Editor—TNFα (tumour necrosis factor-alpha) is a cytokine produced by macrophages and monocytes with a wide range of activities, and polymorphisms within this gene have been postulated to contribute to MHC associations with autoimmune and infectious diseases.1 The role of TNFα in cancer is a controversial matter, because while it plays a key role in the “in vitro” killing of tumour cells by macrophages and lymphocytes, it has also been found in high concentrations in patients with cancer, suggesting that it may be an endogenous tumour promoter “in vivo”.2 3 Different results with several tumour types show that TNFα may have both tumour necrotic and tumour promoting activities.

Recently, several genetic polymorphisms have been described in the human TNFα promoter.4-6Among them, the rare allele at position −308 (TNF308.2) has been proven to be part of a complex haplotype that is involved in higher TNFα levels and has been related to poor prognosis in several diseases.7 The existence of differentTNFα alleles, related to different levels of TNFα, raises the possibility that tumour development is somewhat related to the genetic propensity of the person to produce higher levels of TNFα and, therefore, with the presence of genetic variants in this gene. In fact, an increase in the TNF308.1/TNF308.2 genotype has been reported in different tumour types, with a significantly increased frequency of the TNF308.2 allele in patients with malignant tumours.8 Wilson et al 7 have shown that the polymorphism at −308 has a significant effect on the transcriptional activity of the humanTNFα gene, either because the interaction of the transcription enhancers is increased owing to the different DNA conformations, or because the TNF2 variant is the target for novel binding proteins.

The G to A transition at position –238 (TNF238.2) is also suspected to influence the expression of TNFα, and although its clinical and functional consequences are not clear so far,9 it has been associated with development and prognosis of different diseases as well.10 11

The effects of TNFα on osteoblast differentiation and proliferation are complex but it is generally assumed that it inhibits bone formation and stimulates bone resorption.12 The activity of TNFα in osteosarcoma, Ewing's sarcoma, and primary human osteoblast cultures has been widely studied showing that it has an antiproliferative and cytotoxic role which usually depends on the type of cell line under analysis.13 14

Synergistic cytotoxicity has been described between TNFα and other cytokines, most frequently IFNγ (interferon gamma) and also with certain drugs that are commonly used in the treatment of osteosarcoma and Ewing's sarcoma like the topoisomerase II inhibitors.13 The data available indicate that TNFα and agents that stimulate its production by host macrophages may have a role in the treatment of osteosarcoma and Ewing's sarcoma.

Based on the role of TNFα in bone biology and the growing evidence of the relationship existing between TNFα genetics and cancer, we tried to test the hypothesis of whether genetic polymorphisms of theTNFα promoter contribute to the pathogenesis or prognosis of paediatric bone tumours.

DNA was extracted following standard procedures15 from peripheral blood lymphocytes of 110 paediatric patients (52 females, 58 males) with bone tumours (63 osteosarcomas and 47 Ewing's sarcomas) and 111 healthy children (53 females, 58 males). All the subjects included in the analysis were white, most of them from the region of Navarre (Spain), and the age distribution was very similar in the tumour and control groups (mean (SD): osteosarcoma 13.5 years (SD 3.3), Ewing's sarcoma 12 years (SD 3.7), controls 11.1 years (SD 5.1)).

The human TNFα promoter region between nucleotides −398 and −103 was analysed by PCR-DGGE (polymerase chain reaction coupled to denaturing gradient gel electrophoresis) as previously published16 17 (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

DGGE gel showing polymorphic and normal band patterns. Lanes 1 to 3: genotype normal/−376−238. Lane 4: genotype normal/−308. Lanes 5 and 6: normal, non-polymorphic patterns.

Statistical analyses were performed with the Statistical Package for the Social Sciences (SPSS, version 9.0). The chi-square contingency test, with Yates's modification for small numbers if needed, was used to test for a significant association between disease andTNFα genotypes or haplotypes. Crude odds ratios (ORs) were calculated and given with 95% confidence intervals (CI). To test for association between clinical parameters and genotypes or haplotypes, t tests or one way analysis of variance (ANOVA) were performed. Statistical significance was assumed if p⩽0.05 and highly statistical difference if p<0.01; differences between variables with p>0.05 were considered not statistically different. Osteosarcomas and Ewing's sarcomas were treated independently in the statistical analyses given that they are different clinical entities, both in genetic background and in the clinical and histological aspects.

We have analysed the distribution of the TNFαgene promoter polymorphisms at positions –376, −308, −238, and –163 in 110 bone sarcoma paediatric patients and in 111 paired healthy children, to search for the putative association between theTNFα gene and tumour development or prognosis.

The overall distribution of TNFαpolymorphic alleles in bone cancer paediatric patients and control subjects is shown in table 1. As in previous reports, polymorphisms at −376 and −238 were found to be in linkage disequilibrium.5 To date, there are no consistent data on the frequency of the polymorphic alleles at –376 and –163 in white populations, and larger series should be screened to establish the frequencies for these alleles in both normal and disease related populations.

The frequencies of the alleles at −308 and –238 in the Spanish population (0.87 and 0.92, respectively, deduced from table 2) were found to be very similar to those described for European white populations.10 18

Tables 2 and 3 show the allele and genotype frequencies of the genetic variants of the TNFα gene promoter at –308 and –238 in both cancer patients and healthy subjects. The frequency of the TNF238.2 allele (adenine at position –238) and TNF238.1/TNF238.2 heterozygote genotype was significantly lower in the osteosarcoma group, but not among Ewing's sarcomas. We did not find any outstanding difference in the distribution of the TNF308.2 allele (adenine at position –308) between bone sarcomas and controls.

No relationship was found between the presence of genotypes for theTNFα gene promoter and any of the clinical parameters tested: tumour stage, tumour location or size, development of metastasis or relapse, and age at diagnosis (data not shown).

Surprisingly, the percentage of males carrying polymorphisms of theTNFα promoter was statistically higher than in females (p=0.025, OR=4.05, CI=1.13-14.43). We did not detect this difference in the group of healthy controls, in which the distribution of polymorphic patterns was similar in both sexes.

Several reports have indicated that different HLA (human leucocyte antigen) products and related genes may be risk factors for and also protective factors against cancer. TheTNFα gene is of particular interest because of its involvement in tumour immunity and cancer pathogenesis and the relationship existing between certainTNFα genetic variants and human tumours.

Although the exact regulatory mechanisms altered by theTNFα promoter polymorphisms are not completely delineated, studies on these polymorphisms have shown that those at −308 and −238 are associated with the development and even prognosis of certain types of cancer. Chouchane et al 8 detected a marked decrease of the TNF308.1 homozygous genotype in patients with non-Hodgkin's lymphoma, breast carcinoma, and in a group of different malignant tumours. Nevertheless, it must be taken into account that the results of this study were obtained with adult patients in the Tunisian population and the polymorphisms of TNFαare dependent on ethnicity.

In the present study, we report a decrease in the TNF238.2 rare allele among osteosarcoma paediatric patients and no difference from the control population in the distribution of the TNF308.2 variant in either of the tumour groups analysed, while other authors have described a decreased representation of the TNF308.2 allele in cancer populations (for example, in chronic lymphocytic leukaemia).19

Although the exact influence of the TNF238.2 variant on TNFα function and expression is not fully understood to date, the localisation of this polymorphism in the regulatory Y box20 suggests that it may contribute to the optimal function and regulation of theTNFα promoter. However, it has been proven, in transfection assays, that there are no differences in TNFα production after stimulation of TNF238.2 heterozygous or normal homozygous cells; therefore TNF238.2 is not likely to be of functional relevance for transcriptional activation, and the actual meaning of the –238 promoter polymorphism remains a controversial matter.

There is evidence that the −308 TNF2 allele is overexpressed in diseases where TNFα levels are associated with poor prognosis. In our bone sarcoma series, we did not find a relationship between this or any other polymorphic variant of theTNFα promoter and tumour prognosis, disease free survival or development of metastasis or relapse (data not shown). One hypothesis is that these polymorphisms may serve as markers for additional polymorphisms or mutations in neighbouring genes that may be involved in the disease.

With regard to the increased number ofTNFα polymorphic alleles in male osteosarcoma patients, several other authors have reported gender differences and increased TNFα levels in male patients affected by type II diabetes mellitus.21 A possible explanation for this finding is that the increased plasma levels of TNFα in males are the consequence of the presence of an increased number of the TNF308.2 allele, which has been associated with higher expression of theTNFα gene. In fact, in our series of osteosarcoma, there is a tendency for a higher number of TNF308.2 alleles in male patients, without reaching, however, statistical significance (p=0.079).