A copy number variation morbidity map of developmental delay

Abstract

To understand the genetic heterogeneity underlying developmental delay, we compared copy number variants (CNVs) in 15,767 children with intellectual disability and various congenital defects (cases) to CNVs in 8,329 unaffected adult controls. We estimate that ∼14.2% of disease in these children is caused by CNVs >400 kb. We observed a greater enrichment of CNVs in individuals with craniofacial anomalies and cardiovascular defects compared to those with epilepsy or autism. We identified 59 pathogenic CNVs, including 14 new or previously weakly supported candidates, refined the critical interval for several genomic disorders, such as the 17q21.31 microdeletion syndrome, and identified 940 candidate dosage-sensitive genes. We also developed methods to opportunistically discover small, disruptive CNVs within the large and growing diagnostic array datasets. This evolving CNV morbidity map, combined with exome and genome sequencing, will be critical for deciphering the genetic basis of developmental delay, intellectual disability and autism spectrum disorders.

Figure 1. CNV size distributions in affected and unaffected individuals

The population frequency of the largest CNV in a sample is displayed as a survivor function with the proportion of samples carrying a CNV of a given size displayed as a curve, with 95% confidence intervals indicated by dotted lines. (A) The distribution of large CNVs in the Signature set (filtered to only contain events detectable by the Illumina 550K array) versus our control population (downsampled to only events detectable by the Signature 97K array) is indicated for the overall population. After corrections for different array densities, we observed a >13.5% increase in CNV burden beyond 500 kbp in cases with a proportion of the burden representing potentially novel loci. (B) We also performed a similar analysis on subphenotypes; in this analysis, we included all Signature CNVs in conjunction with downsampled control CNVs as we are highlighting interphenotype differences rather than case versus control frequencies. This is demonstrated here for the autism, cardiovascular and craniofacial phenotypes, which represent fairly distinct sample sets and show an increased burden for the cardiovascular and craniofacial phenotypes, even after exclusion of karyotypically visible (>10 Mbp) events.

Figure 2. Maps of CNV locations for chromosomes 15 (top) and 17 (bottom)

CNVs (>400 kbp) in affected individuals are shown in the upper portion for each chromosome with control CNVs shown in the lower portion. Disease enrichment p-values are plotted just below the control CNV maps, computed in 200 kbp windows along each chromosome (step size of 50 kbp). Deletions and duplications are red and blue, respectively, with the p-value wiggle plots colored accordingly and plotted on a negative log scale. In the middle of each plot, chromosomal features are colored as depicted. Significantly enriched regions are numbered and named on the right-hand side.

Figure 3. Discovery of novel microdeletions associated with genomic disorders

(A) A novel microdeletion on chromosome 15q25.2q25.3. Array CGH analysis for three individuals with a 660 kbp (chr15:82,889,423–83,552,890) deletion is shown. This microdeletion maps within a genomic hotspot flanked by high-identity SD blocks. Intrachromosomal SDs of high similarity relevant to this hotspot region are depicted as red (69.8 kbp, 98.6% identity) and green (17.6 kbp, 98.6% identity) block arrows. Note that the directly orientated SDs (red block arrows) likely mediate the underlying 15q25 rearrangements by non-allelic homologous recombination (NAHR). This region also contains a 60 kbp (chr15:82,775,465–82,835,495) gap in the current builds (build36 and build37) of the reference genome assembly. (B) Atypical 17q21.31 microdeletions refine critical interval genes. High-density array CGH for the 17q21.31 microdeletion region is shown for three individuals. Probes with log2 ratios below a threshold of 1.5 standard deviations from the normalized mean log2 ratio denote deletions (red). The typical deletions (top panel) were identified in 23 individuals while atypical deletions were identified in three individuals. Note that the smallest deletion (blue box) refines the phenotype-associated critical region (chr17:41,356,798–41,631,306) to encompass only five RefSeq genes. (C) Photographs of two individuals (9888884 and 648) with atypical deletions are shown. Patient #9888884 is a 5-year-old female child with clinical features typical of 17q21.31 microdeletion syndrome and includes distinctive dysmorphic features with a bulbous nasal tip, upslanting and almond-shaped palpebral fissures, long face, strabismus, epicanthal folds, and prominent ears; DD with limited speech; hypotonia in infancy; and a friendly disposition. Additional features are low birth weight, short stature, microcephaly, long fingers, and heart defects. She also presented with postaxial polysyndactyly, neonatal cholestasis, resolved leucopenia, dry skin with some hyperpigmented lesions, and an anteriorly split tongue. Patient #648 is 9-year-old male child and has a clinical history of generalized hypotonia, seizures, autism, mental retardation, motor DD, and dysmorphic features consistent with the 17q21.31 microdeletion syndrome (epicanthal folds; ptosis; long, pear-shaped nose; long, tapering fingers). Informed consent was obtained to publish the photographs.

Figure 4. Discovery of novel, exon-altering CNVs using the Signature CGH data

(A) For each coding exon (red bar), the three probes (black rectangles) nearest the exon for any given individual are used to define a cassette score. (B) Distribution of cassette intensities for exon 6 of PARK2 are sorted from lowest to highest (measured in standard deviations, Y-axis) across all samples (X-axis). Red points correspond to known, large deletion events that span the exon. (C) Validation results for the most strongly negative samples from (B) not previously known to carry deletions. Log2 ratio values (Y-axis for each individual row) for PARK2 (coordinates on the X-axis) in each of six tested samples are shown. Probes with very low intensities (< −0.5) are colored red, while those with moderately low values (< −0.3) are gray. Locations of PARK2 exons and probes on two of the most commonly used original oligonucleotide arrays are shown at the top.