🧬 PART THREE CHROMOSOME VARIANTS

17 NORMAL CHROMOSOMAL VARIATION One definition of human genetics is “the study of inherited human variation.” Variation can be normal: traits such as height, blood pressure, and intelligence. Abnormal variation may be clear-cut: dwarfism, hypertension, and intellectual deficiency. But the distinction may blur at the edges: short stature, borderline blood pressure, and low-normal IQ. There is somewhat of a parallel in the study of chromosomes. Some variation is quite normal, and well understood as such. And of course an observation such as a large deletion is abnormal. But some chromosomal variation does not admit of straightforward interpretation. In this short chapter, we review normal variation in chromosomal morphology as it was understood in the classical era of cytogenetics, which is to say during the period when chromosomal analysis was done on microscope study, looking at stained cell preparations on glass slides, and making a direct visual assessment. The turn of the 20th century provides quite a neat bookending, as molecular methodologies came to take precedence in this present century. So, while the material that follows is largely last-century learning, the counselor will occasionally have the need to consult the rich resource of earlier literature, and thus a broad understanding of this “old-fashioned” knowledge is still useful. We may consider normal variation within two major categories, essentially reflecting analysis due to either classical or molecular methodology: heteromorphisms ancient and modern, so to speak. In this chapter we deal with variations in size, staining qualities, and certain other attributes, from the microscopic analysis of chromosomes. Chromosomal copy number variants (CNVs) of small size, detectable only upon molecular karyotyping, are covered in Chapter 18. CLASSICAL CYTOGENETICS Microscopists from the era of classical cytogenetics became very familiar with the appearances of chromosomes and learned readily to distinguish normal structural variation. The counselor of the 21st century may yet need to refer to historic literature and should have at least some familiarity with these classical concepts. Homologs could differ in the respects as follows. Banding Pattern: Heterochromatin Heterochromatin is made up of highly repetitive DNA that has been distinguishable from euchromatin for the larger part of a century (Heitz 1928).1 Heterochromatic variants are best seen on C-banding, which specifically stains the extensive tracts of heterochromatin 1 The seminal contributions of Emil Heitz to the science of cytogenetics are reviewed in Passarge (1979). 510  CHROMOSOME VARIANTS adjacent to the centromeres of each chromosome (hence, the C), substantially comprising alpha-satellite DNA consisting of hundreds of thousands of copies of a 171 base pair repeat. Certain chromosomes show quite marked differences in their C-band pattern, particularly chromosomes 1, 9, 16, and the Y, and the large blocks of heterochromatin thus stained are labeled 1qh, 9qh, 16qh, and Yqh.2 They are of no phenotypic effect.3 Acrocentric Short Arms The short arms of the acrocentric chromosomes (13, 14, 15, 21, and 22) can vary quite considerably in their lengths. Indeed, some p arms are apparently completely absent, and others are several times the typical length. This reflects variation in the three components of the short arm: the centromeric heterochromatin, the satellite stalk, and the satellite material, identified as bands p11, p12, and p13, respectively (Figure 7–4). Band p12 contains multiple copies of genes coding for ribosomal RNA; because the nucleolus of the cell is formed by an aggregation of rRNA, this region is also called the nucleolar organizing region (NOR). Acrocentric short arm variation appears to be without any phenotypic effect. Banding Pattern: Euchromatin Most of a chromosome consists of euchromatin, which contains the active genetic material, resident in greater amount in G-light bands (pale-staining on Giemsa banding) than in G-dark bands. The light microscope cannot reliably enable detection of alterations of less than 3 Mb–5 Mb, and most deletions and duplications of more than this size can be presumed to have phenotypic consequences. Exceptions to this rule include first, euchromatic variants that involve common copy-number variable regions that become visible when copy number is high enough, or when the size of the copy-number variable tract is large enough. Second, there are chromosomal segments whose deletion or duplication has no phenotypic consequence. Euchromatic Variants Euchromatic variants (EVs) due to copy-number variable tracts (Table 17–1) can be considered, in a sense, as extreme forms of CNVs, either because their copy number is at the high end or higher than the normal range, or because their size is greater than 3 Mb (at which point they are excluded from the Database of Genomic Variants; see below). Thus, EVs and the molecular CNVs (Chapter 18) essentially form a continuum, with no fundamental genetic distinction. For example, Tyson et al. (2014) analyzed the REXO1L1 gene and pseudogene cluster that resides within a 12 kb tandem repeat in band 8q21.2, and of which the diploid copy number ranges from approximately 100 to 2 Variation in the size of Yqh in an extended Canadian kindred could inferentially be traced back over three centuries, allowing Genest (1973, 1981) to claim that it was “the oldest known chromosome aberration.” 3 This has been the prevailing, if not universal view, for quite some time. Reproduction may, however, be a vulnerable sphere; and Tempest and Simpson (2017) review the reported associations with infertility and unfavorable reproductive outcomes.

Normal Chromosomal Variation  511 200. This repeat may account for almost half of band 8q21.2 and, at the upper end of this range, additional G-light material is discernible (Figure 17–1). While D’Apice et al. (2015) proposed that deletion of this segment (but with several copies yet remaining) could be the basis of a new microdeletion syndrome, Barber et al. (2016) argue that more likely, it may typically be a benign EV. The same interpretation of innocuousness likely applies to the other EVs listed in Table 17–1. Some of these EVs may have been confused, on classical karyotyping, with pathogenic imbalances. On microarray analysis, however, the distinction should be clear; and in fact many microarrays exclude the repetitive regions that EVs involve. Imbalances of Chromosomal Segments with No Apparent Phenotypic Effect In a review in 2005, Barber found only 23 examples of families with directly transmitted autosomal segmental imbalance in which two or more carriers were unaffected, and a few have since been published (Table 17–2). These cases were often ascertained for incidental reasons, such as prenatal diagnosis for maternal age. The gene content is often lower than the genome average, and the lack of phenotype is attributed to the absence of dosage-sensitive genes, or to dosage compensation by related genes. Similar imbalances with no phenotypic consequence are recorded in more than one family for the gene-poor G-dark bands 2p12, 5p14, 13q21, and 16q21. Most of the families listed in Table 17–2 remain as isolated examples, and may yet turn out to reflect segmental incomplete penetrance. This may apply, for example, to the distal 3p cases, as other families with similar deletions are more often phenotypically affected. This question of penetrance in these few cases of cytogenetically visible imbalances is somewhat of a harbinger of the immense challenge that came to be presented by the flood of CNVs of 21st-century molecular analysis, as we discuss at length in Chapter 18. Table 17–1.  Euchromatic Variants Due to Copy-Number Variable Tracts EUCHROMATIC VARIANT (EV) REPEAT/SEGMENT SIZE CONTROL COPY NUMBER EV COPY NUMBER dup 8p23.2 2.5 Mb 2 3 amp 8p23.1 >260 kb 2-9 8–12 amp 8q21.2 12 kb 97–277 265–270 amp 9p12 ~1 Mb 1–3 7–12 del/dup 9p11.2p13.1 ~5 Mb 4 3–5 dup/trp/ins 9q12 ~5 Mb 4 5–6 del/dup/trp/amp 9q13q21.1 ~5 Mb 4 3–8 amp 15q11.2 ~1 Mb IGVH 1–3; NF1 1–4 IGVH 4–9; NF1 5–10 amp 16p11.2 692–945 kb 3–8 8–10 Abbreviations: amp = amplification; dup = duplication; EV = euchromatic variant; ins = insertion; IGVH = immunoglobulin variable heavy chain; NF1 = neurofibromatosis 1; trp, triplication. Source: From Tyson et al. (2014) 512  CHROMOSOME VARIANTS Inversions.  We mention normal variant inversions seen in certain chromosomes (1, 2, 3, 5, 9, 10, 16, and Y) in Chapter 9. Fragile Sites.  Under certain stressed culturing conditions some, indeed most, chromosomes show apparent rupture in one, or less commonly both chromatids (Sutherland 2003; Mirceta et al. 2022). This is almost always without phenotypic implication. The spectacular exception is the fragile site FRAXA at Xq27, and indeed this laboratory observation lent its name to the well-known fragile X syndrome, originally referred to as a “marker” X (Lubs 1969). The fragile site observed by the microscopist reflected the trinucleotide expansion within the FMR1 gene. Three other sites in the same region are FRAXB, FRAXD, and FRAXE, of which only the latter is pathogenic. Otherwise, only two fragile sites may be of clinical import. FRA11B, at 11q23.3, is possibly the basis of some (not all) Jacobsen syndrome 11q deletions (Michaelis et al. 1998; and see p. 419). In the case in Clang and LaBaere (2023), neither parent of an affected child showed the fragile site. A single case of a man with 46,XY,fra(16)(q22.1), the fragile site classed FRA16B/C, in whom 1% of sperm and two out of 10 PGT embryos showed chromosome 16 imbalance, is to be noted (Martorell et al. 2014). We mention the fragile site FRA10A at 10q23, which may or may not be relevant at prenatal diagnosis, on p. 699. The full range of fragile sites is presented in Mirceta et al. (2022). Figure 17–1.  The likely benign euchromatic variant at 8q21.3, which reflects copy number variation of the REXO1L1 gene and pseudogene cluster. This observation could be viewed, in a sense, as an intermediary between classical and molecular cytogenetic variation. Source: From C Tyson et al., Expansion of a 12-kb VNTR containing the REXO1L1 gene cluster underlies the microscopically visible euchromatic variant of 8q21.2, Eur J Hum Genet 22: 458–463. 2014. Courtesy JCK Barber and C Tyson, and with the permission of Springer Nature.

Normal Chromosomal Variation  513 GENETIC COUNSELING A person carrying a classical chromosome variant has, practically by definition, no increased risk for having abnormal offspring, pregnancy loss, or any other reproductive problem. Some view it as at best pointless and at worst counterproductive even to mention to the individual that a variant chromosome has been found; others feel obliged to pass on the observation. If it is discussed, it must be made clear that it is a normal finding—perhaps interesting but of no practical importance. For the heterochromatic Table 17–2.  Euchromatic Duplications and Deletions CHROMOSOME Del Dup qdp 1 p21-p31 q31.1-q32 2 p12-p12 (x2) q13-q14.1 3 p25.3-pter (×2) q28-q29 4 q34.1-q34.3 p16.1-p16.1 q12q13.1 5 p14.1-p14.3 (×2) 6 q22.31-q23.1 7 p22.3-pter (×2) 8 p23.1/2-pter p22-p22 q24.13-q24.22 p23.1-p23.3 9 p21.2-p22.1 p12-p21.3 10 q11.2-q21.2 p11.1-q11.22 p13-p14 11 p12-p12 11p15.3p15.1 q14.3-q22.1 12 q21.31-q22 13 q14.3-q21.33 q13-q14.3 q21-q21 q14-q21 q21.1-q21.31 q21.1-q21.33 16 q13q22 (×4) 18 p11.31-pter p11.2-pter q11.2-q12.2 22 q11.21-pter Notes: These del/dups (and one quadruplication) are detectable by microscope cytogenetics and without phenotypic effect, as inferred from the observation of transmission from phenotypically normal parent to normal child. The estimated sizes of the deletions and duplications range from 4.2 Mb to 16.0 Mb (del) and from 3.4 Mb to 31.3 Mb (dup). The numbers of studied families, where more than one, are shown in parentheses. Source: From JCK Barber, Directly transmitted unbalanced chromosome abnormalities and euchromatic variants, J Med Genet 42:609–629, 2005, and the Chromosome Anomaly Collection website at http://www. ngrl.org.uk/wessex/collection (updated information is posted in the “What’s New” section). Additional material due to Chen et al. (2011b), Coussement et al. (2011), Kowalczyk et al. (2013), Liehr et al. (2009), and Singer et al. (2021). 514  CHROMOSOME VARIANTS size variants (C-band and NOR) and euchromatic variants, the point can simply be made that some chromosomes come in short, medium, and long forms, and where a chromosome happens to fit in this continuum is without significance. For segmental imbalances ascertained in apparently unaffected individuals, careful clinical assessment should be made, if practicable, of carriers from the same family; incomplete penetrance and variable expressivity should be borne in mind in assessing innocuousness, or not, of the variant. Fragile sites are practically always normal findings. The primacy of molecular karyotyping in the 21st century in fact means that discovery of variants such as these will be rather infrequent events. There is considerable potential for iatrogenic anxiety, whereas in reality the biology of the supposed anomaly has no pathogenic implication. The counselor may thoroughly understand the presumed harmlessness of a variant chromosome, but the person in whose family it has been discovered may react “non-scientifically.” To put a stark setting on it, the worst possible response might be for a couple to choose to terminate a pregnancy because of an overinterpreted variant chromosome, as has actually happened with the 16p11.2 euchromatic variant (López Pajares et al. 2006). Primum non nocere: First do no harm.

COPY NUMBER VARIANTS  517 spastic ataxia of Charlevoix and Saguenay (ARSACS) in a patient with a SACS mutation at 13q12.12 on one chromosome, and a 1.33 Mb CNV deletion encompassing the SACS locus on the other. We have seen a very similar case in a woman with ataxia and a Charcot-Marie-Tooth-like neuropathy inheriting a (normally nonpathogenic) paternal 0.2 Mb CNV deletion which removed SACS, and an accompanying maternal SACS mutation on the other homolog, thus enabling a diagnosis of ARSACS. DATABASES The counselor dealing with a family in which a CNV has been shown has formidable resources to which to appeal. Collaborative efforts from around the world bring together data, and repositories are assembled to which enquiry may be made. DECIPHER.  This name comes from Database of Chromosomal Imbalance and Phenotype in Humans Using Ensembl Resources. It lists known or possibly pathogenic variants and VUSs. The Internet link is http://deciphergenomics.org. A panel displays CNVs that either match or overlap with a segment of interest. Duplications are shown in blue, and deletions are shown in red. The distinction between pathogenic variants and VUSs is indicated by the differing color intensity (the darker, the more likely to be pathogenic). The user will note that many cases show DDD as the data source: This is the database Deciphering Developmental Disorders. The DECIPHER browser also includes CNV data from gnomAD, DGV, ClinVar, and pHaplo/pTriplo scores. ClinGen.  This is a “National Institutes of Health–funded resource dedicated to building an authoritative central resource that defines the clinical relevance of genes and variants for use in precision medicine and research.” It provides a comprehensive evaluation of dosage sensitivity at the levels both of the gene and of the genomic region, based on data from the published literature and public and internal databases. Each gene and genomic region evaluated receives a separate haploinsufficiency (HI) score and triplosensitivity (TS) score, according to whether copy number loss (HI) or copy number gain (TS) results in a clinical phenotype. Within each category, a score of “3” denotes sufficient evidence for dosage sensitivity; “2” denotes emerging evidence; “1” Figure 18–2.  CNV Carrier Frequency by Ancestry Group. Notes: These data are taken from the UK Biobank. Deletion (del) carrier prevalence (left) is significantly higher than expected (+) for the South Asian ancestry group, and duplication (dup) carrier prevalence is lower than expected (-) for the African ancestry group. When considering only recurrent CNVs (right), there were fewer (-) African del and dup carriers than expected. Source: From LM Schultz et al., Copy-number variants differ in frequency across genetic ancestry groups, HGG Adv 5:100340, 2024. Courtesy LM Schultz and L Almasy, and with the permission of Elsevier. 518  CHROMOSOME VARIANTS is equivalent to little evidence; and “0” means no evidence for dosage sensitivity (Riggs et al. 2020).5 Each listed variant has a thumbnail sketch of condition and available evidence alongside; the link is http://clinicalgenome.org. Genome-Wide Dosage Sensitivity.  Collins et al. (2022) analyzed several large genomic datasets in order to predict the probabilities of haploinsufficiency (pHaplo) and triplosensitivity (pTriplo). A score of 0 indicates a low probability of dosage sensitivity, whereas a score of 1 indicates a high probability of dosage sensitivity. Applying this scoring to all human genes, including those not yet associated with a phenotype, they identified 2,987 as haploinsufficient (corresponding to pHaplo ≥ 0.86) and 1,559 as triplosensitive (pTriplo ≥0.94). Of note, 648 genes were uniquely triplosensitive. pHaplo and pTriplo scores are provided in the DECIPHER browser. DGV.  A more encyclopedic collection, including the smallest normal variants, is that due to the Database of Genomic Variants, which is curated at The Center for Applied Genomics at the Hospital for Sick Children, Toronto. The Center records “genomic alterations that involve segments of DNA that are larger than 50 bp. . . . The content of the database only represents structural variation identified in healthy control samples.” The data derive from upwards of eight million entries from worldwide populations (MacDonald et al. 2014; Zarrei et al. 2015). The database is accessed directly at http:// dgv.tcag.ca/dgv/app/home. Continual refining and addition of sequence-based CNV data leads to increasing accuracy and confidence, and a special track within the DGV lists “gold standard structural variants” (GSSVs). UCSC Genome Browser.  The University of California, Santa Cruz genome browser site at http://genome.ucsc.edu can be configured to display information from DECIPHER, ClinGen, OMIM, RefSeq genes, GeneReviews, gnomAD, and other useful datasets. It can include the Copy Number Variation Morbidity Map of Developmental Delay, a dataset which displays CNVs ascertained from nearly 30,000 children with various developmental disabilities, derived from Cooper et al. (2011) and Coe et al. (2014). Comparison data from nearly 20,000 controls allows the user to assess whether a particular CNV is enriched in the clinically ascertained population. Gene Constraint Metrics.  “Constraint scores” are epidemiological measures of genetic fitness in human populations, without consideration of gene function. Intolerant genes are those for which loss of a single copy has been reproductively disadvantageous over human history, either due to lethality or reduced reproductive fitness. Constraint metrics are features of the gnomAD database [https://gnomad.broadinstitute.org/] and can be used to interpret the clinical significance of copy number variants; they are also available in the UCSC browser. The key metrics for CNV interpretation are those that measure “intolerance to haploinsufficiency”: that is, the extent to which loss-of-function (LoF) variants in a given gene are depleted in the general population.6 LoF constraint is measured by two different scores. The “probability of being loss-of-function intolerant” (pLI) score divides genes into those that are intolerant to LoF (pLI >0.9) and those that are tolerant to LoF (pLI ≤0.9) (Lek et al. 2016). The “Loss-of-function Observed/ Expected Upper bound Fraction” (LOEUF) is a continuous metric that reflects the full spectrum of intolerance to pLoF. LOEUF ranges from 0 to 2, with values below 0.35 suggestive of intolerance (Karczewski et al. 2020). 5 Following Riggs et al. (2020), the suggested corresponding clinical classification is: 3, Pathogenic; 2, Likely Pathogenic; 1, Variant of Uncertain Significance (VUS); 0, VUS or Likely Benign. 6 Constraint metrics are also available for missense and synonymous variants, but these are seldom relevant for assessing the pathogenicity of CNVs.

COPY NUMBER VARIANTS  519 COPY NUMBER VARIANTS AND THE BRAIN The most complex organ, the brain, is the most susceptible to CNV imbalance, typically presenting as cognitive/behavioral/psychological dysfunction (Molloy et al. 2023; Mollon et al. 2023a); indeed, in most CNV imbalances there is no observable physical phenotypic abnormality. In a large study of children with intellectual disability/developmental delay, an excess of those with a CNV compared to controls emerged significantly at a CNV size of 400 kb and became more evident at 1.5 Mb (Cooper et al. 2011). At least some of these CNVs, therefore, would have been pathogenic. Again unsurprisingly, larger (>1.5 Mb) CNVs were overrepresented in de novo cases; presumably this reflects a reduced reproductive fitness of those with larger, and pathogenic, CNVs. Similar conclusions are reached in Coe et al. (2014). McCormack et al. (2016) recorded frequencies of benign versus pathogenic CNVs in an abnormal population (Figure 18–3). A subtler study was performed in Estonia (Männik et al. 2015); in this study, the CNV status of a large population was shown to correlate with educational attainment (Table 18–1). These subjects had been selected due to attendance at a general medical practice, and could be considered as quite close to a true random population sampling. Of the CNVs analyzed in these subjects, only smaller (0.25–1 Mb) duplications appeared to be consistently benign. In considering clinical outcomes relating to brain function, the effects of CNVs do not fit neatly into a “pathogenic vs benign” binary, but rather are quantitative. The study of genotypes in large, well-phenotyped population cohorts has shed light on the subtler manifestations of CNVs that may not be immediately apparent in the clinic. Figure 18–3.  Population Frequencies of CNVs. Notes: Total number of pathogenic CNVs, compared to total CNV frequencies, in an abnormal population. These data derive from a series of 5,369 postnatal (single or multiple congenital abnormalities, neurodevelopmental delay with or without neuropsychiatric disorders) and prenatal (two or more abnormalities detected on ultrasound) samples. Source: From A McCormack et al., Microarray testing in clinical diagnosis: An analysis of 5,300 New Zealand patients, Mol Cytogenet 9: 29, 2016. Courtesy DR Love and AM George, and with the permission of BioMed Central. 520  CHROMOSOME VARIANTS Kendall et al. (2019) studied 33 recurrent CNVs in more than 400,000 individuals from the UK BioBank, and found that 24 were associated with reduced performance on at least one cognitive test or measure of functioning. The changes on the cognitive tests were modest (an average reduction of just 0.13 standard deviation) but were heterogeneous across the different CNVs (Figure 18–4). These effects confer significant disadvantages in educational attainment and ability to earn income in adult life (Burghel et al. 2020). Another finding from large-scale genomic studies is that the effect of deletions and duplications on cognitive ability can be predicted using constraint scores (see above) of the included genes. Huguet et al. (2021) showed that each intolerant gene (as measured by LOEUF), when deleted, decreases general intelligence by IQ 2.6 points. For duplications, the effect is one-third that of deletions, corresponding to a reduction of 0.8 IQ points for each intolerant gene. In the setting of CNVs containing multiple genes, the overall effect on intelligence can be estimated as the sum of the effect of individual genes, regardless of the size of the CNV or of number of genes involved. This model explains nearly 80% of the effect of CNVs on intelligence7 and can be applied to both recurrent Table 18–1.  Educational Attainment in Three Estonian Cohorts with Respect to Copy Number Variant Carriage GROUP TOTALS EDUCATIONAL ATTAINMENT* NOT REACHING SECONDARY EDUCATION NO. % Estonian population 7,877 4.08 2,000 25 DECIPHER-listed CNV carriers 56 3.64 28 50 Deletion carrier by CNV size >1 Mb 37 3.51 17 46 500 kb–1 Mb 47 3.93** 16 34.0 250 kb–500 kb 164 3.84 50 30.5 Duplication carrier by CNV size >1 Mb 115 3.69 45 39.1 500 kb–1 Mb 149 4.10 43 28.9 250 kb–500 kb 319 4.14 78 24.5 Notes: In the general population, the average attainment score is 4.08. In those with DECIPHER-listed CNVs, it is less, at 3.64. The averages in those with other deletion CNVs is also less, ranging from 3.51 to 3.93. Likewise, the score is less in the larger duplication category (>1 Mb), but in those with smaller (0.25–1 Mb) duplications it is essentially the same as that for the general population. These average figures match those of the fractions of those not reaching secondary education. *The mean educational attainment score is derived from these levels, based on the Estonian education curriculum: less than primary, 1; primary, 2; basic, 3; secondary, 4; professional or college, 5; university or academic, 6; and scientific degree, 7. **This fraction, although slightly less than that of the general population, does not reach statistical significance. Source: From K Männik et al., Copy number variations and cognitive phenotypes in unselected populations, JAMA 313:2044–2054, 2015). 7 This model highlights the tight relationship between genetic fitness and cognitive abilities. For severe intellectual disability, it is intuitive that individuals are unlikely to reproduce—but for CNVs with much milder effects, it is unclear why a small reduction in intelligence may lead to reduced fitness. COPY NUMBER VARIANTS  521 Figure 18–4.  Copy Number Variants and Cognitive Capacity. Notes: Shown here is the association of 24 recurrent CNVs with seven cognitive tests, and four measures of functioning. The 0 point for each vertical line denotes an average performance. Points to the left of 0, the minus numbers (the negative coefficients), show a poorer performance/functioning for that particular trait. Point estimates are shown with 95% confidence intervals. Source: From KM Kendall et al., Cognitive performance and functional outcomes of carriers of pathogenic copy number variants: analysis of the UK Biobank, Br J Psychiatry 214:297-304, 2019. Courtesy G Kirov, and with the permission of The Royal College of Psychiatrists and the Cambridge University Press.

522  CHROMOSOME VARIANTS and nonrecurrent CNVs. Huguet and colleagues provide a prediction tool at https:// cnvprediction.urca.ca. Autism spectrum disorder (ASD) is a clinical diagnosis for which molecular karyotyping is the first genetic investigation.8 The counselor may deal rather frequently with the challenge of interpreting a finding of a CNV or CNVs, and which may be de novo or inherited. In a segment for which a causal link is well established, such as the del(16) (p11.2) (p. 432), and others shown in Figure 18–5, the expression microdeletion/duplication may be more apposite, and counseling may be (relatively) straightforward. For less well-understood segments, and especially when in combination, our understanding is a work in progress (p. 774). Other neuropsychiatric disorders with a CNV-causative component include schizophrenia, bipolar disorder, major depressive disorder, and attention deficit hyperactivity disorder (Mollon et al. 2023b) (Figure 18–6). Persons with CNVs at 1q21.1, 15q11.2, 15q13.3, 16p11.2, and 17p12 are at an increased risk of psychiatric disorders, although the effect is modest, with most hazard ratios between 2 and 5 (Calle Sánchez et al. 2022). Intriguingly, CNVs at the 17p12 locus are associated with a decreased risk of psychiatric diagnosis. Within the UK BioBank data, Mollon et al. (2023b) identified 18 CNVs that were associated with higher scores for mood disorders and anxiety disorders, including CNVs at 1q21.1, 8p23.1, 15q11.2, 15q13.3, 16p11.2, 16p12.1, 16p13.1, and 22q11.2. 8 It is necessary to distinguish “idiopathic autism” from neurogenetic syndromes that in some may include autistic-like features (e.g., Rett syndrome, fragile X syndrome, tuberous sclerosis). Harris (2016) offers a useful commentary; he refers to CNVs in idiopathic ASD as “common variation, individually of small effect, [which] may have substantial impact en masse.” See also p. 774. Figure 18–5.  CNVs and Autism. Notes: The frequencies are depicted of recurrent CNVs (de novo and inherited) in Autism Spectrum Disorder cases, compared with recurrent CNVs in the population-based cohort of the UK BioBank (UKBB). Deletion CNVs are in red, duplication CNVs in blue. The UKBB frequencies reflect those of an essentially normal population. A CNV at lower right (e.g. dup 16p13.11) is not much more frequent in an ASD population than in the normal population, and therefore very often harmless; whereas a CNV at upper left (e.g. del 15q11.2q13.1) is seen almost entirely only in the ASD population, and thus almost always harmful. The 95% confidence interval of the Odds Ratio is overlaid in grey. The x and y axis scales are logarithmic. Source: From JM Fu et multi al., Rare coding variation provides insight into the genetic architecture and phenotypic context of autism, Nature Genet 54, 1320–1331, 2022. Courtesy ME Talkowski, and with the permission of Springer Nature.

COPY NUMBER VARIANTS  523 Chawner et al. (2021) provide guidance on the management of adults identified to have CNVs with neuropsychiatric risk. Beyond the brain, the complement of CNVs may affect health later in life through their role in influencing many complex traits (Hujoel et al. 2022). Auwerx et al. (2024a) studied data from the UK BioBank identifying 73 CNV loci that contributed to the risk of 40 different diseases, including cardiovascular, respiratory, neurological hematological, liver, skin, and eye disorders, as well as cancer. Some of this increased disease risk was mediated via an increased body mass index. PENETRANCE AND EXPRESSIVITY The concepts of variable penetrance and expressivity,9 more traditionally invoked in Mendelian genetics, impose a real concern with respect to the CNV (Grayton et al. 2012). A CNV may be, in one genomic environment (e.g., in a parent), of no clinical effect, but it may be pathogenic in a child if a different CNV—a “second hit”—comes from the other parent. Subtle examples come from studies in autism (Coe et al. 2014). Or, a Figure 18–6.  CNVs and Neuropsychiatric Disorders. Notes: Odds ratios for selected CNVs across different neuropsychiatric disorders: major depressive disorder, attention deficit hyperactivity disorder, schizophrenia, and intellectual disability. The scale on the x axis is logarithmic. These data could be interpreted, for example, to show that the del 1q21.1 does not lead to a major depressive disorder (Odds Ratio practically = 1.0), whereas the risk for intellectual deficit is tenfold that of the general population. Source: From E Rees and G Kirov, Copy number variation and neuropsychiatric illness, Curr Opin Genet Dev 68:57–63, 2021. Courtesy E Rees and G Kirov, and with the permission of Elsevier. 9 Penetrance refers to the proportion of individuals with an imbalance that shows any trait resulting from that imbalance, whereas expressivity refers to the variability in phenotype of those who carry the imbalanced region. 524  CHROMOSOME VARIANTS microduplication or microdeletion of recognized incomplete penetrance may become penetrant in the company of a CNV (Figures 18–7 and 18–8). The concept of “digenic inheritance” may, in some, understate the genetic complexity: oligogenic or even polygenic mechanisms may be the basis of some CNV combinations determining a boundary beyond which phenotypic abnormality appears. Or, to use the common terminology, a two-hit or more-hit scenario may apply. The other issue to add into this mix is the matter of defining a boundary of abnormality, which can be a subtle question in the case of intellectual and behavioral traits. The Estonian and UK BioBank studies noted above lead to an inference that earlier assumptions that some heterozygotes for “syndromic CNVs” could be unaffected may be incorrect, albeit that the degree of effect is quite or indeed very mild (Lupski 2015; Männik et al. 2015). The counterpoint is that if the boundary of abnormality is taken to be intellectual disability, previous studies are likely to have overestimated penetrance by inclusion of background risk of disability unrelated Figure 18–7.  Deletions, Duplications, and Second-Hits. Notes: A display of duplications and deletions (left), alongside concomitant second-hit CNVs (right) that may influence phenotype, typically for the worse. Duplications and deletions are ranked, from top down, according to the frequency with which second-hit CNVs are observed. Those at the top of the list can sometimes be (apparently) non-penetrant, and thus the second-hit CNV may be necessary to lead to overt pathogenicity. Those further down the list have second-hit CNVs at no greater frequency than in the control population, and are “stand-alone” pathogenic. The fractions of microduplications and microdeletions due to parental or de novo origin are indicated in the left panel, according to the shading of the bars. Source: From S Girirajan et al., Phenotypic heterogeneity of genomic disorders and rare copy-number variants, N Engl J Med 367:1321–1331, 2012. Courtesy S Girirajan, and with the permission of the Massachusetts Medical Society.

COPY NUMBER VARIANTS  525 to the genetic variant, as well as non-intellectual disability phenotypes (Goh et al. 2025).10 IN PRACTICE The following is a very common situation the counselor faces: An imbalance is detected on molecular karyotyping, and the segment concerned contains CNVs and possibly known genes. An example is a 268 kb triplication at Xq27.1, trp chrX:139,332,751-139,601,714 that we have seen in a child with epilepsy and intellectual disability. The extent of the segment is displayed in Figure 18–9, according to the UCSC browser. Two known genes are included, Factor IX (F9) and MCF2; the latter is incompletely present and thus unlikely to be of pathogenic significance. If this sequence is interrogated in DECIPHER, a list is displayed of several segments of larger and smaller size, which overlap with the sequence of interest (Figure 18–10). The closest in this example is a case of dup X:139,474,090-139,682,289, and clicking on this segment displays further detail in a separate box, noting paternal inheritance. The case is colored purple for unspecified pathogenicity, the curators leaving this interpretation open. But the assessment is not inconsistent with the CNV being, at least in terms of brain function, benign. A summary of the genes resident within a region and a commentary on dosage sensitivity status where applicable, with links to synoptic data about each locus, are accessible through ClinGen (https://www.clinicalgenome.org). The display according to the Figure 18–8.  The 16p12.1 Deletion in a Proband, and in the Carrier and Non-Carrier Parent. Notes: The thresholds are shown for developmental and psychiatric manifestation, according to an individual’s genetic load: the presence of the 16p deletion; the carriage of second-hit variants; and the load otherwise of variants relating to an autism polygenic risk score. The deletion is carried by the proband and the carrier parent. Some carrier parents may manifest a relatively mild psychiatric phenotype. The non-carrier parent may have a number of different rare variants (blue), which of themselves, in this setting, are harmless, along with a variable polygenic load. But the combination, then, in the child, of the 16p deletion, along with a number of rare second-hit variants (yellow), combined with a polygenic load (polygenic risk score, blue), combine to cross a threshold sufficient to convey a developmental phenotype. 10 Goh et al. also highlight a range of other factors that might bias CNV penetrance estimates, including different CNV prevalences in different racial groups, age-related onset, variable expressivity, imperfectly collected phenotype data for both cases and controls, and publication biases.

526  CHROMOSOME VARIANTS trp(X) under discussion is shown in Figure 18–11. Checking the DGVa link11 mentioned above, and entering the coordinates of the trp(X)(q27.1), the CNVs contained therein are displayed (Figure 18–12). The largest is a 316,901 bp duplication, and smaller dels and dups are listed. But these, by definition in DGVa, are normal variants and can therefore be dismissed as pheno-contributory. The important next step is a family study, if feasible. In the example just given, it transpired that the brother and mother both had the same trp(X) and were both normal intellectually. But interestingly, they had both suffered thrombotic episodes with elevated levels of Factor IX. The conclusion to be drawn is that the neurological compromise12 in the presenting child is likely coincidental. A genetic diagnosis yet awaits, if indeed there is one. (The reader may well have similar stories to tell.) GENETIC COUNSELING The data from the laboratory which, as described above, can be rather sophisticated, needs to be clearly conveyed to those for whom the report is intended, which will very often be the genetic counselor, and through the counselor to parents and patients themselves. Novel CNVs The distinction between harmful and harmless variants can be a subtle exercise in the case of novel CNVs. As outlined above, interrogating databases such as UCSC browser, DECIPHER, and DGVa may be a first court of appeal. If a parent or other family member has the same variant and is of normal phenotype, the CNV may be adjudged a “likely benign” variant; or, other data—and especially pedigree data in the public domain—may be sufficiently powerful to indicate indeed a nonpathogenic CNV. A qualitative assessment of the genic content, as discussed above, may well be valuable. A de novo CNV may Figure 18–9.  The UCSC Browser. Notes: An example of a CNV display using the UCSC (University of California, Santa Cruz) browser, based on the trpX:139,332,751-139,601,714 bp described in the text. The Factor IX gene (F9, green) is completely contained within the segment; the MCF2 gene (grey) is partially included (Case of J Watt). Green denotes an OMIM gene, grey is a gene not yet linked to a phenotype. 11 Or, if a segment is identified in the University of California Santa Cruz (UCSC) browser, and the track “DGV Struct Var” under the Variation category is chosen, the region will be displayed and segments of deletion (red) and duplication (blue) within the vicinity indicated. Clicking on to a CNV within the segment of interest will link to the DGV database. 12 Brain imaging was normal, and there was no evidence that cerebral vascular thromboses could have been the basis of her abnormality. Figure 18–10.  DECIPHER. Notes: An example of an interrogation using the DECIPHER database, based on the [hg38]trpX:139,332,751-139,601,714 bp described in the text. Entering these coordinates, and then scrolling through a list of cases that DECIPHER presents with some degree of overlap, and looking at duplications (the blue lines), the closest variant is Patient 284123 who has [hg38] dupX:138394719-155996408. Clicking on this bar gives a dialog box with detailed information. In this example, the imbalance in the DECIPHER case is too large to usefully apply to the case in question. 528  CHROMOSOME VARIANTS Figure 18–11.  ClinGen. Notes: The display according to the ClinGen database, of the [hg38] trpX:139,332,751-139,601,714 described in the text. The display highlights F9 as an OMIM Morbid gene contained within the region, and MCF2 as a non-OMIM morbid gene that is overlapping the region. ClinGen dosage sensitivity scores indicate evidence for haploinsufficiency, and no evidence for triplosensitivity. DECIPHER haploinsufficiency score and gnomAD pLI and LOEUF scores are consistent with known pathogenicity of factor IX deficiency, but do not provide information about the effect of duplication (or triplication). Links within the table take the reader to synoptic data about each locus. Figure 18–12.  Database of Genomic Variants (DGVa). Notes: An example of an interrogation using the Database of Genomic Variants (DGVa) based on the trpX:139,332,751-139,601,714 bp described in the text. The website is accessed at http://dgv.tcag.ca/gb2/ gbrowse/dgv2_hg38. Nucleotide numbers are entered according to the appropriate “build” chosen (here, hg38). Duplicated CNV segments (blue) and deleted CNV segments (red) are presented. These are mostly labeled nsv and esv (sv, structural variant, archived and accessioned by dbVAR and DGVa, respectively). The largest duplication is nsv4048303, with breakpoints very similar to the triplication investigated in the example case. Clicking on each entry links to detail about the variant, including the lengths.

COPY NUMBER VARIANTS  529 need to be considered as “likely pathogenic” unless there is solid evidence otherwise; a data resource in this respect is the website http://denovo-db.gs.washington.edu (Turner et al. 2017). A detailed format for the practical assessment of a CNV is outlined in Di Gregorio et al. (2017), who assessed a little over 1,000 individuals with developmental delay/intellectual disability (Figure 18–13). These variants were classified into CNVs of size greater than 3 Mb (which we might equally call microdeletion/duplications); del/dups associated with known syndromes; CNVs spanning known Mendelian disease genes; likely pathogenic CNVs, and noting the genes contained within them; and VUSs/likely benign. These authors referred to the databases of a number of publicly available sources, in order to judge the possible pathogenicity of abnormal copy number of the loci contained within CNVs. In some, a diagnosis was clear enough at the outset, with a number of known syndromes seen. In others, it required a detailed weighing of the nature of the loci and appealing to information from the several sources. The reader wishing further demonstration of the rationale in CNV assessment is referred to this paper. The question of non-penetrance, or at least reduced expressivity of a CNV, is a challenging one. Attempting to dissect out “micro-phenotypes” in a parent may prove rather fraught. Adding to this is the problem of “second-hit” CNVs and the degree to which they may modify or exacerbate a phenotype. In the meantime, in advising about the risk to a future child, the counselor will need to consult current sources and to seek expert advice. A problem of long familiarity in genetic counseling, that of dealing with uncertainty, certainly applies here (Wilkins et al. 2016). Conveying the information about a novel CNV to counselees is an exercise to which genetic counselors are becoming more accustomed, which is not to say that they find it straightforward. Finally, a question of well-considered clinical judgment, and of which the answer might differ between Figure 18–13.  A schema for the analysis of copy number variants. Notes: The data from a series of 1,015 cases of developmental delay/intellectual disability were assessed, and in 10%, a pathogenic CNV was identified. The criteria by which the CNVs were judged are set out in fine detail in the Tables S1-S7 in the original paper. Sources referred to: DGV, DECIPHER as noted above; OMIM, Online Mendelian Inheritance in Man; HGMD, Human Gene Mutation Database; SFARI, Simons Foundation Autism Research Initiative; NDAR, National Database for Autism Research; NDD, neurodevelopmental disorders; ExAC, Exome Aggregation Consortium; GO, Gene Ontology Consortium; VOUS, Variant of Uncertain Significance. Source: From Di Gregorio et al., Copy number variants analysis in a cohort of isolated and syndromic developmental delay/intellectual disability reveals novel genomic disorders, position effects and candidate disease genes, Clin Genet 92: 415–422, 2017. Courtesy A Brusco and GB Ferrero, and with the permission of John Wiley & Sons.