Patients with rheumatoid arthritis are at higher risk for serious infections and death from infection than the general public. Prednisone and biologic agents increase this risk, although the risk associated with biologics can be mitigated when such agents act as prednisone-sparing therapies. Some of the important causes of infectious morbidity in this setting are preventable with screening (eg, tuberculosis) or vaccination (eg, herpes zoster).
Key Points
- •
Patients with rheumatoid arthritis are at higher risk for serious infections and death from infection than the general public.
- •
Prednisone and biologic agents increase this risk, although the risk associated with biologics can be mitigated when such agents act as prednisone-sparing therapies.
- •
Some of the important causes of infectious morbidity in this setting are preventable with screening (eg, tuberculosis) or vaccination (eg, herpes zoster).
Increased infectious risk in rheumatoid arthritis
Patients with rheumatoid arthritis (RA) have long been recognized to suffer a greater burden of serious infection. The precise immune derangements of RA that predispose to infection are not clearly known. Patients with RA seem to have reduced capacity to generate new T lymphocytes, and their T-lymphocyte repertoire becomes severely contracted over time, a phenomenon perhaps akin to the immunosenescence observed in a normal aging host. However, it is likely that a variety of RA-associated host factors predispose patients toward infection, including the physical derangements (eg, destruction of articular surfaces, airway inflammation) that might also impair local immunity or provide a respite for circulating pathogens. In the prebiologic era, Doran and colleagues reported this heightened risk within an RA cohort study in Minnesota’s Mayo Clinic patient population ( Table 1 ). Compared with matched non-RA controls, the investigators documented serious infections to occur nearly twice as frequently in patients with RA, at a rate of 9/100 patients per year. This increased risk remained even after controlling for the effects of important comorbidities and other infectious risks, and similar to historical reports, these investigators documented pulmonary and skin/soft tissue infections to be the most common sites of RA-related infectious morbidity.
| Infection Type | Incidence per 100 Patient-Years | Incidence per 100 Patient-Years | Relative Risk (95% Confidence Interval) |
|---|---|---|---|
| RA | Non-RA | ||
| Pneumonia | 4.0 | 2.4 | 1.7 (1.5–1.9) |
| Skin | 3.0 | 0.9 | 3.3 (2.7–4.1) |
| Sepsis | 0.78 | 0.51 | 1.5 (1.1–2.1) |
| Septic joint | 0.40 | 0.02 | 14.9 (6.1–73.7) |
| Intra-abdominal | 0.22 | 0.08 | 2.8 (1.4–6.2) |
| Osteomyelitis | 0.17 | 0.01 | 10.6 (3.4–126.8) |
Increased infectious risk in rheumatoid arthritis
Patients with rheumatoid arthritis (RA) have long been recognized to suffer a greater burden of serious infection. The precise immune derangements of RA that predispose to infection are not clearly known. Patients with RA seem to have reduced capacity to generate new T lymphocytes, and their T-lymphocyte repertoire becomes severely contracted over time, a phenomenon perhaps akin to the immunosenescence observed in a normal aging host. However, it is likely that a variety of RA-associated host factors predispose patients toward infection, including the physical derangements (eg, destruction of articular surfaces, airway inflammation) that might also impair local immunity or provide a respite for circulating pathogens. In the prebiologic era, Doran and colleagues reported this heightened risk within an RA cohort study in Minnesota’s Mayo Clinic patient population ( Table 1 ). Compared with matched non-RA controls, the investigators documented serious infections to occur nearly twice as frequently in patients with RA, at a rate of 9/100 patients per year. This increased risk remained even after controlling for the effects of important comorbidities and other infectious risks, and similar to historical reports, these investigators documented pulmonary and skin/soft tissue infections to be the most common sites of RA-related infectious morbidity.
| Infection Type | Incidence per 100 Patient-Years | Incidence per 100 Patient-Years | Relative Risk (95% Confidence Interval) |
|---|---|---|---|
| RA | Non-RA | ||
| Pneumonia | 4.0 | 2.4 | 1.7 (1.5–1.9) |
| Skin | 3.0 | 0.9 | 3.3 (2.7–4.1) |
| Sepsis | 0.78 | 0.51 | 1.5 (1.1–2.1) |
| Septic joint | 0.40 | 0.02 | 14.9 (6.1–73.7) |
| Intra-abdominal | 0.22 | 0.08 | 2.8 (1.4–6.2) |
| Osteomyelitis | 0.17 | 0.01 | 10.6 (3.4–126.8) |
Prednisone
Although conflicting data (to be discussed later) foster debate regarding the infectious risks of biologic therapies, there is little debate regarding the ability of prednisone to cause infectious harm. Although a recent meta-analysis of glucocorticoid trials in rheumatic disease found no increased risk of serious infection associated with steroids, observational studies consistently report such an association. The prebiologic era observational study of Doran and colleagues documented a significant 1.5-fold to 2-fold increase in risk with corticosteroid therapy, with risk noted even at a low dose (eg, <15 mg/d) of prednisone. In 2006, Wolfe and colleagues produced similar findings with increased risks (hazard ratio 1.4, 95% confidence interval [CI] 1.1–1.6) noted even at doses of prednisone less than 5 mg/d. More recent registry data further attest to the association of prednisone with serious infection. In the United States, a national collaboration of observational databases (the Safety Assessment in Biologic Therapy [SABER]) compared the risk of serious bacterial infection in patients starting biologics with methotrexate-treated patients who start an additional nonbiologic disease-modifying antirheumatic drug (DMARD). These researchers also documented dose-dependent increases in risk, with relative risk (RR) up to 3-fold higher in patients using doses greater than 10 mg/d ( Table 2 ). The Consortium of Rheumatology Researchers of North America (CORRONA) registry identified 1.5-fold higher risks with prednisone use for opportunistic infections, and other US studies support the idea that risk is increased even at low daily doses of 5 mg or less (1.3-fold–1.5-fold), escalating with increasing doses to 5-fold higher risk at doses of 20 mg or grerater. Within Europe, the results have been no different. The British Society for Rheumatology Biologic Register (BSRBR), a UK cohort of patients with inflammatory arthritis, observed corticosteroids to double the risk of serious infection. In Germany’s biologic registry for patients with RA (RABBIT), the independent increase in relative serious infection risk observed with corticosteroids varied between 2-fold and nearly 5-fold at dosages less than 15 mg and 15 mg or greater daily, respectively.
| Exposures | Events, Number | Person-Years, Number | Rate, per 100 Person-Years | Hazard Ratio (95% CI) for Propensity Score-Matched Cohorts | Adjusted Hazard Ratios (95% CI) |
|---|---|---|---|---|---|
| RA Nonbiologic regimens | 326 | 4192 | 7.78 | Ref. | Ref. |
| Tumor necrosis factor α antagonists | 497 | 6089 | 8.16 | 1.05 (0.91–1.21) | 1.05 (0.91–1.21) |
| Baseline glucocorticoid use, prednisone equivalents | |||||
| None | Ref. | ||||
| >0–<5 mg/d | 1.32 (1.10–1.58) | ||||
| 5–10 mg/d | 1.78 (1.47–2.15) | ||||
| >10 mg/d | 2.95 (2.41–3.61) |
Beyond bacterial infections, corticosteroids increase the risk of tuberculosis (TB), nontuberculous mycobacterial disease (NTM), and other opportunistic infections. This risk has been particularly well documented with herpes zoster, in which the risk is increased 1.5 to 2 times in patients with RA who use corticosteroids. Pulmonary NTM disease, of which patients with RA are at higher risk, has been linked to systemic prednisone and even inhaled corticosteroid use. Disease caused by Pneumocystis jiroveci occurs in the setting of high-dose prednisone use, although the threshold of dose and duration of prednisone use for causing Pneumocystis has not been well described.
Collectively, these data suggest that systemic corticosteroid use is an important and potentially modifiable risk factor for serious infection. Further, the serious infection risks observed within these studies are similar and sometimes exceed those observed with biologic therapy. Despite these infectious risks and the advent of newer biologic therapies, corticosteroid use in RA remains common. Patients who use both biologic therapy and prednisone concurrently have an even higher risk of serious infection than patients who use only biologic therapy ( Fig. 1 ). Clearly, one of the potential benefits of DMARD therapy, either synthetic or biologic, is the opportunity to mitigate risks associated with corticosteroids by allowing for the reduction or elimination of prednisone.
Biologic Therapies
In little more than the last decade, biologic therapies targeting tumor necrosis factor α (TNF-α), T and B lymphocytes, and various interleukins (IL) (including IL-6 and IL-1) have been developed and approved for use in RA and selected other rheumatic diseases. Additional drugs with novel targets are either approved or in trials for RA and other inflammatory conditions, including those that inhibit IL-12 to IL-23 and IL-17a. Novel small molecules (technically not biologics) that target the janus activating kinases (JAK) and splenic tyrosine kinases are also in development, and at least one JAK inhibitor (tofacitinib) is under consideration by the US Food and Drug Administration (FDA) for approval.
Infection risk
Data from randomized controlled trials and long-term extension studies of biologic therapies have generally (but not always) suggested some increase in infectious risks associated with these compounds. For anti-TNF therapies, where abundant registry and administrative health care-based observational studies have been conducted, higher absolute infection rates among anti-TNF users have generally been observed, particularly in the 6 to 12 months after drug start. These rates observed in such real world settings have typically been higher than those observed during clinical trials and long-term extension studies. However, the assessment of infection risk with anti-TNF therapies in population-based studies is complex, and only in the last 1 to 2 years perhaps, a clearer picture has emerged. For the more recently approved biologics in RA with different mechanisms of action (ie, abatacept, tocilizumab, rituximab), rates of serious infection observed in clinical trials have largely been similar to those observed in anti-TNF trials. However, for these newer compounds, real world infection rates and risk estimates are largely missing, because such observational registry or database studies with these therapies have yet to be conducted.
Anti-TNF therapy
Five TNF blockers currently fill the marketplace: etanercept (Enbrel), infliximab (Remicade), adalimumab (Humira), golimumab (Simponi), and certolizumab (Cimzia). As a group, these drugs inhibit TNF-α, a proinflammatory cytokine expressed by activated macrophages, T lymphocytes, and other immune cells, known to play a crucial role in the host response against a variety of infections. TNF directly activates macrophages to phagocytose and kill mycobacteria and a variety of other pathogens, and the failure to control TB during TNF blockade seems intrinsically linked to an inability to control intracellular TB growth in macrophages residing within granulomas, rather than inhibiting the development granulomas per se. Murine studies suggest the importance of TNF in protection against other intracellular organisms such as Listeria and fungi, as well as against extracellular bacterial organisms like Klebsiella pneumoniae and Streptococcus pneumoniae.
Risk of infection with anti-TNF therapy
The overall risk of serious infections attributable to anti-TNF therapy is not precisely known, and our understanding of this issue depends largely on the study settings and methodology used to assess these risks. Further, much of the variability in an individual’s infection risk is not attributable to drug therapy, but rather to patient factors (eg, age, chronic lung disease, and other comorbidities) that are frequently not modifiable. Although conflicting evidence from a variety of study designs exists, it is our opinion that anti-TNF therapy independently increases the risk of serious infection, but that this risk is mitigated by a variety of measurable (and unmeasurable) factors after drug start. As suggested earlier, TNF blockers can, and should, serve as prednisone-sparing therapies, allowing for a reduction in prednisone-associated infection risk. Other physician-patient decisions such as TB screening, vaccination, counseling with regard to infection prevention, and other factors likely also serve to mitigate infectious risk associated with starting anti-TNF therapy. It is likely that patients starting anti-TNF therapies are treated differently with regard to these and other factors compared with patients starting nonbiologic DMARDs. This confounding by indication can be difficult to measure and control for in “real world” observational studies that seek to measure differences in risk between anti-TNF and nonbiologic DMARD users.
Our understanding of infection risk derives from individual randomized controlled trials and their meta-analyses, long-term open-label extension studies, and the observational (registry and database) studies, as discussed earlier. Meta-analyses of TNF-blocker trials generally observed RRs of serious infection to be slightly higher in those patients treated with anti-TNF therapy, but not always so, and frequently the observed effect has been mild and not statistically significant. Bongartz and colleagues analyzed monoclonal antibody trials (infliximab and adalimumab only) in RA and observed a 2-fold increase in serious infection risk. Leombruno and colleagues evaluated serious infection risk within RA trials of etanercept, infliximab, and adalimumab and found no statistically increased risk (RR 1.08 [0.81–1.43]) in those treated with anti-TNF therapy versus placebo overall, but did find a significant 2-fold increase in risk for those studies using higher-dose infliximab or adalimumab. More recently, Singh and colleagues preformed a Cochrane review of anti-TNF therapies across disease indications and found infliximab (odds ratio [OR] 1.45 [0.99–2.1]) and certolizumab (OR 3.5 [1.6–7.8]) to have the highest RR of serious infection compared with placebo, whereas no other TNF blockers were shown to have significant increased risk. There are great difficulties in understanding the infectious profiles relying on data from randomized controlled trials (eg, statistical power is low for individual trials, careful selection of patients), and even greater difficulties in comparing drugs across trials (eg, differences in inclusion criteria, populations, disease indications, others), such that it is unclear from such meta-analytical studies if drugs like infliximab or certolizumab carry any greater risk than other TNF blockers.
Lastly, it is worth noting that trials involving patients with ages and conditions associated with lower baseline infectious risk, such as psoriasis, have observed lower overall rates and relative risks of serious infection in patients using these therapies.
Several large observational studies and registries have also assessed the risk of infection with these compounds. Although these studies have heterogeneity in methods and cannot fully overcome issues of confounding (eg, confounding by indication), these studies have reported similar estimates of serious infection in patients using these therapies. The incidence of serious infection for patients with RA using TNF blockers generally hovers between 3 and 8/100 patient-years ( Table 3 ), and both absolute and relative rates seem dependent on when they are measured after drug start. The BSRBR found a 4-fold increase in risk for serious infection in the first 90 days after anti-TNF start among patients with RA, and Curtis and colleagues reported a similar increased risk in the first 6 months after anti-TNF start. Both studies produced smaller RR estimates when longer periods of exposure were assessed. This finding is likely because of a survivor effect, but also potentially secondary to benefits of improved disease management, and changes in concomitant therapies, which occur after anti-TNF therapy start.
| Country, Year | Crude Incidence per 100 Patient-Years Anti-TNF Treated | Crude Incidence per 100 Patient-Years Nonbiologic Comparator | Adjusted RR a (95% CI) |
|---|---|---|---|
| Germany, 2005 | 6.4, ETN 6.2, INF | 2.3 | 2.2 (0.9–5.4) ETN 2.1 (0.8–5.5) INF |
| United Kingdom, 2007 | 5.5 | 3.9 | 1.3 (0.9–1.8) c 4.6 (1.8–11.9) b |
| United States, 2007 | 2.9 d | 1.4 d | 4.2 (2.0–8.8) d 1.9 (1.3–2.8) |
| Sweden, 2007 | 4.7 | NR | 1.4 (1.2–1.7) e |
| United States, 2007 | 4.9 | 3.8 | 1.3 (0.8–2.1) |
| United States, 2011 | 8.2 | 7.8 | 1.05 (0.9–1.2) |
| Germany, 2011 | 4.8 | 2.3 | 1.8 (1.2–2.7) f |
| Japan, 2011 | 6.4 | 2.6 | 2.4 (1.1–5.05) g |
a Relative rate using nonbiologic users as the referent.
b When restricted to the first 90 days of therapy, and adjusted for age, sex, disease duration, and severity, extra-articular RA, baseline steroid use, diabetes, chronic obstructive pulmonary disease, pulmonary disease, and smoking history.
c Adjusted relative rate when not restricted to the first 90 days of therapy.
d Analysis restricted to the first 6 months after initiation of anti-TNF therapy.
e Rate calculated at 1 year after starting treatment and adjusted for RA severity and comorbidities associated with infections.
f Adjustment for time-varying risk factors, treatment adaptations, and dropout.
Grijalva and colleagues recently published results from the SABER collaboration, a population-based study that compared only new users in both groups: anti-TNF or nonbiologic DMARDs. Within RA, this comparison was restricted to those patients failing methotrexate who subsequently started either an anti-TNF drug (for the first time) or a new nonbiologic DMARD. Although the influence of baseline corticosteroid use, comorbidities, and disease severity was controlled for in modeling, changes taking place after drug start (eg, changes in prednisone, concomitant nonbiologic DMARD use, improved disease control) could not be assessed. Although investigators reported high rates of serious infection (8/100 patient-years for RA), no increased risk was associated with anti-TNF therapy start (see Table 2 ) and contrary to the previous observational studies discussed earlier, no short-term increased risk was noted in the first 3 to 6 months after drug start. Within their RA cohort, these investigators found that infliximab starters were 20% to 25% more likely to suffer serious infection than those who started with either etanercept or adalimumab. However, patients starting infliximab were more likely to be on methotrexate after the index date than those starting etanercept or adalimumab, perhaps contributing to this increase in risk. Other population-based studies have failed to find a similar increase in RR for bacterial infections with infliximab compared with the other TNF blockers. The study reiterated the importance of patient factors to overall infection risk. For example, patients treated with anti-TNF with a history of chronic obstructive pulmonary disease (COPD) had rates of nearly 17 versus 7 per 100 patient-years for those without COPD.
Collectively, the observational studies conducted to date suggest that the overall or net infectious risk of biologic therapy is not straightforward to understand or calculate, and that one must account for time-varying risk factors during such study. Strangfeld and colleagues examined this issue using the RABBIT registry and were able to control for several factors that either increase or decrease the risk of infection after drug start. Their data suggest that anti-TNF therapy start improves disease control and decreases prednisone use, both lowering infectious risk, but that even when controlling for these changes, the start of anti-TNF therapy significantly increases the risk of serious infections 1.8-fold. This study arguably represents our most complete understanding of the dynamic risk profile presented by these therapies. We know less regarding serious infectious risks of the 2 newer anti-TNF therapies certolizumab and golimumab, because large population-based studies of these compounds are lacking.
Opportunistic infections with anti-TNF therapy
Several intracellular and other opportunistic pathogens have been reported in the setting of anti-TNF therapy ( Tables 4 and 5 ). These pathogens include severe and sometimes lethal infections with Histoplasma , Coccidioides, Listeria, Salmonella, Aspergillus, Nocardia , and nontuberculous mycobacteria. Of these infections, TB is most clearly associated with TNF blockade, and a clear distinction in risk can be made between the fusion receptor construct etanercept, and that of the monoclonal antibodies infliximab and adalimumab. Although TB risk has been documented to be clearly increased with all 3 drugs, it is 3-fold to 4-fold more common in adalimumab-treated and infliximab-treated patients relative to etanercept. Several animal and in vitro studies support the biological plausibility of these observations. Potential explanations range from differential granuloma penetration to differential downregulation of antigen-stimulated interferon-γ to differential affects on antimicrobial-producing CD8 effector cells. Cases of TB have been reported from clinical trial experience of certolizumab and golimumab, but few population data exist by which to compare their RR with the other TNF blockers.
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
