Abstract
Purpose
Functional DNA polymorphisms affecting gene expression and serum or saliva levels of interleukins IL-1β,-4,-6,-8,-10 and tumor necrosis factors TNF-α,-β have been associated with increased risk for the development of oral squamous cell carcinoma (OSCC). The present retrospective case–control study examines possible interactions between seven cytokine genotype polymorphisms and their combinatory effect in predicting the occurrence of OSCC in Caucasians.
Methods
Three hundred and thirty Greeks and Germans were studied, consisting of 162 OSCC cases and 168 healthy controls of comparable age, gender, and ethnicity. A series of multivariate logistic regression models, adjusted for age and gender, was constructed in order to assess the contribution of homozygous or heterozygous variant genotypes of polymorphisms IL-1β (+3953C/T), IL-4 (-590C/T), IL-6 (-174G/C), IL-8 (-251A/T), IL-10 (-1082A/G), TNF-α (-308G/A) and TNF-β (+252G/A) upon overall, early and advanced stages of OSCC development.
Results
The contribution of TNF-α and IL-6 was consistent and robust in almost all models constructed. Furthermore, when the mode of inheritance of each variant allele was taken into account in a “biological” multivariate logistic regression model, four polymorphisms emerged as primary predictors for overall stages of OSCC: TNF-α (OR = 15.27; 95% CI = 7.30–31.96), IL-6 (OR = 8.33; 95% CI = 3.95–17.58), IL-8 (OR = 3.54; 95% CI = 1.69–7.43) and IL-10 (OR = 2.65; 95% CI = 1.28–5.46). Finally, IL-1β, IL-4 and TNF-β polymorphisms were not primary predictors of OSCC development in all constructed models.
Conclusions
This study revealed the highly significant contributions of two out of seven studied cytokines (IL-6 and TNF-α) in the occurrence of OSCC. Based on these findings and previous reports, possible stoichiometrical interactions of cytokines leading to OSCC development are discussed.
Keywords: Oral cancer, Single nucleotide polymorphisms, Cytokines, Interleukins, Tumor necrosis factors
Introduction
Oral squamous cell carcinoma (OSCC) encompasses more than 90% of all oral malignancies (Neville and Day 2002; Vairaktaris et al. 2006, 2007a, b, c, 2008; Yapijakis et al. 2008). OSCC remains the eighth most common malignancy in the world, and is a notable cause of morbidity and mortality worldwide (Williams 2000; Massano et al. 2006). Carcinogenesis in the oral region is a multistep process in which mutations in oncogenes and tumor suppressor genes, as well as environmental factors such as smoking, alcohol abuse and certain viruses are involved (Williams 2000; Neville and Day 2002).
In European Union countries, such as Belgium, Denmark, Greece, Portugal, and UK, a rising incidence of OSCC cases has been documented (Massano et al. 2006). Moreover, a rise in mortality rates attributed to OSCC during the past two decades has been observed in Eastern European countries (Massano et al. 2006). Such epidemiological observations suggest that the current clinical screening protocol, which includes thorough medical history, as well as extra- and intra-oral examination, seems to be inadequate for prevention (Neville and Day 2002). As a result, novel strategies for preventing oral cancer should focus on detection of individuals at risk for developing OSCC, allowing greater intervention and follow up, even before suspicious oral lesions appear.
Single nucleotide polymorphisms (SNPs) in genes encoding for susceptibility factors may influence gene expression, protein function and disease predisposition in certain individuals (The International HapMap Project 2003). Association studies have shown great promise for linking such subtle but also functional genomic differences with increased risk for specific diseases (Song et al. 2001; Loktionov et al. 2003; Bergman et al. 2005; Xu et al. 2005; Jin et al. 2005; Matsumura et al. 2005).
Recently, certain functional polymorphisms influencing expression of genes encoding for cytokines and other factors implicated in inflammation, immune response, angiogenesis and thrombosis, have been correlated with an increased risk for developing oral malignancies, by our group and others (Tsai et al. 2005; Ku et al. 2005; Boccia et al. 2006; Wong et al. 2006; Vairaktaris et al. 2006, 2007a, b, c, 2008; Yapijakis et al. 2008). Such association studies confirm previous observations that serum levels of inflammatory cytokines (including tumor necrosis factor alpha and beta [TNF-α and TNF-β], interleukin [IL]-6, IL-8 and members of the IL-1 group) and anti-inflammatory cytokines (such as IL-10, IL-4) seem to be correlated with carcinogenesis through multiple and often controversial pathways, therefore they may have a prognostic significance in OSCC and other types of cancer (Myers et al. 1996; Jablonska et al. 1997; Fujieda et al. 1999; Berghella et al. 2002; Watanabe et al. 2002; Lejeune 2002; St John et al. 2004). However, it is difficult to assess which polymorphic alleles are of primary importance without the concurrent analysis of all existing data and possible combinatory effects.
The present analysis consists of a retrospective case–control study for investigating the combinatory effects of SNPs affecting the gene expression of serum cytokines upon the occurrence of OSCC. In the same core population of patients with OSCC and healthy controls the following polymorphisms were investigated: IL-1β (+3953 C/T), IL-4 (-590 C/T), IL-6 (-174 G/C), IL-8 (-251 A/T), IL-10 (-1082 A/G), TNF-α (-308 G/A) and TNF-β (+252 G/A). All examined polymorphisms are quite common and informative (with allele frequency range of 10–52% in Caucasians), and known to influence gene transcription, leading to specific changes in serum levels of cytokines (Vairaktaris et al. 2006, 2007a, b, c, 2008; Yapijakis et al. 2008).
The primary objective of the present multivariate logistic regression analysis is to assess possible interactions between cytokine genotype polymorphisms and their combinatory effect in predicting the occurrence of OSCC, in an effort to identify individuals at risk. The secondary objective of the present study is to evaluate how the aforementioned predictive model for OSCC varies according to overall, early, and advanced cancer stages. It is important to note that both aforementioned study objectives attempt to elucidate the underlying biological predisposition model, and not an environmentally influenced model, for predicting the occurrence of OSCC according to cytokine genotype polymorphisms.
Materials and methods
Study population
A retrospective case–control study design was applied for the present analysis. The sample population (N = 330) included Caucasians of Greek and German origin, as described in our previous studies (Vairaktaris et al. 2006; 2007a, b, c, 2008; Yapijakis et al. 2008). The case group consisted of 162 patients, who were diagnosed with OSCC, based on histological confirmation and operated between 1 September 1996 and 30 June 2006 at the University Departments of Oral and Maxillofacial Surgery in Athens, Greece and in Nürnberg, Germany. All cases were divided into two subgroups according to OSCC status: (1) patients with early cancer stage (including stages I and II) and (2) patients with advanced cancer stage (including stages III and IV), according to the American Joint Committee for Cancer Staging and End-Results reporting, 1992. The control group included 168 healthy blood donors who were recruited among health care professionals and volunteers. Controls were matched to cases with regard to ethnicity, gender, age, and a low-risk working environment.
The protocol applied was approved by the ethical review boards of the Departments of Oral and Maxillofacial Surgery of the Universities of Athens and Erlangen. All participants under study signed a detailed informed consent form.
Inclusion and exclusion criteria
Since genotype frequencies may vary among ethnic groups, all subjects who formed the cohort of OSCC cases and controls were unrelated adults of Caucasian European descent. Healthy individuals with a positive history of cancer were excluded from serving as controls. All included cases presented with primary squamous cell carcinoma of the oral cavity. Cases with a former positive history of other type of cancer or with oral histological lesions other than OSCC were also excluded.
Genotyping
Blood samples were collected from all participating individuals, and genomic DNA was isolated (Vairaktaris et al. 2006, 2007a, b, c, 2008; Yapijakis et al. 2008). Molecular detection of SNPs for IL-1β (+3953 C/T), IL-4 (-590 C/T), IL-6 (-174 G/C), IL-8 (-251 A/T), IL-10 (-1082 A/G), TNF-α (-308 G/A) and TNF-β (+252 G/A) genes was performed blindly by restriction fragment length polymorphism (RFLP) typing of DNA samples. This involved a combination of PCR amplification and digestion with the appropriate restriction endonuclease, followed by gel electrophoretic analysis. The PCR conditions, the sequence of primer pairs and the restriction enzymes used, have been previously described (Vairaktaris et al. 2006, 2007a, b, c, 2008; Yapijakis et al. 2008). Many of the samples were tested twice for verification of the results of molecular analysis.
The less common alleles of all cytokine polymorphisms, which are referred as “variant” alleles in this text, confer a high gene expression phenotype in almost all cases. The only exception is the IL-6-174C allele, which increases gene expression conditionally in case of an inflammatory stimulus (Vairaktaris et al. 2006).
Statistical analyses
The obtained genotyping data have been presented in previously published genetic association studies (Vairaktaris et al. 2006, 2007a, b, c, 2008; Yapijakis et al. 2008). Statistical comparisons were performed using the SAS version 9.0 (SAS Institute Inc., USA) software package. The Fisher’s exact test was applied in order to compare the genotype distribution and variant allele frequencies of each cytokine polymorphism observed between the two subcategories, according to ethnic status, of the control group. The Fisher’s exact test was also applied in order to compare the genotype and allele frequencies between the control group and all case categories (i.e. all OSCC cases, cases with early stage OSCC, and cases with advanced stage OSCC). The Mantel–Haenszel method was applied for the calculation of age and gender adjusted odds ratios (OR) and their respective 95% confidence intervals (CI). A P value of ≤0.05 was considered as the criterion for statistical significance for all analyses.
Pearson correlation coefficients (r) were calculated between all dichotomous risk factor variables (i.e. age ≥ 55 years, and gender) and OSCC stages. In addition, Pearson correlation coefficients were also calculated between all carrier, variant, recessive, and dominant cytokine genotype polymorphisms and OSCC stages. Moreover, in order to evaluate possible interactions and potential confounding effects, Pearson correlation coefficients were also calculated between all genotype polymorphisms for all cytokines.
Stepwise backwards unconditional multivariate logistic regression analyses were applied in order to evaluate the contribution of selected demographic variables (such as age and gender) and cytokine genotypes upon the risk for presenting with early, advanced, or any stage of OSCC, respectively. A statistical significance criterion of P < 0.05 was applied for the variables retained in the final multivariate logistic regression model. Multivariate logistic regression analyses were also applied in order to calculate OR and 95% CI of the variables retained in the final regression models whereas controlling for age and gender. All multivariate models included the matching terms age (binomial) and gender.
Nine separate multivariate logistic regression analyses were performed comparing the variant homozygous and heterozygous genotypes versus the homozygous wild type genotypes. These analyses included the cancer stage outcome (overall, early and advanced stages) for each of the three following models: (1) a “homozygous variant model” using homozygous variant genotypes, (2) a “carrier model” using heterozygous variant/wild type genotypes and (3) a “biological model” using either homozygous variant/variant or a combination of homozygous variant/variant and heterozygous variant/wild genotypes according to the recessive or dominant character of the variant allele of each cytokine polymorphism. The mode of inheritance of the variant allele has been previously determined by the significant difference between cases and controls with regard to either only homozygous variant genotypes (recessive) or both homozygous variant and heterozygous genotypes (dominant) (Table 3). The risk for OSCC development, as well as the risk for presenting with early or advanced stage OSCC, was calculated separately. The Hosmer and Lemeshow Goodness of Fit test was applied in order to quantify the chi-square value and statistical significance of all regression models constructed. Data on environmental risk factors such as smoking and alcohol consumption were not available for all healthy controls, due to the retrospective nature of the study. Since the main objective of the present analysis was to assess the underlying biological predisposition model for OSCC occurrence, these factors were not included in the multivariate regression models constructed.
Table 3.
Cytokine genotypes of healthy controls and patients and their subgroups with early and advanced stages of OSCC (Vairaktaris et al. 2006, 2007a, b, c, 2008; Yapijakis et al. 2008)
| Cytokine genotypes | Controls (n = 168) | Patients with OSCC | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Overall stages (n = 162) | Early stage (n = 90) | Advanced stage (n = 72) | ||||||||
| N (%) | N (%) | P* | ORa (95% CI) | N (%) | P* | ORa (95% CI) | N (%) | P* | ORa (95% CI) | |
| IL-1β + 3953 | ||||||||||
| T/T | 15 (9.6) | 10 (9.3) | 0.671 | 0.83 (0.33–2.07) | 6 (10.0) | 1.000 | 0.94 (0.32–2.81) | 4 (8.3) | 0.781 | 1.01 (0.29–3.57) |
| C/C | 81 (51.9) | 66 (61.1) | 1.00 (reference) | 38 (63.3) | 1.00 (reference) | 28 (58.3) | 1.00 (reference) | |||
| C/T | 60 (38.5) | 32 (29.6) | 0.138 | 0.61 (0.35–1.08) | 16 (26.7) | 0.104 | 0.52 (0.25–1.05) | 16 (33.3) | 0.489 | 0.77 (0.37–1.60) |
| IL-4 − 590 | ||||||||||
| T/T | 15 (9.3) | 26 (16.7) | 0.057 | 1.54 (0.62–3.85) | 24 (27.9) | <0.001 | 3.26 (1.13–9.41) | 2 (2.9) | 0.097 | 0.40 (0.09–1.69) |
| C/C | 99 (61.1) | 84 (53.8) | 1.00 (reference) | 36 (41.9) | 1.00 (reference) | 48 (68.6) | 1.00 (reference) | |||
| C/T | 48 (29.6) | 46 (29.5) | 0.703 | 1.31 (0.76–2.26) | 26 (30.2) | 0.209 | 1.61 (0.84–3.09) | 20 (28.6) | 0.753 | 1.15 (0.57–2.33) |
| IL-6 − 174 | ||||||||||
| C/C | 6 (3.8) | 18 (11.1) | <0.001 | 4.62 (1.46–14.55) | 10 (11.1) | 0.003 | 4.08 (1.17–14.23) | 8 (11.1) | <0.001 | 11.57 (2.37–56.56) |
| G/G | 90 (57.7) | 42 (25.9) | 1.00 (reference) | 28 (31.1) | 1.00 (reference) | 14 (19.4) | 1.00 (reference) | |||
| G/C | 60 (38.5) | 102 (63.0) | <0.001 | 3.38 (2.06–5.56) | 52 (57.8) | <0.001 | 2.54 (1.41–4.57) | 50 (69.5) | <0.001 | 5.16 (2.46–10.84) |
| IL-8 − 251 | ||||||||||
| A/A | 0 (0.0) | 14 (8.9) | <0.001 | 20.68 (3.57–119.81) | 14 (16.3) | <0.001 | 48.48 (7.49–313.81) | 0 (0.0) | NC | NC |
| T/T | 84 (53.8) | 56 (35.4) | 1.00 (reference) | 26 (30.2) | 1.00 (reference) | 30 (41.7) | 1.00 (reference) | |||
| T/A | 72 (46.2) | 88 (55.7) | 0.011 | 1.74 (1.08–2.78) | 46 (53.5) | 0.015 | 1.89 (1.05–3.40) | 42 (58.3) | 0.117 | 1.52 (0.85–2.74) |
| IL-10 − 1082 | ||||||||||
| G/G | 0 (0.0) | 2 (1.4) | 0.137 | 7.13 (0.33–155.37) | 0 (0.0) | NC | NC | 2 (3.1) | 0.038 | 15.95 (0.71–359.28) |
| A/A | 81 (57.4) | 46 (31.9) | 1.00 (reference) | 28 (35.0) | 1.00 (reference) | 18 (28.1) | 1.00 (reference) | |||
| A/G | 60 (42.6) | 96 (66.7) | <0.001 | 3.05 (1.84–5.05) | 52 (65.0) | 0.002 | 2.53 (1.40–4.55) | 44 (68.8) | <0.001 | 3.48 (1.71–7.09) |
| TNF-α − 308 | ||||||||||
| A/A | 13 (8.5) | 75 (46.9) | <0.001 | 17.58 (8.63–35.78) | 44 (49.4) | <0.001 | 16.11 (7.44–34.87) | 31 (43.7) | <0.001 | 21.65 (8.30–56.45) |
| G/G | 121 (79.1) | 36 (22.5) | 1.00 (reference) | 20 (22.5) | 1.00 (reference) | 16 (22.5) | 1.00 (reference) | |||
| G/A | 19 (12.4) | 49 (30.6) | <0.001 | 7.47 (3.76–14.86) | 25 (28.1) | <0.001 | 6.62 (3.03–14.48) | 24 (33.8) | <0.001 | 10.88 (4.29–27.58) |
| TNF-β + 252 | ||||||||||
| G/G | 3 (2.0) | 82 (51.3) | <0.001 | 59.03 (19.99–174.29) | 52 (59.1) | <0.001 | 70.32 (22.51–219.68) | 30 (41.7) | <0.001 | 43.85 (12.24–157.08) |
| A/A | 108 (70.6) | 28 (17.5) | 1.00 (reference) | 14 (15.9) | 1.00 (reference) | 14 (19.4) | 1.00 (reference) | |||
| A/G | 42 (27.5) | 50 (31.3) | <0.001 | 3.98 (2.18–7.28) | 22 (25.0) | <0.001 | 3.84 (1.79–8.28) | 28 (38.9) | <0.001 | 3.91 (1.79–8.54) |
NC Non calculable
* Two-tailed Fisher’s exact test
aAge and gender adjusted Odds Ratio and 95% CI
Results
Study population demographics
The frequency distribution of selected characteristics of controls and OSCC cases are presented in Table 1. The mean age was 58.5 years (±10.1 years; range 40–84 years) for OSCC cases and 54.7 years (±11.9 years; range 31–83 years) for controls. The age distribution between controls and cases was not statistically significant (P = 0.077). A non-statistically significant distribution according to gender was also observed between controls and OSCC cases (P = 0.291).
Table 1.
Frequency distributions of selected demographic variables among OSCC cases and controls
| Variables | Cases (n = 162) | Controls (n = 168) | P* | ||
|---|---|---|---|---|---|
| n | % | n | % | ||
| Age (years) | 0.077 | ||||
| Mean ± SD | 58.5 ± 10.1 | 54.7 ± 11.9 | |||
| Range | 40–84 | 31–83 | |||
| <55 | 84 | 50.0 | 65 | 40.1 | |
| ≥55 | 84 | 50.0 | 97 | 59.9 | |
| Gender | 0.291 | ||||
| Female | 32 | 19.8 | 42 | 25.0 | |
| Male | 130 | 80.2 | 126 | 75.0 | |
* Two-tailed Fisher exact test for the difference in frequency distribution between cases and controls
Genotype distributions of all cytokine polymorphisms among control groups of different ethnic origin
The genotype frequencies of all polymorphisms among Greek and German controls are compared in Table 2 according to the two-tailed Fisher’s exact test. There was no significant difference of genotype distributions for each examined polymorphism between the two ethnic categories of the control group (P range from 0.179 for IL-8 genotypes, to 0.950 for TNF-α genotypes). Similarly, the allele frequencies between the two ethnic categories of the control group were also observed to be non-significant (all P were greater than 0.05). As a result, a single control group, consisting of both Greek and German controls, was used for all further analyses. The resulting allele frequencies in the combined control group were compatible with the Hardy–Weinberg equilibrium (Vairaktaris et al. 2006, 2007a, b, c, 2008; Yapijakis et al. 2008).
Table 2.
Frequency distributions of genotypes and variant allele frequencies (F), concerning all cytokine polymorphisms between Greek and German controls
| Cytokine polymorphisms | Greeks (n = 112) | Germans (n = 56) | P* | ||
|---|---|---|---|---|---|
| n | % | n | % | ||
| IL-1β + 3953 | 0.596 | ||||
| T/T | 9 | 8.7 | 6 | 11.5 | |
| C/C | 57 | 54.8 | 24 | 46.2 | |
| C/T | 38 | 36.5 | 22 | 42.3 | |
| T allele F | 0.269 | 0.327 | 0.700 | ||
| IL-4 − 590 | 0.624 | ||||
| T/T | 9 | 8.3 | 6 | 11.1 | |
| C/C | 69 | 63.9 | 30 | 55.6 | |
| C/T | 30 | 27.8 | 18 | 33.3 | |
| T allele F | 0.222 | 0.278 | 0.274 | ||
| IL-6 − 174 | 0.459 | ||||
| C/C | 5 | 4.8 | 1 | 1.9 | |
| G/G | 62 | 59.6 | 28 | 53.8 | |
| G/C | 37 | 35.6 | 23 | 44.2 | |
| C allele F | 0.226 | 0.240 | 0.777 | ||
| IL-8 − 251 | 0.179 | ||||
| A/A | 0 | 0.0 | 0 | 0.0 | |
| T/T | 60 | 57.7 | 24 | 46.2 | |
| T/A | 44 | 42.3 | 28 | 53.8 | |
| A allele F | 0.212 | 0.269 | 0.258 | ||
| IL-10 − 1082 | 0.286 | ||||
| G/G | 0 | 0.0 | 0 | 0.0 | |
| A/A | 57 | 60.6 | 24 | 51.1 | |
| A/G | 37 | 39.4 | 23 | 48.9 | |
| G allele F | 0.197 | 0.245 | 0.359 | ||
| TNF-α − 308 | 0.950 | ||||
| A/A | 9 | 8.7 | 4 | 8.0 | |
| G/G | 82 | 79.6 | 39 | 78.0 | |
| G/A | 12 | 11.7 | 7 | 14.0 | |
| A allele F | 0.146 | 0.150 | 0.999 | ||
| TNF-β + 252 | 0.503 | ||||
| G/G | 2 | 2.0 | 1 | 2.0 | |
| A/A | 75 | 73.5 | 33 | 64.7 | |
| A/G | 25 | 24.5 | 17 | 33.3 | |
| G allele F | 0.142 | 0.186 | 0.322 | ||
* Two-tailed Fisher exact test for the difference in frequency distribution between controls of Greek and German origin
Genotype distributions of all cytokine polymorphisms among cases and controls
The genotype frequencies of IL-1β (+3953 C/T), IL-4 (-590 C/T), IL-6 (-174 G/C), IL-8 (-251 A/T), IL-10 (-1082 A/G), TNF-α (-308 G/A) and TNF-β (+252 G/A) polymorphisms, as calculated according to the two-tailed Fisher’s exact test, are summarized in Table 3. IL-6 (-174 C/C), IL-8 (-251 A/A), TNF-α (-308 A/A) and TNF-β (+252 G/G) homozygous variant genotypes are strongly associated with overall stages of OSCC, compared to respective homozygous wild type controls (P < 0.001). Likewise, a strong association was also observed in the IL-6 (-174 G/C), IL-8 (-251 T/A), IL-10 (-1082 A/G), TNF-α (-308 G/A) and TNF-β (+252 A/G) carrier genotypes [P range from <0.001 to 0.011 for IL-8 (-251 T/A)]. All aforementioned variant and carrier genotypes, as well as the IL-4 (-590 T/T) variant genotype, have also been associated with early stages of OSCC [P range from <0.001 to 0.015 for IL-8 (-251 T/A)]. Regarding advanced OSCC stages, IL-6 (-174 C/C & G/C), IL-10 (-1082 G/G & A/G), TNF-α (-308 A/A & G/A), and TNF-β (+252 G/G & A/G) genotypes, have already demonstrated a strong association [P < 0.001, for all variant and carrier genotypes, with the exception of IL-10 (-1082 G/G) variant genotype (P = 0.038)]. Interestingly, the IL-1β (+3953 C/T) polymorphism is the sole studied cytokine polymorphism that does not correlate with OSCC, regardless of cancer stage.
Combined analysis for the association between cytokine homozygous variant genotypes and OSCC risk (model A)
Pearson correlation coefficients (r) for interleukin and TNF homozygous variant genotypes, according to OSCC stage, are presented in Table 4. Older age (≥55 years) was not associated with overall, early, and/or advanced OSCC stages. On the other hand, female gender was inversely associated solely with early OSCC stage. The homozygous variant genotypes of IL-6, TNF-α, and TNF-β were significantly associated with overall OSCC stages, as well as early and advanced cancer stages. It is worth noting that the IL-4 and IL-8 variant genotypes were significantly correlated with overall and early OSCC stages, while the IL-10 variant genotype was significantly correlated with that of advanced cancer stage. Once again, the IL-1β variant genotype was not associated with OSCC.
Table 4.
Pearson correlation coefficients (r) for age, gender and cytokine genotypes in association with OSSC stages
| Variables | Overall stages (n = 330)a | Early stage (n = 258)a | Advanced stage (n = 240)a | ||||||
|---|---|---|---|---|---|---|---|---|---|
| N | r | P* | N | r | P* | N | r | P* | |
| Age (≥ 55 years) | 330 | 0.099 | 0.072 | 258 | 0.117 | 0.061 | 240 | 0.064 | 0.326 |
| Gender (Female) | 330 | −0.063 | 0.255 | 258 | −0.137 | 0.028 | 240 | 0.029 | 0.654 |
| Cytokine homozygous variant genotypes | |||||||||
| IL-1β + 3953 T/T | 172 | −0.035 | 0.651 | 140 | −0.026 | 0.762 | 128 | −0.038 | 0.670 |
| IL-4 − 590 T/T | 224 | 0.136 | 0.043 | 174 | 0.306 | <0.001 | 164 | −0.138 | 0.077 |
| IL-6 − 174 C/C | 156 | 0.320 | <0.001 | 134 | 0.279 | 0.001 | 118 | 0.363 | <0.001 |
| IL-8 − 251 A/A | 154 | 0.346 | <0.001 | 124 | 0.517 | <0.001 | 114 | NC | NC |
| IL-10 − 1082 G/G | 129 | 0.163 | 0.065 | 109 | NC | NC | 101 | 0.286 | 0.004 |
| TNF-α − 308 A/A | 245 | 0.600 | <0.001 | 198 | 0.610 | <0.001 | 181 | 0.575 | <0.001 |
| TNF-β + 252 G/G | 221 | 0.738 | <0.001 | 177 | 0.795 | <0.001 | 155 | 0.721 | <0.001 |
| Cytokine carrier genotypes | |||||||||
| IL-1β + 3953 C/T | 239 | −0.100 | 0.123 | 195 | −0.119 | 0.099 | 185 | −0.054 | 0.469 |
| IL-4 − 590 C/T | 277 | 0.029 | 0.633 | 209 | 0.089 | 0.202 | 215 | −0.032 | 0.637 |
| IL-6 − 174 G/C | 294 | 0.310 | <0.001 | 230 | 0.238 | <0.001 | 214 | 0.349 | <0.001 |
| IL-8 − 251 T/A | 300 | 0.150 | 0.009 | 228 | 0.165 | 0.013 | 228 | 0.113 | 0.088 |
| IL-10 − 1082 A/G | 283 | 0.252 | <0.001 | 221 | 0.216 | 0.001 | 203 | 0.262 | <0.001 |
| TNF-α − 308 G/A | 225 | 0.465 | <0.001 | 185 | 0.423 | <0.001 | 180 | 0.453 | <0.001 |
| TNF-β + 252 A/G | 228 | 0.350 | <0.001 | 186 | 0.275 | <0.001 | 192 | 0.332 | <0.001 |
| Cytokine “Mode of Inheritance” genotypes | |||||||||
| IL-1β + 3953 T/T | 172 | −0.035 | 0.561 | 140 | −0.026 | 0.762 | 128 | −0.038 | 0.670 |
| IL-4 − 590 T/T | 224 | 0.136 | 0.043 | 174 | 0.306 | <0.001 | 164 | −0.138 | 0.077 |
| IL-6 − 174 G/C & C/C | 318 | 0.322 | <0.001 | 246 | 0.256 | <0.001 | 228 | 0.357 | <0.001 |
| IL-8 − 251 A/T & A/A | 314 | 0.185 | 0.001 | 242 | 0.227 | <0.001 | 228 | 0.113 | 0.088 |
| IL-10 − 1082 A/G & G/G | 285 | 0.257 | <0.001 | 221 | 0.216 | 0.001 | 205 | 0.272 | <0.001 |
| TNF-α − 308 G/A & A/A | 313 | 0.566 | <0.001 | 242 | 0.554 | <0.001 | 224 | 0.540 | <0.001 |
| TNF-β + 252 A/G & G/G | 221 | 0.738 | <0.001 | 177 | 0.795 | <0.001 | 155 | 0.721 | <0.001 |
NC Non calculable
* Two-sided χ2 test for the evaluation of r
aNumber of observations (controls and patients) used for the calculation of r
Multivariate logistic regression analyses for homozygous variant genotypes and overall OSCC stages (n = 125) were modeled utilizing the following variables: IL-4, IL-6, TNF-α and TNF-b variant genotypes, age, and gender (Table 5). Of the aforementioned variables, only the IL-6 variant genotype [OR: 11.48; 95% CI (3.21–41.00)] and TNF-α variant genotype [OR: 19.43; 95% CI (6.39–59.11)] met the P < 0.05 significance criterion and were thus retained as independent predictors of OSCC in the final regression model (Percent concordance: 83.6%; Hosmer and Lemeshow goodness of fit test χ2 value: 3.295; P = 0.771).
Table 5.
Multivariate logistic regression models adjusted for age and gender for overall, early and advanced stages of OSCC risk in association with cytokine genotypes
| OSCC overall stages | OSCC early stage | OSCC advanced stage | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Estimatea | P* | OR (95% Wald CI) | Estimatea | P * | OR (95% Wald CI) | Estimatea | P * | OR (95% Wald CI) | |
| Model A—cytokine homozygous variant genotypes | |||||||||
| IL-6 − 174 C/C | 2.440 | <0.001 | 11.48 (3.21–41.00) | 3.207 | <0.001 | 24.71 (4.59–133.02) | 1.945 | 0.010 | 6.99 (1.37–35.65) |
| TNF-α − 308 A/A | 2.967 | <0.001 | 19.43 (6.39–59.11) | 2.806 | <0.001 | 16.55 (4.28–64.05) | 3.698 | <0.001 | 40.38 (7.44–219.25) |
| Model B—cytokine carrier genotypes | |||||||||
| Age (≥55 years) | 0.984 | 0.022 | 2.68 (1.15–6.23) | ||||||
| Gender (Female) | 1.271 | 0.021 | 3.56 (1.21–10.49) | ||||||
| IL-6 − 174 G/C | 1.402 | <0.001 | 4.06 (1.83–9.04) | 1.696 | 0.002 | 5.45 (1.85–16.09) | |||
| IL-10 − 1082 A/G | 1.167 | 0.005 | 3.21 (1.41–7.32) | ||||||
| TNF-α − 308 G/A | 2.721 | <0.001 | 15.19 (6.01–38.39) | 2.047 | <0.001 | 7.75 (3.52–17.05) | 2.230 | <0.001 | 9.30 (3.48–24.81) |
| Model C—“Mode of Inheritance” genotypes | |||||||||
| Gender (Female) | 1.193 | 0.023 | 3.30 (1.18–9.18) | ||||||
| IL-6 − 174 G/C & C/C | 2.120 | <0.001 | 8.33 (3.95–17.58) | 1.054 | 0.002 | 2.87 (1.47–5.61) | 1.991 | <0.001 | 7.32 (2.89–18.55) |
| IL-8 − 251 T/A & A/A | 1.265 | <0.001 | 3.54 (1.69–7.43) | ||||||
| IL-10 − 1082 A/G & G/G | 0.973 | 0.009 | 2.65 (1.28–5.46) | 0.969 | 0.034 | 2.64 (1.08–6.46) | |||
| TNF-a − 308 A/G & G/G | 2.726 | <0.001 | 15.27 (7.30–31.96) | 2.568 | <0.001 | 13.04 (6.62–25.69) | 3.002 | <0.001 | 20.12 (7.69–52.66) |
* Two-sided χ2 test
aMaximum likelihood estimates
Identical variables were applied in multivariate logistic regression analyses in order to assess the predictive value of variant genotypes for the occurrence of early OSCC stage (n = 107; Table 5). Of the aforementioned variables, only the IL-6 [OR: 24.71; 95% CI (4.59–133.02)] and TNF-α [OR: 16.55; 95% CI (4.28–64.05)] homozygous variant genotypes met the P < 0.05 significance criterion and were thus retained as independent predictors of early cancer stage in the final regression model (Percent concordance: 86.7%; Hosmer and Lemeshow goodness of fit test χ2 value: 4.057; P = 0.541).
Analogous multivariate logistic regression analysis for homozygous variant genotypes and advanced OSCC stages (n = 103) was undertaken. The final regression model (Table 5) indicated that the IL-6 [OR: 6.99; 95% CI (1.37–35.65)] and TNF-α [OR: 40.38; 95% CI (7.44–219.25)] homozygous variant genotypes were independent predictors of advanced OSCC stage (Percent concordance: 79.9%; Hosmer and Lemeshow goodness of fit test χ2 value: 7.368; P = 0.195).
Hence, it may be concluded that IL-6 and TNF-α homozygous variant genotypes were independent predictors for overall, early and advanced stages of OSCC. The statistical analyses undertaken do not reveal that any other genotype significantly contributed to the risk of developing advanced OSCC stage. Nevertheless, any wide variances in OR and 95% CI values may reflect limitations due to the small sample size of the study population.
Combined analysis of association between the cytokine carrier genotypes and OSCC risk (model B)
The calculated r values for interleukin and TNF carrier genotypes, according to OSCC stage, are presented in Table 4. Carrier genotypes of IL-6, IL-10, TNF-α, and TNF-β were significantly associated with overall, as well as early and advanced stages of OSCC. The IL-8 carrier genotype was associated only with overall and early OSCC stages. Neither IL-4 nor IL-1β carrier genotypes were associated with overall, early or advanced OSCC stages.
Multivariate logistic regression analyses for carrier genotypes and overall OSCC stages (n = 186) were modeled utilizing the following variables: IL-6, IL-8, IL-10, TNF-α, TNF-β carrier genotypes, age, and gender (Table 5). Of the aforementioned variables, only age (≥55 years) [OR: 2.68; 95% CI (1.15–6.23)], the IL-6 carrier genotype [OR: 4.06; 95% CI (1.83–9.04)], the IL-10 carrier genotype [OR: 3.21; 95% CI (1.41–7.32)] and the TNF-α carrier genotype [OR: 15.19; 95% CI (6.01–38.39)] met the P < 0.05 significance criterion and were thus retained as independent predictors of overall OSCC stages in the final regression model (Percent concordance: 86.2%; Hosmer and Lemeshow goodness of fit test χ2 value: 8.093; P = 0.325).
Multivariate logistic regression analyses for carrier genotypes and early OSCC stage (n = 185) were modeled utilizing all variables entered in the previous model (Table 5). Of these variables, only the TNF-α carrier genotype [OR: 7.75; 95% CI (3.52–17.05)] met the P < 0.05 significance criterion and were thus retained as independent predictor of early cancer stage in the final model (Percent concordance: 68.3%; Hosmer and Lemeshow goodness of fit test χ2 value: 2.927; P = 0.570).
Analogous multivariate regression analysis was undertaken in order to evaluate the association between carrier genotypes and advanced OSCC stage (Table 5). Due to lack of statistical significance, the IL-8 carrier genotype was not entered into the initial regression model. The final regression model (n = 161) indicated that gender (female) [OR: 3.56; 95% CI (1.21–10.49)], the IL-6 [OR: 1.70; 95% CI (1.85–16.09)], and the TNF-α [OR: 9.30; 95% CI (3.48–24.81)] carrier genotypes were independent predictors of advanced OSCC stage (Percent concordance: 80.2%; Hosmer and Lemeshow goodness of fit test χ2 value: 7.648; P = 0.365).
Hence, it may be concluded that while the TNF-α carrier genotype was an independent predictor for overall, early and advanced OSCC stages, the IL-6 carrier genotype was an independent predictor for overall and late OSCC stages. Moreover, the IL-10 carrier genotype and age (≥55 years) were independent predictors solely for overall OSCC stages, while gender (female) solely for the advanced OSCC stage. IL-1β, IL-4, TNF-β and IL-8 carrier genotypes were not significant predictors of either overall or early and advanced OSCC stages in the regression models. Once again, any wide variance in OR and 95% CI values may reflect limitations due to the small sample size of the study population.
Combined analysis of association between the “Mode of Inheritance” cytokine genotypes and OSCC risk (biological model C)
The calculated r values for cytokine dominant or recessive genotypes, according to OSCC stage, are presented in Table 4. Recessive genotypes which were significantly associated with both overall and early stage OSCC included those of IL-4 and TNF-β. On the contrary, only the TNF-β recessive genotype was significantly associated with advanced OSCC stage. Dominant genotypes of all cytokine polymorphisms were significantly associated with all stages of OSCC, with the exception of the IL-8 dominant genotype, which was not correlated with advanced OSCC stage.
A biological model (based on “Mode of Inheritance”) predicting the risk for OSCC was constructed according to the multivariate logistic regression procedure (n = 267) utilizing the following variables: age, gender, IL-4 recessive, IL-6 dominant, IL-8 dominant, IL-10 dominant, TNF-α dominant and TNF-β recessive genotypes (Table 5). In the final regression model (Percent concordance: 89.1%; Hosmer and Lemeshow goodness of fit test χ2 value: 9.866; P = 0.275) the following variables were shown to independently predict the risk for developing OSCC: IL-6 dominant genotype [OR: 8.33; 95% CI (3.95–17.58)], IL-8 dominant genotype [OR: 3.54; 95% CI (1.69–7.43)], IL-10 dominant genotype [OR: 2.65; 95% CI (1.28–5.46)] and TNF-α dominant genotype [OR: 15.27; 95% CI (7.30–31.96)]. With the exception of age and the IL-8 dominant genotype, this model agrees with those derived from the aforementioned statistical analyses for both carrier and homozygous variant genotypes as independent predictors for OSCC.
A second biological model for predicting the risk for early OSCC stage was constructed according to the multivariate logistic regression procedure (n = 233) utilizing the following variables: age, gender, IL-4 recessive, IL-6 dominant, IL-8 dominant, IL-10 dominant, TNF-α dominant and TNF-β recessive genotypes (Table 5). In the final regression model (Percent concordance: 80.9%; Hosmer and Lemeshow goodness of fit test χ2 value: 8.852; P = 0.355) solely the IL-6 [OR: 2.87; 95% CI (1.47–5.61)] and the TNF-α [OR: 20.08; 95% CI (8.80–45.86)] dominant genotypes were shown to independently predict the risk for developing early OSCC stage. With no exceptions, this model agrees with those derived from the aforementioned statistical analyses for both carrier and variant genotypes as independent predictors for early OSCC stage.
Finally, a biological model for predicting the risk for developing advanced OSCC stage (Table 5) was constructed using multivariate logistic regression analysis (n = 190). The following variables were included in this analysis: age, gender, IL-6 dominant, IL-10 dominant, TNF-α dominant and TNF-β recessive genotypes. In the final regression model (Percent concordance: 89.0%; Hosmer and Lemeshow goodness of fit test χ2 value: 11.792; P = 0.161) the following variables were shown to independently predict the occurrence of advanced OSCC stage: gender (female) [OR: 3.30; 95% CI (1.18–9.18)], IL-6 [OR: 7.32; 95% CI (2.89–18.55)], IL-10 [OR: 2.64; 95% CI (1.08–6.46)], and TNF-α [OR: 20.12; 95% CI (7.69–52.66)] dominant genotypes. With the exception of IL-10 dominant genotype, this model agrees with those derived from the aforementioned statistical analyses for both carrier and variant genotypes as independent predictors for advanced OSCC stages.
Hence, it may be concluded that the IL-6 and TNF-α dominant genotypes were independent predictors for overall, early, and advanced OSCC stages. On the other hand, independent predictors were the IL-10 dominant genotype solely for overall and advanced OSCC stages, the IL-8 dominant genotype solely for overall OSCC stages and gender (≥55 years) solely for late OSCC stage. Once again, any wide variance in OR and 95% CI values may reflect limitations due to the small sample size of the study population.
Discussion
Functional DNA polymorphisms that affect gene expression or activity of inflammation-related factors may confer susceptibility to disease, its progression and severity (Bergman et al. 2005; Xu et al. 2005; Tsai et al. 2005; Boccia et al. 2006; Cozar et al. 2007; Kleinrath et al 2007; Yapijakis et al. 2007). The study of combined effects in a series of gene polymorphisms of factors participating in the same biological pathway may illustrate the relative significance of each polymorphic site on the final clinical outcome.
In light of the above, the present retrospective case–control study in the same core population of patients with OSCC and healthy controls utilized multivariate logistic regression analysis in order to assess possible interactions between seven common cytokine genotype polymorphisms [IL1β (+3953 C/T), IL-4 (-590 C/T), IL-6 (-174 G/C), IL-8 (-251 A/T), IL-10 (-1082 A/G), TNF-α (-308 G/A), and TNF-β (+252 G/A)] and their combinatory effect in the risk for developing OSCC. Furthermore, the contribution of each cytokine polymorphism upon the development of OSCC was estimated for overall, early, and advanced cancer stages.
Despite the modest sample size of the studied population, a number of significant findings derived from the analysis. It was observed that the independent contributions of two out of seven studied polymorphisms, specifically IL-6 and TNF-α, were consistent and robust in almost all models constructed, both for carrier and dominant variant genotypes, with the exception of early OSCC stage in model B. Two more cytokine polymorphisms were found to significantly contribute only in certain constructed models. In regard to the IL-10 polymorphism, the carrier genotype contributes only to overall OSCC stages in model B, while the dominant genotypes contribute both to overall and advanced stages in model C. On the other hand, the IL-8 dominant genotype contributes only to overall OSCC stages in Model C. Finally, significant effects of age and gender on OSCC development were observed only in models B and C of multivariate logistic regression analysis. Gender contributes to advanced OSCC stage in models B and C, while age contributes solely to overall stages in model B.
As we observed previously, the IL-1β carrier and homozygous variant genotypes do not contribute to the occurrence of OSCC (Vairaktaris et al. 2007c). Interestingly, although the IL-4 homozygous variant genotype was previously found to be significantly associated with the overall and early OSCC stages by itself (Vairaktaris et al. 2007b), the present multivariate regression analysis indicated that it is not a primary predictor when other stronger factors are included in the models. In addition, the overall best fitted logistic regression models do not include the TNF-β polymorphism as one of the primary predictors of OSCC. This probably reflects the very strong correlation between the TNF-α and TNF-β genotype polymorphisms (r values ranged from 0.914 to 0.969 for the homozygous variant genotypes of both cytokines, P < 0.001), because the two genes are closely linked on chromosome 6p21.3. Moreover, the common TNF-β (+252A) allele is associated with increased TNF-α gene expression, since it is located within a phorbol ester-responsive DNA element with high affinity for the AP-1, jun and c-fos heterodimer transcription factor family (Messer et al. 1991).
Although it is rather uncommon to find individuals with a homozygous variant or carrier genotype of all cytokine polymorphisms, we constructed multivariate logistic regression models A and B in order to approximate the highest and the lowest risk to develop OSCC, respectively, for population fractions that carry the variant alleles under study. On the other hand, the biological model C is based on the mode of inheritance of each variant allele, comparing either dominant or recessive genotypes versus the wild type genotypes.
We propose that the prediction of OSCC occurrence in the general population is most likely to be upheld according to model C, which seems to better correspond to genetic contribution of each cytokine. Similar combined models have already been studied for various diseases (Lindpaintner et al. 1995; Jia et al. 2007). Therefore, in order to estimate the risk (OR) for a member of the studied European population to develop oral cancer, according to Table 5 variables, the following formula may be used:
 |
where
is the total OR, e is the base of the natural logarithm, and 
is the sum of each estimate (E) multiplied by Δ (0 or 1: 0 for wild type homozygous genotype or male gender or age <55 years and 1 in all other cases).
Based on these findings, a possible mechanism that may explain the pathway and stoichiometrical interactions of cytokines leading to OSCC occurrence is depicted in Fig. 1. An inflammatory stimulus, such as a bacterial infection, may cause an increment in serum and/or saliva levels of pro-inflammatory cytokines TNF-α, IL-6 and anti-inflammatory cytokine IL-10 (Miettinen et al. 1996). TNF-α has been found to induce angiogenesis by stimulation of IL-6, IL-8 and VEGF (Schindler et al. 1990; Yoshida et al. 1997). Lower levels of IL-6 in serum are associated with the (-174C) variant allele, which, nevertheless, increases IL-6 gene expression after an inflammatory stimulus (Fishman et al. 1998). Therefore, individuals carrying the variant C allele may be more susceptible to OSCC after a sudden elevation of IL-6 based on an inflammatory onset in contrast to individuals carrying the constitutive high expression G allele. In accordance to the present and our previous genetic association studies (Vairaktaris et al. 2006; Yapijakis et al. 2008), several studies have already reported increased serum or saliva levels of IL-6 and/or TNF-α in patients with OSCC (Jablonska et al. 1997; St John et al. 2004; Rhodus et al. 2005). Finally, the respective contribution of IL-10 in OSCC development seems to be less important. Although IL-10 inhibits TNF-α and IL-6 in lymphocyte cell cultures, it can act as a cooperative growth factor with IL-6 for non-Hodgkin’s lymphoma cells (Voorzanger et al. 1996). Such a cooperative role in oral malignancies may be potent since higher levels of IL-10 have been shown to contribute to a negative clinical outcome in patients with OSCC (Fujieda et al. 1999). In accordance to this notion, our previous genetic association data (Vairaktaris et al. 2008) and the present analysis clearly indicate that the IL-10 high gene expression allele is strongly associated with advanced OSCC stage.
Fig. 1.
Possible pathway and stoichiometrical interactions of cytokine variants leading to oral cancer. The label size of each cytokine is analogous to its contribution revealed by the ‘mode of inheritance’ model (overall OSCC stages)
One of the limitations of the present study includes the inability to assess the effect of smoking habits and alcohol consumption upon the occurrence of OSCC, according to the various genotype polymorphisms examined. However, the scope of the current paper was to evaluate the biological predisposition model of interleukin and tumor necrosis factor interactions and their combined effect upon the prediction of OSCC occurrence. A future nested case-control study may assess the influence of lifestyle and other risk factors (including those of smoking habits, alcohol consumption, family history of cancer and thrombophilia) upon the biological predisposition model for the occurrence of OSCC presented in the current study.
In conclusion, this study revealed the highly significant contributions of two out of seven studied cytokines, specifically IL-6 and TNF-α, which seem to have an imperative role in the occurrence of OSCC. Prospective studies with a larger sample population are necessary in order to further evaluate the contributions of all possible cytokine polymorphisms upon the occurrence of OSCC.
Acknowledgments
This work was supported in part by EPEAEK “Pythagoras” grant 70/3/7391 of the Greek Secretariat of Research and Technology to E.V.
Footnotes
Eleftherios Vairaktaris and Christos Yapijakis have contributed equally in this work.
References
- Berghella AM, Contasta I, Pellegrini P, Del Beato T, Adorno D (2002) Peripheral blood immunological parameters for use as markers of pre-invasive to invasive colorectal cancer. Cancer Biother Radiopharm 17:43–50 [DOI] [PubMed] [Google Scholar]
- Bergman M, Ahnstrom M, Palmeback Wegman P, Wingren S (2005) Polymorphism in the manganese superoxide dismutase (MnSOD) gene and risk of breast cancer in young women. J Cancer Res Clin Oncol 131:439–444 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Boccia S, Cadoni G, La Torre G, Arzani D, Volante M, Cattel C, Gianfagna F, Paludetti G, Almadori G, Ricciardi G (2006) A case–control study investigating the role of sulfotransferase 1A1 polymorphism in head and neck cancer. J Cancer Res Clin Oncol 132:466–472 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Cozar JM, Romero JM, Aptsiauri N, Vazquez F, Vilchez JR, Tallada M, Garrido F, Ruiz-Cabello F (2007) High incidence of CTLA-4 AA (CT60) polymorphism in renal cell cancer. Hum Immunol 68(8):698–704 [DOI] [PubMed] [Google Scholar]
- Jablonska E, Piotrowski L, Grabowska Z (1997) Serum Levels of IL-1b, IL-6, TNF-a, sTNF-RI and CRP in patients with oral cavity cancer. Pathol Oncol Res 3:126–129 [DOI] [PubMed] [Google Scholar]
- Jia C, Liu T, Liu Z, Li M, Hu M (2007) Joint effects of eNOS gene T-786C and ADH2 Arg47His polymorphisms on the risk of premature coronary artery disease. Thromb Res 120:679–684 [DOI] [PubMed] [Google Scholar]
- Jin Q, Hemminki K, Enquist K, Lenner P, Grzybowska E, Klaes R, Henriksson R, Chen B, Pamula J, Pekala W, Zientek H, Rogozinska-Szczepka J, Utracka-Hutka B, Hallmans G, Forsti A (2005) Vascular endothelial growth factor polymorphisms in relation to breast cancer development and prognosis. Clin Cancer Res 11:3647–3653 [DOI] [PubMed] [Google Scholar]
- Fishman D, Faulds G, Jeffery R, Mohamed-Ali V, Yudkin JS, Humphries S, Woo P (1998) The effect of novel polymorphisms in the interleukin-6 (IL-6) gene on IL-6 transcription and plasma IL-6 levels, and an association with systemic-onset juvenile chronic arthritis. J Clin Invest 102:1369–1376 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Fujieda S, Sunaga H, Tsuzuki H, Fan GK, Saito H (1999) IL-10 expression is associated with the expression of platelet-derived endothelial cell growth factor and prognosis in oral and oropharyngeal carcinoma. Cancer Lett 136:1–9 [DOI] [PubMed] [Google Scholar]
- Kleinrath T, Gassner C, Lackner P, Thurnher M, Ramoner R (2007) Interleukin-4 promoter polymorphisms: a genetic prognostic factor for survival in metastatic renal cell carcinoma. J Clin Oncol 25(7):845–851 [DOI] [PubMed] [Google Scholar]
- Ku KT, Wan L, Peng HC, Tsai MH, Tsai CH, Tsai FJ (2005) Vascular endothelial growth factor gene-460 C/T polymorphism is a biomarker for oral cancer. Oral Oncol 41:497–502 [DOI] [PubMed] [Google Scholar]
- Lejeune FJ (2002) Clinical use of TNF revisited: improving penetration of anti-cancer agents by increasing vascular permeability. J Clin Invest 110:433–435 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lindpaintner K, Pfeffer MA, Kreutz R, Stampfer MJ, Grodstein F, LaMotte F, Buring J, Hennekens CH (1995) A prospective evaluation of an angiotensin-converting-enzyme gene polymorphism and the risk of ischemic heart disease. N Engl J Med 332:706–711 [DOI] [PubMed] [Google Scholar]
- Loktionov A, Watson MA, Stebbings WS, Speakman CT, Bingham SA (2003) Plasminogen activator inhibitor-1 gene polymorphism and colorectal cancer risk and prognosis. Cancer Lett 189:189–196 [DOI] [PubMed] [Google Scholar]
- Massano J, Regateiro FS, Januario G, Ferreira A (2006) Oral squamous cell carcinoma: review of prognostic and predictive factors. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 102:67–76 [DOI] [PubMed] [Google Scholar]
- Matsumura S, Oue N, Nakayama H, Kitadai Y, Yoshida K, Yamaguchi Y, Imai K, Nakachi K, Matsusaki K, Chayama K, Yasui W (2005) A single nucleotide polymorphism in the MMP-9 promoter affects tumor progression and invasive phenotype of gastric cancer. J Cancer Res Clin Oncol 131:19–25 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Messer G, Spengler U, Jung MC, Honold G, Blomer K, Pape GR, Riethmuller G, Weiss EH (1991) Polymorphic structure of the tumor necrosis factor (TNF) locus: an NcoI polymorphism in the first intron of the human TNF-beta gene correlates with a variant amino acid in position 26 and a reduced level of TNF-beta production. J Exp Med 173:209–219 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Miettinen M, Vuopio-Varkila J, Varkila K (1996) Production of human tumor necrosis factor alpha, interleukin-6, and interleukin-10 is induced by lactic acid bacteria. Infect Immun 64:5403–5405 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Myers JN, Yasumura S, Suminami Y, Hirabayashi H, Lin W, Johnson JT, Lotze MT, Whiteside TL (1996) Growth stimulation of human head and neck squamous cell carcinoma cell lines by interleukin 4. Clin Cancer Res 2:127–135 [PubMed] [Google Scholar]
- Neville BW, Day TA (2002) Oral cancer and precancerous lesions. CA Cancer J Clin 52:195–215 [DOI] [PubMed] [Google Scholar]
- Rhodus NL, Ho V, Miller CS, Myers S, Ondrey F (2005) NF-kappaB dependent cytokine levels in saliva of patients with oral preneoplastic lesions and oral squamous cell carcinoma. Cancer Detect Prev 29:42–45 [DOI] [PubMed] [Google Scholar]
- Schindler R, Mancilla J, Endres S, Ghorbani R, Clark SC, Dinarello CA (1990) Correlations and interactions in the production of interleukin-6 (IL-6), IL-1, and tumor necrosis factor (TNF) in human blood mononuclear cells: IL-6 suppresses IL-1 and TNF. Blood 75:40–47 [PubMed] [Google Scholar]
- Song C, Xing D, Tan W, Wei Q, Lin D (2001) Methylenetetrahydrofolate reductase polymorphisms increase risk of esophageal squamous cell carcinoma in a Chinese population. Cancer Res 61:3272–3275 [PubMed] [Google Scholar]
- St John MA, Li Y, Zhou X, Denny P, Ho CM, Montemagno C, Shi W, Qi F, Wu B, Sinha U, Jordan R, Wolinsky L, Park NH, Liu H, Abemayor E, Wong DT (2004) Interleukin 6 and interleukin 8 as potential biomarkers for oral cavity and oropharyngeal squamous cell carcinoma. Arch Otolaryngol Head Neck Surg 130:929–935 [DOI] [PubMed] [Google Scholar]
- The International HapMap Project (2003) Nature 426:789–796 [DOI] [PubMed] [Google Scholar]
- Tsai MH, Chen WC, Tsai CH, Hang LW, Tsai FJ (2005) Interleukin-4 gene, but not the interleukin-1 beta gene polymorphism, is associated with oral cancer. J Clin Lab Anal 19:93–98 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Vairaktaris E, Yiannopoulos A, Vylliotis A, Yapijakis C, Derka S, Vassiliou S, Nkenke E, Serefoglou Z, Ragos V, Tsigris C, Vorris E, Critselis E, Avgoustidis D, Neukam FW, Patsouris E (2006) Strong association of interleukin-6 -174 G > C promoter polymorphism with increased risk of oral cancer. Int J Biol Markers 21:246–250 [DOI] [PubMed] [Google Scholar]
- Vairaktaris E, Yapijakis C, Serefoglou Z, Derka S, Vassiliou S, Nkenke E, Vylliotis A, Wiltfang J, Avgoustidis D, Critselis E, Neukam FW, Patsouris E (2007a) The interleukin-8 (-251A/T) polymorphism is associated with increased risk for oral squamous cell carcinoma. Eur J Surg Oncol 33:504–507 [DOI] [PubMed] [Google Scholar]
- Vairaktaris E, Yannopoulos A, Vassiliou S, Serefoglou Z, Vylliotis A, Nkenke E, Critselis E, Avgoustidis D, Yapijakis C, Neukam FW, Patsouris E (2007b) Strong association of interleukin-4 (-590 C/T) polymorphism with increased risk for oral squamous cell carcinoma in Europeans. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 104(6):796–802 [DOI] [PubMed] [Google Scholar]
- Vairaktaris E, Serefoglou Z, Yapijakis C, Vasiliou S, Derka S, Nkenke E, Vylliotis A, Ragos V, Neukam FW, Patsouris E (2007c) The interleukin-1 beta gene polymorphism +3953 C/T is not associated with risk for oral cancer. Anticancer Res 27(6):3979–3984 [PubMed] [Google Scholar]
- Vairaktaris E, Yapijakis C, Serefoglou Z, Derka S, Vasiliou S, Nkenke E, Vylliotis A, Spyridonidou S, Neukam FW, Patsouris E (2008) Gene expression polymorphism of interleukin-10 is associated with risk for oral cancer. Anticancer Res 28(1A):309–314 [PubMed] [Google Scholar]
- Voorzanger N, Touitou R, Garcia E, Delecluse HJ, Rousset F, Joab I, Favrot MC, Blay JY (1996) Interleukin (IL)-10 and IL-6 are produced in vivo by non-Hodgkin’s lymphoma cells and act as cooperative growth factors. Cancer Res 56:5499–5505 [PubMed] [Google Scholar]
- Watanabe H, Iwase M, Ohashi M, Nagumo M (2002) Role of interleukin-8 secreted from human oral squamous cell carcinoma cell lines. Oral Oncol 38:670–679 [DOI] [PubMed] [Google Scholar]
- Williams HK (2000) Molecular pathogenesis of oral squamous carcinoma. Mol Pathol 53:165–172 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wong YK, Chang KW, Cheng CY, Liu CJ (2006) Association of CTLA-4 gene polymorphism with oral squamous cell carcinoma. J Oral Pathol Med 35:51–54 [DOI] [PubMed] [Google Scholar]
- Xu J, Lowey J, Wiklund F, Sun J, Lindmark F, Hsu FC, Dimitrov L, Chang B, Turner AR, Liu W, Adami HO, Suh E, Moore JH, Zheng SL, Isaacs WB, Trent JM, Gronberg H (2005) The interaction of four genes in the inflammation pathway significantly predicts prostate cancer risk. Cancer Epidemiol Biomarkers Prev 14:2563–2568 [DOI] [PubMed] [Google Scholar]
- Yapijakis C, Vairaktaris E, Vassiliou S, Vylliotis A, Nkenke E, Nixon AM, Derka S, Spyridonidou S, Vorris E, Neukam FW, Patsouris E (2007) The low VEGF production allele of the +936C/T polymorphism is strongly associated with increased risk for oral cancer. J Cancer Res Clin Oncol 133:787–791 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Yapijakis C, Vairaktaris E, Serefoglou Z, Vylliotis A, Nkenke E, Derka S, Vassiliou S, Avgoustidis D, Neukam F, Patsouris E (2008) Association of polymorphisms affecting gene expression of tumor necrosis factor alpha and beta with increased risk for oral cancer. Int J Biol Markers (in press)
- Yoshida S, Ono M, Shono T, Izumi H, Ishibashi T, Suzuki H, Kuwano M (1997) Involvement of interleukin-8, vascular endothelial growth factor, and basic fibroblast growth factor in tumor necrosis factor alpha-dependent angiogenesis. Mol Cell Biol 17:4015–4023 [DOI] [PMC free article] [PubMed] [Google Scholar]