The variability of response to treatment in rheumatoid arthritis (RA) precludes any single measure of disease activity, severity or therapeutic effectiveness from providing comprehensive information for a broad spectrum of patients. Treatment guidelines recommend that the choice of RA therapy is based on symptoms and indices of disease activity, along with clinical markers, including rheumatoid factor, erythrocyte sedimentation rate and C-reactive protein levels.1 In addition, patient-reported outcomes (PROs) in RA are increasingly recognised as reliable quantitative measures to evaluate health status and monitor therapeutic response.2

PROs are especially relevant in RA because complete remission is rare, and thus, the primary therapeutic goals are to relieve pain, preserve function for activities of daily living and work, and maximise health-related quality of life (QoL).1 3 PROs provide valuable information on these effects from the perspective of individual patients, to complement physician assessment and clinical measurement.4 PRO instruments are incorporated in routinely used disease activity composite measures: the Disease Activity Score (DAS28)5 includes a patient assessment of general health scored on a visual analogue scale (VAS), while the American College of Rheumatology (ACR) core criteria6 7 include patient assessments of pain, disease activity and physical function. PROs proven be less prone to placebo response in RA clinical trials than physician-generated assessments.8 Furthermore, assessments of function in patients with RA are strong predictors of mortality.9 10

Patients with RA are considered candidates for biological therapy after they have failed to respond to or tolerate at least one or two disease modifying antirheumatic drugs (DMARDs), usually including methotrexate (MTX).11 12 The fully human anti-tumour necrosis factor receptor (anti-TNFR), etanercept (ETN), has been shown to yield important clinical improvement among patients who had an inadequate response to either MTX or other DMARDs.13^–16 In a double blind, randomised clinical trial (RCT) of patients with active RA despite MTX treatment, combination therapy with ETN and MTX improved function and disease activity measures compared with MTX alone.17 In a later RCT among patients who were MTX naïve or had an adequate response to MTX, combination therapy yielded superior efficacy to monotherapy with either MTX or ETN, as measured by several PROs18 and objective clinical endpoints. However, no previous trial has compared combination therapy versus ETN monotherapy for patients with an unsatisfactory response to MTX. Examining this patient population is important because in “real world” clinical settings, approximately 60% of patients fail initial DMARD therapy,19 and half of patients receiving MTX discontinue within 40 months.20 The Add Enbrel or Replace MTX (ADORE) study compared the efficacy and safety of substituting ETN for MTX versus adding ETN to MTX, in patients with RA who had an inadequate response to MTX therapy. Primary clinical outcomes of ADORE have been published separately.21

This study reports the effect of treatment in ADORE on patient-reported assessments of function, pain, general health status and disease activity.

METHODS

The PRO assessments reported here were secondary assessments in a 16-week, prospective, randomised, open-label, parallel-group, outpatient study conducted between March 2003 and May 2004 at 60 centres in Denmark, Finland, France, Germany, the Netherlands, Spain, Turkey and the UK. Study methods were detailed in the previously published report of primary assessments,21 and are summarised briefly below.

Patients

Eligible patients were 18 years of age or older, in ACR Functional Class I, II or III, with active RA despite treatment with MTX 12.5 mg/week or more for 3 months or longer. Active RA was defined by either a DAS285 of 3.2 or greater or a combination of five or more swollen joints, five or greater painful joints and an ESR of 10 mm/h or greater. The study complied with the Helsinki Declaration, and the protocol was approved by an institutional review board or ethics committee in each country. All subjects provided written, informed consent.

Three hundred and fifteen patients were randomised to ETN (n = 160) or ETN+MTX (n = 155). One patient in the ETN group did not receive any study medication and was withdrawn from the study, leaving a total of 314 patients who received at least one dose of study medication and were included in the intention-to-treat population. Baseline demographic and disease characteristics have been reported previously.21

Treatment groups

Patients were randomly assigned to either monotherapy with ETN 25 mg twice weekly by subcutaneous injection (ETN group) or combination therapy with ETN added to the baseline dose of MTX (12.5 mg/week or more taken orally or by injection) (ETN+MTX group). Patients receiving ETN initially received MTX concurrently with ETN, but MTX was tapered off over 4 weeks, after which patients were given ETN alone for 12 additional weeks.

Outcomes

Five PRO instruments were administered to study patients: the Stanford Health Assessment Questionnaire (HAQ), pain VAS, EuroQoL five-dimensional assessment of current health state (EQ-5D) including general health VAS (EQ-5D VAS), and patient global assessment of disease activity (PGAD). The HAQ disability index (HAQ DI) was calculated from the eight HAQ subscales: dressing, arising, eating, walking, reaching, gripping, hygiene and daily activities.22 The HAQ DI and subscales range from 0 (no difficulty) to 3 (unable to do). Pain was scored on a horizontal scale from 0 (no pain) to 100 (pain as bad as it could be). EQ-5D is a generic measure of a patient’s current health status and includes five dimensions (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) each with three levels (no problem, some problem and extreme problem) and a time frame of “today”. By combining different levels from different dimensions, EQ-5D defines a total of 243 health states. Each health state can be associated with a numeric score on which full health has a value of 1 and death has a value of 0. In addition, the EQ-5D includes a 20 cm vertical VAS ranging from 0 (worst imaginable health state) to 100 (best imaginable health state).23 Patient global assessment of overall RA activity was assessed on an 11-point numeric rating scale ranging from 0 (none) to 10 (extreme).

The HAQ was administered at baseline and week 16, while the other PRO assessments were administered at baseline and weeks 4, 8, 12 and 16.

Data analysis

All PRO measures were secondary planned analyses of a trial for which clinical data were published earlier.21 Primary comparison for all PROs was mean improvement from baseline to week 16 within and between the ETN and ETN+MTX groups. The mean change from baseline was analysed with analysis of covariance (ANCOVA) with treatment as main effect and baseline value as covariate; 95% confidence intervals were determined for the difference in mean change from baseline using the least square means from the final ANCOVA model. The significance of the within-group mean changes from baseline was assessed using paired t-tests. This analysis was also implemented for each of the five dimensions of EQ-5D, ie, mobility, self-care, usual activities, pain/discomfort and anxiety/depression.

Treatment groups were compared by proportion of patients within clinically meaningful categories. For the HAQ DI, categories were determined to be: patients achieving an improvement from baseline to week 16 in HAQ DI score greater than 0.22 units (a clinically meaningful improvement24 25) and patients achieving a HAQ DI improvement greater than 0.5 (a threshold used in other studies14 26). Additionally, patients achieving a HAQ DI score at week 16 of less than 1.0 are reported. As a HAQ DI ⩾1.0 has been used to describe moderate functional disability,27 28 we used a cut-off of 1.0 to characterise patients with mild disability at the end of study. Treatments were also compared by the proportion of patients achieving an improvement in EQ-5D utility ⩾0.05 (a clinically meaningful improvement).29 Similarly, the percentage of patients achieving an EQ-5D VAS greater than 82 was also reported, representing population norms.30 χ² analysis was used. An asymptotic 95% confidence interval for the difference in proportions between the two treatments was calculated using the normal approximation to binomial distribution.

Cumulative percentage plots for all PRO measures were presented to characterise treatment effects on PROs.

The last observation carried forward (LOCF) method was used to impute missing data for analysis of the entire intention-to-treat population.31 Screening data were used for baseline values when baseline data were missing. All tests were two-sided using 5% significance, leading to a two-sided 95% confidence interval.

RESULTS

Stanford Health Assessment Questionnaire disability index

Both treatment arms had similar HAQ DI scores at baseline and experienced similar improvements from baseline to week 16, with a mean change of −0.59 in both groups (table 1). No significant difference existed between treatment groups in the proportion of patients achieving HAQ improvements >0.22 or >0.5, or achieving a HAQ score of <1.0 (table 2).

EuroQoL five-dimensional assessment of current health state visual analogue scale

Both treatment groups had similar EQ-5D VAS scores at baseline and showed comparable improvements from baseline to week 16 (mean change, ETN: 19.76 and ETN+MTX: 21.00 (table 1). Improvement was greater from baseline to week 4 than between subsequent 4-week assessments (fig 1A). The proportion of patients in both groups attaining an EQ-5D VAS score near the population norm of 8230 was similar (table 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Time course of change in patient-reported outcomes. ETN, etanercept; MTX, methotrexate; PGAD, patient global assessment of disease activity; EQ-5D, EuroQoL five-dimensional assessment of current health state; VAS, visual analogue scale.

EuroQoL five-dimensional assessment of current health state

A significant impairment was seen at baseline for all five of the EQ-5D health states in both treatment groups (table 1). There was greater improvement in the first 4 weeks of the study than between subsequent 4-week assessments (fig 1B). The majority of patients reported problems in mobility, usual activities, pain/discomfort at baseline and approximately half the patients reported problems related to anxiety/depression at baseline (fig 2). There was significant improvement in each EQ-5D health state within each treatment group (p<0.0001) (fig 2), with clinically meaningful improvements for approximately two-thirds of patients. The category of self-care was the only health status utility that showed a significant difference between treatment groups (p = 0.0154) (table 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Percentage of subjects reporting no problem in five dimensions of EuroQoL. ETN, etanercept; MTX, methotrexate. (A) Usual activities*; (B) Self-care*†; (C) Pain discomfort*; (D) Mobility*; (E) Anxiety/depression. *Within-group change from baseline in each of the five dimensions was statistically significant (p<0.0001). †Statistically significant difference between treatment groups (p = 0.0154).

Disease Activity and Pain visual analogue scale

No significant differences between treatment groups were found for PGAD (fig 1C) or pain (fig 1D) improvements from baseline to week 16 (table 1). A trend for greater improvement in pain VAS scores in the ETN+MTX group did not reach statistical significance (p = 0.058). As with EQ-5D VAS and EQ-5D health status, all three of these PRO, showed greater improvement in the first 4 weeks than between subsequent 4-week assessments.

Characterisation of treatment effect on patient-reported outcome measures

Cumulative percentage graphs for all PRO measures are shown in fig 3. The mean change from baseline to week 16 was similar for both treatment arms, demonstrating a comparable trend in both groups (fig 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 Cumulative percentage graphs showing distribution of change of patient-reported outcomes among patients between treatment groups from baseline to week 16 (last observation carried forward method). ETN, etanercept; MTX, methotrexate; EQ-5D, EuroQoL five-dimensional assessment of current health state, EQ-5D VAS, EQ-5D, including general health visual analogue scale; PGAD, patient global assessment of disease activity.

DISCUSSION

QoL assessment using PROs is valuable for patients with RA, as PROs highlight important information on the effects of different treatments from the patient perspective, complementing physician assessment and objective clinical measurement.4 A study by Strand et al found that for RA, PROs are objective, producing findings that correlate to laboratory measures of inflammation and are less vulnerable to the placebo response compared with physician assessments.8 The meaning of PROs in clinical studies is further revealed by the finding that physicians and patients differ in their assessment of symptom severity and importance.4 For RA specifically, patient and professional assessments of physical and mental function differ even though the same scale was used by both groups.32 33 Thus, adding measures of the patient’s perspective to physician assessment may allow for more accurate interpretation of clinical and trial data.4

In clinical practice, insurance restrictions often require patients with RA to fail treatment with MTX before proceeding to treatment with ETN or other biological agents. In this situation, it is necessary to know whether differences between switching to ETN alone or adding ETN to MTX. In previous trials involving late established RA, partial or inadequate responders to MTX were randomised to receive either placebo or active treatment in addition to MTX.16 17 34 35 These trials have shown that among patients with an unsatisfactory response to MTX, combination of anti-TNFR agents with MTX provides a significantly greater improvement in PROs than MTX alone. However, a limitation of this trial design is that it does not provide empirical evidence to inform clinically relevant decisions on whether to add or switch to a different agent.36 A previous trial37 demonstrated that combination therapy yielded superior PROs compared either monotherapy but was not conducted in patients with sustained unsatisfactory response to previous DMARDs. In our study, patients with an unsatisfactory response to previous DMARDs demonstrated comparable improvement in all PRO measures for both treatment arms (table 1; figs 1 and 3). Thus, monotherapy with ETN may be an equally effective option for patients intolerant to MTX.

In this study, both treatment arms exhibited 0.59 improvement in HAQ DI, a 36–37% improvement from baseline. This finding is consistent with previous studies of ETN in a similar patient population, which found an improvement of 39% in HAQ DI at 6 months.15 Remarkably, the improvement in HAQ DI at week 16 was also similar to improvements seen in the ETN monotherapy arm at week 16 in the TEMPO (Trial of Etanercept and Methotrexate with Radiographic Patient Outcomes) trial, albeit smaller than the improvements seen in the combination group.37 The magnitude of improvements seen in EQ-5D VAS is also consistent with results from a previous 26-week study of patients with RA receiving either ETN or placebo.38 In that study patients in the ETN group reported an improvement of approximately 18 points in EQ-5D-VAS compared with 22- and 23-point improvements seen in the monotherapy and combination groups in our study, respectively. Because patients with RA frequently suffer from pain, depression, anxiety and loss of mobility; we used EQ-5D health state score to characterise the impact of RA on a patient’s QoL. At baseline, the EQ-5D health state scores were 0.44 and 0.39 in the ETN monotherapy and combination groups, respectively. An earlier study by Hurst et al39 reported an EQ-5D score of 0.38 for 233 subjects with RA in England, Wales and Scotland. Thus, our study confirms the findings of that earlier study demonstrating a considerable impact of RA on patients’ QoL. Similarly, the frequency distribution of responses on each dimension of EQ-5D in the current study also appear comparable with those reported in the study by Hurst et al.39

Previous studies have determined that a change in the HAQ DI score of 0.13 to 0.24 may be considered clinically significant.24 25 In our study, 71% and 73.8% of patients achieved clinically meaningful improvements in HAQ DI in the monotherapy and combination groups, respectively. This was consistent with findings that similar proportions of patients also reported clinically meaningful improvements in EQ-5D general health VAS. In the TEMPO trial,18 approximately 77% and 86% of patients achieved clinically meaningful improvement in HAQ DI in the ETN monotherapy group and combination groups, respectively.18 Although, similar to this study, the variance in proportions of patients for the two treatment arms from the TEMPO trial18 to this study may be attributed to differences in the study population.

Another method to facilitate interpretation of changes in PRO measures is to compare the findings with norms40 from a population of subjects with no known history of specified illness; the proportion of study patients with higher scores than the population norm can be interpreted as having been restored to normal QoL. In the current study, although mean EQ-5D VAS scores were well below the population norm of 82,30 approximately 30% of patients reported EQ-5D VAS scores higher than the population norm at the end of the study. This finding is consistent with the TEMPO study, where 31% of patients in the ETN monotherapy group reported EQ-5D VAS scores greater than 82.18

A few study limitations merit discussion. This was not a blinded study and as such, patients’ preconceived notions of the drug may have affected their perceptions of therapeutic results.41 However, at the time of this study, results of the TEMPO trial18 were not yet available; therefore, patients may not have known what to expect from the treatment option they were receiving. Another possible limitation is that the LOCF imputation of missing data may introduce bias if scores change over time,31 but results of the observed (per protocol) population are consistent with LOCF results. Lastly, this study only included patients judged to be appropriate candidates, ie, patients with an unsatisfactory response to MTX; therefore, these results may not be consistent with those found in other patient populations.

CONCLUSIONS

PROs provided valuable efficacy information in this study, reflecting patient perceptions of therapeutic success. For patients with active RA and intolerance or unsatisfactory response to MTX, substituting ETN for MTX or adding ETN to MTX are both effective at reducing disability, pain, disease activity and morning stiffness, as well as improving general health. Studying this group of patients is imperative because in “real world” clinical settings, roughly 60% of patients fail initial DMARD therapy19 and half of patients receiving MTX discontinue within 40 months.20 Switching to ETN appears to be equally effective as combination therapy in this patient population, and appears to represent a valid treatment option.