Secondary Schizophrenia (114 page)

penetrant (and hence of large effect) and rare in the
r

population, being specific to individual cases or sin-Further progress will depend on rigorous
gle families
[16].
It is probably better to assume that
phenotyping (including the use of alternative
there is a spectrum of risk variants of varying effect
phenotypes) and comprehensive LD mapping

sizes, including both common and rare alleles
[17].

of large, ethnically homogeneous samples.

r

In any case, the exact number of alleles, the degree
To guard against false-positive findings,

of allelic interaction, and each allele’s contribution to
replication is essential.

r

overall disease risk remain unknown
[18].
Moreover,
Functional studies of replicated variants

there may be individual differences in disease allele fre-should help clarify underlying molecular
quency within a gene (allelic heterogeneity), between
mechanisms that may lead to the

genes (locus heterogeneity), between samples (popu-development of targeted molecular
lation heterogeneity), and within phenotypes (clinical
treatments for SZ.

heterogeneity), making initial detection and replication difficult
[19].

Despite such formidable challenges, after 20 years
of intensive international effort, progress is occurring.

Genetic complexity of schizophrenia

Replicated SZ linkages are accumulating and converg-Psychiatry has always suspected that schizophrenia
ing on a modest number of loci across the genome,
(SZ) has a genetic predisposition. Kraepelin wrote in
and association studies have identified and confirmed
1919 that “dementia praecox not at all infrequently is
a number of susceptibility genes. Before discussing
familial, often appearing in brothers and sisters”
[1];

these findings, it is important to mention the clinical,
in 1916, Schulz used the family study method to eval-molecular, and analytic developments that have trig-

288

uate the validity of Kraepelin’s subtyping system for
gered this progress.

Chapter 23 – The status of genetic investigations of schizophrenia

Clinical phenotype

ceptibility variants may not respect diagnostic bound-aries
[35].
Thus, efforts to refine the SZ phenotype are
Diagnostic categories

justified. These include: (i) the application of alternative diagnostic systems, for example, Leonhard’s clas-Phenotype definition and measurement are funda-sification
[36]
with the catatonic SZ phenotype attract-mentally important for the success of gene identifica-ing some support from genetic studies
[37, 38];
(ii)
tion
[20]
. This represents a challenge for psychiatric
the derivation of symptom dimensions using instru-genetics, given the lack of biologically valid measures
ments such as the Operational Criteria Checklist for
currently available
[21].

Psychotic and Affective Illness (OPCRIT)
[39]
and the
Since its initial description, SZ has been defined
Lifetime Dimensions of Psychosis Scale (LDPS), a 21-on the basis of clinical criteria. In the fifth edi-item scale for rating the lifetime duration and severity
tion of his textbook Psychiatrie, Kraepelin described
of positive, bizarre, negative, disorganized, and mood
“dementia praecox” as a syndrome of psychotic symp-symptoms of psychotic disorders
[40];
(iii) identify-toms and chronic deterioration and contrasted it with
ing clinical characteristics shared by clusters of related
manic-depressive insanity, characterized by remission
individuals, using techniques such as latent class anal-between periods of mood disturbance
[1].
In 1911,
ysis (LCA); and (iv) the study of SZ endophenotypes.

Bleuler coined the term “schizophrenia” denoting a
Statistical studies provide support for several
“split between thinking and perception,” emphasiz-alternative phenotypes. A recent factor analysis of
ing blunted emotions, abnormal association of ideas,
OPCRIT symptoms identified five factors (mania, real-and impaired volition
[22],
and in 1933 Kasanin intro-ity distortion, depression, disorganization, and neg-duced “schizoaffective psychosis”
[23].
Schneider’s
ativity) that explained more of the disease charac-

“first rank” symptoms such as delusions of control,
teristics than diagnosis but the explanatory power
thought broadcast, and commentary hallucinations
of diagnosis was also high
[41].
A prominent latent
were also influential
[24].Thes
e concepts, combined
class analysis found six latent classes (classic SZ,
with a landmark study demonstrating significant
major depression, schizophreniform disorder, bipolar-international (US–UK) differences in SZ diagno-schizomania, schizodepression, and hebephrenia), all
sis
[25],
contributed to the development of more
of which, except for depression, showed increased risks
specific diagnostic criteria to increase reliability. Suc-for SZ in relatives
[42];
the most marked risk occurred
cessive waves of progress have produced the cur-in hebephrenic probands who were linked to chro-rent sets of diagnostic criteria, DSM-IV
[26]
and
mosome 8p significantly more often than others
[43].

ICD-10
[27].
Parallel advances have included: (i)
There is also some support for endophenotypes, quan-semistructured, standardized interviews such as the
titatively measurable traits situated along the pathway
Diagnostic Interview for Genetic Studies
[28]
and the
between the clinical phenotype and the distal geno-Family Interview for Genetic Studies
[29, 30];
and (ii)
type
[44].
Candidate endophenotypes for SZ
[45, 46]

comprehensive diagnostic approaches such as the Best
include (i) neurophysiological markers
[47],
for exam-Estimate Final Diagnosis (BEFD) procedure
[31]
in
ple, sustained attention deficits
[48],
P50 suppression
which two experienced research diagnosticians inde-deficits associated with the alpha7 nicotinic acetyl-pendently review all available data and then meet to
choline receptor gene, CHRNA7
[49];
(ii) neuroimag-determine a consensus BEFD
[32].

ing results
[50],
for example, bilateral fronto-striato-thalamic and left lateral temporal grey-matter volume
Alternative clinical phenotypes

deficits
[51];
and (iii) neurocognitive markers
[52],
for
A comprehensive approach to the SZ phenotype has
example, spatial working memory
[53],
and a perva-served the psychiatric genetics community well, as evi-sive cognitive deficit linked to chromosome 6p25–24

denced by the progress outlined later. Nevertheless,

[54].
One large-scale application of the endophenotype
it lacks precision and may represent, as Bleuler sug-paradigm is the multisite Consortium on the Genetics
gested, a group of diseases
[22].
In addition, quan-of Endophenotypes in Schizophrenia (COGS) study
titative traits may have greater power than categori-

[55]
. However, the endophenotype paradigm is not
cal diagnoses in genetic analyses, especially if analy-without difficulties
[56],
with little convincing evises only include individuals with extreme phenotypic
dence for any candidate showing a simpler genetic

values
[33, 34];
moreover, evidence suggests that sus-architecture than SZ, the requirement for special
Organic Syndromes of Schizophrenia – Section 3

measurement procedures constraining sample size,

underlying haplotype structure of the human genome
and the need to address the challenges of interlabora-consisting of discrete haplotype blocks (of tens to
tory variation
[57].

hundreds of kilobases) disrupted by sites of ancestral
recombination
[66, 67, 68]
. These data have revealed
profound variation in haplotype block length across

the genome, reflecting regional differences in the fre-The molecular genetic revolution has greatly acceler-quency of recombination events. The HapMap con-

ated the ability to search for disease-associated vari-sortium has also generated a vast and growing set of
ants. Before the mid-1970s, the only genetic mark-genetically validated markers for use in disease studies,
ers available for mapping diseases were blood groups,
together with bioinformatics tools such as Haploview
serum proteins, and HLA tissue types relating to only
[69]
, which provides a visual representation of LD –
a handful of chromosomal locations. With the discov-the nonrandom association of alleles at neighboring
ery of restriction enzymes, which cut double-stranded
markers – in regions of interest. Moreover, by utiliz-DNA at specific recognition sequences, restriction
ing LD, the HapMap enables the representation of the
fragment length polymorphisms (RFLPs) were used

majority of genomic variation in a region by typing
as genomewide markers for mapping disease genes.

only a subset of “tagSNPs” – SNPs that are highly cor-Polymerase chain reaction (PCR) methodology sub-

related with untyped SNPs in the same LD block and
sequently made it possible to amplify highly poly-provide similar information – thus reducing the sub-morphic microsatellite markers (STRPs), which super-stantial cost of conducting such experiments.

seded RFLPs. With the completion of the human

One of the assumptions underlying the HapMap

genome sequence
[58],
gene mapping now increas-approach is the common variant/common disease
ingly utilizes single nucleotide polymorphisms (SNPs)
hypothesis, which proposes that most risk variants for
involving a single base change. Compared with

common diseases are relatively common (frequency

RFLPs and STRPs, SNPs are: (i) easily detectable;
> 0.01–0.10) in the general population and are of mod-

(ii) amenable to high throughput, automated typing,
est effect size
[70, 71].
This hypothesis is supported by
with low error rates; (iii) abundant, occurring one
a meta-analysis of genetic association studies
[72]
and
in every 1000 base pairs, with several million, well-by multiple recent large-scale studies identifying sus-characterized SNPs now available for disease studies
ceptibility variants in a range of common diseases (dis-

[59];
(iv) uniformly distributed across the genome and
cussed later).

present within exons, introns, promoters, enhancers,
In addition to SNPs, previously unrecognized

and intergenic regions; and (v) stable, being less prone
larger-sized (non-SNP) structural variants (size range:
to mutation than other types of polymorphism
[60].

∼
1kb – 3Mb) have recently been identified
[73].
These
In the wake of the Human Genome Project, other

include copy number variations (CNVs), inversions,
projects are making substantial contributions to clari-and insertion/deletion variants (InDels). CNVs, for
fying the genetic basis of disease. The National Human
example, can contain entire genes and may influence
Genome Research Institute’s ENCODE (Encyclopedia

nearby genes, making them important candidates for
of DNA Elements) was launched in 2003 to identify all
disease gene investigations.

functional elements in the human genome sequence

[61]
. The International HapMap Consortium is devel-

oping a map of common patterns of DNA sequence

variation by determining the genotypes of more than
one million SNPs, their frequencies, and the degree
Linkage analysis

of association between them, in DNA samples from

Linkage describes the nonrandom segregation of dis-populations with African, Asian, and European ances-ease and marker loci, leading to co-inheritance of the
try
[62].
This exploits: (i) the concept of the “haplo-same marker allele by multiple affected family mem-type” (the linear arrangement of closely linked alleles
bers; the linked marker allele is consistent within fam-inherited as a unit on one member of a chromosome
ilies but can differ across families
[20,
74].
Linkage
pair), which may provide statistical power advantages
has been widely applied in complex genetic disorders
over the single locus approach to identify genetic vari-such as SZ, following its success in identifying the vari-