SEARCH

SEARCH BY CITATION

Keywords:

  • glucose transport;
  • oral carcinoma;
  • prognosis;
  • [18F]-2-fluoro-2 deoxy-D-glucose;
  • positron emission tomography;
  • Glut-1 expression

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

BACKGROUND

The overexpression of glucose transporters, especially of Glut-1, is a common characteristic of human malignancies, including head and neck carcinoma. Recently, the assessment of glucose metabolism in the tumor with [18F]-2-fluoro-2 deoxy-D-glucose (FDG) and positron emission tomography (FDG-PET) has been used to identify particularly aggressive tumors. The authors tested the hypothesis that both glucose transport and its metabolism play a key role in the progression of oral squamous cell carcinoma (OSCC).

METHODS

Retrospective analysis of Glut-1 expression was performed by immunohistology in 118 patients with OSCC, and a Glut-1 labeling index (LI) was established for each. A separate group of 44 patients with primary OSCC was evaluated prospectively by FDG-PET prior to surgery. To link the expression of Glut-1 with glucose metabolism, both FDG-PET and immunohistology were determined in a subgroup of 31 patients, and the results were correlated with overall survival.

RESULTS

The patients who had OSCC with a low LI for Glut-1 survived significantly longer compared with patients who had OSCC with a high LI (138 months vs. 60 months; P = 0.0034). It was found that Glut-1 expression was an independent marker of prognosis in patients with OSCC. In patients who were evaluated by FDG-PET, the standardized uptake value (SUV) below the median split value of 5.6 was predictive of a longer survival (P < 0.027), whereas an SUV > 5.6 was associated with an increased hazard of death. In combination, a high Glut-1 level and a high SUV predicted shorter survival (P < 0.005) for patients with OSCC. Patients who achieved a complete response to preoperative radiation tended to have tumors with low glucose metabolism, as defined by both the Glut-1 LI and the SUV.

CONCLUSIONS

Both glucose transport and glucose metabolism determine the glycolytic tumor phenotype, which is a significant negative biomarker of prognosis and overall survival in patients with OSCC. Cancer 2003;97:1015–24. © 2003 American Cancer Society.

DOI 10.1002/cncr.11159

Since the basic studies of Warburg,1 increased glycolytic metabolism has been established as a characteristic of malignant cells, although some neoplasms never increase glucose uptake, and others may contain cells with a variable glycolytic phenotype. Among the mechanisms that contribute to the glycolytic phenotype, overexpression of glucose transporters, and especially of Glut-1, has been reported for a large variety of tumors,2, 3 including squamous cell carcinoma of the head and neck (SCCHN).4, 5 Glut-1, a glucose transporter, belongs to the sugar porter family, which currently includes 133 individual members. Glut-1 is not the only glucose transporter expressed in SCCHN, but it is a dominant one. All of these proteins share a 12-transmembrane-spanner (TMS) protein topology, which is believed to be a result of intragenic duplication of a primordial 6-TMS unit.6 In vitro, Glut-1 overexpression is an early event after ras and src transformation,7 and a promotive effect of ras through the HIF-1α-binding site of the Glut-1 promoter region recently has been identified.8

Although the metabolic consequences of increased glucose transport still are not understood completely, a clinical significance of glucose transporter expression has been suggested for several human solid tumors. In patients with nonsmall cell lung carcinoma,9 colorectal carcinoma,10 and gastric carcinoma,11 overexpression of Glut-1 was associated with enhanced tumor aggressiveness and poor survival. Although Glut-1 overexpression is quite common in patients with SCCHN, the prognostic value of this parameter has not been analyzed systematically for this tumor type. To the best of our knowledge, only Baer and coworkers,12 considered the prognostic significance of Glut-1 in head and neck carcinoma (HNC). By and large, however, the significance of Glut-1 overexpression in oral carcinomas remains unexplored.

Clinical introduction of positron emission tomography (PET) provided a means for the noninvasive quantitative assessment of tumor glucose metabolism in vivo using [18F]-2-fluoro-2-deoxy-D-glucose (FDG).13 The in vivo uptake level of glucose is a prognostic marker in patients with lung and breast carcinoma.14–17 Several recent studies have explored the prognostic value of FDG-PET in patients with HNC.18–21 These studies suggest that FDG-PET may be useful in identifying aggressive tumors, which require intensive treatment protocols. However, there is a need to better understand why some tumors do not seem to be FDG avid and to understand the clinical meaning of high uptake values.

In this study, the hypothesis was tested that Glut-1 expression and FDG-PET scans used together have a significant prognostic value in patients with oral squamous cell carcinoma (OSCC). The study was designed initially to measure Glut-1 expression with immunologic techniques in archival specimens of OSCC. Based on a Glut-1 labeling index (LI), a cut-off value was established that divided the patients into those who survived for 138 months (low LI) and those who survived for 60 months (high LI). In a subsequent prospective study, this cut-off value was used together with the results of FDG-PET scans to establish the glycolytic tumor phenotype. The findings verify the importance of this phenotype for predicting prognosis of patients with OSCC.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Immunohistochemistry

Specimens for immunohistochemistry

For immunohistochemical evaluation, the routinely processed paraffin blocks of formalin fixed SCHNN specimens were obtained from 30 female patients and 88 male patients who were seen at the Maxillofacial Unit of the University Hospital of Mainz between 1980 and 1993. The specimens were retrieved retrospectively from the pathology archives.

The criteria for inclusion in this study were as follows: a diagnosis of OSCC, no preoperative therapy, potentially curative radical resection, follow-up data available, and satisfactory tissue preservation (see below). In patients with incomplete resection margins (positive resection margins on histopathologic examination), lymphangiosis carcinomatosa, or extracapsular spread of involved lymph nodes, postoperative radiotherapy (60 grays [Gy]) was given. No other patients received postoperative therapy. The mean age was 58 years (range, 34–88 years). At the time of the survival analysis, 60 of 118 patients (51%) had died. The follow-up period for surviving patients ranged from 1 months to 172 months, with a median of 74 months. The sites of the primary tumors were as follows: maxilla, including the palate, 14 tumors; floor of the mouth, 50 tumors; tongue, 24 tumors; gingiva of the mandible, 19 tumors; retromolar trigone, 7 tumors; and all other locations, 9 tumors. The TNM staging categories were determined according to the criteria established by the American Joint Committee On Cancer and the International Union Against Cancer (UICC).22 Stage grouping of the patients was as follows: Stage I, 28 patients; Stage II, 26 patients; Stage III, 6 patients; and Stage IV, 58 patients). On histopathologic evaluation of lymph node involvement, 71 tumors were classified as N0, 10 tumors were classified as N1, and 37 tumors were classified as N2.

Immunohistochemical staining procedure

In every specimen, the quality of antigen preservation was assessed by immunostaining for vimentin.23 Sections were immunostained for Glut-1 using an avidin-biotin technique. Briefly, 6-μm sections were deparaffinazed, rehydrated, and microwaved for 3 cycles of 5 minutes each (600 W) in 1 mM ethylene diamine tetraacedic acid buffer, pH 8.0, for antigen retrieval. When they were cooled to room temperature, sections were quenched for endogenous peroxidase activity by 0.1% H2O2 (20 minutes), then incubated with 10% normal goat serum (30 minutes) to block nonspecific binding of immunoglobulins to the tissue, followed by incubation with the polyclonal rabbit antihuman Glut-1 antibody (MYM AB 1351; Chemicon, Temecula, CA) at a final dilution of 1:1000 for 60 minutes at room temperature. Visualization involved the LSAB II® kit (DAKO Diagnostics) developed with peroxidase and 3,3′-diaminobenzidine. Nuclei were counterstained with hematoxylin. Negative controls were obtained by omitting the primary antibody. Erythrocytes, which were present in every section, served as internal controls for Glut-1 and for constant immunostaining intensity.

Evaluation of Stained Sections

Glut-1 immunostaining was evaluated by the same investigator, who was blinded to the clinical and follow-up data. The intensity of staining, cellular patterns of staining as well as numbers of positive cells were recorded for every specimen. For quantitative evaluation, a Glut-1 LI was established for each tumor based on the percentage of tumor cells that expressed the protein. For example, if 20% of tumor cells were positive, then the LI was 20. Based on the quantitative evaluation of Glut-1 LI in the entire population of 118 patients with OSCC, the tumors were subdivided in two groups: tumors with a low Glut-1 LI (< 50%) and tumors with a high Glut-1 LI (≥ 50%) (Fig. 1).

thumbnail image

Figure 1. The top histogram shows two distinct subpopulations based on the distribution of the Glut-1 labeling index (LI) in 118 patients with oral squamous cell carcinoma. (A,B) The paraffin sections demonstrate Glut-1 expression in representative areas of tumors in the low LI (< 50%) subgroup (A) and the high LI (≥ 50%) subgroup (B). Sections were immunostained for Glut-1 antibody using an avidin-biotin technique. Original magnification × 400.

Download figure to PowerPoint

FDG-PET Study Group

Patients

In the second study, 44 patients who underwent potentially curative surgical therapy for primary OSCC from 1995 to 1999 were included in FDG-PET evaluation. In addition, 31 of these patients provided tumors for Glut-1 immunostaining. All patients had an FDG-PET investigation as part of their routine preoperative staging procedure and had signed separate informed consent forms for the diagnostic procedure and for therapy. Inclusion criteria were as follows: carcinoma of the oral cavity, PET emission and transmission data available for quantitative assessment of glucose metabolism prior to therapy, potentially curative radical resection, and follow-up data available.

These patients were treated according to the Essen protocol,24 receiving preoperative radiation therapy (36 Gy) and subsequent radical resection with a safety margin of at least 10 mm in patients with T2–T4N0–N3M0 disease. All 44 patients completed primary treatment and were classified histopathologically as having undergone R0 resection. In patients with extracapsular spread of involved lymph nodes or carcinomatous lymphangiosis, postoperative radiotherapy was completed up to 60 Gy. For different reasons (patient objection, tumors initially classified as T1, small but intraosseously extending tumor [T4] in a dentigerous cyst), nine patients were not given preoperative radiation.

The mean patient age was 59 years (range, 35–80 years). At the time of survival analysis, 14 of 44 patients (31%) had died. The follow-up period for surviving patients ranged from 7 months to 60 months, with a median of 38 months. Tumors were localized as follows: maxilla (including the palate), 6 tumors; floor of the mouth, 13 tumors; tongue, 10 tumors; gingiva of the mandible, 5 tumors; retromolar trigone, 8 tumors; and all other locations, 2 tumors. Stage grouping of the patients were as follows: Stage I, 1 patient; Stage II, 5 patients; Stage III, 10 patients; and Stage IV, 28 patients. On histopathologic evaluation of lymph node involvement, 25 tumors were classified as N0, 6 tumors were classified as N1, and 13 tumors were classified as N2.

PET scanning

PET scans were performed on a Siemens ECAT EXACT 922® (Siemens/CTI, Knoxville, TN) whole-body camera. This device acquires 47 planes within an axial field of view of 16.2 cm, yielding an axial resolution of 5.2 mm full width at half maximum in the center of the field. The patients fasted for a minimum of 6 hours before the PET study. During and after administration of FDG, patients were advised to keep at rest; to avoid speaking; and to minimize swallowing for reduction of local, unspecific FDG uptake due to muscular activation. Emission measurements were obtained 45 minutes (± 15 minutes) after the administration of 370 megabecquerels (MBq) 18F-FDG, which lasted for 11 minutes per bed position. Three to four bed positions were scanned for imaging the viscerocranium, the neck, the thorax, and the epigastric region.

Transmission scanning was performed with a 68-germanium ring source for attenuation correction after the emission scanning. For quantitative evaluation, standardized uptake values (SUV) were calculated for a region of interest (ROI) of 1 cm in greatest dimension drawn at the site of maximum FDG uptake of the tumor according to the following equation: SUV = [A(ROI)] / [V(ROI)] × [BM] / [A(total)]25, where [A(ROI)] is the activity of the ROI (in MBq), [V(ROI)] is the volume of the ROI (in mL), [BM] is the body mass (in kg), and [A(total)] is the applied activity (in MBq).

Statistical Analysis

The log-rank test26 was used to obtain P values for the test of statistical significance for the comparison among groups in the univariate analysis. When the log-rank test was used, the analyses were stratified for tumor stage (UICC Stages I and II vs. UICC Stage III and IV). For the multivariate analysis, the P values were obtained from the Wald tests under the Cox model to test for any statistical significance of regression coefficients. All reported P values are two-sided. For determining survival curves, the Kaplan–Meier method was utilized. The definitions for the extent of disease were as follows: advanced disease, UICC Stages III and IV; early-stage disease, UICC Stages I and II.

In the analyses that were performed in the first study, using the 118 patients for whom Glut-1 staining was available, patients were categorized with a low LI (< 50%) or a high LI (≥ 50%). In addition, the Glut-1 LI also was considered as a continuous variable in the Cox regression model. A multivariate survival analysis was performed using the Cox proportional hazards model with the following dichotomized covariates as independent variables: Glut-1 LI (< 50% vs. ≥ 50%), UICC stage (Stages I and II vs. Stages III and IV), tumor grade (Grade 1 and Grade 2 vs. Grade 3), pattern of invasion (types I and II vs. types III and IV), age (< median vs. ≥ median), and peritumoral lymphatic infiltration (shallow vs. dense).

In analyses that were performed in the second study, using the 44 patients for whom FDG-PET results were available, patients were categorized into a low SUV group (SUV < median split value) or a high SUV group (SUV > median split value). The dichotomized covariates that were included in the multivariate analysis were SUV (< median split value vs. ≥ median split value), UICC stage (Stages I and II vs. Stages III and IV), age (< median split value vs. ≥ median split value), and preoperative irradiation (not administered vs. administered). Analyses exploring the linkage between Glut-1 expression and the specific uptake of FDG in 31 patients who were included in the second study used Spearman rank-based correlations and contingency tables after dichotomizing each variable followed by a Fisher exact test.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Glut-1 Expression in OSCC and Surrounding Peritumoral Tissue

The expression of Glut-1 was positive in all OSCC specimens that were examined as part of this study. Both the intensity of staining and the proportion of stained cells were evaluated and found to be highly variable among tumors. The LI, based on the percentage of positive cells, had a higher prognostic value compared with the staining intensity and was selected as the endpoint for data analysis. Figure 1 shows that it was possible to segregate the tumors into two groups, which were defined by high LI and low LI, using an LI of 50% as a benchmark. Various cellular patterns of Glut-1 staining were observed within individual tumors, including membranous, cytoplasmic, and intermediate Glut-1 expression. These patterns of staining were not associated with survival. Along with Glut-1 expression in tumor cells, specific Glut-1 expression was seen regularly in ductal epithelium of the major salivary glands, in the germinal zone of lymphatic follicles, in the perineurium of peripheral nerves, and in erythrocytes (Fig. 2).

thumbnail image

Figure 2. Immunohistochemistry for Glut-1 in peritumoral soft tissues. Expression of Glut-1 is seen in ductal epithelium of salivary glands (A), the germinal zone of lymphatic follicles (B), and the perineurium (C). Erythrocytes (D) constantly exhibit an intense Glut-1 staining and can be used as internal positive controls. Sections were immunostained for Glut-1 antibody using an avidin-biotin technique. Original magnification × 200 (A–C); × 400 (D).

Download figure to PowerPoint

Association Between Glut-1 Tumor Expression and Patient Survival

During follow-up, 9 of 31 patients (29%) in the low LI group died, and 51 of 87 patients (59%) in the high LI group died. Univariate estimation of the median survival according to a Kaplan–Meier plot was 138 months (95% confidence interval, 90–186 months) in the low LI group and 60 months (95% confidence interval, 43–77 months) in the high LI group. The difference in survival was highly significant (P = 0.0034; log-rank test). The data presented in Figure 3 suggest that a high Glut-1 LI predicts an increased hazard of death in patients with both advanced disease (i.e., UICC Stages III and IV: estimated hazard ratio, 2.6; P = 0.029) and patients with early-stage disease (i.e., UICC Stages I and II: estimated hazard ratio, 3.5; P = 0.045).

thumbnail image

Figure 3. Kaplan–Meier survival estimates in patients with oral squamous cell carcinoma (OSCC) according to high (> 50%) versus low (< 50%) Glut-1 labeling index (LI) scores. Both in patients with early (Stages I and II) OSCC (A) and in patients with advanced (Stages III and IV) OSCC (B), a high Glut-1 LI was associated with poor survival.

Download figure to PowerPoint

Glut-1 expression also was confirmed as a significant independent prognostic marker in a proportional hazards regression analysis in which the data were controlled for tumor stage, grading, pattern of invasion, age, and density of peritumoral lymphocytes. High Glut-1 expression was assigned a hazard ratio of 2.65 (P = 0.010; Wald test). Hazard ratios, 95% confidence intervals, and P values from the Wald test for all variables are shown in Table 1. When covariables with P > 0.1 were eliminated, three factors of significant predictive value remained: tumor stage (hazard ratio, 1.99; P = 0.013), patient age (hazard ratio, 2.45; P = 0.001), and Glut-1 expression (hazard ratio, 2.74; P = 0.007). Glut-1 LI and T status as well as N status also were investigated separately after dichotomizing the T-status variable into T1, T2–T3, and T4 and the N-status variable into N0–N1 and N2. The simple, one-sided t tests were performed for the alternative hypothesis that the true difference in the mean Glut-1 values between early and late T or N stages was < 0. We found a strong, positive correlation of Glut-1 LI with T status (P = 0.002) but no significant relationship with N status (P = 0.12).

Table 1. Analysis of Glut-1 Overexpression in Tumors Controlled for Other Prognostically Important Variables in Patients with Oral Squamous Cell Carcinoma
VariableDefinition of groups (Group 1 vs. Group 2)HR (Group 2 relative to Group 1)95%CIP value (Wald)
  1. HR: hazard ratio; 95%CI: 95% confidence interval; LI: labeling index.

Tumor stageStage I–II vs. III–IV2.101.18–3.750.012
GradeGrade 1–2 vs. 31.480.79–2.770.220
Pattern of invasionTypes I–II vs. III–IV0.740.42–1.300.300
Peritumoral lymphocytesShallow vs. dense1.200.69–2.060.520
Age< Median vs. ≥ median2.441.41–4.230.002
Glut-1 (LI)< 50% vs. ≥ 50%2.651.23–5.590.010

The Glut-1 LI also was considered as a continuous variable in the Cox regression model and was identified as a significant prognostic marker of survival, with a P value of 0.02. Thus, an LI of 80% can be considered a worse prognostic characteristic compared, for example, with an LI of 60%. The Cox regression model that was used after categorizing LI by its quartile values also yielded a hazard ratio of 1.2 with P < 0.04. Overall, these data indicate that Glut-1 may be considered a negative marker of prognosis in patients with OSCC, with high Glut-1 expression associated with poor survival, and vice versa.

Association Between Tumor Glucose Metabolism and Patient Survival

In the second study, FDG-PET was used to clearly identify each tumor site. SUVs ranged from 2.7 to 14.9. The median split SUV value was 5.6. Thus, the low SUV group was defined by the uptake below the median split value, and the high SUV group was defined by the uptake above the median split value. In the follow-up period, 3 of 22 patients in the low SUV group died, and 11 of 22 patients in the high SUV group died. The 3-year survival rates were 85% for the low SUV group and 61% for the high SUV group. The estimated survival rates were reduced significantly in the high SUV group, with the difference remaining statistically significant when patients were stratified according to tumor stage (P = 0.027; log-rank test). Representative examples of high and low glucose metabolism as well as survival curves according to the Kaplan–Meier plot are shown in Figures 4 and 5, respectively. Because only 6 of 44 patients had UICC Stage I–II disease, the survival curves shown in Figure 5 were plotted without stratification for tumor stage.

thumbnail image

Figure 4. In vivo glucose metabolism in patients with oral squamous cell carcinoma as measured by [18F]-2-fluoro-2 deoxy-D-glucose (FDG) with positron emission tomography. Representative examples show tumors with low (left) and high (right) mean standard uptake values (SUV) calculated for a region of interest of 1 cm in greatest dimension drawn at the site of maximum FDG uptake of the tumor.

Download figure to PowerPoint

thumbnail image

Figure 5. Kaplan–Meier survival curves were generated by univariate analysis to determine the standardized uptake value (SUV) status. The data suggest that glucose metabolism measured by [18F]-2-fluoro-2 deoxy-D-glucose with positron emission tomography has a significant predictive value for overall survival in patients with oral squamous cell carcinoma (P = 0.038; log-rank test).

Download figure to PowerPoint

In the multivariate proportional hazards regression analysis, which controlled for tumor stage, preoperative radiation, and patient age, high SUV was ascribed a hazard ratio of 5.13, implying a prognostic significance at the P = 0.017 level. Hazard ratios, 95% confidence intervals, and P values of the Wald test for all variables are shown in Table 2. When covariables with P > 0.1 were eliminated, two factors of significant predictive value remained: preoperative radiation (hazard ratio, 3.42; P = 0.054), and glucose metabolism (hazard ratio, 5.0; P = 0.018). The relation between SUV and T status as well as N status also were analyzed separately, similar to what is described for Glut-1 above. SUV showed a strong positive correlation with T status (P = 0.013) and a weaker but positive correlation with N status (P = 0.052).

Table 2. Analysis of [18F]-2-Fluoro-2 Deoxy-D-Glucose Uptake by the Tumor Controlled for Tumor Stage, Preoperative Radiation, and Patient Age
VariableDefinition of groups (Group 1 vs. Group 2)HR (Group 2 relative to Group 1)95%CIP value (Wald)
  1. HR: hazard ratio; 95%CI: 95% confidence interval; SUV: standardized uptake value.

Tumor stageStage I–II vs. III–IV2.430.31–19.050.400
Preoperative radiationNo vs. yes3.010.85–10.680.088
Age< Median vs. ≥ median1.620.55–4.810.380
SUV< Median vs. ≥ median5.131.34–19.650.017

Glut-1 Expression and FDG-PET Results

In a subset of 31 patients who were involved in the second study and who had both Glut-1 expression and SUV measurements, the correlation between these two parameters was investigated. Using three different analyses (Spearman rank test and contingency tables for quartiles of each variable or for high and low values of each variable), no significant correlation between the two variables could be established (Table 3), with P values ranging from 0.3 to 0.86.

Table 3. A Lack of Association between the Glut-1 Labeling Index and Results of [18F]-2-Fluoro-2 Deoxy-D-Glucose and Positron Emission Tomography Scans in a Subgroup of 31 Patients with Oral Squamous Cell Carcinoma who were Evaluated for Both Markersa
SUVSUVP value
LowHigh
  • a

    This is a 2 × 2 contingency table listing the numbers of tumors that were categorized according to high and low values for both markers, as described in the text. The Fisher exact test was used to calculate the P value.

Low118< 0.47
High57

Survival analyses also were performed with individual variables as well as both variables included in the model. Using the log-rank test, SUV alone was a marginally significant prognostic factor for survival (P = 0.056). Comparing low (L) and high (H) values for Glut-1 LI and SUV (i.e., LL, LH, HL and HH), the P value was 0.051, and the significance was attributable mainly to the difference between L SUV values and H SUV values. The log-rank test also was used to compare HH values vs. combined (LL, LH, and HL) values to determine whether high values for both Glut-1 LI and SUV have prognostic significance for survival. In this test, the P value was highly significant at P = 0.008, suggesting that a combination of LI and SUV has a negative prognostic value when patients have both high Glut-1 and high SUV.

Glut-1 Expression, Glucose Metabolism, and Response to Radiation

In addition, we evaluated response to radiation in a subset of patients with advanced T3 and T4 tumors who had received preoperative radiation therapy with 36 Gy (n = 26 patients). Because all of these patients subsequently underwent complete resection of their tumors, the histologic parameters of tumor response to radiation (i.e., complete tumor devitalization) could be determined for each patient. PET scans were obtained from all 26 patients, and the Glut-1 LI was available in 22 tumors. All 4 complete responders to 36 Gy radiation initially had a Glut-1 LI < 50%; whereas 11 of 18 tumors with residual, viable, malignant cells (i.e., incomplete radiation responders) belonged to the high Glut-1 LI group (≥ 50% vs. < 50%; P = 0.005; Fisher exact test). The predictive effect was not as pronounced when in vivo glucose metabolism was evaluated by PET. Low SUVs were seen in 4 of 5 complete responders, whereas 12 of 21 patients with residual tumor after radiation belonged to the high SUV group (P = 0.132; Fisher exact test).

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

The current study examined two phenomena associated with the glycolytic phenotype of solid tumors: the enhanced expression of Glut-1 and the increased glucose metabolism within tumor cells. Previously, both parameters were investigated separately in a variety of tumors, and each has been assigned a prognostic significance.9, 10, 14–16, 20, 21, 27 Consistent results for both immunohistologic and in vivo data in a single tumor entity would provide additional strong evidence for the biologic and clinical relevance of glucose transport and metabolism in tumor progression. To this end, we set out to determine whether concomitant measurements of Glut-1 and SUV in a tumor could serve as a surrogate of survival.

The results of the initial retrospective study confirmed the expression of Glut-1 in all examined patients with OSCC. Although a wide range of the Glut-1 LI (5–100%) was observed in patients with oral carcinoma, the tumors could be divided naturally into two distinct groups based on low proportions of Glut-1 positive cells in the tumor (i.e., < 50%) or high proportions of Glut-1 positive cells in the tumor (i.e., ≥ 50%). This split reflected the metabolic heterogeneity of tumors, with the majority containing high Glut-1 LI. We evaluated the functional significance of this phenomenon by relating Glut-1 expression to clinical outcome and found that, in both univariate and multivariate survival analyses, a high Glut-1 LI predicted shorter survival. Our results agree with those described for patients with other types of solid tumors11, 28, 29 and agree with the findings of Baer et al.,12 who recently reported that glucose transporter expression was associated with poor survival in 44 patients with laryngeal carcinoma who were evaluated by immunohistochemistry of tumor biopsies. Unfortunately, the survival analysis by those authors was not adequate, because it lacked stratification into tumor stage, therapy, or follow-up.12 By contrast, our study included a larger, homogeneous population of 118 patients OSCC and made use of long-term follow-up, data with < 8% of patients lost to follow-up within the first 24 months. The criteria established for Glut-1 evaluation were stringent, including an objective examination of Glut-1 staining without any knowledge of the clinical data, confirmed antigen preservation by vimentin staining in every specimen, and careful selection of the antibody specific for Glut-1 as well as its titration. This assures methodologic integrity of the approach, which is necessary for the meaningful interpretation of immunohistology results.

Our immunohistologic results derived from two independent studies strongly support the significance of Glut-1 expression as a negative biomarker of prognosis in patients with OSCC: A high Glut-1 LI was associated with an increased hazard of death. Further support for this conclusion was provided by our PET results, in which a survival analysis with regard to glucose metabolism indicated that the patients with low SUV on FDG-PET scans survived longer compared with patients with high SUV. Thus, FDG uptake also was an independent predictor of overall survival in patients with OSCC. Combining the two markers appeared to improve the predictive potential significantly, particularly when both Glut-1 LI and SUV were high. It is interesting to note, however, that not all patients conform to the general rule, in that prognosis is not apparently worse in some patients with high Glut-1 LI and high SUV, and outcome is not favorable for some patients with low values. Such exceptions may be due to the influence of factors other than glucose metabolism on tumor-host interactions. For example, it has been shown that the immunologic competence of the patient, as exemplified by the presence of functional T-cells or dendritic cells in the tumor bed, is a strong, independent predictors of overall survival in patients with OSCC.30, 31

Two earlier studies were conducted that correlated the in vivo evaluation of glucose metabolism with clinical outcome.18, 19 In a study of 37 patients with squamous cell carcinoma of the oral cavity, the larynx, the hypopharynx, the nasopharynx, and the parotid gland, Minn et al. observed an association between high FDG uptake and poor survival.18 By contrast, Rege and coworkers reported improved response to radiation therapy and improved survival in the high SUV subpopulation.19 However, their study included only 12 patients with tumors located at different sites and 2 deaths occurred that in patients who had maxillary sinus carcinomas, who are known to have a different prognosis compared with patients who have carcinomas of the oral cavity. Neither our data nor those from two recently published studies on FDG-PET in patients with HNC20, 21 support the findings of Rege et al.19 Those two studies both found that high FDG uptake appears to be a useful parameter for identifying patients with HNC who have aggressive tumors, requiring intensive treatment protocols. Our results are consistent with these previous findings. However, we have no evidence for increased susceptibility of tumors with high FDG uptake to radiation therapy. This result appears to be counterintuitive, because tumors with high metabolic rates and actively dividing cells would be expected to be more susceptible to radiation therapy compared with tumors that have low rates. Conversely, it is possible that increased Glut-1 expression may be responsible for the protection of cells from hypoxia-induced apoptosis.32

Taken together, our results based, on immunohistologic and in vivo PET analysis in two distinct but homogeneous groups of patients with OSCC, provide independent lines of evidence in support of the hypothesis that glucose transport and glucose metabolism play a key role in oral carcinoma progression. In addition to providing crucial prognostic information through the noninvasive detection of glucose metabolism by routine FDG-PET investigations prior to therapy, glucose transport and metabolism may be promising future targets for novel therapeutic interventions. In this context, Haberkorn and coworkers recently demonstrated that inhibition of glucose transport by cytochalasin-B significantly increased gemcitabine-induced apoptosis in hepatoma cells.33 If this approach or similar other therapeutic approaches prove effective in further preclinical studies, then inhibitors of glucose transport may become clinically useful as adjuvants to the routine treatment of patients with oral carcinoma.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES