Presence of the full-length KIR2DS4 gene reduces the chance of rheumatoid arthritis patients to respond to methotrexate treatment

Rheumatoid arthritis (RA) is a chronic, relatively prevalent (about 1% of individuals in Caucasians) inflammatory disorder with a T cell-mediated autoimmune component [1]. Natural killer (NK) cells were also reported to contribute to RA [2]. These cells may act both directly, killing target cells, and indirectly, as either inducer or regulatory cells influencing innate and adaptive immunity (a “two-edged weapon”) [3].

The most polymorphic receptors of NK cells are encoded by killer cell immunoglobulin-like receptor (KIR) genes distributed differently in unrelated individuals. Since these receptors, depending on their structure, are either inhibitory or activating, their repertoire in a genotype affects activity of NK cells (and T cell subpopulations which also express KIRs, including T lymphocyte CD4 + CD28-KIR+) and, in consequence, susceptibility to different diseases including autoimmune disorders such as RA [4‐6]. KIR haplotypes (i.e., sets of KIR genes on one chromosome) have been divided into group A, containing mostly inhibitory genes, and group B, containing several activating genes in addition to inhibitory ones. Moreover, the whole KIR region was divided into centromeric (A or B) and telomeric (A or B) halves, depending on the KIR gene content [7].

Ligands of KIRs, in most cases where known, are human leukocyte antigen (HLA) class I molecules which are extremely polymorphic. KIR and HLA genes are located on different chromosomes (number 19 and 6, respectively); therefore they are independently inherited and an individual may have KIRs with no ligands and vice versa. This affects the maturation of NK cells and the susceptibility of an individual to disease [8, 9]. The most important ligands of KIRs are HLA-C molecules, which all fall into two groups: C1, with an asparagine residue in position 80, and C2, with a lysine residue in this position. KIR2DL1 and KIR2DS1 are receptors for the C2 epitope, whereas KIR2DL2 and KIR2DL3 bind C1 and also some C2 molecules. KIR2DS4 recognizes HLA-A*11 molecules as well as some C1 and C2. KIR2DL4 binds the “non-classical” class I molecule HLA-G. KIR3DL1 and possibly KIR3DS1 bind HLA-B with Bw4 epitope and those HLA-A alleles that possess the Bw4 motif. KIR3DL2 binds HLA-A*03,*11, and microbial CpG DNA [10]. In addition, both KIR3DL1 and KIR3DL2 bind HLA-B*27 homodimers [11, 12].

As KIR and KIR ligand genotypes affect the immune response, and seem to influence the response to RA treatment using anti-TNF-α agents [13], we wondered whether these genotypes might also influence the outcome of MTX therapy of RA. MTX as a foliate antagonist is used in therapy of malignant disorders, but also suppresses the immune response in patients and in low doses was introduced for the treatment of RA because of its presumed anti-proliferative, immunosuppressive and anti-inflammatory properties [14]. However, MTX is not sufficient in all RA patients, and approximately 50-60% of RA patients could be classified as MTX responders [15‐18]. It seems that the effectiveness of MTX treatment corresponds to the individual genetic background, particularly in genes encoding key molecules of methotrexate metabolism and toxicity [19, 20]. However, clinically reliable markers of methotrexate efficiency are still sought.

A clinical response to MTX therapy was obtained for 57.3% of RA patients. When the frequencies of the KIR genes were compared between good responders and poor responders of RA patient groups, we observed an effect of only one KIR gene, full-length KIR2DS4 (KIR2DS4f), on the response to MTX therapy. Namely, patients possessing this gene had a lower chance of responding (p = 0.0334, OR = 0.4344, 95% CI = 0.215, 0.987; Table 2). Additionally, patients with erosive disease (ED) and rheumatoid factor (RF) were less likely to respond to MTX, and those positive for KIR2DS4f had an even lower chance than those who were negative for KIR2DS4f but positive for both ED and RF (36.2% and 43.6% of good responders, respectively; Table 3). Similarly, RA patients negative for ED and RF also had a lower chance of responding to MTX therapy if they possessed KIR2DS4f (75% responding in comparison to 88% within KIR2DS4f-negative patients; Table 3). On the other hand, the presence of KIR2DS4f did not affect the formation of RF (p = 0.89), ED (p = 0.78), or other clinical parameters (data not shown).

The KIR2DS4f gene is located in the telomeric half of the KIR locus [7]. However, we did not observe any differences between good and poor MTX responders in the estimated haplotype frequencies for the telomeric KIR repertoire (p = 0.29; Figure 1) despite the association of the KIR2DS4f gene with a poor response to MTX described above. Also, there were no differences in frequencies of centromeric (p = 0.71) or whole KIR locus (p = 0.96) genotypes between good and poor responders.

The effect of the KIR2DS4f gene on the probability of a response to MTX in RF- and ED-positive patients was strongest in individuals with medium values of the erythrocyte sedimentation rate (ESR), that is, log(ESR) about 3.0-3.5, i.e. 20-33 mm/h (Figure 2). Patients with high ESR values had a low probability of responding, and the presence or absence of KIR2DS4f did not affect their outcome, particularly in individuals with low age at onset. Patients with low ESR had a high probability of responding, which was not affected by the presence of KIR2DS4f, particularly in people with high age at onset (Figure 2). The KIR2DS4f effect did not depend on the presence of C1 (p = 0.209) or C2 (p = 0.484). Surprisingly, the age at diagnosis had a negligible influence on the KIR2DS4f effect.

This report confirms that almost one third of patients affected by rheumatoid arthritis do not respond to low-dose MTX treatment [15‐18]. This phenomenon causes that the discovery of (bio)markers of response to the drug is still required for clinical practice. Here, we found a negative association of the KIR2DS4f gene with the response of RA patients to MTX treatment. This effect was visible not only in RF-positive, ED-positive patients who already had a lower chance of responding than negative ones, but also in RF- and ED-negative individuals who otherwise were better responders. Therefore, it is likely that the KIR2DS4f gene is expressed on effectors or positive regulators or inducers of autoimmune response, relatively resistant to the MTX effect. In this context, the CD4+CD28- subpopulation of T lymphocytes should be discussed. This T cell subset is rare in healthy individuals (about 1% of peripheral lymphocytes), but its expansion was observed in patients affected by, e.g., coronary artery disease [26], polycystic ovary syndrome [27], multiple sclerosis [28] and rheumatoid arthritis [4, 29, 30]. In contrast to those isolated from healthy volunteers [31], CD4+CD28- T cells from rheumatoid arthritis patients may exhibit autoreactive properties [4]. Instead of the CD28 molecule, these T cells frequently express activating KIR receptors, as was shown by Yen et al.[32] particularly for KIR2DS2 in RA vasculitis. Interestingly, it was suggested that the KIR2DS4 gene may also be expressed on CD4+CD28- T lymphocytes [33], and could transmit an activating signal to effector cells. Therefore, KIR2DS4f gene-positive patients might have a potential to produce CD4+CD28-KIR2DS4+ T cells contributing to more severe forms of RA, which would be more resistant to therapy, including MTX treatment.

The gene mentioned above encodes the KIR2DS4f polyspecific receptor recognizing both HLA-A*11 and some HLA-C allotypes bearing C1 or C2 epitopes [10]. In Caucasians, including Poles, its gene is present in roughly one third of KIR2DS4-positive individuals, the other two thirds bearing its defective alleles, KIR2DS4d (see http://​www.​allelefrequencie​s.​net database) [34], potentially encoding a soluble molecule which does not exhibit any activating properties [35]. However, in Far East Orientals, Japanese and Chinese, the proportions between KIR2DS4f and KIR2DS4d are reversed [33]. It would be interesting, then, to perform a similar analysis of RA patients’ response to MTX therapy in those populations, especially as the KIR2DS4 gene seems to impact on RA susceptibility in Taiwanese patients [36].

Some studies suggest that ESR level, mentioned as a biomarker for chronic systemic inflammation as well as a parameter of disease activity, correlated with responsiveness to MTX (reviewed by Romao et al.) [37]. In the present study, we see that the effect of KIR2DS4f is strongest in individuals with medium susceptibility to the therapeutic effect of MTX characterized by an intermediate level of ESR. Patients with high ESR are resistant to therapy, and lack of the KIR2DS4f gene cannot help them; those with low ESR respond so well that even the presence of the KIR2DS4f gene does not cause any harm.

This study shows that the presence of the full-length KIR2DS4 gene reduces the probability that patients affected by rheumatoid arthritis will respond to methotrexate therapy. This phenomenon was strongest in individuals with a medium level (20-33 mm/h) of the erythrocyte sedimentation rate. The patient cohort with lower values of ERS responded better to MTX, while patients with a higher ESR level responded worse to therapy, and the presence of KIR2DS4f did not affect their outcome. Therefore, our findings, if confirmed using an independent and larger group of patients, might suggest that KIR typing could help to predict the response to MTX therapy, especially for patients with moderate rheumatoid arthritis.