GSK1034702

One of very few muscarinic candidates to produce a positive human cognitive signal, and it managed it by being far less selective than advertised. A useful reminder that in this class the benefit and the side effects often come from the same lack of discipline at the receptor.

The human safety record is short and it is not reassuring. In the nicotine abstinence study the compound produced gastrointestinal adverse effects consistent with activation of peripheral muscarinic receptors, and the later pharmacology paper states plainly that its intrinsic agonist activity and its lack of subtype selectivity are the likely causes of the adverse effects seen in its clinical trials ^[2][1].

Direct agonism at the peripheral subtypes is the mechanism to watch here rather than the M1 story the compound was presented with; in isolated tissue it slowed rat atrial contraction to the same extent as methacholine and contracted rat ileum to about half the methacholine maximum ^[2]. Whether it carries the seizure liability documented for M1 ago-PAMs is not established, because the convulsion work in this class was done on the modulators rather than on this molecule ^[5]. Reported human exposure is limited to single doses in small phase 1 studies, so nothing is known about what repeated dosing does.

GSK1034702 came from a GlaxoSmithKline benzimidazolone series and went through a set of phase 1 studies between 2008 and 2010 covering single dose safety and tolerability, repeat dosing and drug interaction, brain uptake and cognition. In the cognition study, published by Nathan and colleagues, 20 male nicotine-abstinent smokers received placebo, 4 mg and 8 mg in a randomized double-blind crossover design; the 8 mg dose significantly improved immediate recall but not delayed recall, and no other cognitive domain moved ^[1]. Because no M1 PET tracer existed, GSK radiolabeled the drug itself with carbon-11 and scanned two anaesthetized baboons and four healthy men, finding brain uptake consistent with passive diffusion ^[3]. The public trial registry lists only those phase 1 studies, all of them completed by 2010.

In the muscarinic literature GSK1034702 is remembered as a cautionary example rather than a lost opportunity. The 2018 reanalysis concluded that its bitopic binding, intrinsic agonism and weak subtype selectivity are exactly the properties a clinical candidate should not have, and argued instead for pure positive allosteric modulators with low intrinsic activity ^[2]. Its positive memory result is still cited as evidence that M1 signalling can move human cognition at all ^[5]. It has no presence outside that literature.