Poor prognosis in patients with stage I and II oral tongue squamous cell carcinoma

Abstract

BACKGROUND

The objective of this study was to compare survival in patients with squamous cell carcinoma (SCC) of the oral tongue with that in patients with SCC in other oral cavity subsites.

METHODS

Patients with stage I and II (T1-T2N0M0) SCC of the oral cavity diagnosed between 1988 and 2004 were queried by using the Surveillance, Epidemiology, and End Results (SEER) database. The log-rank test was used to compare the overall survival (OS) and cause-specific survival (CSS) of patients who had oral tongue SCC with those of patients who had SCC of other oral cavity subsites. A Cox proportional hazards multivariate analysis was performed to evaluate the influence of covariates on the risk of CSS and OS.

RESULTS

Between 1988 and 2004, 6791 patients with stage I and II SCC of the oral cavity were identified. Among them, 40% had oral tongue SCC, and 60% had SCC of other oral cavity subsites. The median patient age was 64 years. The 5-year OS and CSS rates were 60.9% and 83.5%, respectively, for patients with oral tongue SCC versus 64.7% and 94.1%, respectively, for patients with SCC of other oral cavity subsites (OS: hazard ratio, 1.24; P < .0001; CSS: hazard ratio, 3.04; P < .0001). On multivariate analysis, OS and CSS were influenced significantly by T classification, age, sex, and oral tongue subsite. The CSS for patients who had stage I and II oral tongue SCC also was unfavorable compared with the CSS for stage-matched patients who had SCC of other head and neck sites.

CONCLUSIONS

Oral tongue SCC was associated with poor survival compared with other oral cavity and head and neck sites. These data suggested a potential benefit for multimodality therapy in this cohort of patients. Cancer 2008. © 2007 American Cancer Society.

The American Cancer Society estimates that approximately 34,360 patients will be diagnosed with oral cavity and oropharyngeal cancers in the United States in 2007.1 The oral tongue is the oral cavity subsite associated most commonly with squamous cell carcinoma (SCC). Previous studies have demonstrated that cancers of the oral tongue are distinct biologically and epidemiologically from other tumors of the oral cavity. Oral tongue cancers are associated with an increased proportion of patients who are women, aged <40 years, and nonsmokers.2–5 Single-institutional reviews have reported conflicting results regarding the prognostic significance of cancers arising from the oral tongue.6, 7 Because of the small number of patients in these series, it is difficult to reach definitive conclusions regarding a prognostic effect. For the current study, we used the Surveillance, Epidemiology, and End Results (SEER) database to compare the actuarial overall and cause-specific survival of patients who had stage I and II cancers of the oral tongue with the survival of patients who had other oral cavity cancers or cancers of other head and neck mucosal sites.

MATERIALS AND METHODS

Data

Data were extracted from the SEER database, which is sponsored by the United States National Cancer Institute (NCI), using the SEER 17-Registries 1973 to 2004 data set (November 2005 edition). The demographic composition of the SEER registries and incidence trends from SEER data are generally considered representative of the United States population.8, 9 The SEER database currently is comprised of patients from 18 geographically defined registries, covering approximately 26% of the United States population. (Detroit, Atlanta, Seattle-Puget Sound, San Francisco-Oakland, Los Angeles, San Jose-Monterey, Utah, New Mexico, Connecticut, Iowa, Hawaii, greater California, rural Georgia, Kentucky, Louisiana, New Jersey, and the Alaskan and Arizona Native populations).10 Data from these registries are deidentified and submitted electronically to the NCI on a biannual basis. The data subsequently are made available in the public-use SEER database. The accuracy in case ascertainment querying the SEER reportedly is 97.5%.11

Patients

The database was searched for patients who had SCC of the oral cavity diagnosed between 1988 and 2004 with tumors classified as T1 or T2, negative lymph node status (N0), and no metastasis (M0). This period was selected, because extent of disease information was available for accurate staging after 1988. All staging was performed according to the American Joint Committee on Cancer (AJCC) Cancer Staging Manual, sixth edition.12 Oral cavity subsites included the SEER site categories of “tongue,” “floor of mouth,” “lip,” and “gum and other mouth.” Cause-specific survival was determined by coding cause of death according to site of disease. Cause-specific and overall survival were compared for primary tumors of the oral tongue versus other oral cavity subsites. For illustrative purposes, cause-specific survival of stage I/II oral tongue cancer also was compared with stage-matched cancers of the oropharynx, hypopharynx, and larynx cancers. For overall survival, the observed survival rates were used for analysis.

Statistics

Patient and tumor characteristics for SCC of the oral tongue and other oral cavity subsites were acquired. The distribution of these characteristics was compared by using the chi-square test.

Overall and cause-specific survival rates were calculated by using the Kaplan-Meier method13 and were compared by using the log-rank test.14 Hazard ratios for overall and cause-specific mortality are provided with 95% confidence intervals (95% CI). Univariate analysis was performed by comparing these sites according to T classification, race, sex, and age. Data regarding pathologic features, such as margin status, perineural invasion, presence of vascular tumor emboli, and depth of tumor invasion, were unavailable in the SEER database and, thus, were not included in the current analysis. The proportion of mortality events that occurred within 2 years and 5 years of diagnosis was calculated. In addition, the proportion of deaths entered as cause-specific events was determined and compared for oral tongue versus other oral cavity subsites by using the Fisher exact test.

Multivariate analysis was performed by using a Cox proportional hazards regression model for patients with data available for all covariates analyzed.15 Separate multivariate analyses were performed for overall and cause-specific survival. Cox hazard ratios (HRs) and P values were obtained for each covariate analyzed. The covariates analyzed included site (tongue vs other oral cavity), T classification (T1 vs T2), age (<50 years vs 50–70 years vs ≥70 years), and sex.

To characterize cause-specific mortality further for patients with early-stage oral cavity cancer, the actuarial cause-specific survival for patients who had stage I and II SCC of the oral tongue was compared with that of stage-matched patients who had SCC of the oropharynx, hypopharynx, and larynx. Statistical comparisons were made by using log-rank test.

Treatment patterns for patients who had SCC of the oral tongue were compared with the patterns for patients who had SCC of other oral cavity subsites. The proportions of patients who underwent surgery alone, received radiotherapy alone, and received combined-modality (surgery and radiotherapy) were analyzed by subsite and by stage. The distribution of treatment modality was compared statistically by using the chi-square test.

SEER*Stat software (version 6.3.5; Surveillance Research Program, NCI, Bethesda, Md) was used to extract case-level data from the SEER Cancer Public-Use Database 1973 to 2004.16 All reported P values are 2-tailed. Statistical significance was defined as P < .05.

RESULTS

Between 1988 and 2004, 6791 patients with AJCC stage I/II SCC of the oral cavity were entered into the SEER database. Oral tongue cancers comprised 40% of all oral cavity cancers. The median patient age was 64 years. Younger patients and female sex were more common in patients with oral tongue cancer. AJCC T1 primary tumors and white race were more common among other oral cavity tumors. Patient characteristics are shown in Table 1.

Table 1. Patient Characteristics

Characteristic	Disease site, %		P
	Tongue	Other oral cavity
Sex
Men	60	70
Women	40	30	<.001
Age, y
<40	7.7	3.9
40-50	13.8	10.7
50-60	23.5	20.6
60-70	23.5	27.7
≥70	31.5	37.1	<.001
Race
White	84.9	91.6
Other	15.1	8.4	<.001
Tumor classification
T1	63.7	72.1
T2	36.3	27.9	<.01

All Patients

Compared with other oral cavity subsites, actuarial survival was significantly lower among patients with oral tongue SCC. At 2 years and 5 years, the overall survival rates were 76.8% and 60.9%, respectively, for oral tongue cancer versus 84.6% and 64.7%, respectively, for cancer of other oral cavity subsites (Fig. 1) (HR, 1.24; 95% CI, 1.14-1.37; P < .0001). Information regarding cause of death was available for 99% of the patients who were evaluated. Sixty-eight percent of cause-specific deaths occurred within 2 years, and 92% occurred within 5 years. Among all deaths, there was a significantly higher proportion of cause-specific deaths among patients with oral tongue cancer versus patients with cancer at other oral cavity subsites. At 5 years, 39% of all deaths were cause-specific for patients with oral tongue cancer versus 15.5% for patients with cancers of other oral cavity subsites (P < .0001). Consequently, cause-specific survival was substantially lower for patients with oral tongue cancer. At 5 years, the cause-specific survival rate was 83.5% for oral tongue cancer versus 94.1% for other oral cavity subsites (Fig. 1) (HR, 3.04; 95% CI, 2.66–3.76; P < .0001).

Figure 1. Actuarial survival of patients with stage I and II squamous cell carcinoma (SCC) of the oral tongue versus patients with stage I and II SCC of other oral cavity sites. (Left) Overall survival. (Right) Cause-specific survival.

Univariate Analysis

Univariate analyses were performed to compare the prognoses of oral cavity subsites according to T classification, age, and sex. On survival analysis by stage, the prognostic influence of oral tongue site on overall survival was most pronounced for stage I tumors. In these patients, the 2-year and 5-year overall survival rates were 82.4% and 67.4%, respectively, for oral tongue versus 88.3% and 70.8%, respectively, for other oral cavity subsites (HR, 1.22; P < .001). For patients with stage II disease, the difference in overall survival was not statistically significant (HR, 1.09; P = .20). Cause-specific survival was significantly lower among patients with oral tongue cancer for both stage I disease (HR, 3.45; P < .0001) and stage II disease (HR, 2.35; P < .0001) patients.

The influence of age on survival was evaluated by comparing the survival of patients with oral tongue SCC versus patients with other oral cavity SCC within each age group. The difference in overall survival comparing oral tongue SCC with SCC at other oral cavity subsites was significant in patients aged <40 years and aged >60 years. The proportion of overall deaths that were cause-specific was higher among patients aged <40 years compared with older patients (47% vs 29%; P < .001) because of the decreased number of deaths from competing causes of mortality. Consequently, the HR for death comparing oral tongue cancer with other oral cavity subsites was most pronounced in patients aged <40 years (HR, 2.40). Cause-specific survival was significantly lower for patients with cancers of the oral tongue in all age groups.

In addition, we compared survival according to age in patients with oral tongue cancer. There was no difference in cause-specific survival by decade for patients aged <70 years. However, the subgroup of patients aged <70 years had significantly higher cause-specific survival than patients aged >70 years (P < .001). Moreover, there was a significant trend toward lower overall survival with increasing age (P < .001; log-rank test for trend).

The adverse prognostic influence of oral tongue subsite varied according to sex. Among men, oral tongue cancer was associated with significantly lower overall survival (HR, 1.35; P < .0001) and cause-specific survival (HR, 3.62; P < .0001). Among women, cause-specific survival was lower in patients with oral tongue cancer (HR, 2.21; P < .0001), but overall survival was not (HR, 1.07; P = .36). The univariate analysis results are shown in Table 2.

Table 2. Univariate Analysis by Oral Cavity Site

Characteristic	5-Year OS, %			5-Year CSS, %
	Tongue	Other OC	P	Tongue	Other OC	P
OS indicates overall survival; CSS, cause-specific survival; OC, oral cavity.
All patients	60.9	64.7	<.001	83.5	94.1	<.001
Sex
Men	59	65	<.001	83.4	94.9	<.001
Women	63.5	64.1	.363	83.6	92	<.001
Age, y
<40	75.4	88.8	.002	86.8	96.4	.006
40–50	71.9	76.8	.056	85.9	94.4	<.001
50–60	71.5	72.5	.292	86.6	94.5	<.001
60–70	59.9	67.4	<.001	84.6	94.9	<.001
≥70	45.5	52.5	<.001	78.3	92.7	<.001
Tumor classification
T1	67.4	70.8	<.001	88.9	96.5	<.001
T2	50.8	49.9	.200	74.5	87.7	<.001

Multivariate Analysis

Cox proportional hazards survival regression analysis was performed for overall and cause-specific survival. Oral tongue primary (P < .001), T2 disease (P < .001), and younger age (P < .001) were associated with increased hazard for death. Male sex was associated with increased risk of overall mortality (P = .034), whereas female sex was associated with an increased risk for cause-specific mortality (P = .037). Multivariate analyses are shown in Table 3.

Table 3. Cox Proportional Hazards Multivariate Analysis*

Variable	OS		CSS
	CHR	P	CHR	P
OS indicates overall survival; CSS, cause-specific survival; CHR, Cox proportional hazards ratio. * A CHR >1 indicates an increased risk of mortality with the first variable listed.
Site (tongue vs other)	1.26	<.0001	2.84	<.0001
T classification (T2 vs T1)	1.96	<.0001	2.86	<.0001
Age (<50 y vs 50–69 y vs ≥70 y)	1.71	<.0001	1.34	<.0001
Sex (men vs. women)	1.10	.034	0.83	.037

Survival Comparison With Other Head and Neck Sites

To characterize the relative risk of death from early stage oral tongue cancer further, the cause-specific survival of patients who had stage I/II oral tongue cancers was compared with that of patients who had stage I/II SCC of the oropharynx, larynx, and hypopharynx. Oral tongue cancer was associated with significantly increased cause-specific mortality (Table 4). In addition, the 5-year survival for patients who had stage II oral tongue SCC was compared with that of patients who had stage III and IV SCC of the oropharynx. Cause-specific survival was significantly lower for patients who had stage II oral tongue SCC compared with patients who had stage III (HR, 1.70; P < .001) and stage IV (HR, 1.56; P < .001) oropharynx cancers. The overall survival observed in patients with stage II oral tongue SCC was similar to that of patients with stage III (HR, 1.11; P = .18) and stage IV (HR, 0.95; P = .36) oropharynx SCC. Furthermore, the 5-year cause-specific survival for patients with stage I oral tongue cancers that measured >1 cm was similar to that for patients with stage III oropharyngeal cancer (HR, 0.92; P = .55). Actuarial survival curves comparing stage II oral tongue SCC with stage III/IV oropharyngeal SCC are shown in Figure 2.

Figure 2. Actuarial survival of patients with stage II squamous cell carcinoma (SCC) of the oral tongue versus patients with stage III and IV SCC of the oropharynx. (Left) Overall survival did not differ significantly between patients with stage II cancer of the oral tongue versus patients with stage III and IV cancer of the oropharynx (P > .05). (Right) Cause-specific survival was significantly lower for patients with stage II cancer of the oral tongue versus patients with stage III (hazard ratio, 1.70; P < .001) and stage IV (hazard ratio, 1.56; P < .001) cancer of the oropharynx.

Table 4. Five-year Cause-specific Survival of Oral Tongue Cancer Compared With Other Head and Neck Sites

	Disease site, %
	OC		Oropharynx	Larynx	Hypopharynx
	Tongue	Other OC
OC indicates oral cavity. * Significant (P < .05) compared with stage-matched patients who had oral tongue cancer.
Stage
I	88.9	96.5*	94.7*	92.2*	91.5
II	74.5	87.7*	84.9*	82.6*	87.8*

Patterns of Treatment

In patients with oral tongue cancers, 70.4% underwent surgery alone, 7.6% received radiotherapy alone, and 22.0% received combined-modality therapy (surgery and radiotherapy). In patients with SCC of other oral cavity sites, 80.9% underwent surgery alone, 5.2% received radiotherapy alone, and 13.9% received combined-modality therapy. The differences in distribution of treatments were statistically significant (P < .0001). In patients with oral tongue cancers, treatment patterns varied by stage. The use of radiotherapy alone (14.6% vs 4.1%; P < .001) and the use of combined-modality therapy (31.9% vs 17.1%; P < .001) were more common in patients who had stage II lesions compared with patients who had stage I disease.

DISCUSSION

The current analysis of survival rates in patients with stage I/II SCC of the oral cavity demonstrates that oral tongue SCC is associated with lower rates of overall and cause-specific survival compared with SCC located in other oral cavity subsites. Furthermore, cause-specific survival for patients who have stage II oral tongue cancer is unfavorable compared with that of patients who have stage III/IV SCC of the oropharynx. The low survival rates observed for oral tongue cancers probably is related to their unique biology and epidemiology and a high propensity for locoregional recurrence compared with other SCC of the head and neck.

Recent single-institutional studies have reported conflicting results regarding the prognostic significance of oral tongue cancer among oral cavity subsites. Yao et al. reported the University of Iowa experience with 55 patients who received intensity-modulated radiation therapy for oral cavity cancer. In that series, oral tongue cancer was associated with significantly worse 2-year locoregional recurrence-free survival compared with floor of mouth cancer (68.8% vs 100%; P = .03), and there was a trend toward a lower rate of locoregional control compared with all oral cavity subsites (P = .096).6 In contrast, the University of Oregon published their experience with 233 oral cavity tumors and reported no difference in overall survival (P = .8) or disease-free survival (P = .7) when comparing cancers of the oral tongue with cancers of other oral cavity subsites.7 In contrast to our analysis, both the Iowa and Oregon experiences were small series and included patients with all stages of disease, making direct survival comparisons between subsites difficult. Moreover, survival rates were not analyzed according to disease stage in either of those reports.

Histopathologic studies of oral tongue tumors have indicated that these lesions are distinct biologically from other oral cavity cancers. Sathyan et al. prospectively compared the expression of cell cycle-regulatory proteins in 147 buccal mucosa and 94 oral tongue tumors.17 In that study, oral tongue cancers were associated with an increased down-regulation of p16 and p21 proteins. Moreover, alterations in gene expression associated with survival in buccal mucosa cancers were not predictive of outcome in patients with oral tongue cancers. Schwartz et al. studied biologic markers of programmed cell death in animal models by comparing cancers of the oral tongue and buccal mucosa. Oral tongue cancers were associated with reduced expression of BCL-2 and wild type p53 compared with similar lesions of the buccal mucosa.18 In addition, oral tongue cancers that occurred in patients aged <40 years without conventional risk factors appeared to differ biologically from those that occurred in older individuals. A histopathologic study of tumors from 21 young patients with SCC of the tongue who had no known risk factors revealed increased expression of wild-type p53 in 81% of patients.19 This pattern differs from the more typical pattern of overexpression of p53 with mutations in exons 5 through 9 observed in older patients with a significant history of alcohol and tobacco abuse.

The SEER database includes only information regarding patient survival. Patterns of failure could not be analyzed, because this information was not available. Previous reports, however, have demonstrated an increased risk of locoregional failure associated with oral cavity cancer compared with other head and neck sites. In a prospective, randomized trial of accelerated radiotherapy performed by the Danish Head and Neck Cancer Study Group, oral cavity tumors were associated with significantly worse locoregional tumor control compared with other sites.20 Adelstein et al. reviewed the Cleveland Clinic experience of patients who were treated with concurrent chemoradiotherapy and reported a decreased rate of local control in patients with oral cavity tumors.21 Similarly, in a randomized trial evaluating radiotherapy dose in patients who were treated in the postoperative setting, Peters et al. reported an increased risk of local recurrence with oral cavity tumors compared with other head and neck sites.22 In addition, it has been demonstrated that local failure is a specific challenge with oral tongue cancers. A patterns-of-failure analysis of oral tongue and floor of mouth cancers demonstrated that recurrence was most common at the primary site.23

Patients with early-stage head and neck cancers, including oral tongue cancer, traditionally have been treated with single-modality therapy. The poor survival rates with stage I/II oral tongue cancer observed in the current series suggest that these patients may benefit from multimodality therapy. Large stage I (>1 cm) and stage II oral tongue SCCs are associated with similar or decreased survival compared with stage III/IV oropharyngeal tumors. Among the limitations of this study is the absence of detailed pathologic information. Recent adjuvant therapy trials have used high-risk pathologic factors as inclusion criteria or for risk stratification. Most of the patients enrolled in those trials, however, had stage III and IV disease. Ang et al. studied the role of accelerated radiotherapy in high-risk patients. In their trial, only 4% of enrolled patients had stage II tumors.24 Similarly, in the Radiation Therapy Oncology Group 95-01 trial of adjuvant chemoradiotherapy, 83% of patients had lymph node extracapsular extension or ≥2 involved lymph nodes, which classified their disease as stage III or IV by definition.25 The other inclusion criteria for that trial were positive surgical margins, which occur infrequently in early-stage oral tongue cancers. Hicks et al. reviewed the Roswell Park Cancer Institute experience of 79 consecutively treated patients with oral tongue cancer, including 69% who had stage I or II disease. None of the patients in their series had positive margins, and only 15% had close margins.26 Sessions et al. described positive margins in 22% of patients with oral tongue cancer who were treated at Washington University; however, their study did not report the incidence by stage.27 These data suggest that the vast majority of patients with stage I and II oral tongue cancers in the current analysis would not have met the inclusion criteria for collaborative group adjuvant therapy trials.

Despite the absence of pathologic information in the SEER database, there remains compelling evidence that stage II oral tongue cancers are associated with a high rate of cause-specific mortality even compared with more advanced lesions of other head and neck sites. Despite these poor survival rates, only 22% of the patients in the current analysis were treated with combined surgery and radiation therapy. The observed survival proportions suggest that stage I/II oral tongue SCC is associated with a high rate of treatment failure and that more aggressive local treatment, such as combining surgery with radiation or chemoradiation, may be considered.

Retrospective studies have reported conflicting results regarding young age on survival in oral tongue cancer. A recent matched-pair analysis of 138 patients with oral tongue cancer revealed increased cancer-specific mortality associated with age <40 years.28 In contrast, Pittman et al. performed meta-analyses of patients with oral tongue cancer aged <40 years and observed similar 3-year disease-free survival rates compared with patients aged >40 years.29 Similarly, Manuel et al. reported their institution's experience with oral tongue cancer in patients aged <45 years and observed overall and disease-free survival rates comparable to the rates in historic controls.30 In the current series, the HR for cause-specific mortality comparing SCC of the oral tongue with SCC of other oral cavity subsites was independent of age. However, we observed an increased hazard for overall mortality in patients aged <40 years with oral tongue cancer compared with other oral cavity subsites because of a decreased number of deaths from competing causes of mortality in this group.

In conclusion, stage I/II oral tongue SCC is associated with poor survival compared with SCC in other oral cavity subsites. We speculate that patients with stage I/II oral tongue SCC may benefit from inclusion in future clinical trials evaluating more aggressive treatment, such as combined-modality therapy.