Is Lutikizumab, an Anti–Interleukin-1α/β Dual Variable Domain Immunoglobulin, efficacious for Osteoarthritis? Results from a bayesian network meta-analysis

Objective Most guidelines recommend the use of nonsteroidal anti-inflammatory drugs (NSAIDs), duloxetine, and tramadol for the nonoperative treatment of osteoarthritis (OA), but the use of them is limited by the tolerability and safety concerns. Lutikizumab is a novel anti–IL-1α/β dual variable domain immunoglobulin that can simultaneously bind and inhibit IL-1α and IL-1β to relieve the pain and dysfunction symptoms. We conducted this network meta-analysis to comprehensively compare the clinical efficacy and safety of lutikizumab with other drugs recommended by guidelines. Methods We conducted a Bayesian network and conventional meta-analyses to compare the efficacy and safety of lutikizumab with other traditional drugs. All eligible randomized clinical trials, in PubMed, CNKI, EMBASE, and Web of Science databases, from January 2000 to January 2020, were included. The Cochrane risk of the bias assessment tool was used for quality assessment. Pain relief, function improvement, and risk of adverse effects (AEs) were compared in this study. Results 24 articles with 11858 patients were included. Duloxetine (DUL) had the largest effect for pain relief (4.76, 95% CI [2.35 to 7.17]), and selective cox-2 inhibitors (SCI) were the most efficacious treatment for physical function improvement (SMD 3.94, 95% CI [2.48 to 5.40]). Lutikizumab showed no benefit compared with placebo for both pain relief (SMD 1.11, 95% CI [-2.29 to 4.52]) and function improvement (SMD 0.992, 95% CI [-0.433 to 4.25]). Lutikizumab and all other drugs are of favorable tolerance for patients in the treatment of OA compared with placebo. Conclusions Lutikizumab, the new anti–Interleukin-1α/β dual variable domain immunoglobulin, showed no improvement in pain or function when compared with placebo. Selective cox-2 inhibitors and duloxetine remain the most effective and safest treatment for OA. More high-quality trials are still needed to reconfirm the findings of this study.


Introduction
Osteoarthritis (OA) is the most common form of joint disease, usually affecting load-bearing joints such as hip and knee joints [1]. Approximately 302 million people suffer from OA worldwide every year [2]. OA can lead to local pain and joint stiffness in its early stages and can cause dysfunction and even disability in the late stages. OA-related pain and dysfunction increase the risk of mortality [3] as well as the societal economic burden [4]. To address the health issue, most guidelines recommend the use of nonsteroidal antiinflammatory drugs (NSAIDs), duloxetine, or tramadol for nonoperative treatment of OA [2]. However, the use of these drugs is limited by tolerability and safety concerns [5].
Previous literature has confirmed that the proinflammatory cytokines, Interleukin-1α, and 1β (IL-1α/β) are pain mediators and play an important role in the pathogenesis of OA [6,7]. Inactive IL-1α is stored in the cell or on the cell membrane. Once the cells are damaged, IL-1α is activated and released, inducing the activation of IL-1β, and finally promoting the progression of OA. [8,9]. IL-1α and IL-1β both bind to the IL-1 receptor 1 (IL-1R1), causing joint pain, inflammation, cartilage destruction, and bone resorption [10][11][12][13]. In addition, researchers have found that the concentration of IL-1 in the serum and joint fluid of patients with OA is elevated [14,15]. Subsequently, numerous IL-1R antagonists and IL-1R1 antibodies have been developed. However, clinical trials utilizing them in patients with OA did not report the desired results [16,17]. Lutikizumab is a new anti-IL-1α/β dual variable domain immunoglobulin that simultaneously binds and inhibits IL-1α and IL-1β without interfering with other IL-1 family members such as IL-1Ra [18]. Multiple animal experiments and clinical trials already have shown the potential of lutikizumab for the treatment of OA [19][20][21].
To comprehensively assess the clinical efficacy, including pain reduction and physical function improvement and the safety of lutikizumab for the treatment of OA, we designed and conducted a Bayesian network meta-analysis. Ten drugs widely used clinically were included in the meta-analysis. Based on these drugs' activity mechanism, we divided them into five groups: anti-Interleukin-1α/β dual variable domain immunoglobulins (lutikizumab), selective Cox-2 inhibitors (celecoxib and etoricoxib), duloxetine, opioid (tramadol), and traditional NSAIDs (ibuprofen, naproxen, diclofenac, and paracetamol/acetaminophen).
Reference lists of relevant systematic reviews and metaanalyses were also reviewed to identify additional eligible studies. Only randomized clinical trials (RCTs) were included, but no restriction was placed on the language of publication 2.2. Eligibility Criteria. The inclusion criteria were as follows: (1) Only randomized clinical trials (RCTs) with prospective parallel-group design; (2) Studies comparing the target drugs with each other or placebo in participants with OA at any joint. The exclusion criteria were as follows: (1) Doseescalation studies of only one drug; (2) Studies on postoperative patients with OA; (3) Reviews, systematic reviews and meta-analyses, case report, conference abstractions, letters, pharmacokinetical or pharmacodynamical studies, and animal experimental studies.

Quality Assessment.
Two authors conducted the methodological quality and bias assessment of included studies with the Cochrane risk of the bias assessment tool strictly. The following indexes were evaluated and ranked as low risk of bias, unclear risk of bias, or high risk of bias: sequence generation, allocation concealment, blinding, incomplete outcome data, selection outcome reporting, and other sources of bias [22]. All disputes were resolved through discussion.

Data Extraction.
Author, publication year, number of patients, mean age, gender ratio (male/female), diseased joint, funded or not, intervention methods, follow-up period, and outcome data were extracted from included studies. We would give priority to select the data from the intention-totreat analysis to reduce the withdrawal bias if available. For studies involving multiple treatment groups with different doses of the same drug, we selected the most effective dose group based on the respective study's recommendations [23].

Outcome
Measures. The primary efficacy endpoint was pain relief, and the secondary efficacy outcome was function improvement. Considering the differences between the baseline value of each included study, which may lower the reliability of the results and conclusions, the change-frombaseline score at the last follow-up (mean ± SD) was used to evaluate the efficacy to minimize the biases caused by heterogeneity of baseline values. No restriction was placed on the types of questionnaire used in pain evaluation. The function subscales of Western Ontario and McMaster Universities Arthritis Index (WOMAC) were used to evaluate the function improvements preferentially. Any other functional measurement scales, such as the Lequesne index, would be used if no WOMAC function score was reported. Standardised mean difference (SMD) was used because results from different scales were included in the same network [22].
The safety outcomes included the withdrawal due to adverse effects (AEs), serious AEs, and any drug-related AEs. Serious AEs included any AEs that resulted in death, was life-threatening, needed for hospitalization, or prolonged the existing hospitalization, caused disability/incapacity, or caused anomaly/birth defect. The odds ratio (OR) with 95% confidence intervals (CI) was used to measure the safety of target drugs versus placebo or against each other.
2.6. Statistical Analysis. Conventional direct meta-analyses comparing the efficacy and safety of treatments with placebo were conducted in Stata/MP (version 14.0, Stata Corp, College Station, Texas, USA). The heterogeneity across studies was tested by the Q and I 2 statistic, in which P < 0:05 or I 2 > 50% implies significantly heterogeneity. If significant heterogeneity across studies was found, a random-effects model would be used. Otherwise, a fixed-effects model would be preferred.
The random-effects Bayesian network meta-analyses were conducted in Aggregate Data Drug Information System (ADDIS, version 1. 16.8). This method can augment the number of studies within each comparison and narrow the CIs' 2 BioMed Research International width, and then increase the reliability of result and conclusion [24][25][26][27]. Noninformative uniform and normal prior distributions were used in this study, then four different sets of starting values were set to fit the model to yield 40000 iterations (10000 per chain) and obtain the posterior distributions of model parameters [28,29]. The thinning interval was set at 20 and the burn-ins at 1000 for each chain. Convergence of iterations was assessed using the Gelman-Rubin-Brooks statistic. Consistency of the network meta-analysis was reconfirmed via global inconsistency tests and node-split tests in Stata/MP (version 14.0). SMDs and ORs with 95% CI would be generated from the posterior distribution medians. Significant differences were considered between treatments being compared when the corresponding 95% CI did not contain 0 for the SMD or 1 for OR. Surface under the cumulative ranking (SUCRA) and the cluster-ranking plots were used to rank the efficacy and safety of different treatments. P < 0:05 was considered statistically significant.
The following subgroup analyses would be performed if available: according to the drug delivery route (topical, oral, or injective) and according to the diseased joint (hip, knee, hand, or ankle).

Study Selection.
This network meta-analysis was conducted strictly with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [30].

Study
Characteristics. 11858 patients were assessed in this study. Most of the 26 trials included studied knee or hip OA. Only two trials with 541 patients studied hand OA.
The details of patient baseline characteristics are presented in Supplementary Appendix Table 1. The methodological quality and bias-risk evaluations of all included studies are presented in Supplementary Appendix Table 2. Based on these results, the main contributing factors to risks of bias were performance bias, selection bias, and attrition bias.   Figure 2 and Table 2). According to the SUCRA value, DUL had the greatest effect on pain relief (SUCRA = 88:7%), followed by SCI (SUCRA = 88:4%), and lastly ALI (SUCRA = 28:6%). The detailed results of the SUCRA rank are presented in Supplementary Appendix Table 3.  (Figure 2 and Table 2). The results of most SUCRA rankings showed that the most efficacious treatment was SCI (SUCRA = 88:4%), and the least effective one was ALI (SUCRA = 29:6%) (Supplementary Appendix Table 3).

Safety Endpoint
3.5.1. Conventional Direct Meta-Analysis. Twenty trials involving all five therapies were analyzed in the conventional direct meta-analyses. There was no significant heterogeneity reported, and a fixed-effects model was used. DUL, NSA, and OPI had greater rates for all of safety endpoints compared with placebo, while ALI and SCI did not show a significantly higher risk for any safety endpoint. The details of the pairwise meta-analysis for all drugs compared with placebo are shown in Table 1.

Network Meta-Analysis.
Twenty-four trials involving all five treatments were analyzed in the safety network. Node-split tests and global inconsistency tests were per-formed, and no inconsistency was reported. The consistency model was preferred rather than the inconsistency model.
No treatment had more withdrawals due to adverse events (AE), nor a higher incidence of serious AEs, nor any drug-related AEs. Based on the results of the network comparisons, SCI had the lowest rate of withdrawal due to AEs (SURCA 92.6%, OR -0.11, 95% CI [-0.40 to 0.17]), the lowest rate of serious AEs (SURCA 80.2%, OR -0.01, 95% CI [-0.70 to 0.68]), and the lowest rate of drug-related AEs (SURCA 75.5%, OR 0.07, 95% CI [-0.10 to 0.24]). The cluster rank plots showed that SCI was the optimum treatment from the perspective of safety and efficacy (The results of clusterrank plots can be seen in Supplementary Appendix Figure 2). The relative safety between different treatments is presented in Table 3. The SURCA and relative safety compared to placebo are presented in Supplementary Appendix Table 4.
3.6. Subgroup Analysis. Two subgroup analyses were conducted.
The first subgroup analysis conducted exploited the impacts of different drug delivery routes. Three of the 26 trials used topical drug delivery methods. After excluding these studies, no substantial change was revealed. DUL had the largest efficacy for pain relief (

Discussion
This is the first network meta-analysis comparing the efficacy and safety of lutikizumab, the new anti-Interleukin-1α/β dual variable domain immunoglobulin, for treating OA with drugs recommended by guidelines [2]. We included all available evidence from randomized clinical trials (RCTs) directly or indirectly comparing lutikizumab with traditional treatments for OA and used the Bayesian method to increase the number of comparisons to enhance the power of the study. As mentioned above, considering the difference in the baseline values from different study populations and their influence on the results, we chose the change-from-baseline score as the outcome measure and only included the literature that reported the results of the change-from-baseline score. Our main findings are (1) ALI (lutikizumab) is not associated with pain relief or functional improvement of OA compared with placebo; (2) DUL, SCI, and NSA therapies all can improve every symptom of OA effectively and have a significant advantage over OPI and ALI; (3) SCI, ALI, DUL, NSA, and OPI are tolerated well for patients in long-term treatment of OA compared with placebo. These

BioMed Research
International results indicate that lutikizumab is not suitable for the treatment of OA because it provides no improvement in joint pain and dysfunction, while selective Cox-2 inhibitors (such as celecoxib and etoricoxib) are the ideal choice for the treatment of OA from the perspective of safety and efficacy. Combined with the results from other clinical trials [16,17], the inhibition of IL-α/β does not seem to be a new way to treat OA in the future.
There are several limitations in this study. Considering the unmanageable confounding factors in non-RCTs and their unpredictable influences on the results of network meta-analysis, only RCTs were included. Nevertheless, non-RCTs, especially observational studies, can provide valuable insight into the long-term effectiveness and safety of treatment for OA. To enhance the credibility of this meta-analysis, only high-quality studies were included. This may have contributed to the small number of studies included. Publication bias could be a significant problem for this study, especially the funnel plots that showed a dubious asymmetry. We tried to adjust the publication bias using the trimming and filling method. However, a previous study suggested that the results of the trimming and filling method should be interpreted as a sensitivity analysis rather than a corrected estimate of publication bias [55]. So, the results of this study should be interpreted cautiously, particularly for ALI in which the number of included studies is smaller compared with other treatments. Although we have conducted two subgroup analyses to reduce the impact of potential confounding factors, there are still many other factors that could affect the reliability of the results, such as the differences in comorbidities, duration of OA, and grade of OA in the study populations. For instance, comorbidities usually cause worse symptom management and consequentially affect the results of analgesic effectiveness assessment. Paradoxically, research on analgesics often excludes people with clinically significant comorbidities and does not systematically describe the distribution of comorbidities in the study population. Most of the included studies, coincidentally, failed to report an accurate grade or duration of OA. We were unable to adjust for these factors because of the insufficiency of the related data, and thus those results should be interpreted with caution. More high-quality trials are needed.

Conclusion
24 studies, involving 26 trials assessing 11858 patients, were included in this network meta-analysis. The results show that lutikizumab, the new anti-Interleukin-1α/β dual variable domain immunoglobulin, did not improve pain or function in the comparison with placebo. Selective cox-2 inhibitors and duloxetine remain the most effective and safest treatment for OA. More high-quality trials are needed to reconfirm the findings of this study.

Data Availability
Data generated or analyzed during this study are included in this published article. The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethical Approval
This article does not contain any studies with human participants or animals performed by any of the authors.

Disclosure
The study funders/sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Conflicts of Interest
The authors declare that they do not have any competing interests.

Authors' Contributions
Pengcheng Dou and Ren Wu conceived the study, participated in its design and coordination, and critically revised the manuscript. Ziqin Cao, Wanchun Wang, and Shuo Jie had full access to all of the data collection, analysis, and interpretation. Ziqin Cao and Yajia Li drafted the manuscript. Jian Zhou and Xuantao Hu were study investigators and contributed to the process of data collection. Yihan Li and Zeling Long contributed to the process of study research and selection. Tong Wu and Dilihumaer Aili contributed to the risk of bias assessment. All authors read and approved the final manuscript.