Field of the invention

The present invention belongs to the field of rheumatoid arthritis. More particularly the present invention provides biomarkers which can be used to predict the joint destruction progression rate of RA patients. In addition, the biomarkers can also be used as a theranostic for guiding the therapeutic decisions of a subgroup of RA patients. Introduction to the invention

Rheumatoid arthritis is a chronic disorder for which there is no known cure. Fortunately in the last few years, a shift in strategy toward the earlier institution of disease modifying drugs and the availability of new classes of medications have greatly improved the outcomes that can be expected by most patients. The current goal of treatment aims towards achieving the lowest possible level of arthritis disease activity and remission if possible, minimizing joint damage, and enhancing physical function and quality of life. Treatment options include medications, reduction of joint stress, physical and occupational therapy, and surgical intervention. There is still a need for the design of an optimal treatment scheme of RA patients and we aimed at identifying genetic polymorphisms and biomarkers which can aid in defining a more optimal treatment scheme for a subgroup of RA-patients. More than 600 genome-wide association studies (GWAs) have been published, generally with respect to disease in a case-control design. These studies provide an interrogation of the susceptibility loci of the genome. However, most therapeutic strategies are targeted to affect pathways of disease progression. Therefore, both from the hope of new drug discovery and personalized medicine, interrogation of the genome with disease progression as an outcome is attractive. Rheumatoid Arthritis (RA) is an auto-immune disorder that affects 1 % of the population and is associated with significant morbidity, disability and costs for society. Radiographic joint destruction reflects the cumulative burden of inflammation and is conceived as an objective measure of RA severity. Susceptibility studies revealed that patients with and without anti-citrullinated protein antibodies (ACPA) represent two subtypes of RA with differences in associating genetic variants as well as in disease outcome; patients with ACPA had more severe joint destruction. The variance in joint destruction between ACPA-positive RA-patients is however large and incompletely understood. The contribution of genetic variants to joint destruction is unknown and, apart from the MHC Class II region, no genetic severity factor is widely replicated. The notion that variation in joint destruction is partly genetically determined is supported by our finding that monozygotic twins have comparable levels of joint destruction in contrast to dizygotic twins [see Ref. 1 ]. We aimed to increase the understanding of the processes underlying the inter- individual differences in joint destruction in ACPA-positive RA. To this end we performed a two-staged GWAs on the progression of joint destruction in RA. ACPA was determined by anti- CCP2 antibodies. Fourteen of the top 20 strongest associated SNPs identified in a two-staged GWAS were mapped to the region of SPerm associated AntiGen16 (SPAG16). SPAG16 is important for sperm tail structure and fertility. Several other GWAs found associations between SNPs in SPAG16 and other quantitative traits such as height, weight, lipid levels and heart rate. Within the autoimmunity field, SPAG16 has been described in relation to multiple sclerosis (MS), see WO2008/125651 . The prior art is however silent on the association between SPAG16 and RA. These identified SNPs in the SPAG16 gene were associated with the rate of joint destruction. SPAG16 transcript was demonstrated in a cDNA expression library generated from RA synovial tissue and immunohistochemical staining for SPAG16 indicated expression of the protein in human synovial tissue from RA-patients. Based on the reported presence of antibodies to intracellular antigens in RA, it was hypothesized that an increased anti-SPAG16 antibody response might associate with severity of ACPA-positive RA. To this end a recombinant SPAG16 protein ELISA was performed. Whereas no significant difference in anti-SPAG16 antibody positivity between RA-patients and controls was observed, a strong association for anti-SPAG16 antibody-positivity with rate of joint destruction was detected. RA- patients with positive anti-SPAG16 antibody reactivity had 1 .53 times higher progression rates over 7-years than patients without this antibody response (95% CI 1 .26-1 .87; P=2.8x10 ^"5 ).

Figures

Figure 1 : Genome-wide association results in the NARAC. (a) Manhattan-plot showing the - log-ιο (P-value) of the trend-test. P-values are plotted against position on each chromosome, (b) Genomic structure (upper) and -log-i ₀ of SNPs in the GWAs around SPAG16 and LD map (lower), (c) Summary of SNPs associated with rate of joint destruction in the North-American and Dutch patients, β is the regression coefficient back-transformed from the log-scale. In the North-Americans, β is the coefficient of the estimated SHS-progression per year. In the Dutch, β is the coefficient of the slope of SHS per year.

Figure 2: Radiological joint destruction during 7-years of follow-up for RA-patients with and without anti-SPAG16 antibodies. Depicted are the measured Sharp-van der Heijde scores.

Figure 3: Measured radiological progression of best hit rs7607479 (a) and promotor SNP rs13013558 (b) in the North-American and the Dutch patients. For the North-American patients, only one X-ray at a certain time-point was available. Depicted is the progression rate, calculated by dividing the total Sharp-van der Heijde score (SHS) by the disease duration at time of the X-ray. The red line represents the mean estimated progression of the genotypic group. Patients heterozygous and homozygous for the minor allele had an 0.77 (95%CI 0.70- 0.85) and 0.59 (95%CI 0.49-0.72) higher progression rate per year compared to the common (P=4.55x 10 ^"7 ). For the Dutch patients, serial radiological measurements were available. Depicted are the mean measured SHS for the different genotypic groups. The progression rates over 7-years for patients heterozygous and homozygous for the minor alleles at rs13013558 were 0.81 (95%CI 0.68-0.96) and 0.65 (95%CI 0.46-0.92) times that of patients with two common alleles (P=2.2*10 ^"2 ). For rs13013558 this was 0.81 (95%CI 0.67-0.97) and 0.65 (95%CI 0.45-0.94) (P=1 .5* 10 ^"2 ).

Figure 4: Expression of SPAG16 in synovial tissue of a RA-patient. PCR for the presence of SPAG16 transcript variant 2a in a RA cDNA expression library. The predicted band of 1 ,092 base pairs (bp), corresponding to SPAG16 transcript variant 2a cDNA was visualized by agarose gel electrophoresis.

Figure 5: Comparison of presence of anti-SPAG16 antibody response between RA-patients and healthy controls and between genotypes within RA-patients. The graphs represent the mean arbitrary units (AU) of RA-patients and healthy controls (HC) measured in recombinant SPAG16 protein ELISA. Depicted are the mean levels (red line) and the cut-off for positivity defined using the standard curve (blue dotted line), (a) No significant difference in antibody- positivity between RA-patients and Healthy controls (HC) was observed (2% versus 1 % respectively, P=0.32). (b-c) Neither was there a statistical difference in antibody-positivity between genotypes observed (39%, 19% and 28%, P=0.09 for rs7607479 and 26%, 19% and 30%, P=0.17 for rs13013558).

Figure 6: Anti-SPAG16 reactivity in successive dilutions of plasma from a RA-patient. Each plate consisted of a successive dilution of plasma of the same RA-patient. Depicted is the standard curve of plate one on normal scale (a) and on the log-scale (b), demonstrating the start of the steeper part of the curve at the dilution of 200, which was used as cut-off point. Figure 7: Variance between AU values for all RA patients obtained at 2 independent ELISA experiments. Comparison of AU values for all tested RA patients obtained at 2 independent experiments performed on different days.

Detailed description of the invention

In the present invention we have identified a set of biomarkers which are associated with the SPAG16 gene and SPAG16 gene product; said biomarkers can be used for the detection of the joint destruction progression rate in rheumatoid arthritis patients (RA patients). As such the present invention presents a dual biomarker, i.e. one based on the presence of allelic variants in the SPAG16 gene and the other part based on the presence of auto-antibodies directed against the SPAG16 gene product. In particular a set of variant alleles were identified in the SPAG16 gene which when present in a rheumatoid arthritis patient correlates with an enhanced joint destruction progression rate in these patients. More particularly, the variant alleles in the SPAG16 gene are a set of single nucleotide polymorphisms (SNPs) which in combination with anti-citrullinated-protein antibodies, present in a rheumatoid arthritis patient, are a diagnostic prediction for an enhanced joint destruction progression rate in said patient. In addition, the present invention has found that the presence of auto-antibodies directed against the SPAG16 protein present in a patient sample in combination with the presence of anti- citrullinated-protein antibodies in the same patient sample is a diagnostic prediction for an enhanced joint destruction progression rate in said patient.

Accordingly the present invention provides in a first embodiment a method for predicting the joint destruction progression rate in a rheumatoid arthritis patient, said method comprising: a) determining in a rheumatoid arthritis patient sample the presence of anti-citrullinated- protein antibodies,

b) determining in said same patient sample the presence of auto-antibodies directed against the SPAG16 protein and/or determining from said same patient sample if one or more variant alleles of the SPAG16 gene are present,

c) wherein the combined presence of said auto-antibodies and anti-citrullinated-protein antibodies and/or the combined presence of one or more of said variant alleles and anti- citrullinated-protein antibodies predicts a higher joint destruction progression rate compared to patients without said auto-antibodies and with anti-citrullinated-protein antibodies and/or without said one or more variant alleles and with anti-citrullinated-protein antibodies.

The measurement for the presence of antibodies directed to citrullinated peptides or proteins is well known in the art. The clinical tests are usually denominated as the anti-CCP test, which measures the presence of antibodies - in a patient sample - directed to cyclic citrullinated peptides. Thus the testing for the presence or absence of anti-CCP autoantibodies is widely accepted as a tool for diagnosis of rheumatoid arthritis patients. A current state of the art review for the determination of anti-citrullinated protein antibodies (ACPAs) is described by van Schaardenburg D and Dijkmans BA (2009) Nat. Rev. Rheumatol. 5(1 1 ): 627-633.

In yet another embodiment the present invention provides a method for predicting the joint destruction progression rate in a rheumatoid arthritis patient possessing anti-citrullinated- protein-antibodies, said method comprising: a) determining from said patient sample the presence of auto-antibodies directed against the SPAG16 protein and/or determining from said patient sample if one or more variant alleles of the SPAG16 gene are present,

b) wherein the presence of said auto-antibodies and/or the presence of one or more of said variant alleles predicts a higher joint destruction progression rate compared to patients without said auto-antibodies and/or without said one or more variant alleles.

In yet another embodiment the biomarkers (i.e. the presence of SPAG16 protein antibodies in a patient sample and/or the presence of one or more alleles of the SPAG16 gene) can be used for the subtyping of rheumatoid arthritis patients. In particular this subtyping refers to a classification of rheumatoid arthritis patients which have a predicted joint destruction progression rate. Accordingly, the latter patients would benefit for the administration higher amounts (or doses schemes) of rheumatoid arthritis drugs, such as disease modifying antirheumatic drugs.

The measurement for the presence or absence of autoantibodies against SPAG16 can be determined by various methods described herein. The cut-off between the presence or absence is further described in the materials and methods section.

In yet another embodiment a method is provided for selecting a rheumatoid arthritis patient eligible for receiving the maximal tolerated dose of disease modifying anti-rheumatic drugs comprising predicting the joint destruction progression rate in said patient comprising the following steps: a) determining (or analyzing) in a patient sample the presence of anti-citrullinated-protein antibodies,

b) determining (or analyzing) in the same patient sample of step a) the presence of autoantibodies directed against the SPAG16 protein and/or the presence of one or more variant alleles of the SPAG16 gene,

c) wherein the combined presence of anti-citrullinated-protein antibodies and the presence of auto-antibodies directed against the SPAG16 protein and/or the combined presence of anti-citrullinated-protein antibodies and the presence of one or more variant alleles of the SPAG16 gene selects a rheumatoid arthritis patient eligible for receiving the maximal tolerated dose of disease modifying anti-rheumatic drugs.

In particular embodiments the use of the biomarkers is an "in vitro" use. The latter implies a diagnostic method with no direct interaction with the patient.

The term 'rheumatoid patient sample' is a body fluid derived from a patient suffering from rheumatoid arthritis. Such a patient sample includes blood, blood serum, blood plasma, saliva, urine, tears, bone marrow fluid, cerebrospinal fluid (CSF), synovial fluid (SF), lymphatic fluid, amniotic fluid, nipple aspiration fluid and the like. Preferred body fluids for analysis are those that are conveniently obtained from patients, particularly preferred body fluids include blood serum or blood plasma or synovial fluid (SF). In yet another embodiment the invention provides a method for evaluating the prognosis/disease severity of rheumatoid arthritis in a patient by use of the above described biomarkers.

Implications of the presence of the biomarkers of the invention for therapeutic treatment of rheumatoid arthritis: There are three general classes of drugs commonly used in the treatment of rheumatoid arthritis: non-steroidal anti-inflammatory agents (NSAIDs), corticosteroids, and disease modifying anti-rheumatic drugs (DMARDs). NSAIDs and corticosteroids have a short onset of action while DMARDs can take several weeks or months to demonstrate a clinical effect. DMARDs comprise methotrexate, sulfasalazine, leflunomide (Arava®), etanercept (Enbrel®), infliximab (Remicade®), adalimumab (Humira®), certolizumab pegol (Cimzia®), golimumab (Simponi®), abatacept (Orencia®), rituximab (Rituxan®), tocilizumab (Actemra®), anakinra (Kineret®), antimalarials (e.g. Plaquenil®). Other immunomodulators are occasionally used including azathioprine (Imuran) and cyclosporine. Because cartilage damage and bony erosions frequently occur within the first two years of disease, it is important for rheumatologists to be able to decide whether to use more aggressive DMARD agents early in the course of disease. The present biomarkers of the invention satisfy this need and helps in deciding to apply higher doses of DMARD agents or in other words maximal tolerated doses of DMARD agents for the treatment of RA patients having ACPA and the biomarkers of the invention in a body sample derived from said RA patients. While it is generally up to the choice of the practitioner to estimate the application of the maximum tolerated doses of DMARD agents when the biomarkers of the invention are identified in an RA patient, the following gives a short overview of the choice of DMARD agents and their documented maximal tolerated doses regimes.

Methotrexate is now considered the first-line DMARD agent for most patients with RA. It has a relatively rapid onset of action at therapeutic doses (6-8 weeks), good efficacy, favorable toxicity profile, ease of administration, and relatively low cost. Methotrexate is effective in reducing the signs and symptoms of RA, as well as slowing or halting radiographic damage. Dosing typically begins at 12.5-15 mg once per week. A dose escalation to 20 mg within the first three months is now fairly well accepted in clinical practice. Maximal tolerated dose is between 25-30 mg per week and can be applied by the practitioner when the biomarker is identified in an RA patient sample. These doses can be given orally or by subcutaneous

Hydroxychloroquine is an antimalarial drug which is relatively safe and well-tolerated agent for the treatment of rheumatoid arthritis. Chloroquine is another antimalarial agent that is also sometimes used. Hydroxychloroquine can be combined with methotrexate for additive benefits for signs and symptoms or as part of a regimen of "triple therapy" with methotrexate and sulfasalazine. Hydroxychloroquine (Plaquenil®) is the drug of choice among antimalarials. Chloroquine is not commonly used because of greater toxicity on the eye. The usual dose of Plaquenil is 400mg/day but 600mg/day is considered to be the maximal tolerated doses and can be administered to an RA patient having the biomarkers of the invention. It may be prescribed as a single daily dose or in divided doses twice per day.

Sulfasalazine (Azulfidine®) is an effective DMARD for the treatment of RA. Its effectiveness overall is somewhat less than that methotrexate, but it has been shown to reduce signs and symptoms and slow radiographic damage. It is also given in conjunction with methotrexate and hydroxychloroquine as part of a regimen of "triple therapy" which has been shown to provide benefits to patients who have had inadequate responses to methotrexate alone. The usual dose is 2-3 grams per day in a twice daily dosing regimen. The dose may be initiated at 1 gram per day and increased as tolerated towards a maximal tolerated dose which can be administered to an RA patient having the biomarkers of the invention in an RA patient sample. Leflunomide is also an effective DMARD. Its efficacy is similar to methotrexate in terms of signs and symptoms, and is a viable alternative to patients who have failed or are intolerant to methotrexate. Leflunomide has been demonstrated to slow radiographic progression. Studies have demonstrated that it can also be carefully combined with methotrexate in patients with no preexisting liver disease, as long as the liver function tests are carefully monitored. The half-life of the active metabolite of leflunomide is very long. Leflunomide and its metabolites are extensively protein bound and undergo further metabolism before excretion. When initially approved, the medication was given using a loading dose of 100mg daily for three days then followed by 20 mg daily. Higher doses can be used as an aggressive therapy in combination with the presence of the biomarkers of the invention in rheumatoid arthritis patients.

Tumor necrosis factor alpha (TNF) is a pro-inflammatory cytokine produced by macrophages and lymphocytes. It is found in large quantities in the rheumatoid joint and is produced locally in the joint by synovial macrophages and lymphocytes infiltrating the joint synovium. TNF is one of the critical cytokines that mediate joint damage and destruction due to its activities on many cells in the joint as well as effects on other organs and body systems. TNF antagonists were the first of the biological DMARDS to be approved for the treatment of RA. There are currently five TNF inhibitors FDA approved for the treatment of RA (listed in order of their approval for RA); etanercept (Enbrel®), infliximab (Remicade®), adalimumab (Humira®), certolizumab pegol (Cimzia®), and golimumab (Simponi®). Etanercept if a soluble TNF receptor-Fc immunoglobulin fusion construct; infliximab, adalimumab, and golimumab are monoclonal antibodies; and certolizumab pegol is an anti-TNF antigen binding domain- polyethylene glycol construct. The maximal tolerated doses of etanercept, adalimumab, infliximab, certolizumab pegol, golimumab are respectively 50 mg subcutaneous/week, 40 mg subcutaneous/week, 10 mg/kg every four weeks, 200mg/week, 50 mg every four weeks. It is up to the practitioner to decide whether to apply these maximal tolerated doses when the biomarkers of the invention are identified in a sample derived from an RA patient.

Abatacept (Orencia®) is the first of a class of agents known as T-cell costimulatory blockers. These agents interfere with the interactions between antigen-presenting cells and T lymphocytes and affect early stages in the pathogenic cascade of events in rheumatoid arthritis Abatacept is administered either via IV or subcutaneously. When given by intravenous infusion it is used once per month after initial doses at baseline, 2 weeks, and 4 weeks. The IV dose is based on body weight, with patients <60 kg receiving 500 mg, 60-100 kg receiving 750 mg, and >100 kg receiving 1000 mg. The medication is administered over a period of approximately 30 minutes to one hour. The subcutaneous version, a fixed dose of 125 mg regardless of weight, is administered once weekly with or without an intravenous loading dose based on body weight as above. Maximal doses can be applied when the biomarkers of the invention are identified in an RA patient sample.

Rituximab (Rituxan®). B cells are important inflammatory cells with multiple functions in the immune response. They serve as antigen presenting cells, directly interact with T-cells and others, can secrete cytokines, and differentiate into antibody-forming plasma cells. The depletion of B cells has been shown to be effective in reducing signs and symptoms of RA and in slowing radiographic progression. One B cell depleting agent, Rituximab, is currently available for the treatment of rheumatoid arthritis. Rituximab (Rituxan®) was originally developed to treat non-Hodgkin's lymphoma. Rituximab causes a rapid and sustained depletion of circulating B cells in the circulation with clinical improvements in many patients. Clinical trials have demonstrated that Rituximab is effective in decreasing signs and symptoms and in slowing radiographic progression in RA patients who have failed other DMARD therapies. The agent is currently approved in the US, however, only in patients who have failed TNF antagonists. Rituximab is a chimeric monoclonal antibody that binds to the CD20 molecule on the B cell surface leading to the removal of B cells from the circulation. The currently approved dose is 1000 mg administered intravenously over 3-4 hours with two doses given 2 weeks apart. Patients receive intravenous corticosteroids 30 minutes prior to each infusion. The optimal time for re-administration is not yet clear. Some have advocated treatment every 6 months, while others wait for a return of symptoms to redose. Doses of 500 mg have also been studied and appear to have similar clinical efficacy in patients who have failed to respond to DMARDS. Higher doses can be applied by the practitioner when the biomarkers of the invention are identified in an RA patient sample.

Tocilizumab (Actemra®). Tocilizumab is the first approved drug in a class of IL-6 inhibitors. Clinical studies have shown that tocilizumab is effective in decreasing signs and symptoms and in slowing radiographic progression in RA patients who have failed other DMARD therapies. The agent is currently approved in the US, however, only in patients who have failed TNF antagonists. When used in combination with DMARDs or as monotherapy the recommended starting dose is 4 mg/kg followed by an increase to 8 mg/kg based on clinical response. The high doses can be applied by the practitioner when the biomarkers of the invention are identified in an RA patient sample.

IL-1 is another proinflammatory cytokine implicated in the pathogenesis of RA. IL-1 receptor antagonist (IL1 ra) is an endogenous blocker of the cytokine. Evidence supporting an anti- inflammatory role of IL-1 ra in vivo is demonstrated by the observation that IL-1 ra deficient mice spontaneously develop autoimmune diseases similar to rheumatoid arthritis as well as vasculitis. IL1 has effects on cartilage degradation leading to damage as well as inhibiting repair, and is a potent stimulus to osteoclasts leading to bone erosion. One IL1 antagonist, anakinra (Kineret®), is currently approved for the treatment of RA. Other agents have been studied as well in RA.

Anakinra (Kineret™), a human recombinant IL-1 receptor antagonist (hu rlL-1 ra), is approved for the treatment of RA. Anakinra can be used alone or in combination with non-biologic DMARDs. Anakinra is a recombinant human IL-1 ra that differs from native IL-1 ra by the addition of an N-terminal methionine. Anakinra blocks the biologic activity of IL-1 by binding to IL-1 R type I with the same affinity as IL-1 beta. The recommended dose of anakinra is 100 mg/day administered daily by subcutaneous injection. Higher doses schemes can be applied by the practitioner when the biomarkers of the invention are identified in a rheumatoid arthritis patient sample.

In yet another embodiment the invention provides maximal tolerated doses of disease modifying anti-rheumatic drugs for use in reducing the joint destruction progression rate in rheumatoid arthritis patients possessing anti-citrullinated-protein-antibodies and possessing auto-antibodies directed against the SPAG16 protein and/or possessing anti-citrullinated protein-antibodies and one or more variants in the SPAG16 gene wherein said disease modifying anti-rheumatic drugs are selected from the list consisting of methotrexate, hydroxychloroquine, sulfasalazine, leflunomide, tumor necrosis factor inhibitors such as etanercept, infliximab, adalimumab, certolizumab pegol and golimumab, T-cell costimulatory blockers such as abatacept, B-cell depletion agents such as rituximab, interleukin 6 inhibitors such as tocilizumab and interleukin 1 receptor antagonists such as anakinra.

In one aspect the use of the protein biomarker of the invention relates to a process for detecting antibodies directed against SPAG16 protein (i.e. auto-antibodies) which are related to the presence of a worsening of the disease course of RA (i.e. an enhanced joint destruction progression rate in a rheumatoid arthritis patient) in a biological sample of a rheumatoid arthritis patient, more particularly in biological sample derived from a rheumatoid arthritis patient that has anti-citrullinated-protein antibodies. This process comprising contacting the biological sample with a composition comprising a SPAG16 protein or a peptide derived thereof of at least 5 consecutive amino acids present in the SPAG16 protein or a conformational epitope of the SPAG16 protein which is recognized by a SPAG16 antibody under conditions enabling an immunological reaction between said composition and the antibodies which are possibly present in the biological sample and the detection of the antigen/antibody complex which may be formed. The detection can be carried out according to any classical process. By way of examples immunoenzymatic processes according to the ELISA technique or immunofluorescent or radioimmunological (RIA) or the equivalent ones (e.g. LINE blot or LINE assay) can be used. Thus the invention also relates to SPAG16 polypeptides or immunological fragments derived thereof according to the invention labeled by an appropriate label of the enzymatic, fluorescent, biotin, radioactive type. Such a method for detecting antibodies related to RA comprises for instance the following steps: deposit of determined amounts of a SPAG16 polypeptidic composition according to the invention on a support (e.g. into wells of a titration microplate), introduction on said support (e.g. into wells) of (increasing dilutions of) the body fluid (e.g. plasma or serumor SF) to be diagnosed, incubation of the support (e.g. microplate), repeated rinsing of the support (e.g. microplate), the introduction on the support of labeled antibodies which are specific for immunoglobulins present in the body fluid, the labeling of these antibodies being based on the activity of an enzyme which is selected from among the ones which are able to hydrolyze a substrate by modifying the absorption of the radiation of this latter at least at a given wave length, detection by comparing with a control standard of the amount of hydrolyzed substrate. In yet another embodiment the invention also relates to a process for detecting and identifying the SPAG16 antigen in a body fluid derived from an RA patient liable to contain them. This process comprising: contacting the biological sample with an appropriate antibody directed against SPAG16 or an immunological fragment thereof (i.e. antibodies with a specificity for a SPAG16 polypeptide) under conditions enabling an immunological reaction between said antibody and the SPAG16 polypeptide which are possibly present in the biological sample derived from the RA patient and the detection of the antigen/antibody complex which may be formed.

Antibodies, in particular auto-antibodies, which recognize the SPAG16 polypeptides of the invention, can be detected in a variety of ways. One method of detection is uses enzyme- linked immunosorbant assay (ELISA) of the SPAG16 polypeptide of the invention displayed by phages (i.e. phage-ELISA technology). The latter technology is fully described in Somers V. et al (2005) J. of Autoimmunity 25: 223-228, wherein paragraph 2.6 on page 225 is herein specifically incorporated). In other ways in the detection in ELISA a SPAG16 polypeptide or an immunological fragment thereof is bound to a solid support. In some cases, this will be a microtiter plate but may in principle be any sort of insoluble solid phase (e.g. glass, nitrocellulose). In one embodiment a suitable dilution or dilutions of for example synovial fluid or serum or plasma derived from an RA patient to be tested is brought into contact with the solid phase to which the polypeptide is bound. In another embodiment "a solution hybridization" is carried out in which high affinity interactions occur (eg. biotinylated SPAG16 polypeptides are pre-incubated with plasma or serum). The incubation is carried out for a time necessary to allow the binding reaction to occur. Subsequently, unbound components are removed by washing the solid phase. The detection of immune complexes (i.e. auto-antibodies present in for example human plasma or serum binding to a SPAG16 polypeptide or immunological fragment thereof) is achieved using antibodies which specifically bind to human immunoglobulins, and which have been labeled with an enzyme, preferably but not limited to either horseradish peroxidase, alkaline phosphatase, or beta-galactosidase, which is capable of converting a colorless or nearly colorless substrate or co-substrate into a highly colored product or a product capable of forming a colored complex with a chromogen. Alternatively, the detection system may employ an enzyme which, in the presence of the proper substrate(s), emits light. The amount of product formed is detected either visually, spectrophotometrically, electrochemically, fluorescently or luminometrically, and is compared to a similarly treated control. The detection system may also employ radioactively labeled antibodies, in which case the amount of immune complex is quantified by scintillation counting or gamma counting. Other detection systems which may be used include those based on the use of protein A derived from Staphylococcus aureus Cowan strain I, protein G from group C Staphylococcus sp. (strain 26RP66), or systems which make use of the high affinity biotin-avidin or streptavidin binding reaction.

The SPAG16 polypeptide of the invention may be either labeled or unlabeled. Labels which may be employed may be of any type, such as enzymatic, chemical, fluorescent, luminescent, or radioactive. In addition, the SPAG16 polypeptide may be modified for binding to surfaces or solid phases, such as, for example, microtiter plates, nylon membranes, glass or plastic beads, and chromatographic supports such as cellulose, silica, or agarose. The methods by which polypeptides can be attached or bound to solid support or surface are well known to those skilled in the art.

The SPAG16 polypeptide of the invention can be prepared according to the classical techniques in the field of peptide synthesis. The synthesis can be carried out in homogeneous solution or in solid phase. The SPAG16 polypeptide of the invention can also be prepared in solid phase according to the method described by Atherton & Shepard in their book titled "Solid phase peptide synthesis" (Ed. IRL Press, Oxford, NY, Tokyo, 1989). Synthesis protocols in the art generally employ the use of t-butyloxycarbonyl- or 9-fluorenylmethoxy-carbonyl- protected activated amino acids. The procedures for carrying out the syntheses, the types of side-chain protection, and the cleavage methods are amply described in, for example, Stewart and Young, Solid Phase Peptide Synthesis, 2nd Edition, Pierce Chemical Company, 1984; and Atherton and Sheppard, Solid Phase Peptide Synthesis, IRL Press, 1989.

In yet another embodiment antibodies raised to a SPAG16 polypeptide of the invention (or carrier-bound polypeptides) can also be used in conjunction with a labeled SPAG16 polypeptide for the detection of (auto)-antibodies present in plasma serum or synovial fluid by a competition assay. In this case, antibodies raised to a SPAG16 polypeptide or an immunological fragment thereof are attached to a solid support which may be, for example, a plastic bead or a plastic tube. A labeled SPAG16 polypeptide is then mixed with suitable dilutions of the fluid (e.g. plasma or serum) to be tested and this mixture is subsequently brought into contact with the antibody bound to the solid support. After a suitable incubation period, the solid support is washed and the amount of labeled polypeptide is quantified. A reduction in the amount of label bound to the solid support is indicative of the presence of (auto)-antibodies in the original sample. By the same token, the polypeptide may also be bound to the solid support. Labeled antibody may then be allowed to compete with (autoantibody present in the sample (e.g. plasma or serum) under conditions in which the amount of polypeptide is limiting. As in the previous example, a reduction in the measured signal is indicative of the presence of (auto)-antibodies in the sample tested.

In a particular embodiment a test for giving evidence of the fact that the SPAG16 polypeptide are recognized by antibodies present in for example plasma, serum or SF (for example autoantibodies present in plasma, serum or SF of RA patients) is an immunoblotting (or Western blotting) analysis. In the latter case a SPAG16 polypeptide can be chemically synthesized or an immunological fragment thereof (or the complete SPAG16 protein) can be produced via recombinant techniques. In short, after sodium dodecyl sulfate-polyacrylamide gel electrophoresis, polypeptides of the invention are blotted onto nitrocellulose membranes (e.g. Hybond C. (Amersham)) as described by Towbin H. et al., 1979, "Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications", Proc. Natl. Acad. Sci. USA 76:4350-4354. In order to identify selective recognition of the SPAG16 polypeptide (or immunological fragments thereof) of the invention by plasma, serum or SF, nitrocellulose sheets are incubated overnight with each of these samples (e.g. diluted 1 :50) (after blocking a-specific protein-binding sites). Reactive areas on the nitrocellulose sheets are revealed by incubation with e.g. peroxidase conjugated goat anti-human immunoglobulin G antibody (e.g. diluted 1 :200) for 4 h, and after repeated washings, color reaction is developed by adding for example alpha-chloronaphtol (Bio-Rad Laboratories, Richmond, Calif.) in the presence of hydrogen peroxide.

It goes without saying that the free reactive functions which are present in some of the amino acids, which are part of the constitution of the SPAG16 polypeptide of the invention, particularly the free carboxyl groups which are carried by the groups Glu and Asp or by the C- terminal amino acid on the one hand and/or the free NH2 groups carried by the N-terminal amino acid or by amino acids inside the peptidic chain, for instance Lys, on the other hand, can be modified in so far as this modification does not alter the above mentioned properties of the polypeptide. The polypeptides which are thus modified are naturally part of the invention. The above mentioned carboxyl groups can be acylated or esterified. Other modifications are also part of the invention. Particularly, the amine or carboxyl functions or both of terminal amino acids can be themselves involved in the bond with other amino acids. For instance, the N-terminal amino acid can be linked to the C-terminal amino acid of another peptide comprising from 1 to several amino acids. Furthermore, any peptidic sequences resulting from the modification by substitution and/or by addition and/or by deletion of one or several amino acids of the polypeptides according to the invention are part of the invention in so far as this modification does not alter the above mentioned properties of said polypeptides. The polypeptides according to the invention can be glycosylated or not, particularly in some of their glycosylation sites of the type Asn-X-Ser or Asn-X-Thr, X representing any amino acid.

Variations of the SPAG16 polypeptide are also possible depending on its intended use. For example, if the SPAG16 polypeptide is to be used to raise antisera, the polypeptide may be synthesized with an extra cysteine residue added. This extra cysteine residue is preferably added to the amino terminus and facilitates the coupling of the polypeptide to a carrier protein which is necessary to render the small polypeptide immunogenic. If the polypeptide is to be labeled for use in radioimmune assays, it may be advantageous to synthesize the protein with a tyrosine attached to either the amino or carboxyl terminus to facilitate iodination. This polypeptide possesses therefore the primary sequence of the polypeptide above-mentioned but with additional amino acids which do not appear in the primary sequence of the protein and whose sole function is to confer the desired chemical properties to the polypeptide.

In yet another embodiment the invention provides for a kit to diagnose RA or to use to above described methods. To carry out the diagnostic method for RA, the following necessary or kit can be used, said kit comprising: a composition (comprising a SPAG16 polypeptide or an immunological fragment thereof (e.g. a fragment of SPAG16 protein comprising at least 5 consecutive amino acids derived from SPAG16), reagents for making a medium appropriate for the immunological reaction to occur, reagents enabling to detect the antigen/antibody complex which has been produced by the immunological reaction, said reagents possibly having a label, or being liable to be recognized by a labeled reagent, more particularly in the case where the above mentioned polypeptide is not labeled.

In yet another aspect of the invention allelic variants present in the SPAG16 gene are detected and its presence is associated with an enhanced joint destruction progression rate in a patient, more particularly in an RA patient possessing anti-citrullinated-protein antibodies. With the term 'SPAG16 gene' it is meant to include the promoter region of SPAG16, the intronic and exonic regions of SPAG16 and the terminator region. The human SPAG16 gene is known as the sperm associated antigen 16 and is located on chromosome 2q34. The RefSeq DNA sequence of the SPAG16 gene is available as NC 000002.1 1 or NT 005403.17. The GC identifier for this gene is GC02P214149. The GeneLoc location for GC02P214149: start is 214,149,1 13 bp from pier and the end is 215,275,225 bp from pter, which makes the total size of the SPAG16 gene of 1 ,126,1 13 base pairs.

SPAG16 encodes 2 major proteins that associate with the axoneme of sperm tail and the nucleus of postmeiotic germ cells. SPAG16 isoform 2, of 183 amino acids, was previously identified as an autoantibody target in multiple sclerosis (see Somers V et al (2008) J. of Immunol. 180(6):3957-3963).

In this second aspect of the SPAG16 biomarker, variant alleles of SPAG16 were identified in the present invention which presence in a rheumatoid arthritis patient sample is predictive for determining the joint destruction progression rate in a rheumatoid arthritis patient, in particular in a rheumatoid arthritis patient possessing anti-citrullinated-protein antibodies in a body sample.

In a particular embodiment said variant alleles of SPAG16 are selected from the group consisting of rs7607479 (SEQ ID NO: 1 ), rs6435780 (SEQ ID NO: 2), rs1400015 (SEQ ID NO: 3), rs10932481 (SEQ ID NO: 4), rs6435770 (SEQ ID NO: 5), rs13013558 (SEQ ID NO: 6). In yet another embodiment said variant alleles of SPAG16 are selected from the group consisting of: a) a sequence corresponding to rs7607479 (SEQ ID NO: 1 ) or having at least 80% homology, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology to SEQ ID NO: 1 having a substitution, deletion or addition of at least one nucleotide corresponding to position 27 of SEQ ID NO: 1 ; b) a sequence corresponding to rs6435780 (SEQ ID NO: 2) or having a least 80% homology, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology to SEQ ID NO: 2 having a substitution, deletion or addition of at least one nucleotide corresponding to position 27 of SEQ ID NO: 2; c) a sequence corresponding to rs1400015 (SEQ ID NO: 3) or having a least 80% homology, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology to SEQ ID NO: 3 having a substitution, deletion or addition of at least one nucleotide corresponding to position 27 of SEQ ID NO: 3; d) a sequence corresponding to rs10932481 (SEQ ID NO: 4) or having a least 80% homology, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology to SEQ ID NO: 4 having a substitution, deletion or addition of at least one nucleotide corresponding to position 27 of SEQ ID NO: 4; e) a sequence corresponding to rs6435770 (SEQ ID NO: 5) or having a least 80% homology, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology to SEQ ID NO: 5 having a substitution, deletion or addition of at least one nucleotide corresponding to position 27 of SEQ ID NO: 5; f) a sequence corresponding to rs13013558 (SEQ ID NO: 6) or having a least 80% homology, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology to SEQ ID NO: 6 having a substitution, deletion or addition of at least one nucleotides.

In yet another embodiment said variant alleles of the SPAG16 gene are selected from the group consisting of: a) a sequence corresponding to SEQ ID NO: 1 or having at least 80% homology, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology to SEQ ID NO: 1 , wherein position 27 is a C or a T, b) a sequence corresponding to SEQ ID NO: 2 or having at least 80% homology to SEQ ID NO: 2, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology, wherein position 27 is an A or a G, c) a sequence corresponding to SEQ ID NO: 3 or having at least 80% homology to SEQ ID NO: 3, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology wherein position 27 is a C or an A, d) a sequence corresponding to SEQ ID NO: 4 or having at least 80% homology to SEQ ID NO: 4, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology, wherein position 27 is an A or a G, e) a sequence corresponding to SEQ ID NO: 5 or having at least 80% homology to SEQ ID NO: 5, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology, wherein position 27 is a G or an A, f) a sequence corresponding to SEQ ID NO: 6 or having at least 80% homology to SEQ ID NO: 6, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology, wherein position 27 is an A or a G.

In yet another embodiment said variant alleles of the SPAG16 gene are selected from the group consisting of: a) a sequence corresponding to SEQ ID NO: 1 or having at least 80% homology, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology to SEQ ID NO: 1 , wherein position 27 is a T, b) a sequence corresponding to SEQ ID NO: 2 or having at least 80% homology to SEQ ID NO: 2, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology, wherein position 27 is a G, c) a sequence corresponding to SEQ ID NO: 3 or having at least 80% homology to SEQ ID NO: 3, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology wherein position 27 is an A, d) a sequence corresponding to SEQ ID NO: 4 or having at least 80% homology to SEQ ID NO: 4, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology, wherein position 27 is a G, e) a sequence corresponding to SEQ ID NO: 5 or having at least 80% homology to SEQ ID NO: 5, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology, wherein position 27 is an A, f) a sequence corresponding to SEQ ID NO: 6 or having at least 80% homology to SEQ ID NO: 6, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology, wherein position 27 is a G.

In a particular embodiment the presence of variant alleles of the SPAG16 gene selected from the group consisting of: a) a sequence corresponding to SEQ ID NO: 1 or having at least 80% homology, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology to SEQ ID NO: 1 , wherein position 27 is a C, and/or b) a sequence corresponding to SEQ ID NO: 2 or having at least 80% homology to SEQ ID NO: 2, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology, wherein position 27 is an A, and/or c) a sequence corresponding to SEQ ID NO: 3 or having at least 80% homology to SEQ ID NO: 3, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology wherein position 27 is a C, and/or d) a sequence corresponding to SEQ ID NO: 4 or having at least 80% homology to SEQ ID NO: 4, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology, wherein position 27 is an A, and/or e) a sequence corresponding to SEQ ID NO: 5 or having at least 80% homology to SEQ ID NO: 5, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology, wherein position 27 is a G, and/or f) a sequence corresponding to SEQ ID NO: 6 or having at least 80% homology to SEQ ID NO: 6, at least 85% homology, at least 90% homology, at least 95% homology, at least 96% homology, at least 97% homology, at least 98% homology, at least 99% homology, wherein position 27 is an A correlates with a reduction of the joint destruction progression rate in rheumatoid arthritis patients comprising anti-citrullinated-protein antibodies in a body fluid. A reduction of the joint destruction progression in RA patients having anti-citrullinated-protein antibodies in a body fluid correlates with the occurrence of the herein before described protective alleles. It is understood that the protective alleles can be present in the heterozygous or in the heterozygous state in a patient as can be inferred from Figure 1 , panel C. In yet another embodiment besides de SNPs herein described, also within the scope of this invention are SNPs that are linked to the SNPs herein described. Linkage disequilibrium is defined in the context of the present invention as having a D'>0.8. The person skilled in the art will understand that several well-accepted and/or related methods of determining linkage equilibrium may be applied to determine additional SNPs linked to the SNPs herein disclosed. In the context of the present invention, "homology" is understood to refer in the context of "variant alleles" to a sequence identity of at least 80%, particularly an identity of at least 85%, preferably at least 90% and still more preferably at least 95% over the full length of the sequence as defined by the SEQ ID NOs provided herein. In the context of this invention, a skilled person would understand that homology covers further variation(s) in the "variant alleles" in different ethnic groups. In accordance with the methods of the present invention, the herein defined SNPs are predictive of rheumatoid arthritis patients having an increased joint destruction progression rate. The identified SNPs can be used for the identification of rheumatoid arthritis patients who are eligible for receiving more aggressive doses of disease modifying anti-rheumatic drugs (or rheumatoid arthritis patients eligible for receiving maximal tolerated doses of disease- modifying anti-rheumatic drugs). In still other words the herein defined SNPs can be used for to select a subset of rheumatoid arthritis patients, in particular to select a subset of rheumatoid arthritis patients having anti-citrullinated-protein antibodies in a body sample; said subset of rheumatoid arthritis patients would benefit from a more aggressive therapy. The SNPs of the present invention can be identified by designing appropriate primers as commonly known in the art. The design is based on the sequences of the SNPs provided herein. Primers are preferably 10 to 45 nucleotides long and more preferably 15 to 30 nucleotides long. In a preferred aspect when more than one SNP is to be detected the amplification of the SNPs is carried out with a multiplex-PCR reaction. Primers are designed to bind upstream of a SNP in combination with a second primer designed to bind downstream of the same SNP. As will be understood by the person skilled in the art, a multitude of probes can be designed for each SNP identified herein. For example, probes can be extendable primers, which may be extended in a termination extension reaction.

Thus the present invention also provides oligonucleotides or polynucleotides useful for determining the presence of at least one or more variant alleles of the SPAG16 gene.

A skilled person will also have the ability to use probes, such as hybridization probes, that can be labelled by standard labelling techniques such as with a radiolabel, enzyme label, fluorescent label, biotin-avidin label, chemiluminescence, and the like. After hybridization, the probes can be visualized using known techniques. The present invention also relates to a diagnostic composition or kit comprising any of the mentioned oligonucleotides and optionally suitable means for detection.

The SPAG16 gene kit of the invention may advantageously be used for carrying out a method of the invention and could be, inter alia, employed in a variety of applications, e.g. in the diagnostic field or as a research tool. The parts of the kit of the invention can be packages individually in vials or in combination in containers or multicontainer units. Manufacture of the kit follows preferably standard procedures which are known to the person skilled in the art. The kit or diagnostic compositions may be used for detection of the one or more variant alleles of the SPAG16 gene in accordance with the herein-described methods of the invention, employing, for example, amplification techniques as described herein.

Accordingly, in a further embodiment of the present invention provides a kit useful for carrying out the methods herein described, comprising oligonucleotides or polynucleotides capable of determining the presence of at least one or more variant alleles of the SPAG16 gene. The oligonucleotides or polynucleotides may comprise primers and/or probes.

The following examples are intended to promote a further understanding of the present invention. While the present invention is described herein with reference to illustrated embodiments, it should be understood that the invention is not limited hereto. Those having ordinary skill in the art and access to the teachings herein will recognize additional modifications and embodiments within the scope thereof. Therefore, the present invention is limited only by the claims attached herein.

Examples 1 . Genome-wide association study on the progression of joint destruction rate in RA patients

The initial stage of the genome-wide association study on the severity of joint destruction was performed in a cohort of 398 ACPA-positive North-American RA-patients whom had a mean follow-up of 13.9±10.5 years. In a second stage these findings were replicated in a phenotypically well-characterized Dutch cohort of early RA-patients with yearly radiographs during 7-years.

Clinical characteristics of the subjects and the methodology used in both cohorts are summarized below.

All Rheumatoid Arthritis (RA) patients fulfilled the 1987 ACR criteria for RA. Ethical approval was given by the national ethics committee of both cohorts. Written informed consent was obtained from all participants.

A two staged approach was applied. In the first stage North-American patients from the NARAC (North American Rheumatoid Arthritis Consortium) were studied. The NARAC "family collection" samples were from multiplex families (primarily affected sibling pairs) in which at least one sibling had documented erosions, as seen on radiography of the hand, and at least one sibling had an onset of disease between the ages of 18 and 60 years. In addition, all NARAC patients were ACPA-positive. For the present study patients with radiographs of hands were studied, selecting only one patient of the sib-pairs (n=398). The North-American patients were diagnosed with RA between 1953 and 2002 and had obtained a conservative treatment. Presence of anticyclic citrullinated peptide antibodies (CCP) were determined using a second- generation enzyme-linked immunosorbent assay (ELISA) manufactured by Inova Diagnostics [San Diego, CA]. In a second stage (stage 2) the identified variants were replicated in a Dutch group of RA- patients. For replication, Dutch RA-patients included in the Leiden Early Arthritis Clinic between 1993 and 2006 were studied. This cohort is an inception cohort that includes consecutively referred patients with recent-onset arthritis from 1993 and onwards. Treatment changed in time and differed for inclusion periods. Patients included between 1993-1995 were initially treated with NSAIDs, patients included between 1996-1998 were initially treated with chloroquine or salazopyrine and patients included after in 1999 with methotrexate or salazopyrine. Patients treated early with biologicals or combination therapy in clinical trials were excluded from analysis. From all 602 RA-patients, 301 were anti-CCP-positive. Anti- CCP2 antibodies were measured using stored baseline serum samples (Immunoscan RA Mark 2; Euro-Diagnostica, The Netherlands)

The patient characteristics of the North-American and Dutch RA patients are shown in Table 1.

GWAs genotyping was conducted on lllumina BeadChips (HumanHap 550k, version 1 ). After rigorous genotyping and quality control, 392,337 SNPs were selected for further analysis. 10 patients were excluded to prevent population stratification. Since the genomic inflation factor was 1.009, we did not apply any form of adjustments for substructure. The threshold for genome-wide significance was set at P<5x10 ^"7 with 5 ^χ 10 ^"7 <Ρ<9.5 ^χ 10 ^"5 considered as highly suggestive. The radiographs were scored quantitatively using the Sharp-van der Heijde score (SHS) and the estimated progression per year was determined [reference 2]. Single SNPs were tested in relation to log-transformed radiological data with a linear regression analysis assuming an additive effect of each minor allele, adjusting for age and gender.

The strongest association was observed for a cluster of SNPs located at 2q34 (Fig. 1 and Table 2). Fourteen of the top 20 strongest associated SNPs were mapped to the region of SPerm associated AntiGen16 (SPAG16). Of these, rs7607479 achieved genome-wide significance (P=4.55x10 ^"7 ). Patients heterozygous and homozygous for the minor allele had a 0.77 (95%CI 0.70-0.85) and 0.59 (95%CI 0.49-0.72) higher progression rate per year compared to the common genotype (Fig. 3). The other SPAG16 SNPs displayed comparable P-values and effect sizes. Two additional SNPs suggestive for an association with progression rates were rs9301836 and rs2038326. ACPA-positive early RA-patients from the Dutch cohort with yearly radiographs during 7-years were studied for replication (n=301 ). The SNPs with associations of P<9.5*10 ^"5 in phase I and with LD of R ² <0.8, were genotyped by Sequenom. To take optimal advantage of the within patient correlation in longitudinal radiological data, a multivariate normal regression analysis was performed, testing the relative difference in radiologic progression rates between groups during follow-up. An additive model was applied with an interaction of time and with the log- transformed SHS as dependent variable; adjustments were made for age, gender, and treatment strategy [Ref. 3].

Also in this cohort, SNPs in 2q34 were associated with the rate of joint destruction (Fig. 1 c and Table 3). The strongest associations were seen for rs13013558 and rs7607479 (P=1 .5x10 ^"2 and P=2.2x10 ^"2 ). For both SNPs, the progression rate over 7-years for patients with one and two minor alleles was 0.81 (95%CI 0.67-0.97 and 0.68-0.96) and 0.65 (95%CI 0.45-0.94 and 0.46-0.92) times that of patients with the common genotype (Fig. 3a).

The variance in joint destruction explained by rs7607479 and rs13013558 were respectively 5.8% and 3.6% in the initial cohort, and 3.8% and 3.3% in the replication cohort.

Five of the six replicated SPAG16 SNPs, located on an intron, were in close LD in the Dutch RA-patients (R ² >0.8, D'=1 ). Haplotype analysis on the Dutch dataset resulted in only two haplotypes with a frequency >0.10 (AGAGT (0.61 ), GACAC (0.23)), making it hard to distinguish whether the GACAC haplotype contributes more to disease severity compared to the individual SNPs. The sixth SNP, rs13013558, is situated in the promoter region of SPAG16 (80kb upstream). Rs13013558 is also included in a LD block that encompasses the promoter region of IKZF2; a gene located 50kb downstream of rs13013558.

In the art several other GWAs found associations between SNPs encoding for SPAG16 and other quantitative traits such as height, weight, lipid levels and heart rate [Ref. 4 and 5]. Numerous transcript variants and corresponding protein isoforms have been reported for SPAG16 with expression in brain, testis, and also leucocytes, bone marrow, thymus and several others (http://genome.ucsc.edu) [Ref. 6]. We therefore studied whether SPAG16 is expressed in synovial tissue of the human joint. SPAG16 transcript was demonstrated in a cDNA expression library generated from RA synovial tissue (see Fig. 4a). Moreover, immunohistochemical staining for SPAG16 indicated expression of the protein in human synovial tissue from RA-patients (Fig. 4b).

Within the autoimmunity field, SPAG16 has been described in relation to multiple sclerosis (MS). In a previous study, antibodies against SPAG16 were identified in MS-patients. Moreover, anti-SPAG16 antibodies were shown to have disease-exacerbating effects in an animal model for MS, experimental autoimmune encephalomyelitis [Ref. 7]. Based on these findings and on the reported presence of antibodies to intracellular antigens in RA, such as vimentin, we hypothesized that an increased anti-SPAG16 antibody response might associate with severity of ACPA-positive RA. To this end a recombinant SPAG16 protein ELISA was performed on 257 Dutch ACPA-positive RA-patients with plasma at disease onset and 120 healthy controls. Auto-antibody reactivity above 50.7 arbitrary units was considered positive. No significant difference in anti-SPAG16 antibody positivity between RA-patients and controls was observed (Fig. 5). In contrast, a strong association for anti-SPAG16 antibody-positivity with rate of joint destruction was detected (Fig. 2). RA-patients with positive anti-SPAG16 antibody reactivity had 1.53 times higher progression rates over 7-years than patients without this antibody response (95%CI 1 .26-1 .87; P=2.8x10 ^"5 ). The variance in joint destruction explained by anti-SPAG16 antibody-positivity was 3.5%. The genotypes of rs7607479 and rs13013558 were not associated with the anti-SPAG16 antibody response. A multivariate normal regression analysis including both genotypes and anti-SPAG16 antibody positivity revealed that genotype and antibody response were independently associated with joint destruction. The combined analysis resulted in an explained variance of 10.1 %.

In the present invention we demonstrated that SPAG16 is expressed in synovial tissue and genetic variants in SPAG16 as well as antibody response against SPAG16 are associated with progression of joint destruction in ACPA-positive RA.

Materials and methods

1 . Genotyping and Quality-control filtering

The first stage (stage 1 ) consisted of 545,080 SNPs genotyped in 398 samples from the North- American subjects. The samples were genotyped by SNP assay with Infinium HumanHap550, version 1 .0 (lllumina). All lllumina genotyping was performed at the Feinstein Institute for Medical Research Samples according to the lllumina Infinium 2 assay manual (lllumina, San Diego), as previously described (Padyukov L et al (2004) Arthritis Rheum. 50: 3085-92). To assess lllumina 550K genotype accuracy for the SNPs advanced to replication, a subset of the samples (ca. 50%) were re-genotyped using a different genotyping platform (Sequenom iPLEX Gold at the Broad Institute). 99.605% concordance between the two genotyping platforms was observed.

The data set was filtered on the basis of SNP genotype success rates (>98% completeness), minor allele frequency (>0.1 ), and Hardy-Weinberg equilibrium (P<1 *10 ^~5 ). In total 392,337 SNPs were included in the Genome-wide analysis. Patients were excluded if more than 10% of the SNPs failed (n=0), if there was evidence of relatedness by identity-by-decent analysis (n=0) and if there was evidence of population substructures N=10). Subsequently, 388 patients were studied. Since the genomic inflation factor was 1 .009 we did not apply any form of adjustments for substructure. None of the genotyping data were imputed.

In the second stage (stage 2), the strongest associated SNPs from the GWAs (P<9.5* 10 ^~5 ) provided that the R ² <0.8, were typed in the Dutch cohort. Genotyping was done using Sequenom iPLEX, according to the protocols recommended by the manufacturer (Sequenom, San Diego, California). SNPs that were in close linkage disequilibrium, defined by R ² >0.8, were excluded. Software supplied by the same manufacturer was used to automatically identify the genotypes. Each iPLEX consisted of eight positive and eight negative controls, which were indeed positive and negative in all analyses. Clusters were evaluated and all doubtful calls were checked; after manually evaluating the spectra of each cluster, the genotypes were accepted, recalled or rejected. At least 10% of the genotypes were assessed in duplicate, with an error rate of <1 %. The overall success rates was >97%. None of the SNPs were out of Hardy-Weinberg equilibrium (P<0.001 ).

2. Radiograph scoring

Radiographs in both cohorts were scored at the Leiden University Medical Centre (LUMC) with the Sharp-van der Heijde method (SHS). SHS is a validated quantitative method to score joint destruction, which is sensitive to change and evaluates both fractions of joint destruction; bone erosions and cartilage damage (visualized as joint space narrowing). The total SHS was used for analyses.

Stage 1

Of each patient, a radiograph of the hands was available at one time-point. The disease durations at this time-point differed between the patients. All radiographs were scored by one reader (JvN) whom was unaware of genetic or clinical data. 25% of the radiographs were rescored. The intra-observer reliability (intraclass correlation coefficient (ICC)) was 0.99. To calculate the progression rate, the total SHS was divided by disease duration at time of the radiograph, resulting in an estimated progression per year.

Stage 2 Patients had hand and feet radiographs taken at baseline and on consecutive years. Although the maximum follow-up duration in this inception cohort was 15-years, the radiological data were restricted to 7-years of follow-up, because of increasing frequency of missing radiographs later on. All radiographs were chronologically scored by one experienced reader (MvdL) whom was unaware of genetic or clinical data. 499 radiographs (belonging to 60 randomly selected RA-patients) were rescored. The ICC was 0.91 for all scored radiographs, and 0.97 for the radiographic progression rate.

3. SPAG16 ELISA 258 of the 301 ACPA-positive Dutch RA-patients with plasma available at disease onset and 123 Dutch healthy controls were used to investigate antibody response against SPAG16. Baseline characteristics of patients with and without plasma antibodies were statistically significantly different (data not shown).

SPAG16 was identified as an autoantibody target by profiling the auto-antibodies present within MS cerebrospinal fluid through affinity selection of a cDNA phage display library generated from MS plaque tissue (Ref. 7). In this study, the protein product of SPAG16 transcript variant 2a, SPAG16 isoform 2, was identified as autoantibody target. This protein encodes 183 amino acids of which the first 178 are identical to an amino acid stretch that is present in SPAG16 isoform 1 and in at least several of the other isoforms. Recombinant SPAG16 protein isoform 2 was produced as described. [ref. Somers K, J of Autoimmmunity, 2010] The recombinant protein was expressed as a fusion protein to thioredoxin at the N- terminal side and a His ₆ -tag at the C-terminus. Recombinant protein ELISA was performed as described (Somers K et al (2010) J of Autoimmunity). Background reactivity was measured by incubating plasma with recombinant thioredoxin synthesized in an identical manner. Each sample was tested in duplicate in three independent experiments performed on different days. Each ELISA plate contained an anti-SPAG16 antibody-positive MS sample, an antibody- negative healthy control sample and a blanc (no serum). A series of successive plasma dilutions (1 :25, 1 :50, 1 :100, 1 :200, 1 :400, 1 :800, 1 :1600) of an RA-patient with high anti- SPAG16 antibody response was used as a reference standard in all plates, in order to calculate arbitrary units (AU) for each sample (see figure 6). After log-transformation, the linear part of the standard curve was between the 1 :25 and 1 :200 dilution. As the linear part of the standard curve ended at 50,7 AU (1 :200 dilution), this was used as cut-off for anti-SPAG16 positivity. A patient sample was considered to be anti-SPAG16 antibody-positive when the AU- value on the SPAG16 protein was above the cut-off point and the OD difference (=OD for SPAG16 protein - OD for thioredoxin protein) was at least 0.1 . Patients with a high response on thioredoxin were excluded from the analysis, since it could not be determined whether their antibody-response to SPAG16 was false-negative or -positive (RA n=1 ; Healthy Controls n=3). Patients were defined as anti-SPAG16 antibody-positive when their samples were positive in at least 2 out of 3 independent experiments. The variance between AU values obtained in independent experiments performed on different days, as exemplified in figure 7, was very low. All blancs were negative. The negative and positive control samples were indeed tested negative and positive on all plates.

Anti-SPAG16 antibody positivity was not correlated with ACPA-levels (data not shown). 4. SPAG16 immunohistochemistry

Mouse monoclonal antibodies against SPAG16 were produced by application of hybridoma technology. In brief, 6 to 8 week old female BALB/c mice (Harlan, Belgium) were immunized by intraperitoneal injection of 50 μg of full-length SPAG16 isoform 2 recombinant protein in complete Freund's adjuvant (Sigma-Aldrich, Bornem, Belgium), followed by a second injection after 14 days of the same antigen amount in incomplete Freund's adjuvant (Sigma-Aldrich). At sufficient mouse antibody titer, spleen cells were fused with a mouse myeloma cell line (Sp2/0) followed by culture in selective media allowing selection of immortal hybridoma cell lines. Recombinant protein ELISA was used to select anti-SPAG16 antibody-producing cell lines. Moreover, we selected for those anti-SPAG16 isoform 2 monoclonal antibodies that competed with antibody-positive MS serum for binding to recombinant SPAG-16 isoform 2, as these antibodies are representative for the immunoreactivity of MS patient antibodies against SPAG16 isoform 2. Based on those selection criteria, 2 hybridomas producing lgG1 (1 F1 and 5F10) monoclonal antibodies against SPAG16 were selected. Antibodies were purified from tissue culture supernatant by POROS A (Applied Biosystems, Halle, Belgium) affinity- chromatography (BioCAD). Eluted antibodies were purified and dialysed (Slide-A-Lyzer, Thermo Fisher Scientific, Erembodegem, Belgium) against PBS. Purity of the produced monoclonal antibodies was analyzed by SDS-PAGE. Isotypes of the produced antibodies were determined (both lgG1 ) with the mouse monoclonal antibody isotyping kit according to the manufacturer's recommendations (Hycult Biotechnology, Uden, the Netherlands). Anti-human chorion gonadotrophin lgG1 antibody produced and purified in the same manner was used as an isotype control antibody. Formalin-fixed, paraffin-embedded tissue sections from synovial tissue obtained from a total knee-replacement operation of RA-patients were stained to study SPAG16 expression. Paraffin-embedded sections were dewaxed and rehydrated. Endogenous peroxidase activity was blocked by incubation of the slides in 0.3% (v/v) hydrogen peroxide in methanol during 10 minutes. After blocking with rabbit serum (10% (v/v) in TBS, Tris-buffered saline, 20 mM Tris-CI, 150 mM sodium chloride, pH 7.5), incubations with primary mouse monoclonal antibodies (5F10 and 1 F1 mix, 30 μg/ml in TBS) were performed overnight at room temperature. Bound antibodies were detected with the Dako Envision anti-mouse detection system (Dako, Heverlee, Belgium) (30 minutes room temperature), followed by colour development with DAB substrate (Sigma-Aldrich). Counterstaining was performed with Gill's haematoxylin (Klinipath, Olen, Belgium). Sections were coverslipped with Clarion Mounting Medium (Sigma-Aldrich). TBS and TBS supplemented with 0.05% (v/v) TritonX-100 were used for in between washing steps. An identically produced isotype control monoclonal antibody (anti-human chorion gonadotrophin lgG1 ) was used as a negative control. 5. SPAG16 PCR

SPAG16 transcript variant 2a PCR was performed on RA cDNA library plasmid (M13 pSPVI- A B/C phagemid) (Somers K et al (2010) J of Autoimmunity). 100ng plasmid was used in PCR with previously described forward primer (5'-ctgctcagcgagggatgcc-3') (SEQ ID NO: 7) and reverse primer (5'-gaaggtacagagatgtccca-3') (SEQ ID NO: 8) (Eurogentec, Ougree, Belgium) (ref Pennarun). After an initial denaturation step for 10 minutes at 95°C, SPAG16 isoform 2 cDNA was amplified by 35 PCR cycles (30 seconds 94°C, 30 seconds 55°C and 1 minute 72°C). PCR product was visualized by agarose gel electrophoresis followed by ethidium bromide staining.

6. Statistical analysis Stage 1

In the North-Americans, the estimated radiological progression per year between groups was compared using a linear regression model assuming an additive effect, correcting for age and gender. To obtain normal distribution, radiological scores were transformed to the log naturalis scale. The Bonferroni correction for multiple testing is frequently used to determine the threshold for genome-wide significance. This cut-off may be too conservative if there is substantial dependence between the association tests at the various SNPs (Nelson Freimer N, Sabatti C. Nature Genetics 36, 1045 - 1051 (2004). In addition, the number of patients studied affects the level of significance as well. Hoggart CJ et al (2008) Genet. Epidemiol. 32(2): 179-85) calculated that the threshold for significance is 2.9x10 ^"7 in a study on 200 subjects with 660k genotypic data. Extrapolating these calculations to our number of SNPs (n=392,337) and patients (n=388) resulted in a threshold of 5x10 ^"7 . Therefore, in the present study the p-value for genome-wide significance was set at P<5x10 ^"7 . This is comparable to the threshold used in the WTCCC for the analysis of 500,000 SNPs tested on 2,000 cases versus 3,000 controls. [The Wellcome Trust Case Control Consortium. Nature 447, 661 -678 (2007)]

The explained variance of the significant SNPs was determined by the Nagelkerke's R2 of the linear regression analysis. Stage 2

To take advantage of the prospective character of the data, consisting of repeated measurements, and to avoid multiple testing by performing statistical tests for each time point, a multivariate normal regression analysis for longitudinal data was used with radiological score as response variable. Because of non-normal distribution, radiological scores were log- transformed. The effect of time was entered as factor in the model, allowing a mean response profile over time. The effect of time was assumed to be linear in the interaction terms. For each SNP an analysis with the SNP and its interaction with time in the model were conducted. The SNPs were presumed to have an additive effect and therefore entered as continuous variables. The analysis was corrected for age, gender and treatment strategy. The latter served as proxy for treatment strategy. For the replication stage, a P-value <0.05 was considered significant.

In addition to the single SNP analyses, a haplotype analysis of the significant SNPs which were in close LD was performed. LD-blocks were determined in Haploview according to the Gabriels method. [ref] Haplotypes of the individual patients were obtained from PLINK. Haplotypes with a probability >0.95 and a frequency >0.10 were further studied. Haplotype analysis was performed similar to the single SNP analysis; the number of haplotypes present was tested in a multivariate regression analysis presuming an additive effect correcting for age, gender and inclusion period with log-SHS as dependent variable. The variance of the residual (o ² ) of the multivariate regression analysis reflects the unexplained within persons variance. The explained variance is calculated by the difference in residual variance o ² of the model with time and inclusion period and of the model that also includes additional variables such as age and gender

7. Anti-SPAG16 antibody ELISA Positivity of anti-SPAG16 antibody response between healthy controls and RA-patients and among genotypes within RA-patients was compared with a Pearson's Chi-square test. Secondly, the rate of joint destruction between anti-SPAG16 antibody-positive and negative patients was compared using a multivariate normal regression model, as described previously in this application. For these analyses a P-value <0.05 was considered significant. 8. Software

The genome-wide linear regression analyses were calculated using PLINK 1 .0.6. [ref 2] Identification of haplotypes was performed in Haploview 4.1 and individual haplotypes were obtained from PLINK 1.0.6. All other statistical analyses were performed using Statistical Package for Social Sciences version 17.0 (SPSS, Chicago, IL).

Tables

Table 1 Baseline characteristics of patients passing quality control and principle components.

North-American Dutch

N 388 301

Age at onset (mean ± sd) 40.8±11.9 years 55.8±14.8 years

Female (%) 77 68

ACPA-positivity (%) 100% 100%

Year of Diagnosis/inclusion (mean (min-max)) 1985 (1953-2002) 1999 (1993-2006)

Follow-up duration (meanisd) 13.9±10.5 5.8±3.6

Table 2: Results of GWAs in North-Americans: SNPs with P<9.5x10 ^"5

Chr SNP Position Beta 95% CI P

2 rs7607479 214251829 0.77 0.70-0.85 4.55E-07

2 rs2161820 214179111 0.77 0.70-0.85 4.61 E-07

2 rs7584873 214228332 0.77 0.70-0.86 6.38E-07

2 rs1400015 214171061 0.77 0.70-0.85 8.90E-07

2 rs6435777 214034224 0.77 0.70-0.86 1.12E-06

2 rs12997815 214066898 0.77 0.70-0.86 1.20E-06

2 rs6435780 214221392 0.78 0.70-0.86 1.71E-06

2 rs6435770 213922542 0.77 0.70-0.86 2.21 E-06

2 rs6749651 214277740 0.78 0.71 -0.87 2.24E-06 Chr SNP Position Beta 95% CI P

^~ 2 rs10932483 214076656 0.78 0.69-0.86 2.87E-06

2 rs10932481 213988295 0.77 0.70-0.86 3.03E-06

2 rs7580474 213871829 0.77 0.70-0.86 3.57E-06

2 rs12472571 214276034 0.79 0.71 -0.87 3.79E-06

2 rs4132244 214264772 0.78 0.71 -0.87 5.05E-06

13 rs9301836 92275312 1.37 1.20 - 1.57 5.30E-06 22 rs2038326 24519026 0.76 0.68 -0.86 7.07E-06

5 rs4868934 164940665 0.79 0.71 -0.88 2.34E-05

6 rs765466 72181715 0.70 0.59-0.82 2.37E-05 1 rs12023478 245517188 1.30 1.15- 1.46 2.49E-05

14 rs7156170 65484357 1.25 1.13- 1.39 2.70E-05 22 rs761878 48635681 0.73 0.63-0.85 3.06E-05 22 rs4820656 24522077 0.78 0.70-0.88 3.11E-05 13 rs9556229 92302532 1.38 1.19- 1.60 3.44E-05 9 rs2821204 11673901 1.31 1.15- 1.48 4.14E-05 6 rs9485739 96677858 0.80 0.72-0.89 4.63E-05 1 rs6690864 245557330 1.28 1.14- 1.44 4.71 E-05

6 rs17642884 133874977 0.74 0.64-0.85 4.77E-05 22 rs767249 42445146 0.76 0.67-0.87 5.15E-05

7 rs2177549 54151555 0.77 0.68-0.87 5.26E-05 7 rs6593122 54149426 0.77 0.68-0.87 5.26E-05 22 rs138845 48571196 0.75 0.65-0.68 5.29E-05 1 rs6426239 245553359 1.28 1.14- 1.44 5.57E-05 Chr SNP Position Beta 95% CI P

1 rs12023326 245537744 1.28 1.14- 1.44 5.67E-05 18 rs555448 64399863 0.78 0.69-0.88 5.92E-05

9 rs2417682 107819499 1.23 1.11 - 1.36 5.93E-05 15 rs2120968 20483952 0.79 0.70-0.89 6.29E-05 15 rs6493430 48430208 0.82 0.74-0.90 6.41 E-05

10 rs2912795 123240238 1.23 1.11 - 1.35 6.68E-05 6 rs933042 97241898 0.81 0.73-0.90 6.69E-05 14 rs8013361 65483143 1.24 1.12- 1.37 6.70E-05

2 rs13013558 213777709 0.80 0.72-0.89 7.15E-05

11 rs486058 78226894 1.29 1.14- 1.47 8.11 E-05 6 rs1410664 72570991 1.31 1.15- 1.50 8.49E-05 22 rs3859874 24539678 0.80 0.72-0.89 9.19E-05

Table 3: Results of stage 2

Chr SNP Position Beta 95% CI P

2 rs13013558 213777709 0.81 0.68 - 0.96 0.015

2 rs7607479 214251829 0.81 0.67 - 0.97 0.022

2 rs10932481 213988295 0.82 0.68 - 0.98 0.029

2 rs1400015 214171061 0.82 0.68 - 0.98 0.034

2 rs6435770 213922542 0.83 0.69 - 0.99 0.039

2 rs6435780 214221392 0.83 0.70 - 1.00 0.044

6 rs9485739 96677858 1.15 0.95 - 1.38 0.152

22 rs767249 42445146 1.17 0.92 - 1.49 0.203 Chr SNP Position Beta 95% CI P

6 rs765466 72181715 1.17 0.88 - ^■ 1.55 0.274

11 rs486058 78226894 1.11 0.91 - ^■ 1.35 0.317

6 rs933042 97241898 0.92 0.77 - ^■ 1.09 0.333

22 rs138845 48571196 1.09 0.87 - ^■ 1.36 0.450

1 rs12023478 245517188 1.08 0.88 - ^■ 1.32 0.463

22 rs2038326 24519026 1.06 0.87 - ^■ 1.29 0.544

15 rs2120968 20483952 1.78 0.86 - ^■ 1.30 0.577

9 rs2417682 107819499 1.04 0.88 - ^■ 1.22 0.637

14 rs8013361 65483143 0.97 0.81 - ^■ 1.15 0.698

7 rs2177549 54151555 0.96 0.77 - ^■ 1.20 0.708

6 rs17642884 133874977 1.04 0.82 - ^■ 1.31 0.744

7 rs6593122 54149426 0.96 0.77 - ^■ 1.21 0.746

14 rs7156170 65484357 0.97 0.81 - ^■ 1.16 0.751

10 rs2912795 123240238 0.97 0.82 - ^■ 1.15 0.757

22 rs761878 48635681 0.97 0.77 - ^■ 1.23 0.822

6 rs1410664 72570991 1.02 0.84 - ^■ 1.24 0.862

18 rs555448 64399863 1.01 0.82 - ^■ 1.23 0.953

5 rs4868934 164940665 1.00 0.85 - ^■ 1.19 0.974

1 rs6690864 245557330 1.00 0.83 - ^■ 1.22 0.977

13 rs9556229 92302532 1.00 0.8 - ^■ 1.24 0.987

Replication of stage 1 SNPs with P-value below 9.5x10 ^"5 . If SNPs had a R ² >0.8, one SNP was chosen for stage 2. Rs10932483, rs12023326, rs12472571, rs12997815, rs2161820 were not part of the replication phase, since they were in close LD with other SNPs that were typed in this phase. The typing failed for rs2821204. References

1. Van der Helm-van Mil, A. et al. Arthritis Rheum. 54, 2028-2030 (2006).

2. Van der Helm-van Mil, A. et al. Rheumatology. 49, 1429-1435 (2010),

3. van der Linden, M. et al. Arthritis Rheum. 60, 2242-2247 (2009)

4. Johnson, D. et al. BMC Medical Genetics 10, 6 (2009).

5. Rankinen, T. et al. Circ Cardiovasc Genet 3, 294-299,(2010)

6. Pennarun, G. et al. Am. J. Respir. Cell Mol. Biol. 26, 362-370 (2002).

7. Somers V, Govarts C, Somers K, Hupperts R, Medaer R, Stinissen P. Autoantibody profiling in Multiple Sclerosis reveals novel antigenic candidates. J. Immunol. 2008:180(6):3957-3963.

8. Somers K, Geusens P, Elewaut D, De Keyser F, Rummens JL, Coenen M, Blom M, Stinissen P, Somers V. Novel autoantibody markers for early and seronegative rheumatoid arthritis. Journal of Autoimmunity: 201 1 Feb;36(1 ):33-46. Epub 2010 Nov 10.

9. Somers, K. et al. Journal of Neuroimmunology 228, 162 (2010) Abstract.