Article Text

Download PDFPDF

Research paper
Plasma homocysteine and MTHFRC677T polymorphism as risk factors for incident dementia
  1. Andrew H Ford1,
  2. Leon Flicker2,
  3. Helman Alfonso1,
  4. Graeme J Hankey3,
  5. Paul E Norman4,
  6. Frank M van Bockxmeer5,
  7. Osvaldo P Almeida1
  1. 1WA Centre for Health & Ageing, Centre for Medical Research & School of Psychiatry and Clinical Neurosciences, University of Western Australia, Perth, Western Australia, Australia
  2. 2WA Centre for Health & Ageing, Centre for Medical Research & School of Medicine and Pharmacology, University of Western Australia, Perth, Western Australia, Australia
  3. 3School of Medicine and Pharmacology, University of Western Australia, Perth, Western Australia, Australia
  4. 4School of Surgery, University of Western Australia, Perth, Western Australia, Australia
  5. 5School of Pathology & Laboratory Medicine, University of Western Australia, Perth, Western Australia, Australia
  1. Correspondence to Dr Andrew H Ford, School of Psychiatry & Clinical Neurosciences (M573), University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia;{at}


Background Elevated total plasma homocysteine (tHcy) has been associated with increased risk of dementia. The C677T polymorphism of the 5,10-methylenetetrahydrofolate reductase gene (MTHFR) increases tHcy and provides a means of studying the association between tHcy and dementia while not being as susceptible to the common biases and confounding of observational studies. The authors designed this longitudinal study to determine if high tHcy and the MTHFR C677T polymorphism increase the risk of incident dementia among older men.

Methods The authors studied 4227 men aged 70–89 years from the Health in Men Study cohort and established the diagnosis of dementia (International Classification of Diseases—10th edition) using morbidity and mortality records. Information on tHcy, MTHFR gene status, lifestyle and clinical variables were obtained using postal and face-to-face assessments.

Results 230 men (5.4%) developed dementia during the mean follow-up period of 5.8±1.6 years (range 0.1–8.2 years). The hazard of dementia increased with a doubling of tHcy concentration (adjusted HR 1.48, 95% CI 1.10 to 2.00) and was higher in men with tHcy >15 μmol/l (adjusted HR 1.36 95% CI 1.03 to 1.81, p=0.032). Men with the TT genotype had a HR of dementia of 1.25 (95% CI 0.81 to 1.92).

Conclusions The results of this prospective study are consistent with a causal link between high tHcy and incident dementia, but the study lacked power to determine an effect of the MTHFR genotype.

  • Dementia
  • Alzheimer disease
  • genetics

Statistics from


Homocysteine is an amino acid derived from the essential amino acid methionine. High total plasma homocysteine (tHcy) has been associated with numerous medical conditions,1–3 including dementia.4

A handful of clinical trials have investigated the association between tHcy lowering treatment and cognitive decline in non-demented older people, and their results have been equivocal.5–10 Data derived from cross-sectional, case-control and cohort studies have found that high tHcy is associated with dementia,4 11 12 although others have not reported such an association.13 14 The inability of some of these studies to take into account the possibility of reverse causality and confounding creates uncertainty about the validity of the reported associations.15

Recently, investigators have used the principles of Mendelian randomisation to establish causal links between specific proteins and diseases. The term describes the random assortment of alleles that occurs at the time of gamete formation and whose genetic variants (polymorphisms) lead to greater or lower expression of the relevant protein (eg, tHcy) throughout people's lives. This is yet another informative approach to investigate the possible causal link between high tHcy and dementia that is less subject to confounding or reverse causality.16 Homocysteine metabolism is dependent on a number of enzymes and vitamin cofactors, including the 5,10-methylenetetrahydrofolate reductase (MTHFR) enzyme involved in the remethylation of homocysteine to methionine. A common genetic polymorphism of the MTHFR gene (C677T), which is present in around 10% of the population (in the homozygous form), reduces MTHFR activity and increases basal tHcy by about 20%.17 The outcomes of Mendelian randomisation studies, however, are predicated on the concept that the genetic polymorphism has no other effect than its effect on the concentration of homocysteine.18

We designed this prospective study to determine if high tHcy is associated with incident dementia once potential confounding is taken into account and to clarify if the MTHFR C677T polymorphism increases the hazard of dementia in this population.


Study design

This is a prospective cohort study of men aged 70–89 years at time of enrolment.

Study population

The Health in Men Study originated from a population-based trial of screening for abdominal aortic aneurysm in Perth, Western Australia that began in 1996. Details about the design of the study and the recruitment of participants have been described elsewhere.19 Briefly, between 2001 and 2004, 4248 men returned for reassessment and completed a comprehensive health questionnaire and physical examination and donated a blood sample for the measurement of tHcy and for the extraction of DNA. Men were community-dwelling Australian citizens initially recruited via the electoral roll (voting being compulsory in Australia).

These men were followed via health records and death certificates obtained from the Western Australian Data Linkage System (WADLS) ( until the end of September 2009. WADLS brings together all death records, acute hospital admissions, hospital movements, cancer registry, as well as psychiatric outpatient contacts for all residents of Western Australia since 1980.20 We excluded from our analyses 21 men who had a recorded diagnosis of dementia in WADLS before the assessment for Health in Men Study (prevalent cases), leaving 4227 participants at the time of entry into the study. The Human Research Ethics Committee of the University of Western Australia approved this study protocol, and all men offered written informed consent to participate. The research was conducted in accordance with the Declaration of Helsinki recommendations for the conduct of clinical research.

Outcome of interest: diagnosis of dementia

The diagnosis of dementia retrieved from WADLS followed the International Classification of Diseases (ICD) coding system: ICD-10 codes F00 (Dementia in Alzheimer disease), F01 (Vascular Dementia), F03 (Unspecified Dementia) and G30 (Alzheimer disease) and ICD-9 codes 290.0-290.4 and 331.0. We considered that the date of onset of dementia corresponded to the first date when the diagnosis of dementia appeared in WADLS. Otherwise, participants were censored at the date of their death or on 30 September 2009, whichever occurred first.


We collected fasting blood samples from participants at baseline between 08:30 and 11:00. Plasma and serum were separated from blood cells within 1 h of collection and stored at –80°C until assayed. tHcy concentration was determined by reverse-phase high-performance liquid chromatography after treatment with tributylphosphine, deproteinisation and fluorogenic derivatisation using the method of Araki and Sako.21 The coefficient of variation ranged from 3% to 7%. Men with tHcy >15 μmol/l were considered to have elevated tHcy.22

Genomic DNA was isolated from nucleated blood cells via a Triton X-100 method and the C677T polymorphism determined using PCR and HinfI restriction enzyme digestion as previously described.23 HinfI digestion (1.5-U/25 μl reaction mixture) was performed directly in the PCR tube at 37°C for 4 h before analysis of restriction fragments by polyacrylamide gel electrophoresis (12% T, 3.3% C).

Other measured factors

We recorded the age of participants at the time of the 2001–2004 assessment (in years), as well as the highest level of education they had achieved. Age was categorised into four groups: 70–74 years, 75–79 years, 80–84 years and 85+ years. We dichotomised participants' education at the level of high school education or better. We asked participants about their lifestyle, including smoking, and classified them into those who had never smoked, ex-smokers and current smokers. In addition, we retrieved information about physical activity and alcohol use collected during the 1996–1999 assessments. Men recorded the number of minutes of “vigorous exercise that makes you breathe harder or puff and pant (such as fast walking, jogging, aerobics, vigorous swimming) in a usual week” and of “non-vigorous exercise that does not make you breathe harder or puff or pant in a usual week (such as slow walking, slow cycling, Tai Chi or yoga)”. Men who reported a total of 150 min or more of vigorous or non-vigorous activity were considered physically active.24 Finally, participants recorded the number of standard drinks of alcohol that they consumed each day of the week in a usual week. We considered that men who drank 14 or more standard drinks per week were risky drinkers (NHMRC Australian Guidelines to Reduce Health Risks from Drinking Alcohol:

The men completed a self-reported health questionnaire ascertaining the presence of the following medical conditions (treated or untreated): hypertension, diabetes, ischaemic heart disease (angina or past history of myocardial infarction) and strokes. Finally, we used standard procedures to measure participants' blood pressure (within 2 mm Hg), height (in centimetres) and weight (in kilograms, to 0.2 kg). We used the last two measures to calculate the body mass index (BMI) in kg/m2.

Other biochemical analyses

We measured serum glucose, thyroid stimulating hormone, total cholesterol, low-density lipoprotein, high-density lipoprotein and serum triglycerides with a Roche Hitachi 917 analyser. Serum triglycerides were estimated by the Friedewald equation.25 Serum creatinine was measured by Jaffe kinetic reaction on a Roche 917 analyser and the estimated glomerular filtration rate was calculated using the Cockcroft–Gault equation: ((140−age)×wt(kg))/(plasma creatinine×0.8136).

Statistical analysis

We used the statistical software Stata 11.1 (StataCorp 2009, Texas, USA) to manage and analyse our data. We initially compared men with high tHcy (>15 μmol/l) to those with normal tHcy (≤15 μmol/l). We used descriptive statistics to summarise the data (mean and SD, median and IQR or count and proportions). We used Student t test for variables with a normal distribution, and Mann–Whitney U (z statistic) tests were used to compare variables without a normal distribution. Pearson's χ2 was used to analyse categorical variables. tHcy values were log transformed as these were skewed.

Cox proportional-hazards regression models were used to examine the relationship between incident dementia and high tHcy, as well as the association between incident dementia and the MTHFR polymorphism (CC homozygotes were the reference group). A univariate analysis was first performed on all relevant independent variables. We investigated the association between tHcy and dementia in three different ways: as a doubling of tHcy concentration (by dividing the natural logarithm of tHcy by the natural logarithm of 2), as quartiles (lowest quartile used as reference) and as according to whether tHcy was >15 μmol/l. Variables that were significantly associated (p<0.05) with incident dementia in the univariate analysis were included in the multivariate model, provided they had an effect of at least 10% on the hazard of dementia.

Poisson regression was used to calculate the dementia incident rates (IR) and incident rate ratios (IRR), and this was represented graphically. These were also adjusted for relevant covariates. Finally, we repeated our analyses after excluding men who received the diagnosis of dementia within 1 and 2 years of the index assessment, as they could represent prevalent cases of undiagnosed dementia.


Population characteristics

The baseline characteristics of our 4227 men are shown in table 1. Men with high tHcy were 1.5 years older (t=−11.89, p<0.001), and more likely to be past smokers (χ2=17.51, p<0.001) than men with tHcy ≤15 μmol/l. Compared with men with tHcy ≤15 μmol/l, a larger proportion of men with high tHcy reported medical comorbidities such as stroke (14.3% vs 8.8%, χ2=25.49), ischaemic heart disease (31.3% vs 23.6%, χ2=24.45) and hypertension (54.8% vs 44%, χ2=36.75). These men had lower mean high-density lipoprotein (t=5.27, p<0.001) and higher triglyceride (z=−5.03, p<0.001) and thyroid stimulating hormone concentrations (z=−2.28, p=0.023) and were also more likely to have died during follow-up (28.9% vs 17.2%, χ2=68.34, p<0.001).

Table 1

Population characteristics according to tHcy status

Dementia and plasma homocysteine

Two hundred and thirty men (5.4%) were diagnosed with dementia during the mean follow-up period of 5.8±1.6 years (range 0.1–8.2 years). Men diagnosed with dementia had a mean baseline tHcy concentration of 13.9 μmol/l (SD 1.4 μmol/l) compared with 12.6 μmol/l (SD 1.4 μmol/l) in men remaining free of dementia (t=−4.65, p<0.001, after natural logarithmic transformation). The hazard of dementia increased with increasing age, and such a rise was particularly pronounced among those with high tHcy (figure 1). However, we found no statistical evidence of an interaction between high tHcy and age stratum (HR for the interaction: 1.02, 95% CI 0.75 to 1.38). Participants with a history of stroke (HR 2.23, 95% CI 1.59 to 3.13) and of ischaemic heart disease (HR 1.62, 95% CI 1.22 to 2.15) had a higher hazard of dementia as well (table 2). Age, history of stroke and ischaemic heart disease were found to influence the relationship between tHcy and incident dementia and were therefore included in the multivariate model. Body mass index was not included in the multivariate analysis as the inclusion of this as a covariate did not have much of an effect on the hazard of dementia, that is, <10%.

Figure 1

Dementia incidence. The figure depicts the differing annual rates of dementia by age in men according to whether they had elevated tHcy (>15 μmol/l). The adjusted incidence rate ratio of dementia associated with high tHcy was 1.38 (95% CI 1.04 to 1.83, p=0.026; adjusted for age, and history of ischaemic heart disease and stroke).

Table 2

Univariate association with incident dementia

The hazard of dementia nearly doubled (HR 1.94, 95% CI 1.51 to 2.50, p<0.001) with the doubling of tHcy concentration, and this association remained significant (HR 1.48, 95% CI 1.10 to 2.00) after adjustment for confounding (table 3). The adjusted hazard of dementia for men with tHcy>15 μmol/l was 1.36 (95% CI 1.03 to 1.81, p=0.032). The crude IRR of dementia associated with high tHcy was 1.65 (95% CI 1.26 to 2.16, p<0.001). The IRR decreased to 1.38 (95% CI 1.04 to 1.83, p=0.026) after adjustment for age and a history of stroke and ischaemic heart disease.

Table 3

HRs (95% CI) of dementia according to tHcy and gene status

The crude hazard of dementia was 83% higher for men in the highest (>15.1 μmol/l) than in the lowest quartile (≤10.3 μmol/l) of tHcy (HR 1.83, 95% CI 1.26 to 2.67, p=0.002). This difference was no longer significant after adjustment for age, stroke and ischaemic heart disease (HR 1.20, 95% CI 0.80 to 1.79, p for trend=0.201). These data are summarised in table 3. There was no evidence of interaction between tHcy and the MTHFR C677T polymorphism on the crude and adjusted risk of dementia (data not shown).

Analysis of potentially undiagnosed prevalent dementia cases

Nine men were diagnosed with dementia within a year of their baseline assessment. We re-ran all analyses after excluding these men, as they could be considered unidentified prevalent cases of dementia. The adjusted dementia hazard associated with the doubling of tHcy did not differ much from the hazard in the complete sample (HR 1.47, 95% CI 1.08 to 1.99, p=0.015). Similarly, there was minimal difference in men with tHcy>15 μmol/l (adjusted HR 1.37, 95% CI 1.03 to 1.84, p=0.031) and in the highest quartile of tHcy (adjusted HR 1.22, 95% CI 0.81 to 1.85, p=0.339). Twenty-eight additional men were excluded when the censoring of cases of dementia was extended to 2 years. Again, the associations reported above remained significant and largely unchanged in terms of effect size (data not shown).

MTHFR genotype and dementia risk

Four hundred and twenty-one men (11.2%) were homozygous for the C677T MTHFR polymorphism with a mean tHcy of 13.1 μmol/l compared with 12.4 μmol/l for those with the CC genotype (t=−3.25, p=0.001; after natural logarithmic transformation). Compared to men with the CC genotype, CT participants had a dementia HR of 1.03 (95% CI 0.76 to 1.39), whereas TT men had a HR of dementia of 1.25 (95% CI 0.81 to 1.92), p value for trend=0.403. We estimated that the study would need 5637 men (631 with the MTHFR TT genotype) to declare as significant an association of modest effect size (HR=1.25) between the MTHFR C677T polymorphism and incident dementia.


The findings of this prospective cohort study indicate a significant association between tHcy and the risk of incident dementia in older men. The hazard of dementia increased by 48% with a doubling of tHcy concentration. Men with tHcy concentrations greater than 15 μmol/l were also at greater risk of dementia during follow-up. The observed associations were independent of risk factors commonly associated with dementia such as age, education, smoking, alcohol use and vascular risk factors. Finally, our calculations indicate that the effect of the MTHFR TT genotype on the risk of dementia was small, and our study lacked power to declare as significant such an association.

This study has a number of merits and weaknesses worth mentioning. Our findings were derived from a large cohort of community-representative elderly men at risk of developing dementia. These men had a wide range of tHcy concentrations and over a quarter had hyperhomocysteinemia. We excluded prevalent cases of dementia and potentially undetected prevalent cases through repeating the analysis on newly diagnosed men. The men were followed up an average of over 5 years, and we controlled our analyses for a number of factors that may have confounded the relationship between tHcy and dementia.

We concede, however, that the diagnosis of dementia relied on health records and death certificates and, consequently, under-diagnosis is likely.26 This could explain the slightly smaller proportion of incident cases of dementia in our study (5.4%) compared with others,4 13 27 but this would not undermine the specificity of our findings (few false-positive cases would be recorded in WADLS). Elevated tHcy is also associated with cardiovascular and cerebrovascular disease, and the use of hospital morbidity data could thus have introduced bias (Berkson's bias) in favour of the association between incident dementia and tHcy.

We were also unable to account for loss to follow-up through reasons other than death (eg, men moving interstate/overseas), although moves to and from Western Australia in this age group are exceedingly rare (Australian Bureau of Statistics; The restriction of the sample to men may affect the generalisability of our findings, but we are not aware of any evidence suggesting that tHcy has a gender-specific effect on cognition. Moreover, our study relied on a single estimation of tHcy at baseline, and variations in the concentration of tHcy over time could not be taken into account in the analyses. Nevertheless, current evidence suggests that a single baseline measurement may underestimate the risk of the outcome because of regression dilution bias,28 and as a result, our findings might be conservative. Finally, we did not have access to data on B vitamin concentrations or APOE genotype, and these factors could conceivably have confounded some of the associations that we observed.

Elevated tHcy increases cerebrovascular disease burden and its neurotoxic activation of the N-methyl-d-aspartate receptor may further contribute to neuronal death.29 There is also some evidence that homocysteine facilitates the aggregation of the β-amyloid peptide and the phosphorylation of the microtubule-associated protein τ, which have been implicated in the aetiology of Alzheimer disease.30 31 Moreover, high tHcy accelerates cell ageing,32 and methionine-deficient diets extend the lifespan of rodents by 30%.33 Taken together, these findings indicate that the association between high tHcy and dementia is biologically plausible.

Seshadri and colleagues4 evaluated 1092 participants recruited from the Framingham cohort. They were followed up for a median period of 8 years, and there were 111 incident cases of dementia. The authors reported that the RR of dementia was 1.4 (95% CI 1.1 to 1.9) for each 1 SD in the log-transformed homocysteine value after adjustment for age, sex, plasma levels of B vitamins, APOE genotype, education, alcohol use and other vascular risk factors. Further data from the same cohort34 indicates that increasing age enhances the association between high tHcy and cognition. Although our findings seem consistent with such a possibility (figure 1), we found no evidence of a statistical interaction between age grouping and high tHcy on dementia risk. This may be due to the fact that both plasma tHcy and dementia increase with increasing age.

Similarly, an Italian study27 reported the outcome of 816 elderly participants followed for an average of 4 years. Dementia developed in 112 subjects, and those with elevated tHcy (>15 μmol/l) had more than twice the risk compared to subjects with tHcy ≤15 μmol/l (HR 2.08, 95% CI 1.31 to 3.30, p=0.002). These results are compatible with a recent randomised controlled trial of homocysteine-lowering B vitamin therapy for patients with mild cognitive impairment. The authors reported that participants in the B vitamin arm experienced a mean rate of brain atrophy of 0.76% per year (95% CI 0.63 to 0.90) compared to 1.08% (95% CI 0.94% to 1.22%) in the placebo group (p=0.001).35

A number of studies have reported an association between dementia and the C677T MTHFR polymorphism,36 37 but results have not been consistent.38 39 A meta-analysis of genetic association studies found a small but significant association between the T allele and Alzheimer disease (OR 1.13, 95% CI 1.04 to 1.23) supporting the notion that our study was not sufficiently powered to declare this association as significant.40 In our cohort, men with the TT genotype had 41% increased risk of having high tHcy (OR 1.41, 95% CI 1.11 to 1.79), and men with high tHcy had a 36% increased risk of dementia (HR 1.36, 95% CI 1.03 to 1.81). This would equate to the TT genotype increasing the risk of dementia by about 15%. This is a small effect that would require several thousand individuals in this cohort to investigate the possible causal link between the MTHFR C677T polymorphism and dementia. Our study was clearly underpowered to establish such an association. In addition, it is worth noting that our investigation only included older men, and consequently, the effect of the MTHFR C677T polymorphism may have been over-estimated, given that tHcy increases with age and is higher in men than in women.41

In conclusion, our results are important because they confirm the findings from previous observational studies linking high tHcy to dementia and are the first to explore this link by using the principles of Mendelian randomisation. The association between high tHcy and dementia is also biologically plausible29–33 and is consistent with the hypothesis that high tHcy causes dementia. To the best of our knowledge, no randomised trials have attempted to prevent the onset of dementia by reducing tHcy in people at risk, although evidence from a trial of cognitive outcomes in people with hyperhomocysteinemia seems to support such an effect.6 We would argue that the best way to establish whether tHcy-lowering therapy can indeed prevent dementia would be by means of a large randomised trial of older people with high tHcy. A trial of this type would require large numbers of participants to be supplemented over an extended period of time, given that B vitamins, if effective, are likely to produce modest effects that take a prolonged interval to become apparent.42 At this point in time, there is not enough evidence to support the use of tHcy-lowering therapies to prevent dementia.


The authors would like to thank all the men who participated in the trial.



  • Funding The Health in Men Study is supported by National Health and Medical Research Council of Australia project grants (964145, 139093, 403963 and 455811) with additional funding from the National Heart Foundation and the Western Australian Health Promotion Foundation (Healthway).

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the University of Western Australia.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.