Commentary

The cholinergic hypothesis of memory loss has to contend with the glutamatergic hypothesis of amnesia and dementia.¹ These studies by Imbimbo et al and Winblad and Poritis are based on these alternative approaches.

Acetylcholinesterase inhibitors (ACh-EIs) have been the most widely investigated drugs in mild to moderate AD. Measures of their efficacy should tap 4 areas: cognition, functional activity, non-cognitive behaviours, and independent ratings of global change. Compared with placebo, the AChEIs tacrine, donepezil, metrifonate, and rivastigmine show a modest improvement of cognitive function of 2–4 ADAS-Cog points over 6 months, with a return then to baseline values or less.² Ratings of global impression of change improve 1/3–1/2 point. Non-cognitive dysfunction has rarely been studied, and improvements in activities of daily living (ADL) have been hard to detect, except for rivastigmine.³ Overall, up to one quarter of patients do well on AChEIs, but the majority see little to no benefit. Which quarter improves, we cannot predict. The decrease in burden of care has been studied less.

The study by Imbimbo et al is well designed and was done with a low dropout rate. Unfortunately, the magnitude of the cognitive benefit was in the lower half of the range reported for other AChEIs and the number with the change normally considered clinically significant (≥4 ADAS-Cog points) is not reported. But eptastigmine joins rivastigmine in showing substantial benefit on ADL. The dose response effect and the consistency of modest improvements in cognition and daily functioning support the clinical relevance of the results. Although the cholinergic tolerability of eptastigmine was favourable, 4.5% of patients developed a mild transient neutropenia, which was dose dependent and included 1 case of severe neutropenia in the eptastigmine 20 mg group (3 times daily). It is surprising that patients with neutropenia continued in the trial and that there is no comment on incident infections. Another patient on eptastigmine, 20 mg (3 times daily), however, developed aplastic anaemia, a risk of 0.3% (95% CI 0.01% to 1.7%), whereas the risk in the general population is 2 per million per year. The risk on eptastigmine, similar to that of clozapine, will probably limit the use of the drug, certainly if there is no blood monitoring and with safer alternative drugs available. Patient or carer convenience was also less with thrice daily medication compared with rivastigmine's twice and donepezil's once daily regimens.

Severe dementia imposes a great challenge to carers. Functional and global clinical endpoints are more relevant here than cognitive ones. Memantine is an N-methyl-D-aspartate antagonist, which blocks the exitotoxic effects of excess glutamate release, considered central to memory and behavioural dysfunction and to the pathology of AD and vascular dementia.¹ The results of the memantine study by Winblad and Poritis appear to be unbiased. The coincidence of treatment effects in independent nursing and medical assessments supports the validity and clinical significance of the results, but the design allowed no examination for dose response effect. The reported benefits and tolerability need to be established in a much longer duration trial, with health economic analysis. Adverse effects (not described) and serious adverse effects (mainly cardiac deaths) were equivalent in the 2 groups. Memantine seems useful not only in severe AD but also in vascular and mixed dementias, and in those without substantive behavioural problems (neuroleptics, hypnotics, and some antidepressants were excluded in trial patients), found in approximately 40% of patients with AD. Although glutamate antagonists may induce transient psychoses, memantine is thought not to do so.

These 2 contrasting pharmacological approaches to AD, here studied at different levels of severity, leave key issues unanswered: who benefits? Is response genotype specific? How substainable and valuable is a 2 point drop in carer dependence (memantine) in progressive dementia? Are there longer term side effects? Are skills regainable without rehabilitation? And how do the modest gains translate into quality adjusted life years and reduce burden of care and costs? In the future, non-cognitive symptoms should be assessed because they increase disease progression and disability, predict high family burden and costs, and are the primary cause of institutionalisation.¹ Memantine's possible advantage of neuroprotection suggests that treatment combinations with AChEIs might enhance the benefits of only 1 approach.