Abstract
Objective. To perform a Cochrane systematic review of benefit (American College of Rheumatology 50% improvement criteria; ACR50) and safety (adverse events and withdrawals) of golimumab in patients with rheumatoid arthritis (RA).
Methods. We searched the Cochrane Central Register of Controlled Trials (CENTRAL), OVID Medline, CINAHL, Embase, Science Citation Index (Web of Science), and Current Controlled Trials databases for randomized or controlled clinical trials of golimumab compared to placebo or disease-modifying antirheumatic drug in adults with RA. Two authors independently selected appropriate studies and abstracted study characteristics and safety and efficacy data and performed risk-of-bias assessment. We calculated mean differences for continuous measures, and relative risks for categorical measures.
Results. Four randomized controlled trials with 1231 golimumab-treated and 483 placebo-treated patients were included. Of these, 436 were treated with golimumab at 50 mg every 4 weeks [a dosage approved by the US Food and Drug Administration (FDA)]. At an average of 4–6 months, compared to patients treated with placebo and methotrexate (MTX), patients treated with the FDA-approved dosage of golimumab and MTX were 2.6 times more likely to reach ACR50 (p = 0.005, 95% CI 1.3, 4.9; absolute percentage, 38% vs 15%) and 0.5 times as likely to have overall withdrawals (p = 0.005, 95% CI 0.3, 0.8; absolute percentage, 5% vs 10%). Golimumab-treated patients were significantly more likely than those taking placebo to achieve remission (22% vs 4%; p < 0.00001), and to have improvement in functional ability on the Health Assessment questionnaire [0.2 points lower (p < 0.00001, 95% CI 0.25, 0.15); absolute risk difference, −20% (95% CI −25% to −15%); relative percentage difference, −11% (95% CI −14% to −8.3%)]. The studies were too small and short to be powered sufficiently for safety outcomes, but no substantive statistically significant differences were noted between golimumab and placebo regarding adverse events, serious adverse events, infections, serious infections, lung infections, tuberculosis, cancer, withdrawals due to adverse events, and withdrawals due to inefficacy and deaths.
Conclusion. At the approved dosage, in patients with active RA taking background MTX, golimumab is significantly more beneficial than placebo. The short-term safety profile is reasonable. Longterm surveillance studies are needed for safety assessment.
Rheumatoid arthritis (RA) is a systemic disease characterized by destructive, inflammatory polyarthritis affecting small and large joints1. RA has a significant effect on both patient-reported quality of life and function (HRQOL)2,3 and can sometimes lead to serious work disability4,5. Inflammation in RA is characterized by the activation of many immune cells, including but not limited to T cells, B cells, and macrophages6. Such immune cells secrete various cytokines7, i.e., tumor necrosis factor-α (TNF-α) and interleukins7,8, which in turn lead to inflammation, and joint and bone destruction. Lately, much of the focus of treatment of RA has been on inhibition of TNF-α, since it is thought to play a central role in joint inflammation9.
The pharmacological therapeutic strategies for RA include nonsteroidal antiinflammatory drugs (NSAID), traditional disease-modifying antirheumatic drugs (DMARD) such as methotrexate (MTX), and newer biologic DMARD. The biologic DMARD include TNF-α inhibitors and non-TNF biologic DMARD10. TNF-α inhibitors include etanercept11, infliximab12, and adalimumab13; recently, golimumab has been approved for RA in several countries (USA, Canada, European Union, Japan, and others). Non-TNF biologic DMARD (and respective targets) include anakinra (interleukin 1)14, abatacept (costimulatory molecule, CD28)15, and rituximab (B cells)16.
Golimumab is a humanized inhibitor of TNF-α17. It neutralizes TNF-α by binding to it, thus interfering with its binding to the TNF-α receptors on cell surfaces. TNF-α plays a key role in the inflammation and joint destruction that are the hallmarks of RA9. The approved dosage for treatment of RA with golimumab is 50 mg subcutaneous injection given once a month18. This frequency of administration is lower than all the other biologics except for abatacept (monthly subcutaneous), infliximab [intravenous (IV) every 8 weeks], and rituximab (3 doses over 6 weeks intravenously, then no infusions for 6 months). It is important to know whether, with this low frequency, the biologic therapy results in improvement that is judged important by patients and their clinicians.
MATERIALS AND METHODS
Criteria for considering studies for this review
All studies had to be randomized controlled trials (RCT) or controlled clinical trials (methods of allocating participants to a treatment that are not strictly random, e.g., date of birth, hospital record number, or alternation). There was no language restriction on included studies. An expert librarian searched the following databases for randomized or quasirandomized trials of golimumab in RA: (1) The Cochrane Central Register of Controlled Trials (CENTRAL), through The Cochrane Library, Wiley InterScience (www.thecochranelibrary.com), issue 2, 2009; (2) OVID Medline, 1966–2009; (3) CINAHL (through EBSCOHost), 1982–2009; (4) Embase 1980–2009; (5) Science Citation Index (Web of Science) 1945–2009; and (6) Current Controlled Trials.
We included studies involving adults age 18 years or older, with RA, meeting the 1987 American College of Rheumatology (ACR) classification criteria for RA19 (Table 1). Interventions compared were golimumab alone or in combination with DMARD or biologics versus placebo plus MTX or golimumab alone or in combination with DMARD or biologics compared to other DMARD or biologics. There were no restrictions with regard to dosage or duration of intervention. We assessed safety and efficacy at all dosages including the approved dosage of 50 mg every 4 weeks.
Characteristics of patients randomized to the golimumab treatment arm at the FDA-approved dosage in included studies.
Major outcomes
Outcomes were chosen based on recommendations regarding 7 key outcomes for the summary of findings table by the Cochrane Musculoskeletal Group. Major outcomes included the ACR 50% improvement criteria (ACR50), Disease Activity Score (DAS) remission (DAS < 1.6 or DAS 28-joint count < 2.6), function measured by Health Assessment Questionnaire (HAQ) score or modified HAQ calculated as score changes20,21, the proportion achieving minimally clinically important difference on HAQ ≥ 0.2222, number of adverse events, number of serious adverse events (SAE), number of withdrawals due to adverse events, and all withdrawals. ACR50 is defined as 50% improvement in both tender and swollen joint counts and 50% improvement in 3 of these 5 variables: patient global assessment, physician global assessments, pain scores, HAQ score, and acute-phase reactants [erythrocyte sedimentation rate (ESR) or C-reactive protein]23,24.
Secondary outcomes
Secondary outcomes included (1) ACR20 and ACR7024; (2) changes in either DAS, a composite index of tender and swollen joint counts, patient global assessment, ESR25, or DAS28 score26; (3) proportion achieving a “good state”, i.e., a good European League Against Rheumatism (EULAR) response27,28 defined by a decrease in DAS or DAS28 of ≥ 1.2 from baseline with a final DAS < 2.4 (or DAS28 < 3.2), and low disease activity defined by DAS < 2.4 or DAS28 ≤ 3.2; (4) quality of life, measured by Short-Form 36 (SF-36; i.e., continuous data, 8 domains, and 2 physical and mental component summary scores); (5) radiographic progression, as measured by Larsen/Sharp/modified Sharp scores29–31; and (6) withdrawals due to lack of efficacy.
Safety was assessed by the type of adverse effects and SAE, including infections, serious infections, lung infections, tuberculosis, cancer, etc., and death.
Assessment of risk of bias in included studies
We independently assessed the risk of bias using 5 criteria: random sequence generation; allocation concealment: presence of patient, clinician, and assessor blinding in the studies; incomplete outcome data; and sensitive outcome reporting32. For each criterion, we assessed the risk as: Yes (low risk of bias), No (high risk of bias), or unclear (either lack of information or uncertainty over the potential for bias). Disagreements were resolved by discussion between the 2 reviewers who carried out the risk assessment. Overall rating of evidence was done by using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) working group approach33, in which the following ratings were used to reach a summary quality rating score: (1) high (randomized trials or double-upgraded observational studies); (2) moderate (downgraded randomized trials or upgraded observational studies); (3) low (double-downgraded randomized trials or observational studies); or (4) very low (triple-downgraded randomized trials, downgraded observational studies, or case series/case reports). The level for randomized trial evidence can be downgraded by 1 or 2 levels depending on the presence of 5 factors: serious (−1) or very serious (−2) limitations to study quality; important inconsistency (−1); some (−1) or major (−2) uncertainty about directness; imprecise or sparse data (−1); high probability of reporting bias (−1).
We calculated relative risks with 95% CI for dichotomous outcomes. For continuous measures, we calculated mean differences when possible because results presented in this form are more readily interpreted by clinicians. We performed an I-squared to quantify heterogeneity. An I-squared of 0–40% might not be important; 30%–60% may represent moderate heterogeneity; 50%–90% may represent substantial heterogeneity; and 75%–100% considerable heterogeneity34. The random-effects model was the default model for pooling outcomes in the metaanalysis of studies as a conservative approach to account for clinical heterogeneity. The number needed to treat to benefit and the number needed to harm were calculated as the inverse of the absolute risk difference, using the Cates calculator35.
RESULTS
Included studies and risk of bias
The initial search yielded 187 articles and the updated search an additional 29 articles (Figure 1). Of these, 4 qualified for inclusion36–39. Overall, there were 1231 patients treated with golimumab and 483 patients treated with placebo included in the studies, which made 1714 overall patients in all arms. There were 436 patients taking golimumab 50 mg every 4 weeks (the FDA-approved dosage); the rest of the patients taking active drugs were on other dosage schedules.
Selection of studies for this review. RCT: randomized controlled trial.
All studies reported adequate methods of randomization, allocation concealment, and blinding. The overall quality rating of evidence was judged to be high based on the GRADE approach, with a low possibility of bias (Figure 2).
Methodological quality summary for each study. “+”: quality standard was present; “?”: quality standard was unclear.
Comparison of golimumab to placebo
Four studies36–39 provided data for ACR50. Golimumab-treated patients were 2.6 times more likely than those treated with placebo to reach ACR50 (Table 2). There was statistically significant heterogeneity with I-squared value of 76% (p = 0.005). Golimumab-treated patients were also significantly more likely to have DAS remission, more improvement in HAQ, and lower overall withdrawal rates. While all 4 studies provided data on safety, no statistically significant differences were noted in the number of adverse events, number of SAE, or withdrawals due to adverse events.
Summary of findings for the US Food and Drug Administration (FDA)-approved dosage of golimumab (50 mg every 4 weeks). Studies compared golimumab + MTX to placebo + MTX for RA. Patients had failed MTX therapy (and other medications in some cases).
Table 3 provides details on primary and secondary outcomes. There were no statistically significant differences between golimumab-treated patients and placebo-treated patients in the number of adverse events (absolute percentaqe, 73% vs 69%, p = 0.44), SAE (5.9% vs 5.6%, p = 0.85), infections (30% vs 29.5%, p = 0.8), serious infections (1.8% vs 1.6%, p = 0.9), tuberculosis reactivation (0.2% vs 0%, p = 0.5), lung infections (7.1% vs 7.3%, p = 0.9), and cancer (0.5% vs 0.6%, p = 0.8). No significant heterogeneity was noted for safety outcomes except for moderate-substantial heterogeneity for the number of adverse events, I-squared of 55% (p = 0.09).
Efficacy and short-term safety outcomes with FDA-approved dosage of golimumab of 50 mg every 4 weeks subcutaneously. Risk ratio was calculated using Mantel-Haenszel method for all categorical measures except for HAQ scores, change in HAQ scores, and change in DAS scores (continuous outcomes); for which mean differences were calculated between golimumab and placebo groups. Significant results are in bold type.
Patients treated with golimumab were half as likely to withdraw compared to those taking a placebo (5.2% vs 10%; p = 0.005; Tables 2 and 3). There was no statistically significant difference between golimumab and placebo-treated patients for withdrawals due to inefficacy (1.8% vs 3.7%, p = 0.1), withdrawals due to adverse events (2.8% vs 4.9%, p = 0.2), or number of deaths (0.2% vs 0.2%, p = 0.99).
Golimumab-treated patients were 1.5 times more likely to reach ACR20, 2.8 times more likely to reach ACR70, 1.5 times more likely to achieve good EULAR response, 1.6 times more likely to achieve low disease activity, and 5.1 times more likely to reach DAS remission, compared to placebo. Patients treated with golimumab had a significantly greater change in DAS28 scores, more decrease in HAQ scores, and were 1.8 times more likely to achieve a HAQ change ≥ 0.22.
None of the studies provided data on radiographic progression. Outcomes for other dosages of golimumab versus placebo are available in the Cochrane Review40. The results of all dosages of golimumab mirrored the ones for 50 mg of golimumab taken every 4 weeks.
Comparison of golimumab to MTX
Two studies compared 100 mg of golimumab taken every 4 weeks to MTX in MTX-naive patients39 or in patients who had failed MTX36. The studies involved 292 patients treated with golimumab and 293 treated with MTX. The treatment arms included 100 mg of golimumab every 4 weeks + placebo (oral) versus placebo (injections) + MTX36,39. Neither study was powered to examine the differences between these arms, but only to compare 2 doses of golimumab to placebo.
There was a statistically significant greater improvement (decrease) in HAQ scores in golimumab-treated patients compared to placebo, with a mean difference of −0.12 (95% CI −0.16, −0.08, p < 0.00001; Table 4). There were no statistically significant differences between control and golimumab groups for ACR50 rates (27% vs 21%, p = 0.24), number of adverse events (66% vs 67%, p = 0.7), SAE (3% vs 5%, p = 0.7), infections (33% vs 29%, p = 0.3), serious infections (1% vs 1%, p = 0.7), cancers (1% vs 1%, p = 0.8), total withdrawals (4% vs 6%, p = 0.5), withdrawals due to adverse events (25% vs 3%, p = 0.4), and the number of deaths (0.03% vs 0%, p = 0.5). With 1 study providing data39, there was no statistically significant difference between the number of lung infections between the golimumab and control groups (6% vs 9%, p = 0.3). No significant between-group differences were noted for good EULAR response, HAQ change ≥ 0.22, and DAS remission (Table 4).
Comparison of golimumab to methotrexate: golimumab 100 mg every 4 weeks + placebo (oral) vs placebo (injections) + methotrexate. Risk ratio was calculated using Mantel-Haenszel method for all categorical measures except for HAQ scores, change in HAQ scores, and change in DAS scores (continuous outcomes), for which mean differences were calculated between golimumab and placebo groups. Significant results are in bold type.
DISCUSSION
In this review of RCT of golimumab for treatment of active RA, 4 RCT of 436 patients receiving the approved (in USA, Europe, Japan, and Canada) dosage of 50 mg every 4 weeks and 483 controls were included. We found that when used with MTX for treatment of RA, golimumab at the approved dosage resulted in important benefit that was significantly better than placebo for achieving ACR20/50/70, lower DAS28 scores, RA disease remission, and improvement in physical function at 5–6 month followup. The studies were too short and underpowered for comparing safety outcomes, but the risks were similar to placebo of adverse events, SAE, infections, serious infections, tuberculosis, lung infections, cancer, or death. Withdrawals for any reason averaged 5% in the golimumab group, half as frequently as in the placebo group (10%). Although not statistically significant, withdrawals due to adverse events and withdrawals due to inefficacy were also lower in the golimumab group (2.8% and 1.8%) compared with the placebo group (4.9% and 3.7%, respectively). None of the RCT were powered to have safety as a major outcome, which makes them liable to type II error, i.e., lack of differences between treatment and placebo arms may be due to lack of power to detect differences (too few patients). We are conducting an overview of all Cochrane Reviews of biologics to assess whether a much larger sample would make clinically useful estimates of harms possible. Longterm surveillance studies and RCT with safety as the primary outcome are needed to provide these important data.
The dramatic effect of the biologics upon slowing or stopping progression of radiographic joint damage is pivotal information for patients and clinicians making decisions on whether, when, and which biologic to use. Potential advantages of golimumab are less frequent administration (every 4 weeks) compared to other subcutaneous biologics for RA, including etanercept (twice weekly), adalimumab (every 2 weeks), and anakinra (every day), and the subcutaneous route of administration, which patients can do at home, compared to IV administration in infusion units for infliximab, abatacept, and rituximab. These studies have not reported on radiographic outcomes.
Many findings in this review deserve further discussion. The ACR20/50/70 rates in the golimumab group (with MTX) in the approved dosage were 1.5–2.8 times higher than in the placebo + MTX group, and the DAS scores lower. The ACR50 rates are similar to those reported in the systematic reviews of other TNF blockers including etanercept11, infliximab12, and adalimumab13 and to other approved biologics for the treatment of RA, such as rituximab41 and abatacept15. Thus, in the absence of direct comparison studies, based on these data, golimumab seems to have an efficacy similar to the other biologics used for the treatment of RA. These efficacy data need to be confirmed for patients who fail current biologics, especially the TNF blockers, and who may be candidates for golimumab, since studies have shown that when one TNF blocker is not effective, another may be42,43.
The improvements in functional limitation as measured by HAQ were clinically meaningful and statistically significant in golimumab-treated patients. This is important since one of the goals with effective therapy is to improve the function of patients with RA. Mean improvement in HAQ scores was 0.20–0.25. Almost twice as many golimumab-treated patients achieved the minimally clinical important change (MCID) of > 0.22 units on HAQ, compared to the placebo-treated patients. Thus, at the approved dosage, golimumab seems to provide clinically meaningful improvements in function in patients with RA. Generic quality of life data have not been reported in these studies. Since these publications lag behind the original publications, we hope these data will be published in the near future.
The short-term safety for golimumab at the approved dosage of 50 mg every 4 weeks used in combination with MTX seems acceptable. The odds of withdrawals due to adverse events in golimumab-treated patients with RA compared to placebo were 0.80 (95% CI 0.26, 2.42). This is similar to the range reported for withdrawals due to adverse events with other commonly used biologics (in the range of 0.82–2.2)11,13,15,44,45. Golimumab was not associated with any more risk of infections, serious infections, lung infections, tuberculosis, cancer, or death than placebo. Included RCT were short-duration studies powered for efficacy, but not safety outcomes. Longterm surveillance studies are needed to assure patients and physicians of its safety. The FDA Website warns against the risk of serious infections, invasive fungal infections, hepatitis B reactivation, malignancies, heart failure, demyelinating disease, and new-onset psoriasis. In a recently issued safety alert, the FDA also warns against the “increased risk of lymphoma and other malignancies in children and adolescents treated with TNF-blockers.”
Only 2 RCT provided the data on head-to-head subgroup comparisons of golimumab to MTX, and in 1 study, patients had already failed MTX before being randomized. The rates of ACR20/50/70, DAS remission, good EULAR response, and proportion achieving HAQ MCID were not different between these groups. The safety profile of golimumab was similar to that of MTX with regard to number of adverse events, SAE, infections, serious infections, cancer, all withdrawals, and withdrawals due to adverse events. Golimumab-treated patients achieved a significantly greater improvement in mean HAQ score than MTX-treated patients (−0.25 vs −0.13). However, these results must be interpreted with great caution since this was a subgroup analysis, only 2 RCT provided the data, and in 1 study, patients had already failed MTX before being randomized to MTX or golimumab, thereby giving an efficacy advantage to the golimumab group36.
The quality of evidence found in the trials included in this review appears to be high because the studies reported adequate methods of allocation concealment, sequence generation using the interactive voice-response system, and adequate methods of blinding. Only 1 study lacked in addressing incomplete outcome data adequately36. All trials were funded by the manufacturers of golimumab.
Our study had several strengths and limitations. Review of studies, bias assessment, and data abstraction were performed independently by 2 reviewers to avoid errors. We are limited in our ability to comment on outcomes not reported in the published reports, such as assessments of radiographic data or HRQOL. Short study duration and study design with efficacy (but not safety) as primary RCT outcome make analysis of safety difficult. Some studies described the number of patients with more than 1 adverse event and others, the total number of adverse events. We combined these somewhat heterogeneous outcomes to make the data and results easy to interpret.
We found high heterogeneity (moderate to substantial) in ACR50 and number of adverse events for comparison of golimumab to placebo. Heterogeneity was low for other efficacy and safety outcomes for placebo. The heterogeneity in the ACR50 estimate was primarily due to estimates of Emery, et al39. Removal of this study decreased the heterogeneity to 0%. Heterogeneity in number of adverse events was primarily due to Smolen, et al38. Removal of this study decreased heterogeneity to 0%. This implies that our confidence in these estimates is not as high as for the other outcomes, and these results must be interpreted with caution. On the other hand, since all estimates are in the same direction, we doubt that newer studies will change the interpretation regarding these outcomes.
FDA warnings are important when considering the risks of these medications. Many risks associated with TNF blockers as a class are also applicable to golimumab. To our knowledge there are no published metaanalyses or systematic reviews of golimumab in patients with RA. Four recently published reviews summarized results from various studies of golimumab without performing metaanalyses46–49.
The favorable safety and efficacy profile of golimumab used in combination with MTX in patients with RA offers a new treatment option to patients with active disease who have failed MTX or other biologics including anti-TNF-α biologics. Direct head-to-head comparison studies are needed between golimumab and other biologics, and between golimumab and triple therapy (MTX, sulfasalazine, and hydroxychloroquine). RCT with safety as the primary outcome and phase IV postmarketing surveillance studies are likely to provide patients and physicians with much needed longer-term safety data.
Acknowledgments
We thank Louise Falzon of the Cochrane Library for designing and performing the electronic search.
Footnotes
-
This is a version of a Cochrane Review, which will be available in The Cochrane Library, Issue 1, 2010.
-
The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs.
- Accepted for publication February 3, 2010.