Reporting Bias: Infections with TNF Inhibitors

A recent article (Clin Infect Dis 2013; 57: 1318-30 -thanks to Jeff Lewis for this reference) summarized the pediatric literature on infectious complications associated with tumor necrosis factor-α (TNF) inhibitors for both Juvenile Idiopathic Arthritis (JIA) and Inflammatory Bowel Disease (IBD).

In total 33 studies for JIA were included and 39 studies for IBD.  Many others were excluded due to overlapping cohorts or lack of sufficient data.

For JIA, the authors identified 296 infliximab (IFX) patients, 2465 etanercept patients, and 242 adalimumab patients.  Most infectious were mild and mainly viral etiology.  For example, 1016 upper respiratory illnesses with etanercept were reported.  However there were a significant number of more serious infections which included lower respiratory infections (n=37), cellulitis/abscess (n=15), histoplasmosis (n=2), and meningitis (n=4). Four patients had infectious fatalities.

Similarly, for IBD, most patients had mild infections.  Among 1407 IFX patients and 241 adalimumab (ADA) patients, there were 105 URIs noted.  Again, more serious infectious were noted in many.  Four fatalities were reported; 1 was due to disseminated CMV, 1 due to bacterial sepsis, and 2 were due to central line infections.

When examining this report, the question of reporting bias cannot be avoided. The various reports that were summarized included 30 prospective studies, 23 retrospective studies, and 19 various reports (case reports, case series, and FDA reports).  The composite, in my view, likely overestimates the risk of serious infections.  In addition, many of the infections may have been due to concurrent immunosuppressive therapy, but the details for this are lacking.

With regard to microbiology:

  • 5 JIA patients had tuberculosis; there were no tuberculosis cases reported in the IBD cohort
  • Varicella/zoster was the most frequent viral infection and was frequently severe.  In JIA cohort, there were 39 VZV cases (11 severe); among IBD cohort, there were 16 VZV cases (3 severe).

Bottomline: Given the frequent use of anti-TNF agents, better prospective pediatric data are needed.  In addition, careful analysis of the data is needed for better attribution; the risk for many of these infections is likely due to concurrent medications like corticosteroids.

Also Noted:

Clin Gastroenterol Hepatol 2013; 11: 826-31.  In a prospective cohort of 200 anti-TNF-naive adult patients (100 treated with IFX and 100 with ADA), the effectiveness was similar for IFX and ADA at both 1 and 2 year followup.  Improved efficacy was noted when these agents were combined with immunomodulators, though this was statistically significant for IFX. The total patient response was 63.5% at 1 year and 45% at 2 years.

Related posts:

Liver Injury from Anti-TNF Agents

While anti-TNF agents have been associated with drug-induced liver injury (DILI), it has been difficult to get a handle on how much importance to place on this.  A recent study provides more data and some reassurance (Clin Gastroenterol Hepatol 2013; 11: 558-64).

The authors searched the U.S. DILI Network database from 2003 to 2011 and describe 6 cases of anti-TNF DILI; in addition, they searched PubMed for articles related to anti-TNF agent associated hepatotoxicity and identified an additional 28 cases. Other causes of liver disease were excluded in these patients, including reactivation of hepatitis B, and acute viral hepatitis (eg. hepatitis C, hepatitis E).

Results of anti-TNF hepatotoxicity:

  • 26 cases due to infliximab, 4 cases due to etanercept, and 4 due to adalimumab.
  • Based on scoring system, the anti-TNF agent was considered a definite cause of DILI in 1 (3%), very likely in 21 (62%) and probable in 12 (35%).
  • Median latency (duration of therapy before onset of DILI) was 13 weeks with a range of 2-104 weeks.
  • 22 (67%) had positive anti-nuclear and/or smooth muscle antibodies.  15 of 17 of these patients had liver biopsy features consistent with autoimmunity.
  • Among those 22 with autoimmune features, there was a higher peak alanine aminotransferase compared with the 12 without these features (784 vs 528 U/L)
  • Favorable outcome: all but one patient improved after discontinuation of the implicated drug; 12 received corticosteroids. One patient with underlying cirrhosis underwent liver transplantation after infliximab-induced liver injury.

While the authors note the potential for a class effect of anti-TNF agents, in studies from patients with psoriasis, there was a lack of cross-toxicity between etanercept and infliximab.

Take-home messages:

The risk of hepatocellular injury from anti-TNF agents is very low.  DILI due to anti-TNFs often have autoimmune features. The prognosis is favorable, and alternative anti-TNFs can be given after resolution.

Related blog links:

TNF-α antagonists and infections

In our pediatric patients who receive tumor necrosis factor-α (TNF-α) antagonists, fortunately we see few infectious complications.  In older patients, infections are much more important source of morbidity.  The main TNF-α inhibitors in clinical use include infliximab, etanercept, adalimumab, and certolizumab. Two large studies help quantify this risk:

  • Grijalva CG et al. JAMA 2011; 306: 2331-39.
  • Strangeld A et al. Ann Rheum Dis 2011; 70: 1914-20.

The first study assembled retrospective cohorts between 1998-2007 with rheumatoid arthritis (RA), inflammatory bowel disease (IBD), and psoriasis/psoriatic arthritis or ankylosing spondylitis (group 3).  This study’s acronym is SABER: Safety Assessment of Biologic Therapy. This data was compiled from 4 large US automated databases.  In total, there were 10,484 RA, 2323 IBD, and 3215 group 3 patients.  1172 serious infections were identified, mostly (53%) pneumonia or skin/soft tissue infections.  Among IBD patients, hospitalization rates were 10.91 (per 100 person-years) for TNF-α antagonists and 9.6 for comparison group.  The rates of hospitalization were similar in RA (8.16 with TNF-α antagonists) and lower in the group 3 patients (5.41 with TNF-α antagonists).   In all groups, baseline glucocorticoid use was associated with a dose-dependent increase in infections.  Overall, there was not an increase risk of hospitalizations with TNF-α antagonists compared with nonbiologic treatments.

The second cited reference examined patients from Germany with RA, enrolled in the RABBIT registry.  Data was available for 5044 patients.  There were 392 serious infections in this cohort with fewer infections noted after 3 years.  Risks for infection included age (>60), chronic lung or renal disease, history of serious infections, and treatment with glucocorticoids.  Treatment with 7.5-14mg conferred at relative risk (RR) of 2.1; treatment with ≥15mg conferred a RR of 4.7.  The rates of serious infections has an exponential change when risk factors are added together.  In Figure 3, estimated risk of serious infections for patients receiving ≥15mg  glucocorticoids along with three additional risk factors was 45% per year; with two risk factors the risk was approximately 20% and with one additional risk factor approximately 10%.

While these studies confirm significant risks of infections with biologic agents, the absolute risk is low particularly when other risk factors are not present.  In pediatric populations, glucocorticoids are the most prominent risk factor.  In addition, the risk of serious infections may be reduced by effective therapy.  In the SONIC study, serious infectious complications were less frequent in patients on combination therapy (infliximab and azathioprine) than with either monotherapy.  This result was likely due to the decreased need for glucocorticoids.

Additional references/relevant previous blogs:

  • NEJM 2010; 362: 1383. Sonic study. Combination AZA/IFX with greater efficacy. 56.8% remission in combo Rx.
  • -IBD 2008; 14: 721.  Pneumocystis jiroveci (carinii) w infliximab -review of 84 cases.  Dig Dis Sci 2007; 52: 1481-84.  PCP most likely to occur on average 3 weeks after 2nd infusion (possibly due to concommitant drug use)
  • -Gastroenterology 2008; 134: 929.  n=100 consecutive IBD patients with opportunistic infections.  Any of drugs associated w ~2.9 OR in adutls (greatest in >50yrs).  OR 14.5 when multiple immune drugs. Steroids more associated with Candida. AZA/6MP more with viral: HSV, VZV (shingles), CMV.  IFX less commonly with infections -though increased histoplasma and atypical mycobacterium
  • -Gastroenterology 2009; 136: 1182.  Review of biologics.
  • -IBD 2007; 13: 769.  Review of safety of wide range of biologics
  • Clin Gastroenterol Hepatol. 2006; 5:621-30.  TREAT registry.  Steroids but not biologics associated with increased mortality risk.
  • Only one chance to make first impression