82 | THE INFIRMITIES OF PSYCHIATRIC DIAGNOSIS

STEVEN E. HYMAN

The major diagnostic systems in psychiatry, the Diagnostic and Statistical Manual of Mental Disorders, 4th edition (American Psychiatric Association [APA], 1994) and the closely related International Classification of Diseases, 10th edition (World Health Organization, 1992), chapter on mental and behavioral disorders are currently undergoing revision. Both the general scientific press (Miller and Holden, 2010) and the lay press have focused a great deal of attention on these revisions because the resulting manuals will exert significant influence on the putative boundaries between health and mental illness, the treatments that patients will receive, reimbursement for services, determinations of disability, and even the sentencing of some individuals convicted of crimes. To the credit of the American Psychiatric Association, each subsequent recent edition of the DSM system has included a disclaimer in the front matter. For example, the current DSM-IV-TR (APA, 2000), states that the diagnostic criteria are “offered as guidelines” because “use of such criteria enhances agreement among clinicians and investigators.” The authors of the manual make no truth claims for the classification overall or for any disorder within it. Yet in practice it would be hard to imagine any caveat or disclaimer more widely ignored than this one.

Like any disease classification, the DSM is a cognitive structure imposed on scientific data to make it useful for clinical applications and research. Even the most scientifically useful theoretical frameworks or classification systems eventually outlive their beneficial role. Like Ptolemaic astronomy, a broadly accepted theoretical schema that has outlived its ability to explain emerging data can produce intellectual stagnation by excessively narrowing the focus and even the imagination of investigators (Kuhn, 1962). The DSM system is a central organizer of psychiatric research because it provides a shared language, and more than that, it delineates those specific disorders that can generally be studied within the mainstream psychiatric community. The profound influence of the DSM system grew directly from the success of applying the DSM-III (APA, 1980) to enhance comparability across studies, in marked contrast with the diagnostic free for all (Pope and Lipinski, 1978) that existed before its publication in 1980. Without comparability of subject populations in translational or clinical research, it is simply not possible to replicate results or falsify specific hypotheses with any rigor. The obverse of this benefit is that the DSM system came to dominate thinking about psychiatric disorders at a time when the relevant science was embryonic. Thus the DSM system enhanced the ability of researchers, clinicians, patients, and families to communicate with each other, but also produced premature intellectual closure—despite the American Psychiatric Association’s disclaimer (Hyman, 2007, 2010).

The DSM-III remains—even with the impending publication of DSM-5—the fundamental current prototype for psychiatric disease classification and diagnosis. The DSM approach has been to delineate a large number of distinct diagnostic categories and to define them based on operationalized diagnostic criteria. Given the early state of psychiatric science, all of the current sets of criteria are perforce comprised of clinical descriptions (i.e., phenomenology). It would be possible to imagine that objective tests and biomarkers could be discovered for the existing categories, and their careful incorporation into current sets of criteria would ultimately lead to a modern descendent of the DSM-III. This chapter argues that in contrast to this hopeful view, much clinical and laboratory evidence suggests that there are fundamental structural problems with the DSM system and that the current diagnostic structure may be impeding progress (Hyman, 2007, 2010). For example, essentially all disease-focused grant applications, if they are to be awarded, must select patient populations based on DSM criteria. Similarly, manuscripts submitted for publication must also categorize patients according to the DSM system. Patients entered in clinical trials almost invariably must meet criteria for specific DSM disorders. Indeed, with respect to development of new treatments, DSM-based diagnoses, which are taken to represent the psychiatric community’s consensus, strongly influence regulators. Thus, for example, the US Food and Drug Administration takes the DSM system into account in deciding what constellation of symptoms represents a valid indication in psychiatry. Deviations from DSM nomenclature may require special efforts before development of a new drug treatment commences (Buchanan et al., 2005). In addition, psychiatry residents and clinical psychology interns often find themselves memorizing the most recent version of the DSM for certification examinations. It is no wonder that, despite the American Psychiatric Association’s disclaimers, the lay press often describes the DSM as psychiatry’s bible.

DSM-III ANTEDATED IMPORTANT DEVELOPMENTS IN THE LIFE SCIENCES

Even today we do not understand the etiology or pathophysiology of depression, bipolar disorder, schizophrenia, autism, or any other major disorder under the purview of psychiatry. There are still no objective medical tests that are sensitive or specific enough to use for diagnoses, and there are no validated biomarkers for monitoring treatment. Despite this seemingly dismal picture, which contrasts with significant progress in many other areas of medicine, there is good reason to believe that the science of psychiatry is maturing in ways that will prove highly fruitful. The section that follows, illustrating the potential of newer technological and scientific approaches to psychiatric research is not meant to understate the challenges that remain, but to provide examples of major developments in science relevant to psychiatric disorders, and to underscore that all of these emerged in the decades after the publication of DSM-III (APA, 1980).

Much has happened in the life sciences since the gestation of descriptive psychiatry in the 1960s and 1970s and its apotheosis with the publication of DSM-III (APA, 1980) in 1980. Molecular biology, perhaps the dominant force in modern biology and medicine only emerged in the mid-1980s following a moratorium on recombinant DNA technology that ended in 1975 (Wade, 1975). Important discoveries about the nervous system date back to the nineteenth century, but modern neuroscience as an interdisciplinary field is relatively new. The first department of Neuroscience was only founded in 1970, and the field only began to involve clinical disciplines effectively in the ensuing decades. One recent subfield emerging from within neuroscience that is highly relevant to the goals of psychiatry is called “connectomics.” This field represents an attempt to provide full descriptions of the brain “wiring diagrams” of both model organisms and humans based on detailed mapping of cells, synapses, and circuits. Beyond a static map, connectomics is also attempting to overlay a dynamic picture of the functional connectivity of the human brain. Because neural circuit function directly underlies thought, emotion, and behavior, in short, the substrate of psychiatric symptoms, the data and understandings that emerge will be critical to understanding psychopathology and ultimately disease classifiction. Connectomics has become a feasible goal for neuroscientists because of advances in such technologies as microscopy, neuroimaging, computation, and new technologies that utilize molecular tools to study specific neural circuits and their relationship to behavior (Fenno et al., 2011). Noninvasive neuroimaging is giving us important structural and functional views of the human brain. Structural magnetic resonance imaging (MRI) permits the detailed characterization of diverse neural structures; for example, measurements of cerebral cortical thickness or hippocampal volume. Diffusion tensor imaging (DTI) permits the accurate tracing of white matter tracts in the brain. Functional MRI (fMRI) and various modalities of positron emission tomography (PET) make it possible to observe the activation of brain circuits in response to specific tasks and to observe patterns of brain connectivity at rest. PET also permits the labeling and visualization of specific proteins, such as neurotransmitter receptors. To give some historical context, the first published papers using fMRI, now a major contributor to psychiatric neuroscience, only appeared in the mid-1990s.

Genetics is also an important approach to understanding the etiology of mental disorders given that many disorders are highly heritable. Using still-developing technologies motivated by the human genome project (Shendure and Ji, 2008), it has been possible only since 2007 to make significant progress in discovering specific genetic risk factors involved in the pathogenesis of autism (Neale et al., 2012) schizophrenia (Lee et al., 2012), and bipolar disorder. Also in 2007, induced pluripotent cell technology was developed that has made it possible to transform fibroblasts obtained from small skin biopsies of psychiatric patients and healthy comparison subjects into human neurons that can be studied in vitro (Takahashi et al., 2007). This technology is already permitting molecular and cellular investigations into the pathogenesis of psychiatric disorders. The intellectual roots of the DSM system lay in the descriptive psychiatry of the 1960s and 1970s (Robins and Guze, 1970), which in turn dates back to the pioneering nineteenth century work of Kraepelin (1899/1991). Descriptive psychiatry was a critically important intellectual movement that helped free the field from the dominant psychoanalytic approaches of the mid-twentieth century that had little place for biology. It should not be surprising, however, that the institutionalization of descriptive psychiatry mediated by the DSM system has provided a poor lens through which to view modern science.

DESCRIPTIVE PSYCHIATRY

Scientifically mature diagnostic systems are generally based on pathophysiology or etiology rather than clinical description in recognition of the fact that multiple factors related to disease processes, patient adaptations, and environmental context may influence the nature and severity of symptoms, functional impairments, and disability. Beginning in the 1950s, new pharmacological treatments had emerged, including antipsychotic drugs, antidepressants, lithium, and benzodiazepines, creating a significant need for a diagnostic system that would permit the selection of homogeneous groups of patients for clinical trials and facilitate the appropriate matching of patients with treatments. One important hurdle was diagnostic reliability, the ability of different observers to reach the same diagnosis for a given patient. Absent understandings of pathophysiology, lacking empirical knowledge of etiology, and even without objective medical tests, careful clinical description would prove to be the best approach available to psychiatry in the mid-twentieth century. In 1970 Robins and Guze (1970) two investigators at Washington University, argued that it would be possible to achieve reliable and valid psychiatric diagnoses based on clinical description, laboratory studies, exclusion of other disorders, follow-up observations, and family studies. They and their colleagues were inspired by the careful phenomenology of Emil Kraepelin (1899/1991), who had argued that psychiatry could identify specific disorders based on careful description of symptoms, signs, and course of illness. Kraepelin used such methods to distinguish what he called dementia praecox (schizophrenia) from manic-depressive illness (bipolar disorder) in the nineteenth century. Following the lead of Robins and Guze, the St. Louis school produced two sets of diagnostic criteria, the Research Diagnostic Criteria (RDC) and the Feighner criteria (Feighner et al., 1972 (1972); Spitzer et al., 1975), which were direct forerunners of the DSM-III (APA, 1980). To enhance inter-rater reliability these diagnostic systems employed operationalized criteria (i.e., criteria that specified well defined clinical observations). Validation (i.e., providing evidence that a diagnosis identified a verifiable entity in nature) represented an insurmountable problem given the state of the science in the 1970s. Robins and Guze (1970) argued that validity of diagnoses could be established using multiple forms of evidence, such as the stability of a person’s diagnosis over the life course, and the transmission of a disorder within families (Robins and Guze, 1970). As will be seen, however, familial transmission would not prove to be a validator as Robins and Guze had hoped: Disorders defined by purely phenomenological criteria do not breed true (Lichtenstein et al., 2009).

One of the original validating criteria listed by Robins and Guze (1970) was delineation of disorders from each other. Across the RDC, Feighner criteria, and the DSM system, psychiatric disorders are conceptualized as bounded categories that are qualitatively different from each other and from health. All of these diagnostic systems treat psychiatric disorders as discontinuous categories in analogy with infectious diseases such as pneumococcal pneumonia or tuberculosis. The mid-century descriptive psychiatrists eschewed an alternative conceptual schema that has turned out to be more consistent with the accumulated data, which would describe psychiatric disorders in terms of quantitative dimensions that are continuous with normal in analogy with hypertension or diabetes mellitus. In dimensional diagnoses, thresholds for illness are determined empirically based on studies of impairment, disability, or longer-term health outcomes (such as risk of stroke given current blood pressure). Evidence that dimensional approaches better capture psychopathology than categories include the inability to discover discontinuities in the distribution of symptoms from healthy to ill people (e.g., in depression; Lichtenstein et al., 2009) and epidemiological studies that find normal distributions of symptoms in the population (e.g., for attention deficit hyperactivity disorder, autism [Kendlerand Gardner, 1998], and schizophrenia) with illness representing extremes of the distribution on multiple dimensions.

An important goal of the early Washington University diagnostic systems (Feighner et al., 1972 (1972); Spitzer et al., 1975) and the DSM-III (APA, 1980) was to identify homogeneous populations for research and treatment. The DSM-III attempted to achieve this goal by subdividing psychopathology into a large number of narrowly defined categorical diagnoses. This approach goes to the heart of the deep structural problems of the DSM system: It is based on narrow, highly specified (and therefore, it was thought, reliable), discontinuous categories. Far from producing homogeneity, the DSM approach has resulted in a highly infelicitous combination of overlapping categories that are yet, internally heterogeneous. By slicing psychopathology into many narrow silos, the DSM manuals have produced a situation in which a large number of patients receive more than one diagnosis (so-called co-occurrence or comorbidity.). For example, an individual who receives a single DSM-IV diagnosis often meets criteria for multiple additional disorders, and the pattern of often changes over the lifespan (Kessler et al., 1996, 2005). Thus, for example, children and adolescents with one (and often more than one anxiety disorder diagnosis) may receive a diagnosis of major depression in their teens or twenties. Individuals with autism spectrum disorders have high rates of attention deficit hyperactivity disorder, obsessive-compulsive disorder, and others (Lichtenstein et al., 2010). Of course, disorders could co-occur at random based on their prevalence or, co-occurrence at a higher rate than predicted by prevalence alone could signify that one disorder is a risk factor for another, just as diabetes mellitus is a risk factor for peripheral vascular disease. Within the DSM system, however, many diagnoses co-occur at frequencies far higher than predicted by their population prevalence, and both emerging understandings of pathogenesis (Kendler et al., 2011; Krueger and Markon, 2006; Krueger and South, 2009) and temporal relationships (Kessler et al., 2005) make it unlikely that one disorder is a risk factor for later onset disorders. The most parsimonious explanation for the plague of comorbidity is that it is largely an artifact of the DSM system, resulting from its having divided shared pathological processes into excessively narrow slices (Krueger and Markon, 2006; Krueger and South, 2009).

The excessively narrow diagnostic silos of the DSM system (which produce comorbidity) do not, unfortunately, generate homogeneity within each category Thus, for example, well-diagnosed schizophrenia does not “breed true”; instead, single families may exhibit schizophrenia schizoaffective disorder, bipolar disorder, and unipolar mood disorders (Craddock et al., 2006; Lichtenstein et al., 2009) even when the etiology of psychopathology appears to rest largely in a single large chromosomal rearrangement (Millar et al., 2000).

Even though, as early as 1967, Gottesman and Shields (1967) recognized that schizophrenia might be polygenic, the Washington University School and the authors of the DSM-III (APA, 1980) understandably did not take this complicating view into account. Had they done so in a sophisticated manner they might have given greater credence to dimensional constructions and might have created a system more tolerant of heterogeneity. What has emerged from family and genetic studies in recent years is that autism (Neale et al., 2012; Sullivan et al., 2012), schizophrenia (Gejman et al., 2011; Lee et al., 2012; Sullivan et al., 2012), bipolar disorder, and indeed all common psychiatric disorders that have been examined (Sullivan et al., 2012), are polygenic. Large numbers of genes contribute to risk of mental disorders in different combinations in different families, where they produce disease phenotypes in combination with stochastic, epigenetic, and environmental factors. In some families genetic risk for mental disorders seems to be caused by many, perhaps hundreds of small variations in DNA sequence (often single nucleotide bases), each causing a very small increment in risk. In other families a copy number variation (CNV), whether a duplication, deletion, or rearrangement of DNA sequence, may play a large role, but still acts against a background of polygenic risk. In some individuals with sporadic autism, rare mutations may occur de novo. In others, rare mutations are inherited. In sum, no gene is either necessary or sufficient for risk of a common mental disorder, and genes may produce different symptoms depending on broad genetic background, early developmental influences, life stage, or diverse environmental factors. The DSM myth of narrow homogeneous diagnostic categories seems ready for burial once we recognize the full implications of polygenicity and the variable penetrance and expressivity of many risk genes (Cross-Disorder Group, 2013). Rather than tightly constrained disorder categories, it might be better, for the near term, to rely on broader clinical diagnoses that likely capture etiologically heterogeneous conditions but that might share essential features of pathophysiology and thus produce related (but not identical) symptoms and course.

The DSM-III was based of necessity on clinical phenomenology, and had many important strengths that must be kept in historical context. What went most wrong with the DSM-III were fairly arbitrary decisions to conceptualize psychopathology as discontinuous categories, as if they were infectious diseases, and then to promulgate a very large number of them. A short-term alternative to the use of individual DSM disorders might be the use of disorder clusters that will be embedded in the organization of the DSM-5 (Hyman, 2010), whereas longer term improvements will require substantially new thinking (e.g., National Institutes of Mental Health, Research Domain Criteria). In the meantime, students who use this textbook should recognize that DSM categories are heuristics based on half-century-old science, and are not to be reified, but rather used pragmatically until better approximations of nature arrive.

EPILOGUE: CLASSIFICATION ACCORDING TO BORGES

These ambiguities, redundancies, and deficiencies recall those attributed by Dr. Franz Kuhn to a certain Chinese encyclopedia called the Heavenly Emporium of Benevolent Knowledge. In its distant pages it is written that animals are divided into (a) those that belong to the emperor; (b) embalmed ones; (c) those that are trained; (d) suckling pigs; (e) mermaids; (f) fabulous ones; (g) stray dogs; (h) those that are included in this classification; (i) those that tremble as if they were mad; (j) innumerable ones; (k) those drawn with a very fine camel’s-hair brush; (l) etcetera; (m) those that have just broken the flower vase; (n) those that at a distance resemble flies.

“John Wilkins’ Analytical Language”, translator Eliot Weinberger; included in Selected Non-Fictions: Jorge Luis Borges”, ed. Eliot Weinberger; 1999, New York: Penguin Books, p. 231. The essay was originally published as “El idioma analítico de John Wilkins,” La Nación, February 8, 1942.

DISCLOSURE

Dr. Steven Hyman serves on the Novartis Science Board and has advised AstraZeneca within the last year. Both advisory roles focus on early stage drug discovery. His research is funded by the Stanley Foundation.

REFERENCES

American Psychiatric Association. (1980). Diagnostic and Statistical Manual of Mental Disorders (3rd edition). Washington, DC: Author.

American Psychiatric Association. (1994). Diagnostic and Statistical Manual of Mental Disorders (4th edition). Washington, DC: Author.

American Psychiatric Association. (2000). Diagnostic and Statistical Manual of Mental Disorders (4th edition, Text Revision). Washington, DC: Author.

Buchanan, R.W., Davis, M., et al. (2005). A summary of the FDA-NIMH-MATRICS workshop on clinical trial design for neurocognitive drugs for schizophrenia. Schizophr. Bull. 31:5–19.

Craddock, N., O’Donovan, M.C., et al. (2006). Genes for schizophrenia and bipolar disorder? Implications for psychiatric nosology. Schizophr. Bull. 32:9–16.

Cross-Disorder Group of the Psychiatric Genome Consortium (2013). Identification of risk loci with shared effects on five major psychiatric disorders: A genome-wide analysis. Lancet. February 28, 2013. Epub ahead of print.

Feighner, J.P., Robins, E., et al. (1972). Diagnostic criteria for use in psychiatric research. Arch. Gen. Psychiatry 26:57–63.

Fenno, L., Yizhar, O., et al. (2011). The development and application of optogenetics. Annu. Rev. Neurosci. 34:389–412.

Gejman, P.V., Sanders, A.R., et al. (2011). Genetics of schizophrenia: new findings and challenges. Annu. Rev. Genomics Hum. Genet. 12:121–144.

Gottesman, I.I., and Shields, J. (1967). A polygenic theory of schizophrenia. Proc. Natl. Acad. Sci. USA 58:199–205.

Hyman, S.E. (2007). Can neuroscience be integrated into the DSM-V? Nat. Rev. Neurosci. 8:725–732.

Hyman, S.E. (2010). The diagnosis of mental disorders: the problem of reification. Ann. Rev. Clin. Psychol. 6:155–179.

Kendler, K.S., Aggen, S.H., et al. (2011). The structure of genetic and environmental risk factors for syndromal and subsyndromal common DSM-IV axis I and all axis II disorders. Am. J. Psychiatry 168:29–39.

Kendler, K.S., and Gardner, C.O., Jr. (1998). Boundaries of major depression: an evaluation of DSM-IV criteria. Am. J. Psychiatry 155:172–177.

Kessler, R.C., Chiu, W.T., et al. (2005). Prevalence, severity, and comorbidity of 12-month DSM-IV disorders in the National Comorbidity Survey Replication. Arch. Gen. Psychiatry 62:617–627.

Kessler, R.C., Nelson, C.B., et al. (1996). Comorbidity of DSM-III-R major depressive disorder in the general population: results from the US National Comorbidity Survey. Br. J. Psychiatry 168(Suppl. 30):17–30.

Kraepelin, E. (1899/1991). Psychiatry: A Textbook for Students and Physicians. Transl. ed. H. Metoui, S. Ayed. Canton, MA: Watson Publishers.

Krueger, R.F., and Markon, K.E. (2006). Reinterpreting comorbidity: a model-based approach to understanding and classifying psychopathology. Annu. Rev. Clin. Psychol. 2:111–113.

Krueger, R.F., and South, S.C. (2009). Externalizing disorders: cluster 5 of the proposed meta-structure for DSM-V and ICD-11. Psychol. Med. 3:2061–2070.

Kuhn, T.S. (1962). The Structure of Scientific Revolutions, 2nd edition. Chicago: University of Chicago Press.

Lee, S.H., DeCandia, T.R., et al. (2012). Estimating the proportion of variation in susceptibility to schizophrenia captured by common SNPs. Nat. Genet. 44:247–250.

Lichtenstein, P., Carlstrom, E., et al. (2010). The genetics of autism spectrum disorders and related neuropsychiatric disorders in childhood. Am. J. Psychiatry 167:1357–1363.

Lichtenstein, P., Yip, B.H., et al. (2009). Common genetic determinants of schizophrenia and bipolar disorder in Swedish families: a population-based study. Lancet 373:234–239.

Millar, J.K., Wilson-Annan, J.C., et al. (2000). Disruption of two novel genes by a translocation cosegregating with schizophrenia. Hum. Mol. Genet. 9:1415–1423.

Miller, G., and Holden, C. (2010). Proposed revisions to psychiatry’s canon unveiled. Science 327:770–771.

National Institutes of Mental Health. Research Domain Criteria (RDoC). http://www.nimh.nih.gov/research-funding/rdoc/index.shtml Accessed December 1, 2012.

Neale, B.M., Kou, Y., et al. (2012). Patterns and rates of exonic de novo mutations in autism spectrum disorders. Nature 485:242–245.

Pope, H.G., Jr., and Lipinski, J.F., Jr. (1978). Diagnosis in schizophrenia and ­manic-depressive illness: a reassessment of the specificity of “schizophrenic” symptoms in the light of current research. Arch. Gen. Psychiatry 35:811–828.

Robins, E., and Guze, S.B. (1970). Establishment of diagnostic validity in psychiatric illness: its application to schizophrenia. Am. J. Psychiatry 126:983–987.

Shendure, J., and Ji, H. (2008). Next-generation DNA sequencing. Nat. Biotechnol. 26:1135–1145.

Spitzer, R.L., Endicott, J., et al. (1975). Research diagnostic criteria. Psychophar­macol. Bull. 11:22–25.

Sullivan, P.F., Daly, M.J., et al. (2012). Genetic architecture of psychiatric disorders: the emerging picture and its implications. Nat. Rev. Genet. 13:537–551.

Takahashi, K., Tanabe, K., et al. (2007). Induction of pluripotent stem cells from adult human fibroblasts by defined factors. Cell 131:861–872.

Wade, N. (1975). Conference sets strict controls to replace moratorium. Science 187:931–935.

World Health Organization. (1992). The ICD-10 Classification of Mental and Behavioural Disorders. Geneva: Author.