Original Article J. St. Marianna Univ.Vol. 10, pp. 39–49, 2019

Division of Rheumatology and Allergology, St. Marianna University School of Medicine

Early Response of DAS28-ESR (3) Predicts Sustained Response

to Tocilizumab Switched from Abatacept

Takayasu Ando, Takeshi Suzuki, Yutaka Gotou, Mitsuru Imamura, Hiroko Nagafuchi, Yoshioki Yamasaki, Seido Ooka, and Kimito Kawahata

(Received for Publication: August 16, 2019)

AbstractObjective To determine the predictors of the therapeutic effects of tocilizumab (TCZ) switched from other

biologics with different mechanisms of action in rheumatoid arthritis (RA).Methods Patients who switched from tumor necrosis factor inhibitors (TNFis) or abatacept (ABT) to TCZ

were analyzed. They were categorized into two groups based on clinical disease activity at week 24 (responsegroup: 28-joint disease activity score with erythrocyte sedimentation rate (DAS28-ESR) (3) ≤ 3.2, and non-re‐sponse group: DAS28-ESR (3) > 3.2). We compared DAS28-ESR (3) at the initiation of TCZ therapy (ΔDAS)in patients switching from TNFis and ABT. We examined whether the therapeutic effect of TCZ switched fromTNFis and ABT could be predicted using clinical parameters.

Results Sixty-seven patients were analyzed (TNFis, 53; ABT, 14); of these, 36 (67.9%) patients who re‐ceived TNFis and 6 (42.9%) who were treated with ABT were considered responders. In patients who switchedfrom TNFis, ΔDAS in the non-response group was significantly lower than that in the response group untilweek 8, and the improvement in the non-response group reached a plateau at week 12. Conversely, in patientstreated with ABT, ΔDAS was significantly different between the response and non-response groups in the earlyphase, i.e., at week 4. In univariate regression analysis, ΔDAS at week 4 was correlated with DAS28-ESR (3) atweek 24 (p < 0.05) in patients switching from ABT. Receiver operating characteristic analyses suggested that0.74 was the optimal ΔDAS cutoff at week 4 to predict response vs. non-response at week 24 (sensitivity:100%, specificity: 87.5%, p < 0.001).

Conclusion The efficacy of TCZ may vary depending on which biologics were used previously. The effec‐tiveness of TCZ switched from ABT could be predicted by the therapeutic response at week 4.

Key WordsRheumatoid arthritis, tocilizumab, abatacept, predictor, ΔDAS28

Introduction

Rheumatoid arthritis (RA) is an autoimmunedisorder characterized by joint pain and swelling, andeventual progression to joint destruction. By the endof the 20th century, the use of anti-tumor necrosisfactor inhibitors (TNFis) raised treatment standardsfor RA by achieving impressive symptomatic relief,significant improvement in quality of life, and sub‐stantially increased work time1,2), as well as prevent‐

ing the progression of joint destruction3). Currently,non-TNFis (tocilizumab [TCZ] and abatacept [ABT])are also available. In addition, the existence of thera‐peutic windows of opportunity and the concept of“treat-to-target” have become widely accepted, andclinical remission and/or low RA disease activity areachievable goals. In general, TNFis are effective inpatients with an inadequate response to conventionalsynthetic disease-modifying antirheumatic drugs(csDMARDs), including methotrexate (MTX). How‐

21

39

ever, TNFis are inadequate or ineffective in approxi‐mately 30% of patients, with two-thirds exhibitingmoderate to high disease activity at 1 year post-treat‐ment4). A previous report that studied 3497 patientswho discontinued first-line TNFis showed that 73.3%were switched to another TNFis and 26.7% wereswitched to a non-TNFis5). According to the recom‐mendations from the European League AgainstRheumatism, patients who do not respond to initialTNFi therapy should be switched to a different TNFior to a different class of biological agents6).

The ability to predict whether patients will re‐spond to their current treatment, particularly in theearly phase, is important to help decide whether tocontinue with the treatment. Some predictors ofDMARD treatments have been proposed. Low dis‐ease activity at baseline was reported to be associatedwith clinical remission in RA patients treated withTNFis7,8). In 214 patients treated with ABT, multivari‐ate logistic regression demonstrated that high diseaseactivity at baseline was an independent predictor forachieving low disease activity at 24 weeks3). Anotherstudy examined the baseline serum ratio of the type Icollagen C-terminal telopeptide to osteocalcin. Thisratio reflects bone turnover and serum matrix metal‐loproteinase-derived fragments of type I, II, and IIIcollagen levels and was found to be a biomarker forpredicting TCZ responders9).

As described above, repots have documentedseveral clinically available biomarkers or predictorsof success after switching biologics. However, thereis currently no consensus on what strategy to adoptwhen switching biologics, particularly when switch‐ing from one non-TNFi to another.

According to a large Japanese RA database, theNational Database of Rheumatic Disease in Japan(NinJa), the age at RA onset has increased signifi‐cantly over the last decade10). The incidence of ad‐verse side effects of ABT therapy was thought to berelativity low, so ABT was preferentially used in eld‐erly patients and/or in patients with comorbidities. Itis not always possible to use an adequate dose ofMTX in elderly patients because of their reducedphysical function. MTX use is also even occasionallycontraindicated in young patients due to comorbidi‐ties associated with ABT. Hence, biologics that pro‐vide an adequate therapeutic response without theneed for concomitant MTX would be appropriatetreatment for some patients switching from ABT. Theanti-IL-6 receptor antibody TCZ achieves a goodtherapeutic response in RA patients without concomi‐

tant MTX administration. Hence, TCZ considered tobe suitable option for continuing treatment whenswitching from ABT. However, the efficacy ofswitching to TCZ from ABT is unknown, and predic‐tors of this efficacy have not been identified. Further‐more, the rate of adverse events (AEs) during TCZtherapy is thought to be slightly higher than duringABT therapy11–14), and the appropriateness of switch‐ing from ABT to TCZ in terms of potential AEs isunknown. In this study, we examined the effective‐ness and safety of switching from the non-TNFi ABTto TCZ, compared with switching from a TNFi toTCZ.

Patients and Methods

PatientsThis study was approved by the St. Marianna

University School of Medicine ethics committee (Ap‐proval No. 4509). Eligible patients had been diag‐nosed with RA according to the 2010 American Col‐lege of Rheumatology-/-European League AgainstRheumatism classification criteria15), were taking bio‐logical agents after being switched from biologicalDMARDs other than TCZ, and had a 28-joint diseaseactivity score with erythrocyte sedimentation rate (3)(DAS28-ESR (3)) that could be calculated at baselineand at week 24 after starting TCZ therapy. Data wereretrospectively collected from clinical records fromSeptember 2008 to April 2018 at St. Marianna Uni‐versity School of Medicine hospital in Japan.

Study designTo detect the predictors of low disease activity,

we compared the baseline characteristics between pa‐tients treated with ABT vs. TNFis. To identify pre‐dictors in the early period, we analyzed the differencein DAS28-ESR (3) scores from the initiation of TCZtherapy (ΔDAS) to week 4. Finally, ΔDAS was com‐pared according to the previous therapy (ABT orTNFis).

In this study, clinical disease activity was evalu‐ated using the DAS28-ESR (3). The DAS28-ESR (3)is calculated as follows: [0.56*sqrt(t28)+ 0.28*sqrt(sw28) + 0.70*Ln(ESR)]*1.08 + 0.16. Pa‐tients who were switched from a TNFi or from ABTwere categorized into two groups according toDAS28-ESR (3) at week 24 (response group:DAS28-ESR (3) ≤ 3.2; non-response group: DAS28-ESR (3) > 3.2).

22

Ando T Suzuki T et al40

Safety assessmentAnother objective of this study was to assess the

safety of TCZ when patients were switched fromother biologics with different mechanisms of action.To evaluate safety, we retrospectively recorded the in‐cidence of AEs.

Statistical analysisDemographic and disease characteristics were

reported using descriptive statistics. Predictors of lowdisease activity were verified using regression analy‐sis and receiver operating characteristic (ROC) analy‐sis. The following data were recorded at the initiationof TCZ therapy (baseline, week 0): sex, age, diseaseduration, rheumatoid factor (RF) titer, anti–cyclic cit‐rullinated peptide antibody titer, matrix metallopro‐teinase-3 level, concomitant treatment (MTX or pre‐dnisolone), and the number of biologics used in thepast. The following disease parameters were recordedat baseline and after 24 weeks of TCZ treatment:number of tender and swollen joints (tender andswollen joint counts) among 28 joints, erythrocytesedimentation rate (ESR), serum c-reactive proteinlevels, and DAS28-ESR (3). All results are expressedas mean ± standard deviation or percentage. The Stu‐dent’s t-test or the Mann-Whitney t-test were used fortwo-group comparisons. All statistical tests were two-sided, and the level of significance was defined as p <0.05. All analyses were performed using SPSS ver‐sion 25.0.0 software (IBM Corp., Armonk, NY,USA).

Results

Patient enrollment and baseline demographicsSixty-seven patients were enrolled in this study.

Patients’ baseline characteristics when they switchedfrom TNFis or ABT to TCZ are shown in Ta‐bles 1, 2, and 3. In the overall sample, the mean agewas 56.8 ± 15.5 years, most patients were women(85.1%), and the mean morbidity was 10.3 ± 9.0years. Compared to patients in the non-responsegroup, those in the response group had a significantlylower RF titer (p = 0.04) and a significantly lowerDAS28-ESR (3) (p < 0.01). In terms of componentsof the DAS 28-ESR (3), the number of tender joints(tender joint count) tender joint count and ESR weresignificantly different between the two groups (p <0.05).

Clinical efficacyAmong the 67 patients, disease activity at week

24 was high in 5 (7.5%) patients, moderate in 20(29.9%) patients, and low or in remission in 42(62.7%) patients (low disease activity: 15 (22.4%),remission: 27 (40.3%)) (Figure 1). The ΔDAS amongpatients who switched from TNFis in the non-re‐sponse group was significantly lower than in the re‐sponse group until week 8. However, the improve‐ment in the non-response group reached a plateau atweek 12, and ΔDAS was showed a significant differ‐ence in the two groups at week 24 (Figure 2a). Onthe other hand, among patients who switched fromABT, ΔDAS was significantly different in the re‐sponse and non-response groups during the earlyphase of TCZ treatment (week 4) (Figure 2b).

Prediction analysisTo determine the predictors of response

(DAS28-ESR (3) ≤ 3.2 at week 24), we used regres‐sion analysis to examine the relationship between theΔDAS from baseline to week 4 (ΔDAS week 4) andthe DAS28-ESR (3) at week 24 in the response group(Table 4). Among the patients who switched fromTNFis, ΔDAS at week 4 could not predict the DA28-ESR (3) at week 24. On the other hand, ΔDAS atweek 4 could predict the DA28-ESR (3) at week 24in patients who switched from ABT. An ROC analy‐sis showed that the best cutoff value for ΔDAS atweek 4 to discriminate between the response andnon-response groups was 0.74 (true-positive fraction,1.0; false-positive fraction, 0.88; area under thecurve, 0.98; 95% confidence interval [0.92–1.00], p <0.01) (Figure 3).

Safety assessmentThere were 144 AEs among the 67 patients. All

AEs are shown in Table 5. These AEs occurredthroughout the course of TCZ therapy. Viral upperrespiratory infection accounted for almost one-thirdof AEs. The rate of AEs in the patients who switchedfrom TNFis and those who switched from ABT wasnot significantly different. Serious AEs, such as se‐vere infections were observed in both treatmentgroups, and there was one patient with a malignanttumor in the TNFis group. The rate of AEs was com‐parable to those in previous reports16,17). On average,TCZ was continued for 109 weeks. All patients sur‐vived during the study period.

Discussion

Among 67 patients treated with TCZ, we wereable to show that 68% achieved low disease activity

23

Prediction of therapeutic response to tocilizumab 41

Table 1. Baseline Demographic and Clinical Characteristics of All Patients and those Switched

from TNFis and ABT

All patients

(n = 67)

From TNFis

(n = 53)

From ABT

(n = 14)

Age (years) 58.8 ± 15.5 55.1 ± 15.3 63.4 ± 14.9

Women, n (%) 57 (85.1) 46 (86.8) 11 (78.6)

Disease duration (years) 10.3 ± 9.0 9.7 ± 9.0 13.1 ± 9.0

RF (mg/dl), positivity 116.2 ± 165.0 (79.1) 131.1 ± 185.7 (73.6) 60.2 ± 56.3 (100)

ACPA (U/ml), positivity 166.6 ± 174.5 (61.2) 169.9 ± 164.2 (67.9) 176.8 ± 251.3 (35.7)

MMP-3 (ng/mL) 217.3 ± 165.1 210.7 ± 136.5 211.8 ± 232.1

Methotrexate (mg/week), (%) 8.7 ± 3.7 (61.2) 8.7 ± 3.7 (67.9) 8.4 ± 4.3 (35.7)

Prednisolone (mg/d), (%) 4.4 ± 2.1 (59.7) 4.7 ± 2.3 (58.5) 3.3 ± 0.9 (64.3)

Previously used Biologics (n) 1.6 ± 0.8 1.6 ± 0.8 1.3 ± 0.6

DAS 28-ESR(3) (0w) 4.7 ± 1.4 4.7 ± 1.4 4.7 ± 1.3

remission, n (%) 3 (4.5) 2 (3.8) 1 (7.1)

low disease activity, n (%) 4 (6.0) 4 (7.5) 0

moderate disease activity, n (%) 38 (56.7) 30 (56.6) 8 (57.1)

high disease activity, n (%) 22 (32.8) 17 (32.1) 5 (35.7)

TJC (0W) 6.5 ± 6.9 6.3 ± 6.9 6.6 ± 7.2

SJC (0W) 5.3 ± 6.1 5.0 ± 5.8 6.1 ± 4.5

ESR (0W) 48.7 ± 34.5 50.5 ± 35.0 41.9 ± 34.2

CRP (0W) 2.1 ± 2.4 2.2 ± 2.5 1.6 ± 2.2

Comorbid diseases

other collagen diseases, n 15 6 9

interstitial pneumonia, n 11 6 5

diabetes mellitus, n 7 5 2

Some data are shown as mean ± SD.RF, rheumatoid factor; ACPA, Anti-cyclic citrullinated peptides antibody; MMP-3, matrix metalloproteinase-3; ESR, erythrocyte sedimentation rate; TJC, tender joint count; SJC, swelling joint count; CRP, C-reactive protein;

at week 24. RF titers before switching predicted dis‐ease activity at week 24 in the patients who switchedfrom TNFis. ΔDAS at week 4 could predict diseaseactivity at week 24 in the patients who switched fromABT. A total of 144 AEs were recorded during thesafety assessment, resulting in a rate that was compa‐rable to prior reports. These findings suggest that it is

safe to switch to TCZ as a subsequent biologic. Dur‐ing our study, 68% of patients treated with TCZ asthe subsequent biologic achieved low disease activityat week 24. In the ADACTA study, the rate of TCZ-treated patients who achieved low disease activity atweek 24 was 51.5%, and the remission rate was39.9%18). Other studies showed remission rates that

24

Ando T Suzuki T et al42

Table 2. Baseline Demographic and Clinical Characteristics of Patients Switching from TNFis

From TNFis Overall (n = 53) Response group

(n = 36)

Non-response group

(n = 17)

Age (years) 55.1 ± 15.3 53.4 ± 15.7 58.6 ± 14.2

Women, n (%) 46 (86.8) 34 (94.4) 12 (70.6)

Disease duration (years) 9.7 ± 9.0 9.3 ± 8.6 10.4 ± 10.0

RF (mg/dl), positivity 131.1 ± 185.7 (73.6) 73.0 ± 91.5 (78.8) 282.9 ± 273.7 (64.7) *

ACPA (U/ml), positivity 169.9 ± 164.2 (67.9) 168.2 ± 153.7 (58.3) 171.1 ± 185.4 (82.4)

MMP-3 (ng/mL) 210.7 ± 136.5 201.2 ± 120.1 229.8 ± 167.1

Methotrexate (mg/week), (%) 8.7 ± 3.7 (67.9) 8.1 ± 3.0 (63.9) 7.41 ± 5.82 (76.5)

Prednisolone (mg/d), (%) 4.7 ± 2.3 (58.5) 4.3 ± 2.2 (52.8) 3.7 ± 3.2 (70.6)

Previously used Biologics (n) 1.6 ± 0.8 1.6 ± 0.7 1.7 ± 1.1

DAS 28-ESR(3) (0w) 4.7 ± 1.4 4.3 ± 1.3 5.4 ± 1.4 *

remission, n (%) 2 (3.8) 2 (5.6) 0

low disease activity, n (%) 4 (7.5) 4 (11.1) 0

moderate disease activity, n (%) 30 (56.6) 21 (58.3) 9 (53)

high disease activity, n (%) 17 (32.1) 9 (25.0) 8 (47)

TJC (0W) 6.3 ± 6.9 5.0 ± 6.0 9.1 ± 7.9

SJC (0W) 5.0 ± 5.8 3.9 ± 4.7 7.18 ± 7.32

ESR (0W) 50.5 ± 35.0 41.1 ± 28.2 69.8 ± 40.3 *

CRP (0W) 2.2 ± 2.5 1.9 ± 2.2 2.9 ± 3.0

Comorbid diseases

other collagen diseases, n 6 3 3

interstitial pneumonia, n 6 4 2

diabetes mellitus, n 5 3 2

Some data are shown as mean ± SD. *: p < 0.05RF, rheumatoid factor; ACPA, Anti-cyclic citrullinated peptides antibody; MMP-3, matrix metalloproteinase-3; ESR, erythrocyte sedimentation rate; TJC, tender joint count; SJC, swelling joint count; CRP, C-reactive protein;

ranged from 34.1% to 43.9%19,20) at 24 weeks after in‐itiating TCZ therapy. These results that are consistentwith our own.

A previous retrospective cohort study reportedthat the drug retention rate of TCZ as a subsequentbiologic was significantly higher (94.7%) than that ofTNFis (59.3%)21). Another study showed that the re‐tention rate was higher for TCZ switched from a

TNFi than for a TNFi switched from TCZ3). Thus interms of retention rate, TCZ may be more suitablethan TNFis as a subsequent biologic.

Few studies have examined the efficacy ofswitching from ABT to TCZ. During this study, 6 outof 14 patients who switched from ABT to TCZ ach‐ieved low disease activity at week 24. This result sug‐gests the usefulness of treatment with TCZ after

25

Prediction of therapeutic response to tocilizumab 43

Table 3. Baseline Demographic and Clinical Characteristics of Patients Switching from ABT

Overall (n = 14) Response group

group (n = 6)

Non-response group

group (n = 8)

Age (years) 63.4 ± 14.9 62.0 ± 19.7 64.5 ± 11.5

Women, n (%) 11 (78.6) 4 (66.7) 7 (87.5)

Disease duration (years) 13.1 ± 9.0 21.0 ± 8.4 8.5 ± 5.7 *

RF (mg/dl), positivity 60.2 ± 56.3 (100) 43.8 ± 51.8 (100) 72.5 ± 59.7 (100)

ACPA (U/ml), positivity 176.8 ± 251.3 (35.7) 1.4 ± 0.3 (33.3) 352.7 ± 255.2 (37.5)

MMP-3 (ng/mL) 211.8 ± 232.1 245.2 ± 196.2 191.0 ± 262.8

Methotrexate (mg/week), (%) 8.4 ± 4.3 (35.7) 9.0 ± 4.2 (33.3) 8.0 ± 5.3 (37.5)

Prednisolone (mg/d), (%) 3.3 ± 0.9 (64.3) 3.4 ± 1.1 (66.7) 3.2 ± 0.8 (62.5)

Previously used Biologics (n) 1.3 ± 0.6 1.3 ± 0.5 1.3 ± 0.7

DAS 28-ESR(3) (0w) 4.7 ± 1.3 4.0 ± 1.3 5.3 ± 0.9 *

remission, n (%) 1 (7.1) 1 (16.7) 0

low disease activity, n (%) 0 0 0

moderate disease activity, n (%) 8 (57.1) 4 (66.7) 4 (50)

high disease activity, n (%) 5 (35.7) 1 (16.7) 4 (50)

TJC (0W) 6.6 ± 7.2 3.3 ± 3.56 9.0 ± 8.4

SJC (0W) 6.1 ± 4.5 5.7 ± 5.2 6.8 ± 4.4

ESR (0W) 41.9 ± 34.2 31.0 ± 1.9 50.0 ± 41.9

CRP (0W) 1.6 ± 2.2 1.5 ± 0.9 1.7 ± 2.9

Comorbid diseases

other collagen diseases, n 9 6 3

interstitial pneumonia, n 5 3 2

diabetes mellitus, n 2 0 2

Some data are shown as mean ± SD. *: p < 0.05RF, rheumatoid factor; ACPA, Anti-cyclic citrullinated peptides antibody; MMP-3, matrix metalloproteinase-3; ESR, erythrocyte sedimentation rate; TJC, tender joint count; SJC, swelling joint count; CRP, C-reactive protein;

ABT. A large retrospective cohort study reported thatthe retention rates for TCZ switched from ABT andfor ABT switched from biologics other than TCZwere not significantly different22). Although the effi‐cacy of TCZ as a subsequent biologic may be lowerthan that observed in treatment-naïve patients16,22,23),TCZ may be effective in patients who have receivedtwo or more biologics.

During our study, the ΔDAS at week 4 predictedthe DAS28-ESR (3) at week 24 in patients switchedfrom ABT, but not in those switched from a TNFi.We could not account for this difference. One poten‐tial explanation is the different mechanisms of actionof these biologics. ABT affects CD80/86, a surfaceantigen on presenting cells required for T cell activa‐tion as the second signal. ABT may suppress ac‐

26

Ando T Suzuki T et al44

Figure 1. Changes in disease activity stratified by DAS28-ESR (3) score.

DAS28-ESR, disease activity score in 28 joints;

Figure 2. Changes in the DAS28-ESR (3) from baseline. Disease activity in both the response and non-response groups was

similar until week 8 among patients who switched from a tumor necrosis factor inhibitor. Only the response group

showed a continuous decrease until week 24. ∆DAS differed significantly between the two groups at week 4

among patients who switched from abatacept.

quired immunity, and predominantly T cell function,more than innate immunity. On the other hand, TCZaffects RA by inhibiting IL-6 receptor signaling. IL-6plays a role in both acquired and innate immunity. Acquired immunity may contribute to the etiology ofRA in some patients with an insufficient response to

ABT more than innate immunity, and ABT might beappropriate in this population.

A previous randomized prospective study re‐ported that the clinical response at week 12 couldpredict low disease activity at 52 weeks in patientstreated with certolizumab pegol24). Another prospec‐

27

Prediction of therapeutic response to tocilizumab 45

Table 4. Single Regression Analysis between ΔDAS from Baseline to Week 4 and DAS-ESR (3) at

Week 24

β (95% CI) Standardization β p

ΔDAS

from TNFis 0.08 (-0.25 to 0.42) 0.08 0.60

from ABT 1.17 (0.75 to 0.59) 0.63 0.02

In patients who switched from abatacept, DAS28-ESR (3) at week 24 could be predicted by ΔDAS at week 4. Prediction was not possible in patients who switched from TNFis.

Figure 3. ROC curves of ΔDAS from baseline to week 4. The best cutoff value for ∆DAS from baseline to week 4

to distinguish between the response and non-response groups among patients who switched from abata‐

cept was 0.74 (sensitivity 1.0, specificity 0.88, 95% CI: 0.92–1.04 (p < 0.01)

TFP, true-positive fraction; FPF, false-positive fraction;

tive study reported that RA patients who exhibited agood response at 1 month after starting TCZ were 5.5times as likely to achieve remission at 36 months23).These reports support the theory that early treatmentresponse can predict the achievement of low diseaseactivity in RA patients.

Our results showed that TCZ efficacy and pre‐dictors of TCZ response differed between patientsswitched from TNFis or ABT. Our results may sup‐port the hypothesis that the efficacy of biologicsagainst RA varies according to the order of their use.A large-cohort (4970 patients), multicenter, retro‐spective study reported that the retention rate of bio‐logics was higher in patients switched from a TNFi toTCZ than in those switched from TCZ to a TNFi22).

Based on these findings, the order of biologic admin‐istration may affect the efficacy of RA treatment. Thereason for this is unknown, however, and larger stud‐ies and biochemical research are necessary to clarifythe issue.

Although it is common for RA patients to switchbiologics, there are few studies about switching be‐tween non-TNFis. Concomitant use of MTX is oftenavoided in the elderly due to reduced renal functionand/or other complications. The large Japanese data‐base NinJa showed that elderly RA patients weremore commonly treated with non-TNFi monotherapythan younger patients10). Switching between ABT andTCZ is common in daily practice. Our study suggeststhat ΔDAS at week 4 can predict the achievement of

28

Ando T Suzuki T et al46

Table 5. All Adverse Events

All patients(n = 67)

From TNFis(n = 53)

From ABT(n = 14)

Total adverse events 144 122 22

Viral upper respiratory tract inflammation

47 42 5

Bacterial infection 23 20 3

Liver injury 13 7 6

Pharyngitis 12 12 0

Diarrhea 9 9 0

Shingles 8 7 1

Dyslipidemia 6 3 3

Stomatitis 5 5 0

Skin rash 4 4 0

Hypertension 3 2 1

Leukocytopenia 2 2 0

Stomach ache 2 1 1

Anemia 2 1 1

Vomiting / Nausea 2 2 0

Thrombocytopenia 1 0 1

Headache 1 1 0

Malignant tumor 1 1 0

Fungal infection 1 1 0

Oral herpes 1 1 0

Viral gastroenteritis 1 1 0

Pneumocystis pneumonia 0 0 0

low disease activity at week 24 in this situation. ABTtends to be used in elderly patients and/or those withcomorbidities. Our findings are useful to aid the deci‐sion of whether to discontinue biologics soon afterstarting their administration and to avoid AEs associ‐ated with ineffective long-term use.

There are at least four limitations in this study.First, the study used a retrospective design, which isassociated with certain inherent limitations. In thistype of study, the analyzed data were not intended foruse in a study; therefore, some data may be inaccu‐rate and even unavailable. Second, the number of pa‐tients was small. Third, the standard tool to evaluatedisease activity in RA patients is the DAS28-ESR. Inthis study, we did not have all of the necessary infor‐

mation to calculate DAS28-ESR, so we used theDAS28-ESR (3) instead. Although the ability to de‐tect to symptomatic changes is similar when theDAS28 is calculated using 3 or 4 variables25), com‐parison between our study and previous studies maybe difficult. Fourth, patients were excluded if theDAS28-ESR (3) could not be evaluated at weeks 0and 24. Non-responders whose therapy was discon‐tinued before week 24 may have been excluded. It isalso possible that we overestimated the positive clini‐cal response to TCZ.

Conclusion

Switching biologics from ABT to TCZ was aseffective and safe as switching from a TNFi. ΔDAS

29

Prediction of therapeutic response to tocilizumab 47

at week 4 predicted the DAS28-ESR (3) at week 24in patients who switched from ABT to TCZ, and thiswill help avoid AEs associated with ineffective long-term use. Switching from ABT to TCZ should be abeneficial strategy in RA.

Conflicts of Interest

The authors have nothing to disclose.

References

1) Ollendorf Da, Klingman D, Hazard E, et al. Dif‐ferences in annual medication costs and rates ofdosage increase between tumor necrosis dactor-antagonist therapies for rheumatoid arthritis in amanaged care population. Clin Ther 2009; 31:825–835.

2) Smolen JS, Breedveld FC, Burmester GR, et al.Treating rheumatoid arthritis to target: 2014 up‐date of the recommendations of an internationaltask force. Ann Rheum Dis 2015; 75: 3–15.

3) Takahashi N, Kojima T, Kabeko A, et al. Clini‐cal efficacy of abatacept compared to adalim‐mab and tocilizumab in rheumatoid arthritis pa‐tients with high disease activity. Clin Rheumatol2014; 33: 39–47.

4) Ostergard M, Unkerskov J, Linde L, et al. Lowremission rates but long drug survival in rheu‐matoid arthritis patients treated with infliximabor etanercept: results from the nationwide Dan‐ish DANBIO database. Scand J Rheumatol2007; 36: 151–154.

5) Baser O, Ganguli A, Roy S, et al. Impact ofswitching from an Initial tumor necrosis factorinhibitor on health care resource utilization andcosts among patients with rheumatoid arthritis.Clin Therap 2015; 37: 1454–1465.

6) Smolen JS, Landewe R, Breedveld FC, et al.EULAR recommendations for the managementof rheumatoid arthritis with synthetic and bio‐logical disease-modifiyng antirheumatic drugs.Ann Rheum Dis 2009; 69: 964–975.

7) Hirich KL, Watson KD, Silman AJ, et al. Pre‐dictors of response to anti-TNF-anpha therapyamong patients with rheumatoid artheitis: resultsfrom the British Society for Rheumatology Bio‐logics Register. Rheumatology 2006; 45: 1558–1565.

8) Van Der Heijde D, Klareskog L, Landewe R, etal. Disease remission and sustained halting ofradiographic progression with combination eta‐nercept and methotrexate in patients with rheu‐

matoid arthritis. Arthritis Rheum 2007; 56:3928–3939.

9) Bay-Jensen AC, Platt A, Siebuhr AS, et al. Earlychanges in blood-based joint tissue destructionbiomarkers are predictive of response to tocili‐zumab in the LITHE study. Arthritis Researchand Therapy 2016; 18: 13.

10) Kato E, Sawada T, Tahara K, et al. The age atonset of rheumatoid arthritis is increasing in Ja‐pan: a nationwide database study. Int J Dis2017; 20: 839–845.

11) Strand V, Ahadieh S, French J, et al. Systematicreview and meta-analysis of serious infectionswith tofacitinib and biologic disease-modifyingantirheumatic drug treatment in rheumatoid ar‐thritis clinical trials. Arthritis Res Ther 2015;17: 362.

12) Rose-John S, Winthrop K, Calabrese L. The roleof IL-6 in host defence against infections: im‐munobiology and clinical implications. Nat RevRheumatol 2017; 13: 399–409.

13) Singh JA, Wells GA, Christensen R, et al. Ad‐verse effects of biologics: a network meta-analy‐sis and Cochrane overview. Cochrane DatabaseSyst Rev 2011; 2: CD008794.

14) Trap S, Eric Furst D, Boers M, et al. Risk of se‐rious adverse effects of biological and drugs inpatients with rheumatoid arthritis: a systematicreview meta-analysis. Rheumatology 2017; 56:417–425.

15) Aletaha D, Neogi T, Silman AJ, et al. 2010rheumatoid arthritis classification criteria: anAmerican College of Rheumatology/EuropeanLeague Against Rheumatism collaborative ini‐tiative. Ann Rheum Dis 2010; 69: 1580–1588.

16) Koike T, Harigai M, Inokuma S, et al. Effective‐ness and safety of tocilizumab: postmarketingsurveillance of 7901 patients with rheumatoidarthritis in japan. J Rheumatol 2014; 41: 15–23.

17) Ishiguro N, Atumi T, Harigai M, et al. Effective‐ness and safety of tocilizumab in achieving clin‐ical and functional remission, and sustain effi‐cacy in biologics-naïve patients with rheumatoidarthritis: The FIRST Bio study. Mod Rheumatol2017; 27: 217–226.

18) Gabay C, Emery P, Van Vollenhoven, et al. To‐cilizumab monotherapy versus adalimumabmonotherapy for treatment of rheumatoid arthri‐tis (ADACTA): a randomized, double-blind,controlled phase 4 trial. Lancet 2013; 381:1541–1550.

30

Ando T Suzuki T et al48

19) Nishimoto N, Miyasaka N, Tamamoto K, et al.Study of active controlled tocilicumab mono‐therapy for rheumatoid arthritis with an inade‐quate response to methotrexate (SATORI): sig‐nificant reduction in disease activity and serumvascular endothelial growth factor by IL-6 re‐ceptor inhibition therapy. Mod Rheumatol 2009;19: 12–19.

20) Baek HJ, Lim MJ, Park W, et al. Efficacy andsafety of tocilizumab in Korean patients with ac‐tive rheumatoid arthritis. Korean J Intern Med2018; 34: 917–931.

21) Kobayakawa T, Kojima T, Takahashi N, et al.Drug retention rates of second biologic agentsafter switching from tumor necrosis factor inhib‐itors for rheumatoid arthritis in Japan patients onlow-dose methotrexate or without. Mod Rheu‐matol 2015; 25: 251–256.

22) Takabayashi K, Ando F, Suzuki T. Comparing

the effectiveness of biological disease-modify‐ing antirheumatic drugs using real-world data.Mod Rheumatol 2019; 29: 87–97.

23) Nakashima Y, Kondo M, Fukuda T, et al. Re‐mission in patients with active rheumatoid ar‐thritis by tocilizumab treatment in routine clini‐cal practice: results from 3 years of prospec‐tively registered data. Mod Rheumatl 2014; 24:258–264.

24) Cutris JR, Luijtens K, Kavanaugh A. Predictingfuture response to certolizumab pegol in rheu‐matoid arthritis patients: features at 12 weeksassociated with low disease activity at 1 year.Arthritis Care Res 2012; 64: 658–667.

25) Madsen OR. Agreement between the DAS28-CRP assessed with 3 and 4 variables in patientswith rheumatoid arthritis treated with biologicalagents in the daily clinic. J Rheumatol 2013; 40:379–385.

31

Prediction of therapeutic response to tocilizumab 49