https://orcid.org/0000-0002-8130-4350
Abstract
Anaemia, a common comorbidity of RA, is related to high disease activity and poor prognosis. It is unknown which biologic/targeted synthetic (b/ts)-DMARDs are optimal for patients with anaemia and RA in regulating anaemia and controlling disease activity.
We investigated the change in haemoglobin (Hb) levels, drug retention rates and disease activities after the administration of b/ts-DMARDs with different modes of action [TNF inhibitors (TNFis), immunoglobulin fused with cytotoxic T-lymphocyte antigen (CTLA-4-Ig), IL-6 receptor inhibitors (IL-6Ris) and Janus kinase inhibitors (JAKis)] in patients with RA stratified by baseline Hb levels using the multicentre observational registry for patients with RA in Japan (ANSWER cohort).
A total of 2093 patients with RA were classified into three groups based on tertiles of the baseline Hb levels (Hblow, anaemic; Hbint, intermediate; Hbhigh, non-anaemic). IL-6Ri increased Hb levels in all groups (the mean change at 12 months in Hblow was +1.5 g/dl, Hbint +0.7 g/dl and Hbhigh +0.1 g/dl). JAKis increased the Hb level in patients with anaemia and RA and retained or decreased the Hb level in non-anaemic patients (the mean change at 12 months in Hblow was +0.6 g/dl, Hbint 0 g/dl and Hbhigh −0.3 g/dl). In patients with anaemia and RA, overall adjusted 3-year drug retention rates were higher in JAKi followed by IL-6Ri, CTLA4-Ig and TNFi (78.6%, 67.9%, 61.8% and 50.8%, respectively). Change of disease activity at 12 months was not different among different b/ts-DMARDs treatments.
IL-6Ri and JAKi can effectively treat patients with anaemia and RA in a real-world setting.
IL-6 receptor inhibitors and JAK inhibitors improved haemoglobin levels in RA patients with anaemia.
JAK inhibitors showed the highest drug retention rates among b/ts-DMARDs in RA patients with anaemia.
Introduction
RA is a chronic autoimmune condition with erosive synovitis. Since the introduction of the first biologic DMARDs (bDMARDs) 2 decades ago, eight bDMARDs and five targeted synthetic DMARDs (ts-DMARDs) are currently available in clinical practice. bDMARDs can be classified into three categories based on modes of action: TNF inhibitors (TNFis) [infliximab (IFX), etanercept (ETN), adalimumab (ADA), golimumab (GLM) and certolizumab-pegol (CZP)], anti-IL-6 receptor inhibitors (IL-6Ris) [tocilizumab (TCZ), sarilumab (SAR)] and immunoglobulin fused with cytotoxic T-lymphocyte antigen (CTLA-4-Ig) [abatacept (ABT)]. ts-DMARDs include Janus kinase inhibitors (JAKis) with different JAK selectivity: tofacitinib JAK1–3, baricitinib JAK1/2, peficitinib JAK1–3 and upadacitinib and filgotinib JAK1. The EULAR has put these b/ts-DMARDs in the same category under the proposed treatment algorithm for RA, although they have different modes of action [1]. RA is a heterogeneous disease with various phenotypes, and clinical response to b/ts-DMARDs may vary depending upon the dominant cytokines working in each clinical phenotype. Thus rheumatologists are searching for the most appropriate drugs for an individual patient with RA with a specific clinical phenotype to promote personalized medicine for RA.
Anaemia is a common comorbidity of RA and is associated with physical disability, impaired quality of life and even mortality in patients with RA [2–4]. Although some cases of anaemia in patients with RA can be caused by iron or folate deficiency, most cases may be attributed to anaemia of chronic disease (ACD), also referred to as anaemia of inflammation [5]. While the exact pathogenesis of ACD is unclear, recent studies have suggested the integral role of IL-6 in regulating hepcidin, a key molecule in the pathophysiology of ACD in patients with RA [6–8]. Hepcidin produced by hepatocytes upon IL-6 signalling alters iron homeostasis and causes ACD. Other inflammatory cytokines such as IFN-γ, TNF-α and IFN-α also affect haematopoiesis. IFN-γ promotes ACD by shifting toward myelopoiesis over erythropoiesis, enhancing macrophages’ phagocytosis of erythrocytes and shortening the erythrocyte lifespan. TNF-α can also cause ACD by inhibiting erythroid precursor proliferation. On the other hand, the haematopoietic growth factor erythropoietin (EPO) promotes erythropoiesis and is also a key factor in determining haemoglobin (Hb) levels.
Given the central role of IL-6 in the development of ACD in patients with RA, IL-6Ri that directly inhibits the IL-6/hepcidin pathway is more effective for treating patients with anaemia and RA than other bDMARDs. Previous studies revealed that TCZ reduced hepcidin levels and subsequently increased Hb. TCZ increased Hb levels more effectively than other biologic agents such as TNFi and CTLA-4-Ig [9–12]. It is also reported that low Hb level (female ≤11.7 g/dl, male ≤13.2 g/dl) can be a predictive marker for a good response to IL-6Ri compared with TNFi [13].
JAKi can also block IL-6 signalling via JAK1 and JAK2. Therefore JAKi can help improve ACD in patients with RA by inhibiting the IL-6/hepcidin pathway. JAKi also inhibit the IFN-γ signal via JAK1 and JAK2, which may also help improve ACD. On the other hand, JAKi blocks the EPO signal via JAK2, which may worsen anaemia. A decreased level of Hb was observed after treatment with JAKi in randomized controlled trials [14, 15]. Therefore there has been concern about using JAKi for patients with anaemia [16, 17]. However, there is no evidence about how these conflicting effects of JAK inhibition on different cytokines and growth factors affect Hb levels in patients with anaemia and RA in the real-world setting. In patients with anaemia and RA who have high disease activity and a poor prognosis, clinical efficacy outcomes, including disease activity score and drug retention rate, are still lacking.
In this study we investigated the change in Hb levels, drug retention rates and disease activity scores after administration of different b/ts-DMARDs (TNFi, IL-6Ri, CTLA-4-Ig and JAKi) in patients with RA stratified with baseline Hb levels using a multicentre observational registry for patients with RA in Japan, the ANSWER cohort.
Materials and methods
Patients
The Kansai Consortium for Well-being of Rheumatic Disease Patients (ANSWER) cohort is a multicentre observational registry of patients with RA in the Kansai district of Japan. Patients’ data from seven institutes (Kyoto University, Osaka University, Osaka Medical College, Kansai Medical University, Kobe University, Nara Medical University and Osaka Red Cross Hospital) were included. All patients suffering from RA who were treated with one of seven bDMARDs (ABT, ADA, CZP, ETN, GLM, IFX, SAR or TCZ) or two JAKis (tofacitinib, baricitinib) whose data of baseline Hb levels and clinical disease activity index (CDAI) were available were included in our study. All bDMARDs and JAKis included in this study were introduced between 2011 and 2020 and administered as i.v. or s.c. agents. Biosimilar agents were excluded from the analysis. In this study, patients fulfilled the 2010 ACR/EULAR RA criteria [18]. The reasons for discontinuation of bDMARDs were divided into four categories: remission, inefficacy, toxic adverse events and other reasons (e.g. patient preference or economic reasons) [19, 20]. Treating physicians were asked to adopt the most appropriate reason for discontinuation from four categories. Baseline demographic data, such as age, sex, CDAI, disease duration of RA, the number of previous bDMARD or JAKi exposures, concomitant use rates and doses of MTX and prednisolone (PSL) and RF- and ACPA-positive rates and titres, were also collected [21].
Patient stratification according to baseline Hb levels
Patients were classified into males and females before stratification using baseline Hb levels because females have lower Hb levels than males. We calculated equal tertiles of baseline Hb levels in male and female groups and patients were stratified according to these tertile levels as Hblow, Hbint and Hbhigh.
Outcome measures
The overall (excluding non-toxic reasons and remissions) and 3-year drug retention rates of three bDMARD categories (TNFi, IL-6Ri or CTLA-4-Ig) and JAKi were assessed in both unadjusted and adjusted models in each Hb group (Hblow, Hbint and Hbhigh). In both unadjusted and adjusted models, the 3-year retention rates were described according to specific discontinuation reasons (inefficacy and adverse events). Covariates considered in adjusted models were as follows: age, RA disease duration, MTX use, PSL use and the number of previous bDMARDs and JAKi exposures. The changes in Hb levels of three categories of bDMARDs and JAKi were assessed 3, 6, 9 and 12 months after b/ts-DMARDs initiation in each Hb group.
In patients whose 3-month CDAI after bDMARD and JAKi initiation were available, the changes in CDAI were assessed 3, 6, 9 and 12 months after b/ts-DMARD initiation in each Hb group using the last observation carried forward method. Because IL-6Ri and JAKi directly inhibit acute phase reactants, such as CRP, we used the CDAI, which does not contain acute phase reactants in its formula, as a composite RA disease activity index to compare serial disease activities.
Ethics
All study participants provided written informed consent and the study design was approved by the appropriate ethics review board. The representative facility of this registry is Kyoto University and this observational study (not clinical trial) was conducted in accordance with the Declaration of Helsinki and approved by ethics committee of each institute (Kyoto University approval number R0357).
Statistical analysis
Continuous data were described as mean and s.d. if normally distributed and median and interquartile range (IQR) if not normally distributed. Categorical data were described as percentages. In terms of statistical analysis, analysis of variance or the Kruskal–Wallis test was used when variates were numeric. The chi-squared test or Fisher’s exact test was used when variates were categorical.
The drug retention curves were described using the Kaplan–Meier method. Adjusted retention curves were also generated, considering potential confounders, including baseline age, RA disease duration, number of previous bDMARDs, PSL use and MTX use. The time to discontinuation of bDMARDs was analysed with a multivariate Cox proportional hazards model. All analyses were two-tailed and P-values <0.05 were considered statistically significant. Statistical analyses were performed with EZR (Saitama Medical Centre, Jichi Medical University, Saitama, Japan), a graphical user interface for R (R Foundation for Statistical Computing, Vienna, Austria) [22].
Results
Patient characteristics of each Hb group
A total of 2093 patients with RA treated with bDMARDs and JAKi, whose baseline Hb levels and CDAI were available, were extracted from the ANSWER cohort database. Patients were divided by their sex (359 males and 1734 females) and tertiles of their baseline Hb levels were calculated for each sex. We stratified patients into three groups (Hblow, Hbint and Hbhigh) according to the calculated tertiles. Regarding males, the Hblow (anaemic) group included patients with baseline Hb levels <12.0 g/dl, the Hbint (intermediate) group included patients with Hb from 12.0 to 13.7 g/dl and the Hbhigh (non-anaemic) group included patients with Hb ≥13.7 g/dl. Regarding females, the Hblow group included patients with baseline Hb levels <11.4 g/dl, the Hbint group included patients with Hb from 11.4 to 12.6 g/dl and the Hbhigh group included patients with Hb ≥12.6 g/dl.
The baseline patient characteristics of three bDMARDs and JAKi in each Hb group are shown in Table 1. The baseline CDAI and Hb levels did not vary among TNFi, IL-6Ri, CTLA-4-Ig and JAKi in all the Hb groups. In all the Hb groups, CTLA-4-Ig was used more frequently in older patients, TNFi was more frequently used concomitantly with MTX and JAKi was less frequently used in bio-naïve patients.
Baseline characteristics of patients stratified by baseline Hb levels (Hblow, Hbint and Hbhigh)
| Characteristics | Hblow | Hbint | Hbhigh | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CTLA-4-Ig (n = 183) | IL-6Ri (n = 182) | JAKi (n = 67) | TNFi (n = 283) | P-value | CTLA-4-Ig (n = 144) | IL-6Ri (n = 162) | JAKi (n = 62) | TNFi (n = 333) | P-value | CTLA-4-Ig (n = 134) | IL-6Ri (n = 135) | JAKi (n = 69) | TNFi (n = 339) | P-value | |
| Age, mean (s.d.) | 70.3 (11.9) | 63.2 (14.1) | 66.7 (11.6) | 61.9 (16.4) | <0.001 | 65.5 (10.8) | 60.1 (14.1) | 58.4 (14.4) | 57.9 (15.9) | <0.001 | 63.6 (11.4) | 57.1 (12.5) | 57.4 (12.4) | 55.6 (14.7) | <0.001 |
| Female, n (%) | 150 (82.0) | 157 (86.3) | 51 (76.1) | 233 (82.3) | 0.292 | 117 (81.2) | 129 (79.6) | 47 (75.8) | 289 (86.8) | 0.063 | 103 (76.9) | 116 (85.9) | 63 (91.3) | 279 (81.6) | 0.047 |
| Hb (g/dL), mean (s.d.) | |||||||||||||||
| Male | 10.8 (1.0) | 10.5 (1.3) | 11.3 (0.9) | 10.9 (1.0) | 0.13 | 12.9 (0.5) | 13.0 (0.5) | 13.0 (0.6) | 13.0 (0.4) | 0.70 | 14.9 (1.0) | 14.5 (0.7) | 15.4 (1.2) | 14.6 (0.8) | 0.09 |
| Female | 10.2 (0.8) | 10.3 (1.1) | 10.3 (1.0) | 10.4 (0.9) | 0.19 | 12.1 (0.3) | 12.1 (0.4) | 12.2 (0.3) | 12.1 (0.3) | 0.31 | 13.3 (0.6) | 13.5 (0.7) | 13.5 (0.6) | 13.4 (0.6) | 0.30 |
| Disease duration, median (IQR) | 7.9 (2.7–18.4) | 8.1 (2.6–16.4) | 8.7 (4.2–18.6) | 6.9 (1.5–16.8) | 0.199 | 7.1 (2.2–16.5) | 6.0 (1.7–13.6) | 6.8 (3.0–13.9) | 3.9 (1.4–12.6) | 0.096 | 5.8 (2.1–14.1) | 9.2 (3.7–19.4) | 7.8 (3.2–17.0) | 5.6 (1.6–12.6) | 0.003 |
| BMI, mean (s.d.) | 21.6 (3.7) | 22.1 (3.4) | 21.4 (3.3) | 21.3 (3.3) | 0.131 | 21.8 (3.6) | 22.1 (3.3) | 22.1 (3.1) | 22.1 (3.4) | 0.828 | 22.4 (3.6) | 22.6 (3.9) | 22.4 (4.6) | 22.5 (3.5) | 0.973 |
| RF, n (%) | 114 (83.8) | 108 (76.6) | 34 (75.6) | 185 (80.8) | 0.405 | 83 (90.2) | 88 (77.2) | 36 (81.8) | 210 (80.8) | 0.103 | 77 (82.8) | 82 (82.8) | 41 (89.1) | 192 (78.0) | 0.287 |
| RF titre, median (IQR) | 107 (29–313) | 75 (17–216) | 58 (16–280) | 91 (22–210) | 0.156 | 74 (30–189) | 47 (18–164) | 62 (24–192) | 48 (19–124) | 0.094 | 74 (26–229) | 65 (24–149) | 109 (27–216) | 62 (19–164) | 0.187 |
| ACPA, n (%) | 139 (88.5) | 112 (78.3) | 42 (84.0) | 203 (82.9) | 0.127 | 108 (88.5) | 112 (82.4) | 36 (70.6) | 221 (81.0) | 0.042 | 98 (84.5) | 93 (85.3) | 43 (82.7) | 219 (76.8) | 0.145 |
| ACPA titre, median (IQR) | 120 (23–327) | 75 (6–289) | 79 (7–341) | 79 (14–302) | 0.251 | 91 (28–265) | 82 (14–349) | 59 (2–434) | 78 (12–300) | 0.692 | 109 (15–436) | 124 (20–308) | 106 (9–472) | 59 (6–240) | 0.04 |
| Previous b/ts-DMARDs, n (%) | |||||||||||||||
| 1st bio | 120 (65.6) | 72 (39.6) | 18 (26.9) | 148 (52.3) | <0.001 | 94 (65.3) | 57 (35.2) | 19 (30.6) | 185 (55.6) | <0.001 | 83 (61.9) | 58 (43.0) | 15 (21.7) | 176 (51.5) | <0.001 |
| 2nd bio | 30 (16.4) | 54 (29.7) | 12 (17.9) | 61 (21.6) | 31 (21.5) | 49 (30.2) | 18 (29.0) | 83 (24.9) | 23 (17.2) | 44 (32.6) | 18 (26.1) | 105 (30.7) | |||
| ≥3rd bio | 33 (18.0) | 56 (30.8) | 37 (55.2) | 74 (26.1) | 19 (13.2) | 56 (34.6) | 25 (40.3) | 65 (19.5) | 28 (20.9) | 33 (24.4) | 36 (52.2) | 61 (17.8) | |||
| MTX, n (%) | 79 (43.2) | 98 (53.8) | 39 (58.2) | 178 (62.9) | <0.001 | 80 (55.6) | 91 (56.2) | 38 (61.3) | 240 (72.1) | <0.001 | 77 (57.5) | 72 (53.3) | 42 (60.9) | 231 (67.5) | 0.019 |
| MTX dose, median (IQR) | 6 (4–10) | 8 (6–10) | 8 (6–10) | 8 (6–12) | 0.001 | 8 (6–10) | 8 (6–11) | 10 (8–12) | 8 (6–12) | 0.036 | 8 (6–10) | 8 (6–10) | 8 (6–10) | 8 (6–10) | 0.507 |
| PSL, n (%) | 90 (49.2) | 81 (44.5) | 44 (65.7) | 124 (43.8) | 0.010 | 69 (47.9) | 72 (44.4) | 23 (37.1) | 126 (37.8) | 0.148 | 56 (41.8) | 54 (40.0) | 29 (42.0) | 109 (31.9) | 0.093 |
| PSL dose, median (IQR) | 5 (4–8) | 5 (4–10) | 5 (3–7) | 5 (3–8) | 0.06 | 5 (3–6) | 5 (4–6) | 5 (3–7) | 5 (3–6) | 0.555 | 5 (3–6) | 5 (4–10) | 5 (3–8) | 5 (4–7) | 0.323 |
| CDAI, median (IQR) | 17 (12–23) | 17 (11–24) | 18 (10–28) | 17 (10–25) | 0.806 | 15 (9–23) | 15 (9–20) | 14 (11–21) | 14 (9–20) | 0.753 | 14 (8–19) | 13 (7–18) | 15 (10–21) | 12 (7–18) | 0.058 |
| Characteristics | Hblow | Hbint | Hbhigh | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CTLA-4-Ig (n = 183) | IL-6Ri (n = 182) | JAKi (n = 67) | TNFi (n = 283) | P-value | CTLA-4-Ig (n = 144) | IL-6Ri (n = 162) | JAKi (n = 62) | TNFi (n = 333) | P-value | CTLA-4-Ig (n = 134) | IL-6Ri (n = 135) | JAKi (n = 69) | TNFi (n = 339) | P-value | |
| Age, mean (s.d.) | 70.3 (11.9) | 63.2 (14.1) | 66.7 (11.6) | 61.9 (16.4) | <0.001 | 65.5 (10.8) | 60.1 (14.1) | 58.4 (14.4) | 57.9 (15.9) | <0.001 | 63.6 (11.4) | 57.1 (12.5) | 57.4 (12.4) | 55.6 (14.7) | <0.001 |
| Female, n (%) | 150 (82.0) | 157 (86.3) | 51 (76.1) | 233 (82.3) | 0.292 | 117 (81.2) | 129 (79.6) | 47 (75.8) | 289 (86.8) | 0.063 | 103 (76.9) | 116 (85.9) | 63 (91.3) | 279 (81.6) | 0.047 |
| Hb (g/dL), mean (s.d.) | |||||||||||||||
| Male | 10.8 (1.0) | 10.5 (1.3) | 11.3 (0.9) | 10.9 (1.0) | 0.13 | 12.9 (0.5) | 13.0 (0.5) | 13.0 (0.6) | 13.0 (0.4) | 0.70 | 14.9 (1.0) | 14.5 (0.7) | 15.4 (1.2) | 14.6 (0.8) | 0.09 |
| Female | 10.2 (0.8) | 10.3 (1.1) | 10.3 (1.0) | 10.4 (0.9) | 0.19 | 12.1 (0.3) | 12.1 (0.4) | 12.2 (0.3) | 12.1 (0.3) | 0.31 | 13.3 (0.6) | 13.5 (0.7) | 13.5 (0.6) | 13.4 (0.6) | 0.30 |
| Disease duration, median (IQR) | 7.9 (2.7–18.4) | 8.1 (2.6–16.4) | 8.7 (4.2–18.6) | 6.9 (1.5–16.8) | 0.199 | 7.1 (2.2–16.5) | 6.0 (1.7–13.6) | 6.8 (3.0–13.9) | 3.9 (1.4–12.6) | 0.096 | 5.8 (2.1–14.1) | 9.2 (3.7–19.4) | 7.8 (3.2–17.0) | 5.6 (1.6–12.6) | 0.003 |
| BMI, mean (s.d.) | 21.6 (3.7) | 22.1 (3.4) | 21.4 (3.3) | 21.3 (3.3) | 0.131 | 21.8 (3.6) | 22.1 (3.3) | 22.1 (3.1) | 22.1 (3.4) | 0.828 | 22.4 (3.6) | 22.6 (3.9) | 22.4 (4.6) | 22.5 (3.5) | 0.973 |
| RF, n (%) | 114 (83.8) | 108 (76.6) | 34 (75.6) | 185 (80.8) | 0.405 | 83 (90.2) | 88 (77.2) | 36 (81.8) | 210 (80.8) | 0.103 | 77 (82.8) | 82 (82.8) | 41 (89.1) | 192 (78.0) | 0.287 |
| RF titre, median (IQR) | 107 (29–313) | 75 (17–216) | 58 (16–280) | 91 (22–210) | 0.156 | 74 (30–189) | 47 (18–164) | 62 (24–192) | 48 (19–124) | 0.094 | 74 (26–229) | 65 (24–149) | 109 (27–216) | 62 (19–164) | 0.187 |
| ACPA, n (%) | 139 (88.5) | 112 (78.3) | 42 (84.0) | 203 (82.9) | 0.127 | 108 (88.5) | 112 (82.4) | 36 (70.6) | 221 (81.0) | 0.042 | 98 (84.5) | 93 (85.3) | 43 (82.7) | 219 (76.8) | 0.145 |
| ACPA titre, median (IQR) | 120 (23–327) | 75 (6–289) | 79 (7–341) | 79 (14–302) | 0.251 | 91 (28–265) | 82 (14–349) | 59 (2–434) | 78 (12–300) | 0.692 | 109 (15–436) | 124 (20–308) | 106 (9–472) | 59 (6–240) | 0.04 |
| Previous b/ts-DMARDs, n (%) | |||||||||||||||
| 1st bio | 120 (65.6) | 72 (39.6) | 18 (26.9) | 148 (52.3) | <0.001 | 94 (65.3) | 57 (35.2) | 19 (30.6) | 185 (55.6) | <0.001 | 83 (61.9) | 58 (43.0) | 15 (21.7) | 176 (51.5) | <0.001 |
| 2nd bio | 30 (16.4) | 54 (29.7) | 12 (17.9) | 61 (21.6) | 31 (21.5) | 49 (30.2) | 18 (29.0) | 83 (24.9) | 23 (17.2) | 44 (32.6) | 18 (26.1) | 105 (30.7) | |||
| ≥3rd bio | 33 (18.0) | 56 (30.8) | 37 (55.2) | 74 (26.1) | 19 (13.2) | 56 (34.6) | 25 (40.3) | 65 (19.5) | 28 (20.9) | 33 (24.4) | 36 (52.2) | 61 (17.8) | |||
| MTX, n (%) | 79 (43.2) | 98 (53.8) | 39 (58.2) | 178 (62.9) | <0.001 | 80 (55.6) | 91 (56.2) | 38 (61.3) | 240 (72.1) | <0.001 | 77 (57.5) | 72 (53.3) | 42 (60.9) | 231 (67.5) | 0.019 |
| MTX dose, median (IQR) | 6 (4–10) | 8 (6–10) | 8 (6–10) | 8 (6–12) | 0.001 | 8 (6–10) | 8 (6–11) | 10 (8–12) | 8 (6–12) | 0.036 | 8 (6–10) | 8 (6–10) | 8 (6–10) | 8 (6–10) | 0.507 |
| PSL, n (%) | 90 (49.2) | 81 (44.5) | 44 (65.7) | 124 (43.8) | 0.010 | 69 (47.9) | 72 (44.4) | 23 (37.1) | 126 (37.8) | 0.148 | 56 (41.8) | 54 (40.0) | 29 (42.0) | 109 (31.9) | 0.093 |
| PSL dose, median (IQR) | 5 (4–8) | 5 (4–10) | 5 (3–7) | 5 (3–8) | 0.06 | 5 (3–6) | 5 (4–6) | 5 (3–7) | 5 (3–6) | 0.555 | 5 (3–6) | 5 (4–10) | 5 (3–8) | 5 (4–7) | 0.323 |
| CDAI, median (IQR) | 17 (12–23) | 17 (11–24) | 18 (10–28) | 17 (10–25) | 0.806 | 15 (9–23) | 15 (9–20) | 14 (11–21) | 14 (9–20) | 0.753 | 14 (8–19) | 13 (7–18) | 15 (10–21) | 12 (7–18) | 0.058 |
Hblow: male Hb <12.0 g/dL, female Hb <11.4 g/dL; Hbint: male ≥12.0–<13.7 g/dL, female ≥11.4–<12.6 g/dL; Hbhigh: male Hb ≥13.7 g/dL, female Hb ≥12.6 g/dL.
Baseline characteristics of patients stratified by baseline Hb levels (Hblow, Hbint and Hbhigh)
| Characteristics | Hblow | Hbint | Hbhigh | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CTLA-4-Ig (n = 183) | IL-6Ri (n = 182) | JAKi (n = 67) | TNFi (n = 283) | P-value | CTLA-4-Ig (n = 144) | IL-6Ri (n = 162) | JAKi (n = 62) | TNFi (n = 333) | P-value | CTLA-4-Ig (n = 134) | IL-6Ri (n = 135) | JAKi (n = 69) | TNFi (n = 339) | P-value | |
| Age, mean (s.d.) | 70.3 (11.9) | 63.2 (14.1) | 66.7 (11.6) | 61.9 (16.4) | <0.001 | 65.5 (10.8) | 60.1 (14.1) | 58.4 (14.4) | 57.9 (15.9) | <0.001 | 63.6 (11.4) | 57.1 (12.5) | 57.4 (12.4) | 55.6 (14.7) | <0.001 |
| Female, n (%) | 150 (82.0) | 157 (86.3) | 51 (76.1) | 233 (82.3) | 0.292 | 117 (81.2) | 129 (79.6) | 47 (75.8) | 289 (86.8) | 0.063 | 103 (76.9) | 116 (85.9) | 63 (91.3) | 279 (81.6) | 0.047 |
| Hb (g/dL), mean (s.d.) | |||||||||||||||
| Male | 10.8 (1.0) | 10.5 (1.3) | 11.3 (0.9) | 10.9 (1.0) | 0.13 | 12.9 (0.5) | 13.0 (0.5) | 13.0 (0.6) | 13.0 (0.4) | 0.70 | 14.9 (1.0) | 14.5 (0.7) | 15.4 (1.2) | 14.6 (0.8) | 0.09 |
| Female | 10.2 (0.8) | 10.3 (1.1) | 10.3 (1.0) | 10.4 (0.9) | 0.19 | 12.1 (0.3) | 12.1 (0.4) | 12.2 (0.3) | 12.1 (0.3) | 0.31 | 13.3 (0.6) | 13.5 (0.7) | 13.5 (0.6) | 13.4 (0.6) | 0.30 |
| Disease duration, median (IQR) | 7.9 (2.7–18.4) | 8.1 (2.6–16.4) | 8.7 (4.2–18.6) | 6.9 (1.5–16.8) | 0.199 | 7.1 (2.2–16.5) | 6.0 (1.7–13.6) | 6.8 (3.0–13.9) | 3.9 (1.4–12.6) | 0.096 | 5.8 (2.1–14.1) | 9.2 (3.7–19.4) | 7.8 (3.2–17.0) | 5.6 (1.6–12.6) | 0.003 |
| BMI, mean (s.d.) | 21.6 (3.7) | 22.1 (3.4) | 21.4 (3.3) | 21.3 (3.3) | 0.131 | 21.8 (3.6) | 22.1 (3.3) | 22.1 (3.1) | 22.1 (3.4) | 0.828 | 22.4 (3.6) | 22.6 (3.9) | 22.4 (4.6) | 22.5 (3.5) | 0.973 |
| RF, n (%) | 114 (83.8) | 108 (76.6) | 34 (75.6) | 185 (80.8) | 0.405 | 83 (90.2) | 88 (77.2) | 36 (81.8) | 210 (80.8) | 0.103 | 77 (82.8) | 82 (82.8) | 41 (89.1) | 192 (78.0) | 0.287 |
| RF titre, median (IQR) | 107 (29–313) | 75 (17–216) | 58 (16–280) | 91 (22–210) | 0.156 | 74 (30–189) | 47 (18–164) | 62 (24–192) | 48 (19–124) | 0.094 | 74 (26–229) | 65 (24–149) | 109 (27–216) | 62 (19–164) | 0.187 |
| ACPA, n (%) | 139 (88.5) | 112 (78.3) | 42 (84.0) | 203 (82.9) | 0.127 | 108 (88.5) | 112 (82.4) | 36 (70.6) | 221 (81.0) | 0.042 | 98 (84.5) | 93 (85.3) | 43 (82.7) | 219 (76.8) | 0.145 |
| ACPA titre, median (IQR) | 120 (23–327) | 75 (6–289) | 79 (7–341) | 79 (14–302) | 0.251 | 91 (28–265) | 82 (14–349) | 59 (2–434) | 78 (12–300) | 0.692 | 109 (15–436) | 124 (20–308) | 106 (9–472) | 59 (6–240) | 0.04 |
| Previous b/ts-DMARDs, n (%) | |||||||||||||||
| 1st bio | 120 (65.6) | 72 (39.6) | 18 (26.9) | 148 (52.3) | <0.001 | 94 (65.3) | 57 (35.2) | 19 (30.6) | 185 (55.6) | <0.001 | 83 (61.9) | 58 (43.0) | 15 (21.7) | 176 (51.5) | <0.001 |
| 2nd bio | 30 (16.4) | 54 (29.7) | 12 (17.9) | 61 (21.6) | 31 (21.5) | 49 (30.2) | 18 (29.0) | 83 (24.9) | 23 (17.2) | 44 (32.6) | 18 (26.1) | 105 (30.7) | |||
| ≥3rd bio | 33 (18.0) | 56 (30.8) | 37 (55.2) | 74 (26.1) | 19 (13.2) | 56 (34.6) | 25 (40.3) | 65 (19.5) | 28 (20.9) | 33 (24.4) | 36 (52.2) | 61 (17.8) | |||
| MTX, n (%) | 79 (43.2) | 98 (53.8) | 39 (58.2) | 178 (62.9) | <0.001 | 80 (55.6) | 91 (56.2) | 38 (61.3) | 240 (72.1) | <0.001 | 77 (57.5) | 72 (53.3) | 42 (60.9) | 231 (67.5) | 0.019 |
| MTX dose, median (IQR) | 6 (4–10) | 8 (6–10) | 8 (6–10) | 8 (6–12) | 0.001 | 8 (6–10) | 8 (6–11) | 10 (8–12) | 8 (6–12) | 0.036 | 8 (6–10) | 8 (6–10) | 8 (6–10) | 8 (6–10) | 0.507 |
| PSL, n (%) | 90 (49.2) | 81 (44.5) | 44 (65.7) | 124 (43.8) | 0.010 | 69 (47.9) | 72 (44.4) | 23 (37.1) | 126 (37.8) | 0.148 | 56 (41.8) | 54 (40.0) | 29 (42.0) | 109 (31.9) | 0.093 |
| PSL dose, median (IQR) | 5 (4–8) | 5 (4–10) | 5 (3–7) | 5 (3–8) | 0.06 | 5 (3–6) | 5 (4–6) | 5 (3–7) | 5 (3–6) | 0.555 | 5 (3–6) | 5 (4–10) | 5 (3–8) | 5 (4–7) | 0.323 |
| CDAI, median (IQR) | 17 (12–23) | 17 (11–24) | 18 (10–28) | 17 (10–25) | 0.806 | 15 (9–23) | 15 (9–20) | 14 (11–21) | 14 (9–20) | 0.753 | 14 (8–19) | 13 (7–18) | 15 (10–21) | 12 (7–18) | 0.058 |
| Characteristics | Hblow | Hbint | Hbhigh | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CTLA-4-Ig (n = 183) | IL-6Ri (n = 182) | JAKi (n = 67) | TNFi (n = 283) | P-value | CTLA-4-Ig (n = 144) | IL-6Ri (n = 162) | JAKi (n = 62) | TNFi (n = 333) | P-value | CTLA-4-Ig (n = 134) | IL-6Ri (n = 135) | JAKi (n = 69) | TNFi (n = 339) | P-value | |
| Age, mean (s.d.) | 70.3 (11.9) | 63.2 (14.1) | 66.7 (11.6) | 61.9 (16.4) | <0.001 | 65.5 (10.8) | 60.1 (14.1) | 58.4 (14.4) | 57.9 (15.9) | <0.001 | 63.6 (11.4) | 57.1 (12.5) | 57.4 (12.4) | 55.6 (14.7) | <0.001 |
| Female, n (%) | 150 (82.0) | 157 (86.3) | 51 (76.1) | 233 (82.3) | 0.292 | 117 (81.2) | 129 (79.6) | 47 (75.8) | 289 (86.8) | 0.063 | 103 (76.9) | 116 (85.9) | 63 (91.3) | 279 (81.6) | 0.047 |
| Hb (g/dL), mean (s.d.) | |||||||||||||||
| Male | 10.8 (1.0) | 10.5 (1.3) | 11.3 (0.9) | 10.9 (1.0) | 0.13 | 12.9 (0.5) | 13.0 (0.5) | 13.0 (0.6) | 13.0 (0.4) | 0.70 | 14.9 (1.0) | 14.5 (0.7) | 15.4 (1.2) | 14.6 (0.8) | 0.09 |
| Female | 10.2 (0.8) | 10.3 (1.1) | 10.3 (1.0) | 10.4 (0.9) | 0.19 | 12.1 (0.3) | 12.1 (0.4) | 12.2 (0.3) | 12.1 (0.3) | 0.31 | 13.3 (0.6) | 13.5 (0.7) | 13.5 (0.6) | 13.4 (0.6) | 0.30 |
| Disease duration, median (IQR) | 7.9 (2.7–18.4) | 8.1 (2.6–16.4) | 8.7 (4.2–18.6) | 6.9 (1.5–16.8) | 0.199 | 7.1 (2.2–16.5) | 6.0 (1.7–13.6) | 6.8 (3.0–13.9) | 3.9 (1.4–12.6) | 0.096 | 5.8 (2.1–14.1) | 9.2 (3.7–19.4) | 7.8 (3.2–17.0) | 5.6 (1.6–12.6) | 0.003 |
| BMI, mean (s.d.) | 21.6 (3.7) | 22.1 (3.4) | 21.4 (3.3) | 21.3 (3.3) | 0.131 | 21.8 (3.6) | 22.1 (3.3) | 22.1 (3.1) | 22.1 (3.4) | 0.828 | 22.4 (3.6) | 22.6 (3.9) | 22.4 (4.6) | 22.5 (3.5) | 0.973 |
| RF, n (%) | 114 (83.8) | 108 (76.6) | 34 (75.6) | 185 (80.8) | 0.405 | 83 (90.2) | 88 (77.2) | 36 (81.8) | 210 (80.8) | 0.103 | 77 (82.8) | 82 (82.8) | 41 (89.1) | 192 (78.0) | 0.287 |
| RF titre, median (IQR) | 107 (29–313) | 75 (17–216) | 58 (16–280) | 91 (22–210) | 0.156 | 74 (30–189) | 47 (18–164) | 62 (24–192) | 48 (19–124) | 0.094 | 74 (26–229) | 65 (24–149) | 109 (27–216) | 62 (19–164) | 0.187 |
| ACPA, n (%) | 139 (88.5) | 112 (78.3) | 42 (84.0) | 203 (82.9) | 0.127 | 108 (88.5) | 112 (82.4) | 36 (70.6) | 221 (81.0) | 0.042 | 98 (84.5) | 93 (85.3) | 43 (82.7) | 219 (76.8) | 0.145 |
| ACPA titre, median (IQR) | 120 (23–327) | 75 (6–289) | 79 (7–341) | 79 (14–302) | 0.251 | 91 (28–265) | 82 (14–349) | 59 (2–434) | 78 (12–300) | 0.692 | 109 (15–436) | 124 (20–308) | 106 (9–472) | 59 (6–240) | 0.04 |
| Previous b/ts-DMARDs, n (%) | |||||||||||||||
| 1st bio | 120 (65.6) | 72 (39.6) | 18 (26.9) | 148 (52.3) | <0.001 | 94 (65.3) | 57 (35.2) | 19 (30.6) | 185 (55.6) | <0.001 | 83 (61.9) | 58 (43.0) | 15 (21.7) | 176 (51.5) | <0.001 |
| 2nd bio | 30 (16.4) | 54 (29.7) | 12 (17.9) | 61 (21.6) | 31 (21.5) | 49 (30.2) | 18 (29.0) | 83 (24.9) | 23 (17.2) | 44 (32.6) | 18 (26.1) | 105 (30.7) | |||
| ≥3rd bio | 33 (18.0) | 56 (30.8) | 37 (55.2) | 74 (26.1) | 19 (13.2) | 56 (34.6) | 25 (40.3) | 65 (19.5) | 28 (20.9) | 33 (24.4) | 36 (52.2) | 61 (17.8) | |||
| MTX, n (%) | 79 (43.2) | 98 (53.8) | 39 (58.2) | 178 (62.9) | <0.001 | 80 (55.6) | 91 (56.2) | 38 (61.3) | 240 (72.1) | <0.001 | 77 (57.5) | 72 (53.3) | 42 (60.9) | 231 (67.5) | 0.019 |
| MTX dose, median (IQR) | 6 (4–10) | 8 (6–10) | 8 (6–10) | 8 (6–12) | 0.001 | 8 (6–10) | 8 (6–11) | 10 (8–12) | 8 (6–12) | 0.036 | 8 (6–10) | 8 (6–10) | 8 (6–10) | 8 (6–10) | 0.507 |
| PSL, n (%) | 90 (49.2) | 81 (44.5) | 44 (65.7) | 124 (43.8) | 0.010 | 69 (47.9) | 72 (44.4) | 23 (37.1) | 126 (37.8) | 0.148 | 56 (41.8) | 54 (40.0) | 29 (42.0) | 109 (31.9) | 0.093 |
| PSL dose, median (IQR) | 5 (4–8) | 5 (4–10) | 5 (3–7) | 5 (3–8) | 0.06 | 5 (3–6) | 5 (4–6) | 5 (3–7) | 5 (3–6) | 0.555 | 5 (3–6) | 5 (4–10) | 5 (3–8) | 5 (4–7) | 0.323 |
| CDAI, median (IQR) | 17 (12–23) | 17 (11–24) | 18 (10–28) | 17 (10–25) | 0.806 | 15 (9–23) | 15 (9–20) | 14 (11–21) | 14 (9–20) | 0.753 | 14 (8–19) | 13 (7–18) | 15 (10–21) | 12 (7–18) | 0.058 |
Hblow: male Hb <12.0 g/dL, female Hb <11.4 g/dL; Hbint: male ≥12.0–<13.7 g/dL, female ≥11.4–<12.6 g/dL; Hbhigh: male Hb ≥13.7 g/dL, female Hb ≥12.6 g/dL.
Hb changes of TNFi, IL-6Ri, CTLA-4-Ig and JAKi in each Hb group
The change in Hb levels (ΔHb) from baseline to 12 months after the treatment with TNFi, IL-6Ri, CTLA-4-Ig and JAKi in each Hb group are shown in Fig. 1. From baseline to 12 months, IL-6Ri increased Hb levels in all the Hb groups and ΔHb was always greatest in patients with RA treated with IL-6Ri during the 12-month follow-up. TNFi and CTLA-4-Ig increased Hb levels in patients with RA in the anaemic and intermediate groups, while in the non-anaemic group, they did not change Hb levels. In terms of JAKi, Hb levels increased in the anaemic group (mean ΔHb at 12 months = +0.6 g/dl), did not change in the intermediate group (mean ΔHb at 12 months = 0 g/dl) and decreased in the non-anaemic group (mean ΔHb at 12 months = −0.3 g/dl).
The mean difference in haemoglobin (ΔHGB) after b/ts-DMARD initiation in each Hb group. In Hblow, the mean ΔHGB at 12 months was as follows: TNFi vs CTLA-4-Ig vs IL-6Ri vs JAKi, 0.9 (s.d. 0.1) vs 0.8 (s.d. 0.1) vs 1.5 (s.d. 0.1) vs 0.6 (s.d. 0.3), respectively. In Hbint, the mean ΔHGB at 12 months was as follows: TNFi vs CTLA-4-Ig vs IL-6Ri vs JAKi, 0.3 (s.d. 0.1) vs 0.2 (s.d. 0.1) vs 0.7 (s.d. 0.1) vs 0 (s.d. 0.1), respectively. In Hbhigh, the mean ΔHGB at 12 months was as follows: TNFi vs CTLA-4-Ig vs IL-6Ri vs JAKi, −0.1 (s.d. 0.1) vs −0.1 (s.d. 0.1) vs 0.1 (s.d. 0.1) vs −0.3 (s.d. 0.1), respectively. ***P < 0.001. Hblow: male Hb <12.0 g/dL, female Hb <11.4 g/dL; Hbint: male ≥12.0–<13.7 g/dL, female ≥11.4–<12.6 g/dL; Hbhigh: male Hb ≥13.7 g/dL, female Hb ≥12.6 g/dL
Anaemia is a prevalent condition not only in the RA population, but also in the general population, with various aetiologies contributing to its occurrence. Foremost among them is iron deficiency, particularly in young females. To minimize the confounding effect of this population, we focused on RA patients >50 years of age and performed the same analysis. A total of 1604 patients (303 males and 1301 females) were enrolled in this analysis. The baseline patient characteristics of three bDMARDs and JAKi in each Hb group are shown in Supplementary Table S1, available at Rheumatology online. The ΔHb from baseline to 12 months after treatment with TNFi, IL-6Ri, CTLA-4-Ig and JAKi in each Hb group are shown in Supplementary Fig. S1A, available at Rheumatology online. In this population, as in RA patients, IL-6Ri increased Hb levels in all the Hb groups and JAKi showed a unique effect on Hb levels (increased Hb levels in the anaemic group, did not change in the intermediate group and decreased in the non-anaemic group).
In Supplementary Fig. S1B (available at Rheumatology online) we compared the effect on Hb levels with different JAK selectivity, tofacitinib for JAK1-3 and baricitinib for JAK1/2, to investigate the effect of EPO/JAK2 inhibition on Hb levels. In the Hbint and Hbhigh groups, tofacitinib showed higher ΔHb than baricitinib, while in the Hblow group, both tofacitinib and baricitinib increased Hb levels similarly.
Drug retention rate of TNFi, IL-6Ri, CTLA-4-Ig and JAKi in each Hb group
The unadjusted 3-year drug retention rates [overall (excluding remission and other reasons (e.g. patient preference or economic reasons), inefficacy defined and adverse events defined] are shown in Supplementary Fig. S2, available at Rheumatology online.
Because the drug retention rate is influenced by several confounders such as age, disease duration, the number of previous bDMARDs or JAKi exposures and concomitant use of MTX or PSL, we generated the confounder-adjusted drug retention curve of bDMARD and JAKi in each Hb group [19]. The adjusted overall 3-year drug retention rates [excluding remission and other reasons (e.g. patient preference or economic reasons)] are shown in Fig. 2A. In the Hblow group, the 3-year drug retention rates were as follows: TNFi vs CTLA-4-Ig vs IL-6Ri vs JAKi, 50.8% vs 61.8 vs 67.9 vs 78.6, respectively. In the Hbint and Hbhigh groups, the 3-year drug retention rates were as follows: TNFi vs CTLA-4-Ig vs IL-6Ri vs JAKi, 64.7% vs 63.6 vs 73.0 vs 64.3 and 57.1% vs 71.4 vs 76.0 vs 60.6, respectively. The adjusted 3-year drug retention rates based on discontinuation due to inefficacy are shown in Fig. 2B. In the Hblow group, similar to the overall analysis, JAKi showed the highest drug retention rates and IL-6Ri was second. In the Hbint group there were no statistically significant differences between bDMARDs and JAKi. In the Hbhigh group, IL-6Ri showed the highest drug retention rates and CTLA-4-Ig was second. The adjusted 3-year drug retention rates based on discontinuation due to adverse events are shown in Fig. 2C. In all the Hb groups, there were no statistically significant differences between bDMARDs and JAKi.
The adjusted 3-year drug retention rate curves. Adjusted confounders were baseline age, RA disease duration, number of previous b/ts-DMARD exposures, PSL use and MTX use. (A) The adjusted overall drug retention rate curve (excluding non-toxic reasons and remission) of three bDMARD categories and JAKis (including tofacitinib and baricitinib) in each Hb group. In Hblow, the 3-year drug retention rates were as follows: TNFi vs CTLA-4-Ig vs IL-6Ri vs JAKi, 50.8% vs 61.8 vs 67.9 vs 78.6, respectively. In Hbint, the 3-year drug retention rates were as follows: TNFi vs CTLA-4-Ig vs IL-6Ri vs JAKi, 64.7% vs 63.6 vs 73.0 vs 64.3, respectively. In Hbhigh, the 3-year drug retention rates were as follows: TNFi vs CTLA-4-Ig vs IL-6Ri vs JAKi, 57.1% vs 71.4 vs 76.0 vs 60.6, respectively. (B) The adjusted drug retention rate curve based on discontinuation due to the inefficacy of three bDMARD categories and JAKis (including tofacitinib and baricitinib) in each Hb group. In Hblow, the 3-year drug retention rates were as follows: TNFi vs CTLA-4-Ig vs IL-6Ri vs JAKi, 63.4% vs 73.8 vs 81.2 vs 90.1, respectively. In Hbint, the 3-year drug retention rates were as follows: TNFi vs CTLA-4-Ig vs IL-6Ri vs JAKi, 75.6% vs 70.1 vs 84.3 vs 75.6, respectively. In Hbhigh, the 3-year drug retention rates were as follows: TNFi vs CTLA-4-Ig vs IL-6Ri vs JAKi, 65.4% vs 75.9 vs 83.8 vs 72.7, respectively. (C) The adjusted drug retention rate curve based on discontinuation due to adverse events of three bDMARD categories and JAKis (including tofacitinib and baricitinib) in each Hb group. In Hblow, the 3-year drug retention rates were as follows: TNFi vs CTLA-4-Ig vs IL-6Ri vs JAKi, 80.6% vs 84.0 vs 85.3 vs 87.8, respectively. In Hbint, the 3-year drug retention rates were as follows: TNFi vs CTLA-4-Ig vs IL-6Ri vs JAKi, 86.2% vs 90.8 vs 86.5 vs 85.6, respectively. In Hbhigh, the 3-year drug retention rates were as follows: TNFi vs CTLA-4-Ig vs IL-6Ri vs JAKi, 88.3% vs 94.3 vs 91.7 vs 85.1, respectively. Hblow: male Hb <12.0 g/dL, female Hb <11.4 g/dL; Hbint: male ≥12.0–<13.7 g/dL, female ≥11.4–<12.6 g/dL; Hbhigh: male Hb ≥13.7 g/dL, female Hb ≥12.6 g/dL
The result of the Cox proportional hazards regression analysis adjusted for age, disease duration, number of previous b/ts-DMARDs exposures, PSL use and MTX use are shown in Table 2. In patients with anaemia and RA, JAKi, IL-6Ri and CTLA-4-Ig showed lower hazard ratios (HRs) for overall and inefficacy-defined discontinuation than TNFi in this order.
HRs for discontinuation due to inefficacy and adverse events in each Hb group (Hblow, Hbint, Hbhigh)
| Variables | Hblow | Hbint | Hbhigh | ||||||
|---|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | P-value | HR | 95% CI | P-value | HR | 95% CI | P-value | |
| Overall | |||||||||
| TNFi (ref) vs CTLA-4-Ig | 0.72 | 0.52, 0.99 | 0.047 | 0.94 | 0.65, 1.35 | 0.73 | 0.63 | 0.44, 0.90 | 0.011 |
| TNFi (ref) vs IL-6Ri | 0.59 | 0.43, 0.83 | 0.002 | 0.72 | 0.51, 1.03 | 0.074 | 0.59 | 0.40, 0.86 | 0.007 |
| TNFi (ref) vs JAKi | 0.45 | 0.27, 0.76 | 0.003 | 0.89 | 0.53, 1.48 | 0.64 | 0.97 | 0.62, 1.50 | 0.88 |
| Inefficacy | |||||||||
| TNFi (ref) vs CTLA-4-Ig | 0.64 | 0.42, 0.96 | 0.031 | 1.11 | 0.72, 1.70 | 0.62 | 0.66 | 0.43, 0.99 | 0.048 |
| TNFi (ref) vs IL-6Ri | 0.51 | 0.33, 0.77 | 0.001 | 0.60 | 0.38, 0.95 | 0.028 | 0.51 | 0.32, 0.81 | 0.005 |
| TNFi (ref) vs JAKi | 0.27 | 0.13, 0.60 | 0.001 | 0.87 | 0.47, 1.61 | 0.66 | 0.87 | 0.51, 1.48 | 0.61 |
| Adverse events | |||||||||
| TNFi (ref) vs CTLA-4-Ig | 0.91 | 0.53, 1.56 | 0.72 | 0.61 | 0.29, 1.25 | 0.18 | 0.55 | 0.26, 1.15 | 0.11 |
| TNFi (ref) vs IL-6Ri | 0.79 | 0.46, 1.36 | 0.39 | 1.02 | 0.57, 1.82 | 0.95 | 0.83 | 0.42, 1.65 | 0.60 |
| TNFi (ref) vs JAKi | 0.85 | 0.41, 1.75 | 0.66 | 0.91 | 0.35, 2.38 | 0.85 | 1.26 | 0.56, 2.81 | 0.57 |
| Variables | Hblow | Hbint | Hbhigh | ||||||
|---|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | P-value | HR | 95% CI | P-value | HR | 95% CI | P-value | |
| Overall | |||||||||
| TNFi (ref) vs CTLA-4-Ig | 0.72 | 0.52, 0.99 | 0.047 | 0.94 | 0.65, 1.35 | 0.73 | 0.63 | 0.44, 0.90 | 0.011 |
| TNFi (ref) vs IL-6Ri | 0.59 | 0.43, 0.83 | 0.002 | 0.72 | 0.51, 1.03 | 0.074 | 0.59 | 0.40, 0.86 | 0.007 |
| TNFi (ref) vs JAKi | 0.45 | 0.27, 0.76 | 0.003 | 0.89 | 0.53, 1.48 | 0.64 | 0.97 | 0.62, 1.50 | 0.88 |
| Inefficacy | |||||||||
| TNFi (ref) vs CTLA-4-Ig | 0.64 | 0.42, 0.96 | 0.031 | 1.11 | 0.72, 1.70 | 0.62 | 0.66 | 0.43, 0.99 | 0.048 |
| TNFi (ref) vs IL-6Ri | 0.51 | 0.33, 0.77 | 0.001 | 0.60 | 0.38, 0.95 | 0.028 | 0.51 | 0.32, 0.81 | 0.005 |
| TNFi (ref) vs JAKi | 0.27 | 0.13, 0.60 | 0.001 | 0.87 | 0.47, 1.61 | 0.66 | 0.87 | 0.51, 1.48 | 0.61 |
| Adverse events | |||||||||
| TNFi (ref) vs CTLA-4-Ig | 0.91 | 0.53, 1.56 | 0.72 | 0.61 | 0.29, 1.25 | 0.18 | 0.55 | 0.26, 1.15 | 0.11 |
| TNFi (ref) vs IL-6Ri | 0.79 | 0.46, 1.36 | 0.39 | 1.02 | 0.57, 1.82 | 0.95 | 0.83 | 0.42, 1.65 | 0.60 |
| TNFi (ref) vs JAKi | 0.85 | 0.41, 1.75 | 0.66 | 0.91 | 0.35, 2.38 | 0.85 | 1.26 | 0.56, 2.81 | 0.57 |
Adjusted confounders included baseline age, disease duration, number of previous bDMARDs administered, PSL use and MTX use.
Hblow: male Hb <12.0 g/dL, female Hb <11.4 g/dL; Hbint: male ≥12.0–<13.7 g/dL, female ≥11.4–<12.6 g/dL; Hbhigh: male Hb ≥13.7 g/dL, female Hb ≥12.6 g/dL.
HRs for discontinuation due to inefficacy and adverse events in each Hb group (Hblow, Hbint, Hbhigh)
| Variables | Hblow | Hbint | Hbhigh | ||||||
|---|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | P-value | HR | 95% CI | P-value | HR | 95% CI | P-value | |
| Overall | |||||||||
| TNFi (ref) vs CTLA-4-Ig | 0.72 | 0.52, 0.99 | 0.047 | 0.94 | 0.65, 1.35 | 0.73 | 0.63 | 0.44, 0.90 | 0.011 |
| TNFi (ref) vs IL-6Ri | 0.59 | 0.43, 0.83 | 0.002 | 0.72 | 0.51, 1.03 | 0.074 | 0.59 | 0.40, 0.86 | 0.007 |
| TNFi (ref) vs JAKi | 0.45 | 0.27, 0.76 | 0.003 | 0.89 | 0.53, 1.48 | 0.64 | 0.97 | 0.62, 1.50 | 0.88 |
| Inefficacy | |||||||||
| TNFi (ref) vs CTLA-4-Ig | 0.64 | 0.42, 0.96 | 0.031 | 1.11 | 0.72, 1.70 | 0.62 | 0.66 | 0.43, 0.99 | 0.048 |
| TNFi (ref) vs IL-6Ri | 0.51 | 0.33, 0.77 | 0.001 | 0.60 | 0.38, 0.95 | 0.028 | 0.51 | 0.32, 0.81 | 0.005 |
| TNFi (ref) vs JAKi | 0.27 | 0.13, 0.60 | 0.001 | 0.87 | 0.47, 1.61 | 0.66 | 0.87 | 0.51, 1.48 | 0.61 |
| Adverse events | |||||||||
| TNFi (ref) vs CTLA-4-Ig | 0.91 | 0.53, 1.56 | 0.72 | 0.61 | 0.29, 1.25 | 0.18 | 0.55 | 0.26, 1.15 | 0.11 |
| TNFi (ref) vs IL-6Ri | 0.79 | 0.46, 1.36 | 0.39 | 1.02 | 0.57, 1.82 | 0.95 | 0.83 | 0.42, 1.65 | 0.60 |
| TNFi (ref) vs JAKi | 0.85 | 0.41, 1.75 | 0.66 | 0.91 | 0.35, 2.38 | 0.85 | 1.26 | 0.56, 2.81 | 0.57 |
| Variables | Hblow | Hbint | Hbhigh | ||||||
|---|---|---|---|---|---|---|---|---|---|
| HR | 95% CI | P-value | HR | 95% CI | P-value | HR | 95% CI | P-value | |
| Overall | |||||||||
| TNFi (ref) vs CTLA-4-Ig | 0.72 | 0.52, 0.99 | 0.047 | 0.94 | 0.65, 1.35 | 0.73 | 0.63 | 0.44, 0.90 | 0.011 |
| TNFi (ref) vs IL-6Ri | 0.59 | 0.43, 0.83 | 0.002 | 0.72 | 0.51, 1.03 | 0.074 | 0.59 | 0.40, 0.86 | 0.007 |
| TNFi (ref) vs JAKi | 0.45 | 0.27, 0.76 | 0.003 | 0.89 | 0.53, 1.48 | 0.64 | 0.97 | 0.62, 1.50 | 0.88 |
| Inefficacy | |||||||||
| TNFi (ref) vs CTLA-4-Ig | 0.64 | 0.42, 0.96 | 0.031 | 1.11 | 0.72, 1.70 | 0.62 | 0.66 | 0.43, 0.99 | 0.048 |
| TNFi (ref) vs IL-6Ri | 0.51 | 0.33, 0.77 | 0.001 | 0.60 | 0.38, 0.95 | 0.028 | 0.51 | 0.32, 0.81 | 0.005 |
| TNFi (ref) vs JAKi | 0.27 | 0.13, 0.60 | 0.001 | 0.87 | 0.47, 1.61 | 0.66 | 0.87 | 0.51, 1.48 | 0.61 |
| Adverse events | |||||||||
| TNFi (ref) vs CTLA-4-Ig | 0.91 | 0.53, 1.56 | 0.72 | 0.61 | 0.29, 1.25 | 0.18 | 0.55 | 0.26, 1.15 | 0.11 |
| TNFi (ref) vs IL-6Ri | 0.79 | 0.46, 1.36 | 0.39 | 1.02 | 0.57, 1.82 | 0.95 | 0.83 | 0.42, 1.65 | 0.60 |
| TNFi (ref) vs JAKi | 0.85 | 0.41, 1.75 | 0.66 | 0.91 | 0.35, 2.38 | 0.85 | 1.26 | 0.56, 2.81 | 0.57 |
Adjusted confounders included baseline age, disease duration, number of previous bDMARDs administered, PSL use and MTX use.
Hblow: male Hb <12.0 g/dL, female Hb <11.4 g/dL; Hbint: male ≥12.0–<13.7 g/dL, female ≥11.4–<12.6 g/dL; Hbhigh: male Hb ≥13.7 g/dL, female Hb ≥12.6 g/dL.
We also compared drug retention rates of IL-6 signal inhibitors (IL-6Ri and JAKi) and non-IL-6 signal inhibitors (TNFi and CTLA-4-Ig) using Cox proportional hazards regression analysis adjusted for age, disease duration, number of previous b/ts-DMARDs exposures, PSL use and MTX use, to study the effect of IL-6 signal inhibition on drug continuation. The HR for overall discontinuation was as follows [IL-6 signal inhibitors (reference) vs non-IL-6 signal inhibitors]: Hblow 1.6 [1.2–2.1] (P = 0.001), Hbint 1.3 [0.9–1.8] (P = 0.10) and Hbhigh 1.2 [0.9–1.7] (P = 0.18). The HR for inefficacy-defined discontinuation was as follows [IL-6 signal inhibitors (reference) vs non-IL-6 signal inhibitors]: Hblow 2.0 [1.3–2.9] (P < 0.001), Hbint 1.5 [1.1–2.2] (P = 0.03) and Hbhigh 1.4 [0.9–2.1] (P = 0.06).
Disease activity changes of TNFi, IL-6Ri, CTLA-4-Ig and JAKi in each Hb group
The changes in CDAI from baseline (ΔCDAI) of TNFi, IL-6Ri, CTLA-4-Ig and JAKi during the 12-month follow-up in each Hb group are shown in Supplementary Fig. S3, available at Rheumatology online. There were no significant differences in ΔCDAI between four b/ts-DMARDs during the 12-month follow-up in all the Hb groups. Taking different patient characteristics into account, we performed multiple linear regression analysis to evaluate ΔCDAI at 12 months among b/ts-DMARDs in each Hb group. Adjusted after potential confounding factors, including age, disease duration, number of previous b/ts-DMARDs exposures, PSL use, MTX use and baseline CDAI, no significant correlation was observed between the type of b/ts-DMARDs and ΔCDAI at 12 months in any of the Hb groups.
Discussion
In this study we compared the changes in Hb levels, drug retention rates and disease activity scores following the administration of b/ts-DMARDs with different modes of action in patients with RA stratified by baseline Hb levels. Compatible with previous study results, IL-6Ri increased the Hb level most effectively during the 12-month follow-up in all the Hb groups [10–12]. Interestingly, JAKi increased Hb levels in patients in the anaemic group (Hblow) and maintained or decreased Hb levels in patients in the non-anaemic group (Hbint or Hbhigh). In patients with anaemia and RA, after adjustment for potential confounders, JAKi, IL-6Ri and CTLA-4-Ig showed higher overall and inefficacy-defined 3-year drug retention rates than TNFi. Surprisingly, JAKi showed the highest drug retention rate compared with other biologics; although not statistically significant, it highlighted the potential beneficial effect of JAKi that inhibits multiple cytokine signalling for controlling RA disease activity [14, 15]. Comparing IL-6 signal inhibitors and non-IL-6 signal inhibitors, non-IL-6 signal inhibitors showed a higher HR for overall discontinuation and inefficacy-defined discontinuation. Our study suggests that IL-6 signal blockade benefits patients with anaemia and RA by improving anaemia and drug retention in the real-world setting.
Recent studies have described the relationship between anaemia and disease activity in patients with RA [23, 24]. Patients with anaemia and RA showed higher disease activity scores and more progressive structural damage. Anaemia was an independent risk factor for radiographic damage during the 4-year follow-up, adjusted for disease activity and even in patients treated with TNFi [25]. Therefore, patients with anaemia and RA are at risk of joint damage and inadequate treatment outcomes. Hence it is important to determine which b/ts-DMARDs are the most appropriate for treating patients with anaemia and RA.
In our analysis, JAKi, IL-6Ri and CTLA-4-Ig showed higher overall and inefficacy-defined 3-year drug retention rates than TNFi, both in unadjusted and adjusted models, in patients with anaemia and RA. Recently the drug retention rate has been considered a surrogate marker of efficacy and safety in cohort-based studies [19, 26–29]. The drug retention rate of b/ts-DMARDs can be affected by several confounders, including age, disease duration, concomitant use of PSL and MTX and the number of previous b/ts-DMARD exposures [28, 30]. JAKis are relatively new, and in Japan they are often used after failures of multiple bDMARDs. Therefore the drug retention rate of JAKi could be low without adjustment. However, although statistically insignificant, even in the disadvantageous setting, JAKi showed the highest drug retention rate in the non-adjusted drug retention model.
Interestingly, patients with RA in different Hb groups showed different patterns of Hb changes after JAKi treatment initiation. JAKis inhibit the IL-6 signal via JAK1/2 and the EPO signal via JAK2. EPO is a hormone secreted from the kidney and stimulates red blood cell production from the bone marrow. While IL-6 blockade may be beneficial for patients with RA with ACD, EPO blockades could potentially be harmful to patients with anaemia and RA. When assessed without baseline Hb stratification, baricitinib, which inhibits JAK1- and JAK2-mediated signal, showed small decreases in Hb in the first 28 weeks and Hb levels returned to baseline or above by 48–52 weeks [31]. In our analysis, patients with RA with low baseline Hb levels showed an increase in the mean Hb level, while those with high baseline Hb levels showed a decrease in the mean Hb level. These data indicate that the ameliorating effect of JAKi on Hb levels via IL-6/hepcidin blockade outweighs the deteriorating effect on Hb levels via EPO signal blockade in patients with anaemia and RA. In contrast, in patients with RA with high baseline Hb levels, inhibitory effects on EPO may have outcompeted the effects on other cytokines. This unique influence of JAKi on Hb level may mitigate the concern about using JAKi in patients with anaemia and RA because JAKi could increase Hb levels in patients with ACD.
It has been shown that multiple cytokine signalling is involved with the pathogenesis of RA [32]. However, it is still difficult to know which cytokines are dominant in each patient in real-world clinical practice. In patients with anaemia and RA, IL-6 signalling may be dominant, considering IL-6-induced hepcidin was a major molecular mechanism that led to ACD. Increased inflammatory cytokines other than IL-6, such as TNF-α or IFN-γ, also play a role in ACD [33, 34]. In our analysis, the 3-year drug retention rate of JAKi based on the discontinuation due to inefficacy was the highest in JAKi-treated patients among four b/ts-DMARDs-treated patients in both adjusted and unadjusted models (Fig. 2B, Supplementary Fig. S2B, available at Rheumatology online). This may be because JAK can inhibit not only the IL-6 signal, but also multiple cytokine signals, including IFN-γ, which are known to cause treatment resistance [35].
This study has some limitations. First, because this is a retrospective cohort study, even after adjusting for potential confounders, there still exists the possibility that the drug retention rate was affected by other confounders that were not included in our adjusted models. Second, we stratified patients with RA only with baseline Hb levels and did not assess aetiologies of anaemia. We mainly discussed ACD, but other aetiologies, such as iron deficiency, folate deficiency and autoimmune-induced haemolytic anaemia, cause anaemia in patients with RA. Third, our analysis did not monitor dose and frequency changes in b/ts-DMARDs. Although all the b/ts-DMARDs have their principally approved dose, in clinical practice they could be modified for an individual patient. Fourth, JAKis other than tofacitinib and baricitinib, such as peficitinib, upadacitinib and filgotinib, are not included in our analysis because they were only recently approved in Japan and were rarely used during the study period. It is interesting to compare the Hb levels after JAKi treatment with different JAK specificities because selective JAK1 inhibitors that do not block JAK2/EPO signalling might show better Hb recovery in patients with anaemia and RA [36].
This is the first study comparing the change in Hb levels and the drug retention rate among different bDMARDs and JAKis in patients with RA stratified by baseline Hb levels. IL-6Ri increased Hb levels regardless of baseline Hb levels. JAKi showed fine-tuning of Hb levels, increasing Hb levels in patients with anaemia and maintaining or decreasing Hb levels in patients without anaemia. This phenomenon was unique to JAKis and may be explained by the inhibition of multiple cytokine/growth factor signalling, including IL-6, INF-γ or EPO. In patients with anaemia and RA, JAKi, IL-6Ri and CTLA-4-Ig showed higher drug retention rates than TNFi. JAKis are sometimes avoided in patients with anaemia and RA because of the concern about worsening anaemia via the inhibition of erythropoiesis signals. However, from our analysis, not only IL-6Ri, but also JAKi can be considered a treatment option in patients with anaemia and RA, considering the high drug retention rates in the real-world setting, with careful monitoring of adverse events.
Supplementary data
Supplementary data are available at Rheumatology online.
Data availability
The data underlying this article will be shared on reasonable request to the corresponding author.
Authors’ contributions
Y.N. and M.H. conceived the idea of the study. Y.N. developed the statistical analysis plan and conducted statistical analyses. Y.N. and M.H. contributed to the interpretation of the results. Y.N. and M.H. drafted the original manuscript. R.W., W.Y., K.E., T.H., T.K., H.S., M.K., Y.S., H.A., A.O., S.J., R.H., K. Murak, K. Murat, H.I., M.T., S.M., A.M. and other authors substantially contributed to the revision of the manuscript.
Funding
This study uses the ANSWER cohort supported by grants from 10 pharmaceutical companies (AbbVie, Asahi-Kasei Pharma, AYUMI Pharmaceutical, Chugai Pharmaceutical, Eisai, Janssen Pharmaceutical, Ono Pharmaceutical, Sanofi, Teijin Healthcare and UCB Japan) and an information technology services company (CAC). This study was conducted as an investigator-initiated study. These companies had no roles in the study design, data collection, data analysis, data interpretation or writing of the report.
Disclosure statement: R.W. has received research grants and/or speaker fees from AbbVie, Asahi Kasei, Chugai, Eli Lilly and Sanofi. K.E. is affiliated with the Department of Musculoskeletal Regenerative Medicine, Osaka University Graduate School of Medicine, which is supported by Taisho; has received research grants from AbbVie, Asahi-Kasei, Eisai, Mitsubishi-Tanabe and Teijin Pharma; and has received payments for lectures from AbbVie, Amgen, Asahi-Kasei, Astellas, Bristol-Myers Squibb, Chugai, Daiichi-Sankyo, Eisai, Eli Lilly, Janssen, Mitsubishi-Tanabe, Ono Pharmaceutical, Pfizer, Sanofi, Taisho and UCB Japan. T.H. received speaker fees from Asahi-Kasei, Astellas, Mitsubishi-Tanabe and Teijin. T.K. is affiliated with a department that is financially supported by six pharmaceutical companies (Mitsubishi-Tanabe, Asahi-Kasei, AbbVie, Chugai, Eisai and Takeda) and has received payments for lectures from AbbVie, Bristol-Myers Squibb, Chugai, Eisai, Eli Lilly, Pfizer and Boehringer Ingelheim. Y.S. has received speaker fees from Actelion, Bristol-Myers Squibb, Chugai Pharmaceutical, Eisai, Mitsubishi Tanabe Pharma and Sanofi. A.O., K.Murata and M.T. belong to the Department of Advanced Medicine for Rheumatic Diseases that is supported by Nagahama City, Shiga, Japan, Toyooka City, Hyogo, Japan and five pharmaceutical companies (Mitsubishi Tanabe Pharma, Chugai Pharmaceutical, UCB Japan, AYUMI Pharmaceutical and Asahi Kasei Pharma). A.O. has received speaker fees from Chugai, Ono Pharmaceutical, Eli Lilly, Mitsubishi-Tanabe, Asahi-Kasei and Takeda. S.J. has received speaker fees from AbbVie, Asahi Kasei Pharma, Bristol-Myers Squibb, Chugai Pharmaceutical, Eisai, Eli Lilly Japan, Janssen Pharmaceutical, Mitsubishi Tanabe Pharma and Ono Pharmaceutical. R.H. has received speaker fees from AbbVie. K.Murata has received speaker fees and/or consulting fees from AbbVie, Eisai, Chugai Pharmaceutical, Mitsubishi Tanabe Pharma, Pfizer, Bristol-Myers Squibb and Asahi Kasei Pharma. H.I. received a research grant from Bristol-Myers Squibb. M.T. has received research grants and/or speaker fees from AbbVie, Asahi-Kasei Pharma, Astellas Pharma, Ayumi Pharmaceutical, Bristol-Myers Squibb, Chugai Pharmaceutical, Eisai, Eli Lilly Japan, Pfizer, UCB Japan, Janssen Pharmaceutical, Mitsubishi-Tanabe Pharma, Novartis Pharma and Taisho Pharma. S.M. has received research grants and/or speaker fees from Takeda, Eisai, Asahi-Kasei, Astellas, Pfizer, Taisho, Mitsubishi-Tanabe and Chugai. A.M. has received honorarium from AbbVie, Chugai Pharmaceutical, Eli Lilly Japan, Eisai, Pfizer, Bristol-Myers Squibb, Mitsubishi Tanabe Pharma, Astellas Pharma and Gilead Sciences Japan and has received research grants from AbbVie, Asahi Kasei Pharma, Chugai Pharmaceutical, Mitsubishi Tanabe Pharma and Eisai. M.H. received research grants and/or speaker fees from AbbVie, Asahi Kasei, Astellas, Ayumi, Bristol-Myers Squibb, Chugai, EA Pharma, Eisai, Daiichi Sankyo, Eli Lilly, Nihon Shinyaku, Novartis Pharma and Tanabe Mitsubishi. The remaining authors have no financial conflicts of interest to disclose. The pharmaceutical companies had no role in the design of the study, the collection or analysis of data, the writing of the manuscript or the decision to submit the manuscript for publication.
Acknowledgements
All study participants provided informed consent and the study design was approved by the appropriate ethics review board. The representative facility of this registry is Kyoto University, and this observational study (not clinical trial) was conducted in accordance with the Declaration of Helsinki and approved by the ethics committee of each institute (Kyoto University approval number R0357).
Comments