As expected, due to the selection on relationship within 10 generations, those included in the initial genome-wide linkage analysis had an almost 10-fold higher average kinship compared with that of all patients. These patients also showed an increased inbreeding coefficient, which may point to a recessive form of disease.
Results of the complete genome-wide scan from both the dominant model and recessive model with shortest and hypothetical loops are illustrated in .
Figure 2 LOD score plots from multipoint analyses of the whole autosomal genome in nine ADHD patients. (a) Shows the LOD scores for dominant model, b and c for recessive model under homozygosity mapping with the shortest loop and hypothetical loop, respectively. (more ...)
There was no genome-wide significant evidence for linkage under either model. The highest LOD score under the dominant model was observed at 6q16 (HLOD=0.91). Other regions that showed weak evidence of linkage include 2q23–24 (HLOD=0.81), 3q24 (HLOD=0.75), and 12p13 (HLOD=0.71). Homozygosity mapping yielded five genomic regions with HLOD ≥1. The strongest evidence of linkage was observed at 18q21–22 (D18S64, HLOD=2.13). Other regions with HLOD ≥1 include 6p23 (D6S470, HLOD=1.68), 6p12 (D6S257, HLOD=1.07), 1p36 (D1S214, HLOD=1.09), 18p11 (D18S59, HLOD=1.15), and 15q25 (D15S205, HLOD=1.19). Details are provided in . Adjusting for multiple inbreeding loops () did not alter our findings, decreasing LOD scores only marginally.
Loci with MLS ≥1 under homozygosity mapping
The patients' haplotypes at chromosome 18 (presented in ) show excess of homozygosity but not at a single marker. Four out of nine patients are homozygous for allele 1 of the marker D18S464. This is, however, the most common allele, with a frequency of homozygosity of 0.42. Also, at marker D18S64, five patients are homozygous: two are homozygous for allele 1, which has a frequency of homozygosity of 0.05; two are homozygous for allele 2, which has a frequency of homozygosity of about 0.002; and one patient is homozygous for allele 5, which has a frequency of homozygosity of 10−9.
The regions of interest on chromosomes 18 and 6 were additionally typed for the remaining distantly related 17 patients, and the data from all 21 inbred patients () were reanalysed. For this analysis, we used only the recessive model of inheritance as it yielded the evidence for linkage in the initial analyses.
Subpedigrees after breaking the complete pedigree for all 26 patients and haplotypes for chromosome 18 for only the inbred patients (n=21) are shown at the bottom.
The LOD scores did not increase by increasing the sample size, but rather decreased due to adding non-informative individuals (). On chromosome 6p, the highest LOD score we observed was HLOD=1.51, at marker D6S470, using homozygosity mapping. This HLOD score increased to 2 at marker D6S1574 when a recessive model with a disease allele frequency of 0.01 and complete penetrance was run separately for the 10 patients with only inattentive type ADHD (). For chromosome 18, the highest LOD score observed was 1.83 at marker D18S1161. The haplotype analysis, however, showed that 11 out of 21 inbred patients were homozygous for the same allele at marker D18S464 (), and 10 patients were homozygous at marker D18S64.
Figure 4 HLOD score plots from the follow-up multipoint analyses with increased sample size for chromosomes 6 and 18 and also the top GAIN hits for the same regions. The horizontal axis depicts the distance in base pairs, the left vertical axis depicts the LOD/HLOD (more ...)
Finally, we compared our findings with those from the genome-wide association analysis of the GAIN consortium. The chromosome 18 region identified in our linkage analysis also showed evidence for association in the GAIN analysis; rs2311120 (P=10−5), rs4149601 (P=10−4), rs9973180 (P=10−4), and rs2006776 (P=10−4) are located in the linkage region we identified (). These SNPs, particularly rs2311120, which was the third most significant SNP, were among the top 100 most strongly associated SNPs in the GAIN consortium. Also for chromosome 6p in the region we identified in our genome-wide scan, there was also some evidence of association in the GAIN study (rs2772387, P=3 × 10−4).