Article Text


The androgen receptor CAG repeat and serum testosterone in the risk of Alzheimer’s disease in men
  1. D J Lehmann1,
  2. E Hogervorst1,
  3. D R Warden1,
  4. A D Smith1,
  5. H T Butler2,
  6. J Ragoussis2
  1. 1Oxford Project to Investigate Memory and Ageing (OPTIMA), Radcliffe Infirmary, Oxford and University Department of Pharmacology, Mansfield Road, Oxford OX1 3QT, UK
  2. 2Genomics Laboratory, Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford
  1. Correspondence to:
 D J Lehmann;

Statistics from

We recently reported1 that the glutamine (CAG) repeat polymorphism of exon 1 of the androgen receptor was associated with Alzheimer’s disease in men from the Oxford region. If this androgen receptor polymorphism indeed affects Alzheimer’s disease risk rather than being in linkage disequilibrium with the true risk factor, then we should see the effect through androgen receptor actions. The association should, for instance, be influenced by ligands of the androgen receptor. Androgen receptor isoforms are the main, perhaps sole, receptors for the principal mammalian androgens, testosterone and 5α-dihydrotestosterone. On binding ligand, the receptor moves from the cytoplasm to specific compartments of the nucleus (“nuclear foci”), where it interacts with coactivators and corepressors in the cell specific regulation of numerous genes. We therefore looked for an interaction in Alzheimer’s disease risk of the androgen receptor CAG polymorphism with serum concentrations of total testosterone.

We studied 207 elderly men, 79 with sporadic Alzheimer’s disease (mean (SD) onset age, 70 (9) years) and 128 controls (mean age, 75 (10) years), all from the cohort of the Oxford Project to Investigate Memory and Ageing (OPTIMA). Of the 79 Alzheimer cases, 45 were necropsy confirmed as having Alzheimer’s disease by CERAD criteria (40 “definite” and five “probable”) and 34 were diagnosed as “probable Alzheimer’s disease” by NINCDS-ADRDA criteria. All 128 controls were without detectable cognitive dysfunction and with CAMCOG scores of more than 80. Serum concentrations of total testosterone and androgen receptor allele lengths were determined as previously described.12 Testosterone concentrations were divided into tertiles (69 subjects in each tertile): the upper tertile was >19 nmol/l, the middle between 13 and 19 nmol/l, and the lower ⩽13 nmol/l. Short androgen receptor alleles were ⩽20 CAG repeats. Logistic regression analysis was by R.


As expected, short androgen receptor alleles1 and low testosterone2 were associated with Alzheimer’s disease. Controlling for age, odds ratios of Alzheimer’s disease were 2.1 (95% confidence interval, 1.1 to 4.1) (p = 0.025) for short androgen receptor alleles, and 2.2 (1.01 to 4.6) (p = 0.04) for the lower v the upper tertile of testosterone. There was a significant trend in the association with Alzheimer’s disease by testosterone tertile (χ2trend, p<0.05), indicating a possible dose related effect of testosterone.

Table 1 shows the unadjusted odds ratios of Alzheimer’s disease for each combination of long or short androgen receptor alleles with each testosterone tertile, taking long alleles with upper tertile testosterone as reference. The odds ratio for short androgen receptor alleles with lower tertile testosterone was 4.2 (1.4 to 13) (p = 0.01). Combining the two at risk subgroups—that is, carriers of short androgen receptor alleles with middle or lower tertile testosterone—versus all the others gave an odds ratio of Alzheimer’s disease of 2.3 (1.1 to 4.8) (p = 0.03), when adjusted for age and for carrier status of apolipoprotein E ε4.

Table 1

Odds ratios of Alzheimer’s disease in men, according to serum level of total testosterone and androgen receptor CAG allele length


A difficulty with the study of complex diseases such as sporadic Alzheimer’s disease lies in the numerous interactions of each risk gene with other genes, with age and sex, and with the environment. This results in a weak overall effect of each gene and in conflicting results of genetic studies. To resolve the difficulty, we need to examine the interactions that define the “relevant subset” of people at risk for each susceptibility gene.3

The association of the glutamine (CAG) repeat polymorphism of the androgen receptor with Alzheimer’s disease appears limited to men.1 This is unsurprising, given the role of the androgen receptor as the main receptor for testosterone and 5α-dihydrotestosterone. We should therefore expect an interaction with testosterone in Alzheimer’s disease risk. Low serum testosterone has been associated with poor cognitive performance in elderly men4 and with Alzheimer’s disease.2 Testosterone exerts various neuroprotective actions56 through the androgen receptor. The CAG polymorphism of the androgen receptor has been associated with the risk of Alzheimer’s disease in men1 and with other brain disorders.7

Polyglutamine tracts play an important role in the activity of many transcription factors. The androgen receptor tract is in exon 1, which carries the transactivation domain. Transcriptional activity of the androgen receptor decreases with increasing length of the tract, even within the normal range.8 This effect is cell specific, which suggests that it may reflect interactions with other proteins. The androgen receptor coactivator, ARA24, has been found to be less effective with expansion of the androgen receptor polyglutamine tract. Binding of testosterone to the androgen receptor changes the latter’s conformation, releasing it from its cytoplasmic compartment and allowing nuclear translocation. This also permits interactions with coactivators, as well as between the N- and C-terminal domains of the androgen receptor. The polyglutamine tract is involved in several of these interactions.

Our results suggest that the combination of short androgen receptor alleles with lower levels of serum testosterone may increase the risk of Alzheimer’s disease for men. Further study is needed to clarify whether these two potential risk factors interact or act independently. There is a growing appreciation of, first, the influence of sex steroids on Alzheimer’s disease risk2 and, second, the role of sex in Alzheimer’s disease genetics.139 This study therefore merits replication in other, carefully characterised, all male cohorts. To demonstrate an interaction, large numbers would be needed, either through a collaborative study, using meta-analytical techniques for pooling, or eventually through a meta-analysis.


We especially thank all patients and volunteers, members of OPTIMA, the Department of Neuropathology, Radcliffe Infirmary, T James, and M Gales. We are most grateful to Bristol Myers Squibb, the Medical Research Council, and the Norman Collisson Foundation for financial support.


View Abstract

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.