To investigate the relationship between serum concentrations of infliximab, a monoclonal anti–tumor necrosis factor α antibody, and clinical improvement from infliximab therapy for rheumatoid arthritis (RA).
To investigate the relationship between serum concentrations of infliximab, a monoclonal anti–tumor necrosis factor α antibody, and clinical improvement from infliximab therapy for rheumatoid arthritis (RA).
Multiple blood samples were obtained from each of 428 subjects with active RA who were enrolled in a multicenter, randomized, double-blind, placebo-controlled trial (ATTRACT [Anti–Tumor Necrosis Factor Trial in Rheumatoid Arthritis with Concomitant Therapy]) evaluating the clinical efficacy and safety of infliximab therapy. Serum levels of infliximab were measured by enzyme-linked immunosorbent assay. Dose-response trends were analyzed using generalized logistic regression techniques. Pharmacokinetic modeling was used to predict the serum concentrations of infliximab after simulated infusions using doses and dosing intervals not evaluated in the trial.
At week 54, 26% of the subjects receiving 3 mg/kg infliximab every 8 weeks had undetectable trough serum levels of infliximab, a significantly greater proportion than in the other 3 treatment groups (P < 0.001). Increased magnitude of American College of Rheumatology (ACR) response (measured by the ACR-N, a continuous measure of clinical improvement derived from the ACR 20% response criteria) and greater reduction from baseline in serum C-reactive protein level were both associated with higher trough serum concentrations of infliximab (P < 0.001), as was less progression of radiographic joint damage (P = 0.004), providing support for a dose-response relationship. Pharmacokinetic models predicted that decreasing the dosing interval from 8 weeks to 6 weeks would yield higher trough serum levels of infliximab than increasing the dose by 100 mg.
These results suggest that some patients with RA may benefit from infliximab given at higher doses than 3 mg/kg or more frequently than every 8 weeks.
Results from the Anti–Tumor Necrosis Factor Trial in Rheumatoid Arthritis with Concomitant Therapy (ATTRACT), a multicenter, randomized, double-blind, placebo-controlled trial in 428 patients with active RA, have shown that infliximab therapy, when added to weekly methotrexate (MTX), significantly improves clinical, laboratory, and radiographic measures of disease activity (1, 2). In ATTRACT, clinical benefit was found using each of the 4 infliximab dosing regimens: 3 mg/kg every 4 or 8 weeks and 10 mg/kg every 4 or 8 weeks. The American College of Rheumatology 20% (ACR20) response rates at week 54 (3) for the infliximab-treatment groups ranged from 42% to 59%, significantly higher than the 17% rate for the placebo group (2). The infliximab-treatment groups also showed comparable reductions from baseline in serum levels of C-reactive protein (CRP) and in the radiographic progression of joint disease (2).
A comparison of response rates across infliximab-treatment groups suggests a dose response at the lower end of the dosing range. Although the ACR20 response rates were not significantly different among the infliximab-treatment groups, subjects receiving 3 mg/kg infliximab every 8 weeks had a significantly lower ACR50 response rate (4) than did subjects in the 3 higher-dosage groups after 54 weeks of treatment (2). Anecdotally, treatment with 3 mg/kg infliximab every 8 weeks has been associated in some cases with disease flares toward the end of the 8-week interval between infusions. Current labeling in the US allows for infliximab dosing of 3–10 mg/kg every 4–8 weeks, but no studies thus far have addressed possible strategies for dose escalation in the event of inadequate clinical improvement.
We hypothesize that inadequate treatment responses can result from incomplete suppression of tumor necrosis factor α (TNFα) activity. To test this hypothesis, we analyzed treatment outcomes from ATTRACT as a function of serum infliximab concentrations. We found that 22–30% of patients in the group receiving 3 mg/kg every 8 weeks had undetectable trough serum levels of infliximab before each infusion from week 22 through week 54. Further analysis showed that clinical, laboratory, and radiographic improvement was greater with higher trough serum concentrations of infliximab, suggesting that some patients may benefit from receiving infliximab at higher doses than 3 mg/kg or more frequently than every 8 weeks. We also used pharmacokinetic modeling to predict trough serum concentrations of infliximab using a range of doses and dosing intervals not directly tested in the clinical trial. These models provide direction for changing the infliximab dose or shortening the dosing interval to increase the trough serum levels of this agent predictably and thus to increase the likelihood of an improved or sustained clinical response.
Patients. The eligibility for ATTRACT has been described elsewhere (1, 2). Briefly, patients were eligible if they had active RA and were taking at least 12.5 mg/week MTX. Active RA was defined as the presence of ≥6 tender joints, ≥6 swollen joints, and at least 2 of the following: morning stiffness of at least 45 minutes' duration, erythrocyte sedimentation rate of at least 28 mm/hour, and a serum CRP level of at least 2.0 mg/dl.
Study protocol. In ATTRACT, 428 patients were randomly allocated to 5 treatment groups: placebo (n = 88), 3 mg/kg infliximab every 8 weeks (n = 86), 3 mg/kg infliximab every 4 weeks (n = 86), 10 mg/kg infliximab every 8 weeks (n = 87), or 10 mg/kg infliximab every 4 weeks (n = 81). Subjects were initially treated with intravenous infusions of infliximab/ placebo at weeks 0, 2, and 6, and then subsequently every 4 or 8 weeks for 54 weeks. All of the subjects maintained their baseline dose of MTX except when the dose was reduced for suspected MTX toxicity. Other disease-modifying antirheumatic drugs were not allowed during the trial. Nonsteroidal antiinflammatory drugs and low doses of prednisone (≤10 mg/day) were maintained at the baseline dose. An institutional review committee from each participating center approved the study, and the subjects provided written informed consent before entry.
Clinical response was evaluated immediately before each infusion using the ACR definition of clinical improvement (3). The effect of therapy on the progression of radiographic damage was assessed using the van der Heijde modification of the Sharp scoring system (5, 6). The ACR-N (a continuous measure of clinical improvement derived from the ACR20 response criteria) was used to quantify the clinical response as a continuous variable for each subject. It was equal to the smallest percentage of improvement from baseline among the following variables: the number of tender joints, the number of swollen joints, and the median improvement from baseline in the other 5 measures of disease activity (7).
Infliximab serum assay. Blood samples for the measurement of serum infliximab concentrations were collected immediately prior to each infusion through week 6 and then every 8 weeks through week 54. In addition, samples were collected 1 hour after the completion of the infusions at weeks 0, 2, 6, and 14. Pharmacokinetic data from individual subjects were excluded from analysis following either a missed infusion or receipt of <80% of a scheduled dose. Table 1 shows the number of samples available for the postinfusion assays. The numbers of samples in each group that were eligible for preinfusion analysis at weeks 2, 6, 14, 22, 30, 38, 46, and 54, respectively, were as follows: 3 mg/kg every 8 weeks (n = 86, 84, 78, 71, 67, 63, 59, and 57), 3 mg/kg every 4 weeks (n = 86, 84, 81, 76, 75, 69, 64, and 63), 10 mg/kg every 8 weeks (n = 87, 86, 84, 80, 77, 71, 68, and 68), and 10 mg/kg every 4 weeks (n = 76, 80, 76, 68, 65, 61, 60, and 58).
|3 mg/kg||10 mg/kg|
|Every 8 weeks (n = 86)||Every 4 weeks (n = 86)||Every 8 weeks (n = 87)||Every 4 weeks (n = 81)|
|No. of samples||86||86||85||80|
|No. of samples||86||85||85||79|
|No. of samples||83||84||86||78|
|No. of samples||75||79||80||72|
Serum infliximab levels were measured using an enzyme-linked immunosorbent assay as previously described (8). The lowest level of infliximab that could be reliably detected was 0.1 μg/ml.
Modeling the serum profiles of infliximab. Each of the dosing regimens was modeled based on the serum infliximab concentrations obtained prior to the scheduled infusion at weeks 0, 2, 6, 14, 22, 30, 38, 46, and 54. The median pharmacokinetic parameters for the study population were used to simulate the infliximab concentration profile in the serum for predictive purposes. Simulations were generated for serum infliximab concentrations at doses of 4.5, 6, and 7.5 mg/kg body weight. In addition, we simulated the serum concentrations for a change in dosing interval from 8 weeks to 6 weeks at the 3 mg/kg dose.
In the group treated with 3 mg/kg infliximab every 8 weeks, we modeled profiles for 21 subjects who had undetectable trough levels of infliximab at week 30. This treatment group had the highest proportion of subjects with undetectable serum levels of infliximab before the next dose. Their serum concentrations of infliximab were profiled using nonlinear modeling software, WINNONLIN (Pharsight, Mountain View, CA). At the 3 mg/kg dose level, the most appropriate model was a monoexponential model with a first order elimination and a zero order infusion input as described in the following equation: Cp(t) = [(D/ ti)/(V · α)] · (e−αt − e−αt*). In this equation, D is the dose; Cp(t) represents the serum concentration of infliximab at time t; ti represents the duration of infusion; t* = t − ti if t is >ti, but if t is ≤ti, then t* is assumed to be = 0; V is the volume of distribution; and α is the first order rate of elimination of infliximab.
Serum infliximab concentration profiles for each subject were individually fit to the equation by an iterative Nelder-Mead technique of error minimization (9). Models were initiated with parameters from prior single dose studies and iterated to minimize intraindividual errors, thus generating a best-fit profile for each subject. Simulations were based on two scenarios: 1) addition of 100 mg infliximab (single vial) to the calculated doses for the subjects at week 22 of treatment and 2) change of the dosing interval from 8 weeks to 6 weeks at a 3 mg/kg dose. To estimate the new dose for each individual, 100 mg was divided by the body weight (kg) of the subject and added to the original dose for that subject. The new dose was used in simulation with the precise parameters originally obtained for that subject.
Statistical analysis. Fisher's exact test was used to compare the proportion of subjects responding to treatment. Continuous response parameters, such as the ACR-N, were compared using one-way analysis of variance and t-test on the van der Waerden normal scores. All statistical testing was done using 2-sided tests. Dose-response trends were tested with the linear component of orthogonal contrast in linear regression or generalized logistic regression. Simple linear regression was used to detect correlation between two variables.
Peak serum concentrations of infliximab. The serum concentration of infliximab obtained 1 hour after infusion approximates the Cmax for infliximab. At week 0, the median serum concentrations of the 1 hour–postinfusion samples were 68.6 μg/ml and 219.1 μg/ml for subjects receiving 3 mg/kg and 10 mg/kg infliximab, respectively, every 8 weeks (Table 1). The median serum concentration was 3.2-fold higher at this time point for the 10 mg/kg dose compared with the 3 mg/kg dose. Based on these findings, we conclude that the maximum infliximab serum concentrations are directly proportional to the intravenous dose over this range. The observed Cmax approximates the predicted serum concentration in which the total dose of infliximab occupies only the plasma. The higher median postinfusion concentrations of infliximab at weeks 2 and 6 compared with baseline can be explained by the residual amounts of the drug from the prior infusion.
Trough serum concentrations of infliximab. The median trough serum concentrations were relatively constant for each of the 4 treatment groups from week 14 through week 54 (Figure 1). The group receiving 10 mg/kg every 4 weeks had the highest median trough serum levels. The group receiving 3 mg/kg every 8 weeks had the lowest trough levels. The median trough levels were comparable for the groups receiving 10 mg/kg every 8 weeks and 3 mg/kg every 4 weeks, and these levels were intermediate with respect to the highest- and lowest-dosage groups. The trough levels represent the nadir concentration measured immediately prior to the next infusion. Thus, the subjects were exposed to much higher concentrations of infliximab for most of the 4- or 8-week intervals between infusions.
The trough serum concentrations of infliximab showed considerable individual variability for subjects in the same dosage group (Figure 1). From week 22 to week 54, we found that 22–30% of subjects in this group had serum infliximab trough levels below the quantifiable limit of the serum infliximab assay (<0.1 μg/ml) (Figure 2). A significantly greater proportion of subjects in this group had undetectable trough serum levels of infliximab than in the other 3 treatment groups (P < 0.001). In comparison, undetectable trough levels were noted for 9–15% of subjects in the group receiving 3 mg/kg every 4 weeks and for 5–9% of subjects in the group receiving 10 mg/kg every 8 weeks during the same time period. Only 2 subjects in the group receiving 10 mg/kg every 4 weeks had undetectable trough serum levels.
Trough serum levels of infliximab and magnitude of ACR response. Although ACR20 response rates were not significantly different among the 4 infliximab-treatment groups, ACR50 and ACR70 response rates showed a trend toward increasing with increasing doses of infliximab (Figure 3). As previously reported, ACR50 response rates were significantly lower in the group receiving 3 mg/kg every 8 weeks (21.5%) than in the groups receiving 3 mg/kg every 4 weeks (34.1%), 10 mg/kg every 8 weeks (40%), and 10 mg/kg every 4 weeks (37.5%) (2). We reasoned that the differential rates of clinical improvement could result from variable elimination of infliximab from the body.
To address this question, we compared clinical responses among groups of subjects with different trough serum levels of infliximab (<0.1, ≥0.1–≤1, >1–≤10, and >10 μg/ml) at week 54. The highest percentage of subjects with less than an ACR20 response was found in the group with undetectable trough serum levels of infliximab (<0.1 μg/ml) (Figure 4). Conversely, the highest proportion of ACR50 and ACR70 responders had the highest trough serum levels (>1–≤10 and >10 μg/ml). Of note, some subjects achieved an ACR50 or ACR70 response despite undetectable trough serum levels of infliximab (<0.1 μ g/ml) (Figure 4), illustrating the imprecise nature of this relationship.
There was a trend toward higher ACR20 response rates with increasing trough serum levels of infliximab (Figure 5). We tested this relationship by analyzing ACR-N in comparison with trough serum levels of infliximab at week 54. The ACR-N is a continuous measure of clinical improvement derived from the ACR20 response criteria. Using regression analysis, we found that a higher ACR-N was significantly associated with a higher trough serum concentration of infliximab (P < 0.001). Figure 5 shows that 75% of subjects with trough serum infliximab levels of >10 μg/ml achieved an ACR20 response at week 54, and that this concentration was only observed for ACR20 responders in the 3 higher dose groups. This ACR20 response rate of 75% for those with the highest trough levels may correspond to the maximal response rate that can be expected with sustained modulation of TNFα activity.
We also used regression analysis to determine the relationship between laboratory and radiographic measures of disease activity and trough serum concentrations of infliximab at week 54. A greater reduction from baseline in serum CRP levels was associated with higher trough serum concentrations of infliximab (P < 0.001). The primary radiographic outcome in this trial was the change in total Sharp score between weeks 0 and 54. An increase in the total Sharp score from baseline corresponds to progression of radiographic joint damage. We found significantly less progression from baseline in the total Sharp score with higher trough serum concentrations of infliximab (P = 0.004). Overall, these results suggest that favorable treatment outcomes depend in part on maintaining a sufficient trough serum level of infliximab.
Pharmacokinetics modeling. Increasing the infliximab dose or shortening the interval between infusions can produce higher trough serum levels of infliximab. Pharmacokinetic modeling was performed to predict the median trough serum concentrations for infliximab doses and dosing intervals not tested in the trial. We initially profiled the predicted serum concentrations using the median values observed for the groups receiving 3 mg/kg and 10 mg/kg every 8 weeks (Figure 6). Excellent correlation was attained between the simulated and observed trough levels of infliximab. This model was applied for a theoretical 70-kg individual receiving 3 mg/kg infliximab every 8 weeks. Accordingly, a 100-mg increase in the infliximab dose would raise the trough level from 0.8 μg/ml to 1.8 μg/ml. In comparison, decreasing the interval from 8 weeks to 6 weeks while maintaining a 3 mg/kg dose would increase the trough level from 0.8 μg/ml to 2.8 μg/ml. Using directly measured concentrations from week 22 through week 54, we can show that further reduction in the dosing interval from 6 weeks to 4 weeks increases trough levels of infliximab to an even greater extent. Evidence for this point is provided by the observed median trough levels of infliximab of subjects in the group receiving 3 mg/kg every 4 weeks; these levels range from 7.6 μg/ml to 8.8 μg/ml, a 3-fold increment in trough levels compared with the 6-week interval.
We also modeled 21 patients receiving 3 mg/kg infliximab every 8 weeks who had undetectable trough serum levels of infliximab at week 30. Subjects in this group were more likely than those with higher trough levels to achieve an inadequate clinical response and potentially would benefit the most from boosting trough levels. Pharmacokinetic modeling was used to predict the trough serum levels of infliximab for each patient, assuming a 100-mg increase in dose or a reduction in dosing interval from 8 weeks to 6 weeks. A 100 mg–higher dose raised the median trough levels of infliximab from 0.03 μg/ml (interquartile range [IQR] 0.005–0.062) to 0.046 μg/ml (IQR 0.007–0.088). In comparison, shortening the infusion interval from 8 weeks to 6 weeks increased the median trough concentrations of infliximab from 0.03 μg/ml (IQR 0.005–0.062) to 0.213 μg/ml (IQR 0.065–0.363). Overall, pharmacokinetic modeling showed that shortening the dosing interval from 8 weeks to 6 weeks in these subjects raised trough concentrations of infliximab more than would have been expected by increasing the dose by 100 mg.
Infliximab is a chimeric anti-TNFα monoclonal antibody that binds to soluble and transmembrane TNFα with high affinity (Ka = 1010M−1), forming a stable complex that blocks the association of this cytokine with its receptor (10). Inhibition of TNFα is achieved in vitro with infliximab concentrations of 0.2–10 μ g/ml (10). Infliximab is distributed primarily into the intravascular space and has a terminal half-life of ∼9.5 days (11), which is less than the reported 21-day half-life of human IgG1 (12). Our results show that infusions of 3 mg/kg and 10 mg/kg infliximab at weeks 0, 2, and 6 afford predictable and dose-proportional serum drug concentrations. At week 54, we found significantly greater improvement in ACR responses, reduction in serum levels of CRP, and less radiographic progression of joint damage with higher trough serum concentrations of infliximab. These findings warrant support for an empirical trial of higher or more frequent infliximab doses for patients with inadequate treatment responses. Our pharmacokinetic models suggest that shortening the dosage interval can increase the trough serum levels of infliximab more than raising the dose by 100 mg, and this approach may therefore be the preferred strategy for dose modification to improve clinical responses.
The response rates among treatment groups did not clearly diverge until week 54, when subjects receiving 3 mg/kg every 8 weeks were found to have a significantly lower ACR50 response rate than subjects in the other 3 infliximab-treatment groups (2). There was also a trend toward lower ACR20 and ACR70 response rates in this group, but the differences in these levels of response across the 4 infliximab-treatment groups did not reach statistical significance. Notably, after week 30, the proportion of ACR20 and ACR50 responders waned in the group receiving 3 mg/kg every 8 weeks, which contrasts with the relatively constant rates of response of the other 3 infliximab-treatment groups over time. However, ACR50 response rates for the combined 3 mg/kg and 10 mg/kg dose groups at week 10 were 21.5% (35/163) and 20% (33/165), respectively. These similar response rates are probably a product of the induction regimen and the high serum levels of infliximab generated for both dosage groups through week 10.
The relatively low proportion of ACR50 responders in the group receiving 3 mg/kg every 8 weeks may have resulted from insufficient exposure to infliximab during the treatment course. At week 54, the trough serum levels of infliximab varied more than 100-fold in this group, implying marked differences among subjects in exposure to infliximab. The basis for this individual variability is unknown, but it may be an effect of antibodies to infliximab for some subjects and of metabolic differences for others. In a previous study, patients with RA who were treated with 1 mg/ kg infliximab (without MTX) developed a high incidence of antibodies to infliximab (8). This group also cleared infliximab more rapidly from the circulation than did other treatment groups with a lower incidence of antibodies to infliximab (8).
It is unlikely that antibodies to infliximab are the sole explanation for the rapid elimination of serum infliximab. Following completion of the present study, antibodies to infliximab were detected in 25 patients (8.5%) with analyzable samples; these antibodies were distributed across all infliximab-treatment groups (Wagner CL: unpublished observations). The appropriate serum samples were available for measuring antibodies to infliximab in 19 of the 21 subjects (from the group that received 3 mg/kg every 8 weeks) previously described as having undetectable trough levels of infliximab. Of these 19 subjects, only 6 tested positive for serum antibodies to infliximab, while 2 had inconclusive results. Thus, metabolic factors probably account for the rapid elimination of infliximab in some patients.
The relationship between clinical improvement and trough serum levels of infliximab has implications for therapy, but also some caveats. The potential adverse consequences of raising the dose of infliximab must be kept in mind when contemplating dosage adjustment. Although there was a trend toward a slightly higher proportion of serious adverse events in the group receiving 3 mg/kg every 4 weeks compared with the group receiving 3 mg/kg every 8 weeks (16% versus 11%), this incidence was no higher than that in the placebo group (21%). We cannot predict which patients might benefit from dose escalation based on any known clinical, laboratory, or radiographic factors. Also, the relationship between trough levels of infliximab and clinical improvement is imprecise. At week 54, 6 of 28 subjects (21%) in the trial achieved an ACR50 response or better despite trough infliximab levels of <0.1 μg/ml. Therefore, trough serum levels of infliximab are not, by themselves, reliable predictors of treatment response. We would not advocate the routine measurement of serum infliximab levels for this reason.
The maximal improvement from raising the dose of infliximab is unknown. The ACR20 response rate of 75% for subjects with trough serum levels of infliximab >10 μg/ml may approximate the best possible efficacy of infliximab therapy. In the remaining cases, lack of treatment response is probably not due to an insufficient dose of infliximab, but rather to a particular disease characteristic (e.g., joint inflammation not primarily driven by TNFα) or other host factor that precludes an adequate response to therapy.
Our results support initial dosing of infliximab at 3 mg/kg for weeks 0, 2, 6, and 10. Beyond week 14, some patients may require empirical adjustment of the infliximab dose or dosing interval to improve the clinical response. Criteria for the appropriate selection of candidates for dose escalation are unclear. However, a reasonable candidate might be a patient who has an initial treatment response, but who later experiences an increase in disease activity with extension of the treatment interval to 8 weeks. Practically speaking, dose increases are usually done in 100-mg increments, the size of a vial of drug. Alternatively, the infusions may be given more frequently than every 8 weeks.
The predicted trough serum concentrations for different doses and dosing intervals may be extrapolated from the pharmacokinetic models. Shortening the dosing interval from 8 weeks to 6 weeks produces a larger increment in trough serum concentration of infliximab (an increase of ∼2.0 μg/ml) than increasing the dose by 100 mg (an increase of ∼1.0 μg/ml). Further shortening of the interval from 6 weeks to 4 weeks without changing the 3 mg/kg dose would generate an even higher increment in trough level (an increase of ∼5 μg/ml). Interestingly, higher trough serum concentrations may be achieved using a regimen that administers slightly less infliximab over time. Patients treated for 6 months with 3 mg/kg every 6 weeks (4 infusions) would receive a total of 12 mg/kg of infliximab. In comparison, patients treated with 3 mg/kg and an additional 100-mg vial during the same period would receive 4.4 mg/kg infliximab (assuming a 70-kg individual) every 8 weeks (3 infusions) for a total dose of 13.3 mg/kg.
These modeling predictions may not generalize to the treatment population as a whole, because nonresponders tend to eliminate infliximab more rapidly from the circulation than do responders. As shown by additional modeling, the 21 subjects (in the group receiving 3 mg/kg every 8 weeks) with undetectable trough levels would require an even larger dose increase or further shortening of the dosing interval to achieve the same trough level than would the average patient.
In summary, these studies provide further evidence that clinical improvement from infliximab therapy is dose dependent and that higher trough levels of infliximab may be beneficial for treatment of certain cases of RA. The pharmacokinetic models afford a useful guide for empirically adjusting the dose of infliximab to achieve the maximal therapeutic effect for an individual patient. Further investigations are needed to more precisely define the efficacy and safety of various dose-escalation strategies.
The authors wish to thank Professor Ravinder N. Maini for his critical review of the manuscript and the other members of the ATTRACT Study Group for their valuable contributions to the conduct of this clinical trial.