Assessment of cognitive safety in clinical drug development

Abstract

Cognitive impairment is increasingly recognised as an important potential adverse effect of medication. However, many drug development programmes do not incorporate sensitive cognitive measurements. Here, we review the rationale for cognitive safety assessment, and explain several basic methodological principles for measuring cognition during clinical drug development, including study design and statistical analysis, from Phase I through to postmarketing. The crucial issue of how cognition should be assessed is emphasized, especially the sensitivity of measurement. We also consider how best to interpret the magnitude of any identified effects, including comparison with benchmarks. We conclude by discussing strategies for the effective communication of cognitive risks.

FIGURE 1

The effect of an antipsychotic and a benzodiazepine drug on reaction times when administered in combination is greater than the sum of each separately, indicating a drug–drug interaction. At the time of maximum impairment (3 h), responses on a choice reaction time test were slowed by approximately 30 ms by the antipsychotic, approximately 50 ms by the benzodiazepine [greater than the effect of blood alcohol concentration (BAC) of 0.05 g/dL, the legal driving limit in many countries], but by approximately 175 ms when these were administered in combination (greater than the effect of BAC of 0.1 g/dL). If driving at a speed of 100 kph (60 mph), a slowing of response of approximately 175 ms is equivalent to an increased stopping distance of approximately 4.9 m (approximately 15 ft), the length of a large sedan car. Source: Unpublished data, kindly provided by Otsuka Pharmaceutical Co., Ltd.

FIGURE 2

Commonly assessed components of cognition, broadly split into the domains of input, storage, and control. Source: Reproduced, with permission, from Cambridge Cognition.

FIGURE 3

Results from a Phase I study of the comparative effect of 5-mg zolpidem (benzodiazepine with sedative properties: left side of figure) and GSK1521498 (µ-opioid inverse agonist in development for obesity and/or addiction: middle–right side of figure) on reaction times during a test of attention. Zolpidem (Zolp) slowed responses, relative to placebo, to approximately the same degree as previously shown at a blood alcohol concentration of 0.05 g/dL (approximately 25 ms, green points [55]). Even at the highest dose tested, the average impairment caused by GSK1521498 (approximately 20 ms, red/orange points) was lower than that caused by 5-mg zolpidem, consistent with a relatively low cognitive risk and supporting the continued clinical development of the compound. Abbreviation: PBO, placebo. Source: Reproduced, with permission, from [54].