PLD3 in non-familial Alzheimer's disease

arising from C. Cruchaga et al. Nature 505, 550–554 (2014); doi:10.1038/nature12825 Interest in the role of rare genetic variants in the aetiology of complex diseases such as Alzheimer's disease is increasing1, 2. Recently, Cruchaga et al.3 provided evidence supporting the role of rare variants in the phospholipase D3 (PLD3) gene in both familial late-onset Alzheimer's disease (age at onset >65 years) and in non-familial Alzheimer's disease. In a follow-up study of 3,568 non-familial Alzheimer's disease cases and 3,867 controls of German or Spanish descent, we failed to replicate the latter finding. Our results therefore cast doubt on the aetiological relevance of rare coding PLD3 variants in non-familial Alzheimer's disease. There is a Reply to this Brief Communication Arising by Cruchaga, C. & Goate, A. M. Nature 520, http://dx.doi.org/10.1038/nature14037 (2015). Four coding PLD3 variants reported by Cruchaga et al.3 with P < 0.15 (p.Met6Arg, p.Pro76Ala, p.Val232Met and p.Ala442Ala) were genotyped in three independent non-familial Alzheimer's disease case–control samples using Sequenom's iPLEX assay. The two Spanish samples comprised: (1) 2,166 cases and 2,754 controls (Fundació ACE)4; and (2) 461 cases and 180 controls (St Pau Hospital)5. The German sample comprised 941 cases and 933 controls from three multicentre studies6, 7, 8. All non-familial Alzheimer's disease cases fulfilled the National Institute of Neurological and Communicative Disorders and Stroke (NINCDS) and the Alzheimer's Disease and Related Disorders Association (ADRDA) criteria, and all participants provided written informed consent. Association was tested using Armitage's test for allelic trend (INTERSNP9). In addition, a burden analysis was performed with the collapsing test COLL10. In the Spanish and German samples, the variant p.Met6Arg was found to be monomorphic. No evidence for an association between non-familial Alzheimer's disease and any of the three polymorphic PLD3 variants was found in the analyses of the three individual study samples or in the analysis of the combined sample (P > 0.05, Table 1). Interestingly, the power of our case–control study was sufficiently large to detect an odds ratio (OR) of 2 or larger, as reported by Cruchaga et al.3. Since the association reported by Cruchaga et al.3 was mainly for late-onset Alzheimer's disease and was stronger in cases with a positive family history3, we stratified our samples according to: (1) age at onset (that is, ≤65 or >65 years); and (2) family history (defined as the presence of at least one self-reported dementia case in the family). However, neither analysis generated evidence for association (P > 0.05). Similarly, the burden analysis revealed no significant differences between non-familial Alzheimer's disease cases and controls in terms of the occurrence of any of the four PLD3 variants (P > 0.05, Table 1). Our analyses in two European populations did not implicate rare PLD3 coding variants in non-familial Alzheimer's disease susceptibility. This calls into question both their genetic epidemiological relevance in the investigated populations, and their importance in the pathogenesis of non-familial Alzheimer's disease. Although we cannot with certainty exclude a false-negative finding secondary to population stratification, this possibility seems unlikely for the following reasons: our findings are consistent with the negative findings reported from the French population in an accompanying Comment by Lambert et al.11; and the allele frequencies obtained in the French controls were very similar to those obtained in our Spanish and German controls. Furthermore, a meta-analysis of 7,565 cases and 18,424 controls, which included the present sample, did not reveal an association with p.Val232Met (see accompanying Comment by Hooli et al.12). Here, the power to detect an association was 96.6% at α = 0.05. In summary, these data suggest that our results are reliable. Notably, our findings do not concur with those of the accompanying paper by van der Lee et al.13 who reported a significant association for p.Val232Met in their meta-analysis of six independent samples (1,914 non-familial Alzheimer's disease patients and 8,021controls). However, the size of the individual samples was—at least in terms of the number of cases—small, and the participants had been recruited from various populations. Both the low frequency of p.Val232Met, and its varying frequency between their small investigated samples, render the van der Lee et al.13 result difficult to interpret. Although our results do not exclude the possibility that these PLD3 variants are implicated in Alzheimer's disease patients with a strong family history of Alzheimer's disease, the accompanying Comment by Hooli et al.12 found no evidence for this hypothesis in an investigation of 439 multiply affected Alzheimer's disease families. In conclusion, our results in non-familial Alzheimer's disease cases from Spain and Germany provide no support for the hypothesis that rare PLD3 variants are implicated in the aetiology of Alzheimer's disease. Clarification of the role of PLD3 in Alzheimer's disease will require the investigation of further large samples from defined populations. Download references