In the first of a series of science commentaries from leading experts in the field of dementia research, Emeritus Professor A David Smith argues that the typical observational study approach in dementia prevention to identify modifiable risk factors should include consideration of subgroups, and interactions between risk factors, to ensure that important approaches to prevention are not missed.

What is next for dementia prevention? Look out for subgroups and for interactions

A. David Smith, OPTIMA, Department of Pharmacology, University of Oxford, UK

A typical approach to dementia prevention has been to carry out observational studies in order to identify modifiable risk factors, followed by intervention trials in which the risk factors are modified. The most successful such trials have been carried out in people at a very early stage of the disease process, such as Mild Cognitive Impairment (MCI). However, not all such trials have succeeded and I wish to draw attention to two possible reasons. First, the intervention may only be effective in a subgroup of the trial population; second, a specific example of a subgroup is that the intervention may only be effective when the agent interacts with another risk factor. I will give examples from our VITACOG trial in Oxford.

VITACOG (‘Homocysteine and B vitamins in cognitive impairment’) was designed to lower plasma homocysteine (tHcy), an established dementia risk factor (1), in people with MCI by administration of high doses of 3 B vitamins (folate, B6 and B12). The primary endpoint was the rate of whole brain atrophy and a highly significant slowing of atrophy by almost 30% was achieved in the B vitamin treated group (2). However, in a pre-specified subgroup analysis the B vitamin treatment was only significant in those with baseline tHcy above the median (11.03 μmol/L): participants below the median showed a non-significant slowing of atrophy of 11.2% whereas in those above the median tHcy the atrophy rate was slowed by 43% and in those above the upper quartile it was slowed by 53%. This subgroup effect was also found when cognitive and clinical outcomes were examined in a secondary analysis. There was no significant slowing of decline in episodic memory, semantic memory or MMSE overall in the B vitamin treated group, but a significant slowing of decline in these cognitive domains did occur in the subgroup with tHcy above the median (3). An even more marked subgroup effect was found for clinical outcomes: there was a significant improvement in the CDR and IQCODE scores only in the B vitamin group whose baseline tHcy was in the upper quartile. The results of the VITACOG trial have been reported as negative in a meta-analysis because the authors ignored the subgroup analysis (4). A similar effect of baseline tHcy was found in analysis of the effect of B vitamin treatment on regional brain atrophy: a slowing of atrophy rate in AD-related brain regions of almost 90% was found, but only in participants with tHcy above the median (5). A Bayesian network analysis revealed the following causal links between treatment and final outcomes in this subgroup: B vitamin treatment (particularly B12) lowered tHcy, which slowed regional brain atrophy, which improved CDR scores, which led to slowing of decline in MMSE scores. It can be concluded that, in a subgroup of MCI subjects with high tHcy, treatment with B vitamins slows the disease process (6).

It is not surprising that the subgroup with high tHcy showed a better response to treatment since this subgroup would have had suboptimal B vitamin status. In any trial using a nutritional intervention, analysis on the basis of baseline nutrient status should always be pre-specified. What was not expected was the second subgroup effect in VITACOG: an interaction with omega-3 fatty acid status. In view of reports that omega-3 fatty acid status can influence brain volume (7, 8), we measured the total plasma omega-3 fatty acids in the baseline samples from the VITACOG participants. A significant interaction was found between baseline omega-3 status and the effect of treatment with B vitamins on brain atrophy rate: participants with omega-3 levels in the bottom tertile (< 390 μmol/L) showed no beneficial response to B vitamin treatment whereas those with omega-3 in the top tertile (> 590 μmol/L) showed a reduction in atrophy rate of 40%. The placebo group showed no effect of omega-3 status on the rate of brain atrophy. A similar interaction was found between omega-3 fatty acid status and B vitamin treatment when several cognitive outcomes were examined in VITACOG: only participants with a high omega-3 status showed cognitive benefit from B vitamin status; there was no benefit in those with omega-3 status in the bottom tertile (9).

In summary, the VITACOG trial has shown the importance of examining subgroups: B vitamins were only beneficial in MCI in those participants with both high baseline tHcy and good omega-3 fatty acid status. There were significant interactions between B vitamins and both tHcy and omega-3 status.

Interactions have been observed in many observational studies on risk factors, most commonly with the ApoE genotype for which there is a large literature. For example, ApoE ε4 interacts with physical inactivity, smoking, alcohol, low intake of omega-3, saturated fat intake, hypertension, atherosclerosis, diabetes, education, vitamin B12, etc. (10-14). In view of the large number of reported interactions in observational studies it is imperative that any trials in which one or more of these risk factors are modified should pre-specify analysis according to ApoE genotype subgroup.

The way forward in dementia prevention should include consideration of subgroups and of interactions between risk factors. Without such considerations some trials might appear to be negative and so important approaches to prevention could be missed.


1. Smith AD, Refsum H. Homocysteine, B vitamins, and cognitive impairment. Annu Rev Nutr 2016;36:211-39. doi: 10.1146/annurev-nutr-071715-050947.
2. Smith AD, Smith SM, de Jager CA, Whitbread P, Johnston C, Agacinski G, Oulhaj A, Jacoby R, Refsum H. Homocysteine-lowering by B vitamins slows the rate of accelerated brain atrophy in mild cognitive impairment. A randomized controlled trial. PLoS ONE 2010;5(9):e12244. doi: 10.1371/journal.pone.0012244.
3. de Jager CA, Oulhaj A, Jacoby R, Refsum H, Smith AD. Cognitive and clinical outcomes of homocysteine-lowering B-vitamin treatment in mild cognitive impairment: a randomized controlled trial. Int J Geriatr Psychiatry 2012;27(6):592-600. doi: 10.1002/gps.2758.
4. Cooper C, Li R, Lyketsos C, Livingston G. Treatment for mild cognitive impairment: systematic review. Br J Psychiatry 2013;203:255-64. doi: 10.1192/bjp.bp.113.127811.
5. Douaud G, Refsum H, de Jager CA, Jacoby R, Nichols TE, Smith SM, Smith AD. Preventing Alzheimer’s disease-related gray matter atrophy by B-vitamin treatment. Proc Natl Acad Sci U S A 2013;110(23):9523-8. doi: 10.1073/pnas.1301816110.
6. Smith AD, Refsum H. Dementia prevention by disease-modification through nutrition. J Prev Alz Dis 2017;in press. doi:
7. Pottala JV, Yaffe K, Robinson JG, Espeland MA, Wallace R, Harris WS. Higher RBC EPA + DHA corresponds with larger total brain and hippocampal volumes: WHIMS-MRI study. Neurology 2014;82(5):435-42.
8. Samieri C, Maillard P, Crivello F, Proust-Lima C, Peuchant E, Helmer C, Amieva H, Allard M, Dartigues JF, Cunnane SC, et al. Plasma long-chain omega-3 fatty acids and atrophy of the medial temporal lobe. Neurology 2012;79(7):642-50. doi: 10.1212/WNL.0b013e318264e394.
9. Oulhaj A, Jernerén F, Refsum H, Smith AD, de Jager CA. Omega-3 fatty acid status enhances the prevention of cognitive decline by B vitamins in Mild Cognitive Impairment J Alzheimer’s Dis 2016;50(2):547-57. doi: 10.3233/JAD-150777.
10. Haan MN, Shemanski L, Jagust WJ, Manolio TA, Kuller L. The role of APOE epsilon 4 in modulating effects of other risk factors for cognitive decline in elderly persons. J Am Med Assn 1999;282(1):40-6.
11. Peila R, White LR, Petrovich H, Masaki K, Ross GW, Havlik RJ, Launer LJ. Joint effect of the APOE gene and midlife systolic blood pressure on late-life cognitive impairment: the Honolulu-Asia aging study. Stroke 2001;32(12):2882-9.
12. Kivipelto M, Rovio S, Ngandu T, Kareholt I, Eskelinen M, Winblad B, Hachinski V, Cedazo-Minguez A, Soininen H, Tuomilehto J, et al. Apolipoprotein E epsilon4 magnifies lifestyle risks for dementia: a population-based study. J Cell Mol Med 2008;12(6B):2762-71.
13. Wang HX, Gustafson DR, Kivipelto M, Pedersen NL, Skoog I, Windblad B, Fratiglioni L. Education halves the risk of dementia due to apolipoprotein epsilon4 allele: a collaborative study from the Swedish brain power initiative. Neurobiol Aging 2012;33(5):1007 e1-7. doi: 10.1016/j.neurobiolaging.2011.10.003.
14. Vogiatzoglou A, Smith AD, Nurk E, Drevon CA, Ueland PM, Vollset SE, Nygaard HA, Engedal K, Tell GS, Refsum H. Cognitive function in an elderly population: Interaction between vitamin B12 status, depression, and apolipoprotein E E4: The Hordaland Homocysteine Study. Psychosom Med 2013;75(1):20-9. doi: 10.1097/PSY.0b013e3182761b6c.

If you would like to provide comment on this article please email us.

Terms and conditions
The recommendations, views or findings expressed in studies, trials or articles are expressly those of the authors. Such recommendations, views or findings do not represent the recommendations, views or findings of the International Research Network on Dementia Prevention (IRNDP), its Leadership Committee or Independent Advisory Group or any collaborating institutions. IRNDP accepts no responsibility for the accuracy or content, or liability in respect of studies and articles submitted to it.