Fast dissociation rates predict neither efficacy nor lack of side-effects

Response to V. Seutin

We thank Seutin for both his enthusiasm for our review¹ and for the opportunity he has presented us to address his proposal that drugs with weak affinities for central nervous system (CNS) receptors might represent superior treatments for a variety of CNS disorders (for example, 'magic softballs'). This general notion came from the observations of others² that some drugs with low equilibrium-dissociation constants (K_d values generally in the high-nM to low-μM range) represent effective treatments for several CNS disorders, including schizophrenia (for example, clozapine and quetiapine), Alzheimer's disease (for example, memantine³), and mood and seizure disorders (for example, lamotrigine and carbamazepine).

The following relationship holds for 'off-rates' (k₋₁), 'on-rates' (k₊₁) and equilibrium-dissociation constants (K_d values), assuming simple bi-molecular kinetics:

Because ion-channels can display extraordinarily complex kinetic schemes with multiple affinity states and binding sites (both othosteric and allosteric), the estimation of the true 'off-rate' is both channel-state- and time-dependent. Nonetheless, in the case of memantine, for instance, it is clear that memantine is a low-affinity blocker of the N-methyl-D-aspartate (NMDA) channel with both fast association and dissociation kinetics³ for the open state of the NMDA channel. Indeed, it is this peculiar property of memantine which yields physiologically relevant levels of blockade with few of the behavioural side effects and toxicities associated with NMDA blockers with slower 'off-rates' (for example, phencyclidine, and MK-801; see Ref. 3 for a discussion). The remarkable selectivity of memantine for the NMDA channel is largely responsible for its relative lack of side effects (Fig. 1).

Figure 1: Memantine: a remarkably selective rapidly dissociating NMDA open-channel blocker.

Shown are mean K_i values for selected molecular targets (see http://kidb.case.edu for details) for memantine in nM. Where bars are absent, memantine showed no appreciable affinity (K_i>>10,000 nM).

In the case of neurotransmitter transporters, the relationship between lack of side effects and fast dissociation rates simply does not hold. Indeed, all therapeutically useful transporter inhibitors — even those which hit multiple transporters — have uniformly high affinities for, and slow dissociation rates from, their presumed molecular targets (see http://kidb.case.edu for affinity constants). The relatively few side effects that are associated with newer drugs (for example, fluoxetine and paroxetine) when compared with older medications (for example, imipramine and amitryptiline) are due to the relative lack of high affinity 'off-target' interactions leading to a relatively absence of anti-muscarinic, anti-adrenergic and quinidine-like cardiotoxicities (see http://kidb.case.edu).

We have also formally examined the relationship between the experimentally determined (or calculated) k₋₁ values for a variety of antipsychotic drugs (Fig. 2; see Ref. 4 for details) with both high and low rates of extrapyramidal side effects (EPS). Our analysis reveals that there is no simple relationship between the lack of EPS and k₋₁ values when many antipsychotic drugs are evaluated (see Ref. 4 for a discussion). Indeed, as Fig. 2 shows, there is considerable overlap between the k₋₁ values for drugs with high and low incidence of EPS. Indeed, one drug which is devoid of EPS (SR46349B) and which has recently been shown to be as effective as haloperidol in treating schizophrenia⁵ has a very slow experimentally determined dissociation rate from its presumed molecular target, the 5-hydroxytryptamine (5-HT)_2A receptor (k₋₁ = 0.0419 min⁻¹; see Ref. 6).

Figure 2: Dissociation rates do not predict side effects for antipsychotic medications.

Shown are k₋₁ values (see Refs 2,4,6 for details) for selected clinically effective antipsychotic drugs. Drugs with low incidence of EPS are in red, those with a high incidence of extrapyramidal side effects (EPS) are in black.

Finally, we emphasize that even 'fast' off-rates are relatively 'slow' from the perspective of a neuron: neuronal transmission and signal transduction takes place in the millisecond–second time period, whereas off-rates are in the minute–hour time period. Additionally, at steady-state levels sufficient drug is present to yield considerable steady-state occupancy levels in vivo (see Ref. 7 for example). Under these circumstances, it is likely that the pharmacokinetic properties of some medications — especially elimination rates and relative volumes of distribution — are responsible for some of their favourable clinical properties⁸.

Therefore, although medications that target ion channels and which have relatively fast un-blocking kinetics and relatively few off-target interactions might have favourable side-effect profiles, this is not likely to be a useful property for most CNS drug classes.