Sustained improvement over two years in physical function, structural damage, and signs and symptoms among patients with rheumatoid arthritis treated with infliximab and methotrexate

Authors


  • Drs. Breedveld, Furst, van der Heijde, Kalden, Maini, Smolen, St.Clair, and Weisman have received consultancies and/or honoraria from Centocor, Inc. Drs. Marsters and Maini have stock ownership or options in Centocor, Inc. In 2002, Dr. Maini received a fee for giving expert testimony to the FDA during a hearing on inhibition of structural damage. The Kennedy Institute of Rheumatology has a patent and a research and licensing agreement from Centocor, Inc., for the use of infliximab in rheumatoid arthritis, for which it has received royalties. As co-inventor, Dr. Maini receives a percentage of the royalties.

Abstract

Objective

To evaluate the efficacy and safety of repeated administration of infliximab plus methotrexate (MTX) over a 2-year period in patients with rheumatoid arthritis (RA) who previously experienced an incomplete response to MTX.

Methods

Four hundred twenty-eight patients were randomly assigned to receive MTX plus placebo or infliximab at a dose of 3 or 10 mg/kg plus MTX for 54 weeks, with an additional year of followup. The protocol was later amended to allow for continued treatment during the second year. Of 259 patients who entered the second year of treatment, 216 continued to receive infliximab plus MTX for 102 weeks. Ninety-four of these 259 patients experienced a gap in therapy of >8 weeks before continuing therapy. Infusions were administered at weeks 0, 2, and 6, followed by treatment every 4 weeks or every 8 weeks (alternating with placebo infusions in the interim 4-week visits) at a dose of 3 or 10 mg/kg for a total of 102 weeks (including the gap in therapy). For safety and efficacy assessments, data on the patients who were randomized to receive treatment, irrespective of whether treatment was administered for 102 weeks, were evaluated using all actual observations available. The efficacy measures included the Health Assessment Questionnaire (HAQ) (physical function), Short Form 36 health survey (SF-36) (health-related quality of life), total radiographic scores (structural damage), and the American College of Rheumatology 20% improvement criteria (ACR20) (signs and symptoms).

Results

The infliximab plus MTX regimens resulted in significantly greater improvement in HAQ scores (P ≤ 0.006) and SF-36 physical component summary scores (P ≤ 0.011) compared with the MTX-only group. There also was stability in the SF-36 mental component summary score among patients who received the infliximab plus MTX regimens. Median changes from baseline to week 102 in the total radiographic score were 4.25 for patients who received the MTX-only regimen and 0.50 for patients who received the infliximab plus MTX regimen. The proportion of patients achieving an ACR20 response at week 102 varied from 40% to 48% for the infliximab plus MTX groups compared with 16% for the MTX-only group.

Conclusion

Throughout 102 weeks of therapy, infliximab plus MTX provided significant, clinically relevant improvement in physical function and quality of life, accompanied by inhibition of progressive joint damage and sustained improvement in the signs and symptoms of RA among patients who previously had an incomplete response to MTX alone.

Treatment of rheumatoid arthritis (RA) with disease-modifying antirheumatic drugs (DMARDs), including weekly methotrexate (MTX), which is the current gold standard for treatment of RA, fails to control disease activity, prevent structural damage, and maintain quality of life in a large proportion of patients (1–3). In patients with RA who have an incomplete response to MTX therapy, the addition of infliximab, a recombinant IgG1κ monoclonal antibody specific for tumor necrosis factor (TNF), has been proven to be efficacious and safe over a 1-year period of time (4–7). More than half of the patients receiving the infliximab plus MTX regimen for 1 year have experienced significant improvement in the signs and symptoms of the disease and also significant inhibition of progressive damage to the articular structures as demonstrated by radiography. The therapeutic effect of infliximab plus MTX over longer periods of time has not been demonstrated previously.

The pivotal multinational phase III clinical study known as the Anti-TNF Trial in Rheumatoid Arthritis with Concomitant Therapy (ATTRACT) was designed to address 3 co-primary end points of prolonged efficacy. The results for 2 of these end points, the control of signs and symptoms of RA at 30 weeks and inhibition of structural damage assessed by radiographs of the hands and feet at 54 weeks, have been published previously (4, 7). In this report, we investigate the ability of a regimen of infliximab plus MTX to provide a sustained benefit in physical function as assessed by the Health Assessment Questionnaire (HAQ) (8), as well as prolonged improvement in the signs and symptoms and inhibition of radiographic progression over a 2-year period of time in patients with RA.

PATIENTS AND METHODS

Eligibility.

The eligibility criteria and design of the study have been described in detail elsewhere (4, 7). Briefly, patients participated in the study from March 31, 1997 through March 9, 2000. Men and women between the ages of 18 and 75 years were eligible for the study if they had active RA despite receiving treatment with at least 12.5 mg/week of MTX. Active RA was defined in the protocol by the presence of 6 or more swollen joints, 6 or more tender joints, and at least 2 of the following criteria: morning stiffness that lasted at least 45 minutes, an erythrocyte sedimentation rate (ESR) of at least 28 mm/hour, and a serum C-reactive protein (CRP) concentration of at least 2 mg/dl. Patients who completed the first 54 weeks of treatment were given the opportunity to receive treatment through 102 weeks based on a protocol amendment instituted in September of 1998.

Study protocol.

The patients remained on the same stable dose of MTX that they had been receiving on a weekly basis since before the time of study enrollment, and this was administered in combination with infusions of placebo or infliximab at doses of 3 mg/kg or 10 mg/kg. The combination regimen was administered through 54 weeks and followup, for a total of 102 weeks. No other DMARDs were permitted. All patients received intravenous infusions at the initiation of treatment (week 0) and at weeks 2 and 6. Two groups of patients were administered infliximab plus MTX (one group receiving 3 mg/kg and the other receiving 10 mg/kg infliximab) and had subsequent infusions every 4 weeks, whereas 2 other groups of patients received infliximab plus MTX (one receiving 3 mg/kg and the other receiving 10 mg/kg infliximab) but had subsequent infusions every 8 weeks, with placebo infusions at the interim 4-week visits. The MTX-only group received placebo infusions every 4 weeks. An amendment to the protocol was approved while the study was in progress but before the study data were unblinded, which thereby allowed patients to continue treatment through the second year.

Patients were allowed to continue the same dose of nonsteroidal antiinflammatory drugs and oral glucocorticoids (prednisone ≤10 mg/day) that they had been taking at study entry. The study protocol was approved by the institutional review committee at each participating center, and each patient gave his or her written informed consent.

Clinical evaluations.

Arthritis-related functional disability was measured every 4 weeks using the HAQ, a validated self-administered form that assesses functional ability in a variety of areas, including the ability to dress, arise, eat, walk, maintain personal hygiene, reach, and grip, on a scale ranging from 0 (no difficulty) to 3 (unable to perform the activity) (8). In the analysis of the primary end point, HAQ values were averaged over time through week 102 and calculated so that positive values indicated a decrease in arthritis-related functional disability from baseline (less disability than at baseline). This value was obtained by subtracting the followup scores from the baseline scores and calculating the average change in the HAQ score over 2 years. Thus, patients who had reduced disability after baseline had positive results, and patients who had increased disability had negative results.

Health-related quality of life was assessed by the Medical Outcomes Study Short Form 36 health survey (SF-36) as described previously (9). Eight aspects of health status were assessed: general and mental health, physical function, social function, physical and emotional health, pain, and vitality; the score on each subscale ranged from 0 (worst) to 100 (best). For the physical and mental component summary scores, a score of 50 is the average score in the US population (10). The physical and mental component summary scores were calculated using standard methods (9). Higher scores indicate a better health-related quality of life. The changes from baseline averaged over time through week 102 were calculated for the SF-36 physical and mental component summary scores and for the 8 individual subscale scores.

For both the HAQ and SF-36 scores, the prespecified analyses mandated that visits after discontinuation of study followup and evaluations that indicated worsening (those with negative scores) were treated as a change from baseline of 0, as a way of imputing the missing results. With this analytic strategy, patients whose study participation was ongoing and whose condition worsened were handled in the same manner as those who had discontinued the study and received alternative therapies. Sensitivity analyses that allowed for worsening were also performed. Actual patient data were used for patients who discontinued the study infusions but completed the study followup. Data on 1 patient who did not have a baseline HAQ score were not included in the primary end point efficacy analysis. Data on 5 patients with missing or incomplete baseline SF-36 questionnaires were not included in the analysis of SF-36 data.

Radiographic evaluations.

The prevention of structural damage in the hands and feet was measured by the change from baseline in the Sharp radiographic damage score (van der Heijde modification of the Sharp scoring system [11]) at week 102. The modified Sharp/van der Heijde score, as previously described (12, 13), consists of a combined total of joint erosion scores and joint space narrowing scores. The joint erosion score is a composite measure of the severity of erosions in 32 joints of the hands and 12 joints of the feet. The scores increase with increasing numbers of erosions and increasing severity. The maximum erosion score obtainable is 280. The joint space narrowing score is a composite measure of the severity of joint space narrowing in 30 joints of the hands and 10 joints of the feet. The maximum joint space narrowing score obtainable is 160. Thus, the total radiographic score can range from 0 to 440.

The variable analyzed for the 2-year radiographic outcome was the average of the total radiographic scores provided by 2 blinded readers at week 102 minus the average of the 2 baseline scores. Agreement between the 2 readers was assessed as previously described (4, 7). Only data on patients with radiographs of both the hands and the feet at baseline and week 102 were included in these analyses. Positive changes from baseline were an indication of the progression of structural damage in the joints of the hands and feet at week 102 compared with that observed at baseline. Changes from baseline to week 102 also were assessed for erosions and joint space narrowing, which are the individual components of the total radiographic score. Prevention of new erosions was assessed by determining the number of newly involved joints (the number of joints per patient that changed from an erosion score of 0 at baseline to any higher score).

American College of Rheumatology (ACR) 20% improvement evaluations.

The numbers of tender and swollen joints were evaluated by an independent assessor who was blinded to the patient's treatment assignment. Sixty-eight joints were assessed for tenderness, and 66 joints were evaluated for swelling. A clinical response at week 102 was defined according to the ACR definition of 20% improvement (ACR20) (14). In the evaluation of ACR20 data, patients who discontinued treatment through week 54 or who had missing or incomplete data at week 102 were considered to be ACR nonresponders. Patients who completed the study through week 54 and chose not to enter the second year of the study had their week 54 data carried forward to the week 102 analysis. For patients who discontinued treatment in the second year of the study but continued through the study followup, and for patients who completed the study treatment, the actual week 102 data were included in the analysis. Patients treated during the second year who had protocol-prohibited changes in medication and/or a surgical joint procedure were considered to be nonresponders, so that any improvement related to these new interventions would not be attributed to their study treatment.

Adverse events.

Patients were monitored for the occurrence of adverse events, including serious adverse events, deaths, infections, infusion reactions, delayed hypersensitivity reactions, and malignancies, throughout their participation in the study. An infusion reaction was defined as any adverse event that occurred during an infusion or within 1 hour following completion of the infusion, or any adverse event that was considered by the investigator to be an infusion reaction. A possible delayed hypersensitivity reaction was defined as arthralgia and/or myalgia with fever and/or rash (that were not signs or symptoms of other recognized clinical syndromes) occurring 1–14 days after an infusion.

Laboratory evaluations.

Samples for measurement of the CRP were collected every 4 weeks in the first year and every 8 weeks in the second year of treatment. Tests to measure antinuclear antibodies (ANA), antibodies to double-stranded DNA (dsDNA), and antibodies to infliximab were performed on blood samples collected at baseline and at selected followup time points. Only patients who tested positive for ANA were tested for anti-dsDNA antibodies. Blood samples for the measurement of serum infliximab concentrations were collected before and following selected infusions.

Statistical analysis.

In order to control the overall Type I error in the study, the analysis plan prespecified a significance level of 0.025 for the test of the week 102 primary analysis, because there were additional primary analyses conducted at earlier time points. The null hypothesis tested in the primary analysis was that there would be no difference among the treatment groups in the population distributions for improvement in the HAQ scores averaged over time. Secondary efficacy analyses were performed at α = 0.05, and nominal P values were reported for secondary analyses. No adjustment for multiple comparisons was made for these nominal P values. All statistical testing was 2-sided.

Analyses of continuous variables were performed by comparing results among the treatment groups using an analysis of variance (ANOVA) on van der Waerden normal scores, a nonparametric, rank-based method (15). To control for Type I error, pairwise comparisons of each of the infliximab plus MTX treatment groups versus the MTX-only group were performed using linear contrasts, only if the test for an overall treatment effect was significant. The number of newly involved joints at week 102 was compared among the treatment groups by using the Cochran-Mantel-Haenszel ANOVA statistic. For categorical responses, the proportions of patients with a response at week 102 were compared among the treatment groups using the chi-square test. If this overall test yielded significant results, pairwise comparisons between the MTX-only group and each infliximab plus MTX group were performed using the same method.

Protocol amendment.

A protocol amendment to extend the duration of study treatment from 54 weeks to 102 weeks (2 years) was approved after the initiation of the study. This amendment was initiated following a change in the US Food and Drug Administration (FDA) guidance for treatment of RA (16), which requires 2 years of followup for assessment of improvement in physical function. Patients who did not enter the second year of the study were followed up through 102 weeks, and their quality-of-life assessments were recorded as the actual values observed, for the purpose of all analyses. Patients who had an increase in their dosage of corticosteroids or MTX or who began treatment with corticosteroids or DMARDs other than MTX after the end of the first year of their study participation were required to have a washout of these medications. To reinitiate therapy, the corticosteroid and MTX doses had to be the same or lower than the baseline doses, and DMARDs other than MTX had to be discontinued.

Approximately one-third of the patients who participated in the second year of the study (94 of 259) had a gap of more than 8 weeks between treatments, because of the timing of the protocol amendment relative to the time of completion of their first year of the study. Two hundred fifty-nine (61%) of the 428 study participants continued with the second year of treatment, comprising 28 (32%) of 88 patients in the MTX-only group and 231 (68%) of 340 patients in the infliximab plus MTX groups. Based on the positive outcome of the analyses at week 54, as well as ethical considerations and the recommendations of the Safety Monitoring Committee, the study protocol was amended to allow for unblinding of treatment, which revealed to patients whether they were receiving placebo or infliximab infusions; however, the actual infliximab dose and regimen for any patient were not unblinded. The implementation of this amendment occurred late in the study but before the database lock and after 2 years of study participation for most patients.

The unblinding of the treatment administered to these patients had a minimal impact on the overall results of the health assessments, because the HAQ and SF-36 scores were averaged over time. There was a total of 27 possible evaluations for patients whose data were included in the analysis of the primary end point. Accordingly, HAQ evaluations were potentially affected by unblinding of treatment for the 14 patients who had not completed 4% of the scheduled HAQ evaluations, 18 who had not completed 7% of the evaluations, 13 who had not completed 11% of the evaluations, and 8 who had not completed 15% of the evaluations. Nevertheless, for 375 patients (88%), the unblinding did not occur prior to completion of their HAQ evaluations. Therefore, the unblinding of treatment groups was not likely to have substantively affected the primary analysis at week 102, which was an evaluation of the change from baseline in the HAQ score averaged over 102 weeks. No statistical adjustment for treatment unblinding was made in the analysis of the HAQ scores, SF-36 scores, or ACR20 data. Because the radiograph readers remained blinded to the treatment allocation, there was no impact on the interpretation of radiographic data.

RESULTS

Disposition and baseline characteristics of the patients.

Four hundred twenty-eight patients were enrolled at 34 study centers in North America (19 US and 3 Canadian) and Europe (12). The status of the patients through 102 weeks is summarized in Figure 1. Of the 428 patients enrolled in the study, 312 patients completed therapy through the first year. While the study was in progress, a change in the US FDA guidance on treatments for RA was incorporated (16), which required a minimum of 2 years of followup for assessment of improvement in physical function. Therefore, the study protocol was amended to allow for an optional extension of treatment for a second year. Patients were offered the opportunity to continue therapy, regardless of whether they had ended their first year of treatment. Notably, however, patients who had not completed their first year of therapy were required to do so before electing to continue into their second year.

Figure 1.

Status of the patients with rheumatoid arthritis through week 102 of the study. Of the 428 enrolled patients, 259 entered the second year of treatment. Completion of treatment through week 102 was achieved by 55–68% of the infliximab plus methotrexate (MTX)–treated patients, compared with 16% of the placebo plus MTX–treated patients. Discont. = discontinued.

Of the 428 original patients who were randomized, 259 patients (61%) elected to continue into the second year of treatment, and 83% of these 259 patients constituted the group who completed treatment through week 102 of the study. Ninety-four of these 259 patients (36%) had a treatment gap of more than 8 weeks before continuing into their second year of therapy; the mean length of time that infliximab treatment was suspended during the treatment gap was 19.4 weeks. The remaining 165 patients continued receiving infusions and had no treatment hiatus.

The patients who had the treatment hiatus may have received other medications during that period. Eight patients were documented as receiving protocol-prohibited medications. Some of these patients may have been receiving multiple medications. Three patients received an increase in MTX dose. Six patients received an alteration in their glucocorticoid regimen. Although patients may have received DMARDs during the treatment hiatus, they did not receive TNF blockers. Patients who had received DMARDs during the hiatus were required to have a washout period before resuming treatment with infliximab plus MTX or MTX only for the second year. No adjustments for the treatment hiatus were made in the analyses.

Forty-three (17%) of the patients entering the second year of treatment discontinued treatment before week 102. A total of 331 patients (77%) who were originally randomized completed the prespecified protocol assessments through the 2-year study period. Nearly half of the overall study population (216 of the 428 patients) voluntarily continued study medication for the full 102 weeks, of whom the majority (202 of 216) were in the infliximab plus MTX groups.

All treatment groups in the study were unblinded, for ethical reasons, after the positive results of infliximab plus MTX treatment at week 54 were known; consequently, the alternating placebo infusions were discontinued. Patients who had been randomized to receive MTX only were allowed to continue receiving MTX or to discontinue therapy. Similarly, patients randomized to receive infliximab plus MTX were allowed to continue this regimen or to discontinue therapy. At the time of unblinding, 375 patients had completed 102 weeks of study assessments; however, 53 patients were still being evaluated. Of those 53 patients, 32 were in the infliximab plus MTX groups and 21 were in the MTX-only group. All 30 patients in the infliximab plus MTX groups completed evaluations through 102 weeks. Of the 21 in the MTX-only group, 14 completed the study followup with MTX only through 102 weeks, while 7 discontinued their participation in the study (after unblinding, the observations for these 7 patients were carried forward through week 102). For comparison, the baseline characteristics of these groups are shown in Tables 1 and 2.

Table 1. Baseline demographics and disease characteristics of patients at various study stages up to and following extension of the study to a second year (102 weeks)*
 All patients randomized (n = 428)Patients who completed 54 weeks of therapy (n = 312)Patients who entered second year of therapy (n = 259)Patients who completed 102 weeks of therapy (n = 216)Patients who did not complete therapy at week 102 (n = 212)Patients who returned for followup through week 102 (n = 331)
  • *

    Except where indicated otherwise, values are the median. VAS = visual analog scale; HAQ = Health Assessment Questionnaire; CRP = C-reactive protein.

Age, years545353545453
Sex, no. (%)      
 Female332 (78)242 (78)203 (78)172 (80)160 (75)258 (78)
 Male96 (22)70 (22)56 (22)44 (20)52 (25)73 (22)
Tender joint count313030313131
Swollen joint count202021211820
Pain VAS (scale 1–10)6.86.76.76.76.96.7
Duration of morning stiffness, minutes120120120120120120
HAQ score (scale 1–3)1.81.61.61.61.91.8
CRP, mg/dl2.62.42.52.42.72.6
Patient global health VAS (scale 1–10)6.36.06.16.06.86.1
Evaluator global health VAS (scale 1–10)6.36.26.16.06.56.2
Radiographic scores      
 Total radiographic score51.3 (n = 310)48.8 (n = 262)45.5 (n = 233)44.5 (n = 205)60.5 (n = 105)51.3 (n = 306)
  Erosion score23.5 (n = 319)21.8 (n = 270)22.0 (n = 240)21.5 (n = 209)26.5 (n = 110)23.5 (n = 315)
  Joint space narrowing score28.3 (n = 310)27.0 (n = 262)26.0 (n = 233)25.2 (n = 205)35.0 (n = 105)28.3 (n = 306)
Table 2. Baseline demographics and disease characteristics of patients grouped according to status (before and after unblinding, missing radiographic data, and change in medication)*
 All patients randomized (n = 428)Patients unblinded before their week 102 evaluation (n = 53)Patients unblinded after their week 102 evaluation (n = 375)MTX-only patients who discontinued study treatment after unblinding (n = 7)Patients excluded from week 102 radiographic analyses because of incomplete data (n = 119)Patients who changed medication during the study (ACR nonresponders) (n = 145)
  • *

    Except where indicated otherwise, values are the median. MTX = methotrexate; ACR = American College of Rheumatology (see Table 1 for other definitions).

Age, years545354495552
Sex, no. (%)      
 Female332 (78)41 (77)291 (78)6 (86)92 (77)108 (74)
 Male96 (22)12 (23)84 (22)1 (14)27 (23)37 (26)
Tender joint count313230203233
Swollen joint count202419201820
Pain (scale 1–10)6.86.66.86.97.17.0
Duration of morning stiffness, minutes120120120120120120
HAQ score (scale 1–3)1.81.61.81.61.91.8
CRP, mg/dl2.61.82.73.12.62.6
Patient global health assessment (scale 1–10)6.35.76.46.56.86.9
Evaluator global health assessment (scale 1–10)6.36.66.36.56.56.4
Radiographic scores      
 Total radiographic score51.3 (n = 310)41.3 (n = 50)53.5 (n = 260)97.0 (n = 5)NA56.0 (n = 92)
  Erosion score23.5 (n = 319)14.3 (n = 50)24.5 (n = 269)34.3 (n = 6)NA24.8 (n = 93)
  Joint space narrowing score28.3 (n = 310)19.8 (n = 50)29.8 (n = 260)37.0 (n = 5)NA32.3 (n = 92)

The baseline characteristics of the overall study population have been reported previously (4, 7). The decisions to extend treatment for a second year and to unblind the treatments for ethical considerations during the second year yielded the subsets of patients shown in Tables 1 and 2. These subsets are as follows: the 428 patients originally randomized, the 312 patients who completed treatment through week 54, the 259 patients who entered the second year of therapy, the 216 patients who completed the second year of therapy, the 212 patients who did not complete the second year of therapy, the 331 patients who completed study followup through week 102, the 53 patients who were unblinded before their week 102 evaluation, the 375 patients who were unblinded after their week 102 evaluation, the 7 patients who discontinued study treatment after unblinding, the 119 patients with incomplete radiographic assessments who were not included in the radiographic analyses at week 102, and the 145 patients who changed medications or had joint surgery during the study and who were therefore considered ACR nonresponders. Notably, the baseline characteristics of the subsets did not differ significantly from those of the overall patient population or from each other for all demographics. Similarly, the baseline characteristics of the patients for whom complete radiographic assessments were not available or who changed medication also were not significantly different from those of the entire population or from each other.

Efficacy results.

Physical function and quality of life.

The median changes from baseline in the HAQ scores and SF-36 physical component scores averaged over time through week 102 are displayed in Figure 2. The results of these intent-to-treat analyses demonstrated a significant improvement following treatment with each infliximab plus MTX regimen relative to the MTX-only regimen (HAQ P ≤ 0.006; SF-36 P ≤ 0.011). No dose response was evident with respect to the improvement in HAQ observed in the infliximab plus MTX groups.

Figure 2.

Median improvement from baseline in the Health Assessment Questionnaire (HAQ) scores and Short Form 36 (SF-36) physical component summary scores averaged over time through week 102. The improvement from baseline in physical function, assessed as the HAQ score averaged over time, was the prespecified primary efficacy end point at week 102. The patient's health-related quality of life was assessed using the SF-36 questionnaire. q4 wks = administered every 4 weeks after initial infusions; q8 wks = administered every 8 weeks after initial infusions. MTX = methotrexate.

The change from baseline in the SF-36 physical component summary score achieved in the overall infliximab plus MTX group was significantly greater than that achieved in the MTX-only group (P < 0.001). Notably, the results of analysis of 3 of the most highly weighted individual subscales of the overall physical component summary score demonstrated a significantly increased change from baseline when the results of the infliximab plus MTX–treated patients were compared with those of the MTX-only–treated patients (physical functioning subscale scores, role—physical subscale scores, and bodily pain subscale scores all P < 0.001) (Table 3). In addition, the physical functioning subscale scores averaged over time through week 102 were comparable for all infliximab regimens, although the group administered 3 mg/kg every 8 weeks showed the smallest improvement. The role—physical subscale scores were 0 for all treatment groups at baseline. The improvements in role—physical subscale scores in those patients receiving 3 mg/kg every 4 weeks and those receiving 10 mg/kg every 8 weeks were higher than those in patients receiving 3 mg/kg every 8 weeks and 10 mg/kg every 4 weeks.

Table 3. Baseline and change from baseline in the Short Form 36 (SF-36) scores over time through week 102*
SF-36 components and subscalesMTX + placeboInfliximab 3 mg/kg + MTXInfliximab 10 mg/kg + MTX
q8 weeksq4 weeksq8 weeksq4 weeks
  • *

    Values are the median. Where pairwise P values are shown, overall treatment effect P values were P ≤ 0.001. MTX = methotrexate; q4 weeks = administered every 4 weeks after initial infusion; q8 weeks = administered every 8 weeks after initial infusions, with placebo administered in the 4-week intervals.

Physical functioning subscale     
 Baseline score30.022.525.025.030.0
 Change from baseline3.48.713.113.610.3
 P versus MTX onlyReferent0.026<0.001<0.0010.001
Role—physical subscale     
 Baseline score0.00.00.00.00.0
 Change from baseline10.712.726.925.916.3
 P versus MTX onlyReferent0.216<0.001<0.0010.008
Bodily pain subscale     
 Baseline score31.022.031.031.031.0
 Change from baseline8.215.321.920.319.8
 P versus MTX onlyReferent0.024<0.001<0.001<0.001
General health subscale     
 Baseline score45.045.042.045.045.0
 Change from baseline2.15.38.87.26.4
 P versus MTX onlyReferent0.003<0.001<0.001<0.001
Vitality subscale     
 Baseline score35.030.030.035.030.0
 Change from baseline4.411.412.511.714.7
 P versus MTX onlyReferent<0.001<0.001<0.001<0.001
Social functioning subscale     
 Baseline score62.550.062.562.550.0
 Change from baseline3.811.712.711.413.5
 P versus MTX onlyReferent0.007<0.0010.002<0.001
Role—emotional subscale     
 Baseline score33.333.366.750.066.7
 Change from baseline0.07.40.012.60.0
Mental health subscale     
 Baseline score70.062.072.068.068.0
 Change from baseline4.26.04.96.26.2
Physical component summary score     
 Baseline score25.725.223.925.725.8
 Change from baseline2.84.66.86.96.7
 P versus MTX onlyReferent0.011<0.001<0.001<0.001
Mental component summary score     
 Baseline score48.546.849.947.647.9
 Change from baseline1.93.82.22.93.7

The general health subscale scores were relatively high at baseline and showed only a small degree of improvement at week 102 (Table 3). However, significant changes from baseline were observed in the vitality and social functioning subscale scores of the SF-36, and pairwise comparisons indicated that these scores were significantly improved in the infliximab plus MTX treatment regimens versus the MTX-only regimen (P ≤ 0.007) (Table 3).

There was no significant difference in the change from baseline in SF-36 mental component summary scores in the infliximab plus MTX–treated patients relative to scores for the MTX-only–treated patients (MTX-only median change 1.9 versus infliximab plus MTX regimens median changes 2.2–3.8; P = 0.191) (Table 3).

Radiographic progression.

Pairwise comparisons of the changes from baseline in the total radiographic score showed that the change from baseline to week 102 in each infliximab plus MTX treatment group was significantly (P < 0.001) lower than that in the MTX-only group (Table 4). These findings indicate that there was less progression of structural damage in patients who received an infliximab plus MTX regimen compared with those who received MTX only over a period of 2 years. Results were similar across the infliximab plus MTX regimens, with no evidence of a dose response. Significant reductions from baseline in both erosions (P < 0.001) and joint space narrowing (P < 0.001 for all infliximab dosages except 3 mg/kg every 8 weeks [P = 0.002]) were seen in the infliximab plus MTX–treated patients as compared with the MTX-only–treated patients (Table 4). In addition, there were fewer newly eroded joints per patient (P < 0.001) in the infliximab plus MTX treatment groups (mean number of newly eroded joints 0.9–1.2) than in the MTX-only group (mean 2.9 joints). There was considerable agreement in the radiographic assessments between the 2 readers, with between-reader correlations of 0.84 for baseline and 0.92 for week 102 measurements of total radiographic scores. Analyses of each reader's assessments independently confirmed the results of the main radiographic analysis.

Table 4. Change from baseline to week 102 in radiographic parameters among patients treated with MTX only compared with those treated with the infliximab plus MTX regimens*
Radiographic scoresMTX + placeboInfliximab 3 mg/kg + MTXInfliximab 10 mg/kg + MTX
q8 weeksq4 weeksq8 weeksq4 weeks
  • *

    See Table 3 for definitions.

  • Total radiographic score of the hands and feet is based on the van der Heijde–modified Sharp score. At baseline, the median radiographic scores were 61.8 (MTX plus placebo), 55.5 (infliximab 3 mg/kg plus MTX q8 weeks), 45.5 (infliximab 3 mg/kg plus MTX q4 weeks), 49.5 (infliximab 10 mg/kg plus MTX 8.8 weeks), and 48.8 (infliximab 10 mg/kg plus MTX q4 weeks).

Total radiographic score     
 No. of patients evaluated5058666966
 Mean ± SD change12.59 ± 20.051.02 ± 7.131.03 ± 11.651.14 ± 4.92−0.42 ± 6.10
 Median change4.250.430.501.000.00
 Interquartile range0.50, 17.50−1.50, 3.67−2.95, 3.05−0.50, 3.73−3.08, 1.93
 PReferent<0.001<0.001<0.001<0.001
Erosion score     
 No. of patients evaluated5159707168
 Mean ± SD change6.65 ± 11.200.25 ± 4.380.50 ± 9.050.43 ± 2.61−0.57 ± 3.85
 Median change2.000.000.130.50−0.04
 Interquartile range0.50, 10.63−1.77, 1.97−1.50, 2.00−1.00, 1.55−1.67, 0.65
 PReferent<0.001<0.001<0.001<0.001
Joint space narrowing score     
 No. of patients evaluated5058666966
 Mean ± SD change5.91 ± 10.590.77 ± 3.680.10 ± 4.810.63 ± 3.250.17 ± 3.60
 Median change2.000.500.000.000.00
 Interquartile range0.00, 8.50−0.50, 2.50−2.00, 1.50−0.50, 2.00−1.50, 1.00
 PReferent0.002<0.001<0.001<0.001

Clinical response.

At week 102, significantly more patients in each of the infliximab plus MTX treatment groups achieved an ACR20 improvement response (P < 0.001 for all infliximab plus MTX treatment groups), as well as ACR 50% (ACR50) (P < 0.001 for 3 mg/kg every 4 weeks and 10 mg/kg every 8 weeks, P = 0.003 for 3 mg/kg every 8 weeks, and P = 0.006 for 10 mg/kg every 4 weeks) and ACR 70% (ACR70) (P < 0.001 for 3 mg/kg every 4 weeks and 10 mg/kg every 8 weeks, P = 0.008 for 3 mg/kg every 8 weeks, and P = 0.011 for 10 mg/kg every 4 weeks) improvement responses, as compared with the MTX-only group (Table 5).

Table 5. Comparison of clinical responses among patients treated with MTX only and infliximab plus MTX regimens at week 102*
ResponseMTX + placeboInfliximab 3 mg/kg + MTXInfliximab 10 mg/kg + MTX
q8 weeksq4 weeksq8 weeksq4 weeks
  • *

    ACR20 indicates a decrease of at least 20% in the number of tender joints and a decrease of at least 20% in the number of swollen joints, along with a 20% improvement in 3 of the following: the patient's global assessment of disease status, the patient's assessment of pain, and the physician's global assessment of disease status, all of which were assessed with the use of visual analog scales (range 0–10 cm, with higher scores indicating poorer status or more severe pain), the Health Assessment Questionnaire estimate of disability, and the erythrocyte sedimentation rate or C-reactive protein concentration. The percentages of patients with an improvement of 50% (ACR50) and 70% (ACR70) according to the American College of Rheumatology (ACR) criteria were assessed in a similar manner. See Table 3 for other definitions.

ACR criteria*     
 ACR20     
  % of responders1642404840
  PReferent<0.001<0.001<0.001<0.001
 ACR50     
  % of responders621303620
  PReferent0.003<0.001<0.0010.006
 ACR70     
  % of responders110212010
  PReferent0.008<0.001<0.0010.011
Median % decrease in no. of swollen joints20.052.565.169.172.2
 PReferent<0.001<0.001<0.001<0.001
Median % decrease in no. of tender joints29.263.868.875.363.2
 PReferent0.005<0.001<0.001<0.001
Median % decrease in serum C-reactive protein8.445.458.755.773.3
 PReferent0.002<0.0010.002<0.001

A sensitivity analysis of the ACR20 data was performed, in which patients who discontinued study treatment because of an adverse event were considered nonresponders. The results of the sensitivity analysis were similar to the results of the analysis described above. Through week 102, the percentage decreases in the numbers of swollen and tender joints were significantly higher among the infliximab plus MTX–treated patients than among the MTX-only–treated patients (Table 5). Patients who received each of the 4 infliximab plus MTX treatment regimens experienced significantly greater decreases in CRP levels over 102 weeks of treatment when compared with patients receiving only MTX.

Safety results.

A summary of adverse events is provided in Table 6. Serious adverse events were reported by similar proportions of patients who received MTX only (33%) and infliximab plus MTX (29%). Approximately 2% of placebo infusions had an associated reaction compared with 3% of infliximab infusions. Infusion reactions led to discontinuation of infliximab plus MTX therapy in only 8 of the 183 infliximab infusions with an associated infusion reaction. Only 1 patient had a serious infusion reaction after receiving 13 previous infliximab infusions that had not been accompanied by a reaction. This patient, who received 3 mg/kg of infliximab plus MTX every 4 weeks, became dizzy and short of breath during the week 86 infusion, which was subsequently stopped. The patient was alert until the development of a seizure that resolved the same day; treatment with infliximab was discontinued. Another patient who received 3 mg/kg of infliximab plus MTX every 4 weeks had a serum sickness–like reaction at week 86, at which time the sixth infusion was administered during the second year of treatment. This patient continued to receive infliximab infusions with no further adverse events through completion of the study.

Table 6. Incidence of adverse events (AEs)*
Adverse eventMTX + placeboInfliximab 3 mg/kg + MTXInfliximab 10 mg/kg + MTX
q8 weeksq4 weeksq8 weeksq4 weeks
  • *

    AEs were classified according to the World Health Organization adverse reaction terminology. Two patients who were randomized to receive MTX only inadvertently received a partial dose of infliximab and thus were included in the lowest infliximab dose group (3 mg/kg q8 weeks) for the analysis of AEs. See Table 3 for other definitions.

  • Serious infections reported by 2 or more patients included skin ulcer (n = 2), sepsis (n = 2), and arthritis (n = 2) in the MTX-only group and pneumonia (n = 8), cellulitis (n = 6), skin ulceration (n = 4), sepsis (n = 4), bacterial infection (n = 3), arthritis (n = 2), pyelonephritis (n = 2), bronchitis (n = 2), and herpes zoster (n = 2) in the infliximab plus MTX groups.

  • Causes of death in the MTX-only group were left ventricle rupture resulting in cardiopulmonary arrest, intestinal gangrene, arrhythmia, and cardiopulmonary failure. In the infliximab plus MTX treatment groups, patient deaths were attributed to bilateral pulmonary embolism, ruptured abdominal aortic aneurysm, sepsis, cardiorespiratory arrest associated with disseminated tuberculosis, pulmonary fibrosis, coccidioidomycosis peritonitis, and presumed cardiac arrhythmia.

  • §

    Squamous cell carcinoma was reported in the patient receiving MTX only; a possible drug-related malignant melanoma was reported for a patient in the infliximab 3 mg/kg plus MTX q8 weeks group; basal cell carcinoma, rectal carcinoma with hepatic metastasis, and squamous cell carcinoma were reported for 3 patients in the infliximab plus MTX 10 mg/kg plus MTX q8 weeks group, and 2 cases of basal cell carcinoma, 1 lymphoma, 1 squamous cell carcinoma plus malignant melanoma, and 1 breast cancer were reported in 5 patients in the infliximab 10 mg/kg plus MTX q4 weeks group.

  • Observed number includes long-term safety followup for up to 3 years after the last study treatment. Expected number based on US cancer rates from the Surveillance, Epidemiology, and End Results (SEER) database for age- and sex-matched individuals; excludes squamous and basal cell skin cancers, which are not included in the SEER database.

No. (%) of patients with serious AEs28 (33)29 (33)20 (23)25 (29)26 (32)
No. (%) of patients with serious infections11 (13)10 (11)11 (13)11 (13)8 (10)
No. (%) of patients with serious infusion reactions001 (1)00
No. (%) of patient deaths4 (5)3 (3)2 (2)1 (1)1 (1)
No. (%) of patients with malignancies§1 (1)1 (1)03 (3)5 (6)
Number of observed cancer cases vs. number expected0 vs. 1.025 vs. 5.15

Four patients (5%) who received the MTX-only regimen died during the 102-week study period. The causes of death in the MTX-only group were left ventricle rupture resulting in cardiopulmonary arrest, intestinal gangrene, arrhythmia, and cardiopulmonary failure. All 4 deaths were judged not to be related to the study drug. Seven patients (2%) who were treated with infliximab plus MTX died during the 102-week study period. In the infliximab plus MTX treatment groups, patient deaths were attributed to bilateral pulmonary embolism, ruptured abdominal aortic aneurysm, sepsis, cardiorespiratory arrest associated with disseminated tuberculosis, pulmonary fibrosis, coccidioidomycosis peritonitis, and presumed cardiac arrhythmia. One additional death caused by fatal myocardial infarction occurred after the patient's study participation and was reported during the long-term safety followup (Table 6).

Malignancies were reported in 1 patient (1%) who received the MTX-only regimen and in 9 patients (3%) treated with the infliximab plus MTX regimens during the course of the study (Table 6). There are currently 3 infliximab plus MTX–treated patients and 2 MTX-only–treated patients who had malignancies diagnosed during the 3-year followup period (1 patient in the MTX-only group had a diagnosis during the followup period, and 1 patient in the MTX-only group had a diagnosis during the study and a second malignancy diagnosed during the followup period).

From 54 weeks to 102 weeks, there was an increase of ∼13% in the percentage of patients with ANA in the infliximab plus MTX group and an increase of ∼5% in the percentage of patients with anti-dsDNA antibodies, compared with a 6% increase in the percentage of patients with ANA and a 0% increase in the percentage of patients with anti-dsDNA in the MTX-only group. Only patients who tested positive for ANA were tested for anti-dsDNA antibodies. No additional cases of lupus-like syndrome were observed in the second year of the study, and only 1 patient experienced a lupus-like reaction during the entire study, as reported previously (7).

Eight percent of patients receiving the infliximab plus MTX regimens developed antibodies to infliximab during the 102-week study period. Despite the presence of antibodies, similar proportions of patients with antibodies to infliximab (40%) and without antibodies to infliximab (38%) achieved an ACR20 response.

Pharmacokinetic results.

Serum concentrations of infliximab were dose proportional, and reproducible patterns of serum infliximab concentrations were observed for all treatment regimens (Figure 3). Median peak and trough infliximab concentrations were stable over time for all treatment groups among patients who continued to receive infliximab infusions per protocol, and were in the ranges reported previously (5).

Figure 3.

Median serum infliximab concentrations over time. Preinfusion and 1-hour postinfusion samples were collected at weeks 0, 2, 6, 14, 62, and 102. Only preinfusion samples were collected at other time points. Arrows indicate infusions. The hash marks in the top panel indicate alternate q4-week infliximab infusions for which no serum samples were collected. The lines connecting the data points following week 22 are used only to compare the nadir concentrations for each treatment group and do not represent intervening serum infliximab concentrations occurring between each infusion. q4 wks = administered every 4 weeks after initial infusions; q8 wks = administered every 8 weeks after initial infusions.

DISCUSSION

The ATTRACT study evaluated 428 patients with RA who had previously failed to respond completely to MTX. There was a sustained and significant increase in physical function, as assessed by the HAQ and physical component summary score of the SF-36, both of which are quality of life measures, over the 2-year duration of the study. The significant inhibition of structural damage, including both erosions and joint space narrowing, which was observed previously (4), continued through 2 years in the subset of patients who continued treatment with infliximab plus MTX. The appearance of fewer newly eroded joints per patient in the infliximab plus MTX treatment groups than in the MTX-only group also indicates that infliximab in combination with MTX may significantly alter the development of new erosions. The infliximab plus MTX regimens were shown to be superior to MTX alone in maintaining control of the signs and symptoms of RA, and also produced a significant, sustained decrease in CRP levels, a biochemical measure of inflammation. Improvement in physical function, improvement of the signs and symptoms of RA, and inhibition of radiographic progression occurred independent of the infliximab dose used in the study. The observed reduction of disability in combination with prevention of joint damage illustrates the potent and durable effect of infliximab plus MTX therapy in patients with RA.

Health-related quality of life is profoundly affected by active RA. Measures such as the HAQ and SF-36 capture the degree to which individuals with RA have compromised functional ability and quality of life (8, 9). The median improvement from baseline in HAQ scores, averaged over time, that was observed among the patients receiving infliximab in combination with MTX in the ATTRACT study was 0.4, in contrast to a median improvement of 0.1 for patients treated with MTX only. The degree of improvement observed with infliximab plus MTX therapy translates into a functional benefit that has been suggested to be clinically meaningful (17, 18). Epidemiologic data suggest that the magnitude of these improvements in HAQ scores is correlated with a meaningful improvement in functional activities that may translate into lower rates of disability, higher employment rates, and lower health care costs (19).

Health-related quality of life as measured by the SF-36 was also significantly improved in the infliximab plus MTX treatment groups over time, compared with the MTX-only regimen. This patient cohort had an average overall baseline physical component summary score of 26.3, in comparison with an average score of 50 for the general population. As a result of therapy with MTX and infliximab, there was a median change from baseline of 4.6–6.9 in the physical component summary score. Previous work has shown that a change from baseline of 4.4 in the physical component summary score of the SF-36 is a minimally important change among patients with RA (20). Therefore, a clinically important improvement from baseline in the SF-36 physical component summary score was observed among patients treated with infliximab plus MTX. Notably, significant improvements in the vitality and social functioning subscales of the SF-36 were also achieved with the infliximab plus MTX regimens over the 2-year study.

Use of the SF-36 is not limited to RA, since this instrument has been validated to measure health-related quality of life in other diseases as well (21). The magnitude of the increase in the SF-36 scores observed in the ATTRACT study compares favorably with the improvement in function observed in rehabilitated patients with chronic obstructive pulmonary disease (22). In addition, changes observed in the physical component summary score in the ATTRACT cohort approximated the degree of change considered clinically significant in patients with malignancies (improvement of nearly 1 standard deviation) (23). Several of the most highly weighted individual components of the physical component summary score were more significantly improved in the infliximab plus MTX regimens as compared with those in the MTX-only regimen (P < 0.001 for the physical functioning subscale, role—physical subscale, and bodily pain subscale scores). Two additional individual scales of the SF-36 also demonstrated a significant improvement with all infliximab regimens compared with MTX only (i.e., vitality subscale scores P < 0.001; social functioning subscale scores P ≤ 0.002). However, mental health subscale scores (P ≤ 0.423) were not significantly improved in the infliximab plus MTX regimens compared with those in the MTX-only regimen, as would be anticipated since these evaluations did not have markedly abnormal results at baseline.

The direct, indirect, and intangible costs of RA represent an enormous burden to patients, to clinicians, and to society in general (24). Functional disability is associated with work disability, joint replacement surgery, and mortality (19). The increases observed in both physical function, as measured by the HAQ, and health-related quality of life, as measured by the HAQ and SF-36, among infliximab plus MTX–treated patients may translate into substantial societal and economic benefits, because physical function and health-related quality of life are consistently worse in RA patients when compared with the general population, independent of patient age (25).

Early in the course of RA, physical function depends primarily on the presence of pain. As the disease progresses over time, functional outcome is more strongly correlated with structural damage (26, 27). In the ATTRACT population at week 102, individuals treated with infliximab plus MTX had a median change from baseline of 0.5 in the total radiographic score compared with a median increase of 4.25 in the MTX-only patients (mean changes were 0.7 and 12.6, respectively). The rate of radiographic progression in the MTX-only group is consistent with other observations of MTX-treated patients over time in this patient population (28, 29). The finding that the infliximab plus MTX combination inhibits radiographic progression over 2 years confirms earlier observations of the benefit of infliximab plus MTX in preventing structural damage and supports the durability of the benefit of the combination of infliximab and MTX relative to the MTX-only control group. Longer-term observations will be necessary to determine whether this benefit is maintained and whether long-term functional outcome is altered.

Improvement in the signs and symptoms of RA was sustained over the 2-year study period. The degree of clinical benefit is likely to be understated, given the conservative intent-to-treat analysis and performance characteristics of the ACR20 criteria. In particular, the individual components of the ACR20 core disease activity set that waned somewhat over time were the subjective measures of patient pain and patient global disease assessment. Conversely, the swollen and tender joint counts remained improved in a majority of the patients treated with infliximab plus MTX. It is, therefore, possible that patient expectations changed over the course of the trial and altered the perceived subjective degree of improvement.

The length of time that patients receive antirheumatic therapy is an important measure of effectiveness, since therapy duration is a composite measure that accounts for therapeutic benefit as well as negative effects such as adverse reactions and loss of efficacy (30). Thus, the higher proportions of infliximab plus MTX–treated patients completing treatment (55–68%) and remaining in the study for 2 years (73–87%) as compared with the MTX-only group (16% and 58%, respectively) are indicative of the substantial benefit and tolerability of infliximab plus MTX therapy.

The combination of infliximab and MTX was well tolerated. The frequency of serious adverse reactions, serious infections, serious infusion reactions, and deaths was comparable between patients treated with infliximab plus MTX and those treated with MTX only. One patient with a long history of RA and concomitant use of infliximab and MTX therapy died as a result of complications of disseminated tuberculosis. Cancer occurred in 1 patient (1%) who received MTX only and in 9 patients (3%) treated with infliximab plus MTX. The overall frequency of cancer is consistent with that predicted from the Surveillance, Epidemiology, and End Results database for age- and sex-matched individuals (Table 6) (31). There are currently 3 patients who received infliximab plus MTX and 2 patients who received MTX only from the ATTRACT study with malignancies diagnosed during the 3 years of the long-term safety followup.

Autoantibody development with the administration of infliximab has been reported previously (5, 7, 32, 33). Consistent with these observations from previous infliximab clinical studies were increased frequencies of ANA and anti-dsDNA, which were detected over the duration of the ATTRACT study. In addition, consistent with the findings in previous reports, lupus-like events occurred rarely over a 2-year duration of TNF blockade (34).

Despite the occurrence of antibodies to infliximab in ∼8% of patients receiving infliximab plus MTX, the overall pharmacokinetic parameters for infliximab were stable through week 102. Given the stable pharmacokinetic profile and sustained clinical efficacy observed among the patients receiving infliximab plus MTX, the presence of antibodies to infliximab does not appear to have a major impact on the benefits of infliximab.

The study and the interpretations of the analyses of the week 102 data have several limitations. In particular, potentially confounding factors include the amendment to the protocol that extended study treatment, the resultant treatment hiatus period for some patients, the use of DMARDs during the hiatus period, and the rules used for handling missing data.

The amendment to the protocol, which was written in compliance with the guidance for development of treatment for RA (16) requiring a minimum of 2 years of followup for assessment of improvement in physical function, made necessary an optional extension of treatment. This extension of the study resulted in patients resuming their treatment for a second year and involved a treatment hiatus in some patients who had already completed the first year of treatment. No adjustments for the treatment hiatus were made in the analyses of the HAQ scores. A treatment gap of 8 weeks or less is unlikely to represent a significant interruption, considering that patients received infliximab plus MTX on an every-4-weeks or every-8-weeks schedule. Furthermore, HAQ scores were averaged over time, representing observations through week 102 rather than at week 102. In addition, any gap in treatment would be expected to lessen the observed changes in clinical response, physical function, and quality of life outcomes, thereby potentially underestimating the benefit of infliximab treatment. Although patients who had a treatment hiatus may have been treated with other medications (such as DMARDs) during that time, it is unlikely that these therapies had a lasting impact, because such treatment was relatively short and initial eligibility to participate in the study required patients to have had an inadequate response to DMARDs. Moreover, a washout period was mandated before patients could resume trial therapy. Importantly, the medications taken by patients during the more than 8-week treatment hiatus did not consist of other anti-TNF agents or biologic drugs, because they were commercially unavailable during this period of time.

As recommended by the Safety Monitoring Committee, all patients in the study were unblinded to treatment for ethical reasons after the positive radiographic results were known at week 54. Patients received no more placebo infusions if they were randomized to the MTX-only group, and were allowed to continue to receive additional infliximab infusions if they were receiving infliximab plus MTX. There was no effect of this unblinding on radiographic scores, since the radiographic reviewers remained blinded to the treatment allocation.

The unblinding of treatment among the patients had a minimal impact on the overall HAQ results, because the HAQ and SF-36 scores were averaged over time. The HAQ evaluations for 375 patients (88%) were not unblinded before completion of their study evaluations. Thus, the unblinding of patients was not likely to have substantively affected the primary analysis at week 102, which was an evaluation of the change from baseline in HAQ averaged over 102 weeks. No adjustment was made for unblinding in the analysis of HAQ, SF-36, or ACR20 data. Because the radiograph readers remained blinded to treatment, it was unlikely that there was an impact on the interpretation of radiographic data.

All patients received HAQ evaluations through 2 years, and the data were included in the week 102 analysis as prespecified in the analysis plan, regardless of whether they received a full 2 years of treatment. Similarly, data on all patients were included in the ACR20 analysis at week 102. Patients with missing data were considered to be ACR20 nonresponders. Data on all patients who had radiographs at baseline and at week 102 were included in the analysis of radiographic scores. Multiple sensitivity analyses were carried out with the radiographic data, to evaluate the possible impact of missing data; this included an analysis showing that in order for the missing data to change the conclusion of the analysis, the subgroup of patients with missing data in the 3 mg/kg every 8 weeks group would have had to have worse results than those in the placebo group, and the subgroup of patients with missing data in the placebo group would have had to have had better results than those in the 3 mg/kg every 8 weeks group. Importantly, all the analyses were preplanned.

In determining the average change from baseline over 102 weeks, missing data for the HAQ and SF-36 scores were imputed by assigning a 0 change from baseline for patients who discontinued the study followup. This method does not add to the improvement in the HAQ scores over 2 years after discontinuation and, therefore, does not enhance the results. For example, if HAQ scores were available for 70% of the visits with an average improvement in HAQ score of 1.0, the change from baseline in HAQ score for the remaining 30% of the visits would be imputed as 0, and the average would then be calculated over the entire 2-year period.

Several additional analyses were carried out that demonstrated the lack of impact on the primary analysis results upon substitution of different methods of handling missing data. The first sensitivity analysis allowed for worsening and set missing observations after the discontinuation of followup to the worst possible score (score of 3), whereas the second sensitivity analysis allowed for worsening and replaced missing observations after discontinuation of followup with the worst scores for individual infliximab-treated patients and the best scores for the placebo-treated patients. The third sensitivity analysis carried forward the week 54 observations for patients who completed 54 weeks of treatment and who declined participation in the second year, and adjusted for protocol-prohibited medication changes and surgeries by carrying forward the value immediately before the medication change or surgery.

Except for the second sensitivity analysis, the results of these analyses show that the conclusions of the primary analysis were unchanged when using different rules for handling data. In each analysis, the overall treatment effect was significant (P < 0.001) and all 4 of the infliximab treatment groups were significantly better than the placebo group in pairwise comparisons. In the second sensitivity analysis, as would be expected, the MTX-only group had better results and the infliximab plus MTX groups had worse results than in the primary analysis, so that none of the treatment group comparisons was significantly different. However, despite the worst-case nature of the analysis, results in the MTX-only group were still numerically worse than the results in all of the infliximab plus MTX groups.

Importantly, baseline demographics for the 216 patients who completed therapy and the 212 patients who did not complete therapy through week 102 were comparable with those of the entire population as well as with the other subgroups. All of these considerations support the interpretations and conclusions of the study despite the limitations inherent in the evolution of this trial.

In summary, the combination of infliximab and MTX over a 2-year period provides significant and clinically relevant improvement in physical function and quality of life, and this improvement is accompanied by inhibition of progressive joint damage and sustained improvement in the symptoms and signs of inflammation in patients with RA who had an incomplete response to MTX alone. Moreover, treatment with infliximab plus MTX is generally safe and well tolerated, as demonstrated throughout the 2 years of the study.

Acknowledgements

We would like to acknowledge the other members of the ATTRACT Study Group, as follows: D. Felson, MD (chairperson; Boston University Medical School, Boston, Massachusetts), F. Wollheim, MD (Lund University Hospital, Sweden), C. Goldsmith, PhD (St Joseph's Hospital and McMaster University, Ontario, Canada), T. Schaible, PhD (Centocor), G. Harriman, MD (Centocor), K. DeWoody, PhD (Centocor), E. Keystone (Mount Sinai Hospital, Toronto, Ontario), J. von Feldt (University of Pennsylvania, Philadelphia), N. Olsen (Vanderbilt University Medical Center, Nashville, Tennessee), M. Speigel (Arthritis Associates of Connecticut/New York, Danbury, Connecticut), A. Kavanaugh and D. Karp (University of Texas Southwestern Medical Center at Dallas), W. Benson (Charlton Medical Centre, Hamilton, Ontario), W. Shergy (Rheumatology Associates of North Alabama, Huntsville), O. Gluck (Phoenix Center for Clinical Research, Phoenix, Arizona), G. Firestein (Thornton Hospital, La Jolla, California), R. I. Jain (North Shore University Hospital, Manhasset, New York), K. Saag and M. Schilling (University of Iowa, Iowa City), M. Lopatin (Rheumatic Disease Associates, Willow Grove, Pennsylvania), D. Yocum (University of Arizona Health Science Center, Tucson), M. Burnette (Tampa Medical Group, Tampa, Florida), A. Russell (University of Alberta Hospital, Edmonton, Alberta, Canada), R. Spencer and K. Hobbs (University of Colorado Health Sciences Center, Denver), E. Sheldon (Miami Research Associates, Miami, Florida), M. E. Wenger (Lancaster Rheumatology Research, Lancaster, Pennsylvania), R. Pope (Northwestern University, Chicago, Illinois), J. Fiechtner and P. Coleman (Good Clinical Practice, East Lansing, Michigan), P. C. Taylor (Charing Cross Hospital, London, UK), P. Emery and K. Ahmed (Leeds General Infirmary, Leeds, UK), L. Klareskog and P. Larsson (Karolinska Sjukhuset, Stockholm, Sweden), G. Burmester (Medizinische Klinik III, Krankenhaus Charite, Berlin, Germany), C. Antoni (Abteilung Innere Medizin III mit Poliklinik und Institut für klinische Immunologie Universitätsklinikum Erlangen-Nürnberg, Erlangen, Germany), B. Leeb (Center for Rheumatic Diseases Krankenhaus, Vienna, Austria), R. D. Sturrock and M. Field (Royal Infirmary, Glasgow, UK), H. Nüβlein (Stadtisches Klinikum Dresden-Friedrichstadt, Dresden, Germany), J. Petersen (Rigshospitalet, Copenhagen, Denmark), M. Dougados and A. Kahan (Groupe Hospitalier Cochin, Paris, France), A. Boonen (University Hospital Maastricht, The Netherlands), and A. Verhoeven (University Hospital Maastricht, The Netherlands).

Ancillary