Article Text

Download PDFPDF

Extended report
Down-titration and discontinuation of infliximab in rheumatoid arthritis patients with stable low disease activity and stable treatment: an observational cohort study
  1. Aatke van der Maas1,
  2. Wietske Kievit2,
  3. Bart J F van den Bemt3,
  4. Frank H J van den Hoogen1,
  5. Piet L van Riel4,
  6. Alfons A den Broeder1
  1. 1Department of Rheumatology, Sint Maartenskliniek, Nijmegen, The Netherlands
  2. 2Department of Rheumatology, Radboud University, Nijmegen Medical Centre, Nijmegen, The Netherlands
  3. 3Department of Pharmacy, Sint Maartenskliniek, Nijmegen, The Netherlands
  4. 4Department of Rheumatology, Radboud University, Nijmegen Medical Center, Nijmegen, The Netherlands
  1. Correspondence to Aatke van der Maas, Sint Maartenskliniek, Rheumatology, Hengstdal 3, Postbus 9011, Nijmegen 6500 GM, The Netherlands; a.vandermaas{at}


Down-titration, or discontinuing infliximab, has proven to be feasible in RA patients. Therefore, our local treatment protocol includes tapering infliximab dose. This observational study describes the prevalence of successful down-titration in daily clinical practice and its effect on costs and quality of life (QoL).

Methods Infliximab was down-titrated with 25% of the original dose (3 mg/kg) every 8–12 weeks without interval change until discontinuation or flare in all RA patients with stable low 28-joint disease activity score (DAS28) and stable treatment for >6 months. During 1 year DAS28, RA medication, outpatient clinic visits, RA related absenteeism and EuroQoL5D (European QoL questionnaire, EQ5D) were documented. Prevalence of successful down-titration and changes in DAS28, QoL and costs were described.

Results In 16% (95% CI 6 to 26) and 45% (95% CI 31 to 59), respectively, infliximab could be discontinued or down-titrated. Mean infliximab dose decreased significantly from 224 mg (95% CI 212 to 236 mg) at start, to 130 mg (95% CI 105 to 154 mg) after 1 year. Median DAS28 increased from 2.5 (p25–75=2.0–2.9) to 2.8 (2.2–3.6) (p=0.002). Extra corticosteroids were given in 8% of the visits. Disease modifying antirheumatic drugs were seldom changed. There was no statistical difference in QoL after down-titration. Mean reduction in the costs was €3474 (95% CI 2457 to 4492) per patient.

Conclusion In the majority of patients with stable low DAS28 and stable treatment, infliximab can be down-titrated or discontinued, which results in a considerable reduction in costs without influencing QoL.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


The objective of treatment with disease-modifying antirheumatic drugs (DMARDs) or biologics in rheumatoid arthritis (RA) is to determine the optimal dose of the drug and thus maximise its effect and minimise dose-related side effects and costs. Frequent monitoring of disease activity, establishing targets for the desired level of disease activity and increasing therapy when the disease is too active, has proved to be more effective than routine outpatient care.1,,6 However, a disadvantage of a hit-hard-and-hit-early strategy is that it may lead to overtreatment.7,,10 This is also the case for anti-tumour necrosis factor (anti-TNF) inhibitors like infliximab, which is an effective treatment for RA; however, it is also associated with high cost and increased risk for adverse effects like infection and possibly malignancies.6 ,11,,15 As a result, overtreatment with infliximab is undesirable.

Reducing overtreatment in RA patients with long-term low disease activity seems feasible since RA patients treated with biologics like infliximab, who reach low disease activity or remission, can theoretically be divided in at least three subgroups: a subgroup of patients with low disease activity or remission caused by an adequate infliximab dose; a subgroup with adequate disease control due to infliximab, but who would also maintain disease control with a lower dose; and finally a subgroup with low disease activity unrelated to infliximab treatment, who would fare equally well without it. The existence of the latter group seems contra-intuitive; however, two possible explanations exist. First the initial apparent ‘response’ may be caused by regression-to-the-mean, as witnessed by response percentages in placebo-treated patients in anti-TNF randomised controlled trials. Second, infliximab was only temporarily necessary to gain control of the disease.7,,10 To find the optimal dose, down-titration and discontinuation are warranted.

The feasibility of down-titration or discontinuation of biologic treatment in RA patients with low disease activity or remission has been demonstrated by several studies.16,,22 Therefore, the local treatment protocol in our hospital includes a combination of disease activity-guided tapering down of the infliximab dose, and ultimately discontinuation in RA patients with long-term stable low disease activity and treatment. Since there is not much data available about the combination of both down-titration and discontinuation in RA patients, or about the prevalence of successful down-titration in daily clinical practice, we conducted an observational study. Furthermore, we evaluated the effects of down-titrating infliximab dose on disease activity, quality of life (QoL) and direct costs.


All RA patients (fulfilling the American Rheumatology Association criteria 1987) of the Sint Maartenskliniek Nijmegen, The Netherlands, treated with infliximab, who have stable low disease activity, defined as a 28-joint disease activity score (DAS28) <3.2 (in accordance with previous infliximab discontinuation studies), and stable anti-rheumatic treatment for at least 6 months are treated according to our local down-titration protocol.21 ,23 ,24 As stated in this protocol, the infliximab dose is tapered down with 25% of the original dose (3 mg/kg) every 8–12 weeks without changing the interval until discontinuation or flare, which is defined as an increase in DAS28≥1.2 compared with baseline on two subsequent visits with at least 2 weeks in-between visits.17 After the threshold of DAS28>3.2 is reached, a more stringent criterion is used, namely an increase in DAS28≥0.6. To ensure that disease activity will remain adequately controlled, all patients of the down-titration protocol are examined before every infliximab infusion. In addition, patients are encouraged to contact the rheumatologist in-between infusions if to their opinion disease activity has deteriorated, which will result in an outpatient clinical evaluation within 1–3 days. If a patient flares, infliximab dose will be increased to the last effective dose. If despite increasing the dose DAS28 remains 3.2 or more, the dose is further increased up to 3 mg/kg, and if necessary, therapy is switched to a different biologic.

From January to April 2010, all patients who started with infliximab down-titration were included in an observational cohort to describe the prevalence of successful down-titration in daily clinical practice. Each patient was followed for 1 year. Demographic data including gender, age, weight, disease duration, rheumatoid factor, anti cyclic citrullinated peptide (anti-CCP), erosive disease, previous and concomitant DMARD use, previous biologic treatment, concomitant non-steroidal anti-inflammatory drug (NSAID) treatment, duration of infliximab treatment and employment were collected. At every visit, DAS28, change in RA medication (DMARD, NSAID, corticosteroid), number of outpatient clinic visits in-between infusions and RA-related work-absenteeism were documented. At every 4 months, QoL was measured by means of the EuroQoL5D.25

At first, infliximab initiation on all patients in our cohort had failed at least on 2 DMARDs, of which 1 was methotrexate, and had active RA (DAS28>3.2). Since DAS28-guided dose down-titration was performed as usual care for all RA patients, no extra venous puncture was necessary, and no demanding questionnaires were administered; this observational study did not require approval of an ethical committee according to Dutch legislation. Written informed consent was given by all patients.


Descriptive statistics were used for demographic data. Primary outcome was percentage of patients successfully down-titrated, and percentage successfully discontinued after 1 year. Per down-titration step, a Kaplan-Meier curve is provided demonstrating the probability of not requiring a dose re-escalation. Secondary outcomes were the mean difference in infliximab dose at baseline, and after 1 year, tested for statistically significant differences with a student's t-test, as well as median difference in DAS28 per down-titration group (no down-titration, partly down-titrated and stopped) with the Kruskal-Wallis non-parametric test. Furthermore, frequencies of DMARD, NSAID and corticosteroid changes were described, next to difference in number of outpatient clinic visits in the year of down-titration compared with the preceding year. Possible predictors for down-titration were investigated by means of univariate logistic regression. Subsequently, multivariate analysis was performed for three possible predictors given the sample size of our cohort. These factors were chosen based on explained variance found in univariate analyses and based on the literature.

Differences in QoL and direct costs preceding and succeeding down-titration were assessed. The cost analysis consisted of two main components: determination of volumes of care related to the down-titration protocol and determination of cost prices for each volume of consumption. Volumes of care were multiplied with the cost prices to calculate costs. As volumes of care were registered, out-patient clinic visits, medication use and work-related absenteeism, cost prices for medication were retrieved from the Dutch National tariff list provided by the Dutch Board of Health Insurances (August 2011). The standard cost prices from the Dutch Guideline for Cost Analyses were used for valuation of hospital-related care and work-related absenteeism (2010).26 Since we postulated that during down-titration, extra outpatient clinic visits or concomitant medication use will have an effect on costs, the reduction in costs was also analysed every 4 months. The last 4 months are presumed to approximate the new steady-state in costs. To investigate change in EQ5D per 4 months an analysis of variance (ANOVA) for repeated measures was carried out.


In total, 94 RA patients were treated with infliximab, of whom 51 met the criteria of the local down-titration protocol and were observed during 1 year (figure 1). Demographic data of these patients is depicted in table 1. Patients were treated with concomitant methotrexate; however, due to side effects, 32% were treated with leflunomide, azathioprine or infliximab monotherapy. Overall data was collected over 421 visits. Protocol violations were observed in 7.1% of all visits.

Figure 1

Flow diagram of rheumatoid arthritis patients treated according to the down-titration protocol. DAS28, 28-joint disease activity score; RA, rheumatoid arthritis.

Table 1

Baseline characteristics

In 16% (95% CI 6 to 26) of the patients, infliximab could be stopped, and in 45% (95% CI 31 to 59) of the patients, the dose could be successfully down-titrated after 1 year. In 39% (95% CI 26 to 53), no down-titration was possible, that is, patients returned to the original dose. The proportion of patients not requiring a dose re-escalation after each down-titration step due to flare during 1 year follow-up is depicted in figure 2. Considering the patients in whom infliximab could be discontinued, we verified that at time of infliximab initiation all patients had active disease (mean DAS28 was 5.9, SD 1.4) and 88% was rheumatoid factor and anti-CCP positive.

Figure 2

Kaplan-Meier curve demonstrating the probability of not requiring dose re-escalation after each down-titration step—75% (A), 50% (B), 25% (C) and 0% (D) of the infliximab dose.

Mean infliximab dose decreased significantly from 224 mg (95% CI 212 to 236 mg) at the start to 130 mg (95% CI 105 to 154 mg) after 1 year. Expressed in mg/kg, the dose decreased from 3.0 mg/kg (SD 0.2) to 1.7 mg/kg (SD 1.1). Median DAS28 increased statistically significantly from 2.5 (p25–75=2.0–2.9) to 2.8 (p25–75=2.2–3.6) after 1 year in the whole group (p=0.002). An increase in DAS28 was most noticeable in patients in whom infliximab dose could not be tapered; however, differences between groups (discontinued, down-titrated and not down- titrated) were not significant (p=0.14). Fifty per cent of the patients in whom infliximab could not be down-titrated were re-treated with the original infliximab dose within the first three visits. Of note, no infusion reactions to infliximab were observed during follow-up, with no changes in infliximab infusion intervals.

In this cohort, no statically significant clinical factors could be found to predict successful down-titration (table 2). Based on explained variance, smoking, disease duration and DAS28 at baseline were investigated in multivariate analyses revealing an explained variance of 10%, p=0.08. In addition, methotrexate use was analysed in different sequences with two of the abovementioned factors without improving explained variance.

Table 2

Predictive clinical baseline factors for successful down-titration

Extra corticosteroids were given in 8% (95% CI 5 to 10) of all 421 visits, which was in one-third of all patients. In 1 patient, prednisolone was stopped. In addition, three patients received corticosteroid injections because of bursitis (two subacromial and one trochanteric) and one patient because of a trigger finger. In 15 visits (3.5%) a DMARD change occurred, which was in 10 patients. In seven patients, the DMARD dosage was decreased or the DMARD was stopped; in two patients the dose was increased; and in one patient methotrexate changed into sulfasalazine due to a pregnancy wish. In three patients, a different biologic was started. All these patients flared after the first down-titration and reached low disease activity again after dose increase, but unfortunately, secondary failure to infliximab warranted a change in therapy (two were treated with rituximab and one with etanercept). In 42 visits (10%) NSAIDs were changed, with NSAID initiation in 36 visits. RA-related work absenteeism was reported by only one patient for two days. Patients did not visit the outpatient clinic more frequently during down-titration than in the year before infliximab down-titration (p=0.16).

There was no statistically significant difference in QoL after down-titration. The mean difference in EQ5D was −0.031 (95% CI −0.067 to 0.005). If the EQ5D scores were analysed per down-titration group no statistically significant difference was found (p=0.152). The mean reduction in costs over the whole year was €3474 (95% CI 2457 to 4492) per patient. A gradual decline in costs per 4 months was observed (figure 3). If costs in the last 4 months are compared with costs prior to down-titration, the reduction would be €5689 per patient per year, extrapolated to subsequent years. There is no statistical difference in EQ5D between the 4 time-points (p=0.19).

Figure 3

Mean direct costs in euros every 4 months. Start of down-titration was at 0.


Our study demonstrated that down-titration or discontinuation of infliximab is possible in the majority of RA patients with stable low disease activity and stable treatment without a relevant change in DAS28 and QoL during a 1-year follow-up. Throughout this year, the direct costs were substantially reduced, compared with the year before, by more than €3000, and this relates to €5689 per patient for the following years if we assume that the direct costs of the last 4 months can be extrapolated. In addition, most flares seem to occur within the first 10–12 weeks after each dose down-titration step, and most patients who cannot discontinue already flared after the dose decreased to 75% or 50%.

However, it can be assumed that down-titration has some undesirable effects. First, disease activity can increase. In our cohort, a small median increase in DAS28 of 0.3 was observed, with a tendency towards a larger increase in DAS28 for patients in whom down-titration failed. However, this was not significant and might also partially be explained by regression-to-the-mean effects, because all patients had a DAS28<3.2 at study start. Second, an increase in the use of other RA medication besides infliximab can occur. We mainly observed a small increase in corticosteroid and NSAID use. DMARD dose-increases or changes were relatively uncommon. Third, down-titration might bring about an increase in radiographic damage as a result of flares.27 Besides radiographic follow-up as incorporated in daily medical care, no structural data was collected and therefore no conclusions can be drawn. In addition, since anti-TNFα treatment is associated with a reduced incidence of cardiovascular events discontinuing therapy might abolish this effect.28 ,29 Then again, whether anti-TNFα therapy itself or minimising the systemic inflammation is the cause of the reduced incidence remains to be settled beyond doubt.

Another noteworthy possible down-side of infliximab down-titration might be anti-infliximab antibody formation, since previous studies demonstrated an association between low serum levels of TNFα-inhibitor and the presence of anti-TNFα-inhibitor-antibodies leading to secondary failure or infusion-related allergic reactions.30,,32 In addition, since in our cohort 20% of the patients are not simultaneously treated with other immunomodulators next to infliximab, the risk of antibody formation is possibly further increased.33 As having anti-infliximab antibodies is expected to lead to loss of therapeutic effect and occurrence of side effects like allergic infusion reactions, it is reassuring that no infusion reactions in our cohort have been observed so far. However, there were three patients in whom biologic treatment had to be changed due to secondary failure. Further follow-up of all patients is required to assess whether in the future more patients will experience secondary failure. Of note, an alternative explanation on antibody formation could be that since the level of anti-infliximab antibodies is the result of antibody formation and (antigen-bound) antibody clearance, lowering the dose of TNFα-inhibitors (antigen) reveals already present antibodies rather than inducing new antibody formation.

If the benefits of down-titration are considered, dose-related side effects of infliximab are expected to decline. Although inconsistent and troubled with a precision problem, there is evidence for an increased risk of side effects with higher doses of anti-TNF blockers suggesting dose-related side effects,11 ,34,,36 implying that a lower dose has a beneficial effect on the occurrence of side effects. Furthermore, down-titrating infliximab yielded a considerable cost-reduction, while preserving adequate disease activity control as demonstrated in our cohort. With ever-growing medical healthcare budgets, sensible allocation of resources is mandatory, and all costs that can be reduced might, for example, be used to improve other healthcare facilities.

When looking at internal validity of our study, a few limitations are noticeable. First, the relatively small group of patients might influence precision in secondary outcome measures. Then again, CI around DAS28 and the utilities seem reasonable. Second, a longer follow-up period is required, to be able to consider secondary ineffectiveness, cost reductions in subsequent years and number of side effects. Another drawback to our observational study is the lack of a control group in which infliximab dose remains unchanged. This hampers accurate conclusions on the effect of down-titration, among others, DAS28 difference, number of flares, other treatment changes, side effects and treatment failure. On the other hand, for the point-estimate of our primary outcome measure (percentage of successful down-titration) and for prediction of successful down-titration, no control group is necessary, because this would have been zero percent in a hypothetical control group in which infliximab is not down-titrated. In addition, since all patients in whom infliximab was tapered had stable DAS28 and were treated for at least 6 months prior to down-titration we can assume that on theoretical ground no flares, treatment changes or DAS28 increases should have occurred if the dose had not been reduced. As a result, one can assume that a control group could only further reinforce our findings compared with the current premise of no flares and medication changes in an usual-care context. In addition, some debate about the sensitivity of the measurement of QoL remains. The EQ5D might not be able to detect subtle differences in QoL due to down-titration. Then again, it has been validated extensively. Lastly, no validated flare criteria are available until now. In our down-titration protocol, the reversed European League against rheumatism (EULAR) response criteria were used, which have been used in previous studies.17 ,22 We modified the criteria to some extent to increase sensitivity.

To our knowledge, this is the first study to combine down-titration and discontinuation of biologic treatment. First, the other study on down-titration of infliximab tapered higher than the registered dose.22 In 89% of the patients, infliximab could be down-titrated from 5 mg/kg to 3 mg/kg. Another study demonstrated that 71% of the patients on adalimumab could down-titrate the dose until a flare occurred. In another study, etanercept down-titration from 2×/week 25 mg to 1×/week 25 mg was investigated and 73% remained in remission.17 ,37 Discontinuation of infliximab in the BeSt and RRR study was successful in 55–56%, which is a higher percentage than in our cohort.18 ,21 However, in the BeSt study patients had early RA opposed to long disease duration. Furthermore, they were methotrexate-naïve. Therefore, it can be postulated that methotrexate was responsible for reaching low disease activity.

Further research is needed to find predictors of successful down-titration to prevent unwarranted dose de-escalation, and to minimise harm to the patients. In addition, if patients who can discontinue biologic treatment are identified, no time-consuming down-titration protocol would be necessary, and treatment can be discontinued straightaway. Analysis in our cohort revealed no significant or relevant predictive factors. Two comments can be made. First, due to a small sample size, large multivariate analyses were not possible and CI for univariate analyses were large. Second, other possible candidates for prediction should be included in future analysis, such as (anti-)infliximab serum trough levels, ultrasonography or positron emission tomography. Besides prediction, future research on down-titration needs the incorporation of an induction phase, consisting of lowering the dose and a short follow-up to monitor flares, and a maintenance phase, to monitor long-term cost and side-effect reductions and secondary failure. Furthermore, down-titration protocols for other biologics, including non-TNFα blockers, can be envisioned.

In conclusion, down-titrating, or even discontinuing infliximab is possible in the majority of the patients with long-standing low disease activity, resulting in a considerable cost reduction and possible safety gains. No relevant deterioration of disease activity compared with start of down-titration was observed in these patients; however, predictors for successful down-titration are warranted.


The authors would like to thank the rheumatologists and specialist nurses of the department of rheumatology of Sint Maartenskliniek, Nijmegen, for their support.



  • Competing interests Piet van Riel has received grants from Merck, Pfizer, Abbott, BMS, Roche. The other authors have no competing interests to report.

  • Patient Consent Obtained.

  • Provenance and peer review Not commissioned; externally peer reviewed.