Uncertainty not helped by meta-analysis
“Given the uncertainty associated with the results from BSRBR-AS, we decided to undertake a meta-analysis to try to accumulate other data that could help us answer this question,” Dr. Macfarlane explained.
However, this didn’t really help clarify things because combining BSRBR-AS data with the results of a couple of observational studies suggested that the odds of of IBD doubled with any biologic treatment versus no biologic treatment (odds ratio, 2.19), and a 2.5-fold higher likelihood considering etanercept versus no biologic treatment, but no difference was seen comparing etanercept to other anti-TNF agents (OR, 0.93).
When the meta-analysis was restricted to RCTs, the rate of IBD per 1,000 person-years was 3.43 for placebo, 5.64 for all biologics, 8.14 for etanercept, 2.35 for other anti-TNFs, and 7.02 for IL-17 inhibitors.
For extensions of RCTs, IBD rates per 1,000 person-years of follow-up were 2.91 for etanercept, 0.83 for other anti-TNFs, 3.61 for IL-17 inhibitors, and 2.79 for all biologics.
“There was only a small difference in IBD incidence between the biologic therapy and the placebo groups” in the RCTs and associated studies, Dr. Macfarlane said, adding that “there was a small excess incidence associated with etanercept, compared to other anti-TNF agents, and [for] IL-17 therapy, compared to nonetanercept, anti–TNF-alpha therapies.”
Of course, the different study designs and durations of exposure to the various treatments raises significant methodological issues.
“Randomized, controlled trials should provide the highest quality evidence as a result of their design and randomizing patients to treatment,” Dr. Macfarlane said. “However, their relatively short follow-up, as well as their restrictive eligibility criteria, may work against finding a difference in IBD incidence if it were to exist.”
Observational studies are very valuable in the data they can provide but are also beset with problems, such as surveillance bias and confounding by indication.
The higher risk of IBD that was observed in observational studies could be an issue with study design, or perhaps, “in routine clinical practice, rheumatologists are taking on board factors that we have not measured, that are negating any slight increased risk,” Dr. Macfarlane said.
Session chair Nicola Goodson, MBChB, PhD, of Liverpool (England) University NHS Foundation Trust, commented: “I think that could well be a very reasonable explanation, because I think as a clinician, you do tend to channel drugs away from some people and channel drugs towards others.
However, Dr. Goodson noted that there was “a glimmer” of signal coming from the RCTs.
“Methodologically, that is what you would have to take as the most robust evidence,” Dr. Macfarlane said, “but even with all the evidence available, it’s still very hard for us to quantify; that has enormous uncertainty.”
Dr. Macfarlane and Dr. Goodson had no relevant conflicts of interest to disclose. The BSRBR-AS is supported by the BSR, which receives funds to support the registry from Pfizer, AbbVie, and UCB.