Influence of −308 A/G polymorphism in the tumor necrosis factor α gene on etanercept treatment in rheumatoid arthritis

Authors


Abstract

Objective

To determine whether the −308 A/G tumor necrosis factor α (TNFα) gene polymorphism can predict the outcome of etanercept therapy in 86 patients with rheumatoid arthritis (RA), as already observed in patients treated with infliximab.

Methods

Eighty-six RA patients treated with etanercept were genotyped for −308 A/G TNFα gene polymorphism by polymerase chain reaction and melting curve analysis, using specific gene primers and probes. Patients were subdivided into group A (G/A genotype) and group G (G/G genotype). We compared clinical responses to etanercept between groups A and G after 6 months, using the Disease Activity Score in 28 joints (DAS28). After 12-month treatment, 48 of 86 patients were evaluated again.

Results

Of 86 patients, 18 (21%) belonged in group A and 68 (79%) belonged in group G. After 6-month treatment, 55.6% of patients in group A and 82.4% of patients in group G had DAS28 improvement >1.2 (P = 0.027 by chi-square). The mean ± SD DAS28 improvement was 1.69 ± 1.31 in group A and 2.23 ± 1.19 in group G (P = 0.098 by t-test). After 1-year treatment 48 patients were tested again: 10 (21%) belonged in group A and 38 (79%) belonged in group G. Forty percent of patients in group A and 87% in group G had DAS28 improvement >1.2 (P = 0.005 by chi-square). The mean ± SD DAS28 improvement was 1.334 ± 1.37 in group A and 2.29 ± 1.47 in group G (Mann-Whitney U test = 115, P = 0.0057).

Conclusion

RA patients with a −308 G/G TNFα genotype respond to etanercept better than patients with a −308 A/G genotype.

INTRODUCTION

Tumor necrosis factor α (TNFα) blockers include a recombinant TNF receptor (etanercept) and 2 anti-TNFα monoclonal antibodies (infliximab and adalimumab). They are highly effective to treat rheumatoid arthritis (RA) but are expensive and may cause side effects. Moreover, 30–40% of patients with RA do not respond well to TNFα inhibitors (1). It would be very useful to be able to determine which patients are likely to respond to TNF inhibitors.

Of interest, the TNFα gene promoter is polymorphic. Polymorphism at position −308 of the TNFα gene is known to influence binding of transcription factors and to control the level of TNFα production after lipopolysaccharide (LPS) stimulation (2–4).

In a previous study, we found that patients with RA with the −308G/G genotype responded to infliximab better than those with the −308 A/G genotype (5). In the present study, we tested whether the −308 A/G polymorphism in the TNFα gene can predict the outcome at 6 months of etanercept therapy in patients with RA.

PATIENTS AND METHODS

Patients and treatment protocol.

A total of 86 white patients fulfilling the 1987 American College of Rheumatology (formerly the American Rheumatism Association) criteria for the classification of RA (6) were enrolled in 2 centers: rheumatology department at University Hospital La Conception, Marseilles, France, and rheumatology department at University Hospital Minjoz, Besançon, France. All patients had active disease, as indicated by a score >3.2 (range 3.20–7.4) on the Disease Activity Score in 28 joints (DAS28) (7), with a mean score of 5.4. All patients were tested for tuberculosis (skin tests, chest radiographs, and Koch's bacillus in sputum) and hepatitis A, B, and C infection. When needed, treatment of latent tuberculosis (rifampin plus isoniazid) was initiated 1 month before etanercept treatment and continued for 6 months. Two patients with hepatitis C were included. All patients received etanercept subcutaneously (25 mg twice per week). Initial treatment with nonsteroidal antiinflammatory drugs and/or oral glucocorticoids was continued. None of the patients received methotrexate during the study. Thirteen (15%) patients had previously received infliximab without success and this theoretically could have induced a bias in our study, as discussed later. Patients were followed every month by a general practitioner and every 3 months by a rheumatologist. Clinical responses were evaluated with the DAS28. All 86 patients were evaluated before the first injection and after 6 months of treatment. Of these 86 patients, 48 were additionally evaluated after 1 year of treatment; 38 patients were not reevaluated at 12 months. Nine patients received methotrexate after month 6 because of lack of efficiency of etanercept alone and therefore were not evaluated later. Thirteen patients stopped taking etanercept between month 6 and month 12 because of side effects. Sixteen patients had not reached 12 months of treatment at the time this study was completed.

Data were collected between 2000 and 2005. For anti–citrullinated fibrin antibody detection, a serum was considered positive for anti–citrullinated fibrin antibody when levels were above a cutoff that allowed 95% specificity as described previously (8).

Genotyping for −308A/G TNFα polymorphism.

Genomic DNA was isolated from 5 ml of heparinized blood with an EZ1 biorobot (Qiagen, Hilden, Germany), according to the manufacturer's instructions. Genotyping for −308A/G TNFα gene polymorphism was performed by polymerase chain reaction (PCR) and melting curve analysis using a Light Cycler (Roche Diagnostics, Mannheim, Germany), as published previously. The following primer sequences were used: for TNFα-371, forward AAGGAAACAGACCACAGACCTG, and for TNFα-251, reverse GGTCTTCTGGGCCACTGAC. The sequence of the sensor fluorescein probe was AACCCCGTCCCCATGCCCC X and the corresponding anchor red 640 probe was CAAAACCTATTGCCTCCATTTCTTTTGGGGAC. Primers and probes were bought from TIB Molbiol (Berlin, Germany). PCR mix was prepared with the Fast-Start hybridization kit (Roche Molecular Biochemicals, Mannheim, Germany). The 20-μl final reaction volume contained 3 mM of MgCl2, 0.5 μM of each primer, 3 pM of each probe, and 100 ng of patient DNA. Carryover contamination was prevented by using 1 unit of heat-labile uracil DNA-glycosylase (Roche Molecular Biochemicals). Amplification was performed as follows: initial denaturation for 10 minutes at 95°C followed by 45 cycles of 5-second denaturation at 95°C, 5-second annealing at 62°C, 10-second primer extension at 72°C. The program for melting analysis was 20-second denaturation at 95°C, 20-second annealing at 40°C, and temperature increase to 85°C at 20°C/second ramping rate. Fluorescence signal was measured during a melting analysis program. Each experiment included 3 positive controls with known genotypes (A/A, A/G, and G/G, respectively) and 1 negative control (no DNA).

Statistical analysis.

Patients were subdivided into group A (G/A genotype) and group G (G/G genotype). We compared clinical responses to etanercept between groups A and G after 6 months and 1 year, using the DAS28. Qualitative data were compared using chi-square test or Fisher's test, and quantitative data were compared using Student's t-test or Mann-Whitney U test. The threshold of statistical significance for all tests was 5%.

RESULTS

Baseline characteristics of patients.

The baseline characteristics of the 86 patients, before the first injection of etanercept, are presented in Table 1. A total of 80% of the patients were women. Mean ± SD age was 56.5 ± 14.2 years and mean disease duration was 11.5 ± 10.3 years. Most patients had active disease as indicated by a mean DAS28 score of 5.4. Two-thirds of patients were rheumatoid factor positive and 80% expressed the HLA–DR shared epitope. Mean ± SD C-reactive protein level was 21.2 ± 26.8 mg/dl. One-third of patients had no radiologic damage. Among the 86 patients included in the 6-month followup, 18 (21%) had the G/A genotype and 68 (79%) had the G/G genotype. No patient had the A/A genotype. In the group of 48 patients evaluated again after 12 months, 10 (21%) had the G/A genotype and 38 (79%) had the G/G genotype. The HLA–DR3 allele was more frequent in patients with the A/G genotype than in patients with the G/G genotype (44% versus 5.8%; P = 0.001). A total of 71% of patients had antibodies against deiminated fibrin (65% of patients with the G/G genotype and 93% of those with the A/G genotype; P = 0.006). Except for frequency of HLA–DR3 and deiminated fibrin antibodies, no difference was found between group G and group A.

Table 1. Baseline characteristics of patients before the first injection of etanercept*
 All patients (n = 86)G group (n = 68)A group (n = 18)P
  • *

    Values are mean ± SD unless otherwise indicated. Values are given for all patients except where indicated (NT = number tested). SE = shared epitope; DAS28 = Disease Activity Score in 28 joints; CRP = C-reactive protein; RF = rheumatoid factor.

  • Between groups A and G.

Demographics    
 Women, no. (%)69 (80)55 (80.9)14 (77.8)0.747
 Age, years56.5 ± 14.256.7 ± 14.755.4 ± 12.60.72
Disease status    
 Disease duration, years11.5 ± 10.311.31 ± 10.212.2 ± 11.10.756
 SE, no. (%)69 (80.2)57 (83.8)12 (66.7)0.196
 HLA–DR3, no. (%)12 (13.8)4 (5.8)8 (44.4)0.001
Clinical assessment    
 Arthralgia12.5 ± 7.413 ± 7.310.8 ± 7.80.27
 Arthritis4.2 ± 4.73.9 ± 4.95.3 ± 40.265
 Rheumatoid nodules present, no. (%)8 (9.3)7 (10.3)1 (5.6)0.468
 DAS28 score5.4 ± 1.05.4 ± 1.05.2 ± 0.90.321
Biologic assessment    
 Serum CRP, mg/dl21.2 ± 26.8 (NT = 85)22.2 ± 28.417.2 ± 19.4 (NT = 17)0.499
 Positive RF, no. (%)58 (67.4)45 (69.2) (NT = 65)13 (72.2)0.807
 Antideiminated fibrin antibodies, no. (%)53 (70.7) (NT = 75)39 (65) (NT = 60)14 (93.3) (NT = 15)0.006
Additional drug treatment    
 Methotrexate, mg/week5.4 ± 6.15.4 ± 6.25.1 ± 6.10.853
 Prednisone, mg/day9.7 ± 8.3 (NT = 59)10.0 ± 9.1 (NT = 45)8.6 ± 4.7 (NT = 13)0.585

Etanercept response and TNFα −308A/G polymorphism.

Good responders (after 6 months and 1 year of treatment) were defined as patients whose DAS28 score improved by at least 1.2 points and bad responders were defined as patients whose DAS28 score improved by <1.2 points, according to the European League Against Rheumatism criteria (9). Overall, most patients were good responders at 6 months (76.7%) and at 12 months (77.1%). The group of 13 patients who had previously been treated unsuccessfully with infliximab did not behave differently, as 4 (30%) were bad responders and 9 (70%) were good responders (data not shown). At 6 months, the percentage of good responders was significantly higher in group G (82.4%) than in group A (55.6%; P = 0.027 by chi-square) and the mean ± SD DAS28 improvement tended to be higher in group G (2.23 ± 1.19) than in group A (1.69 ± 1.31). However, this did not reach significance (P = 0.098 by t-test) (Table 2 and Figure 1). After 12 months of therapy, the percentage of good responders was higher in group G (86.8%) than in group A (40%; P = 0.005 by chi-square, P = 0.0057 by Mann-Whitney) and the average DAS28 score improvement was higher in group G (2.29 ± 1.47) than in group A (1.33 ± 1.37) (Table 2).

Table 2. Responders/nonresponders and DAS28 improvement under etanercept therapy and −308 tumor necrosis factor α polymorphism*
 After 6-month treatmentAfter 12-month treatment
All (n = 86)A (n = 18)G (n = 68)PAll (n = 48)A (n = 10)G (n = 38)P
  • *

    Values are the mean ± SD unless otherwise indicated. DAS28 = Disease Activity Score in 28 joints.

  • t-test.

Responders, no. (%)66 (76.7)10 (55.6)56 (82.4) 37 (77.1)4 (40)33 (86.8) 
Nonresponders,  no. (%)20 (23.3)8 (44.4)12 (17.6) 11 (22.9)6 (60)5 (13.2) 
DAS28        
 Day 05.36 ± 1.025.15 ± 0.915.42 ± 1.05 5.46 ± 1.045.04 ± 0.8055.601 ± 1.0690.129
 Month 6 3.47 ± 1.163.19 ± 0.37    0.288
 Month 12     3.71 ± 0.993.31 ± 0.990.303
ΔDAS28 1.69 ± 1.312.23 ± 1.190.098 1.334 ± 1.372.29 ± 1.470.071
Figure 1.

Improvement in Disease Activity Score in 28 joints (DAS28) with etanercept treatment between month 0 (M0) and month 6 (M6) in 18 patients with tumor necrosis factor α (TNFα) −308 A/G genotype and 68 patients with TNFα −308 G/G genotype. Thick horizontal lines represent median DAS28 scores and thin lines represent mean DAS28 scores.

DISCUSSION

We followed 86 patients with RA who were receiving etanercept without associated disease-modifying antirheumatic drugs. Clinical response was evaluated using the DAS28 index after 6 months of treatment for all patients and again after 12 months of treatment in 48 patients. −308 A/G polymorphism in the TNFα gene was analyzed in every patient.

We found that patients with RA with the −308 G/G genotype had better responses to etanercept than patients with the −308 A/G genotype after 6 months and 1 year of treatment. Of the 86 patients, 13 had previously been treated with infliximab without success. Among these, the percentage with a positive response (70%) to etanercept at 6 months was not different from that of the entire group and the difference of outcome between patients with a G/G or a G/A genotype was not significant. Indeed, among the 4 etanercept nonresponders, 3 had a G/G genotype and 1 had an A/G genotype. Among the 9 etanercept responders, 4 had an A/G genotype and 5 had a G/G genotype. Thus, inclusion of 13 patients who had previously escaped infliximab in this study of etanercept treatment outcome may have made our overall data less significant.

The influence of polymorphisms in the promoter of the TNFα gene on outcome of TNFα suppressive therapy in RA has been studied previously in different populations. A consensus is now building on 2 facts (1). First, polymorphisms in the promoter of the TNFα gene can influence the level of production of TNFα. Second, TNFα polymorphisms associated with high TNFα production are also associated with poor response to TNFα inhibition.

The influence of polymorphisms in the promoter of the TNFα gene on the production of TNFα has been studied in western populations and in Asian individuals. In the West, A/G polymorphism at position −308 of the TNFα gene has been shown to control binding of a transcription factor and to influence TNFα production in whole blood culture after stimulation by LPS (2, 3). In this assay, the −308A allele is associated with high LPS-induced TNFα production (3, 4). In Asian persons, the −308 A/G polymorphism hardly exists, and 100% of patients express the −308G allele. Nevertheless, a C/T polymorphism at position −857 of the TNFα gene controls the binding of the Oct-1 transcription factor and a high (−857C) or low (−857T) production of TNFα.

At least 6 studies have suggested that TNFα gene polymorphisms associated with low production of TNFα are also associated with better outcome of TNF suppression in RA. We found that French patients with RA with the −308 G/G TNFα genotype respond to infliximab better than patients with the −308 A/G genotype (5).

Similar results were found in studies of patients from Hungary (10), Sweden (11), Spain (12, 13), and Switzerland (14). In these 4 populations, patients with −308 G/G homozygotes (low TNFα producers in vitro) respond to infliximab or etanercept better than patients with A/G or A/A genotypes. In the Korean population, the −308A allele (high producer) is absent. However, another polymorphism at position −857 of the TNFα gene is associated with high (−857C) or low (−857T) production of TNFα in vitro. In this population, the −857T allele is also associated with better response to etanercept in patients with RA (15). Thus, our study of the influence of −308 A/G TNFα polymorphism on the outcome of etanercept therapy in RA confirms that genetic markers of high TNFα production are associated with worse outcome of TNF suppressive treatment.

The most straightforward interpretation of these results is that standard etanercept treatment is overwhelmed by high endogenous TNFα production in patients with a −308 A/G TNFα genotype. However, we did not demonstrate this point, which remains a mere hypothesis. An alternative explanation can be that the −308A allele is in linkage disequilibrium with other genes that contribute to severity of RA or resistance to etanercept. Some of these genes may be carried by the HLA–DR3 haplotype, which usually carries a −308A TNFα allele and is rich in autoimmunity-favoring genes (16).

Our data have to be seen under a practical view. Etanercept is an efficient yet expensive treatment for RA. It can cause severe infections. The −308 A/G polymorphism testing enables distinguishing between patients who have an 80% likelihood of responding to etanercept from patients who have <50% likelihood. This information may prove useful in fragile patients and, more generally, in any cost-conscious health care system.

AUTHOR CONTRIBUTIONS

Dr. Jean Roudier had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Guis, Mugnier, Serre, Jean Roudier, Chantal Roudier.

Acquisition of data. Guis, Balandraud, Auger, Toussirot, Wendling, Mattei, Nogueira, Legeron, Landt, Jean Roudier, Chantal Roudier.

Analysis and interpretation of data. Guis, Balandraud, Bouvenot, Nogueira, Serre, Jean Roudier, Chantal Roudier.

Manuscript preparation. Guis, Balandraud, Mugnier, Jean Roudier.

Statistical analysis. Balandraud, Bouvenot, Jean Roudier.

Ancillary