- Email: [email protected]
TGF beta1 polymorphisms are candidate predictors of the clinical response to rituximab in rheumatoid arthritis
Joint Bone Spine 79 (2012) 471–475
Available online at
www.sciencedirect.com
Original article
TGF beta1 polymorphisms are candidate predictors of the clinical response to rituximab in rheumatoid arthritis Claire Immediato Daïen a,1 , Sylvie Fabre a,1 , Cécile Rittore a,b , Stephan Soler a,b , Vincent Daïen c , Gautier Tejedor b , Doris Cadart a , Nicolas Molinari e , Jean-Pierre Daurès c , Christian Jorgensen a,d,∗ , Isabelle Touitou a,b,d a
Inserm U844, unité génétique clinique, département thérapeutique et médecine physique ostéoarticulaire, CHU de Montpellier, 34295 Montpellier, France Unité médicale des maladies auto-inflammatoires (centre de référence), laboratoire de génétique, CHU de Montpellier, 34295 Montpellier, France c Laboratoire de biostatistiques, institut universitaire de recherche clinique, 34295 Montpellier, France d Université de Montpellier-1, UM1, 34295 Montpellier, France e UMR729 MISTEA, Montpellier SupAcro, 34295 Montpellier, France b
a r t i c l e
i n f o
Article history: Accepted 14 October 2011 Available online 29 November 2011 Keywords: Rheumatoid arthritis Rituximab Biomarker TGF beta1 SNP
a b s t r a c t Objective: To evaluate the association between several candidate single-nucleotide polymorphisms (SNPs) and responsiveness to rituximab in patients with rheumatoid arthritis (RA). Methods: Sixty-three RA patients were included. Nine genes (13 SNPs) were subsequently analyzed, including those coding for cytokines involved in synovitis (IL10, LTA, TGF1, TNF-␣, TNF receptor II) and genes associated with RA susceptibility (-C5 TRAF1, STAT4, TNFAIP3 and PTPN22). Results: Forty-four patients were defined as responders and 19 as nonresponders. TGF1 Codon 10 and TGF1 Codon 25 SNPs were both associated with clinical response (probability to respond to treatment with the Codon 10 C/T genotype: OR = 1.6; P = 0.002, and with the Codon 25 G/C genotype: OR = 1.6; P = 0.025). The probability to be a responder when the TGF Codon10 C/T and TGF Codon 25 G/C genotypes were co-inherited, doubled (OR = 2.6; P = 0.008). Conclusion: The TGF1 SNPs are associated with a good response to rituximab therapy and as such could be useful genetic biomarkers in predicting therapy outcome. © 2011 Société franc¸aise de rhumatologie. Published by Elsevier Masson SAS. All rights reserved.
1. Introduction Rituximab is a chimeric monoclonal antibody directed against CD20, a pan- B-cell surface marker, which significantly improves disease symptoms in many patients with active rheumatoid arthritis (RA), including those that do not respond to methotrexate (MTX) or anti-TNFalpha (anti-TNF␣) agents [1–6]. Rituximab has also shown efficacy in treatment of RA complications such as AA-amyloidosis [7]. However, approximately 30% of RA patients do not respond to this treatment [8]. Furthermore, rituximab is expensive and is associated with reasonable but significant risks of infection [9], subsequent to B-cell depletion. Biomarkers differentiating between responders and nonresponders would, therefore, be
∗ Corresponding author. Department of Physical Medicine & Therapeutics of Osteoarticular Diseases, Inserm U844, hôpital Saint Eloi, bâtiment INM, 80, rue Augustin-Fliche, 34295 Montpellier cedex 5, France. Tel.: +33 4 67 33 77 96; fax: +33 4 67 33 72 27. E-mail address: [email protected] (C. Jorgensen). 1 The authors contributed equally to this article.
useful to allow selection of appropriate patients before initiation of the rituximab treatment. Several markers have been investigated to predict response to treatment. Patients with positive rheumatoid factor (RF) or anticitrullinated protein antibodies (ACPA) are more likely to respond to rituximab [10]. Moreover, changes in synovial cells after treatment [11–13] or the number of IgD+ CD27+ memory B-cells at the time of B-cell recovery [14] could also predict therapeutic response, but only after rituximab initiation. Genetic markers could be useful in daily practice because they do not vary with time, as analysis is carried out on patient’s DNA. Moreover, DNA extraction is less invasive than synovial biopsies, requiring just a few drops of blood. Some genetic markers, such as TNF␣ polymorphisms, have been identified to predict responses to anti-TNF␣ treatment in Caucasian and Korean RA patients [15–17]. However, no genetic markers have yet been identified in patients treated with rituximab. This study aimed to assess the potential of a panel of 13 candidate single-nucleotide polymorphisms (SNPs) to act as predictive markers of rituximab response in RA patients. Some of the SNPs were selected on the basis that they had already been associated with responses to biotherapies (anti-TNF␣
1297-319X/$ – see front matter © 2011 Société franc¸aise de rhumatologie. Published by Elsevier Masson SAS. All rights reserved. doi:10.1016/j.jbspin.2011.10.007
472
C.I. Daïen et al. / Joint Bone Spine 79 (2012) 471–475
or anti-IL1) or are located within genes which code for cytokines (Transforming growth factor beta 1: TGF1 Codon 10 [rs1800470] and 25 [rs1800471]; Lymphotoxin alpha: rs1041981 and rs909253; Tumor necrosis factor alpha: rs1799724, rs 1800629, r80267959, their receptor TNFRII: rs1061622; interleukin 10: rs1800896). Others were genes which have previously been shown to be involved in increased RA susceptibility (PTPN22: rs2476601, STAT4: rs7574865, TRAF1-C5: rs1081848 and TNFAIP3: rs6920220), as identified by Genome-wide association studies (GWAS) [18,19]. 2. Methods 2.1. Patients All patients who had received rituximab infusions in the rheumatology unit (1000 mg intravenously once at day 0 and day 15) for RA, satisfying the American College of Rheumatology (ACR) criteria, revised in 1987 [20], were included. Rituximab was given to patients with RA which was resistant to at least one disease modifying antirheumatic drug (DMARD), according to international guidelines. Clinical response was retrospectively determined from medical records, in accordance with the European League Against Rheumatism (EULAR) criteria. Disease activity score (DAS28) values preceding infusion and the lowest DAS28 values between 3 and 6 months after rituximab treatment were used. To be considered as responsive to treatment the post-rituximab DAS28 value had to be lower than 5.1 and the DAS28 variation between baseline and posttreatment had to be higher than 0.6. The study was approved by the local ethical committee (Montpellier, France) in accordance with the Helsinki Declaration of 1975 (as revised in 1983) and informed consent (including genetic testing consent) was obtained from each patient. Sixty-three patients with RA, all Caucasian, were eligible for this study. Baseline characteristics are summarized in Table 1. Mean age at baseline was 54.7 years, with 77.8% of patients being female. The presence of RF was demonstrated in 73% of patients and 77.8% were positive for ACPA. Radiographic erosions were found in 82.5% of cases. An associated DMARD (methotrexate, leflunomide or sulfasalazine) was given to 75% of nonresponders and 64% of responders to rituximab. The mean number of previous treatments did not significantly differ between the two groups (1.5 and 1.7 respectively). 2.2. Determination of RF and ACPA Nephelometry was used to measure RF (Behring Nephelometer Analyzer II; Dade Behring, Inc., Deerfield, IL, USA), with a detection limit of 15 U/mL. A second-generation anti-CCP-2 antibody enzyme-linked immunosorbent assay (ELISA) (Immunoscan RA Table 1 Baseline characteristics of RA patients.
Number Age (yrs) Females + RF + ACPA Erosion RA duration (yrs) Number of previous biologic therapies Initial DAS28
Responders
Nonresponders
44 54.4 (20–76) 30 (68.2) 34 (77.3) 35 (79.5) 38 (86.4) 12.5 (2–57) 1.5 (0–4)
19 55.4 (29–74) 18 (94.8) 12 (63.2) 14 (73.7) 14 (73.7) 12 (1–38) 1.7 (0–3)
5.5 (3.3–6.4)
5.0 (2.7–7.5)
Both quantitative values (range) and qualitative values (percentage) are represented. Median values are given for RA duration and initial DAS28. Other quantitative values correspond to mean values.
Mark 2; Euro-Diagnostica, Arhem, The Netherlands) with a cutoff level of 25 arbitrary U/mL was used to determine the presence of ACPA, in accordance with manufacturer’s instructions. 2.3. Genotyping For each patient, four drops of peripheral blood were collected onto an FTA elute card (WhatmanTM, Kent, England), and genomic DNA was extracted according to the manufacturer’s instructions. All SNPs were PCR amplified using published or in-house designed primers (primer sequences and PCR conditions available upon request). Restriction fragment length polymorphisms (RFLP) were sought to analyze 13 SNPs, (IL10-1087 (EcoNI), LTA+ 249 (NcoI), PTPN22 (RsaI), STAT4 (MseI), TRAF1 (Bsp1286I), TGF1 Codon 10 (MspA1I) and TGF1 Codon 25 (BglI), TNFa-857 (TaiI), TNFa-308 (NcoI), TNFa+ 488 (NlaIII), TNFRII-codon196 (NlaIII) and TNFAIP3 (BslI)), with an appropriate restriction enzyme. After digestion of the amplicons for 3 hours at 37 ◦ C (65 ◦ C for TaiI), the fragments were separated on a 10% (or 15% for TGF1 Codon 10) polyacrylamide gel. LTA+ 720 was analyzed using tetra- primer amplification refractory mutation system PCR, a method that involves a multiplex PCR including two outer (forward and reverse) primers, common to the two alleles, and two inner (forward and reverse) primers specific to the A and C sequence variations. Blind analysis of alleles was carried out by two different investigators. 2.4. Statistical analysis Patient characteristics and responses to treatment were described using median and range or mean and range for continuous variables and frequencies and proportions for categorical variables. Fisher’s exact test was performed to analyze the impact of the 13 SNPs on the response to rituximab. The Cochran-Armitage tendency test was used to detect a trend between each SNP and rituximab response. A multivariate analysis using logistic regression was performed. All SNPs which were significantly associated to a good response to treatment in the univariate analysis were entered in the model. A stepwise selection of the variables was used. The alpha-to-enter and alpha-to-exit values were set at 0.20 and 0.5 respectively. The significance of adding or removing a variable from the logistic model was determined by the maximum likelihood ratio test. The goodness-of-fit of the models was assessed using the Hosmer and Lemeshow Chi-square test. To estimate the relationship between the two SNPs (TGF1 Codon 25 and TGF1 Codon 10) which were significantly associated to the response to treatment, the adjusted relative risk was calculated. To correct for multiple testing, we employed simple Bonferroni adjustment based on the 13 SNPs that were tested for association to treatment response. A Boostrap resampling approach was used to assess the robustness of the results when findings did not pass the overly conservative Bonferroni adjustment. The statistical significance threshold was set at 5%. Statistical analyses were performed using SAS version 9.2 (SAS Institute, Cary, North Carolina). 3. Results 3.1. Response to rituximab Among the 63 patients included, 44 were responders and 19 were nonresponders. Baseline characteristics are summarized in Table 1. Differences in baseline characteristics were not significant between the two groups. The mean age was 54.4 years in responders and 55.4 years in nonresponders with 68.2% and 94.8%
C.I. Daïen et al. / Joint Bone Spine 79 (2012) 471–475
respectively of females in each group. The number of patients that tested positive for RF and ACPA tended to be higher in responders, when compared to nonresponders (77.3% and 63.2% respectively for RF and 79.5% and 73.7% respectively for ACPA). 3.2. Association of candidate SNPs with response status to rituximab The frequency of the different genotypes, according to responder status is shown in Table 2. IL10 (rs1800896), LTA (rs909253 and rs1041981), TNFA (rs1800629, rs80267959 and rs1799724), TNF receptor II (rs1061622), receptor-associated factor 1 (-C5 TRAF1; rs1081848), STAT4 (rs7574865), TNFAIP3 (rs6920220) and PTPN22 (rs2476601) were not associated with response to rituximab. The only two SNPs which could be shown to have an association with therapeutic response, based on EULAR criteria, were Codon 10 (rs1800470) and TGF1 Codon 25 (rs1800471). The TGF1 Codon 10 C/T genotype significantly increased the probability of responding to rituximab treatment (OR = 1.6; CI95% 1.2–2.3; P = 0.002). Among responders, 18% had a T/T genotype and 61% had a C/T genotype. Among nonresponders, 58% had a T/T genotype and 21% had a C/T genotype. C/C alleles were found equally in responders and nonresponders (20.5% and 21% respectively). The association between response to rituximab and TGF1 Codon 10 polymorphism remained significant after Bonferroni correction. The probability of responding to rituximab with the TGF1 Codon 25 G/C genotype also increased similarly (OR = 1.6; CI 95% 1.3–1.9; P = 0.025). All patients with a G/C genotype were responders, as compared to 63% of patients with a G/G genotype. A bootstrap resampling approach was used to assess the robustness of this result. Eighty-five percent of these samples had a significant association. The combination of the two SNPs led to an even better response to rituximab (probability of being responsive to treatment with a TGF1 Codon 25 G/C and/or TGF1 Codon 10 C/T genotype, when compared with Codon 25 G/G and Codon 10 T/T genotypes: OR = 2.6; CI 95% 1.4–4.6; P = 0.008) (Fig. 1). The A/C genotype for the LTA+ 720 SNP also tended to appear more frequently in patients responding to treatment (P = 0.075), with expression in 48% of responders and 37% of nonresponders. Multivariate analysis did not show any significant gene profile. 3.3. Patient characteristics according to TGFˇ1 Codon 10 and Codon 25 genotype In Table 3, the baseline characteristics of patients are compared according to their TGF1 Codon 10 (C/T or T/T) and TGF1 Codon 25 genotypes (G/C or G/G). Patients with the G/C genotype were all
Fig. 1. EULAR response to rituximab according to TGF1 Codon 10, TGF1 Codon 25 and the association of the two SNPs. Black bars represent responders (R); white bars represent nonresponders (NR).
473
responsive to treatment. There were more females in the G/G group. There were no other statistically significant differences. Only 60% of G/C expressing patients were positive for ACPA, compared with 82% of G/G expressing patients. 3.4. TGFˇ1 SNPs predict response to rituximab but not to anti-TNF˛ To investigate the specificity of TGF1 SNPs in predicting the response to rituximab, we similarly studied the association of TGF1 Codon 25 with the degree of responsiveness to anti-TNF␣ (etanercept, adalimumab and infliximab) treatment in 56 patients (not shown). The baseline characteristics within this group were similar to our population. The mean age was 54 years, 83% were female, 68% were positive for RF and 73% were positive for ACPA. The mean RA duration was 12.4 years and 75% were responsive to treatment. The only statistically significant difference between the baseline characteristics of this group and our own population was the average number of previous biologic therapies, since anti-TNF␣ was a first-line therapy in all patients, whereas patients treated with rituximab received an average of 1.6 previous biologic treatments. There was no correlation between the TGF Codon 25 SNP genotype and response to anti-TNF␣ treatment (four responders had the G/C genotype and 37 responders the G/G genotype; one nonresponder had the G/C genotype and 14 the G/G genotype; P = 1.0). 4. Discussion Many biologic treatments have been proposed for patients with severe RA, refractory to methotrexate but, to date, no biomarkers have been identified, which have high predictive values of clinical response. This study aimed to assess the predictive response to rituximab in RA patients by analyzing a panel of nine candidate genes (13 SNPs). Using a noninvasive technique, requiring less than four drops of blood from each patient, we found a significant association between EULAR response to rituximab in RA patients and TGF1 Codon 25 and Codon 10 polymorphisms. These SNPs are involved in TGF1 production, which is known to inhibit B cell activation and immunoglobulin production. The RA population we studied corresponded to that which can be seen in daily rheumatology practice [21], with 69.8% of patients responding to rituximab treatment, a mean patient age of 55 years and a mean disease duration of 12 years. The presence of RF and ACPA were detected more frequently in responders, when compared to nonresponders, but the difference was not significant (RF: 77.3% vs. 63.2% respectively; P = 0.25; ACPA: 79.5% vs. 73.7% respectively P = 0.66). The frequency of C/C, T/T and C/T TGF1 Codon 10 genotype expressions in RA sufferers corresponded to those found in healthy controls (C/C: 21% vs 19% respectively; T/T: 30% vs 29% respectively; C/T: 49% vs 52% respectively) [22]. The frequency of C/C, G/C and G/G TGF1 Codon 25 genotypes in RA sufferers also corresponded to those found in literature for Caucasian controls (C/C: 0 vs. 0% respectively, G/C: 16.9 vs. 15.2% respectively and G/G: 83.1 vs. 84.8% respectively) [23]. Our candidate gene study led to the identification of TGF1 Codon 10 and TGF1 Codon 25 as being potential predictive factors of rituximab response (likelihood of responding to treatment with the Codon 10 C/T genotype: OR = 1.6; CI 95% 1.2–2.3; P = 0.002; likelihood of responding to treatment with the Codon 25 G/C genotype: OR = 1.6; CI 95% 1.3–1.9; P = 0.025). TGF1 SNPs seemed to be specifically associated to the response to rituximab treatment, as we did not find any similar association with a response to antiTNF␣ treatment. No association between TGF1 Codon 25 or TGF1 Codon 10 and ACPA or RF was found. Therefore, TGF1 Codon
474
C.I. Daïen et al. / Joint Bone Spine 79 (2012) 471–475
Table 2 Genotypes of the RA patients according to their EULAR response to rituximab treatment. SNP
Responders
Nonresponders
P value
TGF1-10 (rs1800470) TGF1-25 (rs1800471) LTA 720 (rs1041981) LTA 249 (rs909253) TNFa 308 (rs 1800629) PTPN22 (rs2476601) IL10 1087 (rs1800896) TNFa 488 (r80267959) TNFa 857 (rs1799724) STAT4 (rs7574865) TNFAIP3 (rs6920220) TRAF1 C5 (rs1081848) TNFRII 196 (rs1061622)
T/C = 27, T/T = 8, C/C = 9 G/C = 11, G/G = 33 A/A = 2, C/A = 21, C/C = 19 C/C = 2, T/C = 23, T/T = 19 G/A = 14, G/G = 28 C/C = 33, C/T = 8, T/T = 1 A/A = 11, A/G = 23, G/G = 6 A/A = 0, G/A = 12, G/G = 31 C/C = 28, C/T = 15, T/T = 0 G/G = 18, T/G = 21, T/T = 4 A/A = 1, G/A = 17, G/G = 24 A/A = 7, G/A = 26, G/G = 10 G/G = 4, T/G = 13, T/T = 24
T/C = 4, T/T = 11, C/C = 4 G/C = 0, G/G = 19 A/A = 0, C/A = 7, C/C = 12 C/C = 1, T/C = 6, T/T = 12 G/A = 3, G/G = 16 C/C = 17, C/T = 2, T/T = 0 A/A = 7, A/G = 7, G/G = 5 A/A = 1, G/A = 4, G/G = 14 C/C = 13, C/T = 5, T/T = 1 G/G = 8, T/G = 8, T/T = 3 A/A = 1, G/A = 8, G/G = 10 A/A = 4, G/A = 8, G/G = 7 G/G = 0, T/G = 6, T/T = 13
0.002* 0.025* 0.075 0.084 0.332 0.349 0.417 0.449 0.644 0.678 0.716 0.732 0.771
*
Significant association (P < 0.05).
Table 3 Patient characteristics according to TGF1 Codon 25 and Codon 10 genotypes. TGF1 alleles
Codon 10 C/T
Codon 10 T/T
Codon 10 C/C
Codon 25 G/C
Codon 25 G/G
Number Age Female Responders + RF + ACPA Erosion RA duration Number of previous biologic therapies Initial DAS28
31 55.5 (20–74) 21 (67.7) 27 (86.7) 24 (77.4) 27 (86.7) 27 (86.7) 14.0 (1–57) 2 (0–4) 5.41 (3.8–7.2)
19 57.5 (43–76) 18 (94.7) 8 (42) 11 (57.9) 13 (68.4) 15 (78.9) 14.5 (5–38) 2 (0–3) 5.61 (2.6–7.5)
13 53 (29–71) 9 (69.2) 9 (69.2) 11 (84.6) 10 (76.9) 10 (76.9) 12 (3–40) 2 (0–3) 5.45 (3.5–6.9)
11 56.5 (45–71) 5 (50) 10 (100) 8 (80) 6 (60) 10 (100) 14 (6–57) 2 (0–3) 5.12 (3.8–6.5)
52 53 (20–76) 48 (98) 30 (61) 34 (69) 40 (82) 39 (80) 12 (1–38) 2 (0–4) 5.34 (2.6–7.5)
Both quantitative values (range) and qualitative values (percentage) are represented. Mean values are given for age at RA diagnosis and initial DAS28 and median values for age, RA duration and number of previous therapies.
25 and TGF1 Codon 10 genotypes appear to be novel predictive factors and therefore could help to propose a targeted therapy to RA patients [24]. Both SNPs belong to a signal peptide sequence of TGF1. TGF1 Codon 25 polymorphism exchanges an arginine for a proline (p.Arg25Pro) and TGF1 Codon 10 exchanges a proline for a leucine (p.Pro10Leu). As the two SNPs are known to be in linkage disequilibrium [25], they could be similarly involved in the response to rituximab. However, the association of these two SNPs predicted better response to rituximab (likelihood of responding to treatment with both TGF Codon 25 G/C and TGF Codon 10 C/T genotypes when compared with Codon 25 G/G and Codon 10 T/T genotypes: OR = 2.6; CI 95% 1.4–4.6; P = 0.008), suggesting an additive effect. One could expect that patients with TGF Codon 10 C/C would response more than those with Codon 10 C/T but this was not the case. However, this subgroup was poorly represented (13 patients). Awad et al. [26] and Guo et al. [27] found that in vitro lymphocyte production of TGF1 was lower in Codon 10 T/C expressing Caucasian controls than in T/T genotype controls and in Codon 25G/C expressing than in G/G genotype subjects. Yokota et al. [28] and Yamada et al. [29] showed the opposite result but this work concerned Japanese patients. TGF1 is known to inhibit B cells and to decrease immunoglobulin secretion [30,31]. Thus, a genotype associated with a lower TGF response could be associated with higher IgG and antibody secretion. Elevated IgG levels have been associated with a good response to rituximab [8]. Therefore, if responders to rituximab in this study produce less TGF1, and thus more immunoglobulins, this may explain the enhanced treatment response observed here. However, this needs to be confirmed on a large prospective cohort of patients (Fig. 2). Using a candidate SNP approach, we found a significant association between EULAR response to rituximab in RA patients and TGF1 Codon 25 and Codon 10 polymorphisms. These SNPs are
Fig. 2. Possible explanation of the TGF1 SNP’s influence on rituximab (RTX) response. TGF1 SNPs seem to control the production of TGF1, which influences B cell maturation, activation and immunoglobulin production. Patients with TGF1 Codon 25 G/C allele + Codon 10 C/T allele would have more active B cells, which could explain why RTX would be more efficient than in patients with more quiescent B cells (25 G/G + 10 T/T).
involved in TGF1 production, which is known to inhibit B cell activation and immunoglobulin production. Our results need to be confirmed on a large prospective cohort. If confirmed, TGF1 polymorphisms could become useful predictive markers and guide appropriate prescription of rituximab to RA patients. Disclosure of interest C.I.D, C.R., S.S., V.D., G.T., D.C., N.M., J.-P.D. and I.T. declare that they have no conflicts of interest concerning this article.
C.I. Daïen et al. / Joint Bone Spine 79 (2012) 471–475
S.F. received honorarium from Roche, Neovacs, Novartis and Merk Serono laboratories for consulting, investigator of clinical trials. C.J. participated as principal investigator in clinical trials, but did not receive honorarium. Acknowledgments To Montpellier hospital (France) for promoting this study and to Jean Sibilia (Strasbourg, France) for his advices. References [1] Edwards JC, Szczepanski L, Szechinski J, et al. Efficacy of B-cell-targeted therapy with rituximab in patients with rheumatoid arthritis. N Engl J Med 2004;350:2572–81. [2] Emery P, Fleischmann R, Filipowicz-Sosnowska A, et al. The efficacy and safety of rituximab in patients with active rheumatoid arthritis despite methotrexate treatment: results of a phase IIB randomized, double-blind, placebo-controlled, dose-ranging trial. Arthritis Rheum 2006;54:1390–400. [3] Cohen SB, Emery P, Greenwald MW, et al. Rituximab for rheumatoid arthritis refractory to anti-tumor necrosis factor therapy: results of a multicenter, randomized, double-blind, placebo-controlled, phase III trial evaluating primary efficacy and safety at twenty-four weeks. Arthritis Rheum 2006;54:2793–806. [4] Duddy ME, Alter A, Bar-Or A. Distinct profiles of human B cell effector cytokines: a role in immune regulation? J Immunol 2004;172:3422–7. [5] Roosnek E, Lanzavecchia A. Efficient and selective presentation of antigenantibody complexes by rheumatoid factor B cells. J Exp Med 1991;173:487–9. [6] Takemura S, Klimiuk PA, Braun A, et al. T cell activation in rheumatoid synovium is B cell dependent. J Immunol 2001;167:4710–8. [7] Narvaez J, Hernandez MV, Ruiz JM, et al. Rituximab therapy for AA-amyloidosis secondary to rheumatoid arthritis. Joint Bone Spine 2011;78:101–3. [8] Sellam J, Hendel-Chavez H, Rouanet S, et al. B-cell activation biomarkers as predictive factors of the response to rituximab in rheumatoid arthritis. Arthritis Rheum 2011;63:933–8. [9] Guis S, Balandraud N, Bouvenot J, et al. Influence of -308 A/G polymorphism in the tumor necrosis factor alpha gene on etanercept treatment in rheumatoid arthritis. Arthritis Rheum 2007;57:1426–30. [10] Quartuccio L, Fabris M, Salvin S, et al. Rheumatoid factor positivity rather than anti-CCP positivity, a lower disability and a lower number of anti-TNF agents failed are associated with response to rituximab in rheumatoid arthritis. Rheumatology (Oxford) 2009;48:1557–9. [11] Teng YK, Levarht EW, Hashemi M, et al. Immunohistochemical analysis as a means to predict responsiveness to rituximab treatment. Arthritis Rheum 2007;56:3909–18. [12] Thurlings RM, Vos K, Wijbrandts CA, et al. Synovial tissue response to rituximab: mechanism of action and identification of biomarkers of response. Ann Rheum Dis 2008;67:917–25. [13] Kavanaugh A, Rosengren S, Lee SJ, et al. Assessment of rituximab’s immunomodulatory synovial effects (ARISE trial). 1: clinical and synovial biomarker results. Ann Rheum Dis 2008;67:402–8.
475
[14] Roll P, Dorner T, Tony HP. Anti-CD20 therapy in patients with rheumatoid arthritis: predictors of response and B cell subset regeneration after repeated treatment. Arthritis Rheum 2008;58:1566–75. [15] de Vries N, Tak PP. The response to anti-TNF-alpha treatment: gene regulation at the bedside. Rheumatology 2005;44:705–7. [16] Lee YH, Rho YH, Choi SJ, et al. Association of TNF-alpha -308 G/A polymorphism with responsiveness to TNF-alpha-blockers in rheumatoid arthritis: a metaanalysis. Rheumatol Int 2006;27:157–61. [17] Mugnier B, Balandraud N, Darque A, et al. Polymorphism at position -308 of the tumor necrosis factor alpha gene influences outcome of infliximab therapy in rheumatoid arthritis. Arthritis Rheum 2003;48:1849–52. [18] Coenen MJ, Gregersen PK. Rheumatoid arthritis: a view of the current genetic landscape. Genes Immun 2009;10:101–11. [19] Boissier MC. Cell and cytokine imbalances in rheumatoid synovitis. Joint Bone Spine 2011;78:230–4. [20] Arnett FC, Edworthy SM, Bloch DA, et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988;31:315–24. [21] Assous N, Gossec L, Dieude P, et al. Rituximab therapy in rheumatoid arthritis in daily practice. J Rheumatol 2008;35:31–4. [22] Pokorny V, Chau J, Wu L, et al. Transforming growth factor beta 1 gene (HSTGFB1) nucleotide T869C (codon 10) polymorphism is not associated with prevalence or severity of rheumatoid arthritis in a Caucasian population. Ann Rheum Dis 2003;62:907–8. [23] Oen K, Malleson PN, Cabral DA, et al. Cytokine genotypes correlate with pain and radiologically defined joint damage in patients with juvenile rheumatoid arthritis. Rheumatology 2005;44:1115–21. [24] Prati C, Wendling D. Using targeted therapies alone for rheumatoid arthritis. Joint Bone Spine 2011;78:1–3. [25] Syrris P, Carter ND, Metcalfe JC, et al. Transforming growth factorbeta1 gene polymorphisms and coronary artery disease. Clin Sci 1998;95: 659–67. [26] Awad MR, El-Gamel A, Hasleton P, et al. Genotypic variation in the transforming growth factor-beta1 gene: association with transforming growth factor-beta1 production, fibrotic lung disease, and graft fibrosis after lung transplantation. Transplantation 1998;66:1014–20. [27] Guo Z, Binswanger U, Knoflach A. Role of codon 10 and codon 25 polymorphisms on TGF-beta 1 gene expression and protein synthesis in stable renal allograft recipients. Transplant Proc 2002;34:2904–6. [28] Yokota M, Ichihara S, Lin TL, et al. Association of a T29C polymorphism of the transforming growth factor-beta1 gene with genetic susceptibility to myocardial infarction in Japanese. Circulation 2000;101:2783–7. [29] Yamada Y, Miyauchi A, Goto J, et al. Association of a polymorphism of the transforming growth factor-beta1 gene with genetic susceptibility to osteoporosis in postmenopausal Japanese women. J Bone Miner Res 1998;13: 1569–76. [30] Lebman DA, Edmiston JS. The role of TGF-beta in growth, differentiation, and maturation of B lymphocytes. Microbes Infect 1999;1: 1297–304. [31] Kehrl JH, Thevenin C, Rieckmann P, et al. Transforming growth factor-beta suppresses human B lymphocyte Ig production by inhibiting synthesis and the switch from the membrane form to the secreted form of Ig mRNA. J Immunol 1991;146:4016–23.
Available online at
www.sciencedirect.com
Original article
TGF beta1 polymorphisms are candidate predictors of the clinical response to rituximab in rheumatoid arthritis Claire Immediato Daïen a,1 , Sylvie Fabre a,1 , Cécile Rittore a,b , Stephan Soler a,b , Vincent Daïen c , Gautier Tejedor b , Doris Cadart a , Nicolas Molinari e , Jean-Pierre Daurès c , Christian Jorgensen a,d,∗ , Isabelle Touitou a,b,d a
Inserm U844, unité génétique clinique, département thérapeutique et médecine physique ostéoarticulaire, CHU de Montpellier, 34295 Montpellier, France Unité médicale des maladies auto-inflammatoires (centre de référence), laboratoire de génétique, CHU de Montpellier, 34295 Montpellier, France c Laboratoire de biostatistiques, institut universitaire de recherche clinique, 34295 Montpellier, France d Université de Montpellier-1, UM1, 34295 Montpellier, France e UMR729 MISTEA, Montpellier SupAcro, 34295 Montpellier, France b
a r t i c l e
i n f o
Article history: Accepted 14 October 2011 Available online 29 November 2011 Keywords: Rheumatoid arthritis Rituximab Biomarker TGF beta1 SNP
a b s t r a c t Objective: To evaluate the association between several candidate single-nucleotide polymorphisms (SNPs) and responsiveness to rituximab in patients with rheumatoid arthritis (RA). Methods: Sixty-three RA patients were included. Nine genes (13 SNPs) were subsequently analyzed, including those coding for cytokines involved in synovitis (IL10, LTA, TGF1, TNF-␣, TNF receptor II) and genes associated with RA susceptibility (-C5 TRAF1, STAT4, TNFAIP3 and PTPN22). Results: Forty-four patients were defined as responders and 19 as nonresponders. TGF1 Codon 10 and TGF1 Codon 25 SNPs were both associated with clinical response (probability to respond to treatment with the Codon 10 C/T genotype: OR = 1.6; P = 0.002, and with the Codon 25 G/C genotype: OR = 1.6; P = 0.025). The probability to be a responder when the TGF Codon10 C/T and TGF Codon 25 G/C genotypes were co-inherited, doubled (OR = 2.6; P = 0.008). Conclusion: The TGF1 SNPs are associated with a good response to rituximab therapy and as such could be useful genetic biomarkers in predicting therapy outcome. © 2011 Société franc¸aise de rhumatologie. Published by Elsevier Masson SAS. All rights reserved.
1. Introduction Rituximab is a chimeric monoclonal antibody directed against CD20, a pan- B-cell surface marker, which significantly improves disease symptoms in many patients with active rheumatoid arthritis (RA), including those that do not respond to methotrexate (MTX) or anti-TNFalpha (anti-TNF␣) agents [1–6]. Rituximab has also shown efficacy in treatment of RA complications such as AA-amyloidosis [7]. However, approximately 30% of RA patients do not respond to this treatment [8]. Furthermore, rituximab is expensive and is associated with reasonable but significant risks of infection [9], subsequent to B-cell depletion. Biomarkers differentiating between responders and nonresponders would, therefore, be
∗ Corresponding author. Department of Physical Medicine & Therapeutics of Osteoarticular Diseases, Inserm U844, hôpital Saint Eloi, bâtiment INM, 80, rue Augustin-Fliche, 34295 Montpellier cedex 5, France. Tel.: +33 4 67 33 77 96; fax: +33 4 67 33 72 27. E-mail address: [email protected] (C. Jorgensen). 1 The authors contributed equally to this article.
useful to allow selection of appropriate patients before initiation of the rituximab treatment. Several markers have been investigated to predict response to treatment. Patients with positive rheumatoid factor (RF) or anticitrullinated protein antibodies (ACPA) are more likely to respond to rituximab [10]. Moreover, changes in synovial cells after treatment [11–13] or the number of IgD+ CD27+ memory B-cells at the time of B-cell recovery [14] could also predict therapeutic response, but only after rituximab initiation. Genetic markers could be useful in daily practice because they do not vary with time, as analysis is carried out on patient’s DNA. Moreover, DNA extraction is less invasive than synovial biopsies, requiring just a few drops of blood. Some genetic markers, such as TNF␣ polymorphisms, have been identified to predict responses to anti-TNF␣ treatment in Caucasian and Korean RA patients [15–17]. However, no genetic markers have yet been identified in patients treated with rituximab. This study aimed to assess the potential of a panel of 13 candidate single-nucleotide polymorphisms (SNPs) to act as predictive markers of rituximab response in RA patients. Some of the SNPs were selected on the basis that they had already been associated with responses to biotherapies (anti-TNF␣
1297-319X/$ – see front matter © 2011 Société franc¸aise de rhumatologie. Published by Elsevier Masson SAS. All rights reserved. doi:10.1016/j.jbspin.2011.10.007
472
C.I. Daïen et al. / Joint Bone Spine 79 (2012) 471–475
or anti-IL1) or are located within genes which code for cytokines (Transforming growth factor beta 1: TGF1 Codon 10 [rs1800470] and 25 [rs1800471]; Lymphotoxin alpha: rs1041981 and rs909253; Tumor necrosis factor alpha: rs1799724, rs 1800629, r80267959, their receptor TNFRII: rs1061622; interleukin 10: rs1800896). Others were genes which have previously been shown to be involved in increased RA susceptibility (PTPN22: rs2476601, STAT4: rs7574865, TRAF1-C5: rs1081848 and TNFAIP3: rs6920220), as identified by Genome-wide association studies (GWAS) [18,19]. 2. Methods 2.1. Patients All patients who had received rituximab infusions in the rheumatology unit (1000 mg intravenously once at day 0 and day 15) for RA, satisfying the American College of Rheumatology (ACR) criteria, revised in 1987 [20], were included. Rituximab was given to patients with RA which was resistant to at least one disease modifying antirheumatic drug (DMARD), according to international guidelines. Clinical response was retrospectively determined from medical records, in accordance with the European League Against Rheumatism (EULAR) criteria. Disease activity score (DAS28) values preceding infusion and the lowest DAS28 values between 3 and 6 months after rituximab treatment were used. To be considered as responsive to treatment the post-rituximab DAS28 value had to be lower than 5.1 and the DAS28 variation between baseline and posttreatment had to be higher than 0.6. The study was approved by the local ethical committee (Montpellier, France) in accordance with the Helsinki Declaration of 1975 (as revised in 1983) and informed consent (including genetic testing consent) was obtained from each patient. Sixty-three patients with RA, all Caucasian, were eligible for this study. Baseline characteristics are summarized in Table 1. Mean age at baseline was 54.7 years, with 77.8% of patients being female. The presence of RF was demonstrated in 73% of patients and 77.8% were positive for ACPA. Radiographic erosions were found in 82.5% of cases. An associated DMARD (methotrexate, leflunomide or sulfasalazine) was given to 75% of nonresponders and 64% of responders to rituximab. The mean number of previous treatments did not significantly differ between the two groups (1.5 and 1.7 respectively). 2.2. Determination of RF and ACPA Nephelometry was used to measure RF (Behring Nephelometer Analyzer II; Dade Behring, Inc., Deerfield, IL, USA), with a detection limit of 15 U/mL. A second-generation anti-CCP-2 antibody enzyme-linked immunosorbent assay (ELISA) (Immunoscan RA Table 1 Baseline characteristics of RA patients.
Number Age (yrs) Females + RF + ACPA Erosion RA duration (yrs) Number of previous biologic therapies Initial DAS28
Responders
Nonresponders
44 54.4 (20–76) 30 (68.2) 34 (77.3) 35 (79.5) 38 (86.4) 12.5 (2–57) 1.5 (0–4)
19 55.4 (29–74) 18 (94.8) 12 (63.2) 14 (73.7) 14 (73.7) 12 (1–38) 1.7 (0–3)
5.5 (3.3–6.4)
5.0 (2.7–7.5)
Both quantitative values (range) and qualitative values (percentage) are represented. Median values are given for RA duration and initial DAS28. Other quantitative values correspond to mean values.
Mark 2; Euro-Diagnostica, Arhem, The Netherlands) with a cutoff level of 25 arbitrary U/mL was used to determine the presence of ACPA, in accordance with manufacturer’s instructions. 2.3. Genotyping For each patient, four drops of peripheral blood were collected onto an FTA elute card (WhatmanTM, Kent, England), and genomic DNA was extracted according to the manufacturer’s instructions. All SNPs were PCR amplified using published or in-house designed primers (primer sequences and PCR conditions available upon request). Restriction fragment length polymorphisms (RFLP) were sought to analyze 13 SNPs, (IL10-1087 (EcoNI), LTA+ 249 (NcoI), PTPN22 (RsaI), STAT4 (MseI), TRAF1 (Bsp1286I), TGF1 Codon 10 (MspA1I) and TGF1 Codon 25 (BglI), TNFa-857 (TaiI), TNFa-308 (NcoI), TNFa+ 488 (NlaIII), TNFRII-codon196 (NlaIII) and TNFAIP3 (BslI)), with an appropriate restriction enzyme. After digestion of the amplicons for 3 hours at 37 ◦ C (65 ◦ C for TaiI), the fragments were separated on a 10% (or 15% for TGF1 Codon 10) polyacrylamide gel. LTA+ 720 was analyzed using tetra- primer amplification refractory mutation system PCR, a method that involves a multiplex PCR including two outer (forward and reverse) primers, common to the two alleles, and two inner (forward and reverse) primers specific to the A and C sequence variations. Blind analysis of alleles was carried out by two different investigators. 2.4. Statistical analysis Patient characteristics and responses to treatment were described using median and range or mean and range for continuous variables and frequencies and proportions for categorical variables. Fisher’s exact test was performed to analyze the impact of the 13 SNPs on the response to rituximab. The Cochran-Armitage tendency test was used to detect a trend between each SNP and rituximab response. A multivariate analysis using logistic regression was performed. All SNPs which were significantly associated to a good response to treatment in the univariate analysis were entered in the model. A stepwise selection of the variables was used. The alpha-to-enter and alpha-to-exit values were set at 0.20 and 0.5 respectively. The significance of adding or removing a variable from the logistic model was determined by the maximum likelihood ratio test. The goodness-of-fit of the models was assessed using the Hosmer and Lemeshow Chi-square test. To estimate the relationship between the two SNPs (TGF1 Codon 25 and TGF1 Codon 10) which were significantly associated to the response to treatment, the adjusted relative risk was calculated. To correct for multiple testing, we employed simple Bonferroni adjustment based on the 13 SNPs that were tested for association to treatment response. A Boostrap resampling approach was used to assess the robustness of the results when findings did not pass the overly conservative Bonferroni adjustment. The statistical significance threshold was set at 5%. Statistical analyses were performed using SAS version 9.2 (SAS Institute, Cary, North Carolina). 3. Results 3.1. Response to rituximab Among the 63 patients included, 44 were responders and 19 were nonresponders. Baseline characteristics are summarized in Table 1. Differences in baseline characteristics were not significant between the two groups. The mean age was 54.4 years in responders and 55.4 years in nonresponders with 68.2% and 94.8%
C.I. Daïen et al. / Joint Bone Spine 79 (2012) 471–475
respectively of females in each group. The number of patients that tested positive for RF and ACPA tended to be higher in responders, when compared to nonresponders (77.3% and 63.2% respectively for RF and 79.5% and 73.7% respectively for ACPA). 3.2. Association of candidate SNPs with response status to rituximab The frequency of the different genotypes, according to responder status is shown in Table 2. IL10 (rs1800896), LTA (rs909253 and rs1041981), TNFA (rs1800629, rs80267959 and rs1799724), TNF receptor II (rs1061622), receptor-associated factor 1 (-C5 TRAF1; rs1081848), STAT4 (rs7574865), TNFAIP3 (rs6920220) and PTPN22 (rs2476601) were not associated with response to rituximab. The only two SNPs which could be shown to have an association with therapeutic response, based on EULAR criteria, were Codon 10 (rs1800470) and TGF1 Codon 25 (rs1800471). The TGF1 Codon 10 C/T genotype significantly increased the probability of responding to rituximab treatment (OR = 1.6; CI95% 1.2–2.3; P = 0.002). Among responders, 18% had a T/T genotype and 61% had a C/T genotype. Among nonresponders, 58% had a T/T genotype and 21% had a C/T genotype. C/C alleles were found equally in responders and nonresponders (20.5% and 21% respectively). The association between response to rituximab and TGF1 Codon 10 polymorphism remained significant after Bonferroni correction. The probability of responding to rituximab with the TGF1 Codon 25 G/C genotype also increased similarly (OR = 1.6; CI 95% 1.3–1.9; P = 0.025). All patients with a G/C genotype were responders, as compared to 63% of patients with a G/G genotype. A bootstrap resampling approach was used to assess the robustness of this result. Eighty-five percent of these samples had a significant association. The combination of the two SNPs led to an even better response to rituximab (probability of being responsive to treatment with a TGF1 Codon 25 G/C and/or TGF1 Codon 10 C/T genotype, when compared with Codon 25 G/G and Codon 10 T/T genotypes: OR = 2.6; CI 95% 1.4–4.6; P = 0.008) (Fig. 1). The A/C genotype for the LTA+ 720 SNP also tended to appear more frequently in patients responding to treatment (P = 0.075), with expression in 48% of responders and 37% of nonresponders. Multivariate analysis did not show any significant gene profile. 3.3. Patient characteristics according to TGFˇ1 Codon 10 and Codon 25 genotype In Table 3, the baseline characteristics of patients are compared according to their TGF1 Codon 10 (C/T or T/T) and TGF1 Codon 25 genotypes (G/C or G/G). Patients with the G/C genotype were all
Fig. 1. EULAR response to rituximab according to TGF1 Codon 10, TGF1 Codon 25 and the association of the two SNPs. Black bars represent responders (R); white bars represent nonresponders (NR).
473
responsive to treatment. There were more females in the G/G group. There were no other statistically significant differences. Only 60% of G/C expressing patients were positive for ACPA, compared with 82% of G/G expressing patients. 3.4. TGFˇ1 SNPs predict response to rituximab but not to anti-TNF˛ To investigate the specificity of TGF1 SNPs in predicting the response to rituximab, we similarly studied the association of TGF1 Codon 25 with the degree of responsiveness to anti-TNF␣ (etanercept, adalimumab and infliximab) treatment in 56 patients (not shown). The baseline characteristics within this group were similar to our population. The mean age was 54 years, 83% were female, 68% were positive for RF and 73% were positive for ACPA. The mean RA duration was 12.4 years and 75% were responsive to treatment. The only statistically significant difference between the baseline characteristics of this group and our own population was the average number of previous biologic therapies, since anti-TNF␣ was a first-line therapy in all patients, whereas patients treated with rituximab received an average of 1.6 previous biologic treatments. There was no correlation between the TGF Codon 25 SNP genotype and response to anti-TNF␣ treatment (four responders had the G/C genotype and 37 responders the G/G genotype; one nonresponder had the G/C genotype and 14 the G/G genotype; P = 1.0). 4. Discussion Many biologic treatments have been proposed for patients with severe RA, refractory to methotrexate but, to date, no biomarkers have been identified, which have high predictive values of clinical response. This study aimed to assess the predictive response to rituximab in RA patients by analyzing a panel of nine candidate genes (13 SNPs). Using a noninvasive technique, requiring less than four drops of blood from each patient, we found a significant association between EULAR response to rituximab in RA patients and TGF1 Codon 25 and Codon 10 polymorphisms. These SNPs are involved in TGF1 production, which is known to inhibit B cell activation and immunoglobulin production. The RA population we studied corresponded to that which can be seen in daily rheumatology practice [21], with 69.8% of patients responding to rituximab treatment, a mean patient age of 55 years and a mean disease duration of 12 years. The presence of RF and ACPA were detected more frequently in responders, when compared to nonresponders, but the difference was not significant (RF: 77.3% vs. 63.2% respectively; P = 0.25; ACPA: 79.5% vs. 73.7% respectively P = 0.66). The frequency of C/C, T/T and C/T TGF1 Codon 10 genotype expressions in RA sufferers corresponded to those found in healthy controls (C/C: 21% vs 19% respectively; T/T: 30% vs 29% respectively; C/T: 49% vs 52% respectively) [22]. The frequency of C/C, G/C and G/G TGF1 Codon 25 genotypes in RA sufferers also corresponded to those found in literature for Caucasian controls (C/C: 0 vs. 0% respectively, G/C: 16.9 vs. 15.2% respectively and G/G: 83.1 vs. 84.8% respectively) [23]. Our candidate gene study led to the identification of TGF1 Codon 10 and TGF1 Codon 25 as being potential predictive factors of rituximab response (likelihood of responding to treatment with the Codon 10 C/T genotype: OR = 1.6; CI 95% 1.2–2.3; P = 0.002; likelihood of responding to treatment with the Codon 25 G/C genotype: OR = 1.6; CI 95% 1.3–1.9; P = 0.025). TGF1 SNPs seemed to be specifically associated to the response to rituximab treatment, as we did not find any similar association with a response to antiTNF␣ treatment. No association between TGF1 Codon 25 or TGF1 Codon 10 and ACPA or RF was found. Therefore, TGF1 Codon
474
C.I. Daïen et al. / Joint Bone Spine 79 (2012) 471–475
Table 2 Genotypes of the RA patients according to their EULAR response to rituximab treatment. SNP
Responders
Nonresponders
P value
TGF1-10 (rs1800470) TGF1-25 (rs1800471) LTA 720 (rs1041981) LTA 249 (rs909253) TNFa 308 (rs 1800629) PTPN22 (rs2476601) IL10 1087 (rs1800896) TNFa 488 (r80267959) TNFa 857 (rs1799724) STAT4 (rs7574865) TNFAIP3 (rs6920220) TRAF1 C5 (rs1081848) TNFRII 196 (rs1061622)
T/C = 27, T/T = 8, C/C = 9 G/C = 11, G/G = 33 A/A = 2, C/A = 21, C/C = 19 C/C = 2, T/C = 23, T/T = 19 G/A = 14, G/G = 28 C/C = 33, C/T = 8, T/T = 1 A/A = 11, A/G = 23, G/G = 6 A/A = 0, G/A = 12, G/G = 31 C/C = 28, C/T = 15, T/T = 0 G/G = 18, T/G = 21, T/T = 4 A/A = 1, G/A = 17, G/G = 24 A/A = 7, G/A = 26, G/G = 10 G/G = 4, T/G = 13, T/T = 24
T/C = 4, T/T = 11, C/C = 4 G/C = 0, G/G = 19 A/A = 0, C/A = 7, C/C = 12 C/C = 1, T/C = 6, T/T = 12 G/A = 3, G/G = 16 C/C = 17, C/T = 2, T/T = 0 A/A = 7, A/G = 7, G/G = 5 A/A = 1, G/A = 4, G/G = 14 C/C = 13, C/T = 5, T/T = 1 G/G = 8, T/G = 8, T/T = 3 A/A = 1, G/A = 8, G/G = 10 A/A = 4, G/A = 8, G/G = 7 G/G = 0, T/G = 6, T/T = 13
0.002* 0.025* 0.075 0.084 0.332 0.349 0.417 0.449 0.644 0.678 0.716 0.732 0.771
*
Significant association (P < 0.05).
Table 3 Patient characteristics according to TGF1 Codon 25 and Codon 10 genotypes. TGF1 alleles
Codon 10 C/T
Codon 10 T/T
Codon 10 C/C
Codon 25 G/C
Codon 25 G/G
Number Age Female Responders + RF + ACPA Erosion RA duration Number of previous biologic therapies Initial DAS28
31 55.5 (20–74) 21 (67.7) 27 (86.7) 24 (77.4) 27 (86.7) 27 (86.7) 14.0 (1–57) 2 (0–4) 5.41 (3.8–7.2)
19 57.5 (43–76) 18 (94.7) 8 (42) 11 (57.9) 13 (68.4) 15 (78.9) 14.5 (5–38) 2 (0–3) 5.61 (2.6–7.5)
13 53 (29–71) 9 (69.2) 9 (69.2) 11 (84.6) 10 (76.9) 10 (76.9) 12 (3–40) 2 (0–3) 5.45 (3.5–6.9)
11 56.5 (45–71) 5 (50) 10 (100) 8 (80) 6 (60) 10 (100) 14 (6–57) 2 (0–3) 5.12 (3.8–6.5)
52 53 (20–76) 48 (98) 30 (61) 34 (69) 40 (82) 39 (80) 12 (1–38) 2 (0–4) 5.34 (2.6–7.5)
Both quantitative values (range) and qualitative values (percentage) are represented. Mean values are given for age at RA diagnosis and initial DAS28 and median values for age, RA duration and number of previous therapies.
25 and TGF1 Codon 10 genotypes appear to be novel predictive factors and therefore could help to propose a targeted therapy to RA patients [24]. Both SNPs belong to a signal peptide sequence of TGF1. TGF1 Codon 25 polymorphism exchanges an arginine for a proline (p.Arg25Pro) and TGF1 Codon 10 exchanges a proline for a leucine (p.Pro10Leu). As the two SNPs are known to be in linkage disequilibrium [25], they could be similarly involved in the response to rituximab. However, the association of these two SNPs predicted better response to rituximab (likelihood of responding to treatment with both TGF Codon 25 G/C and TGF Codon 10 C/T genotypes when compared with Codon 25 G/G and Codon 10 T/T genotypes: OR = 2.6; CI 95% 1.4–4.6; P = 0.008), suggesting an additive effect. One could expect that patients with TGF Codon 10 C/C would response more than those with Codon 10 C/T but this was not the case. However, this subgroup was poorly represented (13 patients). Awad et al. [26] and Guo et al. [27] found that in vitro lymphocyte production of TGF1 was lower in Codon 10 T/C expressing Caucasian controls than in T/T genotype controls and in Codon 25G/C expressing than in G/G genotype subjects. Yokota et al. [28] and Yamada et al. [29] showed the opposite result but this work concerned Japanese patients. TGF1 is known to inhibit B cells and to decrease immunoglobulin secretion [30,31]. Thus, a genotype associated with a lower TGF response could be associated with higher IgG and antibody secretion. Elevated IgG levels have been associated with a good response to rituximab [8]. Therefore, if responders to rituximab in this study produce less TGF1, and thus more immunoglobulins, this may explain the enhanced treatment response observed here. However, this needs to be confirmed on a large prospective cohort of patients (Fig. 2). Using a candidate SNP approach, we found a significant association between EULAR response to rituximab in RA patients and TGF1 Codon 25 and Codon 10 polymorphisms. These SNPs are
Fig. 2. Possible explanation of the TGF1 SNP’s influence on rituximab (RTX) response. TGF1 SNPs seem to control the production of TGF1, which influences B cell maturation, activation and immunoglobulin production. Patients with TGF1 Codon 25 G/C allele + Codon 10 C/T allele would have more active B cells, which could explain why RTX would be more efficient than in patients with more quiescent B cells (25 G/G + 10 T/T).
involved in TGF1 production, which is known to inhibit B cell activation and immunoglobulin production. Our results need to be confirmed on a large prospective cohort. If confirmed, TGF1 polymorphisms could become useful predictive markers and guide appropriate prescription of rituximab to RA patients. Disclosure of interest C.I.D, C.R., S.S., V.D., G.T., D.C., N.M., J.-P.D. and I.T. declare that they have no conflicts of interest concerning this article.
C.I. Daïen et al. / Joint Bone Spine 79 (2012) 471–475
S.F. received honorarium from Roche, Neovacs, Novartis and Merk Serono laboratories for consulting, investigator of clinical trials. C.J. participated as principal investigator in clinical trials, but did not receive honorarium. Acknowledgments To Montpellier hospital (France) for promoting this study and to Jean Sibilia (Strasbourg, France) for his advices. References [1] Edwards JC, Szczepanski L, Szechinski J, et al. Efficacy of B-cell-targeted therapy with rituximab in patients with rheumatoid arthritis. N Engl J Med 2004;350:2572–81. [2] Emery P, Fleischmann R, Filipowicz-Sosnowska A, et al. The efficacy and safety of rituximab in patients with active rheumatoid arthritis despite methotrexate treatment: results of a phase IIB randomized, double-blind, placebo-controlled, dose-ranging trial. Arthritis Rheum 2006;54:1390–400. [3] Cohen SB, Emery P, Greenwald MW, et al. Rituximab for rheumatoid arthritis refractory to anti-tumor necrosis factor therapy: results of a multicenter, randomized, double-blind, placebo-controlled, phase III trial evaluating primary efficacy and safety at twenty-four weeks. Arthritis Rheum 2006;54:2793–806. [4] Duddy ME, Alter A, Bar-Or A. Distinct profiles of human B cell effector cytokines: a role in immune regulation? J Immunol 2004;172:3422–7. [5] Roosnek E, Lanzavecchia A. Efficient and selective presentation of antigenantibody complexes by rheumatoid factor B cells. J Exp Med 1991;173:487–9. [6] Takemura S, Klimiuk PA, Braun A, et al. T cell activation in rheumatoid synovium is B cell dependent. J Immunol 2001;167:4710–8. [7] Narvaez J, Hernandez MV, Ruiz JM, et al. Rituximab therapy for AA-amyloidosis secondary to rheumatoid arthritis. Joint Bone Spine 2011;78:101–3. [8] Sellam J, Hendel-Chavez H, Rouanet S, et al. B-cell activation biomarkers as predictive factors of the response to rituximab in rheumatoid arthritis. Arthritis Rheum 2011;63:933–8. [9] Guis S, Balandraud N, Bouvenot J, et al. Influence of -308 A/G polymorphism in the tumor necrosis factor alpha gene on etanercept treatment in rheumatoid arthritis. Arthritis Rheum 2007;57:1426–30. [10] Quartuccio L, Fabris M, Salvin S, et al. Rheumatoid factor positivity rather than anti-CCP positivity, a lower disability and a lower number of anti-TNF agents failed are associated with response to rituximab in rheumatoid arthritis. Rheumatology (Oxford) 2009;48:1557–9. [11] Teng YK, Levarht EW, Hashemi M, et al. Immunohistochemical analysis as a means to predict responsiveness to rituximab treatment. Arthritis Rheum 2007;56:3909–18. [12] Thurlings RM, Vos K, Wijbrandts CA, et al. Synovial tissue response to rituximab: mechanism of action and identification of biomarkers of response. Ann Rheum Dis 2008;67:917–25. [13] Kavanaugh A, Rosengren S, Lee SJ, et al. Assessment of rituximab’s immunomodulatory synovial effects (ARISE trial). 1: clinical and synovial biomarker results. Ann Rheum Dis 2008;67:402–8.
475
[14] Roll P, Dorner T, Tony HP. Anti-CD20 therapy in patients with rheumatoid arthritis: predictors of response and B cell subset regeneration after repeated treatment. Arthritis Rheum 2008;58:1566–75. [15] de Vries N, Tak PP. The response to anti-TNF-alpha treatment: gene regulation at the bedside. Rheumatology 2005;44:705–7. [16] Lee YH, Rho YH, Choi SJ, et al. Association of TNF-alpha -308 G/A polymorphism with responsiveness to TNF-alpha-blockers in rheumatoid arthritis: a metaanalysis. Rheumatol Int 2006;27:157–61. [17] Mugnier B, Balandraud N, Darque A, et al. Polymorphism at position -308 of the tumor necrosis factor alpha gene influences outcome of infliximab therapy in rheumatoid arthritis. Arthritis Rheum 2003;48:1849–52. [18] Coenen MJ, Gregersen PK. Rheumatoid arthritis: a view of the current genetic landscape. Genes Immun 2009;10:101–11. [19] Boissier MC. Cell and cytokine imbalances in rheumatoid synovitis. Joint Bone Spine 2011;78:230–4. [20] Arnett FC, Edworthy SM, Bloch DA, et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988;31:315–24. [21] Assous N, Gossec L, Dieude P, et al. Rituximab therapy in rheumatoid arthritis in daily practice. J Rheumatol 2008;35:31–4. [22] Pokorny V, Chau J, Wu L, et al. Transforming growth factor beta 1 gene (HSTGFB1) nucleotide T869C (codon 10) polymorphism is not associated with prevalence or severity of rheumatoid arthritis in a Caucasian population. Ann Rheum Dis 2003;62:907–8. [23] Oen K, Malleson PN, Cabral DA, et al. Cytokine genotypes correlate with pain and radiologically defined joint damage in patients with juvenile rheumatoid arthritis. Rheumatology 2005;44:1115–21. [24] Prati C, Wendling D. Using targeted therapies alone for rheumatoid arthritis. Joint Bone Spine 2011;78:1–3. [25] Syrris P, Carter ND, Metcalfe JC, et al. Transforming growth factorbeta1 gene polymorphisms and coronary artery disease. Clin Sci 1998;95: 659–67. [26] Awad MR, El-Gamel A, Hasleton P, et al. Genotypic variation in the transforming growth factor-beta1 gene: association with transforming growth factor-beta1 production, fibrotic lung disease, and graft fibrosis after lung transplantation. Transplantation 1998;66:1014–20. [27] Guo Z, Binswanger U, Knoflach A. Role of codon 10 and codon 25 polymorphisms on TGF-beta 1 gene expression and protein synthesis in stable renal allograft recipients. Transplant Proc 2002;34:2904–6. [28] Yokota M, Ichihara S, Lin TL, et al. Association of a T29C polymorphism of the transforming growth factor-beta1 gene with genetic susceptibility to myocardial infarction in Japanese. Circulation 2000;101:2783–7. [29] Yamada Y, Miyauchi A, Goto J, et al. Association of a polymorphism of the transforming growth factor-beta1 gene with genetic susceptibility to osteoporosis in postmenopausal Japanese women. J Bone Miner Res 1998;13: 1569–76. [30] Lebman DA, Edmiston JS. The role of TGF-beta in growth, differentiation, and maturation of B lymphocytes. Microbes Infect 1999;1: 1297–304. [31] Kehrl JH, Thevenin C, Rieckmann P, et al. Transforming growth factor-beta suppresses human B lymphocyte Ig production by inhibiting synthesis and the switch from the membrane form to the secreted form of Ig mRNA. J Immunol 1991;146:4016–23.