VOL. 20, NO. 1, 95–111
ORIGINAL RESEARCH REPORT
The DSM5/RDoC debate on the future of mental health research: implication for
studies on human stress and presentation of the signature bank
S. J. Lupiena,b,c, M. Sassevilleb,c, N. Françoisb, C. E. Giguereb, J. Boissonneaultb, P. Plusquellecb,d, R. Godboutb,c,
L. Xiongb,c, S. Potvinb,c, E. Kouassib, A. Lesageb,c and the Signature Consortiumb
Centre for Studies on Human Stress, CIUSSS Est, Quebec, Canada; bResearch Centre, Montreal Mental Health University Institute, CIUSSS Est,
Quebec, Canada; cDepartment of Psychiatry, Faculty of Medicine, University of Montreal, Montreal, Canada; dDepartment of Psychoeducation,
Faculty of Arts and Sciences, University of Montreal, Montreal, Canada
In 2008, the National Institute of Mental Health (NIMH) announced that in the next few decades, it will
be essential to study the various biological, psychological and social “signatures” of mental disorders.
Along with this new “signature” approach to mental health disorders, modifications of DSM were introduced. One major modification consisted of incorporating a dimensional approach to mental disorders,
which involved analyzing, using a transnosological approach, various factors that are commonly
observed across different types of mental disorders. Although this new methodology led to interesting
discussions of the DSM5 working groups, it has not been incorporated in the last version of the DSM5.
Consequently, the NIMH launched the “Research Domain Criteria” (RDoC) framework in order to provide
new ways of classifying mental illnesses based on dimensions of observable behavioral and neurobiological measures. The NIMH emphasizes that it is important to consider the benefits of dimensional
measures from the perspective of psychopathology and environmental influences, and it is also important to build these dimensions on neurobiological data. The goal of this paper is to present the perspectives of DSM5 and RDoC to the science of mental health disorders and the impact of this debate on
the future of human stress research. The second goal is to present the “Signature Bank” developed by
the Institut Universitaire en Sante Mentale de Montr
eal (IUSMM) that has been developed in line with a
dimensional and transnosological approach to mental illness.
Received 9 August 2016
Revised 9 November 2016
Accepted 18 December 2016
Mental health research in the twenty-first century
The period from 1990 to 2000 has been termed the decade of
the brain (Health, 2015), leading to major discoveries in the
field of genetics, neuroimaging and degenerative disorders
(Cuthbert & Insel, 2010, 2013). Based on these significant discoveries, the National Institute of Mental Health (NIMH) in the
United States announced in its 2008 strategic plan (Health,
2008) that the next decade would be devoted to applying the
acquired knowledge to patient populations suffering from
mental health disorders. In other words, it would be committed
to making this period the “decade of the patient”.
According to the NIMH, mental health research must correspond to the four Ps of personalized medicine – Predictive,
Preemptive, Personalized and Participatory – in order to be
innovative. To accomplish this, it will be essential to study the
various biological, psychological and social “signatures” of
mental disorders as well as the interactions between different
biological signatures, clinical manifestations and psychosocial
determinants of mental health across the lifespan and as a
function of exposure to different environments. The
“signatures” of mental illness is a term formulated by the NIMH
CONTACT Sonia J. Lupien
Montreal H1N 3M5, Canada
Mental health; DSM5; RDoC;
psychopathology; development; environment
to designate the broad range of biological, psychological and
social factors that may “sign” a specific mental health disorder,
depending on an individual’s sex, history, lifestyle habits and so
on (Health, 2008, 2015).
Personalized mental health medicine represents an
approach specifically adapted to the challenges of twentyfirst century health, especially with regard to chronic illnesses.
Chronic illnesses are now recognized as being the leading
causes of morbidity and mortality in contemporary societies,
as well as the main drivers of health service utilization and
associated expenses. In the health care system, mental health
disorders are among the most common and costly chronic illnesses (Reeves et al., 2011). In 2004, an estimated 25% of
adults in the United States reported having a mental illness
in the previous year (Reeves et al., 2011), and this led to a
total cost of $300 billion in 2002. Despite this, mental illnesses do not always receive the attention they deserve, and
those affected by mental illness do not receive all the treatment they need, since current research is still being conducted in a non-concerted manner, with researchers
separately seeking to identify the gene, the biomarker or the
Centre for Studies on Human Stress, Montreal Mental Health University Institute, 7331 Hochelaga,
ß 2017 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-nd/4.0/),
which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way.
S. J. LUPIEN ET AL.
psychological factor that predicts a mental illness categorized
as a function of symptoms rather than validated medical
At the present time, there is significant growth in the area
of developing genetic biobanks that enable long-term storage of DNA samples with the goal of later studying various
diseases such as cancer and cardiovascular disorders.
However, the studies conducted to date on the genetic
causes of mental disorders have had little success, and the
NIMH suggested that this failure may be due to the fact that
researchers are still looking for a single signature of mental illness when they should instead be seeking multiple signatures
(Health, 2008). The fact is, to understand complex chronic illnesses such as mental diseases, it is necessary to be innovative and move away from a reductionist approach that
basically involves identifying one gene or one brain abnormality as the factor explaining a mental disorder.
One of the main reasons that have been given to explain
the failure of biological mental health studies is the absence of
a good phenotypic classification (Cuthbert & Insel, 2010, 2013;
Kapur et al., 2012). Here, it has been suggested that neurobiological studies have been unable to find any genetic or neuroscientific evidence or a single “biological signature” of the
different categories of mental illnesses proposed by DSM over
its various versions (Cuthbert & Insel, 2013; Doherty & Owen,
2014; Insel et al., 2010), because the phenotype used in these
studies, i.e. the diagnostic criteria as determined by DSM categories, is not the best phenotype to use in order to determine
the biological signature of mental illness (Casey et al., 2013;
Cuthbert & Insel, 2010; Insel et al., 2010).
The categorical approach to mental health disorders
Long before the first edition of the DSM was published in
1952, it was recognized that there was a need to distinguish
between diseases – i.e. to differentiate between various pathologies when choosing a treatment. Indeed, the more clearly
various problems were distinguished, the greater the likelihood of successful medical treatment. It was Emil Kraepelin
who developed the first classification system resembling
what is now found in the DSM. His Lehrbuch der Psychiatrie
categorized mental disorders based on their nature, and he
developed hypotheses about their causes, evolution and
prognoses (see Kohl, 1999a,b).
Since DSM-I was published, many revisions have been made
in order to provide caregivers and researchers with the most
up-to-date data. However, across all of the versions of DSM,
there is one common assumption, i.e. such a categorical
approach to mental illnesses assumes that mental disorders are
discrete entities shared by relatively homogeneous populations
that will display similar symptoms of any given disorder. Based
on this definition, a categorical diagnosis can thus only have
two values, i.e. presence or absence of a given disorder. Over
the years, there have been various discussions about the validity of the way these diagnostic categories are constructed.
Three main critics of the categorical approach to psychiatry
have been raised in the scientific literature over the years.
First, it has been stated that the first assumption of the
categorical approach (i.e. “mental disorders are discrete entities
shared by homogeneous populations”) is not valid. Indeed, it is
well known that psychiatric populations are highly heterogeneous and it is highly common to see patients showing more
than one clear psychiatric diagnosis (Regier, 2008; Widiger &
Coker, 2003). This heterogeneity led, over the years, to the
diagnostic category of “Not otherwise specific” (NOS; First,
2010). The NOS category is often referred to as the “catch-all”
category (Millon, 1991). It is generally used when a clinician
determines that a mental illness is present, although the
patient fails to meet the criteria for one of the existing diagnostic category. Although this category should be given
rarely, studies show that the NOS category is used as often as
any of the specific diagnostic categories of the DSM (Cassano
et al., 1999; Fairburn et al., 2007; Wilberg et al., 2008).
Second, although a categorical diagnostic approach considers that patients suffering for a given mental illness display
similar symptoms, we now know that there is a lot of overlap
of symptoms in psychiatric populations, and this explains the
high level of comorbidities observed in patients. “Pure” psychiatric syndromes are extremely rare in psychiatric research,
and it is very common to see patients receiving two, three
and even four types of psychiatric diagnoses (Andrews et al.,
2002; Kessler et al., 1996; Mineka et al., 1998). For example,
an Australian study of 10,000 participants reported that 40%
of the sample met the diagnostic criteria for more than one
current mental health disorder (Andrews et al., 2002).
Moreover, mood disorders show high comorbidity with anxiety disorders (Kessler et al., 1996; Mineka et al., 1998), and
mood and anxiety disorders are comorbid with substance use
disorders, personality disorders and eating disorders (Mineka
et al., 1998; Widiger & Clark, 2000). This suggests that there
might be a problem with the “phenotype” of mental illness,
i.e. an organism's observable characteristics or traits related
to a particular mental disorder (Kupfer, 2005).
Third, studies on childhood and adolescence often report
that psychiatric categories, as defined by the DSM, are unable
to capture a common source of variation in developmental
psychopathology: the stage of development (see Beauchaine
& McNulty, 2013). Indeed, more and more studies report that
the psychiatric diagnostic category seems to change with a
child or teenager’s development (Cuthbert & Kozak, 2013).
Clinicians and scientists have also criticized the fact that certain criteria do not sufficiently account for the full complexity
of the psychopathological phenomena experienced by children and adolescents suffering from various forms of psychopathologies (Haggerty et al., 1996; Rutter, 2003; Rutter et al.,
2003; Rutter & Sroufe, 2000). Consequently, the principle of
categorization limits the ability to capture nuances that could
be used to indicate the intensity of various symptoms or their
evolution over time.
A new dimensional approach to mental disorders
Despite the millions of research dollars that were spent on
trying to find the “biomarker” of a specific mental health disorder as defined by the DSM, not a single study led to the
discovery of a biological test with a clinical utility to diagnose
a given mental illness. Consequently, various scientists started
to question the reason for this failure.
In an important paper published in 2012, Kapur et al.
(2012) suggested that four main factors can explain this failure. First, compared to other medical fields that have biological tests to diagnose disorders, the field of psychiatry did
not find any biological marker that can differentiate between
patients and control and/or between patients. Second,
although there has been a very large number of studies published on neurobiological dysfunctions in various mental disorders, most of these studies have small or moderate effect
sizes, even though group differences are significant at p < .05.
This led to a tremendous number of scientific papers published on various “biomarkers” of mental disorders, but no
single consensus on a single clinical test that could be efficient at distinguishing a given mental illness. Third, replication studies are extremely rare in psychiatry and the majority
of studies that try to replicate a given finding include small
variations in patient populations, measures and/or biological
measures, which lead to “approximate replications”. When
this happens, any difference between two studies is
explained by differences in patients populations, measures or
biological tests, and this prevents the development of significant biological tests for a given mental illness. Finally, most
studies in psychiatric research compare prototypical patients
and “picture-perfect” healthy controls, which leads to significant results that may be difficult to replicate when one tries
to distinguish between different patient types. Based on
Kapur and colleagues review (Kapur et al., 2012), all of these
factors led to a Catch-22 where the current diagnostic system
was not designed to facilitate biological differentiation, and
the biological studies have not been able to propose a clinically viable alternative system.
Based on this analysis, they suggested that in order to
provide clinical tests of psychiatric disorders, the field of biological psychiatry will have to change its mindset. Rather
than having a strict allegiance on DSM or ICD diagnostic categories, the field should collect data across the current diagnostic categories, focus on comparing across disorders as
much as comparing patients to control subjects, and should
collect longitudinal data so that we can move toward the
development of biological clinical tests for mental illness
(Kapur et al., 2012).
The DSM-5 dimensional approach
The first attempt to develop such a transnosological
approach was made by various committees working on the
fifth version of the DSM. In 2000, over 500 world-renowned
researchers and clinicians were invited to be involved in the
DSM revision process. In addition to the general revision of
diagnostic criteria for each disorder and a somewhat different
approach to organizing disorders, modifications were being
proposed that would affect most, if not all, mental disorders
in the DSM.
One of the major modifications in the DSM-5 consisted of
incorporating a dimensional approach to mental disorders,
which would involve analyzing various dimensions or functions
that are associated with many different types of mental disorders. For example, anxious symptoms occur in conjunction
with many mental disorders, and consequently, the presence
of anxiety in a given patient may therefore be enough to “sign”
a mental disorder and cut across various diagnoses.
Consequently, the nature of the anxiety could indicate an individual’s underlying vulnerability to specific disorders.
The DSM would continue to propose diagnostic categories
of mental disorders, but would include this new, dimensional
approach. Various working groups were developed between
2000 and 2012 to work on the revision of the DSM and in
July 2006, a diagnosis-related research planning conference
focusing on dimensional approaches in diagnostic classification was held at the Natcher Conference Center on the NIH
campus in Bethesda, Maryland. Twenty-eight invited scientists
from around the world participated and worked at delineating the advantages of adding a dimensional approach to the
DSM5 (http://www.dsm5.org/research/pages/dimensionalaspectsofpsychiatricdiagnosis(july26-28,2006).aspx). Five general
advantages of adding a dimensional approach to DSM-5 were
described by this committee.
First, the decision to make use of dimensions in combination
with categories was related to the quality of the diagnosis
being made rather than the nature of the disorder being diagnosed. For instance, while a dimensional assessment would
make it possible to evaluate on a scale with at least three
points (e.g. no disorder versus mild disorder versus severe disorder), categorical assessment is by definition limited to two
points (e.g. present versus not present). At the level of clinical
practice, this restricts the extent to which it is possible to provide a detailed diagnosis. Either the disorder is present or it is
not; there are no nuances. At the level of research, being able
to use dimensional assessments as well as categorical measures could make it possible to study the potential modulating
effect of certain aspects of mental disorders (e.g. anxious
symptoms) on the severity of a given mental disorder.
Scientists and clinicians participating in this event suggested that introducing the use of dimensional measures
should be entirely feasible given the current state of clinical
practice and scientific research. First, dimensional tools (e.g.
questionnaires) are already widely used in different fields (e.g.
somatic and cognitive symptoms of anxiety, panic, avoidance), and the dimensional approach is already standard in
research (e.g. asking about the number of panic attacks).
Additionally, there was the possibility of drawing on work
being done on other subjects in medicine (e.g. hypertension,
obesity). In short, the beneficial effects on clinical practice
and research offered by the increased precision and more
complex nuances of dimensional measures were clear to the
research field. They enhanced the ability to determine the
effect of a treatment, enabled greater precision in quantifying
that effect, improved the capacity to detect existing signals
and made it possible to identify more treatment possibilities.
More faithful to reality
Dimensional measures would make it possible to establish an
individual’s risk of a given disease with a greater degree of
S. J. LUPIEN ET AL.
nuance. An example of this is the “metabolic syndrome,” that
is associated with a high risk of cardiovascular disease (CVD).
Originally, a person with three of the five risk factors for CVD
is deemed to have a metabolic syndrome and be at high risk
of CVD. But with two risk factors out of five, an individual is
still at risk of CVD; the risk is only slightly reduced. Adopting
an approach based on “degree of risk” (dimensional
approach) rather than on “at risk” versus “not at risk” dichotomy (categorical approach) therefore would seem appropriate. This approach better represents the reality and makes it
possible to intervene in an incremental manner with less
severe cases before they become more serious. This also
offers a way to resolve the problem of categorical thresholds,
given that individuals in a subclinical state may also be negatively impacted by their condition and could benefit from
This kind of dimensional approach wo ...
Purchase answer to see full