Tumour necrosis factor polymorphisms and susceptibility to follicular lymphoma


Dr Jude Fitzgibbon, ICRF Medical Oncology Unit, St Bartholomew's Hospital, 45 Little Britain, West Smithfield, London EC1A 7BE. e-mail: jfitzgib@hgmp.mrc.ac.uk


Follicular lymphoma is characterized in 85% of patients by the presence of a t(14;18) chromosomal translocation that results in overproduction of BCL2. In this study the distribution of high and low expressing TNF alleles at the TNF (−308) and LTα (+252) polymorphic sites in 121 patients with follicular lymphoma and 88 control individuals has been analysed. A reduction in high expressing haplotypes in patients compared to normal controls was found (P = 0.055), with no significant difference observed in response rate or overall survival between patients with high or low expressing haplotypes. These results suggest that the TNF locus, or an adjacent locus within the MHC region, is an important genetic risk factor in this disease.

The majority of patients with follicular lymphoma have disseminated disease at presentation and, despite the high rate of response to treatment, the illness follows a fluctuating course of regression and progression with patients having an overall median survival of 8–10 years (Gupta & Lister, 1996). It is characterized by the presence (in approximately 85% of tumour biopsies from patients) of the t(14;18) (q32;q21) chromosomal translocation that results in the up-regulation of an anti-apoptotic factor called BCL2. The translocation has also been observed in patients with reactive follicular hyperplasia (Limpens et al, 1991), polyclonal B-cell lymphocytosis (Delage et al, 1997) and in the peripheral blood of healthy individuals (Limpens et al, 1995) although the reason for this has not been determined. This, however, suggests that additional genetic or environmental factors may be required to influence susceptibility to follicular lymphoma.

The major histocompatibility complex (MHC) class III genes, tumour necrosis factor (TNF) and lymphotoxin alpha (LTα) (formerly TNFβ), are primary mediators of immune activation and inflammatory responses and are involved in the pathogenesis of such conditions as septic shock, autoimmune disorders or malignancies and graft-versus-host disease (Gruss & Dower, 1995). Warzocha et al (1998) recently investigated the contribution of polymorphisms at the TNFα (−308) and LTα (+252) loci to the clinical outcome of patients with non-Hodgkin's lymphoma. In their study, high-circulating levels of TNF at the time of diagnosis were related to the presence of particular high producing haplotypes. In patients with diffuse large cell lymphoma the presence of two or more TNF/LTα high-producing alleles was significantly associated with higher rate of relapse and progression. In contrast, the TNF haplotype distribution had no measurable effect on clinical outcome in patients with follicular lymphoma, although a reduction in frequency of the high expressing haplotypes was seen.

In this study the frequency of these TNF polymorphisms has been determined in a separate cohort of follicular lymphoma patients and the results provide further evidence for the existence of a genetic risk factor for follicular lymphoma within the MHC region on human chromosome 6.


489 patients with follicular lymphoma attended the Department of Medical Oncology at St Bartholomew's Hospital during the period 1971–98. DNA from 121 of these patients was readily available, having been previously extracted from diagnostic lymph node biopsy or bone marrow aspirates (performed as part of their initial staging investigations, or at recurrence) for other on-going studies. All histology was reviewed in the Department of Histopathology. The median age at diagnosis was 44 years (range 26–70 years) with 69 male and 52 female patients. 52 patients were previously untreated. The low median age reflects the referral pattern of patients to such a specialist ‘tertiary’ centre.

Controlled DNA obtained from 88 normal donors was kindly donated by Dr J. Bodmer (Imperial Cancer Research Fund, Oxford) and had previously been used in various HLA-related studies.

Genotyping was carried out using previously described primer pairs for TNF and LTα using standard buffer conditions (Warzocha et al, 1998). A cycling regime using 20 s denaturation, annealing and elongation steps (shorter than those previously reported) was used. A single pair mismatch at the 3′ end of the otherwise identical primers representing the G (TNF1) or A (TNF2) base substitution at the −308 polymorphic site of the TNFα formed the basis for an allele-specific amplification of a 184 bp product. Nco1 (NE Biolabs) digestion was carried out directly on the 368 bp LTα PCR products (that span an A or G polymorphic site at position +252) at 37°C for 2 h prior to electrophoresis on a 2% agarose gel, to discriminate between the LTα (5.5 kb) allele represented by digestion products of 133 bp and 235 bp or LTα (10.5 kb) allele, represented by an uncleaved product. A single amplification product of the correct size was visualized for each primer set used.

The remission duration and overall survival of the patients were estimated by the Kaplan-Meier method and statistical differences were assessed using the chi-square test. Univariant analysis of demographic variables were compared using the chi-square test.


The distribution of the TNF (−308) and LTα (+252) polymorphisms in 121 follicular lymphoma patients and control group are shown in 1Table I. TNF2 is associated with high levels of TNFα production and similarly LTα (5.5 kb) with higher LTα production. In both control and patient groups the expected strong association between TNF2 and LTα (5.5 kb) alleles was observed. The presence of two or more of these alleles correlated with increased cytokine production (i.e. high expressing haplotype), whereas the remainder were considered low expressing haplotypes, i.e. TNF1/1 + LTα 10.5/10.5, TNF1/1 + LTα 5.5/10.5 and TNF1/2 + LTα 10.5/10.5. A significant reduction was found in the frequency of the high-expressing haplotype in patients with follicular lymphoma [28/121 (23%) high expressors versus 93/121 (77%) low expressors] when compared to the control sample [31/88 (35%) high expressors versus 57/88 (65%) low expressors] (χ2, P = 0.055). There was no significant difference in allele frequency of the control TNF and LTα polymorphisms compared to other U.K. published reports (Gallagher et al, 1997; Moffat & Cookson, 1997; Newton et al, 1998).

Table 1. Table I. Distribution and allele frequency of TNF (−308) and LTα (+252) polymorphisms in 121 patients with follicular lymphoma and 88 controls.Thumbnail image of

These results suggest that the expression of certain TNF haplotypes influences susceptibility to follicular lymphoma and confirm the observations by Warzocha et al (1998). As the MHC retains a large number of immunologically important genes, many of which show extensive polymorphism, it could also be argued that these TNF polymorphisms may act solely as a marker for an adjacent locus; for example, TNF2 and LTα (5.5 kB) alleles are linked to the HLA haplotype A1-B8-DR3-DQ2 (Wilson et al, 1993). There is also evidence from murine studies that two loci, Msmr1 (MSM resistance 1) and Esl1 (early lymphoma of SL-1), which confer a predisposition to B-cell lymphoma, map to the MHC region on mouse chromosome 17 (Hiai, 1996). These models may assist in the characterization of the locus described here but they also raise the possibility that more than one genetic factor in this region could be influencing susceptibility to follicular lymphoma.

In contrast there was no association between TNF haplotype distribution and other prognostic variables such as gender, age at presentation, presence of extranodal disease or bone marrow infiltration (results not shown). Of the 39 (of 52 previously untreated patients) who achieved a first clinical remission (i.e. complete and partial), 27 were low expressors and 12 high expressors and there was no significant difference in remission duration (P = 0.38) (Fig 1a). The overall survival for all 121 patients between those with high (n = 28) or low (n = 93) expressing haplotypes was also not significant (P = 0.74) (Fig 1b). It is difficult to say therefore whether the full characterization of this susceptibility factor would be beneficial in the clinical management of these patients.

Figure 1.

. (A) Kaplan-Meier plot of remission duration of previously untreated patients in first clinical remission (n = 39) according to TNF haplotype (χ2 = 0.79, P = 0.38). (B) Kaplan-Meier plot of overall survival for all 121 patients according to TNF haplotype (χ2 = 0.11, P = 0.74).

Larger epidemiological studies should, however, be carried out both to confirm the findings of this and other studies and also to ascertain if polymorphisms in other loci in the MHC region are more strongly associated with increased risk of follicular lymphoma. These studies should also be expanded to include other polymorphisms, e.g. rare alleles at the H-ras 1 proto-oncogene locus that have recently been shown to influence susceptibility and clinical outcome in non-Hodgkin's lymphoma, and, in particular, low-grade lymphoma (Calvo et al, 1998). The existence of these modulating genes could provide one explanation for the occurrence of the t(14,18) translocation in the peripheral blood of apparently healthy individuals, follicular lymphoma may therefore develop in genetically predisposed individuals in whom the t(14;18) translocation has sporadically developed in an early B cell.


The authors are grateful to Dr Julia Bodmer for the control DNA samples, to Ms Louise Lyons, Ms Susan Tonks and Mr Ashiq Salam for technical support, and to the Joint Research Board of St Bartholomew's Hospital for supporting Dr D. Grenzelias during this work.