Circulating concentration of infliximab and response to treatment in ankylosing spondylitis: Results from a randomized control study


  • identifier: NCT00439283.



A minority of patients with ankylosing spondylitis (AS) fail to respond to infliximab treatment. This study compared the circulating infliximab concentration and the presence of clinical symptoms in patients continuously treated with infliximab or after treatment interruption.


Patients with active AS were randomly assigned at week 0 to receive infliximab either at weeks 4, 6, 10, and then every 6 weeks (continuous treatment), or at weeks 4, 6, and 10 and then upon symptom recurrence (on-demand treatment). The circulating concentration of infliximab was determined early during treatment and at weeks 46 and 52 for the continuous treatment group or upon relapse for the on-demand group. Response in the continuous treatment group was defined at week 58 using the ASsessment in AS International Working Group Criteria for 20% improvement.


Among the 93 patients in the continuous treatment group, treatment failure was not associated with a low circulating concentration of infliximab, either during early treatment or at 1 year. Eleven (39.2%) of the 28 nonresponders had an infliximab concentration of >10 μg/ml at week 52, whereas 9 (13.8%) of the 65 responders had an infliximab concentration of <1 μg/ml. In the on-demand group, the infliximab concentration at relapse closely correlated with the time to relapse. However, 24 (36.9%) of 65 patients had a resurgence of clinical symptoms at an infliximab concentration of >10 μg/ml, whereas 25 patients (38.4%) had a relapse at an infliximab concentration of <0.5 μg/ml.


Responsiveness to infliximab treatment is highly heterogeneous among individuals with AS, and this parameter overcomes the circulating infliximab concentration to explain treatment success or failure.


Infliximab is an anti–tumor necrosis factor α (anti-TNFα) chimeric monoclonal antibody (mAb) approved for the treatment of rheumatoid arthritis (RA), ankylosing spondylitis (AS), Crohn's disease, ulcerative colitis, and psoriasis. Despite the potent efficacy of infliximab in these disorders, a minority of treated patients fail to improve. Among this minority of patients, monitoring trough serum levels of infliximab could be helpful to distinguish those patients with an insufficient circulation concentration of the antagonist from those with a form of the disease refractory to TNFα neutralization. In RA patients, large interindividual heterogeneity of serum infliximab levels was observed from the first weeks of treatment, and incomplete response was frequently associated with low circulating levels of infliximab (1–5). This explains why increasing the dose of infliximab restored efficacy in many patients (1). Loss of efficacy of infliximab treatment in patients with psoriasis also correlated with low serum concentration of the mAb (6). In contrast, in patients with AS, little is known regarding how circulating infliximab levels correlate with response to treatment. We are aware of 1 published study monitoring circulating infliximab concentrations in AS patients treated with infliximab (7). In this single-center study involving 38 patients, treatment failure was associated with a low concentration of the TNFα antagonist. Furthermore, decreased infliximab concentrations correlated with the presence of antiinfliximab antibodies, which was associated with nonresponse. This suggested that, as in patients with RA, the inefficacy of infliximab in patients with AS is mostly attributable to an insufficient circulating concentration of the therapeutic mAb.

In the present study, we took advantage of having conducted a large, prospective, multicenter study of infliximab treatment in patients with AS (8) to further investigate the relationship between the response to treatment and the circulating infliximab concentration. This relationship was addressed by combining a dual approach, as follows: in continuously treated patients, we compared serum infliximab levels between responders and nonresponders; in patients in whom treatment was interrupted, we determined serum infliximab concentrations at the time of relapse. Both approaches suggest that disease heterogeneity, rather than the circulating infliximab concentration, contributes to the response to treatment.


Adults (>18 years old) in France with a diagnosis of AS were included in a prospective, randomized, multicenter, 1-year, comparative study, as previously described (8). At the inclusion visit (week 0), patients were randomly assigned to either the continuous treatment group (Q6) or the on-demand treatment group. In addition, patients in the on-demand group were randomly allocated to receive or not receive methotrexate (MTX). All patients were scheduled to receive a loading regimen of infliximab consisting of 3 infusions at weeks 4, 6, and 10, respectively. Thereafter, patients in the Q6 group received an infliximab infusion every 6 weeks until week 52, whereas patients in the on-demand group received a supplementary infusion only upon relapse, up to week 54, with a minimum interval of 4 weeks between 2 infusions (see below for the definition of relapse). For each infusion, infliximab was administered intravenously at a standard dose of 5 mg/kg in 250 ml NaCl 0.9%, over a 2-hour period. The dose of infliximab was increased to 7.5 mg/kg per infusion in the Q6 group, starting no earlier than week 40, for those patients exhibiting signs of relapse at 2 consecutive visits.

In the group receiving MTX, it was administered orally as a single weekly dose. A starting dose of 2.5 mg of MTX was given 4 weeks before the first infliximab infusion, and then the dose increased weekly by 2.5 mg, up to a maximum dose of 12.5 mg, which was to be continued throughout the study.

Considering the objective of the study (a comparison between the infliximab dosage and the response to treatment), we decided to measure infliximab concentrations only in patients in whom the efficacy of treatment was evaluated at week 58. As detailed in Figure 1, this group included both patients who completed the 1-year trial and those who discontinued treatment before week 58 because of lack of efficacy (n = 98 patients in the Q6 group and n = 92 patients in the on-demand group). Patients who discontinued treatment because of adverse events (n = 29) or for personal reasons or noncompliance (n = 28) were excluded. Of the 92 patients in the on-demand group, 3 had no relapse during the followup period and were therefore also excluded due to the absence of serum samples after week 10. Of the 187 patients selected, a few could not be analyzed for technical reasons, such as the lack of several serum samples during the followup period or difficulties identifying samples. Based on these circumstances, 5 patients in the Q6 group and 13 patients in the on-demand group were excluded (Figure 1). There is no rationale suggesting that the missing values from these 18 patients biased the findings of the study. It should also be noted that the profile at entry of the 169 patients analyzed in the present study is nearly identical to that of the 247 patients enrolled in the main study (compare Table 1 with Table 1 in ref. 8). Overall, of the 124 patients included in the Q6 group, 93 were analyzed for infliximab concentrations, and of the 123 patients included in the on-demand group, 76 (34 receiving MTX and 42 not receiving MTX) were studied.

Figure 1.

Randomization of the patients with ankylosing spondylitis studied for circulating infliximab concentrations. Q6 = continuous treatment group (patients receiving infliximab every 6 weeks); MTX = methotrexate; W = week.

Table 1. Characteristics of the patients at the time of inclusion, according to treatment group*
CharacteristicQ6 (n = 93)On demand
All (n = 76)MTX (n = 34)No MTX (n = 42)
  • *

    Except where indicated otherwise, values are the number of patients/number of patients assessed. Q6 = continuous treatment; MTX = methotrexate; BASDAI = Bath Ankylosing Spondylitis Disease Activity Index; BASFI = Bath Ankylosing Spondylitis Functional Index; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein.

Male sex, no. (%)69 (74.2)61 (80.3)25 (73.5)36 (85.7)
Age, mean ± SD years39.5 ± 12.040.8 ± 10.038.3 ± 7.442.9 ± 11.4
Disease duration, mean ± SD years12.3 ± 9.715.0 ± 9.213.5 ± 6.316.2 ± 10.9
HLA–B27 positive69/8758/6829/3229/36
History of peripheral arthritis, no. (%)55 (59.1)45 (59.2)24 (70.6)21 (50.0)
History of uveitis, no. (%)25 (26.9)25 (32.9)10 (29.4)15 (35.7)
History of psoriasis, no. (%)14 (15.1)7 (9.2)2 (5.9)5 (11.9)
BASDAI, mean ± SD6.1 ± 1.46.1 ± 1.35.9 ± 1.36.3 ± 1.3
BASFI, mean ± SD5.3 ± 2.15.6 ± 2.05.5 ± 1.95.7 ± 2.1
ESR, mean ± SD mm/hour35.2 ± 23.829.6 ± 19.634.3 ± 22.326.0 ± 16.53
CRP level, mean ± SD mg/liter30.3 ± 24.229.0 ± 20.932.2 ± 21.926.4 ± 19.9

Response and relapse definitions.

The 20% improvement response according to the ASsessment in Ankylosing Spondylitis International Working group criteria (ASAS20) was used to assess clinical outcome (9). An ASAS20 responder was defined as a patient showing ≥20% improvement from baseline (week 0) and absolute improvement from baseline of ≥1 unit (on a scale of 0–10) in 3 of the 4 ASAS domains (patient global assessment of disease activity, global assessment of pain, Bath Ankylosing Spondylitis Functional Index [10], and inflammation [mean of the 2 morning stiffness–related Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) scores] [11]), and the absence of deterioration from baseline by ≥20% and by ≥1 unit in the fourth domain.

Relapse was determined according to the following procedure. Starting at the time of inclusion, all patients were required to call a toll-free automated phone server once every week until completion of the study in order to perform a guided self-evaluation. First, patients gave a positive or negative answer to the following questions: “Since the last connection, do you think that your disease has remained under control?” and “Since the last connection, do you think that your disease has been worsening?” Second, patients answered questions on the French version of the BASDAI, using an 11-point numerical rating scale. Finally, patients entered a value (from 0–10) corresponding to their global estimate of pain over the last week. Relapse was defined as a negative answer to the first question, a positive answer to the second question, and either an increase in the BASDAI score of ≥1 or an increase in the global pain score of ≥2 as compared with the lowest score reached by that patient since the first infliximab infusion. Whenever relapse was detected ≥4 weeks after the last infusion in the on-demand group, the patient was urged to contact the treating center to schedule an infusion as soon as possible. Notification was also sent to the treating center, so that an infusion could be planned for that patient in the coming week.

Measurement of trough serum infliximab levels.

Serum samples were obtained at each visit, before any infusion was administered, and then were stored at −20°C until being processed for infliximab dosage. Serum levels of infliximab (human/mouse chimeric mAb, IgGκ isotype) were determined in duplicate by enzyme-linked immunosorbent assay. Briefly, 96-well plates were coated with mouse anti-human TNFα mAb (clone B-F7), and preincubated with recombinant human TNFα (both from Diaclone, Besançon, France). Clone B-F7 and infliximab recognize different epitopes on the human TNFα molecule. After extensive washing, diluted (1:1,000) serum was added. Captured infliximab was revealed using horseradish peroxidase–conjugated anti-human IgG mAb (clone MK1A6; Binding Site, Birmingham, UK) and 3,3',5,5'-tetramethylbenzidine as the chromogen. After blocking reaction, absorbance was read at 450 nm. The standard curve obtained from serial dilutions of infliximab standard was linear between concentrations of 1 ng/ml and 33 ng/ml. Interassay and intraassay variations of the test were both <20%.

Statistical analysis.

Quantitative variables are presented as the mean ± SD or the median and interquartile range, and categorical variables are presented as the frequency and percentage. If the infliximab concentration value was missing at week 52, the value measured at week 46 was used. If there were 2 groups, concentrations of infliximab were compared using the Mann-Whitney test. P values less than 0.05 were considered significant. If there were 3 groups, 3 pairwise comparisons using the Mann-Whitney test were performed. After Bonferroni correction, P values less than 0.017 were considered significant. All statistical analyses were performed with SAS software, version 9.1 (SAS Institute, Cary, NC).


The characteristics of the 169 patients included in this subgroup analysis of the larger, multicenter study are described in Table 1. Trough concentrations of circulating infliximab were determined in the Q6 group at weeks 6, 10, 16, 46, and 52. In agreement with previous studies in RA (2, 5), the highest concentrations were observed during the first weeks of treatment, and then they decreased to reach a median steady-state concentration of 6.7 μg/ml at the end of the followup period. However, large interindividual variation was observed at all evaluation points (Figure 2A).

Figure 2.

Infliximab concentration in patients with ankylosing spondylitis (AS). Circulating concentration of infliximab was determined in AS patients treated continuously (n = 93) (A), or on-demand (n = 76) (B). Results are expressed as the median, interquartile, 10th and 90th percentiles (A), and as the mean + SEM (B). MTX = methotrexate.

We analyzed whether this heterogeneity was related to clinical outcome. Responders were defined as patients fulfilling the ASAS20 criteria at week 58, without any dose increase of infliximab treatment during the 1-year trial. Unexpectedly, we found that early circulating infliximab concentrations at week 6, 10, or 16, were lower in responders than in nonresponders. This difference was significant at weeks 6 and 10. There was, however, a large overlap between the 2 groups. No determination was available at week 46 or week 52 for 4 patients who withdrew from the study before week 46 because of treatment failure. Infliximab concentrations at week 46 or week 52 in the other 89 patients did not differ between responders and nonresponders (Table 2). Of the 28 nonresponders, 11 (39.2%) had infliximab concentrations of >10 μg/ml at week 52. In contrast, 9 (13.8%) of the 65 responders had a serum infliximab concentration of <1 μg/ml. Therefore, in AS patients continuously treated with infliximab, treatment failure is usually not explained by insufficient concentrations of circulating infliximab. Conversely, treatment success can be achieved even in patients with relatively low circulating infliximab concentrations.

Table 2. Serum infliximab concentrations (μg/ml) in responders and nonresponders in the continuous treatment group*
nMedian (IQR)nMedian (IQR)
  • *

    IQR = interquartile range.

Week 66124.5 (15.1–33.3)2832.5 (26.3–38.3)0.014
Week 106317.8 (9.2–26.4)2822.4 (18.0–32.1)0.050
Week 166210.1 (4.1–16.2)2816.0 (7.5–22.2)0.064
Week 46577.7 (4.0–14.3)208.6 (1.7–14.3)0.986
Week 52627.1 (4.0–12.3)2210.1 (0.4–15.7)0.923

We also determined infliximab concentrations in the on-demand group. During early treatment, serum infliximab concentrations in the on-demand group were similar to those observed in the Q6 group, with no difference between the patients receiving MTX and those not receiving MTX in addition to infliximab (Figure 2B). At relapse, the serum infliximab concentrations closely correlated with the moment of relapse, i.e., the earlier the relapse, the higher the serum infliximab concentration (Table 3). Of the 65 patients experiencing relapse who were tested, 24 (36.9%) had a resurgence of clinical symptoms at a serum infliximab concentration of >10 μg/ml, but 25 patients (38.4%) experienced a relapse at a concentration of <0.5 μg/ml. This indicates that the concentration of infliximab needed to control AS varied within a large range between individuals, and this variance may explain the large differences in the time to relapse after the last infusion of infliximab.

Table 3. Infliximab concentrations (μg/ml) at the time of relapse in the on-demand group*
 Time of relapse
Week 16 or beforeWeek 17 to week 22Week 23 or after
  • *

    MTX = methotrexate; IQR = interquartile range.

  • P = 0.025 versus no MTX.

  • P = 0.025 versus week 17 to week 22, and P < 0.0001 versus week 23 or after.

  • §

    P < 0.01 versus week 23 or after.

 No. of patients8912
 Median (IQR)25.55 (17.26–42.93)4.18 (0.39–11.16)0.40 (0.22–0.67)
No MTX   
 No. of patients16137
 Median (IQR)4.18 (0.39–11.16)1.60 (0.46–9.51)0.26 (0.15–0.40)
 No. of patients242219
 Median (IQR)15.84 (6.79–29.32)2.71 (0.39–11.16)§0.27 (0.21–0.55)

Overall, patients who received MTX and those who did not receive MTX did not differ in terms of the moment of relapse (P = 0.4) or the infliximab concentration at this moment (P = 0.7). However, in the subgroup of patients experiencing a relapse before week 16, the serum infliximab concentration at the time of relapse was higher in patients receiving MTX (P = 0.025) (Table 3).


Determination of the circulating concentration of TNFα antagonists in patients with inflammatory disorders may contribute to understanding the mechanism of treatment failures with these biologic therapies. Monitoring trough serum levels of infliximab in patients with RA and those with psoriasis showed that an incomplete response is frequently associated with low circulating levels of infliximab (1–6). The development of antiinfliximab antibodies is often associated with low circulating concentrations of infliximab, which may predispose a patient to treatment inefficacy (4, 6, 7, 12–15). The addition of MTX increases the frequency of a response to infliximab treatment in patients with RA, and this may in part relate to suppressing the formation of antibodies to infliximab (12, 16, 17). In patients with RA, the serum infliximab concentration correlates with treatment efficacy at the population level, even though a number of patients with RA fail to experience improvement despite reasonably high (1–10 μg/ml) or even high (>10 μg/ml) serum infliximab concentrations. This failure to improve with treatment suggests that, in some patients with RA, the inflammatory process is at least in part independent of TNFα. It would explain why increasing the dose of infliximab administered to patients with RA does not consistently rescue the response to treatment (1) and why, in patients switching from one TNFα antagonist to another, failure with the first treatment predisposes to failure with the second treatment (18).

Among patients with AS treated with infliximab, 61–94% experience a response after 6–12 months of treatment (8, 19–24). In most cases, maintenance therapy allows for the sustained control of disease in patients with an initial response (22), and discontinuation of infliximab results in relapse; the time to relapse varies considerably at the patient level (8, 19, 23). Reintroduction of infliximab in patients experiencing relapse is generally safe and as efficient as the initial treatment (23). The addition of MTX in patients treated with infliximab on demand does not significantly improve the clinical response rate, nor does it increase the number of infliximab infusions administered (8).

As in patients with other inflammatory disorders, the lack of response to TNFα neutralization in patients with AS may result from insufficient circulating levels of the antagonist or from an intrinsic resistance of the disease to TNFα neutralization. Among continuously treated patients, we compared serum infliximab levels between responders and nonresponders. In patients interrupting treatment, we determined serum infliximab concentrations at the time of relapse. Consistent results were obtained in both groups, showing that concentration of the antagonist did not significantly account for treatment success or failure. Therefore, in both groups, many patients were without symptoms despite low levels of circulating infliximab, whereas other patients had persisting symptoms despite a high concentration of infliximab.

In the group of patients who received continuous treatment, we observed considerable heterogeneity in serum infliximab concentrations between individuals, which was observed as early as 2 weeks after the first infusion of infliximab and persisted to the end of the followup period. However, this heterogeneity did not explain the response to treatment, because therapeutic failure, which was evaluated at week 58, was not associated with low concentrations of circulating infliximab at any moment during the followup period. Actually, we observed higher infliximab concentrations early during treatment in nonresponders as compared with responders. The reason for this observation is unclear. At week 58, several patients with AS did not experience a response to treatment, despite having high serum infliximab concentrations. In other words, the sensitivity of their disease to TNFα neutralization appeared to be poor. Conversely, disease in several patients was controlled despite relatively low (<1 μg/ml) serum infliximab concentrations; this demonstrates that, in some patients, AS may be exquisitely sensitive to TNFα neutralization. The finding of controlled inflammation in patients with low levels of circulating infliximab has already been documented in patients with RA (2). However, low levels of circulating infliximab in patients with RA appeared to be the dominant cause of treatment failure. This observation contrasts with our findings in patients with AS. Neutralizing antibodies against infliximab may increase the clearance of infliximab (25, 26). Measuring antiinfliximab titers may therefore be useful to discriminate, among patients who fail to respond to treatment and with low infliximab concentrations, whether or not this latter finding is attributable to the development of antiinfliximab antibodies. The development of antibodies to infliximab was not determined in our study, because most of the nonresponding patients with AS displayed a relatively high concentration of infliximab, indicating that the development of antiinfliximab antibodies may not be a dominant cause of treatment failure compared with disease heterogeneity.

Determination of the infliximab concentrations in patients in whom treatment was interrupted after the first 3 infusions (on-demand group) confirmed that responsiveness to TNFα neutralization differs markedly between patients with AS. Within this difference of responsiveness among patients with AS, as anticipated, there was a close inverse relationship between the time to relapse and the infliximab concentration at relapse. A minority of patients (24 of 65) fulfilled the criteria for relapse within 6 weeks after the third loading infusion (week 10). Actually, these patients could be considered as poor responders rather than true relapsers. Such an early failure seems indeed to reflect resistance of their disease to TNFα neutralization, because high concentrations of circulating infliximab were observed in these patients. In the other patients in whom relapse occurred later, there was considerable heterogeneity in the time to relapse and the infliximab concentration at relapse. This indicates that in AS patients initially controlled by a short course of treatment, persistence of this control is dependent on circulating levels of infliximab. However, the heterogeneity also indicates that the minimal infliximab concentration needed to control the disease markedly differs between individuals. In fact, many patients (25 of 65) experienced disease control until their serum infliximab concentration fell below 0.5 μg/ml. A comparison of disease control in this group with that in the group of patients who experienced early relapse with high infliximab concentrations illustrates the substantial heterogeneity among patients with AS regarding disease responsiveness to TNFα neutralization.

MTX treatment had no impact on the time to relapse, providing evidence against a direct beneficial effect of MTX in AS, as already reported (8, 27). MTX treatment had no effect on serum infliximab concentrations at weeks 6 and 10. However, we observed that among patients experiencing a relapse before week 16, serum infliximab concentrations at relapse were higher in those who received MTX than in those who did not receive MTX. Treatment with MTX and other immunosuppressive agents has been shown to increase serum concentrations of infliximab, possibly by preventing the development of antiinfliximab antibodies (12–16). This may explain the higher infliximab concentrations in early relapsers. However, this effect of MTX on the infliximab concentration did not improve control of the disease, which is consistent with the hypothesis that early relapsers present with a form of AS largely refractory to TNFα neutralization by infliximab.

Taken together, these findings show that responsiveness to infliximab treatment is highly heterogeneous among patients with AS, and that this parameter overcomes the serum infliximab concentration to explain treatment success or failure. Our results are similar to those recently reported for AS treatment with another TNFα antagonist, etanercept (28). Such an interpretation is at variance with that from a former study conducted by the same group of investigators in a smaller population of AS patients treated with infliximab (7). It should be stressed, however, that in the latter study, the frequency of nonresponders (47%) was unusually high. Furthermore, if several of the nonresponders had undetectable serum levels of infliximab, the mean trough serum infliximab concentration 54 weeks after the initiation of treatment was not markedly different between responders and nonresponders (8.2 μg/ml and 6.3 μg/ml, respectively). Larger studies evaluating either etanercept (28) or infliximab (current study) suggest that a minority of patients with AS present a form of the disease largely resistant to TNFα neutralization, whereas many other patients are highly sensitive to TNFα antagonists. Variable mechanisms of the disease may account for such large heterogeneity in the response of AS to this treatment. Further studies are needed to evaluate whether monitoring of the drug concentration improves characterization of AS in patients receiving biologic therapies, and whether pharmacogenetic or pharmacogenomic parameters contribute to explain treatment failures. It may also be useful to measure the trough serum concentration of infliximab in case of treatment failure, to evaluate the opportunity of increasing the infliximab dose. If the serum level is quite high, as observed in most patients failing to respond to treatment in our study, it is probably unnecessary to increase the dose.


Dr. Emilie had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Krzysiek, Breban, Ravaud, Henry, Trape, Dougados, Emilie.

Acquisition of data. Krzysiek, Ravaud, Prejean, Roy.

Analysis and interpretation of data. Krzysiek, Breban, Ravaud, Roy, Emilie.

Manuscript preparation. Krzysiek, Breban, Ravaud, Roy, Dougados, Emilie.

Statistical analysis. Ravaud, Roy.

Development of assay. Krzysiek, Wijdenes, Barbey, Emilie.


Schering-Plough approved the content of the manuscript, and publication of this article was not contingent on the approval of Schering-Plough.