Reimagining psychoses: An agnostic approach to diagnosis
Matcheri S. Keshavana,⁎, Brett A. Clementzb, Godfrey D. Pearlsonc, John A. Sweeneyd, Carol A. Tammingad
aBeth Israel Deaconess Hospital, Harvard Medical School, Boston MA, United States
bUniversity of Georgia, Athens GA, United States
cHartford Hospital, Yale School of Medicine, Hartford CT, United States
dUT Southwestern Medical School, Dallas TX, United States
a b s t r a c ta r t i c l ei n f o
Received 21 November 2012
Received in revised form 12 February 2013
Accepted 19 February 2013
Available online 14 March 2013
Objectives: Current approaches to defining and classifying psychotic disorders are compromised by substantive
heterogeneity within, blurred boundaries between, as well as overlaps across the various disorders in outcome,
treatment response, emerging evidence regarding pathophysiology and presumed etiology.
Methods: We herein review the evolution, current status and the constraints posed by classic symptom-based
diagnostic approaches. We compare the continuing constructs that underlie the current classification of psycho-
ses, and contrast those to evolving new thinking in other areas of medicine.
Results: An important limitation in current psychiatric nosology may stem from the fact that symptom-based
diagnoses do not “carve nature at its joints”; while symptom-based classifications have improved our reliability,
they may lack validity. Next steps in developing a more valid scientific nosology for psychoses include a) agnostic
deconstruction of disease dimensions, identifying disease markers and endophenotypes; b) mapping such
markers across translational domains from behaviors to molecules, c) reclustering cross-cutting bio-behavioral
data using modern phenotypic and biometric approaches, and finally d) validating such entities using etio-
pathology, outcome and treatment-response measures.
Conclusions: The proposed steps of deconstruction and “bottom-up” disease definition, as elsewhere in medicine,
in predicting outcome, treatment response and etiology, and identifying novel treatment approaches.
© 2013 Elsevier B.V. All rights reserved.
The nosology of psychiatric disorders continues to evolve, but
remains embroiled in active controversy. Given the overlapping be-
havioral boundaries of psychiatric disorders, the validity of traditional
psychiatric diagnoses remains uncertain. This has limited the establish-
mentof neurobiologicalmodels that can guidethedevelopment ofnew
and more effective treatment strategies; incorporation of biological/
genetic concepts in efforts such as Diagnostic and Statistical Manual
(DSM) to refine diagnosis continues to move more slowly than needed.
Perhapsnowhere arethese concerns more stark thaninthe ongoing
major dispute, spanning over a century, about how psychotic disorders
are optimally defined and classified. This is not merely a theoretical
question, but one with immense practical implications for research
into the biological basis of psychotic diseases and the development of
novel, more effective diagnostic and predictive tests for clinical practice
as well as of treatment targets. In this paper, we review the history of
this debate, identify the limitations of current approaches, and discuss
possible directions for the future in light of emerging new data on the
neurobiological substrate of psychotic illnesses as well as evolving
approaches to classification of human diseases in the rest of medicine.
2. Classification of psychoses: tracing the past
Psychotic disorders, though not schizophrenia, were recognized at
tions of insanity before the Christian era (Jeste et al., 1985). The identi-
fication of these disorders as medical diseases and various attempts at
classification of psychoses, however, did not begin until the nineteenth
century. One approach, whose proponents were German and French
physicians (the “splitters”), was to divide the psychotic disorders into
paranoides (Kahlbaum, 1874), hebephrenia (Hecker, 1871, Cited in
Sedler, 1985) and folie circulaire characterized by cyclical changes in
mood (Falret, 1854). Another approach, by the “lumpers” such as
Griesinger (1845), was to view all psychoses as reflecting a single
The nosology of psychoses evolved over the first half of the 20th cen-
tury in three phases (Fig. 1). The first phase was dominated by eminent
contemporary theorists; Emil Kraepelin (1899, 1921) (Kraepelin 1921),
who kept meticulous longitudinal notes on every patient in his clinic on
index cards, observed that patients with catatonia, hebephrenia, and
Schizophrenia Research 146 (2013) 10–16
⁎ Corresponding author. Tel.: +1 2488851057.
E-mail address: email@example.com (M.S. Keshavan).
0920-9964/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
Contents lists available at SciVerse ScienceDirect
journal homepage: www.elsevier.com/locate/schres
paranoid dementia were all characterized by an adolescent or early adult
onset, chronic course and a tendency towards functional decline, mental
dullness and dementia. He distinguished this “dementia praecox” group
from manic depressive insanity, which was episodic with a more favor-
able outcome. This dichotomy continues to hold sway till today, though
Kraepelin himself had doubted this distinction (Berrios and Hauser,
1988). Consistent with the medical models prevalent at his time,
Kraepelin argued that dementia praecox, like neurosyphilis, constituted
a unique degenerative disease entity whose core feature was a
persistent, declining course (though he acknowledged in his later writings
when the medical models of psychiatric illness were dominant and the
hope was that a neuropathological basis would eventually be discovered.
Bleuler (1950) considered the course and outcome to be variable, but
defined schizophrenia (a term coined by him) by the core features of
ous affect, ambivalence, and autism (Bleuler's 4 As); delusions and
hallucinations were thought to be “accessory symptoms”. In addition to
declining course, he noted the interesting combination of avolition and
dissociative pathology as defining clinical manifestations. Bleuler viewed
this entity as a collection of disorders, the “Gruppe der Schizophrenien”,
though the field continued to consider this illness as a single disease
entity. Jaspers (1946), a psychiatrist and phenomenological philosopher,
believed that the core impairment in schizophrenia was one of “un-
understandability” and an impairment in empathic communication. Kurt
Schneider (1959), who focused on the form, rather than on the content
this illness. The advent of psychoanalysis (which both raised therapeutic
optimism and led to a focus away from categorical diagnosis), coupled
led to an alternative view, especially in the U.S., that psychoses, like all
dividual to one or other stress, i.e. the psychobiological model of the em-
inent Swiss–American psychiatrist Adolph Meyer (1957). These diverse
definitions of schizophrenia were labeled “schizophrenic reaction” in
the first edition of the diagnostic and statistical manual (DSMl) of
mental disorders, and were variably applied across different countries.
A major weakness of this “eminence-based” approach was that diagnoses
of psychotic disorders lacked reliability, which led to different groups
of psychiatrists utilizing differing diagnoses based on what school of
thought they endorsed. This was evident in a well-known U.S.–U.K.
study which showed a disparity between the U.S./U.K. clinicians in
diagnosing psychosis; when presented with the same cases, American
psychiatrists diagnosed schizophrenia very frequently while British
psychiatrists diagnosed fewercasesasschizophrenia and more asbipolar
illness (Kendell et al., 1971).
This motivated the next phase of nosology: of the development of
criteria (RDC) (Spitzer et al., 1975) and DSM-III (1980), not only for psy-
choses, but also for other psychiatric diagnoses. While the operational
criteria inherent to the DSM-III and then DSM-IV significantly addressed
the challenge of diagnostic reliability, there remained the problem of
validity, i.e. whether the syndromes as defined actually represent truly
distinct and independent disorders (whether they were capturing
“pure types” as envisioned by ancient Greek thinkers such as Plato) or
clinically similar but etiopathologically distinct entities such as “dropsy”.
Unfortunately, unlike many other branches of medicine, psychiatry
lacked clinico-pathological confirmation of symptom-based diagnoses.
The need for an evidence-based approach to classification of psycho-
ses led Robins and Guze (1970) at the Washington University to identify
four sets of external “validators” of a distinct and singular psychiatric
category, a) phenomenology of cross-sectional symptoms; b) course of
illness, (e.g. a chronic, persistent course defined schizophrenia while
a recurrent course more likely accompanied an affective psychosis);
c) family history, (with the provision that patients with a particular
Kraepelin (1889) gouped
psychoses into Dementia
Praecox and Manic
Insel and Cuthbert 2009
Depressive Insanity based
DSM-IIIR (1987) drops
“organic” and “functional”
DSM-5 to be
Kendell et al 1971
(first rank) symptoms
1850s 1900s 1950s 1960s 1970s 1980s 1990s 2000s 2010s and beyond
Greek and Indian
Dementia Praecox renamed as
Schizophrenia (Bleuler 1911)
based on “splitting” of mental
and Guze 1974)
criteria for reliability
Many psychotic disorders:
Catatonia (Kahlbaum 1874);
Hebephrenia (Hecker 1871);
Folie Circulairre (Falret 1854)
At the time
Atheoretical, for reliability
Atheoretical, toward validity
Fig. 1. The history of nosology of schizophrenia.
M.S. Keshavan et al. / Schizophrenia Research 146 (2013) 10–16
psychotic illness tend to have family members with the same illness
type); and d) laboratory tests that would distinguish between separate
diseases, and between the diseased and healthy individuals. Since no
laboratory tests were then (or now) available to distinguish psychiatric
disorders, another criterion was added, that of treatment response. In
developing these criteria, it was observed that a significant proportion
of DSM-III manic subjects exhibited psychotic symptoms, including
Schneiderian first-rank symptoms (Carpenter et al., 1973; Abrams and
Taylor, 1981). For this reason, like Kraepelin, the Washington University
This was reflected in DSM-IIIR's approach to schizophrenia diagnosis
where clinical course became a defining characteristic of this illness,
an approach maintained in DSM-IV (American Psychiatric Association,
3. Constraints of the current approach to the classification
While the evidence-based approach, championed by Robins and
example is that of defining a disorder first based on symptoms and
course, and then using course as an “external” validator. Second, clinical
external validators; for example, psychotic patients with a chronic, per-
sistent course often have prominent affective symptoms and family his-
tories of bipolar disorder (Ivleva et al., 2010). Even in formal familial
characterizations, schizophrenia does not “breed true” (Lichtenstein
et al., 2009). Third, there is now extensive evidence of symptomatic,
neurocognitive, neurobiological and genetic overlap between schizo-
genetic and phenomenological, also exists between schizophrenia and
multiple other neuropsychiatric disorders including attention deficit
disorders, mental retardation and autism (Craddock and Owen, 2010).
Evidence for a “point of rarity” has been little or inconsistent between
the major groups of psychotic disorders (Kendell and Jablensky, 2003;
Keshavan et al., 2011) in fact, quantitative measures have consistently
revealed a frank lack of these “points of rarity” between psychotic
illnesses. This overlap led to the concept of schizoaffective disorder
(Kasanin, 1994), a construct of questionable reliability and validity
(Jager et al., 2011). While it can account for the co-occurring affective
symptoms and yet preserve the categorical approach to classification
of psychotic disorders, the entity of schizoaffective disorder has only
added to nosological confusion. Finally, the focus on categorization has
led to a diagnostic inflation (DSM-I had 106 diagnoses; DSM-II, 182;
DSM-IIIR, 292, DSM-IV, 297). As more sophisticated biological markers
have emerged, there appears to be consistent overlap between-
disorders among various biomarkers (Ethridge et al., 2012; Meda et al.,
These problems may result from the difficulty of basing biological
models in large part on “top-down” constructs based on phenomenolog-
biological data to support nosological models when there is so little
diagnostic specificity of clinical biological observations. These concerns
are important across several major psychiatric disorders, where genetic
and other biological factors are likely particularly important in these
substantially heritable, brain-based illnesses, and where implications for
treatment and prognosis are especially crucial. Often, biomarkers show-
ing initial promise, such as the dexamethasone suppression test (Carroll
of lack of diagnostic specificity (American Psychiatric Association, 1987).
by symptoms (van Praag, 2008). In other words, we might have
inadvertently “reified” (treated as if valid, despite lack of evidence) the
phenomenology-based classifications in a way that symptom criteria
trump biological observations (Hyman, 2010). Let us consider a medical
analogy. In days before modern microbiology, syndromes characterized
by cough, breathlessness and fever were classified based on symptoms
and course into those cases with chronic, persistent symptoms including
weight loss and others defined by acute intermittent symptoms without
weight loss with further sub-typing into “wet” and “dry” variants. If the
newly discovered tuberculin test did not distinguish between these syn-
dromes and was discarded due to “lack of diagnostic specificity”, medical
progress would have clearly been hindered. It is not difficult to see how
the symptom based classification may in fact have contributed to a) the
large co-morbidities across disorders, large proportions of disorders in
lidity such as schizoaffective disorder; b) the failure of finding robust
across-diagnosis gene differences despite high heritability, c) the failure
to identify biomarkers for predictive and diagnostic purposes and d) the
possibility of excluding too many “real-world” patients that do not meet
strict DSM criteria in clinical trials thereby limiting progress in treatment
development. The inability to find biomarkers of predictive utility, which
largely results from the lack of a biological “gold-standard”, has also been
compounded by the accumulation of non-replicated or partially replicat-
ed findings in underpowered studies, “significance chasing” (e.g. p b .05
as the goal rather than seeking clinically meaningful effect sizes), and
the excessive focus on extreme comparisons (e.g. comparing typical
patients to “squeaky-clean” controls) (Kapur et al., 2012).
It is not the case that clinicians in the field have willfully overlooked
have been critical, but rather there has been a dearth of replicated bio-
logical characteristics of functional brain diseases. Indeed, the scientific
evidence of anatomic, physiologic and systems neuroscience for
mammalian brain has developed gradually and only recently, leaving
a ready biological foundation. It is tremendously gratifying to see
this situation shifting under the growth and sophistication of modern
4. Toward an empirically based “reverse” nosology
Clearly,thefundamentalproblem innosologyofpsychotic disorders
is that symptom-based diagnoses fail to mirror nature. What is the way
ahead? While several authors have proposed biologicallybased classifi-
cations of psychiatric disorders (e.g. Kendler, 1990; van Praag, 2008),
few have as yet gained traction, most likely because of a contemporary
lack of adequate data in biological psychiatry, resistance to change in
time-honored diagnostic schemes, or both. Thankfully, in recent years
there has been a groundswell of new data on the neurobiology and
etiology of psychotic disorders (for reviews see Insel, 2010; Keshavan
et al., 2008) that motivate a fresh effort to develop a neuroscience-
based nosology. Several conceptual advances have also occurred that
can guide steps toward a biologically based nosology.
4.1. Dimensions and endophenotypes
The first step in this iterative deconstructive–reconstructive process
is to define the dimensional characteristics of a broad spectrum of
psychiatric disorders (such as “idiopathic” or primary psychoses) in
a manner agnostic to any existing diagnostic categories. It has been
argued by some investigators that a dimensional approach may
better represent clinical reality than categorical approaches (Allardyce
et al., 2007). The categorical structure of psychotic spectrum can be
deconstructed by identifying quantifiable domains of psychopathology
(Strauss et al., 1974) or dimensional measures (e.g. working memory
improvement) that cut across diagnoses; these may better relate to bi-
ological dimensions, heritable and may thus be closer to the underlying
genetic liability (endophenotypes, or intermediate phenotypes, e.g.
reduced power of gamma synchrony) (Gottesman and Gould, 2003;
M.S. Keshavan et al. / Schizophrenia Research 146 (2013) 10–16
Allen et al., 2009) This approach may also reduce heterogeneity and
increase effect sizes, making it easier to identify risk genes of interest.
4.2. Translational mapping across domains
Thesecond stepwould beto relatetheendophenotypes across struc-
tural, functional, physiological and molecular domains. An important
paradigm shift in this regard is the recent initiative to identify transla-
tional behavioral domains that cut across several neurobiological units
of analyses, the research domain criteria (R-DoC) (Insel and Cuthbert,
2009). This approach seeks to define behavioral measures which
are judged to be ‘dimensional’ and can be translated to physiological
alterations in known circuits, genes, or neurotransmitter systems. Such
an approach may yield related endophenotypic characteristics (or
“extended endophenotypes” (Prasad and Keshavan, 2008)). The RDoC
are not designed for or based on traditional psychopathology, but on
normal behavioral constructs which are mediated by identifiable brain
circuit(s). It is worth noting that none of the RDoC behavioral constructs
put forwardto daterelatetofeatures of psychosis suchas hallucinations,
delusions, or thought disorder. However, abnormalities in circuit-based
brain functions underlying RDoC domains may yield cross-cutting, func-
tionally meaningful syndromal entities (e.g. aberrant salience), rather
than a specific diagnosis (such as schizophrenia), that are closer to the
genetic substrate of the disorders. An example would be to relate work-
ing memory alterations to alterations in gamma synchrony; these may
tyric acid (GABA) pathways (Keshavan et al., 2008). Availability of large
datasets with phenotypic, endophenotypic and genotypic information
across the diagnostic spectrum is likely to facilitate such a strategy. Re-
cent large scale NIH funded consortia such as Bipolar-Schizophrenia
Network for Intermediate Phenotypes (B-SNIP) and the Consortium
on the Genetics of Schizophrenia (COGS) have adopted such thinking
in clarifying phenotypic boundaries across psychotic spectrum dis-
orders (Greenwood et al., 2011; Ethridge et al., 2012; Meda et al.,
al diagnostic categories are standardized, integrated and shared across
investigators. Recent initiatives in this direction include the Psychiatric
GWAS Consortium, whichhas soughtto harmonizelarge setsofalready
collected genotypic and phenotypic data (Sullivan, 2010), and the
Connectome project, which has brought together several functional
imaging groups from around the world to better understand functional
brain networks (Biswal et al., 2010).
4.3. Agnostic reclustering
While dimensional approaches offer increased power in identifying
cross-cutting and valid disease markers, clinicians still need some
formal way to categorize patients so asto inform prediction of outcome
and treatment response. The key goal would therefore be to examine
how the endophenotypes identified in the psychotic disorder spectrum
a) cluster among themselves, b) differ from healthy individuals
and from patients who have non-psychotic disorders, and c) define a
homogenous group within the spectrum sharing a unique biological
Several approaches to the agnostic reclustering for psychiatric taxon-
omy are available. First, bootstrap taxometric procedures, pioneered by
Meehl (1995), can be used to test whether a latent structure, e.g., the
structure underlying variation in schizotypal symptoms, is taxonic
(categorical) or is rather dimensional. Using such an approach, Meehl
suggested that it is possible to identify highly correlated features
(schizotaxia) from commingled (rather than pure) samples of schizo-
phrenia andnon-schizophrenia subjects whomayappeartohave “seem-
ingly unrelated” symptoms. Meehl (1995) boldly suggested that further
revisions of psychiatric classificationshould bebased ontaxometric anal-
ysis of empirical data rather than on committee decisions based on clini-
cal impressions. There is some evidence that prediction of behavior and
treatment response may actually be more accurate if based on empirical
(actuarial) data rather than clinical judgment (Dawes et al., 1989).
Meehl's prescient efforts to develop an empirical taxonomy of psychoses
genetic data related to psychiatric disorders that are now becoming
Reducing clinical heterogeneity can potentially increase the power
to identify susceptibility genes. Using latent class analyses, one can
empirically derive homogenous classes of psychotic illness. In a
study using this approach in large Irish pedigrees, Fanous et al.
(2008) identified several latent classes, and found suggestive linkage
in chromosomal regions, which had yielded little evidence of linkage
when examined using traditional clinical criteria. An important step
forward in this approach is to identify disease categories or dimen-
sions across multiple datasets, i.e. in the clinical, pathophysiological
and etiological domains. An example of this approach is the parallel
independent component analysis (Para-ICA). Para-ICA derives clus-
ters of interacting genes and their associated physiologic processes
such as fMRI or EEG networks. This approach is ideal for common dis-
ease/common variant illnesses, where multiple genes of individually
low contribution to disease risk are presumed to interact and under-
pin endophenotypes. The derived clusters can then be examined by
physiologic pathway tools such as Ingenuity (http://www.ingenuity.
com/) or KEGG (http://www.genome.jp/kegg/pathway.html)
illuminate their pathophysiological significance (e.g. dopamine
transmission, axon guidance, DISC1 interactome). The utility of this
approach has been demonstrated in Alzheimer's disease (Meda
et al., 2012b); genetic determinants of the EEG P300 auditory oddball
response were identified by Liu et al. (2009) in healthy individuals
and linked to a shared cluster of genes in the adrenergic and dopami-
nergic pathways. One of the same single nucleotide polymorphisms
(SNPs) reported in this study was independently replicated as the
most significant variant in a larger-scale investigation of schizophre-
nia patients, (who classically show reduced P300 amplitude as an
endophenotype) by Decoster et al. (2012).
4.4. Validation for clinical utility
The goal of any classification of disease is ultimately to be able to
predict outcome and treatment response, and to facilitate discovery of
clusters of disease characteristics by mapping them back to clinical
features and “external” validators such as course, treatment response
and etiological data (i.e. variation in genetic and environmental risk
factors). In doing so, the clinical characteristics that map on to the
biological characteristics (i.e. biotypes) may well not resemble conven-
tional diagnostic categories but rather reflect cross-cutting dimensions
(e.g. fear, impulsivity, and aberrant salience) (Fig. 2).
5. Lessons from elsewhere in medicine
A biologically derived classification does not mean that one needs
to abandon phenotypically based nosology altogether. Broad clinical
categories (e.g. psychosis spectrum disorders or asthma or epilepsy)
will still be needed as “open constructs” for administrative, fiscal and
legal purposes. Moreover, behavioral symptoms will always need to
be identified and managed, and therefore, are informative. However,
finer-grained divisions of psychotic disorders into schizophrenia,
schizoaffective, and psychotic affective disorders may well have outlived
their utility for reasons other than administrative convenience. It may be
useful, while the relationships between symptom-based vs. brain
behavior-based symptoms are being determined, to employ a two-
tiered approach in each patient, with a familiar, conventional clinical
diagnosis alongside a biologically based categorization (ora symptom di-
mensional score). Examples in medicine include classification of tumors
into categories based on disease phenotype/severity (e.g. variations in
M.S. Keshavan et al. / Schizophrenia Research 146 (2013) 10–16
the extent of tumor, node and metastasis, or TNM), histopathological
teristics that may determine treatment response (e.g. breast cancer with
positive estrogen receptor expression). In fact the utility of this approach
hasbeenmostrecentlyillustratedbya furtherrefinementof thisprocess,
where breast cancers with identical histology are further subdivided ge-
netically, leading to a reclassification of breast cancer into approximately
20 unique disorders (Curtis et al., 2012).
Bycomparisontopsychiatry, in therestof medicine, classification of
diseases has historically developed based on analyses of clinical and
pathological data. This has served well to establish syndrome patterns
that allow specific diagnosis and treatment selection. However, there
is increasing recognition of lack of sensitivity in detecting preclinical
disease, and a frequent lack of specificity in defining the phenotype
unequivocally. In fact, a recent report by the National Academies of Sci-
ences (National Research Council (U.S.). Committee on A Framework
for Developing a New Taxonomy of Disease., 2011) called for revising
the classification of all diseases based on emerging molecular data,
clearly a step forward from the traditional focus on symptomatology.
In the post-genomic era with the expanding body of transcriptomic,
proteomic, and metabolomic data sets in health and disease, it is
possible to more precisely characterize the molecular underpinnings of
human disease to better understand the interface between disease sus-
ceptibility and environmental influence; heterogeneity of phenotypic
manifestations of disease, to predict outcome more precisely, and to
individualize treatment for optimal efficacy. To this end, sophisticated
statistical modeling and biometric approaches are being used in the
recent years to delineate illness boundaries integrating data across
phenomic, endophenomic and genomic/enviromic domains of medical
disease (Barabasi, 2007; Loscalzo et al., 2007). Thus, multi- “omics”
data sets in health and disease can be mined to characterize human dis-
ease more precisely at clinical as well as preclinical levels and identify
“disease networks” between multiple seemingly disparate phenotypes
such as schizophrenia, bipolar disorder and autism (Rzhetsky et al.,
2007). Integration of high throughput data from heterogeneous datasets
across multiple domains allows identification of emergent, hitherto un-
known relationships between disease related variables, and thereby
helps redefine complex disease entities. One such approach is “rede-
scription mining”; using such an approach, the heterogeneous chronic
fatigue syndrome has been resolved into etiologically meaningful
subtypes (Waltman et al., 2006). These approaches can elucidate the
relations between genetic and environmental liability and the disease
phenotype, allow more effective prediction of disease prognosis, and
help personalize treatment in each patient for optimal efficacy.
A recent paradigm in medicine is identification of biomarkers that
help stratify broad-illness phenotypes into finite numbers of treatment-
relevant subgroups, even before etiopathology is fully elucidated (strati-
fied medicine, (Trusheim et al., 2007)). A related emerging concept is
“precisionmedicine”: recent advances in medicine are making it possible
to carry out state-of-the-art molecular profiling that can allow precise
vidual patient. A well known example is the use of human epidermal
growth factor subtype 2 (HER2) receptor expression as a marker of
poor outcome in breast cancer (Slamon et al., 1987); this marker also
led to the development of a targeted treatment (Herceptin) in this breast
cancersubtype(Smithet al., 2007). Another example is theidentification
of patients with lung adenocarcinoma with epidermal growth factor
receptor mutations who respond to specific treatment with Afatinib
(Yang et al., 2012). Non-smoking young Asian women with non-small
cell lung cancer with such mutations seem to be good candidates
for this treatment (Bethune et al., 2010), clearly an example of the
value of stratified medicine.
With the increasing knowledge base on susceptibility genes and
is becoming increasingly possible. Several strategies that are already
promising can now be cited.
First,sub-populations ofpsychoticspectrum disorders canbecharac-
terized based on treatment response of psychotic symptoms to standard
dopamine blockers; thus, increased presynaptic dopamine synthesis
capacity as measured by positron emission tomography, appears to
to those that are resistant to treatment (Demjaha et al., 2012; Howes
et al., 2012).
Second, molecular targets, derived from genetics and translational bio-
markers, can be identified that define cross-cutting clinical targets.
Examples include putative procognitive agents, currently in phase 1, 2
or 3 studies that may act on glutamatergic pathways (glycine transporter
inhibitors, AMPA allosteric modulators MGlu receptor agonists), nicotinic
Fig. 2. A proposed approach to disease identification using biomarkers agnostic to conventional diagnostic categories.
M.S. Keshavan et al. / Schizophrenia Research 146 (2013) 10–16
receptors (alpha7 nicotinic agonists), and GABA (GABA8 agonists) (Insel,
2012). Another example in this regard is the recent observations,
in genome-wide association studies, of markers occurring within the
CACNA1C gene encoding the alpha subunit of the calcium channel (such
as the CACNA1C gene) being associated with schizophrenia and bipolar
disorder. A potentially translatable question is whether mood lability in
such patients might preferentially respond to calcium channel blockers,
which have shown early promise in bipolar disorder trials (Keers et al.,
2009). The implication of genes related to the immune system in schizo-
illness might preferentially respond to anti-inflammatory agents (Muller
and Dursun, 2011).
Finally, novel treatment targets may also stem from putative envi-
ronmental risk factors. One example is the potential value of an antiviral
seropositive for herpes simplex virus 1 (Prasad et al., 2012). Such a
stratified approach is much needed for developing tests in psychiatry
of predictive and heuristic value. Such an effort may also help clarify
the current confusion and improve utility of the nosology of psychotic
6. Conclusions and future steps
One may wonder why it has taken so long for psychiatry to discard
single disease models, despite Bleuler's views to the contrary over a
hundred years ago. The reasons might lie in the relative recency of de-
velopments in neuroscience; during much of the 20th century, clinical
observations lacked substantive biological substrates on which to base
biological interpretations of clinical observations. Major developments
in neuroscience concepts of relevance to understanding psychiatric
disease, such as plasticity-induced brain dynamics underlying memory
processing may likely be fundamental to understandingdevelopmental
pathology in psychosis (Malenka and Bear, 2004); new approaches
such as transcranial magnetic stimulation and transcranial direct cur-
rent stimulation can help map and modulate brain plasticity in vivo
(McClintock et al., 2011). Realization of the extraordinary complexity
of genetic variation and the discovery of overlapping and common
ders has led to revisit traditional notions of disease architecture
(Karayiorgou et al., 2012). In vivo non-invasive studies of circuit-level
physiology is now feasible using optogenetic techniques (an approach
to precisely manipulate neuronal activity in moving animals by turning
on or off light sensitive proteins), which will likely revolutionized
mapping of endophenotypes (Deisseroth, 2012). Phenotypic character-
ization of diseases at the neuronal level will accelerate using induced
pluripotent stem cells, (IPSCs); IPSCs, derived from cells of adult
patients, can provide physiologically reliable cellular model systems to
study psychiatric disorders (Brennand et al., 2011).
sufficient understanding to support effective, testable biological models
of clinical phenomena in psychosis and to determine which provide
value in defining mental diseases. However, no major revisions of the
nosology of psychoses are expected in the near future. While diagnostic
reliability has been achieved in good measure by DSM criteria, and as
neuroscience moves the field toward reconstructing valid subtypes of
psychosis construct, an important consideration of any classificatory
system must be clinical utility; its implementability in routine clinical
settings and capability of distinguishing between different forms of
psychiatric illness (Tandon and Maj, 2008; Tandon, 2012) is the
prime consideration of the forthcoming DSM-5 (http://www.dsm5.org
expected to be released in May 2013). A key aspect expected in the
next iteration of DSM is the introduction of easily measurable psycho-
pathological dimensions. Other proposed changes include elimination
of the classic “Kraepelinian” subtypes, longitudinal (rather than cross-
sectional) characterization of schizoaffective disorder, elimination of
the special status of specific positive symptoms (such as bizarre
delusions), and the inclusion of attenuated psychosis syndrome for
further study. DSM5 will remain atheoretical though efforts are being
made to base the classification in an etiopathophysiological framework,
neuroscientific and genetic information that emerges in the future
toward a nosology based on causes and mechanisms.
Role of funding source
This study was supported in part by NIMH grants MH 78113 (Keshavan); R01
MH077945 (Pearlson);MH 77852 andMH77851(Tamminga);andMH077862(Sweeney).
Matcheri Keshavan wrote the first draft of the manuscript. All authors contributed
to and have approved the final manuscript.
Conflict of interest
There are no other conflicts relevant to this work.
The authors benefitted from discussions with Drs Shitij Kapur, Rajiv Tandon and
Henry Nasrallah in formulating some of the concepts in this paper. Neeraj Tandon
and Jai Shah provided helpful comments on the manuscript.
Abrams, R., Taylor, M.A., 1981. Importance of schizophrenic symptoms in the diagnosis
of mania. Am. J. Psychiatry 138 (5), 658–661.
Allardyce, J., Suppes, T., Van Os, J., 2007. Dimensions and the psychosis phenotype.
Int. J. Methods Psychiatr. Res. 16 (Suppl. 1), S34–S40.
Allen, A.J., Griss, M.E., Folley, B.S., Hawkins, K.A., Pearlson, G.D., 2009. Endophenotypes
in schizophrenia: a selective review. Schizophr. Res. 109 (1–3), 24–37.
American, P., Association, 1987. The dexamethasone suppression test: an overview of its
current status in psychiatry. the APA Task Force on laboratory tests in Psychiatry.
Am. J. Psychiatry 144 (10), 1253–1262.
Association, A.P., 1994. The Diagnostic and Statistical Manual of Mental Disorders.
American Psychiatric Association.
Barabasi, A.L., 2007. Network medicine—from obesity to the “diseasome”. N. Engl. J. Med.
357 (4), 404–407.
Berrios, G.E., Hauser, R., 1988. The early development of Kraepelin's ideas on classification:
a conceptual history. Psychol. Med. 18 (4), 813–821.
Bethune, G., Bethune, D., Ridgway, N., Xu, Z., 2010. Epidermal growth factor receptor
(EGFR) in lung cancer: an overview and update. J. Thorac. Dis. 2 (1), 48–51.
Biswal, B.B., Mennes, M., Zuo, X.N., Gohel, S., Kelly, C., Smith, S.M., Beckmann, C.F.,
Adelstein, J.S., Buckner, R.L., Colcombe, S., Dogonowski, A.M., Ernst, M., Fair, D.,
Hampson, M., Hoptman, M.J., Hyde, J.S., Kiviniemi, V.J., Kotter, R., Li, S.J., Lin, C.P.,
Lowe, M.J., Mackay, C., Madden, D.J., Madsen, K.H., Margulies, D.S., Mayberg, H.S.,
McMahon, K., Monk, C.S., Mostofsky, S.H., Nagel, B.J., Pekar, J.J., Peltier, S.J.,
Petersen, S.E., Riedl, V., Rombouts, S.A., Rypma, B., Schlaggar, B.L., Schmidt, S.,
Seidler, R.D., Siegle, G.J., Sorg, C., Teng, G.J., Veijola, J., Villringer, A., Walter, M.,
Wang, L., Weng, X.C., Whitfield-Gabrieli, S., Williamson, P., Windischberger, C.,
Zang, Y.F., Zhang, H.Y., Castellanos, F.X., Milham, M.P., 2010. Toward discovery
science of human brain function. Proc. Natl. Acad. Sci. U. S. A. 107 (10), 4734–4739.
Bleuler, E., 1950. Dementia Praecox, or the Group of Schizophrenias International
Universities. New York, pp. 13–227.
Brennand, K.J., Simone, A., Jou, J., Gelboin-Burkhart, C., Tran, N., Sangar, S., Li, Y., Mu, Y.,
Chen, G., Yu, D., McCarthy, S., Sebat, J., Gage, F.H., 2011. Modelling schizophrenia
using human induced pluripotent stem cells. Nature 473 (7346), 221–225.
Carpenter Jr., W.T., Strauss, J.S., Muleh, S., 1973. Are there pathognomonic symptoms
in schizophrenia? An empiric investigation of Schneider's first-rank symptoms.
Arch. Gen. Psychiatry 28 (6), 847–852.
Carroll, B.J., Feinberg, M., Greden, J.F., Tarika, J., Albala, A.A., Haskett, R.F., James, N.M.,
Kronfol, Z., Lohr, N., Steiner, M., de Vigne, J.P., Young, E., 1981. A specific laboratory
test for the diagnosis of melancholia. Standardization, validation, and clinical utility.
Arch. Gen. Psychiatry 38 (1), 15–22.
Craddock, N., Owen, M.J., 2010. The Kraepelinian dichotomy — going, going… but still not
gone. Br. J. Psychiatry 196 (2), 92–95.
Samarajiwa, S., Yuan, Y., Graf, S., Ha, G., Haffari, G., Bashashati, A., Russell, R.,
McKinney, S., Langerod, A., Green, A., Provenzano, E., Wishart, G., Pinder, S., Watson, P.,
Markowetz, F., Murphy, L., Ellis, I., Purushotham, A., Borresen-Dale, A.L., Brenton, J.D.,
Tavare, S., Caldas, C., Aparicio, S., 2012. The genomic and transcriptomic architecture of
2,000 breast tumours reveals novel subgroups. Nature 486 (7403), 346–352.
Dawes, R.M., Faust, D., Meehl, P.E., 1989. Clinical versus actuarial judgment. Science 243
Decoster, J., De Hert, M., Viechtbauer, W., Nagels, G., Myin-Germeys, I., Peuskens, J., van Os, J.,
van Winkel, R., 2012. Genetic association study of the P300 endophenotype in
schizophrenia. Schizophr. Res. 141 (1), 54–59.
Deisseroth, K., 2012. Optogenetics and psychiatry: applications, challenges, and
opportunities. Biol. Psychiatry 71 (12), 1030–1032.
M.S. Keshavan et al. / Schizophrenia Research 146 (2013) 10–16
Demjaha,A., Murray,R.M.,McGuire,P.K., Kapur,S.,Howes, O.D., 2012.Dopamine synthesis Download full-text
Ethridge, L.E., Hamm, J.P., Shapiro, J.R., Summerfelt, A.T., Keedy, S.K., Stevens, M.C.,
Pearlson, G., Tamminga, C.A., Boutros, N.N., Sweeney, J.A., Keshavan, M.S., Thaker, G.,
Clementz, B.A., 2012. Neural activations during auditory oddball processing discrimi-
nating schizophrenia and psychotic bipolar disorder. Biol. Psychiatry 72 (9), 766–774.
Evans, K., McGrath, J., Milns, R., 2003. Searching for schizophrenia in ancient Greek and
Roman literature: a systematic review. Acta Psychiatr. Scand. 107 (5), 323–330.
Falret, J.P., 1854. Mémoire sur la folie circulaire. Bull. Acad. Natl. Med. 19, 382–415.
Fanous, A.H., Neale, M.C., Webb, B.T., Straub, R.E., O'Neill, F.A., Walsh, D., Riley, B.P.,
Kendler, K.S.,2008.Novellinkage tochromosome 20pusinglatentclassesofpsychotic
illness in 270 Irish high-density families. Biol. Psychiatry 64 (2), 121–127.
Gottesman, I.I., Gould, T.D., 2003. The endophenotype concept in psychiatry: etymology
and strategic intentions. Am. J. Psychiatry 160 (4), 636–645.
Greenwood, T.A., Lazzeroni, L.C., Murray, S.S., Cadenhead, K.S., Calkins, M.E., Dobie, D.J.,
Green, M.F., Gur, R.E., Gur, R.C., Hardiman, G., Kelsoe, J.R., Leonard, S., Light, G.A.,
Nuechterlein, K.H., Olincy, A., Radant, A.D., Schork, N.J., Seidman, L.J., Siever, L.J.,
Silverman, J.M., Stone, W.S., Swerdlow, N.R., Tsuang, D.W., Tsuang, M.T., Turetsky, B.I.,
Freedman, R., Braff, D.L., 2011. Analysis of 94 candidate genes and 12 endophenotypes
for schizophrenia from the Consortium on the Genetics of Schizophrenia. Am. J. Psychi-
atry 168 (9), 930–946.
Griesinger, W., 1845. Mental Pathology and Therapeutics. Hafner Pub. Co., New York.
Howes, O.D., Kambeitz, J., Kim, E., Stahl, D., Slifstein, M., Abi-Dargham, A., Kapur, S., 2012.
The nature of dopamine dysfunction in schizophrenia and what this means for treat-
ment. Arch. Gen. Psychiatry 69 (8), 776–786.
Hyman, S.E., 2010. The diagnosis of mental disorders: the problem of reification. Annu.
Rev. Clin. Psychol. 6, 155–179.
Insel, T.R., 2010. Rethinking schizophrenia. Nature 468 (7321), 187–193.
Insel, T.R., 2012. Next-generation treatments for mental disorders. Sci. Transl. Med. 4 (155),
Insel, T.R., Cuthbert, B.N., 2009. Endophenotypes: bridging genomic complexity and
disorder heterogeneity. Biol. Psychiatry 66 (11), 988–989.
Ivleva, E.I., Morris, D.W., Moates, A.F., Suppes, T., Thaker, G.K., Tamminga, C.A., 2010.
Genetics and intermediate phenotypes of the schizophrenia–bipolar disorder
boundary. Neurosci. Biobehav. Rev. 34 (6), 897–921.
Jager, M., Haack, S., Becker, T., Frasch, K., 2011. Schizoaffective disorder—an ongoing
challenge for psychiatric nosology. Eur. Psychiatry 26 (3), 159–165.
Jaspers, K., 1946. Allgemeine Psychopathology. Springer Verlag, Heidelberg.
Jeste, D.V., del Carmen, R., Lohr, J.B., Wyatt, R.J., 1985. Did schizophrenia exist before
the eighteenth century? Compr. Psychiatry 26 (6), 493–503.
Kahlbaum, K.L., 1874. Die Katatonie oder das Spannungsirresein. Hirschwald, Berlin,
Kapur, S., Phillips, A.G., Insel, T.R., 2012. Why has it taken so long for biological psychiatry
to develop clinicaltestsand whatto doaboutit? Mol.Psychiatry 17(12), 1174–1179.
Karayiorgou, M., Flint, J., Gogos, J.A., Malenka, R.C., 2012. The best of times, the worst of
times for psychiatric disease. Nat. Neurosci. 15 (6), 811–812.
Kasanin, J., 1994. The acute schizoaffective psychoses. 1933. Am. J. Psychiatry 151
(6 Suppl.), 144–154.
Keers, R., Farmer, A.E., Aitchison, K.J., 2009. Extracting a needle from a haystack: reanalysis
of whole genome data reveals a readily translatable finding. Psychol. Med. 39 (8),
Kendell, R., Jablensky, A., 2003. Distinguishing between the validity and utility of psychiatric
diagnoses. Am. J. Psychiatry 160 (1), 4–12.
Kendell, R.E., Cooper, J.E., Gourlay, A.J., Copeland, J.R., Sharpe, L., Gurland, B.J., 1971.
Diagnostic criteria of American and British psychiatrists. Arch. Gen. Psychiatry 25
Kendler, K.S., 1990. Toward a scientific psychiatric nosology. Strengths and limitations.
Arch. Gen. Psychiatry 47 (10), 969–973.
Keshavan, M.S., Tandon, R., Boutros, N.N., Nasrallah, H.A., 2008. Schizophrenia, “just the
facts”: what we know in 2008 part 3: neurobiology. Schizophr. Res. 106 (2–3), 89–107.
Keshavan, M.S., Morris, D.W., Sweeney, J.A., Pearlson, G., Thaker, G., Seidman, L.J., Eack, S.M.,
Tamminga, C., 2011. A dimensional approach to the psychosis spectrum between
bipolar disorder and schizophrenia: the Schizo-Bipolar Scale. Schizophr. Res. 133
Kraepelin, E., 1899/1921. Manic-depressive Insanity and Paranoia E&S Livingstone,
Lichtenstein, P., Yip, B.H., Bjork, C., Pawitan, Y., Cannon, T.D., Sullivan, P.F., Hultman, C.M.,
2009. Common genetic determinants of schizophrenia and bipolar disorder in
Swedish families: a population-based study. Lancet 373 (9659), 234–239.
Liu, J., Kiehl, K.A., Pearlson, G., Perrone-Bizzozero, N.I., Eichele, T., Calhoun, V.D., 2009.
Genetic determinants of target and novelty-related event—related potentials in
the auditory oddball response. Neuroimage 46 (3), 809–816.
Loscalzo, J., Kohane, I., Barabasi, A.L., 2007. Human disease classification in the postgenomic
era: a complex systems approach to human pathobiology. Mol. Syst. Biol. 3, 124.
Malenka, R.C., Bear, M.F., 2004. LTP and LTD: an embarrassment of riches. Neuron 44
McClintock, S.M., Freitas, C., Oberman, L., Lisanby, S.H., Pascual-Leone, A., 2011. Trans-
cranial magnetic stimulation: a neuroscientific probe of cortical function in schizo-
phrenia. Biol. Psychiatry 70 (1), 19–27.
Meda, S.A., Gill, A., Stevens, M.C., Lorenzoni, R.P., Glahn, D.C., Calhoun, V.D., Sweeney, J.A.,
Tamminga, C.A., Keshavan, M.S., Thaker, G., Pearlson, G.D., 2012a. Differences
in resting-state functional magnetic resonance imaging functional network con-
nectivity between schizophrenia and psychotic bipolar probands and their unaffect-
ed first-degree relatives. Biol. Psychiatry 71 (10), 881–889.
Meda, S.A., Narayanan, B., Liu, J., Perrone-Bizzozero, N.I., Stevens, M.C., Calhoun, V.D.,
Glahn, D.C., Shen, L., Risacher, S.L., Saykin, A.J., Pearlson, G.D., 2012b. A large scale
multivariate parallel ICA method reveals novel imaging-genetic relationships for
Alzheimer's disease in the ADNI cohort. Neuroimage 60 (3), 1608–1621.
Meehl, P.E., 1995. Bootstraps taxometrics. Solving the classification problem in psycho-
pathology. Am. Psychol. 50 (4), 266–275.
Meyer, A., 1957. Psychobiology: A Science of Man. Charles C. Thomas, Springfield.
Muller, N., Dursun, S.M., 2011. Schizophrenia genes, epigenetics and psychoneuro-
National Research Council (U.S.). Committee on A Framework for Developing a New
Taxonomy of Disease, 2011. Toward Precision Medicine: Building a Knowledge
Network for Biomedical Research and a New Taxonomy of Disease. National
Academies Press, Washington, D.C.
schizophrenia: do they help construct “extended endophenotypes”? Schizophr. Bull. 34
Prasad, K.M., Eack, S.M., Keshavan, M.S., Yolken, R.H., Iyengar, S., Nimgaonkar, V.L.,
2012. Antiherpes virus-specific treatment and cognition in schizophrenia: a
test-of-concept randomized double-blind placebo-controlled trial. Schizophr.
Bull. (Epub ahead of print).
Robins, E., Guze, S.B., 1970. Establishment of diagnostic validity in psychiatric illness:
its application to schizophrenia. Am. J. Psychiatry 126 (7), 983–987.
Rzhetsky, A., Wajngurt, D., Park, N., Zheng, T., 2007. Probing genetic overlap among
complex human phenotypes. Proc. Natl. Acad. Sci. U. S. A. 104 (28), 11694–11699.
Sedler, M.J., 1985 Nov. The legacy of Ewald Hecker: a new translation of "Die
Hebephrenie". Translated by Marie-Louise Schoelly. Am J Psychiatry. 142 (11),
Schneider, K., 1959. Clinical Psychology. Grune and Stratton, New York.
Slamon, D.J., Clark, G.M., Wong, S.G., Levin, W.J., Ullrich, A., McGuire, W.L., 1987. Human
breast cancer: correlation of relapse and survival with amplification of the HER-2/neu
oncogene. Science 235 (4785), 177–182.
Smith, I., Procter, M., Gelber, R.D., Guillaume, S., Feyereislova, A., Dowsett, M., Goldhirsch, A.,
Untch, M., Mariani, G., Baselga, J., Kaufmann, M., Cameron, D., Bell, R., Bergh, J.,
Coleman, R., Wardley, A., Harbeck, N., Lopez, R.I., Mallmann, P., Gelmon, K.,
Wilcken, N., Wist, E., Sanchez Rovira, P., Piccart-Gebhart, M.J., 2007. 2-year
follow-up of trastuzumab after adjuvant chemotherapy in HER2-positive breast
cancer: a randomised controlled trial. Lancet 369 (9555), 29–36.
Spitzer, R.L., Endicott, J., Robins, E., 1975. Research diagnostic criteria. Psychopharmacol.
Bull. 11 (3), 22–25.
Strauss, J.S., Carpenter Jr., W.T., Bartko, J.J., 1974. The diagnosis and understanding of
schizophrenia. Part III. Speculations on the processes that underlie schizophrenic
symptoms and signs. Schizophr. Bull. (11), 61–69.
Sullivan, P.F., 2010. The psychiatric GWAS consortium: big science comes to psychiatry.
Neuron 68 (2), 182–186.
Tandon, R., 2012. The nosology of schizophrenia: toward DSM-5 and ICD-11. Psychiatr.
Clin. North Am. 35 (3), 557–569.
Tandon, R., Maj, M., 2008. Nosological status and definition of schizophrenia: some
considerations for DSM-V and ICD-11. Asian J. Psychiatry 1 (2), 22–27.
Trusheim, M.R., Berndt, E.R., Douglas, F.L., 2007. Stratified medicine: strategic and
economic implications of combining drugs and clinical biomarkers. Nat. Rev.
Drug Discov. 6 (4), 287–293.
van Praag, H.M., 2008. Kraepelin, biological psychiatry, and beyond. Eur. Arch. Psychiatry
Clin. Neurosci. 258 (Suppl. 2), 29–32.
Waltman, P., Pearlman, A., Mishra, B., 2006. Interpreter of maladies: redescription mining
applied to biomedical data analysis. Pharmacogenomics 7 (3), 503–509.
Yang, J.C., Shih, J.Y., Su, W.C., Hsia, T.C., Tsai, C.M., Ou, S.H., Yu, C.J., Chang, G.C., Ho, C.L.,
Sequist, L.V., Dudek, A.Z., Shahidi, M., Cong, X.J., Lorence, R.M., Yang, P.C., Miller, V.A.,
2012. Afatinib for patients with lung adenocarcinoma and epidermal growth
factor receptor mutations (LUX-Lung 2): a phase 2 trial. Lancet Oncol. 13 (5),
M.S. Keshavan et al. / Schizophrenia Research 146 (2013) 10–16