Tuberculosis in Children with Rheumatic Diseases Treated with Biologic Disease-Modifying Anti-Rheumatic Drugs

Chronic rheumatic diseases entail the use of biologics in children. Immunosuppressive effects of drug therapy put children at risk of various infections including tuberculosis (TB). Even though TB is a major concern among individuals on biological DMARDs, the incidence and distribution among children on these drugs is not known. Hence, we performed a literature search to ascertain the prevalence of tuberculosis amongst children with rheumatic disorders treated with biological agents. Articles available on MEDLINE and SCOPUS published on or after January 1, 2010 to 1 October 2019 were reviewed and collated. We found that published data on TB infections in children with rheumatic disorders on biologics is scant even from regions with highest TB burden. Tuberculosis was reported on occasion (0–5 cases per country) in the developed world with most reports being from Turkey. While most of the retrospective studies suggest that TB risk is minimal in the paediatric rheumatology patients, prospective studies suffer from a short observation period. Most registries focus on response to therapy rather than complications. In this review we have then discussed about the variation in screening strategies for latent TB and the role of bacille Calmette-Guerin (BCG) vaccination. Based on the dearth of data and inconsistency in data collection, we propose a way forward in the form of establishing well-designed long-term prospective national registries from countries with high background prevalence of TB with focus not only on treatment efficacy but also on adverse events and infections.


INTRODUCTION
Despite the advent of glucocorticoids and immunosuppressive therapies, chronic rheumatic diseases of childhood such as Juvenile Idiopathic Arthritis (JIA), Systemic Lupus Erythematosus (SLE), Idiopathic Inflammatory Myositis (IIM), Auto-inflammatory Syndromes (AIS)and Paediatric Vasculitis (PV) result in significant morbidity, and, at times, even mortality. [1][2][3] In the developing world, infections are the leading con-tributors to such morbidity. Tuberculosis (TB) is one such infection, which remains a particular challenge in these parts of the world. 4 The emergence of drug-resistant tubercular strains and polypharmacy, in the setting of chronic illnesses further compounds the problem. 5 Recent estimates suggest the prevalence of TB in India to be 3.2 cases per thousand population. 4 The presence of rheumatic disorders (RDs) entails treatment with glucocorticoids and immunosuppressive drugs for prolonged periods, more so in cases of lupus, vasculitis and myositis. Some patients with JIA, lupus, vasculitis, and, rarely, IIM, also have underlying antibody deficiencies or complement pathway defects, further increasing their infection risk. Over the past years, there have been efforts towards decreasing usage of glucocorticoids in rheumatic disorders and advocating rational use of immunosuppressive agents. In addition to this, public health initiatives have attempted to address the issues of adequate treatment of TB. 2 The changing dynamics of therapeutic practices could have a bearing on the prevalence of TB in these diseases, and also influence the ways this problem can be addressed. Thus, it is important to understand the prevalence, risk factors, and outcomes of TB infection among children with RDs on biologics. In this review, we have performed a literature search on the prevalence, screening strategies, and global reporting patterns of TB across various studies among children with RDs on biological DMARDs. We have then summarised the available literature and discussed the possibilities that could explain our findings. Finally, we have suggested the way ahead to obtain more robust information from underrepresented countries.

REVIEW STRATEGY
The search strategy for writing review articles as proposed by Gasparyan et al. was followed.6 Articles available on MEDLINE and Scopus, published on or after January 1, 2010, until October 1, 2010 were reviewed using search words "juvenile" and "dermatomyositis" and "biologics" (n=71); "paediatric" AND "Lupus" AND "biologics" (n=81); "paediatrics" AND "Vasculitis" AND "Biologics" (n=55). In addition, for the literature review on registry data in paediatric rheumatology, Scopus searches were conducted combining "registry" with each of the following: "paediatric" AND "Lupus" (n=100), "juvenile" and "myositis" (n=40); "juvenile" and "arthritis" (n=368); biologics" AND "Rheumatology" (n=359) and "Autoinflammatory" AND "syndromes" (n=50). Also, select review articles on the subject were cross-referenced to obtain additional references. Figure 1 summarises the search results. Articles with data on outcomes in children of treatment with biologics were included. Review articles, systematic reviews, case reports, and articles without data in children, and those in languages other than English, and where full-text was not available were excluded. Congress abstracts did not feature in the searches. Studies which had TB where there was no clear separation between those receiving bDMARDS vs those on csDMARDs were excluded. Serious infections were defined as per the publishing author's definitions. The Zotero software, an open-source tool, was used for references management and citations.

SELECTION OF ARTICLES Screening by title
The Scopus searches were imported into Zotero, and articles were first screened by title by one author, and those without relevance, systematic reviews, meta-analysis, narratives, and in languages other than English were removed (Figure 1). The exact process of data extraction is elaborated in the supplementary material.

Juvenile Idiopathic Arthritis and Tuberculosis
Juvenile idiopathic arthritis is a chronic rheumatic disorder consisting of polyarticular (rheumatoid factor positive and negative), oligoarticular, systemic-onset JIA, enthesitis-related-arthritis, psoriatic and undifferentiated subtypes. The occurrence of infections is known and associated with poor outcomes. 7 Tuberculosis is a chronic infection that can result in significant morbidity and mortality in children with JIA. 8 Data in JIA consists of mixed cohorts of various subtypes of arthritis. Interestingly, most series report no occurrence of Tuberculosis (Tables 1, 2 and 3). Tuberculosis has been reported in four prospective studies, involving 2 each from Turkey and Portugal, and 1 each from Brazil and a multicentre trial. The follow-up duration in these studies ranged over 1-5 years. Of the various biologic registries screened, the only two cases of Tuberculosis reported are from Turkey. This is in contrast to minimal or no reports of Tuberculosis from UK, most European countries (France, Germany, Italy, and Greece) and Canada. The general prevalence of tuberculosis in Turkey is26/100,000 (2005). Brazil has one of the highest TB burdens with over 70,000 incident cases per year ( Figure 2D). Portugal has the highest TB prevalence in Western Europe at 23 per 10,000 population, which resonates with the 2 cases reported of two studies in 232 patients. 9 Interestingly, a study from India which has one of the highest background prevalence of Tuberculosis in the world, reported no Tuberculosis though the follow-up duration was 11 months. Plotting data available from various studies in paediatric rheumatology on a world map reveals the distribution is primarily limited to regions with low TB prevalence (Figure 2A). There is sizeable risk of confirmation biases regarding the safety of biologics resulting from absence of data from high TB incident parts of the world (Figure 2 C). On the contrary, in adults, there are reports of greater tuberculosis on anti-TNFs, with the risk being highest with IFX(cumulative incidence 0.5% within the first 500 days of registration) as compared with ETA (0.2%). 10,11 It is worthwhile considering if BCG vaccination practices in children could explain differences between children and adults. Usage of biologics also induces an immunosuppressant state, and there is known risk of higher extra-pulmonary forms of TB in such a setting. 12 Diagnosing these could be a challenge, particularly so in the absence of a robust biomarker for extrapulmonary forms of tuberculosis. 13 Juvenile Lupus and Tuberculosis Data on tuberculosis in paediatric lupus is scant, being limited to 7 retrospective and 2 prospective studies (Supplementary Table S1). While most described the use of Rituximab, one prospective study on Belimumab featured 39 cases over 6 months of follow-up. The maximum duration of follow-up was 3 years and the largest series of 104 was from the United States in 2015. Whilst none of the series reported any tuberculosis, the largest series had overall 22 infections, of which 20 were major infections. Of note, most patients were on concomitant immunosuppressants or steroids during the study period. However, literature is replete with case reports of tuberculosis in lupus. 14 We have previously found TB in 6% of children with LN. 5 Thus, poor tuberculosis reporting could be from use of biologics in patients with less severe disease (minor organ manifestations), early mortality or underreporting. Previously use of high-dose cyclophosphamide (CYC) has been identified as a risk factor for infections in lupus. (15)The risk of infections could possibly be lower with biologics such as belimumab and RTX but this needs to be confirmed in larger studies. (16) Juvenile inflammatory myositis and tuberculosis Out of the various studies on inflammatory myositis, none looked at data on Tuberculosis specifically ( Supplementary  Table S2) suggesting dire need to collect information relevant to this in future studies. On the other hand, we have 4 papers previously describing high prevalence of tuberculosis in myositis, suggesting the need for careful assessment of this aspect in prospective cohorts with longer term follow-up. [17][18][19][20] We have described TB in 17.1% children from India with myositis (n=35, unpublished data). Unfortunately, biologics use is limited in this part of the world due to insurance policies and consequent financial constraints further leading to dearth of data.

Juvenile Vasculitis and Tuberculosis
Data on paediatric vasculitis is scant, being limited to 5 retrospective series, most being on Behcet's disease, Takayasu's arteritis, and Polyarteritis nodosa from Turkey, UK and Canada, overall reporting 35 cases (Supplementary Table S3). No serious infections were reported over the longest study period of 2.1 years.

Autoinflammatory syndromes and Tuberculosis
Although there is emerging data from registries including the Eurofever registry on various auto-inflammatory syndromes, most focus on treatment regimens and response to therapy with dearth of data on infections. In the limited studies available (Supplementary Table 4), no Tuberculosis was reported. 21-23

Choice of biologics and risk of TB in children
Children with rheumatic disorders might be predisposed to Tuberculosis due to the intrinsic mechanism of action of biologics, anti-TNFs in particular, as they target TNF-α, the key cytokine for the Th1 axis. Experience from the biologic usage in adult rheumatic diseases has shown higher chances of TB reactivation with anti TNF agents. We identified 37 episodes of TB in 34 patients out of the 14,218 patients treated with anti TNF agents. In the non-TNF biologic group, a single case of TB has been reported with tocilizumab (OR-6.92 95% CI 0.95,50.56) ( Table 4). Anti-TNF therapy may not be a cause for TB reactivation among children with autoimmune diseases on biological agents. The role of TNF in controlling TB infection is reflected by the mice models deficient in TNF. These rodents are unable to control M. tuberculosis infection and form granulomas in their lungs. 24 TNF-a is required in the protective immune response against M.tuberculosis (MTB) in mice. 25 TNF is an important signal for macrophage activation, in conjunction with IFN-γ. This cytokine has a key role in the immune responses to MTB, because it is involved in multiple processes, such as macrophage activation and cell recruitment to the sites of infection (natural killer cells, granulocytes, fibroblasts, and T cells), which either leads to granuloma formation or kills the pathogen. Furthermore, it activates CD8+ T cell that could directly kill the bacteria, TNF-α additionally activates CD8+ cytotoxic T cells (CTLs) that may be important because these cells release granulysin and directly kill intracellular bacteria. TNF-α also promotes the maturation of monocytes to dendritic cells (DCs) and/or macrophages, inducing the antigen presentation of intracellular mycobacteria. TNF-α produced in a local infection site allows macrophages, natural killer (NK) cells and γδ T cells gather at the infection site and bring their activation. 26 The activated CTL cells have the ability to produce perforin protein and TNF-α by itself, which guide TB-infected monocytes to apoptosis, which involves intracellular living TB bacilli, and to induce the autophagy of infected cells via activated. 24 The other possibility is an increased risk due to the presence of an autoimmune disease. The risk of infections seems to be increased in rheumatic diseases not only from the drugs used, but also the presence of T lympho- have shown that lupus patients have suppressed reactive oxygen species and tumour necrosis factor-alpha activity in human monocytes in response to mycobacterium TB. 27 Previously, the risk of TB has been shown to be increased in children with JIA independent of the use of anti-TNFs. 28,29 However, in this study, the risk of TB was equal to the general population for children who either received anti-TNFs, or non TNF biological agents. This in sharp contrast to numerous other reports of TB in adults, suggesting that anti-TNFs might be safer in children than reported adults. Though this could also be attributed to smaller numbers in subgroup analysis, and remains to be confirmed. Presence of other infections can be risk factors for subsequent infections, though there is limited data from the current searches to substantiate that. One of the children who had CMV infection also had TB. In addition, primary immunodeficiencies such as X-linked agammaglobulinemia can mimic JIA and put the children at risk for infections. 30,31 Causes for low TB in children in current data set Low numbers due to studies in regions with low incidence of TB The number of studies from the various countries along with the reported number of TB cases, are plotted on a world map (Figure 2A,B). This pictorial view of the geographic distribution of the data obtained shows the stark distinction where most of the studies are concentrated in the affluent European and North American countries. Understandably, the reports of TB ( Figure 2B) available are also from these countries. It is evident that the countries with the highest burden of TB ( Figure 2C) have hardly any data on the biological use in children with RDs. Our literature search has brought out the inequalities in data availability across the world, and this has resulted in the probable assumption of low risk of TB among children with RDs on bDMARDs. Although the data review here suggested limited cases of TB on biologics, closer examination of the worldwide prevalence of TB makes paucity of data to be a possibility. The data from the PharmaChild registry had 17 episodes of TB in 14 children receiving biologicals for JIA. 32 All the cases were reported from children on TNF inhibitors. TB was most reported from Asian patients -52%, followed by 37% among the European patients, and 11% in the children treated at the centres in the USA. Since the registry covered 32 countries across the globe, the data seem to point at the fact that the low incidence of TB in other studies seen in Figure 2, is due to a concentration of studies from countries which are not endemic to TB. 33 Studies from areas with moderate TB burden like Turkey and Brazil did report tuberculosis (Figure 2).

TB IN PAEDIATRIC RHEUMATOLOGY
Low number of TB cases as consequence of the methodologies used to collect data Moreover, the low reporting of adverse events could be relevant to the kind of data collected. Many articles in paediatric rheumatology focus on response to therapy. Thus, data recording of infections takes a backseat. Two cases of tuberculosis were reported in a single study from Turkey, with the use of etanercept and adalimumab, which focused on collecting infection related data ( Table  4). The total numbers of infections reported were also remarkably higher in this study, suggesting possible geographic influence as well as methods/intent of data collection. Both developed TB despite a negative latent TB infection (LTBI) screen. Recently, a survey was conducted amongst physicians treating children with rheumatic disorders in India, that suggested a high incidence of TB, more so while the children were on biologics than after they were stopped. 34 Thus, it seems here that what we see in Turkey is just the tip of the iceberg, and the problem might be much severe in areas of TB endemicity. In the current era of biosimilars, data from post marketing surveillance records in the developed world can be mined to gain insight into TB incidence rates. 35 Varied screening strategies before administering biologics On a different note, low number of TB cases could also be due to varied TB and LTBI screening strategies before using bDMARDs. However, the recent survey from Indian rheumatologists suggests screening is universally practiced, though there is no consensus on the optimum method of screening. 34 Thus, a closer look into the prevalent practices and cost-benefit ratios of the strategy used for screening might be insightful in the future. Recently, Hassanzadeh et al. established that blanket screening for TB using the TB Spot assay increased the risk of polypharmacy, adverse drug effects and increased cost manifold. 36 Glasgow, an area of low TB prevalence and high BCG vaccination. Chest radiograph and clinical interview were used to identify risk factors for LTBI. The annual risk of TB was calculated using tables from BTS recommendations and then compared to the risk of drug-induced hepatitis. All patients were given a T-SPOT according to current local policy. Indeterminate T-SPOTs were recorded and repeated. Results . For 130 patients, a total of 160 tests were required resulting in a cost of £ 24,000. 99 (76% A recent systematic review confirmed the lack of consensus in screening strategies for TB in the immunosuppressed in guidelines across countries. 37 Thus, region-specific data needs to be gathered before implementing screening strategies in rheumatology as the risk and cost efficacy ratios might differ significantly according to TB incidence rates. 37

Shorter follow-up duration in children
Moreover, studies can be marred by short follow-up TITLE period, as post-marketing surveillance offers best insight into rare adverse effects. 35 Thus, registries are likely to provide a better overview. The PharmaChild registry which involved 32 countries across the globe reported 24 cases (17 on biological DMARDS) of tuberculosis in children with JIA. 38 Similar compilations are particularly needed from parts the world with high background prevalence of tuberculosis. The short window of childhood might limit study periods as children move on to adulthood, as compared with studies in adults, which are likely to have longer follow-up periods.
TB risk in children in comparison with adults TB screening practices could vary in children, as can be the threshold to prescribe biologics. Varied Tuberculosis incidence in different regions call for region specific guidelines in screening keeping the risk benefit ratio in mind. Lack of clarity in current guidelines is likely to accentuate the problem.

BCG vaccination
Difference in TB occurrence in children as compared with adults on anti-TNFs could also be a function of prevalent vaccination practices. Infant BCG vaccination has shown high efficacy of 70%-80% against childhood TB, especially meningeal and disseminated forms. 39 Sara Suliman et al have shown that BCG re-vaccination in adults with LTBI induces long-lived BCG-reactive NK cell responses. 40 This was in contrast to the limited cytokine change by Isoniazid preventive therapy, which was administered in 33 patients (39 in control group). Recently Katelaris et al. found that LTBI prevalence was lower amongst contacts of TB patients even 20 years after the initial vaccination, though vaccine efficacy declined as a function of time since vaccination. 41 In light of waning vaccine efficacy in adulthood, BCG re-vaccination could possibly reduce TB incident rates while on bDMARDs.

CONCLUSION
To conclude, there is dearth of data on incident TB rates in children with rheumatic disorders with exposure to bDMARDs from TB endemic countries. There is a felt need for regional registries to understand the prevalence, patterns, and prevalent screening practices to chalk out cost effective approaches with the intent to prevent long term debility.

AUTHOR CONTRIBUTION
All authors were involved in ideation and manuscript preparation.

Screening by title
The Scopus searches were imported into Zotero, and articles were first screened by title by one author, and those without relevance, systematic reviews, meta-analysis, narratives, and in languages other than English were removed (Figure 1).

Screening by abstract
The list of articles remaining after the initial screen was passed on to another author, to screen the individual abstracts for relevance, type of study and study population (children or adults). During the article screening, the initial rounds of elimination by screening titles and abstracts was done by one author each and subsequent rounds by two different authors.

Screening by reading full-text
The full text of articles obtained after two rounds of exclusions was then accessed. Those deemed irrelevant at this stage or where full-text was not available on the internet were then excluded. Similar screening strategy as delineated above was followed. The approach was mostly inclusive. Randomised placebo-controlled studies or any other controlled trials, retrospective case series, published data from registries, correspondence with data from more than 3 patients were included for data synthesis. Case reports, systematic reviews and metanalysis were excluded. Predesigned data extraction form (DEF), Table 1 was used to record data from the articles obtained after the above three stages of screening. The DEF was devised by two rheumatologists individually who then discussed and merged the variables suggested by each.
Differences were resolved by consensus between two rheumatologists. DOI numbers, year and author names were recorded to avoid duplication of studies. TB was defined as in the individual studies.
Further, the number of studies from the various countries was tabulated and the number of participants, as well as TB cases, reported recorded for each. These were plotted on a world map (Figures 1A,B) to get a pictorial view of the geographic distribution of the data obtained. Multicentre studies were excluded from the above figure as attributions to individual countries were not possible. Tables 1, 2 and 3 summarise data on tuberculosis in different paediatric rheumatic diseases, while Table 4 summarises data obtained from various registries. Furthermore, the quality of evidence of each study was recorded and summarised as a           Table 3. Data from studies on cohorts/registries of children with Juvenile arthritis on biologics. (continued) Table 4. Summary of available data that could be analysed for tuberculosis incidence in paediatric rheumatology with various biologics.