PLoS One. 2015 Jul 15;10(7):e0132055.
Pretreatment Prediction of Individual Rheumatoid Arthritis Patients’ Response to Anti-Cytokine Therapy Using Serum Cytokine/Chemokine/Soluble Receptor Biomarkers.
- 1Department of Organic Fine Chemicals, Division of Biological and Molecular Sciences, The Institute of Scientific and Industrial Research, Osaka University, Suita, Osaka, Japan; Louis Pasteur Center for Medical Research, Kyoto, Kyoto, Japan.
- 2Department of Organic Fine Chemicals, Division of Biological and Molecular Sciences, The Institute of Scientific and Industrial Research, Osaka University, Suita, Osaka, Japan; Tokushukai Medical Corporation, Tokyo, Japan.
- 3Higashi Hiroshima Memorial Hospital, Higashi Hiroshima, Hiroshima, Japan.
- 4Department of Organic Fine Chemicals, Division of Biological and Molecular Sciences, The Institute of Scientific and Industrial Research, Osaka University, Suita, Osaka, Japan.
- 5Louis Pasteur Center for Medical Research, Kyoto, Kyoto, Japan.
The inability to match rheumatoid arthritis (RA) patients with the anti-cytokine agent most efficacious for them is a major hindrance to patients’ speedy recovery and to the clinical use of anti-cytokine therapy. Identifying predictive biomarkers that can assist in matching RA patients with more suitable anti-cytokine treatment was our aim in this report. The sample consisted of 138 RA patients (naïve and non-naïve) who were administered tocilizumab or etanercept for a minimum of 16 weeks as a prescribed RA treatment. Pretreatment serum samples were obtained from patients and clinical measures of their disease activity were evaluated at baseline and 16 weeks after treatment commenced. Using patients’ pretreatment serum, we measured 31 cytokines/chemokines/soluble receptors and used multiple linear regression analysis to identify biomarkers that correlated with patients’ symptom levels (DAS28-CRP score) at week 16 and multiple logistic analyses for biomarkers that correlated with patients’ final outcome. The results revealed that sgp130, logIL-6, logIL-8, logEotaxin, logIP-10, logVEGF, logsTNFR-I and logsTNFR-II pretreatment serum levels were predictive of the week 16 DAS28-CRP score in naïve tocilizumab patients while sgp130, logGM-CSF and logIP-10 were predictive in non-naïve patients. Additionally, we found logIL-9, logVEGF and logTNF-α to be less reliable at predicting the week 16 DAS28-CRP score in naïve etanercept patients. Multiple linear regression and multiple logistic regression analyses identified biomarkers that were predictive of remission/non-remission in tocilizumab and etanercept therapy. Although less reliable than those for tocilizumab, we identified a few possible biomarkers for etanercept therapy. The biomarkers for these two therapies differ suggesting that their efficacy will vary for individual patients. We discovered biomarkers in RA pretreatment serum that predicted their week 16 DAS28-CRP score and clinical outcome to tocilizumab therapy. Most of these biomarkers, especially sgp130, are involved in RA pathogenesis and IL-6 signal transduction, which further suggests that they are highly reliable.
TRIAL REGISTRATION: UMIN-CTR Clinical Trial UMIN000016298.
In clinical practice it has been noted that each anti-rheumatic therapy delivers a different outcome for individual RA patients making it difficult to prescribe them the most efficacious treatment. It is critical to identify molecular biomarkers that can predict patient response to anti-TNF-a or anti-IL-6 based therapies before patients are treated so that non-effective therapies can be quickly eliminated and doctors can prescribe the cytokine therapy that is most efficacious for each RA patient at an earlier stage (1). However, as shown in Figure 1, there are several underlying problems that make this task challenging. We believe that identifying reliable predictive biomarkers will make it easier to follow EULAR’s treat-to-target recommendation by allowing clinicians to know in advance if a selected therapy will achieve the treatment goal (target) that has been pre-determined for each RA patient (2).
In this retrospective observational study, 138 of RA patients were divided into 47 of non-naïve (previous history of anti-cytokine therapy) and 91 of naïve (no history of anti-cytokine therapy) groups. We used a Luminix beads based array method to measure and analyze cytokines, chemokines, and soluble receptors in RA patient’s pretreatment serum. Patients were treated for 16 weeks with either tocilizumab or etanercept. We analyzed samples and data for this cohort and used multiple linear regression analysis to reveal biomarkers that predicted RA patients’ week 16 DAS28-CRP score to tocilizumab or etanercept therapy. Multiple logistic regression analysis revealed whether or not patients would achieve remission at week 16. The details of patients and methods, study procedures, biomarker assay system and statistical analysis are shown in the text of the original paper.
We attempted to predict the week 16 DAS28-CRP score for RA patients using their pretreatment cytokine/chemokine/soluble receptor data. To find biomarkers that may have contributed to the week 16 DAS28-CRP, multiple linear regression analysis of cytokine/chemokine/soluble receptor levels was performed to determine the best equation of DAS28-CRP improvement. We found that sgp130, logIL-6, logIL-8, logEotaxin, logIP-10, logVEGF, logsTNFR-I and logsTNFR-II values were significantly expressed in naïve tocilizumab patients (R²=0.646, p<0.0001). In non-naïve tocilizumab patients, we observed that sgp130, logGM-CSF and logIP-10 values were possible predictive biomarkers (R²=0.486, p<0.0001) (Table1, A). A comparison of multiple linear regression analysis for naïve patients in the MTX with tocilizumab treated group and the tocilizumamb only treated group showed similar tendencies. Although the R² value was not sufficiently high (albeit significant), we observed that logIL-9, logTNF-α and logVEGF values were possible predictive biomarkers in naïve etanercept patients (R²=0.247, p=0.0107) (Table1, A). The predictive biomarkers we identified for tocilizumab and etanercept therapy are quite different, suggesting that the therapeutic mechanism of each anti-cytokine agent is different, therefore using these biomarkers to determine which treatment will be more effective for each patient can deliver great benefits for the patients involved. Multiple logistic regression analysis was used to determine multivariable models as predictive biomarkers of remission and non-remission based on baseline cytokine/chemokine/soluble receptor levels in biologic naïve tocilizumab patients. The best combination of predictive markers is shown in Table1, B. The data strongly suggests that sgp130, logIL-6, logIP-10, and logsTNFR-II values are potential predictive biomarkers in both naïve (p=0.0004, AUC=0.850) and non-naïve patients (p=0.002, AUC=0.892 ) (Table1, B). This process is summarized in Figure 2.
We believe that the predictive biomarkers we identified through quantifying cytokines/chemokines/soluble receptors are more practical and useful than gene analysis. Determining pretreatment serum biomarkers for individual RA patients in this way allows for more targeted treatments that will deliver better outcomes for patients.
Why is IL-6 predictive in tocilizumab patients? In the IL-6 receptor system, sIL-6R increases in the presence of inflammation. IL-6 and sIL-6R complex activates a wide variety of cells that express gp130 on their cell membrane in inflammatory conditions suggesting that sIL-6R and related soluble receptors contribute to IL-6 signaling and may be predictive markers for patient outcome to tocilizumab therapy. This may explain our observation that patients with higher sgp130 levels, a natural IL-6 inhibitor and a key predictive biomarker, were prone to experience higher clinical efficacy to tocilizumab therapy (Figure 3)(3,4). Therefore we believe that in order to find beneficial clinical parameters, it is necessary to analyze the signal pathways of cytokines and their soluble receptors related to the pathogenesis of RA (5,6).
We believe that it is an important finding that the predictive biomarkers we identified for etanercept therapy differed from those for tocilizumab. These results further prove that individual patients tend to have a different response to each agent prescribed for the same disease condition and will discriminately react to some treatments more favorable than others. Finding more reliable markers is essential so that etanercept therapy can be used more efficiently and effectively.
Our ultimate aim is to develop a kit that helps clinicians to choose the anti-cytokine agent that is most suitable for individual RA patients before administering treatment.
These biomarkers may assist doctors to identify in advance patients who will not respond favorably to a treatment protocol thereby sparing patients from being treated with expensive and powerful agents that are not efficacious for them. It also allows individual RA patients to be matched with the anti-cytokine therapy that will be most effective for them or that will allow them to achieve their treatment target. This type of treatment strategy is in line with the personalized therapy now recommended by EULAR in RA field (7,8,9). Our recommended patient outcome prediction process is presented in Figure 4.
Importance of the study: This treatment strategy is line with the treat-to-target recommendation of EULAR and would facilitate personalized clinical treatment of RA patients which would be more effective (10). We believe our report is a critical first step and moving forward an interventional prospective study with a larger cohort should be utilized to confirm the predictive biomarkers that we have identified. In the meantime, examining these cytokine/chemokine and soluble receptor biomarkers before treating patients with biologic therapy is a pursuit that could be highly beneficial for RA patients and the doctors who treat them.
(1) Soken-Nakazawa J S, Yoshizaki K (2015) Tocilizumab for treating rheumatoid arthritis: an evaluation of pharmacokinetics/pharmacodynamics and clinical efficacy. Expert Opin Drug Metab Toxicol 11: 307-316.
(2) Smolen JS, Landewe R, Breedveld FC, Buch M, Burmester G, et al. (2013) EULAR recommendations for the management of rheumatoid arthritis with synthetic and biological disease-modifying antirheumatic drugs: 2013 update. Ann Rheum Dis.
(3) Nowell MA, Richards PJ, Horiuchi S, Yamamoto N, Rose-John S, Topley, N, et al. (2003) Soluble IL-6 receptor governs IL-6 activity in experimental arthritis: blockade of arthritis severity by soluble glycoprotein 130. J Immunol 171: 3202-3209.
(4) Rose-John S (2012) IL-6 trans-signaling via the soluble IL-6 receptor: importance for the pro-inflammatory activities of IL-6. Int J Biol Sci 8: 1237-1247.
(5) Nishikawa T, Hagihara K, Serada S, Isobe T, Matsumura A, Song, J, et al. (2008) Transcriptional complex formation of c-Fos, STAT3, and hepatocyte NF-1 alpha is essential for cytokine-driven C-reactive protein gene expression. J Immunol 180: 3492-3501.
(6) Song SN, Iwahashi M, Tomosugi N, Uno K, Yamana J, Yamana S, Isobe T, Ito H, Kawabata H, Yoshizaki K (2013) Comparative evaluation of the effects of treatment with tocilizumab and TNF-α inhibitors on serum hepcidin, anemia response and disease activity in rheumatoid arthritis patients. Arthritis Res Ther 15: R141.
(7) Lindstrom TM, Robinson WH (2010) Biomarkers for rheumatoid arthritis: making it personal. Scand J Clin Lab Invest Suppl 242: 79-84.
(8) Conaghan PG (2011) Predicting outcomes in rheumatoid arthritis. Clin Rheumatol 30 Suppl 1: S41-47.
(9) van den Broek M, Visser K, Allaart CF, Huizinga TW (2013) Personalized medicine: predicting responses to therapy in patients with RA. Curr Opin Pharmacol 13: 463-469.
(10) Solomon DH, Bitton A, Katz JN, Radner H, Brown EM, Dougados, M, et al. (2014) Review: treat to target in rheumatoid arthritis: fact, fiction, or hypothesis? Arthritis Rheumatol 66: 775-782.
Acknowledgements: A portion of this study was supported by a grant from the Japanese Ministry of Education, Sports, and Research.
Kazuyuki Yoshizaki, MD & PhD
Department of Organic Fine Chemicals
The Institute of Scientific and Industrial Research, Osaka University
8-1 Mihogaoka, Ibaraki,