Having diabetes seems to drive up risk for dementia, including Alzheimer disease, but the jury has been out as to how this occurs. Now, Japanese scientists propose a mechanism—insulin resistance—and find that diabetes-related readouts of this condition associate with AD pathology prior to memory loss. As reported online August 25 in Neurology, researchers led by Toru Iwaki, Kyushu University, Fukuoka, determined that insulin-resistant seniors tend to rack up more brain Aβ than people with normal insulin regulation, and that having the ApoE4 allele further heightens the risk for these brain lesions. Given epidemiological data suggesting that some two-thirds of U.S. elderly ages 65 and older are insulin-resistant (Cowie et al., 2009), the impact of this potential risk factor could be significant.

Longitudinal studies have shown that type 2 diabetics don’t fare as well cognitively (Allen et al., 2004), and develop AD at higher rates than the general population (Arvanitakis et al., 2004). However, the relationships are far from absolute. Some diabetics stay mentally sharp, most AD patients do not have diabetes, and diabetes itself develops in numerous ways, one of which is insulin resistance, which has been linked to brain amyloid accumulation (Craft, 2007). In the current paper, first author T. Matsuzaki and colleagues homed in on insulin resistance, and looked at whether it correlated with specific brain lesions in AD, namely amyloid plaques and neurofibrillary tangles. “It’s a tight study,” said Suzanne de la Monte, Brown University, Providence, Rhode Island, who noted that many prior investigations examined the link between diabetes and dementia without addressing underlying mechanisms, or studied associations between diabetes and cognitive impairment without focusing in particular on AD.

Matsuzaki’s team investigated 135 elderly who were among several thousand enrolled in a long-term prospective cohort study in the southern Japanese town of Hisayama. The seniors in the analyzed subset underwent autopsy after they died around the turn of the twenty-first century. They had taken an oral glucose tolerance test as part of a clinical exam 10-15 years prior. They were demographically similar to others in the cohort who were not autopsied, decreasing the possibility of selection bias. The participants were cognitively normal at the time of their glucose test. By the end of the study, only 21 of the 135 had developed AD-type dementia, whereas 88 had plaque pathology.

By three different readouts (i.e., two-hour post-load plasma glucose, fasting insulin, and a homeostatic index called HOMA-IR), insulin resistance correlated with the presence of neuritic plaques, as judged by CERAD criteria, in autopsy brain specimens. Furthermore, the relationship between insulin resistance and plaque pathology was dose dependent. Those with more severe glucose and insulin measures had correspondingly increased risk for plaque pathology. Plaque risk shot up even further in ApoE4 carriers, about 15- to 30-fold, relative to non-carriers with similar extent of insulin resistance.

Methodologically, the study draws praise for its attention to a slew of other factors that could have accounted for the findings, said Konrad Talbot, University of Pennsylvania School of Medicine, Philadelphia, in an interview with ARF. The association between diabetes-related measures and plaque pathology held after the researchers controlled for age, gender, and other potential confounding variables including blood pressure, body-mass index, cholesterol, and exercise—even after excluding the 21 people who developed AD dementia by the time they died. The data suggest “insulin resistance can affect formation of plaques before you see any clinical symptoms,” Talbot said. “It could be something happening very early on.”

Perhaps just as striking as the link between insulin resistance and plaques was the complete lack of relationship with neurofibrillary tangles. Whether or not people had tau pathology, as judged by Braak staging, did not seem to correlate with the magnitude of their hyperglycemia or hyperinsulinemia.

The data run counter to an earlier study of Japanese-American men that found type 2 diabetics had increased risk for both plaques and tangles (Peila et al., 2002). However, the lack of correlation with tau pathology appears consistent with a large-scale study on insulin signaling that Talbot is preparing to submit for publication (see ARF related news story). In that study, aberrant phosphorylation of insulin receptor substrate 1 (IRS1), a key insulin signaling abnormality found in AD patients, is absent in tauopathies that lack plaque pathology, Talbot said.

The current paper also leaves unaddressed another issue, and that is, Which comes first—insulin resistance or plaque formation? The diabetes-related factors can presumably act upstream of the AD pathological cascade, as the authors suggest. A recent study bears out this idea by showing that insulin signaling can protect neurons from Aβ toxicity (De Felice et al., 2009 and ARF related news story). However, it’s also possible that some upstream event could drive both (see ARF Live Discussion), Talbot said, by causing brain neurons to become insulin resistant and, at the same time, more susceptible to Aβ accumulation. In his view, the methods used in the current study make it particularly challenging to make the call. Pathological reads were done some 10-15 years after assessment of insulin resistance, and yet amyloid plaques are known to develop decades in advance of symptoms.

Ultimately, the answer to whether insulin resistance can cause AD may lie in ongoing clinical trials testing insulin sensitizers in patients with mild cognitive impairment, suggested Jose Luchsinger, Columbia University, New York, in an editorial accompanying the Japanese study.—Esther Landhuis


  1. Careful reading of a very similar work (see Peila et al., 2002) reveals that there is limited new data in this paper by Matsuzaki et al. Like the latter paper, Peila et al. investigated a large population of individuals (i.e., Japanese-Americans) participating in a longitudinal study (i.e., the Honolulu-Asia aging study) on diabetic factors in aging. Peila et al. specifically tested the hypothesis that 1) diabetes is a risk factor for dementia generally and for Alzheimer disease (AD) and vascular dementia in particular; 2) that such risk is enhanced in carriers of at least one copy of the ApoE4 allele; and 3) that diabetes alone or in combination with an ApoE4 allele is associated with increased neuritic plaques and neurofibrillary tangles in the cerebral cortex and/or hippocampus. The last two hypotheses are the focus of this news study, which is in most respects a replication of the study by Peila et al. (2002) on a different population of Japanese origin.

    In a few respects, however, Matsuzaki et al. provide some new information. First, they establish that only relatively high levels of insulin resistance (measured in three different ways, see their Fig. 1) are associated with neuritic plaque densities in the brain. Second, this association is present even in cases without AD. Third, unlike Peila et al., they found that elevated insulin resistance is associated with elevated neuritic plaque densities even in the absence of ApoE4 alleles (see Matsuzaki et al., Fig. 2), though the presence of even one such allele greatly increases the association of insulin resistance and plaque densities as found by Peila et al. (2002).

    Why did Matsuzaki et al. find an association of high insulin resistance with neuritic plaques, but not with neurofibrillary tangles?

    Peila et al. (2002), by studying a large number of autopsied cases (216 vs. 135 in Matsuzaki et al.) tested closer to the time of death (Unger et al., 1989) and insulin-sensitive glucose transporter 4 in the brain (El-Messari et al., 1998).

    But Matsuzaki et al. provide no data on the density of plaques or tangles in specific brain areas. It is implied that they calculated a composite score for neuritic plaques and for neurofibrillary tangles from semi-quantitative scores (+ to +++) in each of at least 12 brain areas in the cerebral cortex, limbic system, basal ganglia, basal forebrain, thalamus, midbrain, and pons. It is not explained how they calculated such composite semi-quantitative scores. With such scores, the authors only tested the hypothesis that global levels of neuritic plaques and neurofibrillary tangles are strongly associated with high insulin resistance. Their data is insufficient to detect weaker, but significant associations globally or in specific brain areas. Detection of such associations requires more objective quantitative measures of neuritic plaques (amyloid load) and neurofibrillary tangle densities in each brain area sampled. Peila et al. also used only semi-quantitative methods, but they present data for specific brain areas.

    Matsuzaki et al. have consequently only shown that the association of high insulin resistance with global levels of neuritic plaques is stronger than with neurofibrillary tangles. This is not inconsistent with Peila et al., who did not assess global levels of those pathologies. It is possible that there is an association with neurofibrillary tangles in certain brain areas. The global findings are, nevertheless, consistent with our unpublished data that a major hippocampal abnormality in AD associated with impaired insulin signaling (i.e., elevated serine phosphorylation of insulin receptor substrate 1 [IRS-1]: see Talbot et al., Alzheimer's and Dementia 2, suppl. 1: S54, 2006) is not elevated with neurofibrillary tangles without an elevation in neuritic plaques (e.g., in corticobasal degeneration [CBD]; Talbot et al., in preparation).

    What might explain the association between high insulin resistance and neuritic plaque levels?

    One of the more interesting and novel findings of Matsuzaki et al., noted above, was that high insulin resistance was associated with higher global neuritic plaque levels even in the absence of AD. This is consistent with the view, gaining increasing acceptance, that insulin resistance precedes clinical onset of that disorder, a view supported by many epidemiological studies finding that type 2 diabetes (characterized by insulin resistance) is a clear risk factor for AD (Biessels et al., 2006). It is not widely appreciated, however, that the risk goes beyond those with a history of diabetes: it is estimated that among those 60-74, in which the incidence of AD shows a clear increase, 66.7 percent are insulin resistant (Cowie et al., 2009). Indeed, more than 80 percent of the AD cases studied by Janson et al. (Janson et al., 2004) were insulin resistant, either in a diabetic or a pre-diabetic state. That the brain itself is often insulin resistant in AD, even without a history of diabetes, is suggested by our finding that about 90 percent of such AD cases we have studied postmortem display large numbers of hippocampal neurons with abnormally phosphorylated insulin signaling molecules consistent with insulin resistance (Talbot et al., in preparation). Many of these abnormalities were seen to a lesser extent in a type of mild cognitive impairment likely to progress to AD. The abnormalities are very highly correlated with memory impairments of the cases studied and are consistent with reduced levels of insulin signaling found in our ex vivo stimulation experiments on hippocampal tissue from AD cases. These findings are being prepared for submission next month.

    It should not be assumed, as Matsuzaki et al. do, that insulin resistance necessarily leads to neuritic plaque formation. The available data are only correlational and thus leave open the possibility that both phenomena are the consequence of other factors. Among these are soluble oligomers of amyloid-β (Aβ), which are better able to account for the cognitive impairments of AD than plaques, due to the affinity of such oligomers for synapses. Aβ oligomer levels rise even in mild cognitive impairment and are generated before plaques appear. They are known to trigger all the mechanisms that lead to inactivation of the insulin receptor and serine phosphorylation of IRS-1. The latter effect is mediated (at least in part) by Aβ oligomer activation of a stress molecule known as JNK, activation of which is highly correlated with levels of IRS-1 serine phosphorylation in hippocampal neurons of our AD cases.

    All of this points to the importance of finding the roots of neuronal insulin resistance in AD cases and mild cognitive impairments leading to AD, because this is an important risk factor suggesting promising new treatments for the disorder.


    . Type 2 diabetes, APOE gene, and the risk for dementia and related pathologies: The Honolulu-Asia Aging Study. Diabetes. 2002 Apr;51(4):1256-62. PubMed.

    . Immunocytochemical localization of the insulin-responsive glucose transporter 4 (Glut4) in the rat central nervous system. J Comp Neurol. 1998 Oct 5;399(4):492-512. PubMed.

    . Risk of dementia in diabetes mellitus: a systematic review. Lancet Neurol. 2006 Jan;5(1):64-74. PubMed.

    . Full accounting of diabetes and pre-diabetes in the U.S. population in 1988-1994 and 2005-2006. Diabetes Care. 2009 Feb;32(2):287-94. PubMed.

    . Increased risk of type 2 diabetes in Alzheimer disease. Diabetes. 2004 Feb;53(2):474-81. PubMed.

    View all comments by Konrad Talbot
  2. Insulin is not required for glucose to enter brain cells. Alzheimer's disease, then, is not a type 3 diabetes. Later in the disease, the oxidation of glucose transporters does reduce glucose levels in brain cells (Mark et al. 1997).

    Insulin resistance does contribute to Alzheimer's disease as it initially increases the amount of glucose in the brain because the glucose is not being "absorbed" in other parts of the body (Jacob et al., 2002). This results in high levels of myo-inositol, the precursor molecule to Alzheimer's disease (Hauser and Finelli 1963; Huang et al., 1995) and to the activation of phospholipase C gamma-gamma, an enzyme implicated in triggering Alzheimer's disease, primarily via the platelet-derived growth factor receptor (Dequin et al., 1998; Okuda et al., 1996). Polyphenols in various fruits, vegetables, and spices, and polyunsaturated fats such as fish oil partially inhibit phospholipase C gamma and thus provide some protection against the disease (Godichaud et al. 2006; Kang et al., 2003; Sanderson and Calder, 1998; Valente et al., 2009).


    . Regional levels of brain phospholipase Cgamma in Alzheimer's disease. Brain Res. 1998 Nov 16;811(1-2):161-5. PubMed.

    . The grape-derived polyphenol resveratrol differentially affects epidermal and platelet-derived growth factor signaling in human liver myofibroblasts. Int J Biochem Cell Biol. 2006;38(4):629-37. Epub 2005 Nov 28 PubMed.


    . High brain myo-inositol levels in the predementia phase of Alzheimer's disease in adults with Down's syndrome: a 1H MRS study. Am J Psychiatry. 1999 Dec;156(12):1879-86. PubMed.

    . Rosmarinic acid inhibits Ca2+-dependent pathways of T-cell antigen receptor-mediated signaling by inhibiting the PLC-gamma 1 and Itk activity. Blood. 2003 May 1;101(9):3534-42. PubMed.

    . Increased production of PDGF by angiotensin and high glucose in human vascular endothelium. Life Sci. 1996;59(17):1455-61. PubMed.

    . Dietary fish oil appears to prevent the activation of phospholipase C-gamma in lymphocytes. Biochim Biophys Acta. 1998 Jun 15;1392(2-3):300-8. PubMed.

    . A diet enriched in polyphenols and polyunsaturated fatty acids, LMN diet, induces neurogenesis in the subventricular zone and hippocampus of adult mouse brain. J Alzheimers Dis. 2009;18(4):849-65. PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Diabetes-Insulin Roundup: Dementia Connection Grows Stronger, Part 2
  2. Peptide Brace Against AD—Insulin, Neuropeptide Y Tame Aβ Toxicity

Webinar Citations

  1. Is Alzheimer's a Type 3 Diabetes?

Paper Citations

  1. . Full accounting of diabetes and pre-diabetes in the U.S. population in 1988-1994 and 2005-2006. Diabetes Care. 2009 Feb;32(2):287-94. PubMed.
  2. . The relationship between type 2 diabetes and cognitive dysfunction: longitudinal studies and their methodological limitations. Eur J Pharmacol. 2004 Apr 19;490(1-3):169-75. PubMed.
  3. . Diabetes mellitus and risk of Alzheimer disease and decline in cognitive function. Arch Neurol. 2004 May 1;61(5):661-6. PubMed.
  4. . Insulin resistance and Alzheimer's disease pathogenesis: potential mechanisms and implications for treatment. Curr Alzheimer Res. 2007 Apr;4(2):147-52. PubMed.
  5. . Type 2 diabetes, APOE gene, and the risk for dementia and related pathologies: The Honolulu-Asia Aging Study. Diabetes. 2002 Apr;51(4):1256-62. PubMed.
  6. . Protection of synapses against Alzheimer's-linked toxins: insulin signaling prevents the pathogenic binding of Abeta oligomers. Proc Natl Acad Sci U S A. 2009 Feb 10;106(6):1971-6. PubMed.

External Citations

  1. HOMA-IR

Further Reading


  1. . Type 2 diabetes, APOE gene, and the risk for dementia and related pathologies: The Honolulu-Asia Aging Study. Diabetes. 2002 Apr;51(4):1256-62. PubMed.
  2. . Different patterns of cerebral injury in dementia with or without diabetes. Arch Neurol. 2009 Mar;66(3):315-22. PubMed.
  3. . Protection of synapses against Alzheimer's-linked toxins: insulin signaling prevents the pathogenic binding of Abeta oligomers. Proc Natl Acad Sci U S A. 2009 Feb 10;106(6):1971-6. PubMed.

Primary Papers

  1. . Insulin resistance is associated with the pathology of Alzheimer disease: the Hisayama study. Neurology. 2010 Aug 31;75(9):764-70. PubMed.