MITOCHONDRIAL DNA SNPS ASSOCIATED WITH SCHIZOPHRENIA EXHIBIT HIGHLY VARIABLE INTER-ALLELIC HAPLOGROUP AFFILIATION AND NUCLEAR GENOGEOGRAPHIC AFFINITY: BI-GENOMIC LINKAGE DISEQUILIBRIUM RAISES MAJOR CONCERN FOR LINK TO DISEASE

Background Mitochondria play a significant role in human diseases. Genetic variants in mitochondrial DNA (mtDNA) – and in nuclear genescoding for mitochondrial function - have been associated with disease. Mitochondrial diseaseis clinically characterised by complex metabolic, neurological, muscular and psychiatric symptoms. SNPsin mtDNA have been proposed as potential disease modifiers. This has been reported in neurological degenerative diseases such as Alzheimer's disease and Parkinson's disease, metabolic diseases and cancers, as well as psychiatric diseases, notably schizophrenia (SZ) and bipolar disease. However, disease associations with mitochondrial DNA (mtDNA) SNPshave proven difficult to replicate. We identified eight mtDNA SNPsthat had previously been associated with SZ and re-analysed the association with SZ in the Danish iPSYCH cohort, and examined the effect of mtDNA variation and variation in nuclear genogeographic affinity, or ancestry, on the association. Methods Persons were recruited through the iPSYCH program, which is a case-cohort study. DNA was isolated from dried blood spotsobtained through the Danish Neonatal Screeningprogram. Genotyping was performed using the PsychChip array, comprising 418 SNPsin mtDNA. Supervised ancestry estimation of the nuclear genome was performed using a reference population of 2,248 persons assigned to one of nine population Groups. The estimation was done with 103,268 nuclear SNPsusing ADMIXTURE 1.3.050. Results A reanalysis of eight schizophrenia-associated mtDNA SNPs, in 23,743 normal Danes and 2,538 schizophrenia patients, revealed marked inter-allelic differences both in haplogroupaffiliation and nuclear ancestry, genogeophraphic affinity (GGA). Only two mitochondrial SNPs, m.15043 A and m.15218 G, were significantly associated with schizophrenia. However, these associations disappeared when corrected for haplogroupaffiliation. Discussion The combined variation in mtDNA affiliation and nuclear genomic ancestry, bi-genomic linkage disequilibrium (2GLD) could entail population stratification. This might result in both false positive and negative associations between mtDNA SNPsand disease. The extensive 2GLD documented for the eight SNPsexamined here is a major concern when interpreting historic as well as designing future mtDNA association studies. The fact that careful control, as here, of mtDNA affiliation and the 2GLD, results in none of eight previously SZ associated mtDNA SNPsbeing associated with SZ in the very large Danish iPSYCH cohort, suggests that previously reported associations could indeed be spurious findings due to cryptic population stratification.