Introduction
Rheumatoid arthritis is a chronic autoimmune disease characterized by inflammatory polyarthritis that mainly affects the small joints [
1]. Biological disease-modifying antirheumatic drugs (bDMARDs) have emerged as an important advancement in the treatment of rheumatoid arthritis [
2]. There are several types of biologics, each of which targets a specific type of molecule involved in the pathogenesis of the disease. These include tumor necrosis factor alpha (TNF-α) inhibitors, such as etanercept, adalimumab, infliximab, certolizumab pegol, and golimumab. Other biologics that target other molecules include abatacept (a selective co-stimulation modulator that inhibits T-cells), rituximab (B-cell inhibitor), tocilizumab (IL-6 receptor antagonist), and anakinra (IL-1 receptor antagonist).
Despite the increasing number of biologics, the ability to achieve complete remission in certain RA patients remains challenging. Approximately 66% of RA patients failed treatment with TNF inhibitors in 6 months of follow-up [
3], and a minimum of 10% who tried a second bDMARD had their medication stopped due to lack of response [
2]. This suggests that there is a significant proportion of patients who do not respond to bDMARDs.
Several observational studies have identified different predictors of remission in RA patients receiving biologics [
4‐
24]. However, many of these predictors remain inconsistent. Some studies showed that old age, female gender, smoking history, obesity, presence of comorbidities, increased disease activity at the time of diagnosis, increased disease duration, and poor functional status at baseline have been associated with a lower response rate to biologics [
4‐
6,
8,
9,
14,
15,
17]. While other studies showed no significant association between age, gender, and remission rate [
15,
18,
23]. Patients with elevated ESR at the time of diagnosis have also shown poor response to biologics in some studies [
9,
14], but there was no significant association in other studies [
8,
17]. A meta-analysis was also conducted in 2018 to assess for the predictors of remission in RA patients regardless of the treatment that the patients received [
4]. In this study, we conducted a systematic review and meta-analysis to assess the strength of association between these predictors and the rate of remission in RA patients treated with bDMARDs.
Methods
We conducted this systematic review and meta-analysis based on the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-analysis [
5], and Meta-analysis of Observational Studies in Epidemiology [
6].
Data sources and search strategy
We performed a comprehensive search for published studies indexed in PubMed, Embase, and Web of Science databases from inception through May 1, 2022. We also performed a manual search for additional relevant studies using references of the included articles. The following search terms were used: “biologics OR Etanercept OR Infliximab OR Adalimumab OR Certolizumab OR Golimumab OR Anakinra OR Tocilizumab OR Sarilumab OR Abatacept OR Rituximab” AND “relapse OR remission” AND “arthritis OR rheumatoid OR rheumatoid arthritis” AND “risk factors OR predictors.” The search was not limited by language, study design, or country of origin. Two investigators (YK and AB) independently performed the literature search, screened using a priori criteria, and shortlisted the studies for final review. The bibliographic software EndNote was used for screening. Any discrepancies were resolved by a third reviewer (SG).
Inclusion and exclusion criteria
Studies meeting the following inclusion criteria were included: (1) full-text peer-reviewed publications of retrospective or prospective, cohort or case–control studies, (2) assessed for predictors to response to different types of biologics in RA patients, and (4) reported odds ratio (OR) for this association after multivariate analysis and adjustment of potential confounding factors. We excluded conference abstracts. We also excluded studies reported data based on hazard ratio or univariate analysis rather than multivariate analysis.
The following data were extracted from the studies: study characteristics (author, publication year, study design, country of origin, and study population), patients’ baseline characteristics, the follow-up duration, and variables that were adjusted in a multivariable analysis. Risk factors that were assessed in at least three studies were included in the meta-analyses. Two investigators (YK and AB) independently extracted the data from the articles, and discrepancies were resolved by a third reviewer (SG).
Statistical analysis
We performed a meta-analysis of the included studies using Review Manager 5.3 (Cochrane Collaboration, Copenhagen) and Comprehensive Meta-Analysis 3.3 software (Biostat, Englewood, USA). Multivariate adjusted odds ratios (OR) for individual studies were pooled using a random-effects model and reported using a 95% confidence interval (CI) for each risk factor where applicable. Pooling was undertaken if at least three studies reported an odds ratio for a given risk factor. A
P value < 0.05 was considered statistically significant. Heterogeneity was assessed using the Higgins
I2 index, where
I2 values > 50% implied the presence of significant heterogeneity [
7].
Sensitivity analysis
To evaluate the robustness of results, leave-one-out analysis was attempted for risk factors reported by ten or more studies.
Bias assessment
We assessed the quality of the included studies using the Newcastle–Ottawa Scale [
8]. Two authors (YK and AB) independently assessed each study for bias. For risk factors reported by ten or more studies, publication bias assessment across studies was performed qualitatively by visualization of the funnel plot [
9] and quantitatively, using Egger’s regression analysis [
10]. A
P value was generated using Egger’s analysis, and a value of < 0.05 was associated with significant publication bias. If bias was present on Egger’s test, further statistics using the Fail-Safe N test and Duval and Tweedie’s “Trim and Fill” test were used to ascertain the impact of the bias.
Discussion
Biologic therapies have successfully revolutionized the management of RA. However, there is a significant proportion of patients who do not respond to the treatment. Identifying the predictors that will affect the treatment response before starting medications with known serious side effects remains challenging. We preformed this systematic review and meta-analysis to investigate the strength of association between different predictors and remission rate in RA patients treated with biologics. In this analysis, 67% of patients achieved complete remission of disease after a follow-up period of 6–12 months. Remission criteria was defined as DAS28 score of less than or equal to 2.6 or SDAI score of less than or equal to 3.3. Old age, female gender, smoking history, obesity, high disease activity at the time of diagnosis, poor functional status, and elevated ESR were associated with lower remission rate. On the other hand, positive ACPA at the time of diagnosis has been associated with higher remission rate. While disease duration, positive RF, prior or concurrent use of steroid, prior or concurrent use of MTX, high TJC, and high SJC score at the time of diagnosis were not significantly associated with lower remission rate. These results were consistent with those treated with TNF-α inhibitors alone.
Many studies supported our findings that women with RA had worse progression of the disease as compared to men despite being on similar treatment [
32]. Similar findings have been reported by other studies [
33‐
35]. It has been demonstrated that men and women respond differently to the same treatment due to physiologic differences. Another explanation to our finding is that we used the DAS28 score, which is highly dependent on pain perception, to assess for disease remission. Men may have a higher threshold for reporting joint tenderness which lowers their score. However, we cannot exclude the possibility that men may have a form of the disease that remits more often in comparison with women. Regarding age, our study showed that patient aged > 55 years old were responding poorly to biologics which contradicts the results of other registries that showed no effect of age on response to biologics [
11,
36]. Older patients are more likely to have long disease duration which may negatively affect the therapeutic efficacy of biologics. Moreover, elderly patients usually have multiple comorbidities at baseline that make biologic agents potentially more dangerous which results in early discontinuation of these medications.
Obesity, defined as BMI > 30, was found to be a poor predictor of remission in patients receiving biologics. Studies showed that the adipose tissue produces pro-inflammatory cytokines such as TNF-α, and IL-6. The higher fat mass, the higher concentrations of these cytokines which may affect the therapeutic response [
37]. Moreover, being a current or former smoker decreases the chances of response to biologics. Smoke acts on both cellular and humoral immunity that leads to a systemic proinflammatory state [
38,
39]. Chronic cigarette smoking appears to trigger various morphological, physiological, and enzymatic changes that impairs inflammatory responses [
38‐
40].
Regarding MTX, only 15–20% of our included patients received biologic drugs without prior or concurrent use of MTX. Our analysis showed that MTX prescription at baseline has no significant association with remission. Results were consistent among patients who received TNF- α inhibitors in combination with MTX. Our findings contradict the outcomes of a randomized controlled trial that was conducted in 1998 to investigate the impact of concurrent use of MTX with infliximab in 101 patients with RA [
41]. That study showed that MTX has been associated with reduced immunogenicity of infliximab after repeated infusions which helped improve the clinical response. Our results also contradict the outcomes of a network meta-analysis that was conducted in 2019 that also showed that combination therapy of MTX with biologics improved clinical response as compared with biologic monotherapy [
42]. Although many studies show that biologic use with MTX improves the clinical outcomes, this should not be considered as a standard of care for different reasons. First, many prescribers require MTX failure before starting biologics. Second, many patients prefer starting MTX prior to biologics because of the cost, and potential side effects. So far, we do not know whether starting biologic treatment rather than MTX improves long-term prognosis given that most of the patients included in the studies were started on MTX prior to biologics. On the other hand, our results should be further investigated by looking at the clinical background of the patients who were started on MTX and those who tried biologics without prior use of MTX. Studies showed that positive RF, younger age at symptom onset, and higher baseline disease activity are associated with higher rates of MTX failure [
43]. Further subgroup analysis should be conducted to eliminate the effect of these confounders before making a conclusion.
Currently, there is no biomarker that is known to predict response to biologics in RA patients. Our analysis showed that RF was not significantly associated with poor response to biologics. However, elevated ESR of more than or equal to 20 mm per hour was found to be a significant poor predictor of remission. While patients with positive ACPA showed high remission rate in response to biologics. Several studies reported no relationship between RF or ACPA positivity and the clinical response to tocilizumab treatment [
18,
44,
45]. In fact, ACPA positivity has emerged as an important predictor of response to biologics. A post hoc analysis of the AMPLE trial in 2016 initially showed that baseline ACPA positivity was associated with a better response to abatacept and adalimumab [
46]. Such association can be explained by the fact that ACPA exert their biological functions by binding to the Fc receptors, expressed particularly by immune cells of the myeloid lineage, and activating the complement system via the classical and alternative pathways [
47]. Given that most of the biologics work on inhibiting T-cells, B-cells, and their products of antibodies and inflammatory cytokines, partially explains their relative effectiveness in patients with positive ACPA [
48].
Several limitations to our meta-analysis should be mentioned. First, our included studies had inherent bias given their observational nature. Second, there was a significant heterogeneity among the studies that investigated several risk factors such as age, female gender, obesity, smoking, prior use of MTX, baseline functional status, positive RF, and elevated ESR. This heterogeneity could be due to difference in remission criteria, variation in patient demographics, and absence of consistent follow-up period among the studies. Despite the use of the random-effects model to assess for heterogeneity, our results should be interpreted carefully. Third, our study included some methodological limitations that need to be considered while interpreting the results. In our included studies, the patients treated with biologics had long-standing disease and had failed several previous DMARDs. The evaluation of disease remission in these patients using the DAS28 scoring system is tricky given that joint pain and swelling could result from structural and permanent damage due to prolonged disease course. In addition to that, we used ESR value of more than 20 mm/h as a poor predictor of biologics. However, ESR level significantly increases with age, so higher cutoff values should have been considered positive given that most of our patients are older than 40 years old. Moreover, patients were followed-up for an average of 6 months in most of the included studies, and only six out of twenty-one studies had a follow-up period of more than one year which may have affected the response rate to biologics [
4]. Finally, some risk factors were excluded given that they were reported in less than three studies such as family history and elevated CRP.
Despite these limitations, our study has several strengths. Up to our knowledge, this is the first meta-analysis that summarizes the available literature and provides a quantitative assessment of different risk factors associated with remission. Moreover, our analysis reported a large cohort of 16,934 patients from twenty-one studies. We also performed sensitivity analysis to the risk factors reported by ten or more studies, and no publication bias was detected in any of them. Finally, our results remained consistent when we preformed subgroup analysis for TNF inhibitors.
In conclusion, RA patients who are females with advanced age, obesity, smoking history, poor functional status, high disease activity, and elevated ESR at the time of diagnosis showed significantly decreased rate of disease remission after receiving biologics. On the other hand, positive ACPA, and prior use of MTX can increase remission rate in these patients. These predictors should be taken into consideration before starting medications with known serious side effects like biologics. Our findings might help develop a clinical prediction model to estimate the rate of remission in RA patients treated with biologics.
Supplementary Fig.
6 Sensitivity analysis for: A, age. B, female gender. C, disease activity. D, disease duration. E, prior use of MTX.
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.