Schizophrenia Research Forum - A Catalyst for Creative Thinking

Live Discussion: New Findings in Psychosis: Breaking Down Diagnostic Barriers

Nick Craddock

Mike Owen

View article

View article

Like many other researchers and clinicians, Nick Craddock and Mike Owen of Cardiff University in Wales feel that the traditional boundaries of psychotic and mood disorders may be getting in the way of progress in basic research, particularly genetics. In the past year, they have published a series of papers, both theoretical and experimental, that argue for moving beyond the Kraepelinian dichotomy of schizophrenia and bipolar disorder as distinct disease entities.

On May 15, 2006, Craddock and Owen led us through a live online discussion of these issues, with an eye toward finding alternative approaches to diagnosis and classification. Mayada Akil of the National Institute of Mental Health served as moderator. Please read the edited transcript, two recent review papers (full text can be accessed by clicking on the journal images above), and the text below and continue this discussion by offering comments.

View Transcript of Live Discussion — Posted 27 June 2006

View Comments By:
Gunvant Thaker — Posted 24 April 2006
William Carpenter — Posted 28 April 2006
Avi Peled — Posted 7 May 2006
Mario Maj — Posted 8 May 2006
Patricia Estani — Posted 10 May 2006
Ian Brockington — Posted 12 May 2006
Avi Peled — Posted 13 May 2006
Assen Jablensky — Posted 13 May 2006
Robin Murray — Posted 14 May 2006
Avi Peled — Posted 15 May 2006
Wayne Fenton — Posted 15 May 2006
Carol Tamminga — Posted 15 May 2006
Herbert Meltzer — Posted 15 May 2006
Francis A O'Neill — Posted 24 May 2006
Karl-Ludvig Reichelt — Posted 29 May 2006
Kiumars Lalezarzadeh — Posted 18 June 2006


by Nick Craddock and Mike Owen

It has been conventional for psychiatric research, including the search for predisposing genes, to proceed under the assumption that schizophrenia and bipolar disorder are separate disease entities with different underlying etiologies. These represent Emil Kraepelin's traditional dichotomous classification of the so-called "functional" psychoses and form the basis of modern diagnostic practice. However, findings emerging from many fields of psychiatric research do not fit well with this model. In particular, the pattern of findings emerging from genetic studies shows increasing evidence for an overlap in genetic susceptibility across the traditional classification categories, including association findings at DAOA(G72), DTNBP1 (dysbindin), COMT, BDNF, DISC1, and NRG1. The emerging evidence suggests the possibility of relatively specific relationships between genotype and psychopathology. For example, DISC1 and NRG1 may confer susceptibility to a form of illness with mixed features of schizophrenia and mania. The elucidation of genotype-phenotype relationships is at an early stage, but current findings highlight the need to consider alternative approaches to classification and conceptualization for psychiatric research rather than continue to rely heavily on the traditional Kraepelinian dichotomy.

The aim of this Schizophrenia Research Forum discussion is to discuss the current utility, or otherwise, of the Kraepelinian dichotomy for research and clinical practice and consider this against alternative approaches to diagnosis and classification. To focus discussion, two specific issues will be addressed, in the form of questions:

1) Do the disadvantages of the Kraepelinian dichotomy now outweigh the advantages?

2) What are the best alternative approaches to diagnosis and classification? (e.g., dimensions, alternative categories, prototypes.…)

Craddock N, O'Donovan MC, Owen MJ. Genes for schizophrenia and bipolar disorder? Implications for psychiatric nosology. Schizophr Bull. 2006 Jan;32(1):9-16. Abstract

Craddock N, Owen MJ. The beginning of the end for the Kraepelinian dichotomy. Br J Psychiatry. 2005 May;186:364-6. Abstract


Mayada Akil, NIH/NIMH
A. G. Alias, Fulton State Hospital
Lars Bertram, Harvard Medical School
William Carpenter, Maryland Psychiatric Research Institute
Laurel A Copeland, University of Texas Health Science Center at San Antonio
Nick Craddock, Cardiff University
Rachel Craddock
Pamela DeRosse, Zucker Hillside Hospital
Patricia Estani, Ministry of Health of Argentina
Jinbo Fan, Broad Institute
Diego Forero, University of Antwerp
Anne Gibbs, McLean Hospital
Irving Gottesman, University of Minnesota
Dilihan Gumus, Cardiff University
Mei-Hua Hall, McLean Hospital
Hakon Heimer, Schizophrenia Research Forum
Perry D. Hoffman, National Education Alliance for Borderline Personality Disorder (NEA-BPD)/Mount Sinai School of Medicine
Tetsufumi Kanazawa, Osaka Medical College
Verena Krause, McLean Hospital
Giri Krishnan, Indiana University
Todd Lencz, Long Island Jewish Medical Center
Douglas F. Levinson, Stanford University School of Medicine
Deborah Levy, McLean Hospital
Maureen Martin, Washington University, St. Louis
Robin Murray, Institute of Psychiatry London
Mike Owen, Cardiff University
Sharilyn Rediess, Johnson and Johnson
Marcella Rietschel, University of Heidelberg
Thomas Schulze, University of Heidelberg
Jean Theberge, McLean Hospital
Ming Tsuang, UCSD

Note: The transcript has been edited for clarity and accuracy.

Mayada Akil
Welcome everyone. I would like to start by introducing our two discussants: Professor Michael J. Owen is professor of Psychological Medicine and head of the Department of Psychological Medicine, Wales College of Medicine, UK. He is a psychiatrist/geneticist and heads the Cardiff Neuropsychiatric Genetics group, which is one of the largest such groups in the world. His research interests relate to genetic aspects of schizophrenia, bipolar disorder, Alzheimer’s disease, ADHD and dyslexia, as well as the relationship between 22q11 deletions and psychosis. Professor Nick Craddock is a professor of Psychiatry at Cardiff University. He is also a psychiatrist whose research focus is the molecular genetic investigation of bipolar spectrum mood disorders and psychosis. He has a specific interest in using molecular genetics to refine the clinical phenotype. The topic for discussion today is very timely. Drs. Owen and Craddock have published on this topic and two of their papers are available at the SRF discussion background page. Perhaps Nick or Mike can start the discussion by briefly outlining their thesis. While they are working on it participants should feel free to put out any questions or comments they have.

Nick Craddock/Mike Owen
Recent genetic findings add to a large body of other evidence that conflicts with the traditional dichotomy. We believe now is the time to move to different approaches to classification. Does anyone disagree with this view?

Doug Levinson
Clearly the current schizophrenia-schizoaffective-mood disorder definitions will not stand forever, but I am not convinced that we have enough genetic evidence to suggest what different view should be adopted.

Nick Craddock/Mike Owen
Doug, agreed, not enough for a long-term alternative yet. But researchers need to embrace the deficiencies and collect data that allow analysis unconstrained by the current categories.

Marcella Rietschel/Thomas Schulze
We agree that the current classification is not perfect, but caution against rushing into new classification without solid biological backing. Or, do you think we do already have it?

Nick Craddock/Mike Owen
Marcella, Thomas, we are not suggesting complete abandonment of current systems. However, research is currently being impeded by sticking to a concept of disease entities that do not map onto the biology. Researchers must think outside the box.

Doug Levinson
I would oppose using our current state of genetic knowledge to change DSM-V or ICD-11, but I would favor developing a Research Diagnostic Criteria, Version 2 (RDC-2) that would do what you suggest—guide researchers to collect the kinds of data that would better inform future decisions.

Patricia Estani
I think that the dichotomy manic-depressive disorder/schizophrenia is not a good, or a currently sustainable point of view, because this classification does not fit current psychiatric genetics evidence.

Mayada Akil
Mike, Nick, are you recommending retaining the existing diagnostic categories for clinical purposes but changing them for research purposes?

Nick Craddock/Mike Owen
Mayada, most UK psychiatrists actually already treat according to domains/dimensions of psychopathology, that is, psychosis, depression, mood instability, etc. The problem is that the current classifications and regulatory bodies perpetuate an inappropriate and unhelpful myth of distinct diseases.

Doug Levinson
I don't think it helps clinicians to be constantly changing the "official" criteria unless data are very clear—it just convinces the average clinician that the criteria are not very meaningful. The RDC (Spitzer et al., Research Diagnostic Criteria. [1975] New York State Psychiatric Institute, New York, NY) accomplished the goal of getting researchers to collect data (namely, a great generation of family studies) that made it possible to create DSM-III and the ICD-10.

Nick Craddock/Mike Owen
Doug, of course, changing to definite distinct systems every few years can cause confusion. So what we need is an approach that is not misleading about the current understanding, is clinically useful, and helps, rather than hinders, researchers to unravel the biological basis of disorders.

Marcella Rietschel/Thomas Schulze
We agree. All the participants, just by their mere presence today, demonstrate their interest in thinking outside the box. We think that we should employ innovative strategies with the current data we have, such as covariate-based approaches, reverse phenotyping (use genetic marker data to drive new phenotype definitions).

Nick Craddock/Mike Owen
Marcella, Thomas, we agree about the benefits of using genetics for "reverse phenotyping." Indeed, we develop this reasoning in our recent review of COMT in Molecular Psychiatry (Craddock et al., 2006).

Irving Gottesman
Nick, Mike, I hope you are not voting for a return to the notion mid-1850s of a Unitary Psychosis. Kraepelin is not really dead is he?

Nick Craddock/Mike Owen
Irv, no.

Maureen Martin
Like Herbert Meltzer wrote, there are probably unique and common genetic (and environmental) risk factors for these disorders. I think it is too early to replace the nosology—it may still be better to use the old definitions in a research setting and to use these definitions to continue to identify common and separable endophenotypes and genetic risk factors of each classification. This might make linking endophenotypes and genetic risk factors a little easier so we can better dissect these disorders (based on genetic/biological markers) in the future (in another 10 to 20 years?).

Mayada Akil
Nick, Mike, what would that approach look like?

Nick Craddock/Mike Owen
Mayada, pretty well all disease classifications are mixtures of defined pathological entities and more or less well-defined clinical syndromes. Thus, it is to be expected that this will be the case in psychiatry as our knowledge develops. Therefore, we might find some relatively discrete syndromes that have discrete biology but others that are better conceptualized on a continuum.

Patricia Estani
I think that the model that Craddock and colleagues presented in their reviews is not a biologically closed box. Instead, I agree that the new diagnostic classifications must be catalyzed by molecular genetic findings.

Irving Gottesman
Nick, Mike, I hope that "reverse phenotyping" incorporates the strategies of endophenotyping discussed in the first SRF online discussion.

Deborah Levy
Irv, I second that.

Marcella Rietschel/Thomas Schulze
Irving, endophenotypes are a major basis, provided your criteria are held up. But actually, we should not overwhelm clinicians with the rapid changes we are hoping to make.

Nick Craddock/Mike Owen
Irv, of course, endophenotypes are desirable for reverse phenotyping. The question is how best to use them in forward studies.

Marcella Rietschel/Thomas Schulze
We think that it is most important to realize that the classifications are temporary, especially when clinicians try to develop guidelines for how to treat patients with given diagnoses. Otherwise, considering the limited therapeutic options we have, it does not really matter what diagnosis you give to patients. As long as they are treated individually, that means that their symptoms are treated adequately.

Robin Murray
It does matter what diagnosis you give to patients, as many psychiatrists believe that we have treatments for diagnostic entities, for example, antipsychotics for schizophrenia. However, we don’t. Instead, the sensible clinician will use a dimensional approach and treat positive, negative, manic, and depressive symptoms.

Nick Craddock/Mike Owen
Robin, we absolutely agree.

Robin Murray
A major question is whether any alternative classification can rely on genetic or biological markers, or whether the best alternative is to use symptom dimensions based on factor analysis.

Nick Craddock/Mike Owen
Robin, a biological basis is best.

Robin Murray
Nick, Mike, obviously a biological basis is better if the measure is reliable and has some validity. The difficulty at present is that susceptibility genes cannot be said to have been sufficiently confirmed.

Nick Craddock/Mike Owen
Robin, we are not in any way suggesting that the current genetic findings are sufficient to decide on precise alternative. Rather, our argument is that they are sufficient to show that we need to be thinking about the possible alternatives and—crucially for researchers—already analyzing our existing data in ways that cut across the traditional dichotomy.

Doug Levinson
As I understand it, a diagnosis should predict familial aggregation, course, and treatment response. Regarding treatment, it is true that most psychiatrists now medicate psychotic patients as if the schizophrenia-like and mood components should be treated separately. But this does not seem to be true; people with chronic schizophrenia-like illnesses do not seem to derive much benefit from the constant additions of antidepressants and mood stabilizers (except for antidepressants in "post-psychotic" major depression). This suggests that we are not quite ready for a unitary model, and that simply using factor analyses would not improve the situation. Robin, Nick, Mike, what evidence exists that when someone with chronic psychosis develops transient mania-like symptoms, addition of a mood stabilizer is effective, versus optimization of the antipsychotic regimen?

Robin Murray
Doug, my view is that if individuals develop manic symptoms at any point in their illness, then they should be treated for this by a mood stabilizer. Such a view would suggest that schizophrenia-like symptoms can arise out of 1) a primarily affective disorder or 2) a primarily developmental disorder.

Diego Forero
I agree that there is a need for experimental-based conceptions of mental disease classifications (going further from historical DSM-IV definitions). However, a lot of empirical data is required to achieve this purpose (to find shared regions in linkage studies or association with a same genetic polymorphism for bipolar and schizophrenia is only a small part). The Owen and Craddock proposal is a good starting point for innovative and important approaches in the field.

William Carpenter
Nick, Mike, what is the easiest explanation for family pedigree studies suggesting separate diseases and linkage/genotype studies suggesting overlap?

Nick Craddock/Mike Owen
Will, 1) lack of power and 2) starting with extreme phenotype.

William Carpenter
Nick, Mike, so, with good power and average phenotype, do we get more or less overlap between current disease classes?

Nick Craddock/Mike Owen
Will, the studies have not been done.

Doug Levinson
Will, I agree with Robin that genetic association findings are not yet sufficiently confirmed to support nosological changes, and I would say that if you compare the largest schizophrenia (Lewis et al., 2003) and bipolar (McQueen et al., 2005) linkage meta-analyses, as power has increased, the results do not look terribly convergent.

William Carpenter
Nick, Mike, so the studies have not been done. But don't you clearly expect the genotypes to beat diagnostic types and result in falsifying the current nosology? DSM-V is due in about 5 years. When will we know enough to redesign classes? Or do we move entirely to domains? If you have avolition pathology, that's that. If you also have cognition impairment, then that’s that, too, etc.

Nick Craddock/Mike Owen
Will, we are currently undertaking Whole Genome Association (500,000 SNPs) in 2000 BP spectrum case and will have results by end of 2006. This will not give any definitive answer but may well indicate likely possibilities. We agree that aiming to identify domains of psychopathology that map onto biological systems is the way to go.

Diego Forero
A question for Lars: what is your opinion about the use of several disease subtypes for genetic explorations of mental disorders? A real approach to discover specific and minor genetic effects or only a statistical trick (multiple testing)?

Lars Bertram
Diego, I think right now it is both, unfortunately.

Diego Forero
Thank you, Lars.

Nick Craddock/Mike Owen
Diego, Lars, there are approaches to take account of multiple testing. And replication is the acid test.

Lars Bertram
Nick, Mike, clearly, but I think the confusing and often disagreeing "evidence" tells us that the current methods are just not there yet. We hope that our soon-to-be-launched "SchizophreniaGene" database project is going to fill that gap. [Edit. Note: The SchizophreniaGene database, patterned on AlzGene, will debut online at SRF some time later this year.]

Nick Craddock/Mike Owen
Lars, perhaps. But the problem is current studies use categories that do not map well onto the biology. In our COMT review (Craddock et al., 2006) we show that using the wrong phenotype category causes enormous power loss.

Lars Bertram
Nick, Mike, loss in power can be cured by an increase in sample size; meta-analysis is one easy way to do that. Merging or combining phenotypes may lead to spurious results, at least when there is no real pathogenetic connection between them. Some of these efforts may amount to mere "trial and error." I do realize that that's sometimes what it takes, though.

Patricia Estani
I would like to ask Dr Carpenter, if any point of beginning exists in a phenotype's classification, at a behavioral level, which point is the first one? What type of classification could be the first?

William Carpenter
Patricia, explain a little more what you mean by the beginning point. Is this pathology as a dimension which includes normal behavior?

Patricia Estani
At a behavioral level, what kind of classification could be used, for example, in the case of schizophrenia, that would be more adaptable to genetic findings. Into how many categories could we define schizophrenia in the most simple way, at a behavioral level. For example, is avolition a dimension or symptom of a determined symptomatic category? Or do you explain that avolition itself is a category or dimension?

William Carpenter
Patricia, I would break schizophrenia into a number of symptom categories (e.g., negative, reality distortion, disorganization, obsessive-compulsive, motor), a number of cognitive and physiologic phenotypes (some of which are endophenotypes). Each domain would be of interest for study and application in treatment development. But I don't know what "recombinations" would then make sense for a disease classification scheme.

Mayada Akil
Wayne Fenton, in his preliminary comments, raised the question of whether the dichotomy holds up when we consider schizophrenia with deficit syndrome. Any reactions?

Nick Craddock/Mike Owen
Mayada, Dr. Fenton suggested that deficit state was an important variable. We agree. This highlights the need for studies with detailed phenotyping and a longitudinal perspective.

William Carpenter
I would add that one recombination of interest to us is the use of primary negative symptoms to define deficit schizophrenia which looks like a disease entity within the schizophrenia syndrome, separated from other forms of schizophrenia on many clinical features, but also biomarkers and etiologic risk factors.

Nick Craddock/Mike Owen
Does anyone know of good longitudinal data sets that could be used for genetic analysis?

Irving Gottesman
Nick Mike, three cheers for a longitudinal perspective sensitive to the fact that diagnoses/phenotypes will not stand still during a lifetime of variable exposures to stressors and good stuff as well. See Angst for patients with multiple depressive episodes who then develop mania (Angst et al., 2005), and Shields/Gottesman twins at the Maudsley whose initial diagnosis was depression but who on follow-up had clear schizophrenias (Gottesman, I.I. and Shields, J. [1972]. Schizophrenia and Genetics: a twin study vantage point. New York: Academic Press; Clark and Mallett, 1963). And see Odegaard's Norwegian pedigrees with plenty of crossovers among the psychoses (Gottesman and Shields, 1976, Table 6).

Doug Levinson
I think it will take three or four adequately powered (2,000-3,000 cases) whole genome association studies of schizophrenia and of bipolar disorder, and follow-ups to each study, to start to get a clearer picture. I agree with Nick and Mike that new thinking about phenotyping should be developed now and applied to some of these studies. It would be better, for example, to collect all psychoses using a richer phenotyping scheme. But, it is difficult to collect thousands of subjects longitudinally, and it is incredibly difficult to collect truly valid data retrospectively about the temporal course of psychotic, mood, cognitive, etc., symptoms. Perhaps an RDC-2 would serve as a model for efforts to get clinicians to begin recording more information about symptoms over time, or perhaps we can develop better technologies for doing it.

Deborah Levy
Power could be substantially improved by incorporating endophenotypes into linkage/association studies.

Nick Craddock/Mike Owen
Deborah, perhaps.

Irving Gottesman
Deborah, for sure.

Nick Craddock/Mike Owen
Doug, an appropriately constructed RDC-2 is a possible way forward. The crucial issue is to get researchers to collect the same sufficiently detailed information and to think across existing categories. At the moment, conferences have "schizophrenia" sessions and "bipolar" sessions. Not very useful for joined up thinking!

Doug Levinson
I used the term "RDC-2" because I thought that the principle of RDC is worth remembering: instead of trying to embody a single point of view, RDC tried to capture any categories (given that it was a categorical system) that anyone thought were worth studying, letting the future data speak for themselves. (Keep in mind, though, that the RDC did provide many categories for mixed syndromes, none of which has held very well compared with the old dichotomy—we should be prepared for any outcome, even one which is closer to Kraepelin.) An RDC-2 would have to include a purely dimensional approach as well as a variety of syndromal approaches.

Nick Craddock/Mike Owen
Doug, we believe an RDC-2 should be sufficiently flexible to include both dimensional and categorical approaches.

Hakon Heimer
Mike, Mayada, you were recently at a meeting to discuss definitions of psychosis for DSM-V, and some others of you may have been part of that discussion. Were there any useful concepts/objections, etc., that came up in that meeting?

Carol Tamminga
The APA sponsored a meeting to debate the dimensional versus diagnostic approaches to psychotic diseases from multiple perspectives, including genetics, molecular, and pharmacologic. There was considerable enthusiasm for pursuing a dimensional approach to psychosis for the purpose of research into its molecular basis. There was some caution expressed at the idea of fully changing diagnostic boundaries for practical reasons.

Robin Murray
My impression of the DSM-V preliminary meeting was that there was little research support for the present dichotomy but that none of the biologists present was prepared to say that his or her field, for example, genetics, imaging, neuropsychology, etc., would be superior. The only system that has been shown to be superior to categories is a symptom dimension one as demonstrated by Van Os (Van Os et al., 1996). Such a system is readily combined with genetics.

Nick Craddock/Mike Owen
Hakon, there was a majority opinion that it was time to move on. We agree with Robin that the Van Os dimensional approach received favorable reception at the meeting. And I agree with Carol that most were cautious about abandoning the current system for clinical purposes. An important point we would like to make is this: we do not argue that all the current details will prove to be correct. Rather, our comments refer to the big picture. It is clear from genetic findings to date that the dichotomy is 1) unsupported and 2) unhelpful to discovery.

Diego Forero
Lars, when will the initial results of the SchizophreniaGene database be available?

Lars Bertram
Diego, we hope to be presenting the first preliminary data in October or November of this year.

Diego Forero
Very good, Lars.

Mayada Akil
Everybody, since we will be wrapping up soon, please type in any final questions or comments. Note, however, that you can post comments after the discussion.

Patricia Estani
I think that molecular genetics specialists must work more closely with behavioral specialists, like psychologists interested in behavioral genetic applications to psychiatric research.

Diego Forero
A question for Thomas and Marcella: aside from the ethnic differences, what do you think about the differences between the findings of your paper in AJP 2005 and those from the Williams et al. paper in AGP 2006 (Williams et al., 2006; see SRF news/commentary)? I think that it is an interesting point.

Marcella Rietschel/Thomas Schulze
Sorry, Diego, to which differences do you refer?

Diego Forero
I mean the association of DAOA with specific psychosis phenotypes versus major depression.

Marcella Rietschel/Thomas Schulze
Actually, we have found the DAOA associated with schizophrenia, bipolar disorder (especially those with persecutory delusions), panic disorder, and major depression.

Nick Craddock/Mike Owen
Thank you, everyone—an excellent discussion. Apologies for brief answers and erratic typing. Apologies also if we haven't answered some questions. It has been a bit like playing simultaneous chess against world grandmasters!

Hakon Heimer
That's what Irv said after the last one!

Stephen Shanfield
This is such a complex topic that it is worthy of other chat sessions.

Doug Levinson

Ming Tsuang
Thanks. Worthwhile discussion.

Mayada Akil
This was fun. Thanks for coming.

Nick Craddock/Mike Owen
Bye, everyone.

Patricia Estani
It was a good discussion. Good afternoon from Argentina. Until the next discussion, Hakon and Nick.

Diego Forero

Marcella Rietschel/Thomas Schulze
Auf Wiedersehen.

Comments on Online Discussion
Comment by:  Gunvant Thaker
Submitted 23 April 2006
Posted 24 April 2006

Nick Craddock and Michael Owen in their editorial in the British Journal of Psychiatry, and both authors with Michael O’Donovan in their recent review paper in Schizophrenia Bulletin, present an interesting and convincing set of arguments supporting the presence of an overlap between schizophrenia and bipolar disorders (Craddock and Owen, 2005; Craddock et al., 2006). Even though the diagnostic dichotomy has been debated ever since Emil Kraepelin proposed the distinction between dementia praecox and manic-depressive insanity, this discussion is very timely in view of recent genetic findings.

Craddock and Owen present four sets of evidence summarized in their two papers: 1) there is familial co-aggregation of schizophrenia and bipolar disorders; 2) a recent twin study that was not constrained by diagnostic hierarchal method showed overlap in genetic susceptibility (Cardno et al., 2002); 3) there is an overlap in linkage “hot spots” (Berrettini, 2003; Craddock et al., 2006); and 4) several genes including DAOA (G72)/G30), BDNF, DISC1, NRG1, and dysbindin show significant associations with both disorders (Craddock et al., 2006). This evidence forms the basis of the argument for breaking down the diagnostic barriers between these two and perhaps among other psychotic illnesses. Although I find their arguments convincing, in my comments below I introduce a note of caution and put forward an alternative proposal in place of breaking down the boundaries.

First, the findings from the family studies provide the weakest argument for an extensive overlap between the two psychotic disorders. The majority of family studies carried out in the 1980s and 1990s, as well as a couple of studies published in the past few years, support the view that schizophrenia and bipolar disorders breed true with minimal increase in cross diagnostic risk. Ken Kendler reviewed data from the Danish Adoption Study, the Iowa 500 non-500 family study, and the Roscommon Family Study and reached the same conclusion (Kendler and Gardner, 1997). A recent family study from India that included bipolar and schizophrenia probands, and another family study that was population-based including a cohort of 2.4 million individuals born after 1952 in Denmark, support this view (Somnath et al., 2002; Laursen et al., 2005). There is a minority of studies that show familial coaggregation (reviewed by Craddock et al., 2006), and it’s unclear whether these findings are specific to a sample or a region affected by unusual factors (e.g., assortative mating across diagnostic categories). The single twin study that showed shared susceptibility used cross-sectional diagnosis (in order to obtain non-hierarchal diagnosis), which tends to be very unreliable—the interrater reliability in that study was modest, ranging from 0.58 to 0.68 (kappa).

In contrast to the findings from the family studies, data from the linkage and candidate gene association studies are relatively robust. This leads to my second comment—there are likely genes or environmental factors unique and necessary to develop bipolar disorder or schizophrenia (thus explaining why these disorders breed true). Other genes may determine the clinical manifestation of the two disorders, for instance, whether there is a strong affective component in schizophrenia pedigrees. This view is supported by secondary analyses of linkage studies in bipolar and schizophrenia families, where the investigators focused on subgroups of families based on psychosis or affective symptoms (Potash et al., 2001; Hamshere et al., 2005). This provides a method of reducing heterogeneity within both schizophrenia and bipolar disorders, and helps the search for susceptibility genes with small effects. Thus, rather than break down the Kraepelinian barriers, I would propose to develop methods that would identify homogenous subgroups of families based on phenotypic characteristics.

This leads to my third and last comment. It’s unlikely that clinical symptoms may be adequate in identifying all of the subgroups within the two disorders, including within the clinical phenomena shared by bipolar and schizophrenia disorders. Under these circumstances, the use of biological measures that are stable and heritable (so-called endophenotypes) may help in reducing heterogeneity within schizophrenia and bipolar disorders, and better define boundaries among various sub-syndromes within these disorders. (See the recent interesting SRF discussion organized by Irving Gottesman [University of Minnesota] and Mayada Akil [NIMH].) Although not extensively studied, data suggest that many of these endophenotypes mark independent aspects of schizophrenia risk, and some overlap with the bipolar illness (Hong et al., in press; Sweeney et al., 1998; and Strasser et al., 2005). The use of endophenotypes will provide a useful approach to diagnosis and classification in our research, identify homogenous subgroups, and help identify genes and other etiological factors unique to schizophrenia and bipolar disorder as well as those shared by the two disorders. This has a tremendous implication for novel treatment development that is likely to be specific.


Berrettini W. Evidence for shared susceptibility in bipolar disorder and schizophrenia. Am J Med Genet C Semin Med Genet. 2003 15:59-64. Abstract

Cardno AG, Rijsdijk FV, Sham PC, Murray RM, McGuffin P. A twin study of genetic relationships between psychotic symptoms. Am J Psychiatry. 2002 159:539-45. Abstract

Craddock N, O'Donovan MC, Owen MJ. Genes for schizophrenia and bipolar disorder? Implications for psychiatric nosology. Schizophr Bull. 2006 32:9-16. Abstract

Craddock N, Owen MJ. The beginning of the end for the Kraepelinian dichotomy. Br J Psychiatry. 2005 186:364-6. Abstract

Hamshere ML, Bennett P, Williams N, Segurado R, Cardno A, Norton N, Lambert D, Williams H, Kirov G, Corvin A, Holmans P, Jones L, Jones I, Gill M, O'Donovan MC, Owen MJ, Craddock N. Genomewide linkage scan in schizoaffective disorder: significant evidence for linkage at 1q42 close to DISC1, and suggestive evidence at 22q11 and 19p13. Arch Gen Psychiatry. 2005 62:1081-8. Abstract

Hong, LE, Summerfelt A, Wonodi I, Adami HM, Buchanan RW, Thaker GK: Independent Domains of Sensory Gating in Schizophrenia and the Effect of Stimulus Interval. American Journal of Psychiatry, in press.

Kendler KS, Gardner CO. The risk for psychiatric disorders in relatives of schizophrenic and control probands: a comparison of three independent studies. Psychol Med. 1997 27:411-9. Abstract

Laursen TM, Labouriau R, Licht RW, Bertelsen A, Munk-Olsen T, Mortensen PB. Family history of psychiatric illness as a risk factor for schizoaffective disorder: a Danish register-based cohort study. Arch Gen Psychiatry. 2005 62:841-8. Abstract

Potash JB, Willour VL, Chiu YF, Simpson SG, MacKinnon DF, Pearlson GD, DePaulo JR Jr, McInnis MG. The familial aggregation of psychotic symptoms in bipolar disorder pedigrees. Am J Psychiatry. 2001 158:1258-64. Abstract

Somnath CP, Janardhan Reddy YC, Jain S. Is there a familial overlap between schizophrenia and bipolar disorder? J Affect Disord. 2002 72:243-7. Abstract

Strasser HC, Lilyestrom J, Ashby ER, Honeycutt NA, Schretlen DJ, Pulver AE, Hopkins RO, Depaulo JR, Potash JB, Schweizer B, Yates KO, Kurian E, Barta PE, Pearlson GD. Hippocampal and ventricular volumes in psychotic and nonpsychotic bipolar patients compared with schizophrenia patients and community control subjects: a pilot study. Biol Psychiatry. 2005 57:633-639. Abstract

Sweeney JA, Strojwas MH, Mann JJ, Thase ME. Prefrontal and cerebellar abnormalities in major depression: evidence from oculomotor studies. Biol Psychiatry. 1998 43:584-594. Abstract

View all comments by Gunvant ThakerComment by:  William Carpenter, SRF Advisor (Disclosure)
Submitted 27 April 2006
Posted 28 April 2006

Very important topic coming at a good time. Over the next five years we will have the task of formulating classification and criteria for DSM-V. I have a responsibility for psychosis and schizophrenia, and anticipate substantial interest in pathologic domains which identify subgroups within each major diagnostic class, and determine how these domains cut across diagnostic lines. In regard to the genetic vulnerabilities, it is difficult to imagine entirely distinct "genetics" for similar psychopathology found in different disease classes. It seems to me more likely that a domain such as restricted affect breeds true than a diagnostic class. If restricted affect is common in schizophrenia, but less common in some other disorders, it will bias schizophrenia towards "breeding true." But restricted affect (in this example) would be the psychopathology of interest. What will it mean for classification if we find that genetic vulnerability for restricted affect in schizophrenia is the same as found with restricted affect in other psychotic disorders? I think it will mean a great deal for classification, and hope we have enough data and wisdom to know what it means!

Guni Thaker's comments are very informative. The Craddock/Owen publications are a terrific basis for this discussion.

View all comments by William CarpenterComment by:  Avi Peled
Submitted 5 May 2006
Posted 7 May 2006

Allen Frances, the chairperson of the task force that produced DSM-IV, and Helen Egger have written, “We are at the epicycle stage of psychiatry where astronomy was before Copernicus and biology before Darwin” (Francis and Egger, 1999). This is so true when you consider our current levels of discussion, for example, in this online discussion. I apologize if this saying seems provocative but it is time for psychiatry to go beyond its primitive current state. We are still debating mental disorders at levels which have no future; at the molecular level we talk about genes and try to understand the level of emergent properties from whole-brain organizations, that is, symptoms. In between genes (molecular level) and symptoms (emergent properties from whole-brain organizations) there are 1010 neurons, each connected (nonlinear transmission) with 107 connected neurons. This level of synaptic neural network is ignored, offering a black box for psychiatrists. Frustrating is the fact that our colleagues, the neuroscientists, have already embraced the science that best deals with the brain's higher functions—the emerging (and already established) science of neural computation.

According to insights from neural computation, the best explanatory level for understanding mental disorders as brain disorders is the synaptic connectivity neural-network level. It is an intermediate level between the molecular level and the brain organization level, and it has explanatory power going down to the molecular level, as well as to the whole brain organization level. Neural computation teaches us that synaptic dynamics involve fast, millisecond range plasticity (i.e., connectivity changes) and slow plasticity (e.g., BDNF-related spine formations and deletions). Slow plasticity ranges over days and weeks, the time serotonin specific reuptake inhibitors (SSRIs) take to produce BDNF-related changes and remission from depression. There is initial evidence that psychosis involves breakdown of fast integrative processes dependent on fast plasticity. In this way, for example, brain systems involving heard speech can act disconnected and independent from the rest of the brain, allowing for auditory hallucinations to be experienced. The slow plasticity processes, also defined as changes in neuronal resilience (Charney and Manji, 2004), have been related to mood changes. Since fast and slow processes act in parallel and are interdependent, it should not be surprising that psychosis and mood disorders appear together clinically. Their clinical coexistence will not be explained by genes. Rather, the best explanation will come at the plasticity neural network level, and the genes will have relevance if they explain activity at the plasticity neural network level. It is no wonder that the DSM-IV and earlier formulations are misleading—they are a fruit of our ignorance. In the Research Agenda for DSM-V, it has been stated that the DSM-defined syndromes have failed to describe distinct classifiable entities, despite enormous research efforts; none of the DSM-defined syndromes correlates with any neurobiological phenotypic marker or gene that could have etiological relevance. The efficacy of medications cuts across the DSM-defined categories, as both anxiety disorders and depression respond to SSRIs. The DSM-V Agenda states, “Reification of DSM-IV entities, to the point that they are considered equivalent to disease, is more likely to obscure than to elucidate research findings.” The Agenda for DSM-V calls for a “paradigm shift” in psychiatric diagnosis. I believe the paradigm shift will come when insights from neural computation are finally integrated into psychiatry. (More about this can be found in my discussion in SRF Idea Lab), in my book entitled Brain Dynamics and Mental Disorders, and the website Theoretical-Psychiatry). As for psychosis, once schizophrenia and bipolar disorders are reconceptualized as disturbances of synaptic plasticity, it is just natural that their clinical manifestations will overlap. The degree of one or the other manifest in clinical observation is simply the degree of changes in the time domain, that is, millisecond range versus days and weeks. The neural-computation approach is not my original idea; it has been already discussed in various forms by distinguished authors such as Bernard J. Baars, Michael Breakspear, Nancy Andreasen, Ed Bullmore, Jonathan D. Cohen, David Copolov, Karl Friston, Ralph Hoffman, Assen Jablensky, Martha Koukkou, Peter Liddle, James L McClelland, Eitan Ruppin, Giulio Tononi, and others. I try to take these ideas and put them to clinical relevance.


Frances AJ, Egger HI: Whither psychiatric diagnosis Aug NZJ Psychiatry 1999, 33:161-165. Abstract

Charney DS, Manji HK. Life stress, genes, and depression: multiple pathways lead to increased risk and new opportunities for intervention. Sci STKE. 2004 Mar 16;2004(225):re5. Review. Abstract

Peled A. Brain dynamics and mental disorders, a project for a scientific psychiatry. Yozmot Heiliger 2004.

View all comments by Avi PeledComment by:  Mario Maj
Submitted 8 May 2006
Posted 8 May 2006

I think that what we call "schizophrenia" and "bipolar disorder" are the extremes of a continuum of psychotic conditions. At present, bipolar disorder is reasonably well characterized in the two main diagnostic systems, whereas schizophrenia is not. The various mixed and intermediate syndromes have been described on the basis of an artificial balance between the "affective" and "psychotic" components rather than on a careful characterization of what we see in clinical practice (the literature on "cycloid psychoses" is a significant exception).

I think that a dimensional approach is not mature yet. "Positive symptoms" in a typical case of schizophrenia are different from those of a typical case of psychotic depression, and have different therapeutic implications. Therefore, I do not see how a "positive" dimension could be applied in both cases without generating misunderstandings.

In my opinion, in the forthcoming edition of the DSM and the ICD, we should still try to refine the categorical definitions of schizophrenia, bipolar disorder, and cycloid psychosis/bouffée delirante. Some specifiers could be introduced in the category of schizophrenia. A dimensional approach, with precise guidelines for the clinician, could be proposed in the final appendix of the system, in order to facilitate the systematic collection of information about the feasibility and utility of the approach in clinical settings.

View all comments by Mario MajComment by:  Patricia Estani
Submitted 10 May 2006
Posted 10 May 2006

I would like to express my general agreement with the opinions cited by the authors in both articles. The model of classification that they propose in the second article is attractive and well defined (Craddock et al., 2006). This proposal is a model that adapts the classifications to the new neurobiological theories of psychiatry, especially complex psychiatric genetics.

I think that the current classifications are not useful anymore, not only in a research but also in a clinical sense. They are so phenomenological and so extensive in their repertoire of symptoms. These classifications do not fit with actual advances in neurobiological research in psychiatry, especially with the field of psychiatric genetics.

I think that if the article has any fault, this would be that more emphasis must be placed on the behavioral components of the different disorders. Dr. Gottesman, in his review of the endophenotype concept, discussed the need for reviewing our concepts of diagnosis in psychiatry. Thus, this topic is certainly not new. In the next decade, a real change in our concepts will be realized.

I would also like to remark on the excellent comment of Dr. Avid Peled. The functioning of neural networks in computational models is one of the most interesting paradigms of understanding the brain's pathology. Dr. Peled presented the symptoms of schizophrenia in an interesting new paradigm of distributed, parallel networks processing. I think that this shift in our point of view is very important.

View all comments by Patricia EstaniComment by:  Ian Brockington
Submitted 12 May 2006
Posted 12 May 2006

In my editorial in the journal European Psychiatry (1992, volume 7: 203-207) entitled “Schizophrenia: yesterday’s concept,” I set out the case against this concept. Briefly summarized, it’s a gross oversimplification. It’s a fiction, with no natural (as opposed to socially imposed) boundaries. It has no agreed central principle, since there are three alternative essential or defining features—progression to defect, “nuclear symptoms” (disorders of will and self), and the solipsism (autism) that underlies delusion-formation and thought disorder. Its history is one of semantic wrangles. This half of the Zweiteilungsprinzip is fundamentally flawed, and should be abandoned.

As for the alternative, great progress has already been made in reclassifying patients who used to be termed “schizophrenic.” In terms of categories, there are a number of possibilities, including various types of delusional psychoses, psychoses based on auditory hallucinosis or disorders of will and self, and hebephrenia. A particularly important concept is that of cycloid or acute polymorphic psychosis, which has a hundred years of history and applies to a large subgroup of patients, but is at present only used by a few European centers. In terms of dimensions, factors or vectors, delusion-formation, auditory hallucinosis, and various components of defect are the most obvious candidates for study.

The other half of the dichotomy—manic depressive psychosis—has a completely different status. (It is important not to use the term “affective psychosis,” because “affect” means so much more than mania or depression, and the inclusion of depression shifts the focus from bipolarity to mood disorders.) Bipolar disorder is a strongly supported concept. The fact that an illness exists in two opposite forms indicates a disorder of brain function that must be common to all members of the group. In our statistical study “Bipolar disorder, cycloid psychosis and schizophrenia: a study using lifetime psychopathology ratings, factor analysis and canonical variate analysis,” (European Psychiatry 1991, volume 6: 223-236), bipolar disorder emerged as a distinct group again and again, whatever criterion groups were used. Bipolar psychoses could be the cornerstone of the nosology of the psychoses. But it is unsatisfactory that typical mania is so important in the definition. The group includes cyclothymia (possibly also hyperthymia), seasonal affective disorder, catatonia, possibly cycloid (acute polymorphic) psychosis, some recurrent familial endogenous depression, non-organic puerperal and menstrual psychoses, steroid and postoperative psychoses and 48-hour cyclers, in addition to mania, schizoaffective mania, and hypomania provoked by various drugs. This category is a valid object for scientific inquiry, but its boundaries need clarification by empirical studies.

View all comments by Ian BrockingtonComment by:  Avi Peled
Submitted 13 May 2006
Posted 13 May 2006

I agree with Ian Brockington that schizophrenia is yesterday’s concept; still, his alternative remains in the descriptive realm. It is time to have an etiologic diagnosis for schizophrenia and, for that matter, for all mental disorders.

View all comments by Avi PeledComment by:  Assen Jablensky
Submitted 13 May 2006
Posted 13 May 2006

Despite an ever increasing volume of research data, the apparent stalemate of the search for specific genetic causes of schizophrenia and bipolar disorder raises doubts about the validity of the dichotomous definition of the clinical phenotypes, leading to proposals to replace it with a unitary clinical continuum of “psychosis.” Given the poor coherence of the clinical, genetic and other biological findings in both schizophrenia and bipolar disorder, such doubts are justified and the critical discussion initiated by Craddock and Owen is timely.

As a prefatory note, I should like to point out that Kraepelin was clearly aware of the extent of overlap between the two broad syndromes when he wrote (in 1920) that dementia praecox and manic-depressive insanity do not necessarily imply distinct pathological processes, but rather indicate different “areas of our personality” affected by a process (or processes) that are primarily genetic. The “disease entity” was a working hypothesis, provisionally based on marked differences in their long-term outcomes, due to be eventually replaced by specific neuropathology and pathophysiology. This is still a remote goal. Both schizophrenia and bipolar disorder remain broad clinical syndromes, diagnosed primarily by reported, characteristic subjective phenomena. The inconsistent results of genetic linkage and association studies using solely the DSM-IV diagnosis as the phenotype suggests a “mission impossible“ situation, where we seek to discover specific genes contributing to overinclusive diagnostic categories for which no specific biological substrates have yet been identified, possibly due to extensive genetic heterogeneity and an admixture of different underlying disease subtypes.

Simply dismantling the presumed dichotomy in favor of a general “psychosis” spectrum is unlikely to result in an alternative model that would substantially increase the power of genetic analysis to disambiguate the genotype-phenotype relationships in these disorders as long as we retain a poorly measurable, qualitative phenotype. The possible existence of etiologically different subtypes of schizophrenia is rarely considered in genetic studies, which implicitly assume a unitary view of the disorder. Although locus and allelic heterogeneity is generally acknowledged and genetic linkage analysis often performed under the assumption of heterogeneity, this is usually done post hoc, that is, after the data have been collected, or by default, when difficult to interpret results have been obtained. Phenotype refinement through “splitting,” that is, disaggregation into biologically anchored endophenotypes, or extension by covariate quantitative traits, has been a successful strategy in the genetic dissection of other complex diseases, including cancers, asthma, type I diabetes, Parkinson’s disease, or dementia. This approach has had limited following in schizophrenia research. The failure to address the unresolved heterogeneity in schizophrenia and bipolar disorder continues to be a serious obstacle to the effective harnessing of novel technologies, such as whole-genome association analysis, or joint association and expression studies, into an effort to “deconstruct” schizophrenia.

View all comments by Assen JablenskyComment by:  Robin Murray, SRF Advisor
Submitted 14 May 2006
Posted 14 May 2006

I largely agree with the argument put forward by Craddock and Owen. The evidence which convinces me that schizophrenia and bipolar disorder are not genetically distinct is 1) the Maudsley Twin study of Cardno et al (2002), whose findings Craddock and Owen cite, and 2) studies of families multiply affected with schizophrenia or with bipolar disorder which show certain similar white matter and ERP abnormalities in both sets of patients and relatives (McDonald et al., 2004).

Assuming the existence of some common susceptibility genes, it is then important to understand why schizophrenia and bipolar disorder are not identical. Neuropsychological and grey matter deficits are much more noticeable in schizophrenia, as are neurological soft signs. When do these differences originate? Cohort studies show that children who later develop schizophrenia have an excess of subtle neuromotor and cognitive impairments. Children who later develop bipolar disorder do not share these; indeed, some studies suggest that they appear to be superior to the normal population in motor development and in examinations at school (Cannon et al., 2002). Furthermore, the risk-increasing effect of obstetric complications appears to be confined to schizophrenia. Exposure to perinatal hypoxia is known to result in smaller volume of the amygdala and hippocampus, which are reduced in schizophrenia but not in bipolar disorder; familial predisposition to schizophrenia has also been associated with decreased volume of these structures.

For these reasons, my colleagues and I suggested (Murray et al., 2004) that while the most likely explanation for the similarities between schizophrenia and bipolar disorder is shared susceptibility genes, the most likely explanation for the differences is that the former but not the latter is subject to additional genes or early environmental hazards which impair neurodevelopment. Any alternative dimensional classification will need to take these developmental differences into account.


Cardno AG, Rijsdijk FV, Sham PC, Murray RM, McGuffin P. A twin study of genetic relationships between psychotic symptoms. Am J Psychiatry. 2002 Apr;159(4):539-45. Abstract

Cannon M, Caspi A, Moffitt TE, Harrington H, Taylor A, Murray RM, Poulton R. Evidence for early-childhood, pan-developmental impairment specific to schizophreniform disorder: results from a longitudinal birth cohort. Arch Gen Psychiatry. 2002 May;59(5):449-56. Abstract

McDonald C, Bullmore ET, Sham PC, Chitnis X, Wickham H, Bramon E, Murray RM. Association of genetic risks for schizophrenia and bipolar disorder with specific and generic brain structural endophenotypes. Arch Gen Psychiatry. 2004 Oct;61(10):974-84. Abstract

Murray RM, Sham P, Van Os J, Zanelli J, Cannon M, McDonald C. A developmental model for similarities and dissimilarities between schizophrenia and bipolar disorder. Schizophr Res. 2004 Dec 1;71(2-3):405-16. Review. Abstract

View all comments by Robin MurrayComment by:  Avi Peled
Submitted 15 May 2006
Posted 15 May 2006

As Assen knows, I agree and support the idea that the current nosology has brought to a halt relevant advancement in the finding of the etiology of these disorders. Other then grouping symptoms diversely or trying to link specific phenomena to a gene (like was popular recently with impulsivity instead of with a DSM-related syndrome) I believe a theoretical framework should reconceptualize for us what mental disorders really are. If this framework is brain-related, obeying our knowledge about the brain as a complex non-linear system, then findings will start to emerge. In such a framework, psychosis is actually a disorder of neural complexity, i.e., a complicated disturbance to the balance between over-connectivity and disconnectivity in the brains of the poor patients. If there is ever linkage from genes to psychosis, it must first be made between synaptic plasticity regulation and the gene, and later on the synaptic disturbance when applied to neural networks spread in the brain will have the clinical explanatory power. Directly it will not work. I have devised a framework that could be plausible. (See my first comment to this discussion.)

View all comments by Avi PeledComment by:  Wayne Fenton
Submitted 14 May 2006
Posted 15 May 2006

I believe findings of shared genetic determinants across psychotic conditions are most consistent with a dimensional view of psychosis in which presence or absence of deficit psychopathology and the deficit syndrome will be a more salient categorical distinction than presence or absence of affective syndromes.

Population-based studies quantify the co-occurrence of schizophrenia and depressive syndromes. In the National Institute of Mental Health Epidemiological Catchment Area Study (ECA), 1.5 percent of the U.S. adult population met criteria for a lifetime diagnosis of schizophrenia (Robins and Regier, 1991). The odds ratio for comorbidity of unipolar affective disorder among patients with a lifetime diagnosis of schizophrenia was 14 (Judd, 1998), indicating a high probability that an episode of unipolar affective disorder will occur in the lifetime of an individual with schizophrenia. Using improved diagnostic methods, the National Comorbidity Study (NCS) yielded a prevalence of schizophrenia of 0.85 percent (66/8,000) (Kessler et al., 1994). Eighty-one percent of patients with schizophrenia were also diagnosed with a mood disorder some time in their lives; 59 percent of schizophrenic patients had a diagnosis of unipolar depression, and 22 percent had a diagnosis of bipolar disorder at some point in their lifetimes.

The deficit syndrome is present in approximately 10 percent of schizophrenia patients at illness onset, 25 percent after 5 years, and highly stable thereafter (Fenton and McGlashan, 1994). Characterized by profound amotivation, anhedonia, and enduring blunted affect (erosion of capacity for affect), it is associated with a very low risk of suicide and high risk of severe long-term disability Fenton et al., 1997). Carpenter and colleagues have described significant neurobiological and developmental features of this syndrome within schizophrenia (Kirkpatrick et al., 2001).

It is highly likely that non-deficit schizophrenia patients (e.g., approximately 80 percent) are those found by epidemiologists with a substantial lifetime risk of one or more affective episodes: When they present in an affective episode they may be diagnosed with schizoaffective disorder; the same patient later seen not in an affective episode may be diagnosed with schizophrenia.

Robin Murray's contention that additional neurodevelopmental insults differentiate schizophrenia from bipolar pathophysiology may be even more true in understanding differential risk for deficit versus non-deficit schizophrenia. The natural history of the deficit syndrome is characterized by greater neurodevelopmental features, including spontaneous movement disorders absent neuroleptic exposure. In addition, this phenotypic distinction is obviously an important candidate for whole genome microarrays and other efforts at molecular dissection of psychotic syndromes.


Fenton WS, McGlashan TH: Deficit syndrome of schizophrenia: Antecedents, symptom progression, and outcome. American Journal of Psychiatry, 151:(3) 351-356, 1994. Abstract

Fenton WS, McGlashan TH, Victor BJ, Blyler CB: Symptoms, Subtype and Suicidality in Patients with Schizophrenia Spectrum Disorders. Am J Psychiatry 154: 199-204, 1997. Abstract

Judd LL: Mood Disorders in Schizophrenia: Epidemiology and Comorbidity. J Clinical Psychiatry Monograph 16:1 2-4 February 1998.

Kessler RC, McGonagle KA, Zhao S, et al. 1994. Lifetime and 12 month Prevalence of DSM III-R Psychiatric Disorders in the United States: results from the National Comorbidity Survey. Arch Gen Psychiatry 51:8-19. Abstract

Kirkpatrick B, Buchanan RW, Ross DE, Carpenter WT. A separate disease within the syndrome of schizophrenia. Arch Gen Psychiatry. 2001 Feb ;58(2):165-71. Abstract

Robins LN, Regier DA eds. 1991. Psychiatric Disorders in America: The Epidemiological Catchment Area Study. New York, N.Y.: The Free Press (McMillan).

View all comments by Wayne FentonComment by:  Carol Tamminga, SRF Advisor
Submitted 15 May 2006
Posted 15 May 2006

The interesting position by Craddock and Owen is one that is supported not only by modern genetic data, but also by pharmacologic and phenomenologic findings. A dimensional approach to categorizing serious mental illnesses is an orientation that should now be further tested to examine to what extent, in epidemiological samples, our psychotic diagnoses have similar and different clinical, biological, and molecular characteristics. The outcome of how we "bin" our information in psychiatry is of great importance to the discovery of the molecular bases for these disorders, and subsequently, to the development of rational treatments.

It is the case that those aspects of illness, no matter what diagnosis, respond to antipsychotic drugs. We have drugs effective for combating psychosis, mood destabilization, depression, anxiety—all dimensions of illness, not diagnoses. It is also the case that discussions with affected persons revolve around what their illness manifestations are, and the diagnosis is not highly relevant to the treatment discussion.

View all comments by Carol TammingaComment by:  Herbert Meltzer (Disclosure)
Submitted 15 May 2006
Posted 15 May 2006

A Concept Foretold—Kraepelin vs. Greisinger

As a longtime enthusiast of the "Einheitspsychose" (Psychose unique, or unitary psychosis) of Wilhelm Greisinger (1817-1869), I am delighted to see the field coming back to this view. My early research in neuromuscular abnormalities, which most people know only part of (elevated creatine kinase activity during acute episodes) started out as a means of understanding the etiology of schizophrenia, but when I discovered that abnormalities in all aspects of the neuromuscular system (neurogenic atrophy, muscle development abnormalities, increased sprouting of subterminal motor nerves to compensate for loss of motor neurons, abnormal neuromuscular junctions, increased motor unit territories) were common to patients with schizophrenia, bipolar disorder, and psychotic major depression, as well as some of their first-degree relatives, I adopted the view that these findings were genetically based, were common to these disorders, and absent in patients with major depression without psychotic features. I did not study bipolar patients without psychotic features, who had never had a psychotic episode, but I would predict they would be more like those with major depression. These views were expounded in a number of papers over a decade of research, but are summarized in Meltzer (1975) and Meltzer and Crayton (1975) which were part of a symposium I organized for the Association of Nervous and Mental Disorders in 1974.

The entire symposium was focused on the very issue of this SRF Live Discussion—whether the biology of schizophrenia and bipolar disorder as we knew it at the time was consistent with Kraepelin or Greisinger. My view was that there were genes and pathophysiology that were common to these disorders, as well as some unique features which determined the phenotype. My opinion was that Greisinger had gotten it right in the first place, a view I still hold. Incidentally, I postulated that the abnormalities I and my colleagues found to be present in the motor neurons, subterminal motor nerves, and neuromuscular junctions of the schizophrenic and bipolar patients had to be the result of genes that expressed themselves in some brain neurons, not just motor neurons. Examples of what I had in mind were eventually discovered: neuregulin, dystrobrevin, and reelin. It is likely there are others.

Research into the neuromuscular system to find the expression of these and related genes in patients and their families in skeletal muscle and the nerves which innervate them is quite feasible and can lead to a test of the unitary hypothesis as well as a means to test the phenotype of these abnormalities. A final point is that research in the question of the overlap between schizophrenia and bipolar disorder should distinguish the nonpsychotic bipolars and their family members from those who have psychosis and that major depression with psychotic features should also be included in the research design.


Meltzer, H.Y.: Neuromuscular abnormalities in the major mental illnesses. I. Serum enzyme studies. In: Biology of the Major Psychoses, D.X. Freedman (Ed). Research Publications Association of the Research in Nervous Mental Disorders, Vol. 54, Raven Press, New York, pp 165 188, 1975.

Meltzer, H.Y. and Crayton, J.W.: Neuromuscular abnormalities in the major mental illnesses. II. Muscle fiber and subterminal motor nerve abnormalities. In: Biology of the Major Psychoses, D.X. Freedman (Ed). Research Publications Association of the Research in Nervous Mental Disorders, Vol. 54, Raven Press, New York, pp 189 207, 1975.

View all comments by Herbert MeltzerComment by:  Francis A O'Neill
Submitted 24 May 2006
Posted 24 May 2006

There is no doubt that the current classification system for psychosis is largely unsatisfactory, both from a clinical and a research perspective. However, it seems unlikely that at this stage a classification based on findings from genetic studies will be any better. Several genes affect several cortical and sub-cortical structures, interact and influence complex systems within the brain, to produce a series of symptoms that may fluctuate over time. A simple classification system would appear impossible. For research the future is surely to concentrate on symptom complexes and their neurobiological basis. For clinical purposes this may not have the immediate utility of a categorical approach and some updated version of the old system may be the best compromise. The demands of research and clinical work are different and any efficient system will recognize this.

View all comments by Francis A O'NeillComment by:  Karl-Ludvig Reichelt (Disclosure)
Submitted 29 May 2006
Posted 29 May 2006

There will never be an end to this discussion as long as diagnosis is based on symptoms and not etiology. In untreated schizophrenia, peptide increases and especially opioids of exorphine type have been found increased (1-5). In depression, as expected because of peptidase decreases (6-8), we find peptide increases and some overlapping in depression and schizophrenia (submitted). In spite of having been confirmed in five different countries, these data are largely ignored.

Because these disorders have a considerable hereditary component, there must be chemical changes. If not, they can hardly have any hereditary basis at all. A hypothesis that explains both etiology and symptoms has been forwarded (5,9) and offers an explanation of the findings to date clinically.


1. Hole K, Bergslien H, Jorgensen HA, Berge OG, Reichelt KL, Trygstad OE. A peptide-containing fraction in the urine of schizophrenic patients which stimulates opiate receptors and inhibits dopamine uptake. Neuroscience. 1979;4(12):1883-93. Abstract

2. Drysdale A, Deacon R, Lewis P, Olley J, Electricwala A, Sherwood R. A peptide-containing fraction of plasma from schizophrenic patients which binds to opiate receptors and induces hyper-reactivity in rats. Neuroscience. 1982 Jun;7(6):1567-73. Abstract

3. idet M et al. (1982) Acta Physiol Hung. 60: 121-127.

4. Cade R et al. (2000) Nutr Neurosci. 3: 57-72.

5. Lindstrom LH, Besev G, Gunne LM, Terenius L. CSF levels of receptor-active endorphins in schizophrenic patients: correlations with symptomatology and monoamine metabolites. Psychiatry Res. 1986 Oct;19(2):93-100. (Unfortunately, they used the term endorphin for opioids that may be exorphins.) Abstract

6. Maes M, De Meester I, Vanhoof G, Scharpe S, Bosmans E, Vandervorst C, Verkerk R, Minner B, Suy E, Raus J. Decreased serum dipeptidyl peptidase IV activity in major depression. Biol Psychiatry. 1991 Sep 15;30(6):577-86. Abstract

7. Maes M, Scharpe S, Meltzer HY, Suy E, Cosyns P, Calabrese J. Lower angiotensin I converting enzyme activity in melancholic subjects: a pilot study. Biol Psychiatry. 1992 Oct 1;32(7):621-4. No abstract available. Abstract

8. Maes M, Goossens F, Scharpe S, Meltzer HY, D'Hondt P, Cosyns P. Lower serum prolyl endopeptidase enzyme activity in major depression: further evidence that peptidases play a role in the pathophysiology of depression. Biol Psychiatry. 1994 Apr 15;35(8):545-52. Abstract

9. Reichelt KL, Seim AR, Reichelt WH. Could schizophrenia be reasonably explained by Dohan's hypothesis on genetic interaction with a dietary peptide overload? Prog Neuropsychopharmacol Biol Psychiatry. 1996 Oct;20(7):1083-114. Abstract

View all comments by Karl-Ludvig ReicheltComment by:  Kiumars Lalezarzadeh
Submitted 9 June 2006
Posted 18 June 2006

High prevalence of schizophrenia in cities with high latitudes needs attention when considering "co-morbid" mood disorders, fast or slow cycle bipolar disorder, and seasonal affective disorder (SAD), that is, in relation to hormonal cycles in males and females.


Saha S, Chant DC, Welham JL, McGrath JJ. The incidence and prevalence of schizophrenia varies with latitude. Acta Psychiatr Scand. 2006 Jul;114(1):36-9. Abstract

View all comments by Kiumars Lalezarzadeh