Lessons from the Multidomain Alzheimer Preventive Trial

The Multidomain Alzheimer Preventive Trial (MAPT) investigators should be commended for their work on one of the larger and longer duration multicentre trials to investigate dementia prevention.¹ The results of MAPT showed no significant effects with any of the three treatment interventions compared with placebo on the primary outcomes, the 15 secondary outcomes (except for one comparison of combined intervention with placebo on the ten item Mini-Mental State Examination, p=0·036), or on five of seven prespecified and exploratory subgroup analyses, which compared a multidomain intervention combined with omega 3 polyunsaturated fatty acids or placebo, or omega 3 capsules alone, against placebo capsules in old adults with memory complaints. Analyses showing significant effects favoured multidomain intervention with omega 3 versus placebo in a subgroup at high risk for dementia (based on Cardiovascular Risk Factors, Aging, and Incidence of Dementia [CAIDE] score), but with a small effect size (mean difference 0·131), and multidomain intervention (with either omega 3 or placebo) for amyloid-PET positive subgroups of 16 and 23 participants.

Several design features from MAPT might inform future trials of these interventions. To maintain double-blinded conditions, the multidomain intervention arms should be controlled for and masked by inclusion of an arm with similar interpersonal interactions, but without the multidomain intervention’s putative therapeutic elements of cognitive training, exercise, and nutritional advice. In MAPT, the multidomain intervention, with and without omega 3, provided the only potential cognitive effect compared with either placebo or omega 3 (Article¹ appendix p 5). Given that participants and investigators knew whether or not the multidomain intervention was given, it is not possible to determine whether this effect was due to the intervention, ascertainment bias, or chance. Because this very small effect of 0·078 (Article¹ appendix p 5) was only a fraction of the minimally clinically relevant difference of 0·334 SD that was used to establish the sample size, it might not be clinically meaningful. Furthermore the two unbiased comparisons did not differ (Article¹ appendix p 5)—ie, no effect was detected when omega 3 was compared with placebo, and when multidomain intervention with omega 3 was compared with multidomain intervention with placebo.

The dose of 800 mg docosahexaenoic acid (DHA) per day, given in the omega 3 capsule, was set just above the intake recommended by the French food safety authority, well below the maximum, and was probably associated with minor increases in concentrations of DHA in the CSF. People with cerebral amyloidosis (34% based on 67 positive and 133 negative florbetapir-PET scans) or carriers of an APOE ε4 allele (287 [23%] participants) probably have decreased delivery of DHA to the brain compared with individuals with less amyloidosis or not carrying APOE ε4.² In the Alzheimer’s Disease Cooperative Study-sponsored trial of DHA that used plasma saturating doses of 2 g DHA per day, APOE ε4 carriers showed only a 37% increase in DHA in the CSF whereas APOE ε4 non-carriers showed a 68% increase compared with placebo after 18 months.² On this basis, the DHA dose in MAPT likely would have increased CSF concentrations by less than 20%. Most clinical trials using 1 g or lower doses of omega-3 have had negative results for cognitive outcomes.³ Even larger DHA doses have been ineffective in trials for mild cognitive impairment⁴ and mild Alzheimer’s disease,⁵ possibly because brain pathology is quite advanced in these patients.

Beyond dose considerations, the 3 year treatment duration was probably too brief to show benefit because the half-life of DHA in brain phospholipids is about 2·5 years.⁶ Results from rodent studies show improvement in Alzheimer’s disease pathology with omega 3 supplements when given early and for at least 10% of the animal’s lifespan, with attenuated effects reported if supplementation starts at a later age and disease stage.⁷ Any benefit linked to DHA presence in the brain phospholipids would take years of treatment to show. The need for long-term exposure is consistent in results from longitudinal studies which show that polyunsaturated fat intake during mid-life⁸ and over several years³ is associated with decreased Alzheimer’s disease risk.

These considerations and the MAPT outcomes suggest that it would not be practical to implement new trials with participants similar to and as heterogeneous as those in MAPT and to expect different outcomes. Rather than replicating MAPT or doing prodromal Alzheimer’s disease trials, we suggest that omega 3 supplementation would be better used for younger adults than those included in MAPT with higher, brain-penetrant doses of DHA and for much longer periods to ensure that the intervention has a chance for success. Particular populations that could benefit from early and long-term omega 3 interventions are APOE ε4 carriers in mid-life and people with PSEN1 and APP mutations associated with the dominantly inherited form of Alzheimer’s disease.

Given that the most compelling outcome of MAPT is the potential for multidomain intervention (with or without omega 3 treatment) to show early and sustained cognitive effects, research could also focus on identifying the therapeutic elements of multidomain intervention and trying to maximise any effect in future studies.⁹