Replication of linkage at chromosome 20p13 and identification of suggestive sex-differential risk loci for autism spectrum disorder

Abstract

Background: Autism spectrum disorders (ASDs) are male-biased and genetically heterogeneous. While sequencing of sporadic cases has identified de novo risk variants, the heritable genetic contribution and mechanisms driving the male bias are less understood. Here, we aimed to identify familial and sex-differential risk loci in the largest available, uniformly ascertained, densely genotyped sample of multiplex ASD families from the Autism Genetics Resource Exchange (AGRE), and to compare results with earlier findings from AGRE.

Methods: From a total sample of 1,008 multiplex families, we performed genome-wide, non-parametric linkage analysis in a discovery sample of 847 families, and separately on subsets of families with only male, affected children (male-only, MO) or with at least one female, affected child (female-containing, FC). Loci showing evidence for suggestive linkage (logarithm of odds ≥2.2) in this discovery sample, or in previous AGRE samples, were re-evaluated in an extension study utilizing all 1,008 available families. For regions with genome-wide significant linkage signal in the discovery stage, those families not included in the corresponding discovery sample were then evaluated for independent replication of linkage. Association testing of common single nucleotide polymorphisms (SNPs) was also performed within suggestive linkage regions.

Results: We observed an independent replication of previously observed linkage at chromosome 20p13 (P < 0.01), while loci at 6q27 and 8q13.2 showed suggestive linkage in our extended sample. Suggestive sex-differential linkage was observed at 1p31.3 (MO), 8p21.2 (FC), and 8p12 (FC) in our discovery sample, and the MO signal at 1p31.3 was supported in our expanded sample. No sex-differential signals met replication criteria, and no common SNPs were significantly associated with ASD within any identified linkage regions.

Conclusions: With few exceptions, analyses of subsets of families from the AGRE cohort identify different risk loci, consistent with extreme locus heterogeneity in ASD. Large samples appear to yield more consistent results, and sex-stratified analyses facilitate the identification of sex-differential risk loci, suggesting that linkage analyses in large cohorts are useful for identifying heritable risk loci. Additional work, such as targeted re-sequencing, is needed to identify the specific variants within these loci that are responsible for increasing ASD risk.

Figure 1

Schematic of three-stage study design for linkage analyses. Linkage analyses were performed using a three-stage approach. (1a) A discovery sample (Stage 1) and (1b) previous linkage results were used to identify loci of suggestive linkage (logarithm of odds (LOD) ≥2.2). (2) These suggestive loci were tested for linkage in an extended sample. (3) For loci achieving genome-wide significant linkage (LOD ≥3.6) in either the discovery, extended, or previously published studies, the non-overlapping subset of families was tested for independent replication of linkage. AGRE, Autism Genetics Resource Exchange; SNP, single nucleotide polymorphism.

Figure 2

Genome-wide linkage: stage 1 sample. Logarithm of odds (LOD) scores from genome-wide, non-parametric, multipoint linkage analysis for autism spectrum disorder affection status in the stage 1 sample are plotted for autosomes (Merlin) and chromosome X (Minx). Top panel: all multiplex families (ALL); middle: male-only families (MO); bottom: female-containing families (FC). Dashed lines mark LOD thresholds 2.2 for suggestive and 3.6 for significant linkage [39]. Arrows note signals with LOD ≥2.2.

Figure 3

Independent replication of genome-wide significant linkage at 20p13. Genome-wide significant linkage signal at 20p13 from the combined sample (ALL fams; green, solid line), Autism Genetics Resource Exchange families (AGRE fams) analyzed by Weiss et al.[33] (orange, dashed line), and all AGRE families not previously analyzed by Weiss et al. [33] (Ind. Fams; purple, solid line). Top: linkage across the full chromosome 20; middle: linkage across a 2-logarithm of odds (LOD) interval from the peak LOD; bottom: RefSeq gene alignment in the 2-LOD interval. Dashed lines mark LOD thresholds corresponding to linkage P-values of 0.01, 0.001, and 0.0001 [35].

Figure 4

Linkage in regions of interest from previous studies: combined sample. Logarithm of odds (LOD) scores from non-parametric, multipoint linkage analysis for autism spectrum disorder affection status in the combined sample (Merlin) are plotted for chromosomes with suggestive linkage peaks (LOD ≥2.2) from previous studies. Colored bars and black marks indicate the spans and peaks of linkage regions of interest, respectively; text indicates the linkage region source (L, Liu et al., 2001 [26]; Y, Yonan et al., 2003 [34]; St, Stone et al., 2004 [28]; C, Cantor et al., 2005 [25]; M, McCauley et al., 2005 [27]; Sz, Szatmari et al., 2007 [32]; W, Weiss et al., 2009 [33]; green and ALL, all multiplex families; blue and MO, male-only families; red and FC, female-containing families). Dashed lines mark LOD thresholds of 2.2 for suggestive and 3.6 for significant linkage.

Figure 5

Significant sex-differential linkage peaks. Regions of sex subgroup-specific suggestive linkage (logarithm of odds (LOD) ≥2.2) with empirically significant signal enrichment from randomization testing are plotted. (A) 1p31.3, MO; top: linkage across full chromosome 1 from all stage 1 family groups; upper-middle: linkage from all stage 1 (solid lines) and combined sample (dashed lines) family groups and association signal from transmission disequilibrium test (TDT; black dots, EMP1 is empirical P-value from TDT) in male-only (MO) families across 2-LOD interval from peak LOD; lower-middle: empirical P-values from test of linkage subgroup specificity (black = stage 1, gray = combined), dotted lines indicate thresholds corresponding to empirical P-values of 0.05, 0.01, and 0.005; bottom: RefSeq gene alignment in 2-LOD interval. (B) Linkage across full chromosome 8 from all stage 1 family groups. (C) 8p21.2, female-containing (FC), (D) 8p12, FC; for each, top: linkage from all stage 1 family groups and TDT association in FC families across 2-LOD interval from peak LOD; middle: empirical P-values from stratification permutation tests; bottom, RefSeq gene alignment in 2-LOD interval.