The prognostic value of hypoxia markers in T2-staged oral tongue cancer

Introduction 

Approximately 30% of malignant tumors of the head and neck are oral cavity cancers, with nearly 95% of these being squamous cell carcinomas (SCCs). In the United States, approximately 22,560 new cases of oral cavity cancers were diagnosed in 2007, with approximately 5370 deaths attributable to this disease.1 Regional metastases are present in approximately 50% of patients at diagnosis, and these metastases are associated with poor patient survival.2, 3 Among the other factors that affect survival of patients with oral cavity cancer are stage of disease at the time of diagnosis, size and site of tumor, depth of invasion, and histological type.4, 5 Molecular markers linked to the occurrence and progression of malignancies have been used in molecular tumor staging or grading, or for molecular targeting of therapy, but little is known regarding the prognostic and therapeutic value of biomarkers in the diagnosis and treatment of oral cavity cancer.6, 7

Hypoxia is characteristic of solid tumors, arising from a less ordered vasculature and necrosis.8, 9 Clinically relevant hypoxia has been detected in approximately 50% of all solid tumors.10 Tumor hypoxia is an important factor determining the response of head and neck SCCs to surgery, chemotherapy, and radiotherapy.8, 9 Less is known, however, about hypoxia biomarkers in oral cavity cancer, particularly in T2-staged oral tongue SCC, which has a high propensity for regional metastasis and locoregional failure after radical surgery or radiotherapy, affecting patient survival.11 We therefore assessed whether hypoxia markers and clinicopathologic variables could predict survival of patients with T2-staged SCCs of the oral tongue.

Patients and methods 

Patients, treatments, and study design 

The medical records of all consecutive patients treated for primary oral tongue cancer at our institution between January 1997 and December 2005 were reviewed. Patients were identified through a computer-assisted search of our institutional patient database. Initially, 47 patients with T2-staged oral tongue cancer were identified, but 4 patients with prior radiotherapy or concurrent second primary malignancies were excluded. The clinicopathologic, treatment, and follow-up data of the 43 eligible patients (27 men, 16 women) who underwent radical surgery and neck dissection were reviewed (Table 1). The study protocol was reviewed and approved by our Institutional Review Board.

Table 1. Clinical characteristics of 43 patients with T2 oral tongue squamous cell carcinoma

	Variable	No. of patients	% of total
	Gender, male/female	27/16	63/37
	Age,a years	59 (22–87)
	Primary tumor
	Size,a cm	2.5 (2.1–4.0)
	Thickness,a cm	1.0 (0.2–2.6)
	Differentiation, well/moderate/poorly	30/9/4	70/21/9
	Nodal classification (AJCC, 2002)
	N0/N1/N2	21/12/10	49/28/23
	Treatment
	Neck dissection, no/yes	0/43	0/100
	Radiotherapy, no/yes	19/24	44/56
	Follow-up period of survivors,a months	60 (37-123)

a Median (range).

All patients underwent glossectomy with proper tumor-free margins >1cm with neck dissection on the involved side with/without reconstruction for defects of the mobile tongue. The 31 patients with clinically negative nodal involvement underwent supraomohyoid neck dissection (levels I–III), and the 12 patients with clinically positive nodal involvement underwent modified radical neck dissection (levels I–V). Four patients with midline involvement of primary tongue cancer underwent bilateral neck dissection. Twenty-four patients (56%) received postoperative locoregional radiotherapy (median dose, 60Gy; range, 45–70Gy), using single daily fractions. Indications for radiotherapy included close or positive resection margins, multiple cervical lymph node involvement, and extracapsular spread of positive nodes.

The surgical pathology results of all patients were reviewed for evidence of size, invasion thickness and differentiation of primary tumors, surgical margin status, and regional metastases of the neck dissection specimen. Tumors were staged according to the American Joint Committee on Cancer (AJCC) staging system.12 All patients were followed up with clinical and radiologic examinations.

Results 

The median size of the 43 primary oral tongue cancers was 2.5cm (range, 2.1–4.0cm) and the median tumor thickness was 1.0cm (range, 0.2–2.6cm) (Table 1). All tumors were SCCs, with most (70%) being well differentiated. Twenty-two patients (51%) had positive cervical lymph nodes. Resection margin was positive in one patient and <0.5cm in 3 patients.

On immunohistochemical staining of a tissue microarray, most tumors were negative or weakly positive for HIF-1α and HIF2-α and moderately or strongly positive for CA-9, GLUT-1, and EPOR (Fig. 1, Table 2). On correlation analyses between clinicopathologic findings and hypoxia marker staining, tumor differentiation was correlated with nodal classification and tumor thickness was correlated with expression of CA-9 and GLUT-1 (P<0.05 each) (Table 3). Of the five hypoxia markers, only GLUT-1 was correlated with nodal classification. In addition, the expression of hypoxia markers was generally well correlated (P<0.05).

View Large Image Download to PowerPoint

Figure 1. Immunohistochemical staining of the hypoxia-related markers (A) Hypoxia-inducible factor (HIF)-1α (nuclear staining); (B) Carbonic anhydrase (CA)-9 (cytoplasmic membrane staining); (C) Glucose transporter (GLUT)-1 (cytoplasmic membrane staining); and (D) Erythropoietin receptor (EPOR) (cytoplasmic staining). Original magnification, ×400 (A) and ×200 (B–D).

Table 2. Immunohistochemical results of hypoxia markers in 43 T2 oral tongue cancers

	Marker	Interpretation (n)a					Median value
		0 (0%)	1 (1–10%)	2 (11–50%)	3 (51–80%)	4 (>80%)
	HIF-1α	37	2	2	2	0	0
	HIF-2α	33	4	5	1	0	0
	CA-9	17	7	7	10	2	1
	GLUT-1	12	2	6	14	9	3
	EPOR	11	1	5	7	19	3

Abbreviations: CA, carbonic anhydrase; EPOR, erythropoietin receptor; GLUT, glucose transporter; HIF, hypoxia-inducible factor.

a Graded by percentage of stained tumor cells.

Table 3. Correlations between clinicopathologic findings and hypoxia marker staining

		P	ra
	Tumor differentiation vs. N classification	0.025	0.341
	Tumor size vs. nodal ECS	0.038	0.318
	Tumor thickness vs. CA-9	0.017	0.363
	Tumor thickness vs. GLUT-1	0.041	0.314
	N classification vs. nodal ECS	0.010	0.387
	N classification vs. GLUT-1	0.026	0.339
	HIF-1α vs. HIF-2α	0.074	0.275
	HIF-1α vs. CA-9	0.005	0.418
	HIF-1α vs. GLUT-1	0.015	0.368
	HIF-2α vs. CA-9	0.029	0.334
	CA-9 vs. GLUT-1	<0.001	0.551
	CA-9 vs. EPOR	0.001	0.482
	GLUT-1 vs. EPOR	<0.001	0.653

Abbreviations: CA, carbonic anhydrase; ECS, extracapsular spread; EPOR, erythropoietin receptor; GLUT, glucose transporter; HIF, hypoxia-inducible factor.

a Spearman’s rho correlation coefficients.

Among the 43 patients, 14 had died of disease, 4 died of other causes, and 25 were alive at last follow-up with a median survival of 60 months (range, 37–123 months). Fifteen patients presented with recurrent locoregional disease and five with distant metastases. Four patients had other cancers (hepatocellular carcinoma, cholangiocarcinoma, lung cancer, and papillary thyroid carcinoma) during follow-up. At 5 years, the locoregional control rate was 59%, the DSS rate was 69%, and the OS rate was 59%. Univariate analysis showed that HIF-1α and EPOR expression were significantly correlated with DSS (P<0.05) (Fig. 2), but other clinicopathologic variables, such as tumor thickness, nodal involvement, and resection margin status, were not (Table 4). In multivariate analysis, only EPOR expression remained a significant predictor of DSS (P=0.030) (Table 5).

View Large Image Download to PowerPoint

Figure 2. Actuarial disease-specific survival in 43 patients with T2-staged oral tongue cancers according to hypoxia-related marker staining. (A) Hypoxia-inducible factor (HIF)-1α: low (n=37) vs. high (n=6), P=.035. (B) Erythropoietin receptor (EPOR): low (n=24) vs. high (n=19), P=.019.

Table 4. Univariate analysis for locoregional control and disease-specific survival

	Variable	No. of patients	% 5-year LRC	P	% 5-year DSS	P
	Sex, male/female	27/16	60/59	0.492	66/61	0.429
	Age, <60/⩾60 years	23/20	60/58	0.935	69/69	0.811
	Tumor size, ⩽2.5/>2.5cm	25/18	58/64	0.770	72/65	0.315
	Tumor thickness, ⩽1.0/>1.0cm	23/20	63/50	0.741	78/58	0.204
	Tumor differentiation, W/M/P	30/9/4	67/64/25	0.064	73/65/50	0.668
	Resection margin, neg/close	39/4	61/50	0.401	71/50	0.263
	Nodal involvement, no/yes	21/22	71/47	0.113	81/56	0.184
	Nodal ECS, no/yes	37/6	61/44	0.144	72/50	0.256
	Radiotherapy, no/yes	19/24	63/55	0.675	79/61	0.403
	HIF-1α, low/high	37/6	65/46	0.154	75/33	0.035a
	HIF-2α, low/high	33/10	67/55	0.630	72/60	0.698
	CA-9, low/high	24/19	63/51	0.857	80/53	0.159
	GLUT-1, low/high	34/9	61/53	0.416	76/44	0.060
	EPOR, low/high	24/19	72/43	0.104	83/52	0.019a

Abbreviations: CA, carbonic anhydrase; DSS, disease-specific survival; ECS, extracapsular spread; EPOR, erythropoietin receptor; HIF, hypoxia-inducible factor; LRC, locoregional control; neg, negative; W/M/P, well/moderate/poorly.

a P<0.005.

Table 5. Cox proportional hazards models for disease-specific survival

	Variable	Disease-specific survival
		RR	95% CI	P
	HIF-1α, low/high	0.352	0.109–1.139	0.081
	EPOR, low/high	0.274	0.085–0.880	0.030

Abbreviations: CI, confidence interval; EPOR, erythropoietin receptor; HIF, hypoxia-inducible factor; RR; relative risk.

Discussion 

We analyzed the prognostic significance of demographic, clinical, and histological factors and hypoxia-related biomarkers in 43 patients with oral tongue SCCs. Although mobile tongue cancer is usually detected at a relatively early T stage, cervical nodal metastasis is present in as many as 17% of patients with T1 tumors and 45% of patients with T2 tumors,13 and patients with T2-staged oral tongue cancer have a higher propensity for nodal metastasis and a poorer prognosis than patients with T1 tumors.14 We therefore assessed prognostic variables in patients with T2 oral tongue SCC. Although various demographic, clinical and histological factors have been suggested as prognostic in these patients,2, 3, 4, 5, 14, 15, 16 including tumor thickness, differentiation, and nodal stage, we found that none of these factors was significantly associated with locoregional control or survival. This may be due to the small number of patients or the inclusion of patients with only T2-stage tumors. We found, however, that EPOR expression was an independent prognostic factor in these patients, indicating that this biomarker may be useful in identifying more aggressive tumors at the time of diagnosis.

Several biomarkers have been identified as prognostic predictors in patients with oral tongue cancer.17, 18, 19, 20, 21 For example, cyclin D1 and p16INK4A expression predicted reduced survival in 148 patients with anterior tongue SCC treated by surgical resection,17 and lack of PTEN expression was associated with shorter survival in 41 patients with stage II–IV oral tongue SCCs.18 Increased DSS was associated with high tumor expression of p21WAF1/CIP1, particularly in combination with p53 accumulation or Bax expression, in 80 randomly selected patients with T1 to T4 oral tongue SCCs.19 In addition, apoptosis rate and p53 expression were found to predict poor prognosis in 55 randomly selected patients with tongue SCCs.20 Recently, high expression of CD105 (endoglin) or vascular endothelial growth factor (VEGF) was found to be associated with a more aggressive potential in 94 patients with T1 and T2 tongue cancers.21 Occult cervical metastasis in stage I and II oral tongue SCC may be predicted by reduced expression of E-cadherin,22 by cyclin D1 gene aberrations,23 and by increased expression of cytokeratin.11 In agreement with these findings, we have shown that several hypoxia markers can be used to identify aggressive T2 tongue SCCs.

Tumor hypoxia has been associated with poorer outcomes in patients with head and neck carcinomas.24, 25, 26 Among the endogenous hypoxia markers used to evaluate microenvironmental conditions of head and neck tumors are HIF, CA, GLUT, and EPOR, the overexpression of which has been associated with solid tumor aggressiveness. These hypoxia markers have also been used to noninvasively identify tumors showing poor response to radiotherapy and/or chemotherapy.9 We found that, of these markers, only EPOR was a significant prognostic indicator in patients with T2 oral tongue SCC. The same results can be found in the previous reports showing tumor expression of EPOR expression adversely affected prognosis of the patients with head and neck or esophageal cancers.27, 28 There have been, however, the actual controversies regarding the role of erythropoietin in cancer treatment because the addition of erythropoietin did not improve survival outcomes of head and neck cancer patients, particularly receiving radiotherapy.29, 30, 31 In addition, GLUT-1 expression was associated with nodal metastases, suggesting that the latter may be used as a potential predictor of nodal metastases in patients with oral tongue cancer.

To our knowledge, this study is the first to evaluate the utility of hypoxia markers as prognostic predictors of oral tongue SCC. Our results, however, may be limited by the small number of patients, the inclusion only of patients with T2 oral tongue SCC, and the retrospective study design. These limitations may be associated with the absence of statistical significance for any clinicopathologic factors, in contrast to previous findings.4, 5, 6, 7, 14, 15, 16 The inclusion of only patients with T2-staged tumors, however, may enhance our understanding of the unique biological behavior of these tumors.

In conclusion, we have shown here that EPOR expression may be an independent prognostic predictor for patients with T2-staged SCC of the oral tongue, with a high propensity of regional metastasis and moderate survival rates.

Conflicts of Interest Statement 

None declared.