TNF-α inhibitors can increase the risk of respiratory infection when used in rheumatism: a meta-analysis and systematic review

Background The risk of tumor necrosis factor-α(TNF-α) inhibitors (infliximab, for the treatment of rheumatic diseases leading to infection events has not yet been established. This meta-analysis aims to assess the risk of developing serious infections of three TNF-α inhibitors for rheumatic diseases. Methods A systematic literature search of Pubmed, Embase and Cochrane Library was conducted through December 2018. Selecting the RCTs which subjects were diagnosed as rheumatoid diseases according to ACR criteria or other authoritative diagnostic criteria and over 18-year-old. Finally, RCTs with Jadad score greater than or equal to 4 were included in this meta-analysis. The Odds Ratio (OR), Confidence Interval (CI) and p value were calculated to assess the risk of serious infections. Results 34 RCTs involving 14166 subjects were included, including 11 RCTs for infliximab, 7 RCTs for etanercept, and 16 for adalimumab. Meta-analysis demonstrated that, with the pooled OR of 1.29 (95%CI 1.04 to 1.60), the TNF-α inhibitors group had a higher risk of serious infection than control group. In the subgroup analysis, infliximab and adalimumab had a higher risk of serious infection than control group, and the pooled ORs were 1.48 (p=0.03) and 1.47 (p=0.03), respectively. For other infections including pneumonia, upper respiratory infection, and nasopharyngitis, the risks of these adverse events were higher in experimental group than control group, while the risk of tuberculosis were not, with the pooled OR of 2.31 (p=0.08. Conclusions TNF-α inhibitors, especially infliximab and adalimumab, can increase the risk of infections. Among the infections, pneumonia, upper respiratory infection and nasopharyngitis have higher risks in TNF-α inhibitors group than control group. As a result, we summarized that TNF-α inhibitors can increase the risks of respiratory infection when used in rheumatic disease. It is suggested that clinicians should pay attention to the prevention of respiratory infections when using TNF-α inhibitors, so


Background
TNF-α is a cytokine with many physiological activities. Normal level of TNF-α can regulate the immunity, fight against bacterial, viral and parasitic infections, promote tissue repair, and mediate the apoptosis of tumor cells. However, overexpression of TNF-α will lead to a disorder of immunity, participate in inflammatory reactions, tissue damages, shock and other pathological processes.
Abnormal TNF-α concentration and TNF receptor signaling have been shown to contribute to many diseases, including rheumatoid arthritis, Crohn's disease, atherosclerosis, psoriasis, sepsis, diabetes, obesity, et al [1] .
Since TNF-α inhibitors can specifically bind to overexpressed TNF-α in vivo and reduce its level, it has a widely application in various rheumatic diseases such as rheumatoid arthritis (RA), ankylosing spondylitis (AS), inflammatory bowel disease (IBD), and psoriasis. However, excessive usage of TNF-α inhibitors will form a low level of TNF-α, affect its normal physiological functions, and lead to infections, even tumors.
This meta-analysis searched the published RCTs involving infections, and analyzed OR values of tuberculosis, pneumonia, bronchitis, upper respiratory infection, nasopharyngitis, sinusitis, gastroenteritis, urinary tract infection, cellulitis, sepsis and abscess. Compared with other meta analyses in the past five years [10][11][12][13][14][15][16] , the innovation points were as follows: (1) This meta-analysis focused on specific types of infection. (2) Studies included in this meta-analysis all were RCTs, which avoided the deficiency of observational studies. (3) The number of included RCTs was much more than other two meta analyses [12,13] , so the evaluation of infections was more objective and comprehensive. (4) One meta-analysis had 71 studies [14] , but the funnel plot, Begg's test (p = 0.06) and the Egger's test (p = 0.09) suggested a publication bias, which might overestimate the risk of infections. So far, Bongartz T and leombruno JP published two meta-analysis that systematically and objectively described the risk of infections caused by TNF-α inhibitors in the treatment of rheumatic diseases [15,16] . However, these two papers were published before 2009, and the RCTs after 2009 should be added. Few consensuses on the risk of infections caused by TNF-α inhibitors have been reached. This meta-analysis systematically retrieved studies and assessed the risk of infections caused by three TNF-α inhibitors, and extended into specific types of infections innovatively.

Inclusive Criteria Types of Studies
RCTs about infliximab, etanercept, and adalimumab in the treatment of rheumatic diseases were included.

Types of Subjects
Subjects, whose age was greater than or equal to 18 years old, had been diagnosed RA, AS, IBD or psoriasis according to American College of Rheumatology criteria or other authoritative and recognized diagnostic criteria. Disease progression, race, nationality and complications are not limited.

Types of Interventions
The experimental group was treated with TNF-α inhibitors, with or without disease-modifying antirheumatic drugs(DMARDs). The control group was treated with placebo, with or without DMARDs, or with DMARDs alone.

Exclusive Criteria
RCTs that accord with any of the following criteria will be excluded: (1) studies with no detailed records of infections and no mention of infections, where severe infection is defined as an infection requiring intravenous antibiotic treatment or requiring hospitalization or life-threatening infection; (2) if both the studies are the same patients in different periods, only the study with the longest follow-up time and the most comprehensive record is selected; (3) studies with improper control group, for instance, the study that allocate TNF-α inhibitor monotherapy in experimental group and DMARDs monotherapy in control group;(4) studies with Jadad score less than or equal to 3 points; (5) studies not published in english; (6) studies with the full text not available.

Data Sourses and Search Strategy
We searched MEDLINE, EMBASE and the Cochrane Library with the terms infliximab; etanercept; adalimumab; rheumatoid, arthritis; ankylosing, spondylitis; Crohn' disease; ulcerative, colitis; psoriasis. In a pattern of drug-disease retrieval, using a limitation of clinical trial, we form the retrieval expressions that adapt to different databases. The deadline for retrieval is December 2018.

Assessment of risk of bisa
Two review authors independently assessed each study. The results were classified as low risk, unclear risk, and high risk. Disagreements were resolved by consensus. The assessments included: (1) Random sequence generation (selection bias). (2) Allocation concealment (selection bias). (3) Blinding of participants and personnel (performance bias). (4) Blinding of outcome assessment (detection bias). (5) Incomplete outcome data (attrition bias). (6) Selective reporting (reporting bias).

Statistical analysis
Revman5.3 software was used for data processing. Odds ratios (ORs) and 95% confidence intervals(95%CIs) were used. Heterogeneity analysis used χ 2 test and considered its size in combination with p value. P< 0.05 indicates large heterogeneity. Random effects model is adopted and subgroup analysis and sensitivity analysis are performed. Reversely, p> 0.05 indicates low heterogeneity, and fixed effects model is adopted.

Search Results
After a preliminary screening through browsing the abstracts, 157 articles were obtained.
The incidence rate of serious infections in etanercept group(2.2%) was compared with that in the control group (3.0%), and the results were not statistically significant with the pooled OR of 0.77 (95%CI 0.48-1.23, p = 0.28). Subjects in these studies had been initially screened for a history of tuberculosis through the PPD skin test or chest X-ray. Only one of the 18 subjects with tuberculosis was in control group, and the remaining 17 subjects were all subjects of the experimental group. The consistency test showed a low heterogeneity, but the difference was not statistically significant with a pooled OR of 2. Conservative results of adalimumab subgroup were obtained using the random effect model (OR1.25, p = 0.09). Finally, only infliximab subgroup showed statistical significance with the pooled OR of 1.58(95%CI 1.08-2.31, p = 0.02). What should be noted is that the results of two adalimumab RCTs [2,32] were statistically significant, with OR of 4.04(p = 0.01) and 1.77(p = 0.04), respectively. important to noted that one RCT [5,18,19] showed statistical significance with the OR of 3.41 (p = 0.01).

Summary
The information of infections reported in 34 studies was extracted for meta-analysis. Results showed that the application of TNF-α inhibitors lead to increased risk of pneumonia, upper respiratory infection and nasopharyngitis. Odds ratios and confidence intervals were shown in Table2.

Discussion
In this meta-analysis, a total of 34 RCTs with JADAD score greater than or equal to 4 were included, which were considered as RCTs closely correlated to the research content through full-text browsing.
The significance level α was 0.05.

Serious Infections
Results of 34 RCTs suggested that, when used in rheumatic diseases, TNF-α inhibitors such as infliximab, etanercept, and adalimumab, can elevated the risk of serious infections. The risk of serious infections in the intervention group was 1.29 times higher than that in the control group (p = 0.02).
This result was consistent with some previous meta-analysis [15] . Furthermore, there was a metaanalysis suggesting the recommended dose of TNF-α inhibitor did not upgrade the risk of serious infections [16] . According to the results of subgroup analysis, the risk of serious infections in infliximab and adalimumab treatment group was significantly higher than that in the control group (p = 0.03).
However, when it comes to etanercept, no similar result can be reached(p = 0.28). This result is identical with a real-world study for reasons that are not yet clear [47] . A meta-analysis tried to explain it [48] , and put forward a view that infliximab and adalimumab were TNF-α monoclonal antibodies, while etanercept was TNF-α soluble receptor. This may determine their risk of infections. It can be concluded that etanercept is more secure than infliximab and adalimumab in the risk of serious infections, which is valuable for clinicians making clinical decisions.

Detailed Infections Tuberculosis
TNF-α did not increase the risk of tuberculosis (p = 0.08). However, of the 10 RCTs reporting tuberculosis, 17 out of 18 cases were in the intervention group, suggesting that the use of TNF-α inhibitors may increase the risk of tuberculosis, although the difference could not be measured due to the low incidence rate or the scarce cases. A meta-analysis reported that the risk of tuberculosis using TNF-α inhibitors was 1.94 times as high as that of the control group (p = 0.02). Subgroup analysis revealed that the risk of tuberculosis in the intervention group of RA patients was 2.29 times higher than control group (p = 0.03). However, the evidence quality assessed by GRADE was "Low" [49] . Other real-world studies have shown that the risk of tuberculosis with infliximab and etanercept is 3 to 4 times higher than that with adalimumab, and some patients developed tuberculosis after several years of drug withdrawal [50] . Thus, TNF-α inhibitors are associated with a long-term risk of tuberculosis. Before treatment with TNF-α inhibitors, patients must be screened for tuberculosis. For patients with latent tuberculosis infection, preventive anti-tuberculosis treatment should be carried out. Through screening and preventive treatment, lowering the risk of tuberculosis and improving the prognosis of patients with rheumatism.
Besides the infections mentioned above, we selected pneumonia, upper respiratory infection, nasopharyngitis, bronchitis, sinusitis, gastroenteritis, urinary tract infection, cellulitis, sepsis, abscess for statistical analysis. Pneumonia TNF-α inhibitors mildly increased the risk of pneumonia over time, and only infliximab significantly increased the risk of pneumonia after stratified analysis. Infliximab was 4.48 times more likely to cause pneumonia. Frust DE suggested in his review that the most common sites of infection treated with biologics were respiratory tract (including pneumonia), but the risk was not quantitatively calculated [51] . This meta-analysis further proved that infliximab could increase the risk of pneumonia, providing a basis for clinicians to make decisions and strengthen pharmacovigilance. Similarly, without quantitative analysis, Frust DE in his review concluded that respiratory tract infections were common [51] . In this meta-analysis, quantitative analysis was performed. It is confirmed that TNF-α inhibitors, infliximab in particular, can increase the risk of upper respiratory infection and nasopharyngitis.

Upper Respiratory Infection And Nasopharyngitis
Only pneumonia, upper respiratory infection and nasopharyngitis showed statistical significance, with pooled OR of 1.72 (p = 0.03),1.38 (p = 0.006) and 1.33 (p = 0.002), respectively. It is indicated that TNF-α inhibitors can moderately increase the risk of pneumonia, upper respiratory infection and nasopharyngitis. Regrettably, there is a lack of evidence on bronchitis, sinusitis, gastroenteritis, urinary tract infection, cellulitis, sepsis and abscess; other larger RCTs are needed.
The dose of TNF-α inhibitor also affects the occurrence of infections. A meta-analysis found that there was no significant difference among the medium-and low-dose group with the control group, but the high-dose group had a significant higher risk(p = 0.04) [48] . Another meta-analysis showed that the risk of infections was 1.4 times higher in low-dose group than control group [15] . A systematic review of observational studies had also shown that only when TNF-α inhibitors were used in excess of normal doses will the risk of infection increase [52] . The dose-effect relationship suggests that TNF-α inhibitors are indeed associated with infections.
Limitations exist in this meta-analysis. Firstly, some studies may be missed due to manual retrieve. Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.

Availability of data and materials
All data generated or analysed during this study are included in this published article [and its supplementary information files]. The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
Not applicable.