CNV “Double Whammies” May Account for Variable Neuropsychiatric Phenotypes
22 February 2010. Copy number variants (CNVs) have promised to account for some of the heritability for psychiatric diseases not captured in genomewide association studies (GWASs). While there is little question that these genetic deletions and duplications contribute significantly to disease risk, clear mechanistic conclusions have been difficult to draw because several CNVs studied so far are associated with a wide range of phenotypes.
The authors of a new Nature Genetics paper on a 16p12.1 microdeletion associated with developmental delay say their results suggest that some of the striking phenotypic variation seen in previous CNV studies—whether in the form of distinct clinical features or severity of illness—may be explained by a “two-hit” model: a given CNV may predispose individuals for a neuropsychiatric phenotype, and also exacerbate phenotypic expression or introduce distinct clinical features in conjunction with a second CNV.
In a meta-analysis carried out last year (Itsara et al., 2009), members of the team directing the new study—led by Evan Eichler of the University of Washington, with collaborators at the University of California-San Diego, Harvard, and the University of Auckland, among other institutions—identified the 520 kb deletion in question in five individuals affected by autism or schizophrenia, but there was insufficient statistical power to definitively implicate the CNV in those disorders.
To take a closer look at this CNV, co-first authors Santhosh Girirajan of the University of Washington and Jill Rosenfeld of Signature Genomic Laboratories, Spokane, Washington, and their colleagues assembled large discovery and replication cohorts, with a combined total of more than 20,000 cases of intellectual disability or developmental delay and nearly 15,000 controls. They also studied a smaller schizophrenia cohort of about 3,000 patients and compared it to the same 15,000 controls.
In this study, they found that the 16p12.1 CNV was significantly enriched in the cases of developmental delay, being found in 42 cases versus eight controls (P = 1.18 x 10-4). There was no significant association between the CNV and schizophrenia in the smaller cohort, but the authors could not rule out insufficient statistical power underlying this result.
As in previous CNV studies, the cases exhibited a gamut of phenotypes: the most common observations were developmental delay, speech delay, and craniofacial and skeletal anomalies, but medical records also revealed retarded growth, microcephaly, cardiac defects, seizures, psychiatric diagnoses, and more.
An added burden
However, cases were not only more likely to carry the 16p21.1 deletion; they were six times more likely than controls, including controls carrying some >500 kb CNV other than the 16p21.1 deletion, to carry “double hits”: a second CNV >500 kb or some other significant chromosomal abnormality. Compared to the general population, cases were 60-fold more likely to carry these double hits.
Two-hit carriers had more “severe or distinct” phenotypes than those who carried just the 16p21.1 deletion, say the authors. One case carried the 16p21.1 CNV plus a 22q13 terminal deletion that has been associated with Phelan-McDermid syndrome, featuring autism spectrum disorders and macrocephaly (Phelan et al., 2001; Durand et al., 2007). However, this case exhibited microcephaly and no autistic features (at two years, three months of age). Another case carrying a second deletion hit at 5q15q23.2, which has been associated with craniofacial features and benign intestinal polyps (Lindgren et al., 1992), exhibited more severe craniofacial and cognitive phenotypes than had been reported for the 5q15q23.2 deletion alone. A carrier of an 848 kb duplication on 14q32.1, who also carried a mutation in the BRAF gene consistent with a diagnosis of cardiofaciocutaneous syndrome (CFCS), exhibited a “diverse range of severe clinical features including craniofacial anomalies; complete agenesis of the corpus callosum; renal and cardiac defects; and Hirschsprung disease,” a far more severe clinical picture than that so far described for either CFCS or a case of a de novo 14q32.1 reported in association with schizophrenia (Xu et al., 2008).
Revisiting chromosome 1
Members of the research team are no strangers to the phenotypic puzzles posed by CNVs. In 2008, Eichler and coauthor Heather Mefford, both at Washington, led a study of patients with a 1.35 mb deletion or reciprocal duplication in a region of 1q21.1 spanning seven genes (Mefford et al., 2008; also see SRF related news story and Q&A with Eichler and Mefford). Patients in that study carrying the 1q21.1 deletion had mild to moderate developmental delay accompanied by dysmorphic features, and those carrying the duplication exhibited learning disabilities, mental retardation, or autism. However, a broad spectrum of other phenotypes—skeletal, facial, cardiac, ocular, neurological—that displayed no syndromic pattern led the authors to conclude that the observations “dispel the notion that rare copy-number variants will necessarily follow the one gene- (or one rearrangement-) one disease model.”
In light of the high incidence of two genetic hits in the 16p21.1 carriers in the current study, the researchers re-examined 25 cases carrying the 1q21.1 deletion, and found a 40-fold enrichment for a second large CNV in that group. Again the phenotypes were highly variable, “ranging from severe neurological deficit and craniofacial abnormalities to severe schizophrenia without cognitive impairment.”
An additional analysis of eight other syndromic and non-syndromic genomic disorders revealed that carriers of the 16p21.1 deletion ranked highest for second hits, followed closely by other CNVs for which variable phenotypes have been reported (e.g., 15q13.3, 16p11.2, and 22q11.2). The researchers found an inverse correlation between de novo CNVs and a second, with 16p12.1 as a case in point. Only a single de novo 16p12.1 deletion has been reported, but this CNV had the highest likelihood of the presence of a second hit in inherited CNV cases.
Though the current study focuses on double CNV hits, the authors offer a broad definition of their model, in which “[t]he second hit could potentially be another CNV, a disruptive single-base-pair mutation in a phenotypically related gene or an environmental event that influences the phenotype.” Overall, they write, “Our ‘two-hit’ model might...help to explain the significant comorbidity that exists among cognitive impairment, schizophrenia, and autism, and the underlying phenotypic variability reported for several recurrent deletions.”—Pete Farley.
Girirajan S, Rosenfeld JA, Cooper GM, Antonacci F, Siswara P, Itsara A, Vives L, Walsh T, McCarthy SE, Baker C, Mefford HC, Kidd JM, Browning SR, Browning BL, Dickel DE, Levy DL, Ballif BC, Platky K, Farber DM, Gowans GC, Wetherbee JJ, Asamoah A, Weaver DD, Mark PR, Dickerson J, Garg BP, Ellingwood SA, Smith R, Banks VC, Smith W, McDonald MT, Hoo JJ, French BN, Hudson C, Johnson JP, Ozmore JR, Moeschler JB, Surti U, Escobar LF, El-Khechen D, Gorski JL, Kussmann J, Salbert B, Lacassie Y, Biser A, McDonald-McGinn DM, Zackai EH, Deardorff MA, Shaikh TH, Haan E, Friend KL, Fichera M, Romano C, Gécz J, DeLisi LE, Sebat J, King MC, Shaffer LG, Eichler EE. A recurrent 16p12.1 microdeletion supports a two-hit model for severe developmental delay. Nat Genet. 2010 Feb 14. Abstract
Comments on News and Primary Papers
Comment by: Ben Pickard
Submitted 25 February 2010
Posted 25 February 2010
In their Nature Genetics paper, Girirajan et al. contribute to the slow shift of focus in the field of complex genetic disorders, away from population risks towards the risks specific to the individual. The driving force of this shift is the ongoing discovery of mutations more penetrant than the common single nucleotide polymorphisms (SNPs) studied in case-control association studies. Copy number variants (CNVs) and coding variants are the two principal classes of these mutations, typified by their relative rarity, frequent familiality, and generally higher odds ratio (OR) values indicative of their impact.
Considerable evidence from increased levels of comorbidity, dysmorphic features, brain structural changes, and latent endophenotypes suggests that early neurodevelopmental deficits can predispose to later neuropsychiatric conditions (Ross et al., 2006). This paper demonstrates how the phenotype in a single individual can be more closely linked with the causative genotype when the simultaneous action of two CNVs is considered. This provides some concrete evidence to explain how seemingly disparate diagnoses (for example, epilepsy, autism, schizophrenia and mental retardation) and variable penetrance can occur in different individuals who ostensibly carry the same pathogenic CNV (van Bon et al., 2009; Mefford et al., 2008)—specifically, it appears that a second CNV can translate a generalized neurodevelopmental pathology into a more specific, severe, and reproducible final clinical endpoint. Hence, the authors invoke a “two-hit” hypothesis, as originally applied by Knudson to explain the observed progression of familial forms of cancer (Knudson, 1971).
From a semantic point of view, the use of the “two-hit” terminology is a little loosely applied here. Originally, this described an inherited predisposition to cancer in the form of a tumor suppressor mutation which was coupled with a later, somatic, “second hit” oncogenic mutation that resulted in the tumor. In the current application of the term, both mutational changes are germline, present from the outset, and even a single CNV “hit” can still produce a phenotype. However, the comprehensive data in the paper and the analogy used are thought-provoking, as they highlight three etiological issues, outlined below, which may have wide applicability to complex genetic disorders.
1. A key observation from this paper is that the hypothetical Venn diagrams illustrating the overlapping relationships between the varied diagnoses and also between diagnoses and genomic variation would be highly complex. The phenotypic reach of each CNV may also be considerably greater than we currently suppose—limited by the developmental or neurological conditions that have yet to be comprehensively studied by comparative genomic hybridization. Conversely, the interpretation of shared genetic risk in association and epidemiological studies may have to be reassessed in the light of this paper’s findings. For example, the demonstration of a component of genetic contribution shared between bipolar disorder and schizophrenia may not be such a simple story (Lichtenstein et al., 2009). That component may be just a genetic contribution to generalized neurodevelopmental failure which predisposes to neuropsychiatric disorders in the context of “second hit,” disease-specific mutations (schizophrenia or bipolar disorder, in this case). Can a gene truly be called a “schizophrenia gene” if you haven’t discounted its role in other conditions first? This is reminiscent of the shared genetic contribution to underlying autoimmunity processes which is emerging from the genomewide association studies of diagnostically discrete disorders such as type I diabetes, Crohn’s disease, and rheumatoid arthritis (Baranzini, 2009).
2. The two-hit terminology and the involvement of neurodevelopmental deficits clearly imply a sequential pattern of CNV effect—early predisposition followed by later resolution of diagnosis. Mouse gene knockouts have shown that such staged action is indeed possible, and its confounding effects on animal models of disease led to the development of spatially and temporally controllable transgenic technologies (Gingrich et al., 1998). However, to prove this in the context of the CNV model will require both the individual spatiotemporal expression profiles of the deleted genes and also some measure of individual gene dosage sensitivity to be correlated with the apparent mode of CNV action (developmental or “modifier”).
3. The model’s restriction to just two contributory CNVs as presented here is, in my opinion, a by-product of CNV scarcity, visibility, and sample size, rather than a genuine biological phenomenon. As the authors state, the accuracy of genotype-phenotype correlations at the level of the individual will undoubtedly increase when common and rare SNP variation is also taken into account—potentially to the point where predictive diagnosis is realistic. Likewise, the authors’ suggestion that the second CNV has a “modifier” effect on the foundation CNV is enticing, and backed up by the ubiquity of the 16p12.1 across several genomic disorders, but it is still difficult at this stage to discount the equally compelling explanation that two CNVs just represent a simple increase in mutational load. However, this latter explanation has its own problems, as it suggests a continuum of illness dictated by additive genetic risk. Does this continuum start with mild developmental delay and end with neuropsychiatric illness—or vice versa?
Finally, it must not be forgotten that schizophrenia, bipolar, autism, and epilepsy can exist without comorbid traits and without evidence for developmental issues. Is this, then, an important dimension to be considered in the future sub-categorization of these disorders?
Ross CA, Margolis RL, Reading SA, Pletnikov M, Coyle JT. Neurobiology of schizophrenia. Neuron . 2006 Oct 5 ; 52(1):139-53. Abstract
van Bon BW, Mefford HC, Menten B, Koolen DA, Sharp AJ, Nillesen WM, Innis JW, de Ravel TJ, Mercer CL, Fichera M, Stewart H, Connell LE, Ounap K, Lachlan K, Castle B, Van der Aa N, van Ravenswaaij C, Nobrega MA, Serra-Juhé C, Simonic I, de Leeuw N, Pfundt R, Bongers EM, Baker C, Finnemore P, Huang S, Maloney VK, Crolla JA, van Kalmthout M, Elia M, Vandeweyer G, Fryns JP, Janssens S, Foulds N, Reitano S, Smith K, Parkel S, Loeys B, Woods CG, Oostra A, Speleman F, Pereira AC, Kurg A, Willatt L, Knight SJ, Vermeesch JR, Romano C, Barber JC, Mortier G, Pérez-Jurado LA, Kooy F, Brunner HG, Eichler EE, Kleefstra T, de Vries BB. Further delineation of the 15q13 microdeletion and duplication syndromes: a clinical spectrum varying from non-pathogenic to a severe outcome. J Med Genet . 2009 Aug 1 ; 46(8):511-23. Abstract
Mefford HC, Sharp AJ, Baker C, Itsara A, Jiang Z, Buysse K, Huang S, Maloney VK, Crolla JA, Baralle D, Collins A, Mercer C, Norga K, de Ravel T, Devriendt K, Bongers EM, de Leeuw N, Reardon W, Gimelli S, Bena F, Hennekam RC, Male A, Gaunt L, Clayton-Smith J, Simonic I, Park SM, Mehta SG, Nik-Zainal S, Woods CG, Firth HV, Parkin G, Fichera M, Reitano S, Lo Giudice M, Li KE, Casuga I, Broomer A, Conrad B, Schwerzmann M, Räber L, Gallati S, Striano P, Coppola A, Tolmie JL, Tobias ES, Lilley C, Armengol L, Spysschaert Y, Verloo P, De Coene A, Goossens L, Mortier G, Speleman F, van Binsbergen E, Nelen MR, Hochstenbach R, Poot M, Gallagher L, Gill M, McClellan J, King MC, Regan R, Skinner C, Stevenson RE, Antonarakis SE, Chen C, Estivill X, Menten B, Gimelli G, Gribble S, Schwartz S, Sutcliffe JS, Walsh T, Knight SJ, Sebat J, Romano C, Schwartz CE, Veltman JA, de Vries BB, Vermeesch JR, Barber JC, Willatt L, Tassabehji M, Eichler EE. Recurrent rearrangements of chromosome 1q21.1 and variable pediatric phenotypes. N Engl J Med . 2008 Oct 16 ; 359(16):1685-99. Abstract
Knudson AG. Mutation and cancer: statistical study of retinoblastoma. Proc Natl Acad Sci U S A . 1971 Apr 1 ; 68(4):820-3. Abstract
Lichtenstein P, Yip BH, Björk C, Pawitan Y, Cannon TD, Sullivan PF, Hultman CM. Common genetic determinants of schizophrenia and bipolar disorder in Swedish families: a population-based study. Lancet . 2009 Jan 17 ; 373(9659):234-9. Abstract
Baranzini SE. The genetics of autoimmune diseases: a networked perspective. Curr Opin Immunol . 2009 Dec 1 ; 21(6):596-605. Abstract
Gingrich JR, Roder J. Inducible gene expression in the nervous system of transgenic mice. Annu Rev Neurosci . 1998 Jan 1 ; 21():377-405. Abstract
View all comments by Ben Pickard