Article Text
PDF
Abstract
Objective The polymorphism 158V/F of Fc fragment of IgG (FCGR) type 3A may influence the response to rituximab (RTX) in rheumatoid arthritis (RA). We investigated the FCG3A polymorphism in a large cohort of RA patients treated with RTX, also by considering the possible loss of response from month +4 to +6 after RTX and the presence of established predictors of response.
Methods The study analysed 212 RA patients. European League Against Rheumatism (EULAR) response was evaluated at months +4 and +6 after the first RTX infusion. The FCGR3A polymorphism was analysed by PCR followed by Sanger sequencing.
Results The FCGR3A genotypes were associated with EULAR response (good or moderate) at month +6 (response in 34/38 (89.5%) VV vs 70/106 (66%) VF and in 51/77 (66.2%) FF patients; p=0.01), but not at month +4 (response in 32/37 (86.5%) VV vs 69/102 (67.6%) VF and 53/73 (72.6%) FF patients; p=0.09). Loss of response was observed only in VF and FF carriers ((VV vs VF vs FF: 0/37 (0%) vs 11/102 (10.8%) vs 12/73 (16.4%); p=0.02)).
Probability of response at month +6 was very high when at least two of the three following items selected by multivariate analysis were present: positive rheumatoid factor and/or anticyclic citrullinated peptide antibodies, previous treatment with ≤1 anti-tumor necrosis factor (TNF) agent, and 158VV FCGR3A genotype (p<0.0001; OR 7.9, 95% CI 4.1 to 15.1).
Conclusions The 158VV FCGR3A genotype was associated with response to RTX in a large cohort of RA patients. Patient genotyping may be helpful to plan RTX treatment, and may be integrated with clinical predictors.
- Rheumatoid Arthritis
- Rheumatoid Factor
- Pharmacogenetics
Statistics from Altmetric.com
Introduction
Rituximab (RTX) (MabThera, Rituxan) is a chimeric monoclonal antibody directed against CD20, an antigen expressed by most B cells.1 It is approved as a second line after anti-TNF failure for the treatment of rheumatoid arthritis (RA).2–6 Several factors were suggested to predict the response to RTX, including factors related to target antigen and patient-related factors, notably the genetic background.7–15
Receptors for IgG allow antibody-dependent cell cytotoxicity, and the polymorphism 158V/F (rs396991) of the Fc fragment of IgG type 3A (FCG3A) was investigated as a genetic factor influencing disease response after RTX therapy in haematologic and rheumatic diseases.16–21 In the majority of cases, the 158 V allele, that confers an increased binding affinity for IgG,22 was significantly associated with a positive response to RTX and/or to a reduced dose dependence.
Response to RTX in RA usually occurs from month +4 to +6 after treatment, and the degree of response may vary, being either good or moderate. The response noticed at month +6 is already present at month +4 in most patients, but while some patients show progressive improvement (ie, the degree of response is higher at month +6 than at month +4), others lose response from month +4 to month +6. Thus, the evaluation of a putative predictor of response depends on how response is evaluated, distinguishing both the degree of response and the time to response. At present, some preliminary data, including earlier data from our group23 and a recent paper from French authors,15 identify 158V/F FCGR3A polymorphism as a possible predictor of response to RTX in RA. The aim of this study was to better evaluate of the association between 158V/F FCGR3A polymorphism and response to RTX, by analysing a larger series of patients and by dissecting the results at month +4 and at month +6 after treatment. The pharmacogenetic data were also integrated with established clinical predictors of response to RTX in RA.
Methods
Patients
A cohort of 212 unselected patients with longstanding RA diagnosed according to the 1987 American College of Rheumatology classification criteria,24 was retrospectively evaluated. All patients gave their informed consent to the genetic study according to the Declaration of Helsinki, and the investigation was approved by the local Study Review Board (Far-Reu 001, 06/20/2010). Patients were referred to 13 different rheumatologic centres in Italy. They were unselected and represented the majority of the RA subjects treated with RTX in any single participating centre. Patients were followed for at least 6 months after RTX treatment, from 2004 to 2011, and data were collected after the first cycle of RTX. Patients were taking no glucocorticoids (GC) or <10 mg/day of prednisone equivalents at baseline, and cases where GC were introduced or increased after the beginning of RTX therapy were excluded. Primary or secondary inefficacy, rather than development of side effects, was the major reason for anti-TNF failure. The European League Against Rheumatism (EULAR) response criteria were used to evaluate response rate at months +4 and +6 after the first cycle of RTX, by means of DAS28 using erythrocyte sedimentation rate. Patients who showed a moderate or good EULAR response at month +4, but an absence of response at month +6, were classified as patients with a loss of response to RTX from month +4 to +6.
Molecular analyses
DNA was extracted from blood samples using the Maxwell16 device and a DNA extraction kit (cat: AS1010: Promega, Madison, Wisconsin, USA) according to the manufacturer's instructions. After the amplification of the DNA fragment referred specifically to the 158V/F FCGR3A polymorphism by PCR, genotyping was performed by direct Sanger sequencing, to allow the possible identification of copy number variation.25 According to previously reported methods,26 the following PCR primers were employed: forward 5′-CCCTTCACAAAGCTCTGCACT-3′ and reverse 5′-ATTCTGGAGGCTGGTGCTACA-3′. The amplification programme initiated with 2′ of denaturation at 94°C followed by 10 cycles: 20″ at 94°C, 40″ gradient from 62°C to 57.5°C, 0.5°C decrement at each cycle, 1′30″ at 72°C; then 29 cycles as following: 20″ at 94°C, 40″ at 57°C, 1′30″ at 72°C. Finally 5′ of elongation at 72°C. The PCR product was a fragment of 1.109 kB, that was checked by electrophoresis onto a 2% ethidium-stained agarose gel. Finally, the sequencing PCR was done using the following primer: 5′-CCC CAA AAG AAT GGA CTG AA-3′ and data analysed by Chromas Lite 2.01. The sequence surrounding the polymorphic locus was verified based on National Center for Biotechnology Information Reference Sequence: NM_001127596. A representative illustration of the three genotypes is reported in figure 1. In a percentage of heterozygote (VF) cases (22/102, 21.6%), an unequal heterozygote peak was observed, in particular, a V allele prevalence in 18 samples, and an F allele prevalence in four samples (data not shown). These patterns are possibly related to unequal allele ratios due to copy number variation.25
The panel illustrates the three representative sequencing results: FF homozygosis (the arrow indicates the presence of a single peak corresponding to base A, adenine, in the DNA sequence at the polymorphic site); the VV homozygosis (the arrow indicates the presence of a single peak corresponding to base C, citosine); and the VF heterozygosis (the peaks corresponding to bases A and C have the same height).
Statistics
Variables were reported as mean±SD or median (range), or as frequencies, as appropriate. Baseline demographic and clinical differences between the three genotypes were tested by one-way analysis of variance, with Bonferroni posthoc test if required. The associations between EULAR response at month +4, EULAR response at month +6 and response variation from month +4 and month +6 and FCGR3A (VV, VF, FF genotypes) was tested by χ² tests, as appropriate.
Univariate logistic regression analyses were performed to test the variables associated with EULAR response at month +6, by considering the following covariates: age, sex, presence of rheumatoid factor (RF) and/or anticyclic citrullinated peptides (anti-CCP) antibodies, baseline DAS28, baseline health assessment questionnaire (HAQ), number of TNF blockers previously failed. Selected variables were then tested in a multivariate logistic regression analyses.
Data were analysed with SPSS software V.13.1. Results were considered statistically significant when p<0.05.
Results
Demographic data
The patients consisted of 181 women and 31 men, with mean age 60±13 years, and mean disease duration of 13±10 years. All the RA patients were active at the study entry, showing a mean activity score on 28 joints (DAS28) of 5.9±1.2, and a mean HAQ score of 1.6±0.7. Serologically, patients were 152/212 (71.7%) RF positive and 151/212 (71.2%) anti-CCP positive. All the patients were treated with RTX at the standard dose for RA, in combination with methotrexate or other Disease Modifying Anti-Rheumatic Drugs (DMARDs) (leflunomide, cyclosporin A or hydroxychloroquine) in the majority of the patients (91.5%). The majority of patients (153/212, 72.2%) had been previously treated with one or more anti-TNF agents, while the remaining cases had been unresponsive to methotrexate alone or in combination with other DMARDs for at least 6 months. The baseline characteristics of the patients, according to the FCGR3A genotype, were reported in table 1. All the patients showed a high disease activity, though a slight difference in the baseline DAS28 score between VV, VF, and FF patients was noticed, with statistical significance (p=0.01) only when comparing the VV with the FF cases.
Baseline characteristics of the patients according to the FCGR3A genotype
Association between EULAR response and FCGR3A genotype
EULAR response was observed in 154/212 (72.6%) patients at month +4, and in 146/212 (68.9%) patients at month +6 (tables 1 and 2). The distribution of the EULAR type of responses (good/moderate/no response) among the three FCGR3A genotypes is reported in table 2.
EULAR responses according to FCGR3A genotype at month +4 and at month +6
After RTX therapy, CD19 B cells resulted depleted between month +1 and +3 in all the cases, as assessed by routine PCR analyses. FCGR3A genotypes were not associated with EULAR good/moderate response at month +4 (p=0.09), while they were significantly associated with EULAR good/moderate response at month +6 (p=0.01) (table 2), with a significant difference between VV and VF genotypes (p=0.015), VV and FF genotypes (p=0.018), but not between VF and FF genotypes (p=0.96). The presence of the VV genotype was confirmed to be associated with EULAR good/moderate response at month +6 by univariate logistic regression analysis (p=0.01, OR 4.4, 95% CI 1.4 to 13.5).
Unequal heterozygosis, with apparent V or F allele prevalence (as assessed by the observation of the electropherograms), was observed in a small number of patients (n=18; 8.5%), as expected.25 The percentage of NR cases in VF cases with allele V prevalence (33.3% at month +4, 38.9% at month +6) did not differ from the equal VF patients (32.5% at month +4, 33.8% at month +6). Thus, unequal VF cases remained included in the overall VF series. On the other hand, the number of unequal VF samples with apparent F allele prevalence was very low (n=4) and did not allow us to assess whether extra F alleles might confer a different responsiveness to RTX.
Association between the loss of response to RTX between month +4 and month +6 and the FCGR3A genotype
As regards the response variations from month +4 to month +6, 174/212 (82.1%) patients showed the same type of response, 15/212 (7.1%) patients became responders only at +6 while unresponsive at +4, and 23/212 (10.8%) showed a loss of response from +4 to +6 (table 3). Patients carrying the VV genotype showed a significantly lower frequency of loss of response from month +4 to month +6 than patients carrying the other two genotypes ((VV vs VF vs FF: 0/37 (0%) vs 11/102 (10.8%) vs 12/73 (16.4%); p=0.02)).
Response variation from month +4 to month +6 according to the FCG3A genotype
Evaluation of the response to RTX by the integration of clinical, serological and pharmacogenetic predictors
By univariate regression analyses, the following variables resulted associated with EULAR good/moderate response at month +6: lower baseline HAQ (OR 0.56, 95% CI 0.35 to 0.88; p=0.01), ≤1 anti-TNF previously failed (OR 3.72, 95% CI 2.02 to 6.86; p<0.0001), the presence of RF and/or anti-CCP antibodies (OR 4.89, 95% CI 2.39 to 10.38; p<0.0001) and FCG3A VV genotype (OR 4.53, 95% CI 1.53 to 13.36; p=0.006) (table 4).
Univariate regression analysis considering EULAR good/moderate response at month +6 as dependent variable
By multivariate regression analyses, the presence of the VV genotype (p=0.004; OR 9,43 95% CI 2.08 to 42.7), the presence of RF and/or anti-CCP antibodies (p=0.02; OR 3.01, 95% CI 1.16 to 7.79), and the failure of ≤1 anti-TNF agent (p=0.0003; OR 3.66, 95% CI 1.79 to 7.47) were significantly associated with EULAR good/moderate response at month +6, while the baseline HAQ was excluded (p=0.10, OR 0.66 95% CI 0.39 to 1.08). Notably, patients having at least two of the three selected variables (FCG3A VV genotype, seropositivity, ≤1 previous anti-TNF drug) showed an eight times higher probability of EULAR good/moderate response at month +6 (p<0.0001; OR 7.9, 95% CI 4.1 to 15.1).
Finally, the prevalence of the FCGR3A genotypes in RA patients was similar to a previously reported series of RA and Caucasian controls.25
Discussion
RTX is an effective therapy for RA, but response may vary between patients, as in the case of other biologic agents. The identification of biological markers helpful to predict a positive (or negative) response to treatment is a relevant issue in RA.27
The possible usefulness of the FCGR3A 158V/F genotyping to better manage RTX therapy was assessed in several diseases, mainly in haematological malignancies, reporting in general a significant association between higher rate of response and the VV homozygosis.15–21 ,23 In preliminary studies by our group,23 a better response to RTX was noticed in RA patients carrying the VV genotype, if compared with that observed in VF/FF carriers. The present study extended the evaluation of the association between 158V/F FCGR3A polymorphism and response to RTX, by analysing a larger series of patients, and by dissecting the results both at month +4 and at month +6 after treatment. The results highlight that genetic study of the FCGR3A polymorphism may represent a useful tool to optimise the choice of RTX in RA, also when integrated with other already established clinical predictors.
A significant association between the FCGR3A 158V allele carriage (ie, VV plus VF genotypes) and the response to RTX in RA has been recently reported by Ruyssen–Witrand, assessed at 24 weeks after RTX in a smaller cohort of RA patients unresponsive or intolerant to anti-TNF agents.15 By contrast, in the present series, no difference was found between VF and FF patients, and only the VV homozygosis was associated with response to RTX, in agreement with the published literature.16–21 Notably, Ruyssen–Witrand et al were not able to demonstrate a significant association between a variable widely accepted to be associated with response to RTX in RA, that is, RF-positive status, and response to RTX.15 As also suggested by the authors, a possible selection bias might have been responsible for these contrasting findings, underlying the importance of data replication. Furthermore, technical issues may be hypothesised. Evaluation of unequal VF samples is not reported in the French study,15 while unequal VF with prevalent V alleles may be falsely genotyped as VV, thus underestimating the response rate of VV patients. We estimated the possible presence of copy number variation, resulting in unequal heterozygosis with V allele prevalence, but it did not seem to confer a higher responsiveness to RTX if compared with equal VF patients. It is possible that the FCGR3A 158V/F genetic variants might predispose to a different responsiveness to RTX in RA populations belonging to different geographical areas, that is, Italy and France, and therefore, additional studies in other cohorts are needed. However, as concerns the response to RTX in RA, we recently observed similar pharmacogenetic data in Italian and British patients.28
The association between the FCGR3A 158V/F genetic variant genotype and the time to response to RTX was analysed in detail in this study. While all the patients with the VV genotype who responded at +4 also responded at +6, 10.8% of the VF and 16.4% of the FF patients showed a loss of response from month +4 to month +6. Then, the association between the VV genotype and response to RTX was higher at month +6 than at month +4. Whether patients who respond at +4, but not at +6, deserve either a different RTX treatment schedule or a different biologic treatment remains to be addressed, and the analysis of the FCGR3A 158V/F polymorphism might prove of value to this end.
The disease activity was slightly higher in RA patients carrying the VV genotype in this study, who also showed a better response to RTX. However, all the three groups of VV, VF and FF patients showed a very active disease at baseline. Furthermore, the DAS28 was not associated with response to RTX by multivariate analyses, supporting an independent role of the FCGR3A genotype to predict response to RTX.
Notably, while the FCG3A polymorphism was helpful in predicting a response to RTX in general, it did not allow us to discriminate between good and moderate EULAR response, even if a trend for a higher degree of response in the VV group was observed. On the other hand, since the degree of response may increase after repeated RTX cycles in RA,29 a thorough clinical evaluation of the usefulness of RTX therapy in the particular patient, according to a treatment to target strategy, is probably premature if based only on the evidence of a moderate versus good response at month +6 after RTX. A longer follow-up, including at least one RTX retreatment cycle, might be more informative for further decisions. Future studies will address this issue.
Thabet et al25 associated the 158VV homozygosis to an increased risk for anti-CCP-positive RA. In the present series, however, no difference in the prevalence of the 158VV genotype was found between RF-positive and negative or between anti-CCP-positive and negative patients. This allowed us to exclude that the association between RF (or anti-CCP) positive RA and the 158VV genotype could determine the association between the 158VV genotype and the response to RTX.
Finally, we investigated the predictive power for response to RTX therapy in RA based on the FCGR3A genotype results integrated with the two most recognised clinical predictors of response to RTX in RA,30 ,31 that is, seropositivity, and the lower exposure to previous anti-TNF drugs, as also herein selected by multivariate analysis. Although not representing the primary objective of the study, this analysis was performed to evaluate, in a real-life RA case series, the potential of integrating multiple predictors of response to develop a treatment strategy in RA. The results were encouraging, when also considering that other factors implicated for the degree of response to RTX in RA were not evaluated, for example, immunoglobulin levels, B-cell depletion as detected by sensitive analyses, and B-cell phenotype at baseline and after reconstitution.32 The multivariate statistic model showed an increased predictive capacity for response to RTX by combining the presence of at least two of the following three predictors: the FCGR3A 158VV genotype, seropositivity and treatment with ≤1 previous anti-TNF agent. This combination led to an eightfold probability of response to RTX. The pharmacoeconomic advantages of combining clinical, serological and genetic markers of response to RTX may be hypothesised.33
In conclusion, the 158VV FCGR3A genotype was associated with response to RTX in a large Italian cohort of patients with RA. Patient genotyping may be helpful to plan RTX treatment, and may be usefully integrated with other well established clinical predictors. On the basis of the present and previous retrospective data, prospective studies are needed.
Acknowledgments
We thank Prof Lisamaria Bambara and Prof Maurizio Cutolo for their scientific support; Dr Paola Masolini and Dr Cinzia Fabro for their technical contribution.
References
Footnotes
-
Handling editor Tore K Kvien
-
Contributors LQ and MF equally collaborated in this work. LQ, MF, SDV: conception and design, analysis and interpretation of data; drafting the article and final approval of the version to be published. EP, AZ, SS, MB, MM, DB, VR, FA, PS-P, PM, FF, PT, LB, MS, RP, MC, FS, VC, AS, RC, LC, RB, GP, MG, SB: analysis and interpretation of data; revising paper critically for important intellectual content and final approval of the version to be published.
-
Funding This work was supported by Regione Friuli-Venezia Giulia, Italy (grant number LR26.05.07/art.23).
-
Competing interests Dr De Vita received honoraria from Bristol-Myers Squibb, Glaxo, Roche, and UCB (less than US$10 000 each). Dr Sarzi-Puttini received honoraria from Roche (less than US$10 000).
-
Ethics approval Local Study Review Board (Far-Reu 001, 06/20/2010).
-
Provenance and peer review Not commissioned; externally peer reviewed.