Summations

• • Personalised medicine has finally arrived in psychiatric journals, but these journal articles have mainly focused on the promise of using disease mechanisms to personalise treatment.

Personalised medicine and drug mechanisms

The history of the concept of personalised medicine within pharmacology has been reviewed in a prior article; (Reference de Leon2) therefore, it is only briefly summarised and updated here. At the beginning of the 20th century, ‘chemical individuality’ in drug response was hypothesized (Reference Garrod3), but it was not demonstrated until the 1950s when pharmacologists observed that some severe adverse drug reactions (ADRs) only occurred in a small number of patients who were thought to have peculiar genetic profiles. This led to the term ‘pharmacogenetics’ (Reference Vogel4). The Second World War provided an impetus to drug development and the development of pharmacological laboratories for better understanding drug response (Reference Kanigel5). For the next 20–30 years, these laboratories made progress in describing the pharmacokinetic and pharmacodynamic mechanisms behind drug response (Reference Brodie and Cosmides6). In the 1980s–1990s, the various genes associated with the cytochrome P450 (CYP) isoenzymes were discovered, along with their genetic variations (Reference Meyer7). In the late 1990s, advances in genetics, which allowed parallel genetic testing (testing for multiple genes at the same time) by using the so-called DNA microarrays (Reference Fodor8), led to pharmacogenomics, the idea that many genes may influence drug response. The ability to test many genes led, in 1997, to Science defining ‘personalised prescription’ as ‘tailoring drugs to a patient’s genetic makeup’ and to the prediction that personalised prescription would ‘soon’ reach clinical practice (9). The media, captivated by the race for the human genome, popularised the concepts of personalised medicine and personalised prescription. After the race was over, it became clear that factors other than genetics may be very important in drug response. Therefore, in the view of the author (Reference de Leon10), personalised prescription can be expressed as personalising dosing and/or drug selection and should include genetic as well as environmental (e.g. co-medications, herbal medicines, smoking, foods, beverages) and personal (e.g. age, gender, ethnicity, illness) variables to improve the safety and efficacy of psychopharmacological drugs. Moreover, psychiatry is increasingly combining new pharmacogenomic findings with therapeutic drug monitoring (Reference Hiemke, Baumann and Bergemann11), particularly to guide personalised dosing (Reference Crettol, de Leon, Hiemke and Eap12). In the opinion of the author, the literature has not sufficiently recognised that each drug has its own pharmacokinetic and pharmacodynamic mechanisms; (Reference de Leon, Spina and Diaz13) therefore, ‘each drug is an individual’.

Personalised medicine and disease mechanisms

Further development of DNA microarray technology has allowed the testing of multiple other products besides DNA, including proteins, RNAs and lipids, and led to the development of new diagnostic branches such as ‘protenomics’, ‘transcriptomics’ and ‘metabonomics’. All of these new techniques have resulted in a new concept in drug development, namely, the biomarker. There are many definitions of biomarker; however, Wagner proposed ‘a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacological response(s) to a therapeutic intervention’ (Reference Wagner14). There are different types of biomarkers, including the pharmacogenomic biomarkers described above as the pharmacokinetic and pharmacodynamic mechanisms behind drug response.

There has been a research boom in biomarkers in medicine both for practical and theoretical reasons. From the practical point of view, biomarkers have been particularly successful in oncology (Reference Dienstmann, Rodon and Tabernero15) for two reasons: (1) the cancerous tissue is available for study of the pathogenetic mechanisms, and (2) our understanding of these pathogenetic mechanisms is sufficiently developed to allow us to start selecting drugs according to the mechanisms in each individual patient. From the theoretical point of view, there has been a revolution in bioinformatics, including the development of a new scientific concept called ‘complexity’ (Reference Waldrop16), reflecting the complexity of scientific exploration when there are multiple influences causing a phenomenon and the need exists for complex computer models to integrate all that is known about it. Complexity theory and its accompanying computer models are being introduced in medicine by using the so-called ‘network’ medicine (Reference Barabási, Gulbahce and Loscalzo17). Therefore, personalised medicine using disease mechanisms is becoming a fad in medicine.

If one is a psychiatrist who wants to think that psychiatry is not lagging behind other medical disciplines, one should endorse the idea of using disease mechanisms to personalise psychiatric treatment. As a matter of fact, many of the recent articles in psychiatric journals focus on the future of psychiatry propose using disease mechanism to personalise psychiatric treatment, but these articles usually ignore the less glamorous drug response mechanisms proposed by pharmacologists. There are many of these articles in many different journals, but the author can point out a few of them for the reader (Reference McGorry18–Reference Ozomaro, Wahlestedt and Nemeroff22).

This author proposes that the path to personalised psychiatry, which uses disease mechanisms while forgetting drug response mechanisms, is the wrong road. He acknowledges that our knowledge of the use of pharmacokinetic and pharmacodynamic mechanisms for personalising psychiatric treatment is pretty limited as it applies to a few drugs, and more fundamentally to pharmacokinetic mechanisms. Our understanding of pharmacodynamic mechanisms of psychiatric drugs is too limited to have much clinical relevance (Reference Crettol, de Leon, Hiemke and Eap12,Reference de Leon, Armstrong and Cozza23,Reference Arranz and de Leon24). Personalised medicine based on drug mechanisms is a ‘humble’ pathway not likely to lead to any major breakthroughs but could promote the use of currently available drugs in a better way.

It is not surprising that, if one wants to compete with cancer researchers in marketing and funding, one should propose that psychiatry can cure mental illnesses (Reference Insel and Scolnick25) by personalising treatment (Reference Kapur, Phillips and Insel26), as oncology appears to be on the verge of curing many types of cancers. On the contrary, the author proposes that at the current time, despite the wide endorsement of psychiatric researchers of the disease mechanism to personalise psychiatry, it may be premature and, furthermore, a waste of time and money. This is based on two ideas: (1) psychiatry is lagging behind medicine and many of the psychiatric diseases, such as schizophrenia, are more syndromes than diseases; and (2) even if one focuses on diseases with known brain neuropathology, such as Alzheimer disease, the complexity of brain mechanisms currently appears insurmountable.

The problem with the concept of disease in psychiatry

In a JAMA article in 2005, McHugh (Reference McHugh27) pointed out that psychiatry is 150 years behind medicine, and therefore we are trying to do what 19th century physicians did. At that time, there was a revolution characterised by the convergence of three ways of thinking, which have been called anatomoclinical (relating signs and symptoms with disturbances in specific organs), physiopathological (relating diseases or their symptoms/signs with disturbances in normal physiology) and etiopathological (finding specific causes of some diseases and their lesions) (Reference Lain Entralgo28). This led to the revolutionary success of 20th century medicine.

The problems with the scientific approach in psychiatry have been discussed previously by the author; (Reference de Leon29) however, anyone with historical knowledge of psychiatric disease knows that diseases such as schizophrenia (Reference de Leon30) are at best syndromes and not medical diseases in the traditional sense. Cancers, such as colon cancer, are being divided into different diseases based on pathogenic mechanism. Unfortunately, psychiatry is far from reaching that point. Moreover, we have no way of validating the diagnosis of most psychiatric disorders and establishing their boundaries with other psychiatric disorders.

Problems dealing with brain complexity: Alzheimer disease

There is no doubt that originally Alzheimer disease was a psychiatric disease. In 1907, Alzheimer, a psychiatrist, described the neuropathology in one case of presenile dementia with psychosis, and in 1910, Kraepelin, a psychiatrist, (in the 8th edition of his textbook) baptised it as a new illness (Reference Berrios31). After including ‘senile’ dementia cases within the boundaries of Alzheimer disease, a major development in neuroscience research occurred in 1990; it included genetic findings such as the association of Apolipoprotein E-4 with the age of onset of common late-onset Alzheimer disease.

It is now doubtful that Alzheimer disease can be called a psychiatric disease, as most of the new findings come from neurologists and other neuroscientists. To demonstrate, the author completed a PubMed search on 1 March 2014 comparing articles from the last 10 years of two pairs of similar journals in neurology and in psychiatry. Neurology was compared with the American Journal of Psychiatry and JAMA Neurology (previously Archives of Neurology) with JAMA Psychiatry (previously Archives of General Psychiatry). Neurology published 912 articles on Alzheimer disease (or 8.2%, 912/11 156) versus 54 (or 1.7%, 54/3428) in the American Journal of Psychiatry. JAMA Neurology published 501 articles (or 16.8%, 501/2985) versus 48 (or 3.0%, 48/1560) in JAMA Psychiatry. In both absolute and relative terms, Alzheimer disease is a major topic in neurological journals and a rather minor topic in psychiatric journals.

Let us stretch reality and consider Alzheimer disease, a psychiatric disease with known neuropathology and clearly established boundaries, and a good example of a psychiatric disease that follows the medical model, which is the ideal that Kraepelin proposed for psychiatric diseases 100 years ago (Reference de Leon32). The problem is that 20 years of intensive research using the disease mechanism has led to many strategies targeting different mechanisms, but unfortunately the outcome has been disappointing (Reference Morley33,Reference Crenshaw, Gottschalk and Lutz34). Thus, the attempts of neuroscience to use disease mechanisms to develop new treatments for Alzheimer are not working well; a personalised treatment for Alzheimer disease would appear to be a fantasy owing to the complexity of brain mechanisms. Therefore, even if we assume that Alzheimer disease is a psychiatric disease, the lack of success of the mechanistic disease approach to it is not good news for psychiatry. Alzheimer disease has 100 years of replicated neuropathological findings and 20 years of mechanistic studies based on reasonable understanding of its genetics. If one compares that with schizophrenia, for which Alzheimer completely failed as a neuropathologist, we find 100 years of arguments about schizophrenia’s boundaries (Reference de Leon30) and 60 years of treatment with drugs discovered by serendipity, which are considered not to be specific treatment for schizophrenia but are for all kinds of psychoses. In summary, even if major breakthroughs and ways of validating schizophrenia diagnosis occur in coming decades, our experience with Alzheimer disease tells us that our limited understanding of brain mechanisms is a major problem when compared with the highly developed mechanistic discipline of oncology.

A forgotten way of personalising treatment in psychiatry using clinical characteristics

The author has proposed that consideration of disease mechanisms may not currently be a viable way of personalising psychiatric treatment. However, past attempts to personalise psychiatric treatments based on the clinical profile of the patient remain ignored by current psychiatric researchers.

The idea behind these ignored approaches is that diseases, such as schizophrenia and depression, are not diseases but syndromes that can be carved out by sophisticated use of clinical symptoms in more specific diseases that can be better related to treatment response. These approaches were heavily invested in descriptive psychopathology, which was supposed to be the ‘language of psychiatry’; (Reference Ban35) however, interest in descriptive psychopathology has been decimated by the changes in US psychiatry since the DSM-III (Reference de Leon29,Reference Andreasen36).

Ban has been one of the main traditional psychopharmacologists defending the concept that a disease’s clinical profile can be used to group patients according to response (Reference Ban37,Reference Ban38). He has focused on depression and schizophrenia, but for space reasons this editorial will focus only on schizophrenia. The current concept of schizophrenia presented in US nosological systems as the DSM-III combines the ideas of Kraepelin, Bleuler and Schneider; (Reference de Leon30,Reference Hoenig39) however, there is another model of schizophrenia that is, for the most part, ignored in US psychiatric textbooks and journals. As far as the author can tell, this alternative approach to diagnosing schizophrenia has rarely been described in US journals; two were selected as representative: Journal of Clinical Psychopharmacology and American Journal of Psychiatry. On 27 March 2014, a PubMed search produced 4624 articles from the Journal of Clinical Psychopharmacology, of which 873 were on schizophrenia but none of them used this classification (adding the word ‘Leonhard’ to the prior schizophrenia search provided no articles). On 27 March 2014, a PubMed search produced 23 589 articles from the American Journal of Psychiatry, of which 3731 were on schizophrenia, but no US articles used this classification. Only two articles (one letter to the editor and one twin genetic study) by the same German research group were identified.

Wernicke, one of the Kraepelin’s main competitors, developed an alternative way of classifying psychosis. This approach to psychosis is sometimes labelled the approach of the Berlin school of psychiatry, as it was developed in Berlin. Wernicke trained Kleist, who trained Leonhard, who presented the final version of this alternative model of classifying psychosis. The Leonhard’s psychosis model has been totally eclipsed by the Kraepelinian model. Wernicke died at a relatively early age in an accident (Reference de Leon32). Most importantly, Leonhard worked at in East Germany which was controlled by the Soviets and his textbook was not translated into English until recently (Reference Leonhard40). Leonhard was a very sophisticated clinician who did family and twin studies in hundreds of patients. His two major accomplishments include: (1) his proposal for the first time that familial history separates mono and bipolar depression; and (2) his theory that a group of psychoses lie between schizophrenia and manic-depressive illness; he called them cycloid psychoses (Reference Leonhard40). Leonhard proposed that schizophrenia is a syndrome including several illnesses (Reference Leonhard40), some of which are genetic and others of which are not. Systematic schizophrenia consists of non-genetic illnesses including paraphrenias, hebephrenias and catatonias. Unsystematic schizophrenias are genetic illnesses including cataphasia, affect-laden paraphrenia and periodic catatonia. Astrup (Reference Astrup41), a Norwegian psychiatrist, conducted an important long-term outcome study on schizophrenia patients. He initially studied 189 patients by applying Leonhard’s classification. Fish (Reference Fish42) reclassified, with Astrup, another 285 chronic schizophrenia patients. Finally, Fish included all 474 patients with schizophrenia in an article, which reported that <1/4 of systematic schizophrenia patients versus >4/5 of unsystematic schizophrenia patients responded to first-generation antipsychotics (Reference Fish42). Later on, Guy et al. (Reference Guy, Ban and Wilson43) in an international survey of 768 patients with chronic schizophrenia, found that tardive dyskinesia was overrepresented in patients with systematic versus those with unsystematic schizophrenia (13.3% vs. 4.3%). In a small group of 50 patients with chronic schizophrenia in Germany, Beckmann et al. (Reference Beckmann, Fritze and Franzek44) proposed that the prognosis did not appear to change with antipsychotics when compared with Leonhard’s observations. All of these outcome studies were conducted by sophisticated clinicians who did not use blinding or placebo in their studies but suggested that systematic schizophrenia does not respond well at least to first-generation antipsychotics. There is little reason to think that these clinical researchers were biased towards finding greater response to antipsychotics in unsystematic schizophrenia, since Leonhard had developed his classification in the pre-neuroleptic era.

The author, along with Ban (Reference Ban37,Reference Ban38), believes that schizophrenia and major depressive disorders are syndromes rather than illnesses; however, all of these pre-DSM-III attempts to divide them into more homogenous subgroups that may predict treatment response remain forgotten. They are not considered because they were conducted by sophisticated clinicians who did not use well-controlled designs with blinding and placebo. Owing to these methodological weaknesses, these studies remain forgotten by current psychopharmacologists who only value clinical trials using well-controlled designs. Contemporary psychopharmacologists may have forgotten that psychopharmacological drugs were discovered by sophisticated clinicians without using well-controlled designs. The use of well-controlled designs has not brought any revolutionary changes in psychopharmacology, just some second-generation drugs with possibly some better ADR profiles but of doubtful greater efficacy. The author thinks that it will be interesting to incorporate some of these attempts to subdivide schizophrenia and major depression in future well-controlled pragmatic studies on psychotropic drugs. Unfortunately, this will require intensive clinical training of psychiatrists involved in the diagnosis and assessment of patients. When psychiatrists are heavily invested in using biological mechanisms to personalise psychiatric treatments, it may be too late to try to go back to forgotten attempts to personalise treatment according to sophisticated clinical assessments.