저작자표시 - 비영리 - 변경금지 2.0 대한민국 이용자는아래의조건을따르는경우에한하여자유롭게 이저작물을복제, 배포, 전송, 전시, 공연및방송할수있습니다. 다음과같은조건을따라야합니다 : 저작자표시. 귀하는원저작자를표시하여야합니다. 비영리. 귀하는이저작물을영리목적으로이용할

저작자표시 - 비영리 - 변경금지 2.0 대한민국 이용자는아래의조건을따르는경우에한하여자유롭게 이저작물을복제, 배포, 전송, 전시, 공연및방송할수있습니다. 다음과같은조건을따라야합니다 : 저작자표시. 귀하는원저작자를표시하여야합니다. 비영리. 귀하는이저작물을영리목적으로이용할수없습니다. 변경금지. 귀하는이저작물을개작, 변형또는가공할수없습니다. 귀하는, 이저작물의재이용이나배포의경우, 이저작물에적용된이용허락조건을명확하게나타내어야합니다. 저작권자로부터별도의허가를받으면이러한조건들은적용되지않습니다. 저작권법에따른이용자의권리는위의내용에의하여영향을받지않습니다. 이것은이용허락규약 (Legal Code) 을이해하기쉽게요약한것입니다. Disclaimer

High serum allograft inflammatory factor 1 is associated with poor response to TNFα inhibitors in ankylosing spondylitis patients

Abstract High serum allograft inflammatory factor 1 is associated with poor response to TNFα inhibitors in ankylosing spondylitis patients Eunyoung Lee Department of Internal Medicine, College of Medicine The Graduate School Seoul National University Background: Anti-TNFα therapy has been proven to be highly efficacious in ankylosing spondylitis (AS). Considering its high costs and potential risk for adverse events, early detection of non-responders to anti-tnfα agents is critical. Objectives: To identify serum markers predicting clinical response to TNFα blockers in AS Methods: Baseline gene expression differences were screened by i

pathway focused gene assays of peripheral blood RNA from 6 AS patients (3 responders and 3 non-responders) before initiating anti-tnfα treatment, and selected results were confirmed by qrt-pcr in 18 patients (11 responders and 11 non-responders). Concentration of corresponding serum protein was measured by ELISA and compared in 69 responders and 48 non-responders. No response to TNFα blocker was defined as less than 50% improvement in Bath ankylosing spondylitis disease activity score (BASDAI) at week 14 from baseline. Results: Nine candidate genes were selected from gene assays and validated by qrt-pcr. Among these genes, the expression of allograft inflammatory factor 1 (AIF1) was 3.52 fold higher in non-responders than responders (p=0.032). The serum AIF1 level at baseline was significantly higher in BASDAI 50 non-responders; median 32.8 [IQR 20.6;67.3] pg/ml in responders and 54.2 [28.9;91.0] pg/ml in nonresponders (p=0.033). AIF1 level of 63.5 pg/ml or more was associated with higher risk for BASDAI > 5.0 at week 14 after anti-tnfα treatment (adjusted OR 6.953, p=0.002). Conclusion: Baseline serum AIF1 level was higher in TNFα blocker non-responders. After adjusting age and initial BASDAI, high concentration of baseline AIF1 was associated with high disease activity after TNFα blocker treatment. These results suggest that AIF1 may be a ii

novel serum marker for predicting non-responders to anti-tnfα therapy in AS. Keywords: Ankylosing spondylitis, anti-tnf, allograft inflammatory factor 1, AIF1, BASDAI, serum marker Student Number: 2016-21920 iii

Index Abstract Index List of Tables List of Figures List of Supplementary Tables List of Supplementary Figures I. Introduction II. Subjects and Methods III. Results IV. Discussion V. References iv

List of Tables Table 1. Demographic and clinical parameters of 6 ankylosing spondylitis patients with or without BASDAI 50 response* (3 each) evaluated by initial pathway focused gene assays... 12 Table 2. Nine genes selected from screening gene assays... 13 Table 3. Demographic and clinical parameters of 18 ankylosing spondylitis patients with or without BASDAI 50 response* analyzed by qrt-pcr.... 14 Table 4. Comparison of gene expression at baseline between BASDAI 50 responders* and non-responders... 15 Table 5. Demographic and clinical characteristics of ankylosing spondylitis patients with or without BASDAI 50 response* included in serum AIF1 ELISA.... 18 Table 6. Comparison of baseline AIF1 levels in ankylosing spondylitis patients receiving anti-tnfα agents according to several response criteria (BASDAI 50, ASAS 20 and ASAS 40).... 19 Table 7. Logistic regression analysis for predicting high disease activity* after anti-tnfα treatment for 14 weeks.... 21 Table 8. Univariate analysis for correlation between baseline AIF1 levels and clinical variables... 22 v

List of Figures Figure 1. Overall study flow... 6 Figure 2. ROC curve of baseline serum AIF1 level (A) and final multivariate logistic regression (B) for predicting high disease activity * at week 14... 20 Figure 3.. Changes of serum AIF1 concentration in ankylosing spondylitis patients after anti-tnf treatment according to BASDAI 50 response * at week 14... 24 Figure 4. Comparison of baseline serum concentration of LIGHT, VEGF and leptin between BASDAI 50 responders * (n=10) and non-responders (n=10).... 28 Figure 5 Changes of serum levels of LIGHT, VEGF and leptin after anti- TNF treatment in BASDAI 50 responders * (n=10) and non-responders (n=10).... 29 vi

List of Supplementary tables Supplementary table 1. Demographic and clinical characteristics of ankylosing spondylitis patients analyzed by multiplex ELISA.... 27 vii

List of Supplementary figures Supplementary figure 1. Overall study flow and included subjects... 7 viii

I. Introduction Ankylosing spondylitis (AS) is a common immune-mediated inflammatory arthritis affecting sacroiliac joints, spine and peripheral joints 1,2. Patients suffer from back pain, peripheral arthralgia, stiffness and fatigue. Treatment goals are to reduce the symptoms, sustain functional status and delay the structural damage. 3 The mainstay of pharmacotherapy is nonsteroidal anti-inflammatory drugs (NSAIDs). Anti-tumor necrosis factor (TNF)α therapy should be considered for patients with persistently high disease activity despite conventional treatments. 3,4 TNFα blockers have been proven to show rapid and significant improvements in symptoms and disease activity. 5,6 However, 40-50% of patients do not respond to TNFα blockers 7-9. Considering its high costs 10,11 and potential risk of adverse events such as infections and hypersensitivity reactions 12,13, early detection of non-responders to anti- TNFα agents is critical. Previously suggested serum markers such as serum amyloid A (SAA) 14, leptin, N-terminal propeptide of type 1 collagen (P1NP) and insulin 15 did not have validated cutoff values to predict response to anti-tnfα treatment and most of them were not well correlated with the pathophysiology of AS. Thus, there is unmet needs for serum markers identifying non-responders to anti-tnfα treatment. Recent study demonstrated that the combination of elevated baseline C-reactive protein (CRP) and SAA can predict the clinical response to 1

anti-tnfα therapy. 14 However, these parameters have limitations because other studies showed that some good responders to anti-tnfα treatment also could have elevated baseline CRP or SAA. 16 In a report about serum biomarkers in AS patients treated with golimumab, the combination of baseline levels of leptin, immunoglobulin M and vascular endothelial growth factor (VEGF) had more accuracy in prediction of Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) 50 response compared to CRP alone. Additionally, baseline level of P1NP and insulin were suggested to be the strong predictors of Assessment in Ankylosing Spondylitis 20 (ASAS 20) response and lower baseline leptin level was associated with the ASAS 20 response and improvement in Bath Ankylosing Spondylitis Functional Index (BASFI). 15 However, other reports did not support the associations between leptin levels and disease activity in AS patients. 17,18 Allograft inflammatory factor 1 (AIF1) is a distinctive protein that increases during inflammation 19 and was originally identified and cloned from rat cardiac allografts with chronic rejection. 20 It is encoded within the HLA class III region, chromosome 6p21.3, which is densely clustered with genes associated with inflammatory responses including TNFα, NFκB and complement cascade. 21 AIF1 is known to have roles in various inflammatory conditions such as allograft rejection, rheumatoid arthritis (RA), systemic sclerosis, uveitis and autoimmune 2

encephalomyelitis. 19,20,22,23 AIF1 was reported to have a role in the pathogenesis of RA by increasing synovial cell proliferation and interleukin 6 (IL-6) production. 22 In the present study, we investigated baseline differences of gene expressions between responders and non-responders and evaluated 433 genes with pathway-focused RT-PCR panels. Then we selected and narrowed down the possible markers by additional qrt-pcr and ELISA. The final candidate marker was AIF1 and we investigated whether its serum level could predict the response to treatment in patients with AS receiving TNFα inhibitors. 3

II. Subjects and methods 1. Subjects Overall study flow and included subjects are shown at figure 1 and supplementary figure 1. Patients fulfilling the 1984 modified New York criteria for AS 24 and treated with anti-tnfα agents were enrolled from the rheumatology clinic at Seoul National University Hospital (SNUH). Disease activity was assessed using the BASDAI, which has a possible score of 0 10 with a higher score indicating greater disease activity. 25 The patients had BASDAI score at baseline and 14 weeks after the treatment and were grouped as responders if they achieved 50% or more improvement in BASDAI (BASDAI 50) at week 14 from baseline. High disease activity was defined as BASDAI > 5.0 at week 14. 26 Patients clinical features and laboratory findings such as erythrocyte sedimentation rate (ESR) and CRP were collected from the electronic medical records (EMR) of SNUH. Peripheral venous blood samples for quantitative real time (qrt)-pcr and ELISA were taken before and 14 weeks after the treatment. For initial screening gene assay, 3 responders and 3 non-responders were selected from this cohort. Nine genes were selected from them and the results were confirmed in other 18 patients (11 responders and 7 non-responders) by qrt-pcr from peripheral blood mononuclear cells (PBMC). For AIF1 ELISA, 79 serum samples from SNUH cohort and 38 serum samples from PLANETAS study, a 4

previous phase I clinical study of CT-P13 (Clinicaltrials.org identifier NCT01220518) were used. Details of this trial have been reported previously. 27 The study was approved by the institutional review board of SNUH. 5

Figure 1. Overall study flow * Human TNF signaling pathway and human inflammatory response and autoimmunity assay were used AIF1, allograft inflammatory factor 1; LIGHT, lymphotoxins exhibiting inducible expression; qrt-pcr, quantitative real time-pcr; SNUH, Seoul National University Hospital; VEGF, vascular endothelial growth factor 6

Supplementary figure 1. Overall study flow and included subjects * Confirmation qrt-pcr of the 9 genes selected from gene assays. Two pathway focused assays, Human TNF signaling pathway and human inflammatory response and autoimmunity assay were used. LIGHT, lymphotoxins exhibiting inducible expression; qrt-pcr, quantitative real time-pcr; SNUH, Seoul national university hospital 7

2. Gene array and qrt-pcr RNA was isolated from PBMC using TRIzol reagent or Qiagen RNeasy Mini kit (Qiagen, Valencia, CA). RNA quality was assessed by Nanodrop checking the RNA concentration and purity. Complementary DNA was synthesized using GoScript reverse transcription system (Promega, Madison, WI). For screening, two pathway-focused gene expression analyses, human TNF signaling pathway and human inflammatory response and autoimmunity were conducted on 6 patients using RT2 Profiler PCR Arrays (SABiosciences, Frederick, MD). Human TNF signaling pathway analysis panel included 84 genes and inflammatory response and autoimmunity analysis panel included 370 genes (21 genes were included in both panels). Genes were selected if the difference of threshold cycle (Ct) value was more than 10% between responder and non-responder and the absolute Ct is less than 28 in at least one group. Confirmation of selected genes was done using TaqMan Gene Expression Assays (Applied Biosystems); AIF1, Hs00610419_g1; Rho GDP-dissociation inhibitor 1 (ARHGDIB), Hs00171288_m1; B-cell linker (BLNK), Hs00179459_m1; Caspase 8 (CASP8), Hs01018151_m1; Complement component 3a receptor 1 (C3AR1), Hs00269693_s1; Inhibitor of kappa light polypeptide gene enhancer in B-cell kinase beta (IKBKB), Hs01559460_m1; S100 calcium binding protein A8 (S100A8), Hs00374264_g1; Tumor necrosis 8

factor superfamily member 14 (TNFSF14), Hs00542476_g1; VEGFB, Hs00173634_m1. Ct values above 36 were considered unreliable. 28,29 Gene expression was normalized to an endogenous reference gene, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), Hs02786624_g1) and compared in responders and non-responders using ΔΔCт method. 30,31 3. ELISA Serum AIF1 levels were measured using commercial ELISA kits (Human AIF1/IBA1 CLIA kit, LSbio, Seattle, WA) according to the manufacturer s instructions. Each sample was diluted 1:1.66, analyzed in duplicate and the mean values were used for final analysis. Serum levels of lymphotoxins exhibiting inducible expression (LIGHT), leptin and VEGF were measured by multiplex ELISA using magnetic Luminex screening assay (R&D systems, Minneapolis, MN) in 20 samples from SNUH cohort. The twenty samples were randomly selected (10 responders and 10 non-responders). 4. Statistical analysis All normally and non-normally distributed continuous variables are presented as mean ± SD and median [interquartile range], respectively. 9

Baseline features and laboratory findings between the two groups were compared using Student s t-test or nonparametric Mann-Whitney test (continuous variables) or the Chi-square test (categorical variables). The relationships between serum AIF1 levels and various factors including age, BASDAI, CRP, and ESR were assessed by Spearman s rank correlation analysis. Changes of protein levels between week 0 and week 14 were tested by Wilcoxon signed rank test. Statistical tests were two sided, and p-value <0.05 was considered significant. All statistical analyses were performed using SPSS (version 22.0, IBM SPSS statistical software) and RStudio software (version 3.3.3). 10

III. Results 1. Baseline gene expression qrt-pcr was performed in 3 non-responders and 3 responders (table 1). After literature review and pathway-focused gene expression analyses, we selected 9 candidate genes (AIF1, S100A8, C3AR1, BLNK, VEGFB, ARHGDIB, CASP8, IKBKB and TNFSF14) on a basis of pre-specified criteria as described in method section and their possible involvement in arthritis or autoimmune diseases. More detailed information of these genes is demonstrated in table 2. To narrow down and confirm the most possible markers, qrt-pcr of these candidate genes were additionally performed in 18 patients (11 responders and 7 non-responders) with balanced demographic and clinical characteristics in each group except CRP and BASDAI at week 14 (table 3). CRP at week 14 was slightly higher in responder group than non-responder group (0.2 mg/dl vs. 0.0 mg/dl, p=0.044) but as the upper limit of normal range of CRP in our hospital is 0.5 mg/dl, the clinical significance may be low32,33. Comparison of baseline gene expression between the responders and non-responders are presented in table 4. In non-responders, expressions of AIF1 (fold difference 3.52), BLNK (11.03), CASP8 (7.58) and VEGFB (14.61) were significantly elevated compared to responders. As Ct values of BLNK, CASP8 and VEGFB were over 36, we selected AIF1 as the final candidate for possible biomarker. 11

Table 1. Demographic and clinical parameters of 6 ankylosing spondylitis patients with or without BASDAI 50 response * (3 each) evaluated by initial pathway focused gene assays Subject BASDAI 50 response * Sex Age TNF inhibitor BASDAI CRP ESR Week 0 Week 14 Week 0 Week 14 Week 0 Week 14 1 Yes M 56 Infliximab 9.4 4.5 0.95 0.03 29 3 2 Yes F 21 Infliximab 8.9 3.0 2.78 0.01 93 28 3 Yes F 67 Adalimumab 6.0 0.1 8.79 0.82 107 4 4 No M 64 Infliximab 6.4 5.9 1.51 0.5 20 7 5 No M 40 Adalimumab 9.4 5.9 4.80 0.37 35 7 6 No M 41 Adalimumab 8.7 7.7 5.09 1.57 90 40 * BASDAI 50 response means improvement of 50% or more in BASDAI at week 14 compared to week 0. BASDAI, Bath ankylosing spondylitis disease activity index; CRP, C-reactive protein (mg/dl); ESR, erythrocyte sedimentation rate (mm/hr) 12

Table 2. Nine genes selected from screening gene assays Gene symbol Gene name Putative function Associated disease AIF1 Allograft inflammatory factor 1 Macrophage activation Allograft rejection, RA, IBD, systemic sclerosis ARHGDIB Rho GDP-dissociation inhibitor 1 Cell apoptosis Tumor metastasis BLNK B cell linker B cell receptor signaling Unknown CASP8 Caspase 8 Cell apoptosis Asthma C3AR1 Complement component 3a receptor 1 Activation of complement system Unknown IKBKB Inhibitor of κ light polypeptide gene enhancer in B-cell kinase beta NFκB activation Unknown S100A8 S100 calcium binding protein A8 Neutrophil chemotaxis and adhesion IBD, uveitis, SpA, Experimental OA TNFSF14 Tumor necrosis factor superfamily 14 T cell activation/development RA, IBD, AS VEGFB Vascular endothelial growth factor B Unknown Hyperlipidemia and microvasculopathy in type 2 DM AS, ankylosing spondylitis; DM, diabetes mellitus; IBD, inflammatory bowel disease; OA, osteoarthritis; RA, rheumatoid arthritis; SpA, spondyloarthropathy 13

Table 3. Demographic and clinical parameters of 18 ankylosing spondylitis patients with or without BASDAI 50 response * analyzed by qrt-pcr. BASDAI 50 responders (N=11) BASDAI 50 non-responders (N=7) p-value Male, n (%) 6 (54.5) 5 (71.4) 0.826 Age (years) 39.7 ± 12.4 38.6 ± 13.4 0.854 TNF inhibitor, n (%) 0.372 - Adalimumab 2 (18.2) 3 (42.9) - Etanercept 1 (9.1) 0 (0.0) - Golimumab 2 (18.2) 0 (0.0) - Infliximab 6 (54.6) 4 (57.2) BASDAI Week 0 6.9 [5.9; 7.8] 6.0 [5.9; 7.6] 0.751 Week 14 1.6 [1.4; 2.8] 5.5 [4.5; 5.8] 0.000 ΔBASDAI -4.9 [-5.5;-4.3] -1.6 [-2.8;-0.6] 0.001 CRP (mg/dl) Week 0 2.4 [1.3; 3.8] 0.9 [0.5; 2.0] 0.258 Week 14 0.2 [0.1; 0.4] 0.0 [0.0; 0.1] 0.044 ΔCRP -2.0 [-3.0;-1.5] -0.9 [-2.0;-0.5] 0.266 ESR (mm/h) Week 0 39.0 [33.0;68.5] 19.0 [10.0;43.0] 0.135 Week 14 7.0 [4.0;33.0] 3.0 [ 3.0; 3.5] 0.197 ΔESR -25.0 [-41.0;-9.0] -16.0 [-29.0;-7.0] 0.427 Values represent mean ± standard deviation or median [IQR] * BASDAI 50 response means improvement of 50% or more in BASDAI at week 14 compared to week 0. Difference between two values, value of week 14 - week 0. BASDAI, Bath ankylosing spondylitis disease activity index; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate. 14

Table 4. Comparison of gene expression at baseline between BASDAI 50 responders * and non-responders Gene (ΔCт ) BASDAI 50 responders (N=11) BASDAI 50 non-responders (N=7) 2^(-ΔΔCт ) p-value AIF1 1.49-0.33 3.52 0.031 ARHGDIB 0.71-1.16 3.64 NS BLNK 6.13 2.67 11.03 0.003 CASP8 5.47 2.55 7.58 0.011 C3AR1 2.73 2.90 0.89 NS IKBKB 5.69 4.56 2.20 NS S100A8-1.86-2.69 1.78 NS TNFSF14 6.27 3.64 6.21 NS VEGFB 5.35 1.48 14.61 0.000 * BASDAI 50 response means improvement of 50% or more in BASDAI at week 14 baseline. Ct of the target gene Ct of the housekeeping gene (GAPDH). ΔCт of the responder ΔCт of non-responder. AIF1, allograft inflammatory factor 1; ARHGDIB, Rho GDP-dissociation inhibitor 1; BASDAI, Bath ankylosing spondylitis disease activity index; BLNK, B-cell linker; CASP8, caspase 8; C3AR1, Complement component 3a receptor 1; CRP, C-reactive protein; Ct, threshold cycle; ESR, erythrocyte sedimentation rate; IKBKB, Inhibitor of kappa light polypeptide gene enhancer in B-cell kinase beta; NS, not significant (p>0.05); S100A8, S100 calcium binding protein A8; TNFSF14, Tumor necrosis factor superfamily member 14; VEGFB, vascular endothelial growth factor B. 15

2. Baseline serum AIF1 concentration Baseline serum concentrations of AIF1 were measured using ELISA in 117 patients. The patients included 69 BASDAI 50 responders and 48 non-responders. Clinical characteristics of the patients are described in table 5. Both groups were comparable in all parameters including sex, age, baseline BASDAI and inflammatory markers. The median AIF1 level was 32.8 pg/ml [20.6;67.3] in BASDAI 50 responders and 54.2 pg/ml [28.9;91.0] in non-responders (table 6). The difference between groups was statistically significant (p=0.033). As identifying definite non-responders is the main objective of the present study, we newly defined non-responders as patients who did not satisfy BASDAI 50 response and also continued to have high disease activity (BASDAI > 5.0) at week 14. Baseline serum AIF1 levels were significantly higher in these patients than other patients (76.1 pg/ml [48.2;161.1] vs. 34.7 pg/ml [20.6;67.7], p<0.001). Receiver operating characteristics (ROC) curve of baseline serum AIF1 level for predicting high disease activity at week 14 is shown at figure 2A. The area under the curve (AUC) was 0.757 (95% C.I 0.644-0.871, p<0.001) and estimated cutoff value was 63.5 pg/ml (specificity: 71.9%, sensitivity: 66.7%). Multivariable logistic regression analysis showed that baseline AIF1 level of 63.5 pg/ml or more was associated with higher odds for high disease activity at week 14 after anti-tnfα treatment (adjusted OR 6.953, p=0.002) (table 7). The AUC 16

of the final regression model was 0.822 (95% CI 0.714-0.929, p<0.001, figure 2B). Among 117 patients, ASAS response could be evaluated in 38 patients. Twenty-six patients were ASAS 20 responders and 18 were ASAS 40 responders (table 5). ASAS 20 responders and ASAS 40 responders had significantly lower AIF1 levels at baseline than non-responders (table 6). Also, ASAS 40 responder group had lower level of baseline AIF1 than ASAS 20 responder group. These results suggest the dose relationship between serum AIF1 level and response to TNFα blockers in AS. The correlations between baseline serum concentration of AIF1 and clinical variables at week 0 are shown at table 8. Baseline serum AIF1 level had weak correlation with baseline BASDAI (rho=0.241, p=0.009) and no linear relationship with CRP or ESR, which are inflammatory markers, and this result indicates that AIF1 may act as a different mechanism from currently known inflammatory pathways in ankylosing spondylitis. 17

Table 5. Demographic and clinical characteristics of ankylosing spondylitis patients included in serum AIF1 ELISA. BASDAI 50 responders * (N=69) BASDAI 50 non-responders (N=48) p-value Male, n (%) 55 (79.7) 39 (81.2) 1.000 Age (years) 40.3 ± 12.8 43.7 ± 11.6 0.145 TNF inhibitor, n (%) 0.536 - Adalimumab 16 (23.2) 6 (12.5) - Etanercept 6 (8.7) 5 (10.4) - Golimumab 5 (7.2) 2 (4.2) - Infliximab 42 (60.9) 35 (72.9) BASDAI Week 0 6.6 [5.4; 8.0] 6.5 [5.2; 7.8] 0.692 Week 14 2.0 [1.6; 2.8] 4.6 [3.6; 6.0] 0.000 ΔBASDAI -4.4 [-5.4;-3.6] -1.8 [-2.7;-0.9] 0.000 CRP (mg/dl) Week 0 2.8 [1.2; 6.7] 2.5 [0.9;18.1] 0.920 Week 14 0.2 [0.0; 0.8] 0.6 [0.1; 2.2] 0.008 ΔCRP -2.6 [-6.3;-0.5] -2.2 [-13.8;-0.5] 0.932 ESR (mm/h) Week 0 36.0 [26.0;64.0] 34.5 [19.0;42.0] 0.108 Week 14 6.5 [3.0;14.0] 10.5 [4.0;22.0] 0.158 ΔESR -25.0 [-52.0;-14.0] -19.0 [-29.0;-7.0] 0.044 ASAS20 responder, n (%) 16/16 (100) 10/22 (45.5) 0.001 ASAS40 responder, n (%) 16/16 (100) 2/22 (9.1) 0.000 Values represent mean ± standard deviation or median [IQR] * BASDAI 50 response means improvement of 50% or more in BASDAI at week 14 from baseline. Difference between two values, value of week 14 - week 0. ASAS response data were available in 38 patients. ASAS, assessment in ankylosing spondylitis; BASDAI, Bath ankylosing spondylitis disease activity index; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate. 18

Table 6. Comparison of baseline AIF1 levels in ankylosing spondylitis patients receiving anti-tnfα agents according to several response criteria (BASDAI 50, ASAS 20 and ASAS 40). Responders Non-responders p-value BASDAI 50 * 32.8 [20.6;67.3] (n=69) ASAS 20 53.8 [23.9;75.0] (n=26) ASAS 40 28.9 [20.6;68.1] (n=18) 54.2 [28.9;91.0] (n=48) 101.2 [63.5;163.2] (n=12) 79.2 [53.8;149.2] (n=20) 0.033 0.005 0.005 * BASDAI 50 response means improvement of 50% or more in BASDAI at week 14 from baseline. ASAS response data were available in 38 patients. ASAS, assessment in ankylosing spondylitis; BASDAI, Bath ankylosing spondylitis disease activity index; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; NR, non-responder; R, responder. 19

Figure 2. ROC curve of baseline serum AIF1 level (A) and final multivariate logistic regression (B) for predicting high disease activity * at week 14 *High disease activity at week 14 was defined as BASDAI > 5.0. AIF1, allograft inflammatory factor I; AUC, area under the curve; C.I, confidence interval; ROC, receiver operating characteristics 20

Table 7. Logistic regression analysis for predicting high disease activity * after anti-tnfα treatment for 14 weeks. Serum level of AIF1 Unadjusted OR (95% C.I) p-value Adjusted OR (95% C.I) p-value < 63.5 pg/ml 1.0 0.002 1.0 0.002 63.5 pg/ml 5.111 (1.861, 14.040) 6.953 (2.046, 23.629) *High disease activity is defined as BASDAI > 5.0. adjusted for age, initial BASDAI, initial CRP, initial ESR. AIF1, allograft inflammatory factor 1; C.I, confidence interval 21

Table 8. Univariate analysis for correlation between baseline AIF1 levels and clinical variables Rho * p-value Age (years) -0.007 NS BASDAI at week 0 0.241 0.009 CRP at week 0 0.139 NS ESR at week 0 0.061 NS * Spearman s correlation coefficient. BASDAI, Bath ankylosing spondylitis disease activity index; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; NS, not significant 22

3. AIF1 concentration after anti-tnfα treatment We also measured serum AIF1 levels at 14 weeks after anti-tnfα treatment and analyzed the changes before and after the treatment. Serum AIF1 levels were analyzed in 55 samples selected randomly among 117 samples. There were no significant differences in AIF1 concentration before and after treatment in both groups (Wilcoxon signed rank test, figure 3). This result indicates that blocking TNFα does not affect serum concentration of AIF1. 23

Figure 3. Changes of serum AIF1 concentration in ankylosing spondylitis patients after anti-tnf treatment according to BASDAI 50 response * at week 14 AIF1 (pg/ml) 500 400 300 200 100 (A) BASDAI 50 responder (n=32) p=0.729 AIF1 (pg/ml) 800 600 400 200 (B) BASDAI 50 non-responder (n=22) p=0.071 0 Week 0 Week 14 0 Week 0 Week 14 * BASDAI 50 response means improvement of 50% or more in BASDAI at week 14 from baseline AIF1, allograft inflammatory factor 1 24

4. Serum concentration of LIGHT, VEGF and leptin LIGHT was previously reported to correlate with inflammatory markers, and VEGF and leptin were suggested as predictors of BASDAI 50 response. 15,34 We additionally measured baseline serum concentration of LIGHT, VEGF and leptin by multiplex ELISA in 20 patients (supplementary table 1). Unlike previous report, serum LIGHT 34 and VEGF 35 were not correlated with baseline CRP or ESR. Only leptin had inverse correlation with CRP (rho=-0.6, p=0.007). Also, there were no significant differences of these serum protein levels between responders and non-responders; LIGHT, 186.0 pg/ml [138;239] vs. 157.0 pg/ml [88.1;194.0]; VEGF, 111.0 pg/ml [80.7;200.0] vs. 142 pg/ml [49.3;224.0]; leptin 3665.6 [2410.1;5922.9] pg/ml vs. 2810.1 [2305.0;4515.4] pg/ml (figure 4). Among 20 patients, 14 patients (8 responders and 6 non-responders) were included in AIF1 ELISA and in these patients, baseline AIF1 concentrations were numerically higher in non-responders but there was no statistical significance (32.0 pg/ml [15.5;53.0] in responders and 48.6 pg/ml [34.0;268.0] in non-responders, p=0.142). However, as sample size is not large enough in this analysis, we cannot conclude that baseline levels of LIGHT, VEGF and leptin are not associated with clinical response to TNFα blockers. After anti-tnfα treatment, serum levels of LIGHT significantly decreased in responders but not in non-responders (figure 5). Serum levels of VEGF were 25

decreased after anti-tnfα treatment in both groups, while leptin showed no changes. 26

Supplementary table 1. Demographic and clinical characteristics of ankylosing spondylitis patients analyzed by multiplex ELISA. BASDAI 50 Responders * (N=10) BASDAI 50 Non-responders (N=10) p-value Male, n (%) 6 (60) 6 (70) 1.000 Age (years) 40.3 ± 12.4 43.2 ± 15.2 0.639 TNF inhibitor, n (%) 0.196 - Adalimumab 2 (20.0) 5 (50) - Etanercept 0 (0.0) 1 (10.0) - Golimumab 2 (20.0) 0 (0.0) - Infliximab 6 (60.0) 4 (40.0) BASDAI Week 0 7.1 [6.3;8.2] 7.7 [6.3;9.1] 0.481 Week 14 1.8 [1.03;3.0] 5.7 [5.2;5.9] 0.000 ΔBASDAI -1.8 [-3.2;-0.5] -5.5 [-6.0;-4.7] 0.000 CRP (mg/dl) Week 0 3.0 [1.9;4.5] 1.0 [0.4; 2.9] 0.198 Week 14 0.1 [0.1;0.2] 0.2 [0.0; 0.8] 0.706 ΔCRP -1.0 [-2.8;-0.4] -2.8 [-3.5;-0.9] 0.327 ESR (mm/h) Week 0 49.5 [31.0;77.0] 19.5 [11.0;33.0] 0.089 Week 14 6.0 [4.0; 9.0] 4.0 [3.0;15.0] 0.967 ΔESR -17.0 [-28.0;-8.0] -51.0 [-68.0;-25.0] 0.027 Values represent mean ± standard deviation or median [IQR]. * BASDAI 50 response means improvement of 50% or more in BASDAI at week 14 from baseline. Difference between two values, value of week 14 - week 0. BASDAI, Bath ankylosing spondylitis disease activity index; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate. 27

Figure 4. Comparison of baseline serum concentration of LIGHT, VEGF and leptin between BASDAI 50 responders * (n=10) and non-responders (n=10). * BASDAI 50 response means improvement of 50% or more in BASDAI at week 14 from baseline. LIGHT, lymphotoxins exhibiting inducible expression; VEGF, vascular endothelial growth factor 28

Figure 5. Changes of serum levels of LIGHT, VEGF and leptin after anti- TNF treatment in BASDAI 50 responders * (n=10) and non-responders (n=10). * BASDAI 50 response means improvement of 50% or more in BASDAI at week 14 from baseline. LIGHT, lymphotoxins exhibiting inducible expression; VEGF, vascular endothelial growth factor 29

IV. Discussion In this study, we showed that gene AIF1 had significantly higher expression levels at baseline in the BASDAI 50 non-responder group than in the responder group. Also, baseline serum AIF1 level was significantly higher in BASDAI 50 non-responder group and high serum level of AIF1 was predictive of high disease activity after anti-tnfα treatment. The result was consistent in ASAS responders and nonresponders. In addition, serum AIF1 level did not change after anti-tnfα treatment in both groups. These results suggest that high AIF1 level is associated with poor response to TNFα blockers. In contrast to RA, in which AIF1 has been suggested to be closely associated with the pathogenesis 22,36,37, the role of AIF1 in AS remains uncertain. In the previous study of RA patients, synovial cell proliferation and IL-6 production were significantly increased after treatment with human recombinant AIF1 to RA synoviocytes, while no effect was seen on IL-1β and TNFα production. 22 This result indicates that AIF1 may have a critical role in RA pathogenesis by increasing IL- 6 secretion via IL-1β and TNFα independent pathway. In AS patients, IL- 6 was found to be increased in the serum and sacroiliac joints 38 and was correlated with spinal inflammation detected by magnetic resonance image. 39 Although antagonizing IL-6 was not effective in AS patients 40, IL-6 may have role in inflammatory process of AS and AIF1 may affect 30

IL-6 in a TNFα independent pathway. Another in vitro study about AIF1 and systemic sclerosis reported that AIF-1 isoform 2 up-regulated the expression of IL-4 and IL-17 mrna and increased the production of both cytokines in supernatants of experimental T cells, but not the levels of interferon-γ and TNFα mrna. 41 However, there is also a report that stimulation of macrophage cell lines with proinflammatory cytokines IL- 1β and TNFα increased the expression of AIF-1 and sodium salicylate down-regulated the endogenous level of AIF1 and also reversed the upregulation of AIF1 which was induced by IL-1β or TNFα. 42 Furthermore, infliximab therapy reversed the increase of AIF1 in serum and colonic mucosa in rats with inflammatory bowel disease and AIF1 was suggested as a non-invasive marker of monitoring intestinal inflammation. 43 Taken together, these results suggests that the relationship between AIF1 and other inflammatory cytokines remains uncertain. There is evidence that AS patients have manifestations of neuropathic pain. 44 Studies of sural nerve biopsy and epidermal nerve fiber density revealed association of fibromyalgia and demyelination of peripheral nerve system. 45,46 Interestingly, DNA microarray analysis of nerve biopsies showed upregulation of AIF1 in inflammatory nerve disease such as chronic inflammatory demyelinating polyneuropathy and vasculitic neuropathy. 47-49 In our study, patients who showed no improvement in pain or BASDAI, had high levels of AIF1 and the levels 31

did not change after blocking TNFα. These results suggest that AIF1 may have a role in pain response in AS patients. Comparing serum AIF1 levels in NSAIDs responders and non-responders in AS or measuring in fibromyalgia patients might give more insight to this theory. Serum level of VEGF decreased in both groups after TNFα inhibition and this result was consistent with previous reports. 39 Gene LIGHT was down regulated by infliximab and serum LIGHT correlated well with the changes in inflammatory markers. 34 However, in our study, serum LIGHT levels were decreased only in BASDAI responders after anti- TNFα treatment. Also, serum leptin level was not associated with BASDAI response or disease activity contrary to previous studies. 15,17 These results need to be validated in larger study population. We reviewed AS patients who started anti-tnfα treatment at SNUH from 2010 to 2015, and 121 patients had information including BASDAI, ESR and CRP at baseline and 14 weeks after treatment. Majority of these patients had elevated ESR and CRP before starting anti-tnfα therapy and they were decreased to normal range after the treatment; ESR in 94 patients (77.7%), CRP in 103 patients (85.1%). In contrast, among 25 patients who had less than 50% improvement of BASDAI, only 6 patients had elevated ESR and 2 patients had elevated CRP at week 14. Because anti-tnfα agents are basically anti-inflammatory drug, 32

systemic inflammation, represented by ESR and CRP, can return to normal range despite an inadequate clinical improvement. Also, previous study demonstrated that inflammatory markers do not necessarily reflect disease activity in AS. 40,50-52 Based on these backgrounds, we examined clinical response to TNFα blockers with BASDAI 50. Our study has some limitations. First, we had limited data of response measurements with objective components such as ASDAS. However, when we analyzed patients with ASAS response data, there was dose response relationship between serum AIF1 level and ASAS 20 and ASAS 40 response with statistical significance. Second, we did not assess the level of inflammatory cytokine such as TNFα, IL-6 and IL-17 and therefore, possible mechanism by which AIF-1 regulate the inflammation in AS could not be shown in this study. Third, we did not assess the relationship between the AIF level and radiographic progression in AS. We found consistent results across RNA profile to serum protein level. Also, we focused on baseline differences of gene expression and protein levels. Most of the previous studies targeted markers that showed changes after anti-tnfα treatment. Comparing pre- and post-treatment gene expression or serum proteins can give insights to the pathogenesis of the disease and clues for developing biomarkers. Even if the level of 33

serum marker does not change after treatment, the basal level still can be predictive of treatment outcome. However, we think that this result is not confirmative and should be replicated in larger studies. In conclusion, we showed that serum AIF1 level was higher in anti- TNFα non-responders and was predictive for high disease activity after anti-tnfα treatments. This could be a novel serum marker predicting responses to anti-tnfα agents in AS. Further studies are needed to understand the molecular mechanism of AIF1. 34

V. References 1. Raychaudhuri SP, Deodhar A. The classification and diagnostic criteria of ankylosing spondylitis. J Autoimmun. 2014;48-49(Supplement C):128-133. 2. Brown MA, Kenna T, Wordsworth BP. Genetics of ankylosing spondylitis--insights into pathogenesis. Nat Rev Rheumatol. 2016;12(2):81-91. 3. Braun J, van den Berg R, Baraliakos X, et al. 2010 update of the ASAS/EULAR recommendations for the management of ankylosing spondylitis. Ann Rheum Dis. 2011;70(6):896-904. 4. Escalas C, Trijau S, Dougados M. Evaluation of the treatment effect of NSAIDs/TNF blockers according to different domains in ankylosing spondylitis: results of a meta-analysis. Rheumatology (Oxford). 2010;49(7):1317-1325. 5. Braun J, Brandt J, Listing J, et al. Long-term efficacy and safety of infliximab in the treatment of ankylosing spondylitis: an open, observational, extension study of a three-month, randomized, placebo-controlled trial. Arthritis Rheum. 2003;48(8):2224-2233. 6. van der Heijde D, Dijkmans B, Geusens P, et al. Efficacy and safety of infliximab in patients with ankylosing spondylitis: results of a randomized, placebo-controlled trial (ASSERT). Arthritis Rheum. 2005;52(2):582-591. 35

7. Kiltz U, Heldmann F, Baraliakos X, Braun J. Treatment of ankylosing spondylitis in patients refractory to TNF-inhibition: are there alternatives? Curr Opin Rheumatol. 2012;24(3):252-260. 8. Braun J, Sieper J. What is the most important outcome parameter in ankylosing spondylitis? Rheumatology (Oxford). 2008;47(12):1738-1740. 9. Mansour M, Cheema GS, Naguwa SM, et al. Ankylosing spondylitis: a contemporary perspective on diagnosis and treatment. Semin Arthritis Rheumatism. 2007;36(4):210-223. 10. McLeod C, Bagust A, Boland A, et al. Adalimumab, etanercept and infliximab for the treatment of ankylosing spondylitis: a systematic review and economic evaluation. Health Technol Assess (Winchester, England). 2007;11(28):1-158, iii-iv. 11. Zavada J, Uher M, Sisol K, et al. A tailored approach to reduce dose of anti-tnf drugs may be equally effective, but substantially less costly than standard dosing in patients with ankylosing spondylitis over 1 year: a propensity score-matched cohort study. Ann Rheum Dis. 2016;75(1):96-102. 12. Tong Q, Cai Q, de Mooij T, et al. Adverse events of anti-tumor necrosis factor alpha therapy in ankylosing spondylitis. PloS one. 2015;10(3):e0119897. 36

13. Sampaio-Barros PD, van der Horst-Bruinsma IE. Adverse effects of TNF inhibitors in SpA: Are they different from RA? Best Pract Res Clin Rheumatol. 2014;28(5):747-763. 14. de Vries MK, van Eijk IC, van der Horst-Bruinsma IE, et al. Erythrocyte sedimentation rate, C-reactive protein level, and serum amyloid a protein for patient selection and monitoring of anti-tumor necrosis factor treatment in ankylosing spondylitis. Arthritis Rheum. 2009;61(11):1484-1490. 15. Wagner C, Visvanathan S, Braun J, et al. Serum markers associated with clinical improvement in patients with ankylosing spondylitis treated with golimumab. Ann Rheum Dis. 2012;71(5):674-680. 16. Gerard S, le Goff B, Maugars Y, Berthelot JM. Six-month response to anti-tnf drugs in axial spondylarthropathy according to the fulfillment or not of New-York criteria for ankylosing spondylitis or French recommendations for anti-tnf use. A "real life" retrospective study on 175 patients. Joint Bone Spine. 2008;75(6):680-687. 17. Park MC, Lee SW, Choi ST, Park YB, Lee SK. Serum leptin levels correlate with interleukin-6 levels and disease activity in patients with ankylosing spondylitis. Scand J Rheumatol. 2007;36(2):101-106. 37

18. Toussirot E, Streit G, Nguyen NU, et al. Adipose tissue, serum adipokines, and ghrelin in patients with ankylosing spondylitis. Metabolism. 2007;56(10):1383-1389. 19. Zhao YY, Yan DJ, Chen ZW. Role of AIF-1 in the regulation of inflammatory activation and diverse disease processes. Cell immunol. 2013;284(1-2):75-83. 20. Utans U, Arceci RJ, Yamashita Y, Russell ME. Cloning and characterization of allograft inflammatory factor-1: a novel macrophage factor identified in rat cardiac allografts with chronic rejection. The J Clin Invest. 1995;95(6):2954-2962. 21. Iris FJ, Bougueleret L, Prieur S, et al. Dense Alu clustering and a potential new member of the NF kappa B family within a 90 kilobase HLA class III segment. Nat Genet. 1993;3(2):137-145. 22. Kimura M, Kawahito Y, Obayashi H, et al. A critical role for allograft inflammatory factor-1 in the pathogenesis of rheumatoid arthritis. J Immunol. 2007;178(5):3316-3322. 23. Chinnasamy P, Lutz SE, Riascos-Bernal DF, et al. Loss of Allograft Inflammatory Factor-1 Ameliorates Experimental Autoimmune Encephalomyelitis by Limiting Encephalitogenic CD4 T-Cell Expansion. Mol Med. 2015;21:233-241. 24. van der Linden S, Valkenburg HA, Cats A. Evaluation of diagnostic criteria for ankylosing spondylitis. A proposal for 38

modification of the New York criteria. Arthritis Rheumatol. 1984;27(4):361-368. 25. Garrett S, Jenkinson T, Kennedy LG, Whitelock H, Gaisford P, Calin A. A new approach to defining disease status in ankylosing spondylitis: the Bath Ankylosing Spondylitis Disease Activity Index. The J Rheumatol. 1994;21(12):2286-2291. 26. Rudwaleit M, Listing J, Brandt J, Braun J, Sieper J. Prediction of a major clinical response (BASDAI 50) to tumour necrosis factor alpha blockers in ankylosing spondylitis. Ann Rheum Dis. 2004;63(6):665-670. 27. Park W, Hrycaj P, Jeka S, et al. A randomised, double-blind, multicentre, parallel-group, prospective study comparing the pharmacokinetics, safety, and efficacy of CT-P13 and innovator infliximab in patients with ankylosing spondylitis: the PLANETAS study. Ann Rheum Dis. 2013. 28. Bustin SA, Benes V, Garson JA, et al. The MIQE Guidelines: Minimum Information for Publication of Quantitative Real-Time PCR Experiments. Clin Chem. 2009;55(4):611-622. 29. Burns M, Valdivia H. Modelling the limit of detection in realtime quantitative PCR. Eur Food Res Technol. 2008;226(6):1513-1524. 30. Schmittgen TD, Livak KJ. Analyzing real-time PCR data by the 39

comparative CT method. Nat Protocols. 2008;3(6):1101-1108. 31. Pfaffl MW. A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res. 2001;29(9):e45. 32. Kiltz U, Baraliakos X, Karakostas P, et al. Do patients with nonradiographic axial spondylarthritis differ from patients with ankylosing spondylitis? Arthritis Care Res. 2012;64(9):1415-1422. 33. Vastesaeger N, van der Heijde D, Inman RD, et al. Predicting the outcome of ankylosing spondylitis therapy. Ann Rheum Dis. 2011;70(6):973-981. 34. Haroon N, Tsui FW, O'Shea FD, et al. From gene expression to serum proteins: biomarker discovery in ankylosing spondylitis. Ann Rheum Dis. 2010;69(1):297-300. 35. Danve A, O'Dell J. The ongoing quest for biomarkers in Ankylosing Spondylitis. Int J Rheum Dis. 2015;18(8):826-834. 36. Pawlik A, Kotrych D, Paczkowska E, et al. Expression of allograft inflammatory factor-1 in peripheral blood monocytes and synovial membranes in patients with rheumatoid arthritis. Hum Immunol. 2016;77(1):131-136. 37. Harney SM, Vilarino-Guell C, Adamopoulos IE, et al. Fine mapping of the MHC Class III region demonstrates association of AIF1 and rheumatoid arthritis. Rheumatology (Oxford). 40

2008;47(12):1761-1767. 38. Gratacos J, Collado A, Filella X, et al. Serum cytokines (IL-6, TNF-alpha, IL-1 beta and IFN-gamma) in ankylosing spondylitis: a close correlation between serum IL-6 and disease activity and severity. Br J Rheumatol. 1994;33(10):927-931. 39. Visvanathan S, Wagner C, Marini JC, et al. Inflammatory biomarkers, disease activity and spinal disease measures in patients with ankylosing spondylitis after treatment with infliximab. Ann Rheum Dis. 2008;67(4):511-517. 40. Sieper J, Porter-Brown B, Thompson L, Harari O, Dougados M. Assessment of short-term symptomatic efficacy of tocilizumab in ankylosing spondylitis: results of randomised, placebocontrolled trials. Ann Rheum Dis. 2014;73(1):95-100. 41. Del Galdo F, Jimenez SA. T cells expressing allograft inflammatory factor 1 display increased chemotaxis and induce a profibrotic phenotype in normal fibroblasts in vitro. Arthritis Rheum. 2007;56(10):3478-3488. 42. Yang ZF, Ho DW, Lau CK, et al. Allograft inflammatory factor- 1 (AIF-1) is crucial for the survival and pro-inflammatory activity of macrophages. Int Immunol. 2005;17(11):1391-1397. 43. Roman ID, Cano-Martinez D, Lobo MV, et al. Infliximab therapy reverses the increase of allograft inflammatory factor-1 in serum 41

and colonic mucosa of rats with inflammatory bowel disease. Biomarkers. 2017;22(2):133-144. 44. Wu Q, Inman RD, Davis KD. Neuropathic pain in ankylosing spondylitis: a psychophysics and brain imaging study. Arthritis Rheum. 2013;65(6):1494-1503. 45. Caro XJ, Winter EF, Dumas AJ. A subset of fibromyalgia patients have findings suggestive of chronic inflammatory demyelinating polyneuropathy and appear to respond to IVIg. Rheumatology (Oxford). 2008;47(2):208-211. 46. Caro XJ, Winter EF. Evidence of abnormal epidermal nerve fiber density in fibromyalgia: clinical and immunologic implications. Arthritis Rheumatol. 2014;66(7):1945-1954. 47. Kinter J, Broglio L, Steck AJ, et al. Gene expression profiling in nerve biopsy of vasculitic neuropathy. J Neuroimmunol. 2010;225(1-2):184-189. 48. Renaud S, Hays AP, Brannagan TH, 3rd, et al. Gene expression profiling in chronic inflammatory demyelinating polyneuropathy. J Neuroimmunol. 2005;159(1-2):203-214. 49. Steck AJ, Kinter J, Renaud S. Differential gene expression in nerve biopsies of inflammatory neuropathies. J Peripher Nerv Syst. 2011;16 Suppl 1:30-33. 50. Ruof J, Stucki G. Validity aspects of erythrocyte sedimentation 42

rate and C-reactive protein in ankylosing spondylitis: a literature review. Journal Rheumatol. 1999;26(4):966-970. 51. Claushuis TA, de Vries MK, van der Weijden MA, et al. C- reactive protein polymorphisms influence serum CRP-levels independent of disease activity in ankylosing spondylitis. Clin Exp Rheumatol. 2015;33(2):159-165. 52. Chen C-H, Chen H-A, Liao H-T, Liu C-H, Tsai C-Y, Chou C-T. The clinical usefulness of ESR, CRP, and disease duration in ankylosing spondylitis: the product of these acute-phase reactants and disease duration is associated with patient s poor physical mobility. Rheumatol Int. 2015;35(7):1263-1267. 43

RNA pathway focused gene assay qrt- PCR ELISA BASDAI Gene assay qrt-pcr AIF1 p AIF1 pg/ml [IQR 20.6;67.3], pg/ml [28.9;91.0] p AIF1 pg/ml BASDAI 44

p AIF1 BASDAI AIF1 BASDAI AIF1 AIF1 BASDAI 45