invasion.
In their study of squamous cell carcinoma of the lower lip, Frierson et al. emphasized that important variables were best evaluated in deeper portions of the lesions.9 In addition, in recent immunohistochemical studies, staining of cell-to-cell adhesion molecules, such as E-cadherin,10, 11 and of extracellular matrix components, such as Type IV collagen and laminin,12, 13 in the deep portion were important for evaluating the malignant grade of tumors. However, no other studies have evaluated the usefulness of cellular morphology in the deep portion of a squamous cell carcinoma of the head and neck. Our data clearly demonstrate that cellular characterization in the deep portion is appropriate for evaluating of malignant grade.
In their study of oropharyngeal squamous cell carcinoma, Crissmann et al. showed that the pattern of invasion is predictive of survival.6 This result is in accordance with our current results.
Some investigators have reported that vascular invasion4, 14, 15 and perineural invasion16-18 are important prognostic factors in patients with squamous cell carcinoma of the head and neck. Vascular invasion was regarded as a sign of multiple stage progression in the metastatic cascade.12 Moreover, it correlated with tumor thickness and cervical metastasis. Therefore, vascular invasion was regarded as a confounding factor on multivariate analysis and was thus excluded from this analysis.
Tumor thickness is an important factor in cervical metastasis of squamous cell carcinoma of the tongue and floor of the mouth,3, 19, 20 lower lip,9 soft palate,21 and oral cavity.22 However, its importance has not been substantiated with multivariate analysis in patients with Stage I/II squamous cell carcinoma of the tongue. Multivariate analysis in the current study identified only tumor thickness as having a powerful predictive value for postoperative cervical metastasis. Tumor thickness roughly can be estimated by preoperative palpation with a vernier scale. However, ultrasonography appears to be more accurate for the estimation of tumor thickness and we currently are studying its usefulness for preoperative evaluation.
Another important factor in postoperative cervical metastasis was found to be the cellular morphology of the deep portion of the tumor. This morphology can be predicted from examination of superficial biopsy specimens because the deep portion usually is as differentiated or less so than the superficial layer. Therefore, if the differentiation of a specimen from the superficial layer is moderate or poor, the differentiation of the deep portion is very likely to be moderate or poor. On the basis of the results of the current study, we concluded that tumors > 4 mm with moderately or poorly differentiated squamous cell carcinoma histology as demonstrated by biopsy have a high potential for postoperative cervical metastasis.
Several studies have suggested that elective neck dissection should be performed in patients with N0 squamous cell carcinoma. However, these studies differed widely in their indications. Cunningham et al. recommended elective neck dissection for patients with Stage I/II oral cavity carcinoma.1 Whitehurst et al. advocated elective neck dissection in patients with oral tongue carcinoma > 2 cm.23 Mendelson et al. recommended simultaneous elective neck dissection for all patients with T2 lesions and that elective neck dissection should be considered seriously for patients with T1 lesions higher than Grade 2 of Border's classification24 or with lesions spreading into the floor of the mouth.22 Teichgraeber et al. proposed that more aggressive therapy25 such as ipsilateral elective irradiation or modified neck dissection should be performed in patients with carcinoma of the tongue with poorly differentiated T1 lesions or with T2 or higher lesions. Bilateral neck dissection or irradiation was reserved for midline lesions of T2N0 or higher. In contrast, Spiro et al.26 and Vandenbrouck et al.27 suggested delaying neck dissection for patients with Stage I/II squamous cell carcinoma of the oral cavity until a lymph node is detectable clinically.
We consider radical treatment excessive for patients with Stage I/II carcinoma of the tongue, because there is no recurrence in 60-70% of patients in whom neck dissection is not performed. Therefore, definitive indications for elective neck dissection should be established.
On the basis of univariate analysis alone, Fakih et al. proposed radical neck dissection for patients with carcinoma of the tongue with a tumor depth > 4 mm.28 Weiss et al. suggested that if the probability of occult cervical metastasis is >20%, treatment of the neck is warranted.29 In our study, disease recurred in the neck in 70% (19 of 27) of patients in whom tumor thickness was > 4 mm. Therefore, Stage I/II carcinoma of the tongue > 4 mm is an appropriate indication for elective neck dissection. In addition, a diagnosis of moderately or poorly differentiated squamous cell carcinoma on biopsy is an even stronger indication for neck dissection.
We consider patients with Stage I/II carcinoma of the tongue whose tumors are > 4 mm in thickness to have latent metastasis. Therefore, these patients require early neck dissection. Elective neck dissection reasonably may employ the same procedures as delayed neck dissection in these patients. Survival might be better in patients undergoing elective neck dissection than in patients undergoing delayed neck dissection. In addition, the possibility of preserving nonlymphatic structures might be better in patients undergoing elective neck dissection than in those with delayed (performed only after overt metastasis has developed) neck dissection. In our hospital, conservative supraomohyoid neck dissection30 is indicated in patients with T1N1 and T2N1 carcinoma of the tongue. Thus, conservative supraomohyoid neck dissection may be sufficient for patients with Stage I/II carcinoma of the tongue whose tumors are > 4 mm in thickness.
Multivariate analysis demonstrated that only tumor thickness had a predictive value for cervical metastasis and influenced disease free survival in patients with Stage I/II carcinoma of the tongue. Our results clearly demonstrate that conservative supraomohyoid neck dissection is indicated in patients with Stage I/II carcinoma of the tongue whose tumors are >4 mm in thickness.