Abstract:Back to table of contents Previous article Next article PerspectivesFull AccessThe Long-Term Effect of Schizophrenia on the Brain: Dementia Praecox?John A. Sweeney, Ph.D.John A. SweeneySearch for more papers by this author, Ph.D.Published Online:1 Jun 2013https://doi.org/10.1176/appi.ajp.2013.13030413AboutSectionsPDF/EPUB ToolsAdd to favoritesDownload CitationsTrack Citations ShareShare onFacebookTwitterLinked InEmail Kraepelin’s use of the term “dementia praecox” for the condition we now know as schizophrenia encouraged the view that his patients had a progressive decline in behavioral function associated with (and probably caused by) progressive changes in brain anatomy and function over the course of illness. This view was challenged first by the failure of early postmortem histopathology studies to identify gross anatomic changes. Later, it was challenged by models proposing that neurodevelopmental alterations caused the disorder and that the primary disease-related damage to brain organization was largely in place at illness onset. Enthusiasm for “progressive” and “neurodevelopmental” models has varied over time and across countries (1), and establishing the validity of these models and their mechanisms remains one of the major challenges of schizophrenia research.Interest in demonstrating and establishing the mechanisms for progressive disease was an important reason for several schizophrenia programs to start longitudinal studies, including MRI protocols, of “first-episode” schizophrenia more than two decades ago. In brief, they hoped to identify progressive encephalopathy that might explain a decline in function. These were very challenging studies. The 202 patients reported by Andreasen et al. in this issue (2) were recruited over 18 years and underwent scanning on average three times. Additional challenges in these studies were maintaining patients’ study participation and keeping funding and investigative teams together over long follow-up periods. That 67% of the patients in this study remained involved over so long a period represents a remarkable achievement.The primary focus of the Andreasen et al. study is on clinical associations of atrophic anatomic changes with duration of persistent psychosis and intensity of antipsychotic treatment. Regional hypermetabolic activity during acute psychosis (3) and glutamate excitotoxicity (4, 5) represent potential mechanisms for a neurotoxicity of psychosis.Indirect evidence suggests that antipsychotic medications may have adverse effects on brain anatomy and function. Persistent adverse effects of antipsychotics on some cognitive processes supported by frontostriatal brain systems are well established in preclinical models (6). Studies using similar cognitive paradigms with schizophrenia patients show similar adverse effects (7). Reduced prefrontal connectivity in resting-state fMRI studies has also been demonstrated after initiation of antipsychotic treatment (8). Thus, evidence for potential adverse effects of antipsychotics outside the domain of motor systems has grown in recent years. The greatest reason for concern about adverse neuroanatomic effects of antipsychotic treatment stems from data showing a robust reduction in brain volume in monkeys treated with clinically relevant dosages of conventional and atypical antipsychotics (9). This remains a troubling observation on clinical grounds given the lack of therapeutic alternatives to antipsychotics.Andreasen et al. report that cortical volume loss occurs at a faster rate in schizophrenia patients than in matched healthy comparison subjects. This finding has been reported in this and other samples previously. Combined with multiple demonstrations that neuroanatomic deficits are present at illness onset, this suggests that both progressive and neurodevelopmental pathology may be important in the disorder. The apparent importance of both progressive and neurodevelopmental factors is not simplifying, but it gives us important insight into the complexity of the disorder.Andreasen et al. also report that relapses are more common but briefer near illness onset, and that over time a subgroup of patients experience prolonged relapses. Given the data, it seems likely that this subgroup tended to show the greatest reductions in brain volume over time. Relapse duration but not number of relapses was related to brain volume changes, with the effects most prominent in the frontal lobe and white matter. This observation is consistent with the view that patients with recurring psychosis may experience cumulative, persistent atrophic brain changes—and that there may be a subgroup of patients with poor treatment response and more persistent psychotic symptoms who experience progressive neuroanatomic changes.Antipsychotic treatment intensity was related to brain volume reductions in the frontal and temporal cortex and in parietal white matter. Antipsychotic-related effects on brain volume were notably smaller than those reported in rodent and nonhuman primate models; this difference may reflect species differences, an interaction with disease, more variable dosing clinically, or underreporting of treatment nonadherence. In any case, it is reassuring to see that progressive atrophic effects associated with antipsychotic treatment are less than those seen in animal models. Effects of relapse duration and antipsychotic treatment intensity on anatomic measures were of similar magnitude.Questions about medication and episode effects on brain anatomy were addressed using statistical approaches that controlled for some factors using regression approaches while testing for effects of factors of interest. This approach is relatively straightforward when factors are independent, but when they are correlated, as illness severity and drug dosage are, causal inferences are less certain. It is hard to fully confirm whether sicker patients with greater brain abnormalities receive more medication (and thus the causes of anatomic changes precede or even lead indirectly to higher drug dosages), or whether brain atrophy reflects a pharmacological effect of antipsychotic drugs. The authors recognize this issue, but it is an important caveat. Other statistical issues include potential nonlinearities in the relationships of interest, such that, for example, drug dosage may be associated with brain volume reductions primarily at higher dosages or with longer relapse durations. Another question is whether a subgroup of individuals, perhaps defined by genetic or epigenetic factors, are more likely to show atrophic changes in relation to relapse duration or drug treatment intensity.As with many advances, the findings of Andreasen et al. raise new questions. Do the volume reductions impair role function and cognition? Can preclinical studies clarify the mechanisms of antipsychotic effects on neuronal integrity? If antipsychotics are needed for symptom control and relapse prevention yet with increasing dosage there is a risk of atrophic brain changes, the need to optimize drug dosing would be increased, emphasizing lower effective dosages rather than raising dosage to the level just below a significant side effect threshold (10). This latter issue highlights the direct clinical importance of the findings from this study.This study of a class of drugs used over 50 years highlights the still limited understanding we have of the full range of beneficial and adverse effects of antipsychotic medications. We have learned about clinical efficacy and receptor binding profiles, but the complex and longer-term effects of antipsychotic drugs on functional brain systems and the systems-level neurobiology of psychosis remain far from clear. Armed with a fuller understanding, we could be better positioned to develop new treatments that maximize benefits and minimize adverse effects and be better able to talk to patients and families about long-term treatment options.From the Departments of Psychiatry and Pediatrics, University of Texas Southwestern Medical Center, Dallas.Address correspondence to Dr. Sweeney (john.[email protected]edu).Dr. Sweeney has served on advisory boards for Bristol-Myers Squibb, Eli Lilly, Pfizer, Roche, and Takeda and has received grant support from Janssen. Dr. Freedman has reviewed this editorial and found no evidence of influence from these relationships.References1 Weinberger DR, McClure RK: Neurotoxicity, neuroplasticity, and magnetic resonance imaging morphometry: what is happening in the schizophrenic brain? Arch Gen Psychiatry 2002; 59:553–558Crossref, Medline, Google Scholar2 Andreasen NC, Liu D, Ziebell S, Vora A, Ho BC: Relapse duration, treatment intensity, and brain tissue loss in schizophrenia: a prospective longitudinal MRI study. Am J Psychiatry 2013; 170:609–615Link, Google Scholar3 Bogerts B: The temporolimbic system theory of positive schizophrenic symptoms. Schizophr Bull 1997; 23:423–435Crossref, Medline, Google Scholar4 Coyle JT, Basu A, Benneyworth M, Balu D, Konopaske G: Glutamatergic synaptic dysregulation in schizophrenia: therapeutic implications. Handb Exp Pharmacol 2012; 213:267–295Crossref, Medline, Google Scholar5 Takahashi N, Sakurai T, Davis KL, Buxbaum JD: Linking oligodendrocyte and myelin dysfunction to neurocircuitry abnormalities in schizophrenia. Prog Neurobiol 2011; 93:13–24Crossref, Medline, Google Scholar6 Castner SA, Williams GV, Goldman-Rakic PS: Reversal of antipsychotic-induced working memory deficits by short-term dopamine D1 receptor stimulation. Science 2000; 287:2020–2022Crossref, Medline, Google Scholar7 Reilly JL, Harris MS, Keshavan MS, Sweeney JA: Adverse effects of risperidone on spatial working memory in first-episode schizophrenia. Arch Gen Psychiatry 2006; 63:1189–1197Crossref, Medline, Google Scholar8 Lui S, Li T, Deng W, Jiang L, Wu Q, Tang H, Yue Q, Huang X, Chan RC, Collier DA, Meda SA, Pearlson G, Mechelli A, Sweeney JA, Gong Q: Short-term effects of antipsychotic treatment on cerebral function in drug-naive first-episode schizophrenia revealed by “resting state” functional magnetic resonance imaging. Arch Gen Psychiatry 2010; 67:783–792Crossref, Medline, Google Scholar9 Dorph-Petersen KA, Pierri JN, Perel JM, Sun Z, Sampson AR, Lewis DA: The influence of chronic exposure to antipsychotic medications on brain size before and after tissue fixation: a comparison of haloperidol and olanzapine in macaque monkeys. Neuropsychopharmacology 2005; 30:1649–1661Crossref, Medline, Google Scholar10 Lewis DA: Antipsychotic medications and brain volume: do we have cause for concern? Arch Gen Psychiatry 2011; 68:126–127Crossref, Medline, Google Scholar FiguresReferencesCited byDetailsCited byTripping Over the Same StoneBiological Psychiatry, Vol. 88, No. 2 Volume 170Issue 6 June 2013Pages 571-573 Metrics PDF download History Received 1 March 2013 Published online 1 June 2013 Published in print 1 June 2013

Abstract:Multiple-treatments meta-analysis is thought to be a feasible method to compare the efficacy and safety among different treatments, especially when there was no head-to-head research among some treatments. But sometimes some conclusions are inconsistent with the clinical experience. Recently, we read a multiple-treatment meta-analysis finished by Stefen Leucht et al, which was published in Lancet.1 The authors summarized the results of the RCT studies on 15 antipsychotics commonly used in practice; they also horizontally compared the efficacy and safety profile by the meta-analysis. We believe that the results provide more solid evidence for the rational usage of antipsychotics to the psychiatrists, also for the government to distribute health resources in a more reasonable way. There were several published comments,2–6 which focused on the discussion the probability of potential bias of multiple-treatment meta-analysis in principle level, or the better interpretation of the results from this paper, nevertheless, we want to propose our additional concerns from clinical and clinical epidemiological perspective, and share our thinking how to avoid the bias. At first, as psychiatrists, we found that some results of this study can't match our clinical experience. For example, the Amisulpride showed a second best effecacy.1 But this drug is rarely used in China. The main reason is the well-known side-effect of prolactin increase,7 which had not been fully analyzed since there was no enough available data. Besides, we found the conclusion on Amisulpride was mainly based on old researches. Only 2 of the 15 researches were published in recent 5 years, while the number of researches published before 1999 is 7.1 Thus, based on this interesting point, after rounds of discussions with clinical epidemiologists, expert on evidence-based medicine and other psychiatrists, we found there were some issues in the design of the meta-analysis requiring further reflection, which may lead to bias and affect the results. These reflections are outlined below. We think that the methodology of meta-analysis is on a continuous way to advance, and we would like to take this opportunity to discuss them with the authors, editors, also with the other readers. The authors included the RCTs of all the commonly used antipsychotics in order to evaluate the efficacy and safety profiles. Although studies included are all RCTs, there are fundamental differences among them. As the author did mention in the appendix 2 that fixed-dose studies, they also did some stratified analysis and sensitivity analysis with consideration of this potential bias. We believe this research provides us a very good paradigm, however, there are still some more points requiring taking into account as below. Fixed-dose design and flexible-dose design There are two kinds of dosing designs, the fixed-dose design and the flexible-dose design. In studies with fixed-dose design, subjects get fixed doses randomly without clinicians’ intervention. While, in studies with flexible-dose design, the dosage of each subject is depended on clinicians’ decision. Studies comparing efficacy and safety between different fixed doses of a particular drug should be included with more cautious, since the intervention here is different from those studies which the flexible-dose design is adopted. A further stratified analysis and/or sensitivity analyses that exclude all fixed-dose studies should be necessary. Distribution of the dosage is different In the RCTs evaluating efficacy and safety between post-marketing medications, the subjects were given the flexible dose at the clinician's discretion according to the Evidence Based Practice principle.8,9 Hence the dosage would be nearly normal distribution. When the authors compared efficacy and safety profile between dosages, the subjects would be evenly assigned to different dose groups randomly,10,11 which would be showed as a uniform distribution. It is apparently not very reasonable to ignore the difference in dosage distribution and directly compare the two kinds of studies’ results. Flexible-dose strategy can balance the efficacy and safety better In studies adopting fixed-dose design, the subjects got one dose by randomization,12 but in studies adopting flexible dose design, the subjects got the dose by clinicians’ comprehensive consideration of evidence, clinical experience and individual needs.13,14 This different dosing strategy would affect the evaluation of drug efficacy and safety, and stratified analysis or sensitivity analysis may help to eliminate this bias. With these differences in mind, including the fixed-dose studies into this multiple-treatments meta-analysis would have complicated impact on the final result. As the authors showed in the appendix, for those fixed-dose studies only the ones matching the International Consensus Study of Antipsychotic Dosing15 were included. Moreover, the sensitivity analyses were made on those studies which were more efficacious or less well defined, or had some special design. However, we think these are not enough. In further multiple-treatments meta-analysis, more strict inclusion criteria should be made, and stratified analysis and/or sensitivity analysis that exclude all fixed-dose studies is necessary. Pre-authorization and post-marketing studies Pre-authorization and post-marketing studies should be analyzed separately even with the same or similar inclusion and exclusion criteria, since the representative of the subjects are not the same. In pre-authorization studies, subjects are treated with drugs not being approved. As a result, comparing with the post-marketing studies, both investigators and patients would be more cautious when making the decision to participant in such a study. This difference may cause selection bias and channeling bias,16–18 which is a form of allocation bias, where drugs with similar therapeutic indications are prescribed to groups of patients with prognostic differences. Patients may be more conservative to the new drugs which have similar mechanism to some post marketing ones, not to mention the new drug with different mechanism. So those who have relatively mild symptoms or simple conditions, or those who are from poor economic status are more likely to participate in. However, these considerations are not likely to be involved in the post marketing studies. In order to avoid the influence of the different drug development stages on the result of meta-analysis, we suggest performing subgroup analysis and sensitivity analysis, and data from pre-authorization studies and post marketing ones should be assessed separately and compared with combined analysis result. Also, discussions about study subjects’ representation should be included in this paper. Chronology bias due to different launch time19,20 Chronology bias occurs when comparison with historic group. The difference between latter and former time points on diagnose, treatments or preferred outcome measures may cause inequality within groups. New drugs attracted more attention Investigators would focus on newly launched drugs and pay less attention to “older drugs”. Most investigators believe that no specific safety report is needed because most doctors are very familiar with the safety profile of the older drugs, instead, pay more attention to newer ones, which leads to a publication bias that the newer drugs have worse safety profile than the older ones. Use of older drugs are more precise The longer a drug is on the market, the more clinical trials and experience it would have, and the more accurate and reliable the data and the evidence-based guidance are. As a result, over years after marketing, evidences and guidelines for treatments would change several times.21 Take clozapine for an example.22 At first, clozapine was developed to treat schizophrenia in general. During the post-marketing research, it was found to have serious adverse reactions. After that, the usage of clozapine is confined to refractory schizophrenia patients with aggressive behavior and suicidal tendencies. As a result, in recent years, the report of adverse reactions of clozapine remained at a low level. In this example we could see that, we would gain, gradually, the knowledge and experience of the efficacy and safety of a drug. The usage of the drug would be optimized with time, due to the experience learned from clinical practice and vast numbers of studies after being launched. Also, clinicians would be more familiar with the drug profile, so they are more capable to use the drug to treat patient in an appropriate way. This time advantage would lead to the potential benefit of older medications. We suggest that the time effect should be considered in the analysis. Meta-analysis would all be affected by this way if only being involved different medications or interventions, so that there should be analysis and discussion about time on market. Meanwhile, a new method of data combination should be established to assess the relationship between the efficacy result and the time on market. The efficacy and safety difference between newly-launched drugs and older ones would be more reliable if the timing factor is taken into account when the data are analysed. Generally, multiple-treatments meta-analysis brings us a feasible method to compare the efficacy and safety of the majority treatments for one disease or therapeutic areas. It also provides better evidence for making clinical decisions and, especially, policy decisions. However, multiple-treatments meta-analysis is such a new method and facing a huge number of factors in one analysis, so the result should be applied with more cautions. As the authors showed in the article and appendix, considerable sensitivity analyses were conducted, however, we think the sensitivity analysis is not enough even in this study, which has good design and comprehensive consideration. Besides, although the results of this kind of meta-analysis are based on majority high quality studies, the interpretation of the results should not be made apart from the experience of clinicians. Especially when the results are based only on indirect comparison, the clinician's experience is more important. Furthermore, we think clinician's experience may be helpful for the optimizing of methodology in multiple-treatments meta-analysis. (Received May 22, 2014) Edited by Cui Yi

Dr. Keefe and Colleagues Reply

Response to Sarpal Et Al.: Importance of Neuroimaging Biomarkers for Treatment Development and Clinical Practice.

Eff Ectiveness of Antipsychotic Drugs in Fi Rst-Episode Schizophrenia and Schizophreniform Disorder : an Open Randomised Clinical Trial

Short-term Effects of Antipsychotic Treatment on Cerebral Function in Drug-Naive First-Episode Schizophrenia Revealed by “resting State” Functional Magnetic Resonance Imaging

Antipsychotic Treatments: Who is Really Failing Here? – Authors' Reply

The Long-Term Effect of Schizophrenia on the Brain: Dementia Praecox?

Precise Treatments for Schizophrenia: Where is the Way Forward?

Antipsychotic Drugs and Cognitive Function: A Systematic Review and Pairwise Network Meta-Analysis

Efficiency of the CATIE and BACS Neuropsychological Batteries in Assessing Cognitive Effects of Antipsychotic Treatments in Schizophrenia

Neurocognitive Effects of Atypical Antipsychotics in Patients with First-Episode Schizophrenia.

Effects Of Olanzapine, Quetiapine, And Risperidone On Neurocognitive Function In Early Psychosis: A Randomized, Double-Blind 52-Week Comparison

Oculomotor and Neuropsychological Effects of Antipsychotic Treatment for Schizophrenia

[Cognition, schizophrenia and the effect of antipsychotics]

Dr. Carpenter and Associates Reply

Short Term Neurocognitive Effects of Treatment with Ziprasidone and Olanzapine in Recent Onset Schizophrenia

Neurocognitive Effects of Atypical and Conventional Antipsychotic Drugs in Early-Stage Schizophrenia: A Naturalistic 12-Month Follow-Up Study

Bias in Multiple-Treatments Meta-Analysis: A Case on Comparative Efficacy and Tolerability of 15 Antipsychotic Drugs for Schizophrenia

Cognitive function and treatment response trajectories in first-episode schizophrenia: evidence from a prospective cohort study

Differential Effects of Different Antipsychotic Drugs on Cognitive Function in Patients with Chronic Schizophrenia

Comparative Efficacy and Safety of Atypical and Conventional Antipsychotic Drugs in First-Episode Psychosis: a Randomized, Double-Blind Trial of Olanzapine Versus Haloperidol.

Antipsychotic treatment and cognitive function in patients with schizophrenia