Efficacy of biologic agents in improving the Health Assessment Questionnaire (HAQ) score in established and early rheumatoid arthritis a meta-analysis with indirect comparisons
Fig. 1. Search results.
Efficacy of biologic agents at lowering HAQ in established RA patients failing -0.2; 95%CI: -0.40, 0; and rituximab anti-TNF agents
-0.23; 95%CI: -0.32, -0.14) (Fig. 4). Four of the included trials required fail-ure of an anti-TNF agent at enrollment. Effect of baseline patient
Meta-analysis of these trials revealed characteristics, study design and a study quality on HAQ
95%CI: -0.43, -0.30 (pooled HAQB- HAQC of -0.36; There There were no significant differences in I2=92%) (Fig. 3). HAQ improvement in established RA was a significant difference in the patients previously failing DMARDs at (n=19 trials), patients failing anti-TNF for abatacept of -0.40; 95% CI:-0.51, improving efficacy of HAQ: the different HAQbiologics B- HAQC agents (n=-0.29), rituximab (-0.37; 95% CI: -0.46, als); 4 trials) and ERA (n=5 tri--0.27) and tocilizumab (-0.36; 95% CI: als combined biologic therapy with a p=0.001 (Table II). Most tri--0.42, -0.30).
DMARD agent (21/23 with metho-trexate, 1/23 with sulfasalazine, 1/23 Efficacy of biologic agents at lowering with other DMARD). Five trials used HAQ in early RA (ERA) patients
biologic monotherapy (adalimumab, ERA trials included DMARD-naïve etanercept, golimumab, infliximab and (n=1 abatacept), and MTX-naïve patients (subjects trial investigating infliximab) cious at lowering HAQ as combination which appeared as effica-could have been exposed to other therapy. There was no detectable differ-DMARDs previously; n=4 trials). The ence in HAQ improvement at 6 months pooled HAQ(n=als was -0.23; 95% CI: -0.32, -0.14 B- HAQC of all five tri-(Table II). Using meta-regression, mean 19) compared to 12 months (n=9) (I2difference in change in HAQ was not 95% CI not meeting MCID for change =0%), with the upper limit of the significantly in HAQ (Fig. 4). There was no signifiease duration and baseline HAQ scores. affected by baseline dis-cant difference in HAQ improvement -Similarly, study quality as measured by for the different biologic agents. The the Jadad score, had no significant ef HAQfect on results. Patient cross-over from --0.20; 95%CI: -0.34, -0.06; etanercept B- HAQC for adalimumab was control groups to intervention groups -0.3; 95%CI: -0.52, -0.07; infliximab
and year of publication was also not found to impact HAQ (data not shown).
336
DThis is the first meta-analysis with indiiscussion
rect comparisons to specifically assess -the comparative efficacy of all clinically available biologic therapies in improv-ing physical function as measured by the Health Assessment Questionnaire (HAQ) in RA. Other indirect compari-son studies of HAQ have included only anti-TNF agents in DMARD-failures (6, 8, 10). We also evaluated improvement in HAQ in patients failing anti--TNF agents and DMARD/methotrex-ate -naïve patients (primarily the ERA subgroup). These groups were selected a prioriresented because we felt that they rep-clinical differences, such as disease se-populations with significant verity (including baseline HAQ score), duration of symptoms, extent of ra-diographic damage co-morbidities and potentially different responsiveness to treatment. In our study we found signif-icant differences in HAQ improvement in the group with prior anti-TNF fail-ures, which had higher baseline HAQ scores, and the same per cent change in HAQ compared to the DMARD failure group (data not shown).
We chose the HAQ score as our outcome of interest for several reasons: (1) it is a patient-centred, clinically-relevant vali-dated score that correlates highly with disability and quality of life (3-5); (2) it is a continuous outcome, which is more sensitive to detect change than the traditionally used binary ACR20/50/70 score (2); and (3) it can be used in eco-nomic evaluations, which are crucial when assessing expensive therapies.
All biologics in patients failing DMARDs were efficacious at improving HAQ. The pooled mean difference -in change in HAQ for biologics com-pared to control was -0.22, which meets the minimally clinically important dif-ference (MCID) for HAQ (15). With re-spect to the DMARD-failure group, the detectable heterogeneity could be ac-counted for by differences in the efficacy of the different biologic agents. The -least effective biologic was infliximab, which of other meta-analyses (6-7). Only one is consistent with the findings infliximab DMARD-failure group and it had vari-
study was included in the ous study arms with different doses. If