Tofu, Soy Products, and Dementia
The long-running tofu-dementia connection remains a strong concern. To date, to the best of my knowledge, the ONLY prospective data on the long-term effects of soy products in the human diet are the White et al report (1), which is very rigorous prospective epidemiology in a very well-characterized long-term cohort study, and did indeed find that even what we might think of as pretty moderate increases in tofu consumption were associated with greater brain brain atrophy and poorer late-life cognitive scores.
More recently, there are two reports from Hogervorst et al. (2,3) a weaker (cross-sectional) study which confirmed the finding of brain atrophy and cognitive decline: “weekly tofu intake had a negative association with memory in analyses controlled for age, sex, education and site. … The analyses were not much altered when entering other foods. … Analyses using median splits of soy product intake per week gave similar results to those concerning continuous consumption per week, showing a better memory performance in those who did not consume tofu daily or more than once daily.” The effect was not seen in younger subjects ((3): indeed, oddly, such subjects seemed to have better cognitive function on a higher-tofu diet), but our proper concern is with longer-term effects, not what may be a short-term boost ((by analogy, not equivalence, cf. the effects of amphetamines).
I have also come across (4), a prospective study in a cohort of Japanese Americans in King County, Washington. The results were all over the map (tho’ perhaps suggestive, with the bidirectional relationships with HRT use), and it was an early report from only 2 y of followup in the cohort and after only 1 dietary assessment (dietary reporting instruments really need to be repeated, both because people are highly imperfect at reporting and because people’s diets do actually change over time). I wrote to the authors to inquire about further followup: unfortunately, they won’t be doing “a longer follow-up because of approximately 1200 women in the cohort, only 274 were in this sub-cohort. We likely do not have sufficient statistical power to conduct these newer analyses” (personal communication, Dr. Amy R. Borenstein).
And, finally, there is (5), another cross-sectional study (this one in Hong Kong). It found that in women, and not men, higher consumption of an a posteriori-identified “vegetables-fruits” dietary pattern “rich in vegetables, fruits, soy products and legumes” was associated with reduced risk of cognitive impairment — but so was a “‘snacks-drinks-milk products’ pattern which was a mixture of healthy and unhealthy food groups including fast food, sweets and desserts, nuts, milk products and whole grains”. Obviously, aside from being cross-sectional, there’s no way to determine in that mix if soy products were playing a protective, deleterious, or neutral role in the overall protective association of the “vegetables-fruits” dietary pattern, even if we don’t laugh the whole exercise out of the court of scientific opinion.
Tofu, or Soy Generally?
As Rodney says, so far this has only been found for tofu — but that’s because very few soy foods have even been looked at. (1) found that miso was not associated with brain atrophy, but the serving size is very small and it is a complicatedly-processed food. Hogervorst et al actually found a protective effect of tempeh (2): this could be due to something about its processing neutralizing a toxic factor (formaldehyde was not found in market samples of tempeh, but again its presence in Indonesian tofu has only been intermittent, so that might be a sampling error), or enhancing a protective factor (eg., its higher levels of folate), or it could be a fluke (this is, again, a smaller and weaker study than White et al). We just don’t know, and won’t until we get (a) more studies confirming (or refuting) the link to tofu and a non-link to other soy foods.
White et al fingered isoflavones (soy phytoestrogens) as the most likely culprit for the tofu association, based on some animal data (and, presumably, the sheer fact that their presence is teh main thing to distinguish them from other similar foods). Phytoestrogens are found to improve cognitive function in some short-term studies in (usually menopausal or perimenopausal) women, but that doesn’t mean that there isn’t a long-term deleterious effect, or one specific to their hormonal status (cf. the HRT bidirectionality in (4) and maybe the inverse association of the “vegetables-fruits” dietary pattern of cognitive impairment in women but not men in (5)).
Most soy foods, including most processed soy foods (soy burgers and hot dogs) and soy protein powders, have lots of isoflavones in them, and some are even “fortified” with extra isoflavones. So if those are the culprits, individual foods could be risky or harmless depending on how much isoflavone they contain. However, it could be something else entirely, in which case it would depend on whether the food contains “Contaminant X.” Again, we just don’t know, and won’t until we get (a) more studies confirming (or refuting) the link to tofu, and (b) a non-link to other soy foods, and © a strong mechanistic explanation for any discrepancy. Thus, we shouldn’t (yet) rule in or out any particular soy product on the basis of the presence or absence of phytoestrogens or any other factor. Again, more studies needed!
Following up on a throwaway comment by an unreliable previous poster, Dean diligently found an interview by the lead author of (2) indicating the possible involvement of formaldehyde contamination as a factor in Indonesian tofu — which is, in fact, mentioned in the MS of (2). There is plenty of evidence that formaldehyde exposure is negatively associated with cognitive function, and in experimental studies it can hyperphosphorylate tau (leading to intraneuronal lesions that drive neuronal death and cognitive decline). So that’s a plausible driver — in the Indonesian study. It would not, however, explain the original finding in a cohort of Japanese men living in Hawai’i, which (again) was a stronger study design to begin with.(1)
The one further complication is that the problem of formaldehyde contamination in Indonesia is not specific to tofu: aside from only being intermittently found in tofu, it’s also been found in chicken and noodles, in use by restaurants and food stalls, and “scores of snacks and beverages containing formalin and rhodium-B were sold at a number of schools.”! That might to some extent make its presence in tofu a bit of a wash relative to other foods, effectively taking it off the table as an explanation for the tofu-dementia link: we just don’t know.
After I wrote this post, I realized that I hadn’t addressed the question of aluminum contamination from processing with Al-based pots, or increased Al accessibility from the tofu-making process, which had both been invoked as potential causes of the association — a link that seems to have first been made in this post by Honolulu vegan advocate Dr. William Harris. I was inclined to dismiss this, as the last time I reviewed the evidence on this it seemed as if it had largely collapsed under the weight of negative studies. However, re-reviewing it now (see eg. these studies), the evident confusion about the bioavailability and body distribution of aluminum from different sources, and looking at cautious and responsible sources like the CDC’s Agency for Toxic Substances and Disease Registry and the skeptical Steven Novella’s blog on Alzheimer’s and aluminum, my overall impression is that while there is no slam-dunk, there clearly seems to be a rational basis for concern.
In light of this, people should not be making definitive statements either way about aluminum’s link to AD specifically — and more importantly, to dementia more broadly. In the real world, no one gets pure AD, or vascular dementia, or dementia with Lewy bodies, or hippocampal sclerosis: one gets brain aging featuring elements of all of these, with (in most cases) features of one dominating neuropathologically and corresponding to a clinical dementia phenotype that is more or less characteristic of that neuropathological pattern. Age-related cognitive decline, dementia, and specific dementia phenotypes emerge from the aggregate loss of functional reserve. Aluminum could be neurotoxic in a way that is not specific to AD, and still increase the number of people who emerge with clinical dementia characterized as AD (or not).
In turn, they should not be making definitive statements attributing the tofu-dementia association to aluminum toxicity. We only have the most tentative suggestion (via Dr. Williams) that tofu in one geographic location (Honolulu — which was the site of (1), but not of (2,3)), at one particular point in time (some years after the patients in (1) were eating the tofu that is somehow linked to their brain atrophy) had variable but potentially worrisome levels of aluminum. And, again, an uncertain if worrisome link between oral aluminum exposure and dementia or AD. A mixture of uncertain epidemiology, mixed animal data, and weak circumstantial evidence is not a basis for definitive statements.
I’m not aware, unfortunately, of anyone marketing tofu or other products using methods that specifically minimize aluminum content, or that test and certify to the low levels in their product. And, as indicated, if such products exist, it’s not clear whether that would be sufficient grounds to start liberally consuming soy again with no fear of dementia
With the best available study (1) finding an association with brain atrophy and cognitive impairment; with supportive evidence from a weaker study (2,3); and absent even ONE equally-strong report of a lack of effect or of a protective one, it seems quite prudent to minimize or eliminate consumption of tofu per se (I see you rolling your eyes, Mary!), and likely soy products generally, especially in men.
You must be logged in to reply to this topic.