Article Text


Shifts in angiotensin I converting enzyme insertion allele frequency across Europe: implications for Alzheimer’s disease risk
  1. F Panza1,
  2. V Solfrizzi1,
  3. A D’Introno1,
  4. A M Colacicco1,
  5. C Capurso1,
  6. A Capurso1,
  7. P G Kehoe2
  1. 1Department of Geriatrics, Centre for Aging Brain, Memory Unit, University of Bari, Policlinico, Piazza Giulio Cesare 11, 70124 Bari, Italy
  2. 2Department of Care of the Elderly, University of Bristol, Frenchay Hospital, Bristol, UK
  1. Correspondence to:
 Dr Francesco Panza;{at}

Statistics from

Early studies suggested that angiotensin I converting enzyme (peptidyl-dipeptidase A) 1 (ACE) gene polymorphism is associated with an increased risk of coronary artery disease and, more recently, with sporadic late onset Alzheimer’s disease.1 Studies conducted in northern European populations have considered the ACE*I allele to be a risk factor for various types of cognitive decline.1,2 One such study in a French population found an association between the ACE*D allele and dementia,3 while other studies in southern European populations found either a slight but significantly increased frequency of ACE*I in Alzheimer’s disease patients4 or did not detect any effect of ACE polymorphism.5

Our group recently reported the novel finding that apolipoprotein E (APOE) ε4 allele shows a geographical trend, decreasing in frequency from northern to southern Europe.6 We hypothesised that the variability in the strength of evidence for an association between ACE polymorphism and Alzheimer’s disease was related to similar geographical variations in ACE*I frequency. We investigated whether there was evidence in southern Italy of an association between the ACE polymorphism and increased risk of Alzheimer’s disease. Secondly, we compared our results with the findings from published studies on other European populations.1,2,4

Between June 1998 and October 2001, we consecutively examined in our centre 141 patients with Alzheimer’s disease (51 men, 90 women; mean (SD) age at onset, 71 (8.5) years), and 268 unrelated caregivers, spouses, friends, neighbours, or volunteers (118 men, 150 women; mean age at collection, 72 (7.1) years). A clinical diagnosis of probable Alzheimer’s disease was made according to the criteria of the National Institute for Neurological and Communicative Disorders and Stroke/Alzheimer’s Disease and Related Disorders Association, and the group of non-demented elderly control subjects was sex and age matched. The ascertainment, diagnosis, and collection of cases and controls are described in detail elsewhere.6 The age at onset of Alzheimer’s disease symptoms was estimated from semistructured interviews with the patients’ caregivers. The study protocol was approved by the ethics committee of the University of Bari. After a complete explanation of the study, written informed consent was obtained from all the subjects or their relatives.

APOE genotypes were determined as detailed elsewhere.6 ACE genotypes were produced using established methods, followed by a quality control amplification step necessary in detecting underamplified ACE*I alleles.1

The statistical analysis was performed by Pearson χ2 test to make genotype and allele comparisons as well as test for agreement of data with Hardy-Weinberg principles. Allele frequencies were determined by allele counting. To express variances of the allele frequencies, we used 95% CIs, calculated by Wilson’s formulas. The differences among age at onset of Alzheimer’s disease symptoms in relation to different ACE genotypes were calculated with Mann-Whitney test. To evaluate whether the association between Alzheimer’s disease and ACE genotypes were homogeneous in all APOE strata we used a permutation based exact logistic model by LogXact procedure implemented in the SAS system (Proc-LogXact 4; Copyright 2001 by CYTEL Software Corporation, Cambridge, MA 021139). In order to correct for multiple statistical testing, the results were adjusted according to Bonferroni inequality. The Cochran-Armitage trend test was carried out to evaluate the geographical trend among ACE allele and genotype frequencies in Alzheimer’s disease patients and controls of three European countries (Italy, Spain, and United Kingdom), from published studies.1,2,4 The data were analysed by SAS FREQ procedure (version 8.2).

Table 1 shows ACE allele and genotypes frequencies in Alzheimer’s disease patients and controls in southern Italy. The frequencies of the different ACE genotypes in our population were in Hardy–Weinberg equilibrium (HWE) (cases: Pearson χ2 = 2.09, p = 0.15; controls: χ2 = 2.49, p = 0.11). Moreover, there was no evidence that the genotypic counts of Alzheimer’s disease patients and controls were not under HWE (Alzheimer’s disease patients under HWE, given the controls were under HWE: likelihood ratio, χ2 = 2.18, p = 0.34). No significant differences were found in ACE genotype frequencies between patients with Alzheimer’s disease and controls in this southern Italian population.

Table 1

Angiotensin I converting enzyme (ACE) genotype and allele distributions in Italian, Spanish, and United Kingdom populations*

We did not find any statistically significant differences in rates between ACE alleles and Alzheimer’s disease among APOE allele strata, nor did we observe any lack of homogeneity among response differences (data not shown). Interestingly, Alzheimer patients with the ACE*I/*I genotype were on average 3.6 years younger at onset than those with the ACE*D/*D genotype (mean (SD) age at onset: ACE*D/*D, 72.1 (6.8) years; ACE*I/*D, 70.3 (8.1) years; ACE*I/*I, 68.5 (7.3) years). However, this difference did not reach statistical significance (z = 1.49; Bonferroni p > 0.05).

The ACE*I allele frequency in Alzheimer’s disease patients and controls showed a statistically significant decreasing trend from northern to southern regions of Europe (z = 5.36 p <0.001; z = 4.33 p <0.001, respectively), while there was a concomitant increase in ACE*D allele frequency (table 1). This was reflected by genotype data whereby a decreasing geographical trend from north to south was found for ACE*I/*I genotype (cases, z = 3.92 p <0.01; controls, z = 4.15 p <0.001) and an inverse trend for ACE*D/*D genotype (cases, z = 3.29 p <0.001; controls, z = 3.46 p<0.001). Interestingly, we found a statistically significant decreasing trend from northern to southern regions for the ACE*I/*D genotype, but this was observed only in Alzheimer’s disease patients (z = 3.12 p <0.01).


The present study does not support previous findings that increased Alzheimer’s disease risk is associated with the ACE*I genotype and allele frequencies.1 The age at onset of Alzheimer’s disease patients with the ACE*I/*I genotype appeared to be lower than in those with the ACE*D/*D genotype. Though this was not statistically significant, it suggests that the presence of an ACE*I allele might bring forward the onset of the disease without being linked to an increased overall risk of it occurring. Our findings support those of a previous report in which no evidence of an interaction between ACE alleles and age at onset, sex, and family history was found (data not shown).1

It is becoming apparent that the possible association between the ACE polymorphism and increased Alzheimer’s disease risk is complex. The variation in results between different studies may simply reflect the inherent susceptibility of such association studies to type I and type II statistical errors. Another possible explanation may be the direct result of geographical genetic variation which we have hypothesised. Indeed, as with our previous findings with APOE,6 we report here that the putative association between ACE gene variants and increased risk of Alzheimer’s disease may be influenced by geographical genetic variations (table 1). The different and conflicting patterns of association of ACE polymorphism and Alzheimer’s disease in populations worldwide may be explained by similar geographic trends or indeed another Alzheimer’s disease susceptibility locus located elsewhere in ACE or a nearby gene. Furthermore, the same ACE gene may have pleiotropic age and sex dependent effects on Alzheimer’s disease. Though the strength of association of APOE e4 with Alzheimer’s disease seems not to be influenced by the low prevalence of e4 in southern Europe,6 the decrease of the ACE*I allele frequency could be related to the different patterns of association between this polymorphism and Alzheimer’s disease in various European studies.1–5


View Abstract


  • Competing interests: none declared

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.