Bridging the gap: Assessing safety beyond clinical trials in psoriasis

There is a lack of symmetry between the assessment of efficacy and safety for a given intervention. While efficacy is usually demonstrated through randomized clinical trials (RCTs), questions about safety often remain unresolved. There are multiple reasons for this, the most obvious of which are statistical and related to the artificial conditions of RCTs. The typical sample size in an RCT rarely exceeds a few thousand carefully selected participants, who are evaluated over a relatively short period of time. This is generally sufficient to document common short-term outcomes—such as improvement of a chronic condition or resolution of an acute one in a simplified patient population, especially when compared against placebo—but inadequate for assessing rare or long-latency adverse events (AEs). These AEs, by necessity, can only be reliably evaluated once the drug has entered the market, in the so-called post-marketing phase.

Disease-specific clinical registries offer a valuable means of assessing medical interventions in this post-marketing context.¹ They enable evaluation of efficacy and safety in a broader, less selected patient population and in a real-world setting (‘real-world’ being an often-abused term, sometimes used to simply describe what was once considered a ‘case series’). Several variables, summarized in Table 1, can affect drug assessment in such real-world contexts and should be considered in study design. We are indebted to psoriasis registries such as the BadBIR registry in the United Kingdom,² the Italian registry Psocare³ and the Spanish registry BioBadaderm⁴ for improving our understanding of the factors influencing clinical response and its maintenance over time—and ultimately for enhancing psoriasis management.

In this issue of the Journal, data from the BioBadaderm study are presented, specifically focusing on safety assessments of targeted biologic and some non-biologic therapies in psoriasis.⁵ Various lessons can be learned from the study.

First, a distinction should be made between AEs and adverse reactions (ARs)—a difference that is not merely semantic. AEs were considered in the BioBadaderm registry. According to the Medical Dictionary for Regulatory Activities (MedDRA), an AE is ‘any unfavorable and unintended event temporally associated with the use of a medical treatment or procedure, regardless of whether it is considered related to the treatment’, whereas an AR implies the establishment of a causal relationship. Clinical judgement about the relevance and aetiologic role of a given drug is missing. The analysis of AEs can raise signals, but it does not establish causality.

Second, the limitations of conventional RCTs in assessing safety cannot be overcome by meta-analyses, even those employing a network meta-analysis design. As highlighted by the authors, meta-analyses typically rely on data from RCTs and are usually restricted to the induction period. While statistical power is increased, the fundamental limitations of RCTs, as mentioned above, remain unaddressed.

Third, current registries, even when enrolling substantial numbers of patients, cannot fully address the challenge of assessing risks for rare events such as most cancers. Alternative designs—for example, case–control studies focusing on specific AEs—are more appropriate and allow for control of co-factors and susceptibility variables. With 17,284 patient-years of follow-up in the BioBadaderm study, spread across 13 different drug exposures, only a doubling of risk for AEs with an incidence of ≥1:100 can be reliably evaluated.

Finally, the key lesson from the BioBadaderm study is that clinical research should not be relegated to specialized settings; it should be an integral part of the activity of proficient dermatologists. In the ‘natural laboratory’ of everyday clinical practice, they can raise questions and attempt to address them—thereby contributing to our collective knowledge.

None to disclose.