Corticosteroids and antibodies to tumor necrosis factor (anti-TNF) are mainstays of treatment of acutely active IBD and while anti-TNF therapy is indicated for maintenance therapy, chronic corticosteroid therapy is associated with known significant risks. Nonetheless, chronic or recurrent corticosteroid treatment in IBD is common. In this edition of the journal Lewis et al. use Medicaid and Medicare databases to contrast adverse outcomes in persons with IBD and either corticosteroid or anti-TNF use. Compared to high dose corticosteroid use, anti TNF therapy is associated with less risk for death in Crohn’s disease but no statistical difference in death outcomes in ulcerative colitis. The adverse effects of corticosteroids are well known yet vigilance is still required to ensure their use is limited to the short term.
The most adverse of outcomes is death, yet clinicians do not typically consider death among their concerns when planning an intervention in persons with inflammatory bowel disease (IBD). Crohn’s disease (CD), but not ulcerative colitis (UC), is associated with an increased risk for death compared with the general population (1). However, in both CD and UC there is a significant increased risk for death within 1 year of diagnosis and in both diseases surgery poses an increased risk especially in the short term postoperatively, and, especially, in the elderly (1). Using administrative data to assess the relative risk for death has several strengths. The database used in this Canadian report was population based and could identify diagnoses associated with causes of death. However, it is near impossible to know from administrative data to what extent complex or untreated IBD activity contributed to ultimate death from, say, cancer or infection.
Although it has been shown that corticosteroid (CS) use is associated with an increased risk for sepsis (2, 3) and thiopurine therapy and anti-tumor necrosis factor (anti-TNF) are associated with an increased risk for lymphoma (4) an unanswered question is what those risks would be in persons with IBD not treated at all. A French study recently reported that thiopurines and anti-TNF comparably increase the risk for lymphoma and the combination further increased the risk threefold over monotherapy with either agent (4). It is unknown, however, how those patients, ill enough to require a thiopurine or anti TNF would have fared if they had not been treated with those agents at all. This natural history experiment will likely never be done unless affected persons present in jurisdictions where these medicines can not be accessed. In a Manitoba study prior to the advent of anti-TNF therapy the lymphoma risk was increased in males with CD independent of thiopurine use (5). On the other hand, many patients with IBD do not require IBD-specific therapy at different times in their disease (6, 7). These patients likely have milder disease and likely have lower overall complication and mortality rates. Therefore, understanding the risk of death in IBD and to what extent bad outcomes leading to death can be attributed to medical therapy is difficult to quantify for patients because it is unknown how they might do, poorly or otherwise, without these therapies. There is precedent throughout the gastrointestinal tract that chronic inflammation is associated with cancer and this includes chronic colitis where the colorectal cancer risk is higher with increased chronic inflammation.
In light of the missing information; that is how individuals affected by a complication thought to be treatment-related would have fared without therapy, what is often reported is a comparison of outcomes between different therapies. Until the recent approval of Vedolizumab and Ustekinumab for treatment of IBD this meant comparing chronic CS use, anti-TNF use, thiopurine use, and surgery. These have been the mainstays of treating at least moderately active disease over the past two decades.
In this journal Lewis et al. used Medicaid data (from 2001 to 2005) and Medicare data (from 2006 to 2013) to assess outcomes in persons with IBD who received either at least 3000 mg of prednisone within a 12-month period or a new anti-TNF initiation (8). Of note, over twice as many CD patients and three times as many UC patients received prolonged CS compared to anti TNF. The Medicaid data would be biased by the data being from an era where anti-TNF therapy was not as liberally initiated. A strength of their study was using propensity scoring in an attempt to minimize the inclusion of patients who were too well to have likely received either of CS or anti TNF and also to allow for patients to switch bidirectionally between the treatment arms, contributing follow-up time to the treatment that they had most recently received (considering that most patients with IBD receiving anti TNF will have received CS for some length of time).
The risk of death was significantly lower in patients treated with anti-TNF therapy for CD [odds ratio (OR) 0.78, 95% confidence interval (CI) 0.65–0.93] but was not significantly lower for UC (OR 0.87, 95% CI 0.63–1.22). Although the two treatment groups were balanced for comorbidity, the mean age at death was 67–76 years and hence, we can not be certain if the CS users who never initiated anti-TNF were deemed too risky for anti-TNF either by virtue of cardiac status, recent infection status or other general health issues. In other words, did the CS users, despite best efforts to control confounders, have a higher propensity of death. This could also explain the lower risk of major advanced cardiac events in the anti-TNF group and possibly for hip fracture (both events found to be lower in the anti TNF group). Since the risk of hip fracture is known to be increased in CS users the results should not necessarily be interpreted as anti TNF having bone protective effects. Especially in the Medicaid participants, but even for Medicare participants, it is unknown if there were bureaucratic impediments to the prescription of anti TNF considering they are hundreds times the expense of CS. Despite the possibility that patients at greater risk were not prescribed ant-TNF, in this study there was no significant difference in risk for infection or cancer between CS and anti-TNF users.
It has been suggested that in the early experience with anti TNF use, practitioners and patients often chose CS over anti-TNF because of fears for anti-TNF toxicity (9). However, in general, anti-TNF therapy has been considered very safe and well tolerated (10). Despite the advent of more effective therapies, CS continue to be widely used in the treatment of IBD even in health care systems where access to expensive new drugs is relatively liberal (11). So why do practitioners still widely prescribe chronic CS, as has also been shown by Lewis et al. (8)? For one, most practitioners feel comfortable with CS dosing and knowledge of their potential side effects since they are widely used across specialties and for varied diseases. Second, many practitioners initiate a prescription for a “short” or “intermediate” course of CS and whether due to missed appointments, patient preferences or administrative impediments to initiating biological therapy the “short” course of CS may become a prolonged one. Further, for some patients who use CS they are used to the rapid response, not always evident with anti-TNF use and this rapid response is more conducive to travel, or work productivity regardless of expected side effects. Finally, there is a huge cost differential between CS and anti TNF use. CS have been used for over 60 years in treating inflammatory diseases and unfortunately familiarity of the risks of chronic CS has not sufficiently lessened the reluctance to substitute newer agents such as the biologicals that may carry lesser risks but potentially also less certainty for risk for serious outcomes. Clinicians embrace certainty. When doctors are uncertain it seems we are more likely to fall back to treatments of which we have more certainty, be it for benefits or for toxicity.
The time to change our ibd prescribing habits is overdue
CS are useful drugs that are both inexpensive and easy to administer. Anti TNF therapy which has now been used for nearly 20 years carries a relatively increased risk for infection and cancer but the absolute risks for infection and malignancy are small (4, 10). The elderly are a population that may do worse with infection but this risk could be mitigated by using anti TNF monotherapy without combination immunosuppressive therapy and ensuring levels are adequate to reduce the likelihood of antidrug antibodies formation. Another option is the use of gut specific anti-integrins like Vedolizumab which should have less systemic toxicity, less risk for infection and less risk for malignancy considering its gut specificity (however, those suggested reduced risks have not been proven in head to head comparison with anti TNF). In comparison to the options of CS dependence or surgery, immunomodulating therapy with biologicals should actively be pursued in the elderly as well as in younger IBD patients despite the potential for added risks in the elderly (12, 13). CS should and will remain in our armamentarium for a long time yet as their use in the short term has gotten many patients out of acute trouble. Their prolonged use, however, has gotten many patients into acute trouble and needs to be reassessed vigilantly.