Keywords

Introduction

Tumor necrosis factor-α (TNF-α) plays a central role in the immunopathogenesis of a wide variety of inflammatory conditions from diseases such as rheumatoid arthritis (RA) to inflammatory bowel diseases (IBD). Development of TNF-α inhibitors (TNFI) has revolutionized the ability treat these conditions resulting in substantial improvement in outcomes [1,2,3]. Since the introduction of infliximab and etanercept in 1998, indications for the use of TNFI have expanded, and these medications are predominately prescribed by rheumatologists, dermatologists, and gastroenterologists for moderate to severe inflammatory and autoimmune diseases. Although these drugs have had a substantial impact in the treatment of many diseases, there are important safety concerns, the foremost of which is increased risk of infection caused by bacterial, mycobacterial, fungal, and viral pathogens [4]. Herein, we will review available data on the epidemiology of infectious complications in patients receiving TNFI for the treatment of inflammatory conditions.

Tumor Necrosis Factor-Alpha and the Innate Immune System

Tumor necrosis factor-α, primarily produced by macrophages and T-lymphocytes, is the principal endogenous regulator of inflammation and immune responses. First described in 1975 and named after its ability to cause tumor apoptosis in vitro, TNF-α is found constitutively in macrophages as a 233-amino acid transmembrane protein. Monomeric membrane-bound TNF-α aggregates into metabolically active homotrimers. When cleaved by the membrane-bound metalloprotease TNF-α converting enzyme (TACE), a soluble 157-amino acid TNF-α residue is released into circulation [5]. Only in the homotrimeric form is soluble TNF-α able to bind to its target receptors (Fig. 5.1). The activity of TNF-α is mediated by two types of receptors: tumor necrosis factor receptor 1 (TNFR1, also known as p55) and 2 (TNFR2, also known as p75). Although both receptors are structurally related, they are functionally distinct receptors mediating the activity of TNF-α in cells [6]. TNFR1 is found in a broad array of cells including macrophages, while TNFR2 is expressed predominantly in endothelial cells and lymphocytes [6]. Activation of TNFR1, which contains an intracellular death domain, results in induction of a signaling cascade with pleotropic effect that includes cell proliferation, apoptosis, and cytokine secretion [7]. TNFR2 does not contain a death domain and its stimulation can result in proliferation, migration, and production of cytokines such as interleukins -1 (IL-1) or -6 (IL-6), both important mediators of inflammation [7].

Fig. 5.1
Two images. In the image, a T N F alpha released from T A C E gets attached to T N F R 1, 2. In image b anti T N F alpha bond together into a triangular shape.

Overview of the TNF-α cascade in the presence of TNF-α inhibitors. Left half: Transmembrane TNF-α found on cell membranes of macrophages and other immune cells forms trimers and are released as biologically active homotrimeric soluble form via cleavage by TNF-α converting enzyme (TACE). Both soluble and transmembrane TNF-α homotrimers can bind to their ligand receptors (TNFR1 and TNFR2) found in a wide variety of cells throughout the body. The effect of which is a cascade of cell signaling that includes (1) cytokine and chemokine release; (2) maturation, proliferation, and migration of macrophages and other immune cells; (3) increased phagocytic activity of macrophages; and (4) formation and maintenance of granuloma. Right half: TNF-α inhibitors (TNFI) act by either binding transmembrane (A) and/or soluble (B) TNF-α. TNFI with IgG1 Fc region contains a CH1 domain that in the presence of complements can induce complement-dependent cytotoxicity (CDC) and antibody-dependent cell-mediated cytotoxicity (ADCC) leading to apoptosis of cells expressing transmembrane TNF-α (e.g. macrophages). (Illustration created by authors with BioRender.com)

Full size image

The activation of the innate immune response by an infectious pathogen includes release of TNF-α by activated macrophages into the affected tissue. The subsequent activation of TNFR1 and TNFR2 by binding with the homotrimer TNF-α results in a torrent of inflammatory events that includes release of inflammatory cytokines IL-1β, IL-6, IL-8, and granulocyte-macrophage colony stimulating factor (GM-CSF); upregulation of adhesion molecules, including intracellular adhesion molecule-1 (ICAM-1), vascular cell adhesion molecule-1 (VCAM-1), and E selectin (also known as endothelial leukocyte adhesion molecule-1, or ELAM-1); and increased expression of chemokines (e.g. RANTES, MCP-1, MIP-2) [8,9,10,11,12,13]. The combined effect results in vasodilatation at the infection site, coordinated recruitment, and migration of leukocytes to the target site, and activation of efficient phagocytosis of the pathogens resulting in successful host defense [13, 14].

TNF-α is essential for mounting effective host defense against pathogens that require granuloma formation for control [13]. These pathogens, which include Mycobacterium species including M. tuberculosis (TB), M. avium, and fungal pathogens such as Histoplasma capsulatum, Aspergillus fumigatus, and Cryptococcus neoformans, are not easily eradicated by host defense mechanisms and require sequestration into granulomas [13, 15,16,17,18,19]. TNF-α coordinates the organized formation of granulomas initially with chemokine production, phagosome activation, and leukocyte recruitment and differentiation, and subsequent leukocyte aggregation into function granulomas that can control infectious pathogens [13, 14].

Tumor Necrosis Factor-Alpha Inhibitors (TNFI)

Currently there are five approved anti-TNF-α agents available for various clinical indications (Table 5.1). All are indicated for the treatment of RA, ankylosing spondylitis, and psoriatic arthritis [2, 3, 20]. Except for etanercept, inflammatory bowel diseases (namely, Crohn’s disease and ulcerative colitis) can be effectively treated with TNFI [21, 22]. Other TNFI indications include treatment of inflammation of the skin (plaque psoriasis and hidradenitis suppurativa) and the eye (uveitis) [23,24,25].

Table 5.1 Summary of approved tumor necrosis factor-α inhibitors
Full size table

Etanercept and Infliximab were the earliest developed TNFI for clinical use and both were approved in 1998. Etanercept is a soluble fusion protein consisting of two human TNF-α receptor-2 (TNFR2) bound to the constant (Fc) region of a human IgG1 that acts as a decoy and binds both soluble forms of TNF-α and TNF-β, the latter is a related cytokine that utilizes the same receptors as TNF-α [26]. In contrast, infliximab is a chimeric monoclonal antibody consisting of a mouse anti-TNF-α variable (Fab′) region coupled with a human IgG Fc region. Adalimumab and Golimumab are both fully human IgG monoclonal anti-TNF-α antibodies with both humanized Fab′ and Fc regions [26]. As IgG1 monoclonal antibodies, infliximab, adalimumab, and golimumab can inhibit both soluble and membrane-bound forms of TNF-α but do not neutralize TNF-β [27]. As the Fc region of IgG1 contains a CH2 domain that is responsible for the activation of C1 (first component of the classical pathway of complement activation), both the full chain IgG monoclonal antibodies (infliximab, adalimumab, and golimumab) and etanercept in the presence of complements can induce both complement-dependent cytotoxicity (CDC) and antibody-dependent cell-mediated cytotoxicity (ADCC) with subsequent lysis of membrane-bound TNF-α expressing cells [26,27,28]. Lacking a CH1 domain that serves as the platform for C3 activation (the most vital step in the complement cascade), etanercept induces significantly less CDC on membrane-bound TNF-α-expressing cells [27, 28].

Certolizumab is a PEGylated Fab′ fragment of a humanized monoclonal anti-TNF-α antibody. The attachment of the Fab′ fragment to a 40-kDa polyethylene glycol moiety markedly increases the half-life of certolizumab compared to other TNFIs [29]. Like the full-chain anti-TNF-α monoclonal antibodies, certolizumab inhibits both soluble and membrane-bound TNF-α and lacks activity against TNF-β [29]. But in contrast to other inhibitors, certolizumab does not contain the crystallizable IgG Fc fragment and does not cause complement fixation, thus it does not induce CDC and ADCC in vitro [27,28,29].

Risk of Infection

Data evaluating infection risk are derived from a variety of sources, including clinical trials, meta-analyses, observational studies, and registries [30,31,32,33,34] (Table 5.2). In general, there is increased risk of infection with TNFI use, especially for tuberculosis, bacterial infections, and fungal infections. Several studies report a higher infection risk with infliximab when compared to adalimumab or etanercept [32, 35, 36]. However, there are several important limitations. Clinical trials may be limited by small sample sizes, inclusion of healthier patients, and insufficient statistical power to detect uncommon infection events. In observational studies, due to lack of randomization, confounding factors can impact results. Patients with autoimmune diseases enrolled in trials may be receiving corticosteroids or other medications that increase risk of infection, making attribution of infection risk to a particular TNFI challenging. Another important limitation in identifying TNFI infection risk in populations is related to underlying diseases that may impact infection risk. For example, patients with RA have an increased risk of infection compared with non-RA controls [37].

Table 5.2 Selected studies of infections associated with TNFI
Full size table

Mycobacterial Infections

Effective host immune response against Mycobacterium tuberculosis involves TNF-α-mediated formation of organized granulomas for control and prevention of dissemination. Studies including meta-analyses of randomized controlled trials, retrospective and prospective cohorts, and post-marketing registries have consistently shown increased risk for active tuberculosis (TB) in people on TNFI [30, 33, 38,39,40,41]. Patients with latent tuberculosis infection (LTBI) receiving TNFI therapy for RA, ankylosing spondylitis, or psoriatic arthritis have an estimated fourfold increased risk of TB reactivation as compared to controls [4, 30, 39, 41]. In a recent publication, Cao and colleagues reviewed 23 placebo-controlled clinical trials and similar increased odds of active tuberculosis were seen in patients with Crohn’s disease receiving TNFI [38]. All TB reactivation cases occurred in the TNFI arm and none in the placebo controls, with an odds ratio (OR) of 4.85 with a 95% confidence interval (CI) between 1.02 and 22.99 [38].

The risk of active TB may be different among TNFI [39, 40]. A systematic review in patients with RA treated with TNFI showed higher risk of tuberculosis with the use of adalimumab and infliximab compared to etanercept, with OR of 3.88 and 2.78, respectively [39]. The French RATIO registry reported a significantly higher odds ratio with adalimumab (OR 17.08) and infliximab (OR 13.29) when compared to etanercept [40]. Tuberculosis reactivation occurred five times higher during the first year of initiating TNFI therapy [40]. Latent TB infection screening and treatment for patients who will be receiving TNFI therapy can reduce risk of reactivation by 65% [39].

Nontuberculous mycobacteria (NTM) can cause a variety of human diseases particularly of the lungs in people with underlying lung conditions. Few data exist on the risk of NTM in patients on TNFI. The U.S. Food and Drug Administration (FDA) MedWatch database report in 2009 found 105 cases of NTM related to TNFI use. The majority were women (65%), had rheumatoid arthritis (70%), and most were receiving infliximab (70%) [42]. Half of the NTM infections were due to Mycobacterium avium, and though 56% were lung infections, extrapulmonary infections were not uncommon [42].

Mycobacterial infection rates in patients who used TNFI were evaluated using the Kaiser Permanente database [43]. TNFI-associated rates of NTM were 49 per 100,000 person years, greater than in unexposed RA patients (19.2 per 100,000 person years) or the general population (4.1 per 100,000 person years). NTM rates were lower for users of etanercept, when compared with infliximab or adalimumab [43].

Bacterial and Other Serious Infections

Many studies have reported data on TNFI use and serious (hospitalized) infections, which include a variety of organisms, but most frequently refer to bacterial infections [4, 30,31,32]. However, fewer details have been captured on specific bacteria causing an infection or infectious syndrome. Typically, pneumonia, skin, and soft tissue and urinary tract infections are the most common serious infections observed in adults, similar to the pre-biologic era [37]. In children, skin/soft tissue and respiratory infections are common [44].

In general, when comparing patients on TNFI to those receiving conventional disease modifying anti-rheumatic drugs (DMARDs), there is an increased risk of serious infection, with adjusted rate ratios ranging from 1.5 to 5.0, and infections per 100 person-years ranging from 2 to 15 [32, 33, 45,46,47]. It is important to note that timing of risk assessment is important, as studies focused on the first year of TNFI therapy show adjusted increased rate ratios, where a decline in absolute and relative risk of infection is typically seen after 1 year [32, 46,47,48].

A 2011 network meta-analysis of RCTs and extension studies found that TNFIs increased serious infection risk [30]. Certolizumab pegol was the only individual TNFI agent that significantly increased the risk of serious infection compared to control (OR 3.51; 95% CI 1.59–7.79). Another recent meta-analysis of 106 RCTs of targeted therapies (mostly TNFI) in RA patients demonstrated an increased risk of serious infections (OR 1.31, 95% CI 1.09–1.58) in patients who received standard dose TNFIs compared with traditional DMARDS. The risk was more pronounced (OR 1.9, 95% CI 1.50–2.39) in patients receiving high doses [31].

Studies in other populations have shown a variability in risk estimates. A meta-analysis evaluating patients with psoriatic arthritis reported a crude OR for infection of 1.18 (95% CI: 1.05–1.33) in patients exposed to TNFI (versus controls) [49]. In a meta-analysis among patients with ankylosing spondylitis, the risk of serious infection related to TNFI was low and was not significantly increased compared to untreated controls [50]. Two meta-analyses in patients with inflammatory bowel disease concluded that the risk of serious infection with TNFI was not increased [51, 52].

Many randomized controlled trials and observational studies fail to detail the precise nature of infectious syndromes or the causative agents. However, some series have reported either site-specific infections or data on specific pathogens. Risk for septic arthritis in RA with use of TNFI was evaluated in the British Society for Rheumatology Biologics Register. The adjusted hazard ratio for septic arthritis was 2.3 (95% CI: 1.2–4.4) for TNFI compared with traditional DMARDs. Staphylococcus aureus was the most common cause of septic arthritis [53].

Several studies have evaluated TNFI use and risk of listeriosis, one of which described a fourfold increased risk of severe listeriosis with TNFI in comparison with the general population [54,55,56]. There is a risk for legionellosis and TNFI, with one study finding the incidence rate of legionellosis in patients in TNFI to be 46.7 per 100,000 person-years and greater than the general population [57].

Viral Infections

Hepatitis B

TNF-α stimulates hepatitis B (HBV)-specific T-cell responses, inhibits HBV replication, and mediates HBV clearance in infected hepatocytes [58, 59]. Hepatitis B reactivation is the result of the loss of HBV immune control and is defined as an increase in HBV DNA level of either: (1) ≥2 log (100-fold) compared to baseline, (2) ≥2 log (1000) IU/mL in a previously undetectable level, or (3) ≥4 log (10,000) IU/mL if baseline not available [60]. HBV reactivation is a well-known complication in patients receiving TNFI [61,62,63,64,65]. In a retrospective Taiwanese study, HBV reactivation occurred in 5 (28%) of 18 hepatitis B surface antigen (HBsAg)-positive patients and 1 (25%) in 4 patients with occult HBV infection during the first year of TNFI therapy [62]. In addition, HBV reactivation occurs in previously inactive HBsAg carriers occurs following TNFI therapy [61]. A prospective Japanese cohort of 50 anti-HBc-positive RA patients on TNFI therapy followed up to 32 months, HBV reactivation was seen in 2 of 5 (40%) of HBsAg-positive patients and only 1 of 45 (2%) HBsAg [65]. In patients with previously resolved HBV infection, TNFI therapy was found to be safe, with no HBV seroconversion or reactivation observed [66]. Prophylactic antiviral therapy is effective in preventing reactivation [62, 67].

Guidelines consider use of TNFI as a moderate risk category with regards to HBV reactivation. Patients who are to start TNFI should at least have a baseline HBV serology that includes HBV surface antigen (HBsAg) and total HBV core antibody (anti-HBc) [67]. In HBV-endemic areas, HBV DNA should also be checked at baseline to detect occult HBV infections. In patients with either positive HBsAg or HBV DNA, preemptive anti-HBV antivirals with high barrier to resistance, such as tenofovir or entecavir, should be considered until 6–12 months after the last TNFI dose. Serial monitoring of HBV while on TNFI therapy every 6–12 months even for those with resolved HBV infections (anti-HBc positive but negative for HBsAg and HBV DNA) is recommended.

Hepatitis C

Evidence supports a TNF-α role in mediating inflammatory responses to hepatitis C (HCV) such as enabling apoptosis of infected cells, but it does not appear to play a pivotal role in the control of HCV replication [68]. In addition, TNF-α polymorphism has no significant effect to HCV susceptibility or viral clearance [68]. Data on the safety of TNFI use in patients with chronic HCV is limited and mostly derived from small cohorts and aggregates of case reports and case series. In a small cohort of 29 patients with both active RA and mild chronic HCV, use of etanercept was observed to be safe, with no increased risk of hepatic flare related to HCV replication [69]. A literature review found 216 patients with HCV who received TNFI (either etanercept, infliximab, or adalimumab) with mean observation time of 1.2 years and found only three patients needing TNFI withdrawal due to suspected HCV reactivation [70]. The limited data available supports that TNFI use in HCV patients is at least safe in the short-term [69, 70]. With the availability of safe and effective direct acting antivirals (DAAs) for HCV, treatment should be considered for patients planning to receive or receiving TNFI.

Herpes Zoster

Numerous studies have evaluated risk of herpes zoster with TNFI use, but evidence of risk of herpes zoster and TNFI therapy have been conflicting. A large U.S. multicenter cohort study [71] involving more than 33,000 patients with RA and other inflammatory diseases showed no increased risk of HZ when treated with TNFI. However, European registries [35, 72,73,74] and an Asian case-control study [75] showed an approximate twofold increase risk. Moreover, patients on TNFI had almost a ten times higher rate of hospitalization related to zoster when compared to the general population (32 vs. 3.4 cases per 100,000 patient-years) in a Spanish registry [74]. An international prospective registry study of patients with psoriasis showed that TNFI was not significantly associated with an increased risk of HZ, although the adjusted hazard ratio was 2.73 (95% CI 0.98–7.58) [76]. A British registry study found that zoster was highest among patients on infliximab (hazard risk [HR] of 2.2; 95% confidence interval [CI] 1.4–3.4) and lowest with adalimumab use (HR 1.5; 95% CI 1.1–2.0) [35].

Herpes zoster is a vaccine preventable disease. Shingrix, an adjuvanted recombinant zoster vaccine, significantly reduced risk of shingles by 94–97% compared to placebo in immunocompetent adults 50 years or older [77]. In a pooled post hoc analysis of participants with autoimmune diseases from two phase 3 trials showed overall vaccine efficacy of Shingrix at 90.5% (95% CI: 73.5–97.5%) [78]. This vaccine given in two doses 2–6 months apart is currently recommended for adults age 50 including those who are on low dose immunosuppression or anticipating being on immunosuppressive therapy [79, 80]. Although no head-to-head studies present, Shingrix is preferred over the live attenuated HZ vaccine, Zostavax, due to the latter lower efficacy rates especially in the older at-risk groups [77, 81, 82]. Zostavax is no longer available in the United States.

Fungal Infections

Tumor necrosis factor-α plays an important role in the control of infection due to fungi; however, fungal infections complicating TNFI use are relatively uncommon. Most reports detail impact on histoplasmosis, coccidioidomycosis, aspergillosis, and Pneumocystis jirovecii pneumonia (PCP) [34, 83,84,85,86,87,88]. The precise risk of TNFI use and fungal infection is difficult to acertain, as the concomitant use of other immunosuppressive therapies, especially corticosteroids, renders risk interpretation problematic. A recent meta-analysis reported the risks of opportunistic infections in RA patients from clinical trial data of biologic use, mostly TNFI [34]. Biologic use did not significantly increase the risk for all fungal (superficial or invasive) infections (odds ratio 1.31, 95% CI 0.46–3.72), invasive fungal infections (odds ratio 2.58; 95% confidence interval 0.68–11.91), or PCP (odds ratio 1.77, 95% confidence interval 0.42–7.47). A large US cohort study evaluated new users of TNFI and investigated the incidence of nonviral opportunistic infections among patients with RA, ankylosing spondylitis, psoriatic arthritis, psoriasis, and IBD [89]. Among 33,324 new users of TNFI, 80 nonviral OIs were identified. Of these, 32 (40%) were caused by fungi, with a crude incidence rate of 112 cases per 100,000 person-years. The most common fungal infections were pneumocystosis (16 cases) and histoplasmosis (9 cases).

The estimated incidence of aspergillosis is approximately seven cases per 100,000 persons treated with TNFI [85, 88]. A case series was published by Tsiodras and colleagues, who reviewed publications up to June 1, 2007 to determine the association of fungal infections with TNF-α blockade [90]. Sixty-four cases of aspergillosis, mostly invasive pulmonary disease, were identified. The most common TNFI used was infliximab in 48 cases (75%), followed by etanercept in 14 (22%), and adalimumab in three cases (3%) [90].

The incidence of coccidioidomycosis in patients receiving TNFI is estimated to range up to 5.58 per 100,000 persons treated with infliximab and 0.88 per 100,000 persons treated with etanercept [91]. Bergstrom and colleagues described 13 cases among patients receiving TNFI from in areas endemic for coccidioidomycosis [92]. The interval between TNFI and infection ranged from 1 to 96 weeks (mean, 27 weeks), and two cases were likely due to reactivation. All patients had pneumonia on presentation, with 4 (30.7%) having disseminated disease. The risk of infliximab in development of symptomatic coccidioidomycosis when compared to other agents was greater (RR 5.23, 95% CI 1.54–17.71; p < 0.01).

Taroumian and colleagues described 44 patients with rheumatologic disease treated with nonbiologic DMARDS and/or biologic therapies in Tucson, Arizona [93]. Twenty-nine patients had pulmonary coccidioidomycosis, nine patients had disseminated disease, and six had asymptomatic coccidioidomycosis based on positive serology. With continuation or resuming biologic therapy after treatment, no patients had subsequent dissemination or complications of coccidioidomycosis.

Histoplasmosis is one of the most common fungal infections in patients receiving TNFI [84, 85]. Wallis and colleagues collected data from cases reported to FDA Adverse Event Reporting System (AERS) from January 1998 through September 2002 and identified 40 cases of histoplasmosis. The estimated rate of histoplasmosis per 100,000 patients treated was 18.78 in patients treated with infliximab and 2.65 in patients treated with etanercept [85]. Vergidis and colleagues described 98 patients diagnosed with histoplasmosis while receiving TNF-α inhibitors from January 2000 to June 2011. Seventy-four (76%) patients presented with disseminated histoplasmosis; pulmonary involvement was present in 78 (80%) patients. The median time to diagnosis after TNFI initiation was 15.5 months (range of 1–88 months) [84]. Rheumatoid arthritis was the most common underlying disease, and infliximab (67.3%) was most used. TNFI therapy was initially discontinued in 96.9% of patients but resumed in 33% of patients at a median of 12 months. The recurrence rate at follow-up was 3.2%.

Olson and colleagues found that 15 of 26 patients with RA who developed disseminated histoplasmosis from 1998 to 2009 were on TNFI and had a median time on TNFI to histoplasmosis diagnosis of 15 months (range, 2–132 months) [94]. Most patients were treated with at least 6 months of antifungal therapy. In this study, TNFI were discontinued at the time of infection in 14 patients and was restarted successfully in 4/15 with recurrence of disease in only one patient [94].

Pneumocystis Pneumonia

Pneumocystis jirovecii pneumonia (PCP) complicating patients receiving TNFI is uncommon, with variability in incidence rates depending on the population studied and diagnostic method used. Observational studies have reported incidence rates of up to 8.8 cases per 1000 patient-years [95,96,97]. Takeuchi and colleagues examined the incidence of adverse events in Japanese patients with RA for their first 6 months on infliximab as post-marketing surveillance [95]. The diagnosis of suspected PCP was made in 22 (0.4%) patients, with many cases diagnosed by PCR for P jirovecii DNA from bronchoalveolar lavage fluid.

The National Institutes of Health conducted a population-based study to determine if the incidence of PCP in patients with RA had changed significantly from 1996 to 2007 using data from the Nationwide Inpatient Sample and the California Office of Statewide Health Planning and Development [97]. They found no significant change in the number of patients with RA and PCP diagnoses over this period.

Conclusion

Tumor necrosis factor-α inhibitors have become an important class of drugs and will continue to be used widely in the treatment of autoimmune and inflammatory diseases. Although uncommon, increased risk of infection caused by bacterial, mycobacterial, fungal, and viral pathogens have the potential for increased morbidity and mortality. Risk of infection is often difficult to characterize, as it may differ with underlying patient comorbidities, concomitant medications, and the specific TNFI agent. Use of TNFI will warrant clinician vigilance and continued infection surveillance.

Potential Conflicts of Interest

JVC: Research grants from Gilead Sciences, Inc.

JWB: Consultation for Eli Lilly and Viela Bio.