• IBD

Over the past decade, population genetic studies and animal models have pointed us to the importance of multiple components of the interleukin (IL)-23 receptor signalling pathway in inflammatory bowel disease (IBD). This receptor is expressed on, and is critical for, the function of the T-helper 17 subset of T cells (Th17), upon which an almost doctrinal focus has been placed, along with, by extension, the prototypic Th17 cytokine, IL-17A, as the pathogenic molecule. However, biology appears to be more complex, as Hueber et al report the surprising failure of the IL17A antibody, secukinumab, in Crohn's disease (CD).1 In their multicentre phase IIa study, 59 patients with established moderate to severe CD were randomised to treatment with secukinumab versus placebo. At a scheduled interim analysis of the data from the first 41 patients, prespecified futility criteria were met, and the trial was therefore prematurely terminated.

A number of aspects of the study design merit further comment. The authors chose an unconventional Bayesian approach to study analysis, reporting as their primary end point the probability of the study drug achieving a reduction in the CD activity index (CDAI) from baseline of ≥50 points compared with placebo. Initial study recruitment was poor, which necessitated a number of amendments to the study protocol, and randomisation caused some imbalances in prior medical and surgical treatment. In an uncommon step, the placebo group was bolstered by the use of historical data from 671 placebo patients in six other CD studies, which were accorded equal weight with the 20 patients assigned to placebo in the present study. While this approach has the potential ethical advantage of reducing the number of patients assigned to receive no treatment, it appears particularly difficult to adjust for the vast range in placebo response rates (12–45%), outcome measures and time points of analysis between the historic studies chosen.

Nevertheless, the study results are clear: pharmacological blockade of IL-17A does not improve CD. Indeed, planned area under curve analysis of CDAI scores between weeks 4 and 10 showed statistically significant worsening of disease upon treatment with secukinumab, with post hoc analysis suggesting that this effect was worst among patients with active inflammation at the time of study enrolment. This worsening of disease was reflected in the high rate of serious adverse events, with two further serious adverse events, including one patient requiring an ileostomy, occurring in the treatment group shortly after the end of the study and hence not included in the final analysis. Almost half the patients receiving secukinumab developed an infection, including four fungal infections, which fits with our understanding of a role for IL-17A in mucosal defences against fungi, while no infections were seen in the placebo group.

Why did secukinumab fail? Reconsideration of the evidence for IL-17A pathogenicity may be informative. A few years ago, a glut of IL-23-related discoveries arising from murine IBD models, human genetic studies and studies that reported increased expression of Th17 cytokines such as IL-17A in inflamed IBD lamina propria made IL-17A blockade appear a safe bet for a novel therapeutic in CD. However, the role of IL-17A in preclinical murine models of IBD was not as unambiguous as one may have wished: depending on the model studied, IL-17A has been claimed to show a protective or an exacerbating effect.2–4 Th17 biology is complex, with IL-17A being only one of several cytokines secreted by these cells, with murine data revealing both functional redundancy and reciprocal regulation between IL-17A, IL-17F and IL-22.5 In addition, Th17 cytokine production is not limited to T cells, with other important sources of IL-23-induced IL-17A including innate lymphoid populations.6 Set in this context, blockade of IL-17A could have appeared a myopic way of targeting this complex pathway. That said, there is strong recent evidence that specific targeting of IL-17A is indeed highly effective in the treatment of psoriasis,7 a disease that shares substantial genetic risk with CD (including the same risk haplotypes at IL23R),8 and overexpression of Th17 cytokines within lesional tissue.9 The difference in outcome between trials in psoriasis and CD cannot simply be attributed to pharmacodynamic or pharmacokinetic properties of the antibody used: Secukinumab has been reported to show impressive effects in psoriasis, albeit with small numbers of patients.10 Indeed, the differential response to IL-17A blockade is also interesting in the context of results from trials of ustekinumab and briakinumab in the two diseases. These p40 antibodies block both IL-23 and IL-12 and have been shown to be impressively effective in psoriasis,11 while substantially higher doses of the same drugs appear to yield less profound benefits in patients with CD.12 ,13 A large ongoing ustekinumab phase III trial in CD should yield a final verdict.

Results from these trials have important obvious ramifications for our current concepts of the immunology underlying CD. Moreover, they may be similarly informative for our interpretation of complex disease genetics. The vastly divergent outcomes in CD and psoriasis of a therapeutic intervention targeting a shared genetic risk haplotype and pathway serves to reinforce the inherent limitations of predicting biology from risk genes and highlights the importance of extending our nascent understanding of epigenetic, microbial and environmental contributors to inflammatory diseases. The recent elegant description of a mechanistic interaction between commensal organisms, genome methylation, host immunity and gut inflammation should usher in the next generation of such studies.14

IL-17A blockade must now be added to a growing list of failed therapies in CD, including antibodies directed to interferon γ and CD25. Well-designed, in-depth immunological studies incorporated into these early-phase clinical trials would afford an excellent opportunity to gain insight into real human pathophysiology by setting the targeted molecule in the context of its biological network. Trials in IBD distinguish themselves from those in most other inflammatory conditions in that the site of inflammation is accessible and indeed typically assessed, which allows the study of the downstream effects of novel therapeutics ‘in action’. Such studies would probably also offer answers to why therapies unexpectedly fail, such as in the case of secukinumab. This could guide the further development of ‘failed’ therapeutics for indications outside IBD, and would profoundly refine our mechanistic understanding of IBD and hence foster the development of future therapies. These avenues should be considered, if we are to avoid developing the next generation of failed therapies.