Thesis May Be Reproduced, Distributed, Stored in a Retrieval System, Or Transmitted in Any Form Or by Any Means, Without Prior Written Permission of the Author

UvA-DARE (Digital Academic Repository)

Drug treatment for patients with acute mania: Understanding clinical trials and treatment success

Welten, C.C.M.

Publication date 2016 Document Version Final published version

Link to publication

Citation for published version (APA): Welten, C. C. M. (2016). Drug treatment for patients with acute mania: Understanding clinical trials and treatment success.

General rights It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.

UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)

Download date:05 Oct 2021

DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA

Understanding clinical trials and treatment success

Carlijn C.M. Welten Cover and layout design by The Fat Moose, www.thefatmoose.nl Printing by Uitgeverij BOXPress | Proefschriftmaken.nl

ISBN: 978-94-6295-297-3 Author: Carlijn Welten

- Understanding clinical trials and treatment success -

ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad van doctor

aan de Universiteit van Amsterdam

op gezag van de Rector Magniﬁcus

prof. dr. D.C. van den Boom ten overstaan van een door het College voor Promoties ingestelde commissie,

in het openbaar te verdedigen in de Agnietenkapel

op 15 januari 2016, te 10.00 uur

door Carolina Cecilia Maria Welten

geboren te Utrecht Promotores prof. dr. D.A.J.P. Denys Universiteit van Amsterdam prof. dr. H.G.M. Leufkens Universiteit Utrecht

Copromotores dr. M.W.J. Koeter Universiteit van Amsterdam dr. T.D. Wohlfarth College ter Beoordeling van Geneesmiddelen

Overige leden prof. dr. L. de Haan Universiteit van Amsterdam prof. dr. J.A. Swinkels Universiteit van Amsterdam prof. dr. E.M. Derks Universiteit van Amsterdam dr. N.C.C. Vulink Academisch Medisch Centrum prof. dr. A.W. Hoes Universiteit Utrecht prof. dr. R.W. Kupka Vrije Universiteit Amsterdam prof. dr. D. Deforce Universiteit Gent

Faculteit der Geneeskunde

“No amount of love can cure madness or unblacken one’s dark moods. Love can help, it can make the pain more tolerable, but, always, one is beholden to medication that may or may not always work and may or may not be bearable”

- Kay Redﬁeld Jamison, An Unquiet Mind: A Memoir of Moods and Madness, 1995 -

CONTENTS

Part I General introduction

Chapter 1: General introduction 13

Part II Regulatory questions

Chapter 2: Efﬁcacy of drug treatment for acute mania differs 27 across geographic regions

Chapter 3: Placebo response in antipsychotic trials of patients 45 with acute mania

Chapter 4: Net gain analysis, an addition to responder analysis 63

Part III Clinical questions

Chapter 5: Does insight affect the efﬁcacy of antipsychotics in 83 acute mania?

Chapter 6: Early non-response in the antipsychotic treatment of 99 acute mania; a criterion for reconsidering treatment?

Part IV Summary and discussion

Chapter 7: Summary and discussion 121

Part V Appendix

Nederlandse samenvatting 145

Dankwoord 155

Portfolio 163

About the author 169

Contents | 9

General1 introduction PART

01 CHAPTER 1: GENERAL INTRODUCTION

INTRODUCTION

This thesis is about the treatment of patients with an acute manic episode. An acute manic episode is part of the Bipolar I Disorder, characterized by episodes of acute mania, episodes of depression and periods of remission. Only one acute manic episode in a lifetime is needed to meet criteria for Bipolar I Disorder (1). In this thesis, I focus on the treatment of the acute manic episode and leave the depressive episode of bipolar disorder aside.

An acute manic episode is a dramatic and potentially very harmful period for patients and their loved ones. Patients can feel as if they are on top of the world, can be easily irritated, and may have feelings of grandiosity. Frequently, they have little need for sleep and therefore only sleep few hours a night. They often speak rapidly and continuously, and experience many thoughts running through their minds. They are easily distracted and can be extremely busy with social, work or school related or sexual activities. Unfortunately, they are often excessively preoccupied with activities that have high potential for serious harm, including interpersonal problems, extensive spending leading to serious debts and promiscuous behavior (1). An acute manic episode can last several weeks. When patients recover from an acute manic episode, they are frequently confronted with a personal, interpersonal, occupational and/or ﬁnancial disaster due to the acute manic episode.

Personal experiences as a resident at the psychiatric crisis department of Amsterdam (Arkin, Spoedeisende Psychiatrie) revealed the (self-) destructive behavior of several patients with an acute manic episode. There was a patient who sold his holiday home in Spain for only very little money, opened the door for strangers and gave away the entire furniture; a patient who walked naked on the streets of Amsterdam and had numerous sexual contacts; and a 76-years old patient who drunk day and night while dancing in her house in little and extravagant clothes. It was harrowing to see these patients in an acute manic episode, without boundaries and mostly without any sense of shame, a true motivation to study the treatment of an acute manic episode of bipolar disorder.

Bipolar disorder is a lifelong condition, which unfortunately cannot be cured. Thus, one of the goals of treatment of patients with bipolar disorder is to prevent a new acute

PART 1 - CHAPTER 1 - General introduction| 15 manic or depressive episode. Therefore, patients with bipolar disorder in remission are often treated with the mood stabilizer lithium, and less frequently with (anticonvulsant) mood stabilizers, antidepressants or antipsychotics, to prevent recurrence to an acute manic or depressive episode (2). However, due to serious life events, intense stress, inadequate medication, or discontinuation of medication (frequently occurring after a long episode-free period), a recurrence to mania or depression may still occur.

When a recurrence to an acute manic episode occurs, the immediate treatment goal is to rapidly resolve manic symptoms. As recommended by the NICE Guideline for Bipolar Disorders (2014) and the Dutch Guidline for Bipolar Disorders (2015), the ﬁrst choice in the treatment of acute mania is antipsychotic mono-therapy. If this is not effective, it is advised to switch to another antipsychotic compound and ﬁnally, adjuvant medication (e.g. lithium or valproate) is recommended (2).

Preferably, only those antipsychotics should be prescribed for patients with an acute manic episode that are registered for this specific indication. In order to be registered (and thus obtain market authorization), the drug is tested on efficacy, safety, and quality in phase III randomized controlled clinical trials (RCTs). The drug has to show a significant and clinical relevant effect compared to placebo or to the established treatment and adverse events are assessed. If the registration authority (e.g. EMA, FDA) decides that there is a positive benefit-risk balance for the drug, registration and market authorization can be granted. This assessment can be challenging and the outcomes are often debated; were the patients in the trial really representative for those seen in daily clinical practice, were the outcome measures really the most adequate, was the magnitude of the effect really clinically relevant, and did the benefits really outweigh the risks, etcetera.

To better understand and further improve clinical trials for the registration of treatment for acute mania, the Medicines Evaluation Board (MEB) of the Netherlands embarked on a collaborative project with the the Academic Medical Centre (AMC) at the University of Amsterdam. The MEB is grateful to the pharmaceutical companies for allowing us excess to their database. This database consisted of the raw individual patient data (IPD) of twelve registration studies for the indication of acute mania in an eleven-years period. All studies were double blind, randomized, placebo-controlled trials including

16 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success patients diagnosed with a DSM-IV acute manic episode of bipolar disorder.

In this database, 3,207 patients were included; 1,403 patients received active treatment, 631 patients received an active comparator, and 1,191 patients received placebo. The mean age was 39.21, the mean BMI was 26.34, and 46% of patients was female. In our study, 59.2% of patients was Caucasian, 12.2% African American, 13.2% Asian, and 15.4% had another ethnic background. The studies assessed severity of the acute manic episode with two eleven-item interview-based questionnaires; the Young Mania Rating Scale (YMRS) (scored 0 to 60) and the Mania Rating Scale from the Schedule for Affective Disorders and Schizophrenia – Change Version (MRS from SADS-C) (scored 0-52) (3, 4). The YMRS questionnaire was used in ten studies (N=2764) and the MRS questionnaire in 2 studies (N=443). The mean baseline score was 30.20 on the YMRS and 27.75 on the MRS questionnaire.

The individual patient data in my database offered me a unique opportunity to look at some unresolved issues important for a better understanding and valid interpretation of the results of randomized controlled trials (RCTs) for drug registration in acute mania. An individual patient data (IPD) meta-analysis of registration studies, regardless acceptance for registration, consisting of different compounds, conducted in different geographic regions all over the world and in 3,207 patients with acute mania, enabled us to powerfully study RCT ﬁndings that are beyond single trials, beyond single compounds, beyond single geographic regions and beyond the relatively small sample size of individual studies.

Since regulatory questions are closely connected to clinical questions, the IPD in my database offered me the opportunity to study clinical questions in order to improve the success rate of antipsychotic treatment in patients with acute mania in clinical practice. A successful antipsychotic treatment, with success deﬁned as at least 50% reduction in manic symptoms, only occurs in approximately half of the patients (5) and there is thus ample room for improvement. This improvement can take place through a better understanding and improvement of the processes and predictors involved in a successful treatment of patients with acute mania.

PART 1 - CHAPTER 1 - General introduction| 17 OUTLINE OF THIS THESIS

This thesis has four parts. Part I is the general introduction and presents the aims and the outline of the thesis. Part II presents our ﬁndings with regard to the regulatory questions of this thesis. Part III presents the answers to the clinical questions. In part IV, the main ﬁndings of this thesis are summarized, discussed, and regulatory and clinical implications are given.

Regulatory questions The 20th and the beginning of the 21st century are characterized by globalization of many aspects of life. Following the economic, political, and social world, nowadays clinical trials are also carried out globally. The globalization of clinical trials means that regulatory bodies, such as the European Medicines Agency (EMA) and the USA Food and Drug Administration (FDA), now have to evaluate the findings and the relevance of studies from different geographic regions involving populations that may differ from those in their own region. (6, 7) The question arises whether the results of studies from one geographic region can be extrapolated to another region, given global differences in patient characteristics, variations in health care systems, different environments and cultures. (6, 8) This is especially important in the field of psychiatry where environmental and cultural factors are of great influence. Therefore, in chapter 2 I investigate (1) whether there are differences across geographic regions (USA, Europe, and Other regions) in the efficacy of medications for the treatment of an acute manic episode of bipolar disorder, and (2) whether possible regional differences in effect can be explained by regional differences in baseline characteristics or placebo response. To answer these questions I perform an individual patient data (IPD) meta-analysis of 3,207 patients (twelve registration studies) and stratified patients to three regions; Europe, the USA and the Other region. I use mean change score (by Hedges’ g) and response rate (by odds ratio (OR)) as outcome measures for efficacy.

Apart from regional differences, high inter-study variance and interpersonal-variability, phase II and III clinical trials in psychiatry also suffer from a high failure rate. (5, 9, 10) Efﬁcacy trials of psychiatric medications have a higher failure rate than trials of medicines for other medical disorders (9, 11-14). These failures mostly become manifest in phase III clinical trials, where drugs are tested for efﬁcacy and safety in the target

18 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success group in doubled blinded randomised placebo-controlled studies (11). Unfortunately, this diminishes the likelihood of new medicines becoming available for psychiatric disorders. One of the most frequently cited reasons for this high failure rate is the high placebo response in trials of psychiatric medications (5, 9, 12-16). This makes it important to understand the role of placebo response in psychiatric treatment. Therefore, in chapter 3 I examine (1) whether the magnitude of the placebo response predicts treatment effect (2), which patient and study characteristics predict placebo response, and (3) what is the most parsimonious model to predict placebo response. To answer these questions I use the IPD of 1,019 patients receiving placebo (ten registration studies). I again use mean change score (Hedges’ g) and response rate (OR) as outcome measures for efﬁcacy and performed a series of multilevel mixed effect linear (and logistic) regression analyses to examine the predictive value of study and patient characteristics. Finally, I conduct a series of step-wise multi-level (logistic) regression analyses (backward elimination) to determine the most parsimonious model to predict placebo response.

The high inter-study variance and the interpersonal-variability (5) could also be caused by methodological aspects of the studies. In short-term efficacy studies of patients with an acute manic episode, two outcome measures are most frequently used; the mean change score and the response rate. The mean change is used to assess significant improvement and the response rate to assess clinical relevance, with response defined as at least 50% improvement compared to baseline and a minimal difference in the percentage of responders between the active compound and the placebo or active comparator (17). One may question, however, whether the difference in the percentage responders between the treatment conditions is the best operationalization of clinical relevance. Responder analysis only takes into account the differential probability of a success and does not take into account the differential probability of deterioration. This is of particular importance for diseases like an acute manic episode where the probability of becoming a responder is limited (approximately 50%) and deterioration in the acute phase may cause substantial mental, physical and social harm (5). Therefore, in chapter 4, I propose a new approach called ‘net gain analysis’ with net gain defined as the difference in the percentage responders between the treatment and placebo group minus the difference in the percentage deteriorators between these groups, and compare its results with those of the traditional responder analysis for different antipsychotics in the treatment of patients with acute mania. For this analysis, I use the

PART 1 - CHAPTER 1 - General introduction| 19 IPD of 2,666 patients (ten registration studies) and rank the efﬁcacy of the different antipsychotic compounds by the net gain analysis and by the traditional responder analysis. I use percentage responders, percentage deteriorators, and percentage net gain to determine these rankings.

Clinical questions When an antipsychotic compound finally enters the market, we know that the drug has proven to be relatively safe and effective, and that the benefit/risk balance is considered positive. The clinician can than prescribe the drug to the patient suffering from the indicated disorder. For patients with bipolar mania, it is known that the majority has no or a very limited (impaired) insight in his or her illness (5). Given the fact that an acute manic episode is treated with medication, it is important to know if insight in one’s illness influences the efficacy of treatment. For example, if insight would influence efficacy negatively, the clinician might consider pre-treatment psycho-education or adjuvant therapy (pharmacological and/or psychological) to intensify treatment. Therefore, in chapter 5, I examine whether insight modifies efficacy of antipsychotic treatment in patients with acute mania. To answer this question, I use the IPD of 1,904 patients (seven registration studies) with antipsychotic treatment and data on item 11 (‘Insight’) of the YMRS questionnaire. We used mean change score (β-coefficient), response rate (OR) and remission rate (OR) as outcome measures for efficacy. We performed multilevel mixed effect linear (or logistic) regression analyses to assess the modifying effect of insight on efficacy of antipsychotic treatment.

When the clinician considers the patient suitable for antipsychotic mono-therapy, treatment is started. As described above, an acute manic episode is known to cause serious emotional turmoil for patients and their environment, and therefore the aim of the acute treatment is to rapidly reduce symptom severity in an attempt to counteract the risk on serious harm. Unfortunately, the success rate of pharmacological treatment is less than 60%, leaving more than 40% of the patients to be unsuccessfully treated (5). In order to intervene in an existing treatment of patients who will otherwise have an unsuccessful outcome at endpoint, it is important to know whether early non-response predicts future antipsychotic treatment failure. Therefore, in chapter 6, I examine whether (1) early non-response predicts antipsychotic non-response and non-remission at study endpoint and (2) determined the best combination of cut-off score for deﬁning

20 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success non-response and the time point to predict the lack of effect of treatment. To answer these questions, I use the IPD of patients with data on week one, two, and follow-up (N=1,243, ten registration studies). I calculate the predictive values of non-response at study endpoint (PVnr_se) for different cut-off scores at week one and week two to determine the most adequate criterion for treatment reconsideration.

PART 1 - CHAPTER 1 - General introduction| 21 REFERENCES

1. American Psychiatric Association A: Handboek voor de classiﬁcatie van psychische stoornissen (DSM- 5). Nederlandse vertaling van Diagnostic and Statistical Manual of Mental Disorders. . Fifth Edition ed. Amsterdam, Boom; 2014. 2. NICE: Bipolar disorder: the assessment and management of bipolar disorder in adults, children and young people in primary and secondary care in NICE clinical guideline 185, National Collaborating Centre for Mental Health; 2014. 3. Young RC, Biggs JT, Ziegler VE, Meyer DA. A rating scale for mania: reliability, validity and sensitivity. British Journal of Psychiatry 1978;133:429-435. 4. Spitzer RL, Endicott J. Schedule for Affective Disorders and Schizophrenia - Change Version (3rd ed.). New York: Biometrics Research. 1987. 5. Yildiz A, Vieta E, Tohen M, Baldessarini RJ. Factors modifying drug and placebo responses in randomized trials for bipolar mania. International Journal of Neuropsychopharmacology. 2011;14:863- 875. 6. CHMP: Reﬂection paper on the extrapolation of results from clinical studies conducted outside the EU to the EU-population. Committee for Medicinal Products for Human Use 2008. 7. FDA: Guidance for Industry and FDA Staff - FDA Acceptance of Foreign Clinical Studies Not Conducted Under an IND frequently Asked Questions. Edited by Services USDoHaH2012. 8. ICH: E5(R1): Ethnic Factors in the Acceptability of Foreign Clinical Data 1998. 9. Kemp AS, Schooler NR, Kalali AH, al. e. What is causing the reduced drug–placebo difference in recent schizophrenia clinical trials and what can be done about it? Schizophrenia Bulletin. 2010;36:504- 509. 10. Hurko O, Ryan JL. Translational research in central nervous system drug discovery. Neurotherapeutics. 2005;2:671-682. 11. Nutt D, Goodwin G. ECNP Summit on the future of CNS drug research in Europe 2011: report prepared for ECNP by David Nutt and Guy Goodwin. European Neuropsychopharmacology 2011;21:495-499. 12. Walsh BT, Seidman SN, Sysko R, Gould M. Placebo response in studies of major depression: variable, substantial, and growing. . JAMA. 2002;287:1840-1847. 13. Vieta E, Cruz N. Increasing rates of placebo response over time in mania studies. Journal of Clinical Psychiatry. 2008;69:681-682. 14. Khan A, Detke M, Khan SR, Mallinckrodt C. Placebo response and anti-depressant clinical trial outcome. Journal of Nervous and Mental Disease. 2003;191:211-218. 15. Agid O, Siu CO, Potkin SG, Kapur S, Watsky EJ, Vanderburg D, Zipursky RB, Remington G. Meta- Regression Analysis of Placebo Response in Antipsychotic Trials, 1970–2010. American Journal of Psychiatry. 2013;170:1335-1344. 16. Kinon BJ, Potts AJ, Watson SB. Placebo response in clinical trials with schizophrenia patients. Current Opinion in Psychiatry. 2011;24:107-113. 17. CHMP: Note for guidance on clinical investigation of medicinal product of the treatment and prevention of bipolar disorder. . Edited by Use CfMPfH2001.

22 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success PART 1 - CHAPTER 1 - General introduction| 23

PARTquestions 2Regulatory

02 CHAPTER 2: EFFICACY OF DRUG TREATMENT FOR ACUTE MANIA DIFFERS ACROSS GEOGRAPHIC REGIONS

AN INDIVIDUAL PATIENT DATA META-ANALYSIS OF PLACEBO-CONTROLLED STUDIES

Journal of Psychopharmacology. 2015 Aug;29(8):923-32.

Welten, C.C.M., M.D.1,2*, Koeter, M.W.J., Ph.D1 , Wohlfarth, T.D., Ph.D2, Storosum, J.G., M.D., Ph.D1, van den Brink, W., M.D., Ph.D1, Gispen-de Wied, C.C., M.D., Ph.D2, Leufkens, H.G.M., Ph.D2, Denys, D.A.J.P., M.D., Ph.D1,3 1Dept. of Psychiatry, Academic Medical Centre, University of Amsterdam, Amsterdam, the Netherlands; 2Medicines Evaluation Board, Utrecht, the Netherlands; 3Netherlands Institute for Neuroscience, Royal Netherlands Academy of Arts and Sciences, Amsterdam, the Netherlands ABSTRACT

Given globalization trends in the conduct of clinical trials, the external validity of trial results across geographic regions is questioned. The objective of this study was to examine the efficacy of treatment in acute mania in bipolar disorder across regions and to explain potential differences by differences in patient characteristics. We performed a meta-analysis of individual patient data from 12 registration studies for the indication acute manic episode of bipolar disorder. Patients (n=3207), were classified into one of three geographic regions: Europe (n=981), USA (n=1270), and other regions (n=956). Primary outcome measures were mean symptom change score on the Young Mania Rating Scale (YMRS) from baseline to endpoint and responder status (50% improvement form baseline). Effect sizes were significantly smaller in the USA (g=0.203, 95%-CI 0.062-0.344; OR 1.406, 95%-CI 0.998-1.980) than in Europe (g=0.476, 95%-CI 0.200- 0.672; OR 2.380, 95%-CI 1.682-3.368) or other regions (g=0.533, 95%-CI 0.399-0.667; OR 2.300, 95%- CI 1.800-2.941). Regional differences in age, gender, initial severity, BMI, placebo response, discontinuation rate, and type of compound could not explain the geographic differences in effect. Less severe symptoms at baseline in the US patients did explain some of the difference in responder status between patients in Europe and the USA. These findings suggest that the results of studies involving patients with acute mania cannot be extrapolated across geographic regions. Similar findings have been identified in schizophrenia, contraceptive, and in cardiovascular trials. Therefore, this finding may indicate a more general problem regarding the generalizability of pharmacological trials over geographic regions.

28 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success INTRODUCTION

The function of regulatory authorities is to protect and promote public health by deciding on the market authorization of pharmaceutical products (1). To this end, they assess, among other things, the quality of clinical trial data and the relevance of the results for their domestic market (2, 3). Until a few years ago, clinical trials were mainly carried out in North America, Western Europe, South Africa, and Australia (4), but nowadays clinical trials, like the economy, are global. As discussed by Thiers et al., the globalization of clinical trials has both advantages and disadvantages. Potential benefits include dissemination of medical knowledge and effective medical practice to other parts of the world and greater access to high-quality medical care for patients worldwide. An important disadvantage is the difficulty of drawing valid scientific conclusions based on pooled data from ethnically and culturally diverse populations (4).

The globalization of clinical trials means that regulatory bodies, such as the European Medicines Agency (EMA) and the USA Food and Drug Administration (FDA), now have to evaluate the relevance of studies from different regions and involving populations that may differ from those in their own region (5, 6). The question arises whether the results of studies from one geographic region can be extrapolated to another region, bearing in mind there might be differences in intrinsic factors (patient related) and extrinsic factors (environment and culture related) (5, 7). In a recent study, Mattila et al. found differences in the efﬁcacy of atypical antipsychotics for the treatment of acute psychotic episodes in patients with schizophrenia from North America, Europe, and the rest of the world (8). The question arises whether similar differences exist in the efﬁcacy of treatment for other major psychiatric disorders, e.g., bipolar disorder.

According to the DSM-5, bipolar disorder is characterized by manic, depressive, and mixed episodes, and its diagnosis is based on the occurrence of at least one acute hypomanic episode (9). Bipolar disorder has a lifetime prevalence of approximately 1% (10), and is accompanied by a high level of social dysfunction, comorbidity, and suicide (9, 11-13). Drugs used to treat the acute manic episode in bipolar disorder are commonly grouped into (atypical) antipsychotics and (anticonvulsant) mood stabilizers.

PART 2 - CHAPTER 2 - Efficacy of drug treatment for acute mania differs across geographic regions| 29 Very little is known about differences in the efficacy and effect size of these drugs in patients from different geographic regions. In a recent meta-analysis, Vieta et al. 2011 found significant differences in the baseline characteristics and in the mean change score from baseline to follow-up between patients from the USA, India, and Russia (14). These findings suggest that there are differences in efficacy between countries. However, regulatory bodies such as the EMA are not country oriented but region oriented and organized, and it is thus essential to obtain evidence about inter-regional variations in effect.

The aim of this study was twofold: to investigate whether there are differences across geographic regions (USA, Europe, other regions) in the efﬁcacy and effect size of medications for the treatment of the acute manic episode of bipolar disorder, and to investigate whether possible regional differences in effect can be explained by regional differences in baseline characteristics, placebo response or compound distributed.

EXPERIMENTAL PROCEDURES

Selection of studies We included all studies (n=12) submitted to the Dutch Medicines Evaluation Board during an 11 year period as part of market authorization application for the indication acute manic episode of BD. All studies were double-blind randomized, placebo-controlled trials involving patients diagnosed with DSM-IV bipolar disorder. Pharmaceutical companies provided raw data in order to enable an individual patient data meta- analysis.

The drugs investigated were antipsychotics and (anticonvulsant) mood stabilizers. Active comparators were included and analyzed as treatment. For reasons of conﬁdentiality, medications are referred to as compounds A to G. We restricted the analyses to treatment groups that were given a proven effective dose of the medication, as indicated in the Summary of Product Characteristics (SmPC) if the drug was registered for an acute manic episode. If the drug was not registered for this indication, expert consensus agreed on what would constitute an effective dose taking into account the doses mentioned in SmPCs for related disorders.

30 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success Instruments The severity of the acute manic episode of BD at baseline and at study endpoint was assessed with two interview-based questionnaires. The Young Mania Rating Scale (YMRS) comprises 11 items: seven items are scored on a 0-4 scale and four are scored on a 0-8 scale. Total scores range from 0 (no symptoms) to 60 (severe symptoms). (15) The Mania Rating Scale from the Schedule for Affective Disorders and Schizophrenia – Change Version (MRS from SADS-C) comprises 11 items: one item is scored on a 0-2 scale and ten items are scored on a 0-5 scale (higher score indicates higher severity). Total scores range from 0 (no symptoms) to 52 (severe symptoms) (16).

Outcome measures We used two efﬁcacy outcomes: the standardized difference in mean change score on the YMRS or the MRS from baseline to follow-up and the difference in percentage responders. A patient was considered a responder if his/her score on the YMRS or MRS decreased by 50% or more from baseline to follow-up.

The endpoint was defined as the 3-week post-baseline assessment, since this is the time point recommended in the EMA Committee for Proprietary Medicinal Products (CPMP) guideline on the clinical investigation of medicinal products for the treatment and prevention of bipolar disorder (17). As studies involved patients from more than one region, we categorized the patients (who were from 33 countries) into three “regions”: Europe (n=17), USA (n=1), and other regions (n=15). This classification was based mainly on the country grouping used by the World Health Organization (WHO) (18). We classified studies that included patients from more than one region into separate sub-studies per region. These sub-studies are referred to as USA, European, and other studies.

STATISTICAL ANALYSIS

To answer the ﬁrst research question, we used a two-step, random effects individual patient data meta-analysis. We used random effect rather than a ﬁxed effect meta- analysis because the included studies had been performed by independently operating companies who examined different medications, tested at different times and in different populations.

PART 2 - CHAPTER 2 - Efﬁcacy of drug treatment for acute mania differs across geographic regions| 31 Table 1. Patient characteristics per sub-study .

2 2 2 2 2 2 2 2

715) 827) 580) 744) 251) 242) 162) 521) 527) 003) 981) 179) 674) 175) 230) 718) 786) 661) 408) 494) 596) 638) 429) 233) 245) 276) 591) 641) 062) 992) 793) 900) 805) 661) 541) 479) ...... (7 00 (5 14 (5 29 (6 59 (5 04 (7 91 (7 61 (7 39 (6 20 (6 30 (7 94 (4 03 (5 94 (5 51 (4 16 (5 52 (4 60 (4 17 (4 94 (7 53 (5 50 (4 00 (6 50 (5 84 (7 45 (5 25 (7 29 (6 36 (6 66 (7 60 (6 74 42 (5 69 (7 67 (5 41 (7 27 (5 ...... (Y)MRS baseline 31 25 29 25 28 30 27 30 28 29 28 28 28 25 28 26 26 28 27 26 26 28 32 34 31 34 32 34 33 33 36 30 33 29 33 29

3 3 3 3 6 6 6 6 3 4 3 3 3 6 6 6 3 3 3 3 3 3 12 12 12 12 12 12 12 12 12 12 12 12 12 12 Duration (weeks)

. 5 3 . .

7 2 6, 5, 2, . . 7 8 . . . . .

4 3 . . 1, Other 2 . 6, Other 6 .

5, Other 0 2 0, Other 3 . . . 6, Hispanic 3 . 0, Other 12 0, Other 13 . . 2, Hispanic 3 7, Hispanic 4 2, Hispanic 1 0 7, Native American 5 ...... 4, Other 1 3, Other 1

. . 0 2 . 3, Hispanic 7 9, Hispanic 11 . . 0, Asian 2 8, Asian 2 3, Asian 57 3, Asian 55 1, Hispanic 6 9, Oriental 2 9, Hispanic 1 9, Asian 1 2, Asian 1 0, Asian 7, Hispanic 8 ...... 8, Asian 1 7, Asian 1 9, Asian 26 . . .

3, Other 13 . 6 .

4 .

6, Oriental 25 9, Oriental 25 . . 8 8 9 0 .

. . . 2 . 5, Asian 76 . 4, Other 99 . 2, Other 0 .

0 0 0 0 0 0 0 0 ......

0, African American 13 1, African American 10 4, African American 2 1, African American 20 7, African American 2 3, African American 20 7, African American 20 0, African American 13 9, African American 36 8, African American 38 8, African American 39 2, Asian 0 8, African American 41 6, African American 2 2, Asian 26 6, Asian 25 1, Asian 0 0, Asian 1 1, African American 19 ...... 0, Other 91 .

8, Other 91 0 0 0 0 0 ......

2 9 . .

4 . Other 2 Other 0 Native American 1 1 Ethnicity (%) Caucasian 100 Caucasian 79 African American 10 Caucasian 81 Caucasian 94 Caucasian 10 Caucasian 94 Caucasian 11 Caucasian 72 Caucasian 80 Caucasian 100 Caucasian 61 Asian 100 Caucasian 100 Caucasian 56 Asian 100 Caucasian 100 Caucasian 59 Asian 100 Caucasian 99 Caucasian 56 Caucasian 70 Caucasian 100 Caucasian 34 Caucasian 100 Caucasian 34 Caucasian 99 Asian 100 Caucasian 99 Asian 100 African American 0 Caucasian 100 Caucasian 1 Caucasian 100 Oriental 8 Caucasian 69

25 53 55 52 44 63 41 66 52 50 29 56 55 58 56 48 59 63 44 50 60 41 44 28 39 33 43 73 45 73 64 43 70 44 70 56 ...... % Male 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

764) 513) 059) 380) 962) 153) 377) 111) 846) 009) 964) 264) 073) 911) 096) 005) 584) 149) 984) 651) 361) 766) 136) 219) 374) 879) 941) 197) ......

------(3 775 (9 832 (6 476 (5 478 (5 233 (4 513 (4 928 (4 454 (4 558 (5 602 (6 129 (4 506 (4 976 (7 470 641 (4 065 (8 644 (3 161 (5 233 (6 847 (5 209 (4 987 (4 755 (4 806 (5 213 (5 382 (3 672 (4 442 (4 ...... BMI (mean) 26 27 20 28 25 21 24 21 28 29 24 25 29 26 25 30 25 26 29 25 26 24 25 24 26 20 25 20

199 104) 830) 917) 802) 861) 073) 928) 362) 614) 243) 227) 535) 425) 560) 670) 126) 472) 357) 621) 773) 550) 924) 938) 444) 255) 895) 060) 282) 006) 318) 661) ...... 632) 830) 287) 619) . . . . 25 (7 72 (10 42 (10 81 (11 83 (12 66 (12 75 (13 35 (12 48 (10 63 (10 71 (8 13 (10 85 (11 12 (12 07 (10 70 (6 47 (11 11 (10 78 (11 10 (12 71 (10 88 (8 01 (11 76 (13 60 (11 67 (13 92 (13 64 (10 73 (14 97 (11 36 (11 82 (13 24 (12 28 (13 58 (10 17 (11 ...... Age (mean) 46 38 30 39 43 33 42 31 39 38 49 40 34 40 39 40 38 39 37 40 40 38 44 38 45 38 42 35 44 34 35 40 37 40 34 40

Exp/ contr 3/1 63/75 19/19 36/75 34/38 40/39 37/38 41/39 70/69 55/60 11/6 165/87 58/29 73/38 104/58 18/9 73/38 102/58 18/9 59/68 74/65 17/17 64/57 36/43 60/57 38/43 58/48 49/48 50/48 48/48 111/11 9 78/58 49/48 68/58 43/48 96/92

1 F F F F F C C C C C C C C C E E E E G G G G B B A A A A A D D D D D D D C ) = MRS Questionnaire (rest = YMRS Questionnaire)

E E E E E E E E E E E E E U U U U U U U U U O O O O O O O O O O O O O O Region

) = Compound;

1 Study 1_1 1_2 1_3 1A_2 2_1 2_3 2A_1 2A_3 3_2 4_2 5_1 5_2 5_3 6_1 6_2 6_3 6A_1 6A_2 6A_3 7_1 7_2 7_3 8_1 8_3 8A_1 8A_3 9_1 9_3 9A_1 9A_3 10_3 11_1 11_3 11A_1 11A_3 12_2

32 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success Table 2. Patient characteristics per region.

Region C1 Exp/contr Age (mean) BMI % Ethnicity (%) Duration YMRS baseline MRS baseline Male (weeks) Europe A,C,D, 667/314 42.12 (12.719) 25.901 0.46 Caucasian 99.2, African American 0.3, Asian 0.4, Other 3-12 29.54 (6.159) 27.90 (7.225) E,F,G (4.627) 0.1 USA A-F 764/506 39.68 (10.781) 29.220 0.56 Caucasian 66.8, African American 28.2, Asian 0.6, 3-12 28.24 (5.577) 25.35 (5.848) (7.037) Oriental 0.2, Hispanic 3.1, Native American 0.3, Other 0.8 Other A,C,D, 585/371 35.59 (11.776) 22.860 0.60 Caucasian 8.2, African American 3.1, Asian 43.2, 3-12 33.62 (7.755) 30.27 (6.857) E,F,G (4.988) Oriental 4.4, Hispanic 1.2, Other 40.0 Tot All 2016/1191 39.21 (11.973) 26.300 0.54 Caucasian 59.2, African American 12.2, Asian 13.2, 3-12 30.20 (6.830) 27.75 (6.903) (6.273) Oriental 1.4, Hispanic 1.6, Native American 0.1 Other 12.3 1) = Compound

In the ﬁrst step, we calculated the total scores on the respective questionnaires at baseline and week three. If outcome data at week three were missing, we used data for week four and if data for week four was missing, last observation carried forward analysis was used to impute the missing outcome data for week three. In the second

step, we performed a meta-analysis on the outcomes of step one. We used Hedges’ g (g) as the effect size for the continuous outcome (difference in mean total score from baseline to follow-up) and the odds ratio (OR) as the effect measure for the dichotomous outcome (responder). The interpretation of Hedges’ g is as follows: 0.20- 0.30 a ‘small effect’, around 0.50 a ‘moderate effect’, and >0.80 a ‘large effect’ (19). In addition to the effect size and 95%-conﬁdence intervals (CIs), we calculated the 95%-prediction interval for Hedges’ g and the OR (20). In a random effects meta- analysis, the 95%-prediction interval (95%-PI) indicates the upper and lower bound of the effect that may be expected when a new study is performed.

To answer the second research question, we used individual patient data. To assess the effect of potential explanatory variables such as age, gender, BMI, ethnicity, initial severity, study year, and placebo response on outcome, we used a mixed effects linear regression analysis with a random intercept for study and mixed effect logistic regression analysis with a random intercept for study.

The analyses of step one in the two-step meta-analysis were performed with SPSS version 20 (SPSS20) and those of step two were performed with Comprehensive Meta- Analysis, version two (CMA2). All mixed effects regression analyses for the second research question were performed with the xtmixed and xtmemixed programs of STATA 12.

PART 2 - CHAPTER 2 - Efﬁcacy of drug treatment for acute mania differs across geographic regions| 33 Figure 1. Flow chart of mean change score (g) by region: meta-analysis.

Meta Analysis

Group by Study name Statistics for each study Hedges's g and 95% CI Region Hedges's Standard Lower Upper g error limit Variance limit Z-Value p-Value Europe 2A_1 -0,098 0,234 -0,556 0,055 0,360 -0,421 0,674 Europe 2B_1 0,061 0,229 -0,387 0,052 0,509 0,266 0,790 Europe 5_1 -0,248 0,484 -1,196 0,234 0,699 -0,514 0,608 Europe 6A_1 1,053 0,211 0,640 0,044 1,466 4,994 0,000 Europe 6B_1 0,711 0,204 0,311 0,042 1,112 3,480 0,001 Europe 7_1 0,363 0,178 0,014 0,032 0,712 2,036 0,042 Europe 8A_1 0,266 0,182 -0,090 0,033 0,622 1,462 0,144 Europe 8B_1 0,710 0,190 0,339 0,036 1,081 3,746 0,000 Europe 9A_1 0,830 0,202 0,434 0,041 1,226 4,111 0,000 Europe 9B_1 0,809 0,209 0,400 0,044 1,218 3,878 0,000 Europe 11A_1 0,479 0,175 0,136 0,031 0,822 2,740 0,006 Europe 11B_1 0,276 0,179 -0,074 0,032 0,626 1,548 0,122 Europe 0,476 0,100 0,280 0,010 0,672 4,756 0,000 Other 1A_3 0,100 0,318 -0,523 0,101 0,723 0,315 0,753 Other 2A_3 0,094 0,223 -0,343 0,050 0,531 0,423 0,672 Other 2B_3 0,313 0,223 -0,123 0,050 0,750 1,406 0,160 Other 5_3 0,699 0,232 0,245 0,054 1,153 3,020 0,003 Other 6A_3 1,098 0,423 0,269 0,179 1,927 2,595 0,009 Other 6B_3 0,725 0,408 -0,075 0,166 1,525 1,777 0,076 Other 7_3 0,138 0,335 -0,519 0,112 0,796 0,413 0,680 Other 8A_3 0,378 0,226 -0,064 0,051 0,821 1,676 0,094 Other 8B_3 0,514 0,224 0,074 0,050 0,953 2,291 0,022 Other 9A_3 0,455 0,204 0,055 0,042 0,855 2,227 0,026 Other 9B_3 0,619 0,207 0,213 0,043 1,026 2,987 0,003 Other 10_3 0,783 0,136 0,515 0,019 1,051 5,736 0,000 Other 11A_3 0,720 0,208 0,312 0,043 1,128 3,462 0,001 Other 11B_3 0,653 0,214 0,234 0,046 1,072 3,057 0,002 Other 0,533 0,068 0,399 0,005 0,667 7,804 0,000 USA 1A_2 -0,058 0,170 -0,392 0,029 0,275 -0,343 0,732 USA 1B_2 0,110 0,201 -0,284 0,041 0,505 0,548 0,584 USA 3_2 0,410 0,170 0,076 0,029 0,744 2,405 0,016 USA 4_2 0,546 0,189 0,176 0,036 0,916 2,890 0,004 USA 5_2 0,032 0,132 -0,227 0,017 0,291 0,241 0,809 USA 6A_2 0,048 0,163 -0,272 0,027 0,367 0,292 0,770 USA 6B_2 0,183 0,164 -0,138 0,027 0,505 1,119 0,263 USA 7_2 0,119 0,169 -0,213 0,029 0,450 0,701 0,483 USA 12_2 0,478 0,147 0,189 0,022 0,767 3,242 0,001 USA 0,203 0,072 0,062 0,005 0,344 2,820 0,005 Overal l 0,396 0,044 0,309 0,002 0,483 8,925 0,000 -1,00 -0,50 0,00 0,50 1,00

Fav ours A Fav ours B

Meta Analysis

Figure 2. Flow chart of responder status (OR) by region: meta-analysis.

Meta Analysis

Group by Study name Statistics for each study Odds ratio and 95% CI Region Odds Lower Upper ratio limit limit Z-Value p-Value Europe 2A_1 1,339 0,496 3,614 0,576 0,565 Europe 2B_1 1,494 0,569 3,925 0,815 0,415 Europe 5_1 0,750 0,087 6,468 -0,262 0,794 Europe 6A_1 6,087 2,538 14,598 4,047 0,000 Europe 6B_1 2,441 1,037 5,746 2,043 0,041 Europe 7_1 1,608 0,787 3,287 1,303 0,192 Europe 8A_1 1,412 0,660 3,020 0,889 0,374 Europe 8B_1 2,515 1,176 5,381 2,377 0,017 Europe 9A_1 9,214 3,193 26,587 4,108 0,000 Europe 9B_1 5,271 1,776 15,646 2,994 0,003 Europe 11A_1 1,833 0,845 3,972 1,535 0,125 Europe 11B_1 2,143 0,975 4,710 1,897 0,058 Europe 2,380 1,682 3,368 4,894 0,000 Other 1A_3 1,247 0,339 4,589 0,332 0,740 Other 2A_3 1,333 0,532 3,340 0,614 0,539 Other 2B_3 1,417 0,570 3,522 0,750 0,453 Other 5_3 2,813 1,111 7,117 2,183 0,029 Other 6A_3 4,375 0,705 27,161 1,584 0,113 Other 6B_3 1,750 0,275 11,152 0,592 0,554 Other 7_3 1,607 0,233 11,092 0,481 0,630 Other 8A_3 1,846 0,732 4,656 1,299 0,194 Other 8B_3 2,308 0,929 5,732 1,801 0,072 Other 9A_3 1,718 0,769 3,839 1,320 0,187 Other 9B_3 2,800 1,221 6,423 2,430 0,015 Other 10_3 3,981 2,299 6,895 4,930 0,000 Other 11A_3 2,644 1,163 6,011 2,321 0,020 Other 11B_3 1,910 0,824 4,430 1,508 0,131 Other 2,300 1,800 2,941 6,650 0,000 USA 1A_2 1,029 0,525 2,015 0,082 0,935 USA 1B_2 0,861 0,386 1,924 -0,364 0,716 USA 3_2 3,128 1,508 6,489 3,064 0,002 USA 4_2 3,237 1,505 6,964 3,005 0,003 USA 5_2 0,935 0,547 1,599 -0,245 0,806 USA 6A_2 0,942 0,485 1,830 -0,176 0,860 USA 6B_2 1,398 0,724 2,700 0,999 0,318 USA 7_2 0,959 0,460 2,002 -0,111 0,912 USA 12_2 2,258 1,186 4,297 2,479 0,013 USA 1,406 0,998 1,980 1,950 0,051 Overal l 2,046 1,721 2,432 8,109 0,000 0,01 0,1 1 10 100

Fav ours A Fav ours B

Meta Analysis

34 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success

RESULTS

Of the 12 studies, four were performed in only one region (three in the USA and one in the other regions), and of the seven medications, one was investigated in only one region. Subdividing the 12 studies per medication and region resulted in 36 sub-studies, of which 13 were performed in Europe (Belgium, Bulgaria, Croatia, Czech Republic, Estonia, France, Germany, Greece, Hungary, Latvia, Lithuania, Poland, Romania, Russia, Turkey, Ukraine, United Kingdom), 9 in the USA, and 15 in the other regions (Argentina, Australia, Chile, China, Hong Kong, India, Indonesia, Korea, Malaysia, New Zealand, Philippines, Singapore, South Africa, Taiwan, Tunisia). The total number of patients in the included studies was 3207: 981 in Europe, 1270 in the USA, and 956 in the other in the other regions. Table 1 presents the patient characteristics per sub-study (stratiﬁed by study, medication type, and region). Table 2 presents the overall patient characteristics stratiﬁed by region. Results for the two-stage meta-analysis are presented in the Forest plots of Figures 1 and 2.

Effect in terms of mean change scores The combined Hedges’ g (95%-CI) for all studies was 0.396 (0.306-0.486, tau:0.234), resulting in a 95%-prediction interval of -0.089-0.881. Hedges’ g (95%-CI) differed across the regions: Europe 0.476 (0.253-0.699), USA 0.203 (0.033-0.373), and other regions 0.533 (0.385-0.681). These differences were statistically signiﬁcant (p<0.01). This can be interpreted as a small effect in the USA and a moderate effect in Europe and other regions (Figure 1).

Effect in terms of responder status The combined OR (95%-CI) for all studies was 2.048 (1.721-2.432, tau:0.133), resulting in a 95%-prediction interval of 0.954-4.390. The ORs (95%-CI) differed across the regions: Europe 2.380 (1.682-3.368), USA 1.406 (0.998-1.980), and other regions 2.300 (1.800-2.941). These differences were statistically signiﬁcant (p<0.05) (Figure 2).

Differences in baseline characteristics across regions There were significant differences at baseline in age (p<0.001), BMI (p<0.01), initial severity (p<0.01), gender (p<0.001), and ethnicity (p<0.001) between the US and European studies. Only severity measured with the MRS was not significantly different (n=433, p=0.673). All variables were significantly different in the US and other region studies (p<0.05).

PART 2 - CHAPTER 2 - Efﬁcacy of drug treatment for acute mania differs across geographic regions| 35 Patients in the USA were younger (39.68) than patients in Europe (42.12) and older than the patients in the other regions (35.59). They had a higher BMI (29.22) than patients in Europe (25.90) and the other regions (22.86) and were less severely ill at baseline according to the YMRS and MRS scores (28.24, 25.35) than patients in Europe (29.54, 27.90) and the other regions (33.62, 30.27). The US studies included more men (0.56) than the European studies (0.46) and fewer than the other region studies (0.60). Ethnicity was strongly related to region: the proportion of Caucasians, African Americans, and Asians was 99.2%, 0.3%, and 0.4%, respectively, in Europe, 66.8%, 28.2%, and 0.6%, respectively, in the USA, and 8.2%, 3.1%, 43.2%, respectively, in the other regions (Table 3).

Differences in placebo response across regions Patients in the other regions had a higher placebo response than patients in the USA in terms of both mean change score (-10.132 vs. -7.561; p=0.017) and responder status (0.423 vs. 1.962; p=0.021). There was no signiﬁcant difference in the placebo response between patients in Europe and the USA (mean change score -7.161 vs. -7.561, respectively; p=0.702) or in responder status (0.392 vs. 0.423, respectively; p=0.862).

Differences in study year across regions All studies were performed between 1996 and 2007. Studies including US patients were performed in the period 1996-2007 and studies including patients from Europe and other regions were performed in the period 1998-2007.

Differences in discontinuation rates across regions There was a signiﬁcant difference (p<0.001) in discontinuation rates between patients in the US and patients in Europe and patients in the other regions, showing a higher discontinuation rate in patients in the US (40%) compared to patients in Europe (21%) or the other regions (21%).

POTENTIAL EXPLANATORY VARIABLES

Mean change score After adjusting for the potential explanatory variables age, gender, initial severity, BMI, placebo response, and study year, and discontinuation rate separately, we still

36 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success Table 4. IPD Analyses effect of potential confounders on mean change score (g) by region in YMRS questionnaire b(SE)

Unadjusted Adjusted for variables (SE) Age1 Gender1 Severity1 BMI2 Placebo Study Year1 Discontinuation Combined3 Response1 Rate1 Europe vs USA -1.965 (0.958) -1.965 (0.958) -1.965 (0.958) -1.867 (0.941) -4.074 (1.119) -2.895 (0.096) -2.009 (0.096) -2.727 (0.956) -2.857 (0.936) Other vs USA -4.342 (1.032) -4.342 (1.032) -4.385 (1.032) -4.629 (1.013) -4.946 (1.131) -5.092 (1.031) -4.3667 (1.032) -5.045 (1.032) -5.465 (1.004) LRT χ2(2)a (p-value) 14.47 p<0.001 17.47 p<0.001 17.79 p<0.001 20.48 p<0.001 19.39 p<0.001 24.73 p<0.001 17.73 p<0.001 24.16 p<0.001 29.86 p<0.001 1) Analyses only for studies using YMRS questionnaire. N=2764; 2) BMI could only be analysed in N=2186; 3) Combined adjustment for all variables except for BMI.

Mixed model linear regression analyses with random intercept for study and heterogenetic level one variance. Dependent variable = mean change score. Independent variables are Region and Treatment (control

group is reference category). Region is a categorical variable (USA=0, Europe=1, Other=2) represented in the model by 2 dummy variables Region(1) contrasting Europe vs. USA and Region(2) contrasting Other vs.

USA. In each column the results are presented for an analysis with the variable mentioned in the column header added as covariate.

Likelihood ratio test model with vs. model without Treatment by Region interaction χ2(2) = 6.39, p-value = 0.041

a) LRT = likelihood ratio test comparing model with and without the Treatment by Region interaction terms

Table 5. IPD Analyses effect of potential confounders on effect modiﬁcation response (OR, p-value) by region in YMRS questionnaire.

Unadjusted Adjusted for variables (p-value) Age1 Gender1 Severity2 BMI3 Placebo Study Year1 Discontinuation Combined4 Response1 Rate1 Europe vs USA 1.431 (0.061) 1.430 (0.061) 1.431 (0.061) 1.400 (0.104) 2.044 (0.001) 1.682 (0.008) 1.438 (0.058) 1.404 (0.176) 1.665 (0.016) Other vs USA 1.537 (0.020) 1.541 (0.019) 1.562 (0016) 1.526 (0.037) 1.678 (0.027) 1.676 (0.005) 1.543 (0.019) 2.114 (0.274) 1.721 (0.008) LRT χ2(2)a (p-value) 6.39 p=0.041 6.44 p=0.040 6.71 p=0.035 5.00 (0.082) b) 11.08 p=0.039 10.48 p=0.005 6.53 p=0.038 6.43 p=0.040 8.96 p=0.011 1) Analyses based on all mania studies n=3207; 2)Analyses only based on the YMRS Questionnaire studies N=2764; 3) BMI could only be analysed in N=2186; 4) Combined adjustment for all variables except for BMI.

Mixed model logistic regression analyses with random intercept for study. Dependent variable = Responder. Independent variables are Region and Treatment (control group is reference category). Region is a

categorical variable (USA=0, Europe=1, Other=2) represented in the model by 2 dummy variables Region(1) contrasting Europe vs. USA and Region(2) contrasting Other vs. USA. In each column the results are

presented for an analysis with the variable mentioned in the column header added as covariate.

Likelihood ratio test model with vs. model without treatment by region interaction χ2(2) = 6.39, p-value =0.041

a) LRT = likelihood ratio test comparing model with and without the treatment by region interaction terms, b) these analysis only used studies that used the YMRS subjects (we needed an initial severity score)

found signiﬁcant differences across regions (p<0.001). After simultaneous adjustment

for the combination of the variables age, gender, severity, placebo response, and study year, and discontinuation rate, the signiﬁcant differences across geographic

regions remained (p<0.001). Since ethnicity was strongly confounded with region, we were only able to assess the potential confounding effect of African Americans on YMRS data for 1097 US patients in a post-hoc analysis (Table 4). Adjustment for this potential confounder resulted in an unadjusted effect estimate -2.531 (SE:0.597, p<0.001) versus an unadjusted effect estimate of -2.521 (SE:0.597, p<0.001), showing that the proportion African Americans did not inﬂuence the mean change score in patients in the USA. This indicated that differences in ethnic composition between the regions could not explain the differences in mean change score between the regions.

Responders Age, gender, BMI, and placebo response, and discontinuation rate did not explain the observed geographic differences in effect size. However, after adjustment for initial severity the difference between responders from Europe and the USA was no longer signiﬁcant (p=0.082), but the difference between responders from the other regions

PART 2 - CHAPTER 2 - Efficacy of drug treatment for acute mania differs across geographic regions| 37 and the USA remained significant (p=0.037). Nevertheless, the same tendency for a higher proportion of responders in European studies compared with US studies remained. After simultaneous adjustment for age, gender, severity, placebo response, and study year, the significant differences across regions remained (p<0.01). (Table 5) Since ethnicity was strongly confounded with region (see above), we assessed the potential confounding effect of African Americans in the US studies (n=1270 patients) in a post-hoc analysis. This resulted in an adjusted odds ratio of 1.474 (95%-CI:1.159- 1.875) versus an unadjusted odds ratio of 1.460 (95%-CI:1.149-1.857), indicating that differences in ethnic composition between the regions could not explain the differences in the proportion of responders between the regions.

Sub-group Antipsychotics To analyze whether geographic differences could be caused by a different distribution of compound type to the three regions, we analyzed the main group of compounds, the antipsychotics. This led to an overall combined Hedges’ g (SE) of 0.482 (0.047). Hedges’ g differed signiﬁcantly (p<0.01) across the regions: Europe 0.543 (0.101), USA 0.248 (0.081), and other regions 0.627 (0.070). However, in this subgroup analysis, the difference in responder status (OR, 95%-CI) across regions was no longer signiﬁcant (p=0.134), although the size and direction of the differences was very similar to that found in the main analyses: USA 1.564 (0.915-2.678), Europe 2.465 (1.518-4.002), and other regions 2.562 (1.830-3.588).

DISCUSSION

We investigated whether results from efficacy studies of drugs for the treatment of acute mania can be extrapolated across geographic regions, and whether potential differences in efficacy can be ascribed to regional differences in baseline characteristics, placebo response, discontinuation rate, and compound distributed. We found significant differences in drug effect size across regions, with European studies (g=0.476, OR=2.380) and other region studies (g=0.533, OR=2.300) consistently reporting larger effects than US studies (g=0.203, OR=1.406). Although we found significant regional differences in baseline characteristics (age, gender, ethnicity, BMI, and initial severity), and significant differences in placebo response and discontinuation rates between patients in the USA and other regions, these differences did not explain the regional

38 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success differences in mean change score from baseline to follow-up. However, differences in responder status in the US and European studies could be partly explained by the lower initial severity of manic episodes in the US patients compared with that in the European patients, but this was not the case for differences in responder status between the US and other region studies.

Our results are consistent with the findings of Mattila et al., who showed a tendency to regional differences in the efficacy of atypical antipsychotics in the acute treatment of patients with schizophrenia in North America versus Europe and the rest of the world. They found a smaller effect size in US studies (8). Furthermore, our findings are consistent with those of Vieta et al., who also found that patients treated for acute mania in the USA were younger, had a higher BMI, and had less severe symptoms than patients in Russia and India, and that, as in our study, none of the baseline characteristics was a significant predictor of change in MRS score (14). Unlike Vieta et al but in line with other reports, we did not find a higher placebo response in the US studies (21).

Geographic differences in drug efficacy have been reported for other health issues/ conditions, including cardiovascular disease (22-24) and contraception in contraceptive trials (25, 26). Over the years, efficacy of contraceptives frequently showed lower efficacy in clinical trials conducted in US patients versus trials conducted in European patients (25, 26). Despite efforts of Blair et al. (2008) the EVEREST study to select a fairly homogenous population of patients with heart failure, important differences in etiology, severity, management, and outcomes were found (23). A review by Mentz et al. also found differences in outcome and baseline characteristics of patients with different types of cardiovascular disorders (heart failure, acute coronary syndromes, hypertension and atrial fibrillation) across geographic regions (24). Similarly, efficacy of contraceptives frequently showed lower efficacy in clinical trials conducted in US patients compared to trials conducted in Europe (25, 26), an issue that was addressed by the FDA (27).

The current study had two limitations. First, the p-value of responder status of the US studies was only marginally signiﬁcant (p=0.051), unlike that of the European and other region studies (p<0.001). In order to determine whether this was caused by a

PART 2 - CHAPTER 2 - Efficacy of drug treatment for acute mania differs across geographic regions| 39 difference in drug distributed, we separately analyzed the main group of compounds, the antipsychotics. This led to an even larger difference in the magnitude of the mean change score between the US, European, and other region studies. However, the difference in effect size of responder status across regions was no longer significant, although the differences were similar. The second limitation is the limited ability to explain regional differences, because of the limited availability of information about baseline variables in our database. Possibly potential explanatory factors not investigated in the current study include genetic differences (intrinsic factor), and differences in available treatment medical services, differences in the patient recruitment, benefits of participation, differences in interpretation of disease-specific symptoms, duration and onset of illness (14, 28-30), and history of prior (pharmacological) treatments/ medication and hospitalization. (14, 28, 30) Vieta et al. did not find the last four variables to be significantly associated with treatment outcome (14).

To our knowledge, this is the first study to evaluate geographic differences in the efficacy of drugs for the treatment of acute mania in patients with bipolar disorder, based on individual patient data meta-analysis and meta-regression-analysis. The effects of treatment were significantly smaller in the USA (small effect) than in Europe and other regions (moderate effect), and these differences could not be explained by differences in baseline characteristics, study year, differences in placebo response or compound distributed across geographic regions. These findings may affect the requirements of regulatory authorities with regard to regional extrapolation of study results. The development of new medicines, and not only for psychiatric patients, has become a global activity. This involves many different continents, ethnic populations, or health systems. The interpretation of study results by regulators, payers and clinical practice, therefore implies complex questions regarding the external validity of the findings and the consistency of study results across subsets. These questions will remain on the table, because research, including our study, pertinently reveals that geography matters when it comes to the interpretation of clinical data from studies in different psychiatric settings. With regard to geographic differences in efficacy for drug treatment of patients with acute mania, we found that geographic differences could only partly be explained. But we also saw that albeit differences, overall data showed efficacy of drug treatment in acute manic patients both in US and European/other patients. This would therefore probably result in a positive benefit/risk balance in patients of all three

40 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success regions. However, in drugs for a less severe indication, for chronic use, or with smaller effect sizes and more severe adverse events, this judgment could of course change. Therefore, whether the FDA can and will accept European data or whether European regulators will do vice versa, remains a question of judgment and regulatory policies and future research could help in this judgment. In regulatory terms, we can only aim for a prescription drug label that is rooted in the best science available, which is not the same as the best thinkable science. Our study shows that there is scientiﬁc room for extrapolation, not always, and sometimes with obvious restrictions.

PART 2 - CHAPTER 2 - Efﬁcacy of drug treatment for acute mania differs across geographic regions| 41 REFERENCES

1. Gispen-de Wied CC, Leufkens HG. From molecule to market access: drug regulatory science as an upcoming discipline. European Journal of Pharmacology. 2013;719:9-15. 2. Rehnquist J. The globalization of clinical trials: a growing challenge in protecting human subjects. Washington, DC: Department of Health and Human Services. Journal Tnternational de Bioéthique. 2003;14:165-169. 3. FDA: The Food and Drug Administration’s oversight of clinical trials. . Edited by Services DoHaH. Washington, DC2007. 4. Thiers FA, Sinskey AJ. Trends in the globalization of clinical trials. Nature Reviews Drug Discorvery. 2008;7:13-14. 5. CHMP: Reflection paper on the extrapolation of results from clinical studies conducted outside the EU to the EU-population. Committee for Medicinal Products for Human Use 2008. 6. FDA: Guidance for Industry and FDA Staff - FDA Acceptance of Foreign Clinical Studies Not Conducted Under an IND frequently Asked Questions. Edited by Services USDoHaH2012. 7. ICH: E5(R1): Ethnic Factors in the Acceptability of Foreign Clinical Data 1998. 8. Mattila T, Wohlfarth T, Koeter M, Storosum J, van den Brink W, de Haan L, Leufkens H, Denys D. Geographic variation in efficacy of atypical antipsychotics for the acute treatment of schizophrenia – An individual patient data meta-analysis. European Neuropsychopharmacology. 2014;24:1067-1077. 9. American Psychiatric Association A: Handboek voor de classificatie van psychische stoornissen (DSM- 5). Nederlandse vertaling van Diagnostic and Statistical Manual of Mental Disorders. . Fifth Edition ed. Amsterdam, Boom; 2014. 10. Merikangas KR, Akiskal HS, Angst J, Greenberg PE, Hirschfeld RM, Petukhova M, Kessler RC. Lifetime and 12-month prevalence of bipolar spectrum disorder in the National Comorbidity Survey replication. Archives of General Psychiatry. 2007;64:543-552. 11. Rosa AR, Reinares M, Franco C, Comes M, Torrent C, Sánchez-Moreno J, Martínez-Arán A, Salamero M, Kapczinski F, Vieta E. Clinical predictors of functional outcome of bipolar patients in remission. Bipolar Disorders. 2009;11:401-409. 12. Sanchez-Moreno J, Martinez-Aran A, Tabarés-Seisdedos R, Torrent C, Vieta E, Ayuso-Mateos JL. Functioning and disability in bipolar disorder: an extensive review. Psychotherapy and Psychosomatics. 2009;78:285–297. 13. Sierra P, Livianos L, Rojo L. Quality of life for patients with bipolar disorder: relationship with clinical and demographic variables. Bipolar Disorders. 2005;7:159-165. 14. Vieta E, Pappadopulos E, Mandel FS, Lombardo I. Impact of geographical and cultural factors on clinical trials in acute mania:lessons from a ziprasidone and haloperidol placebo-controlled study. International Journal of Neuropsychopharmacology. 2011;14:1017-1027. 15. Young RC, Biggs JT, Ziegler VE, Meyer DA. A rating scale for mania: reliability, validity and sensitivity. British Journal of Psychiatry 1978;133:429-435. 16. Spitzer RL, Endicott J. Schedule for Affective Disorders and Schizophrenia - Change Version (3rd ed.). New York: Biometrics Research. 1987. 17. CHMP: Note for guidance on clinical investigation of medicinal product of the treatment and prevention of bipolar disorder. Committee for Medicinal Products for Human Use 2001. 18. WHO: WHO Country Groupings. 2013. 19. Cohen J: Statistical power analysis for the behavioral sciences. 2nd ed. . Hillsdale (NJ), Erlbaum; 1988.

42 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success 20. Borenstein M, Hedges LV, Higgins JPT, Rothstein H: Introduction to Meta-Analysis. , Wiley; 2009. 21. Watsky EJ, Schwartz JH, Kremer CME: The impact of geography on precision in clinical trials : experience with placebo and an active control in a phase 2a study in patients with acute symptoms of schizophrenia. San Diego (CA)2009. 22. Mentz RJ, Kaski JC, Dan GA, Goldstein S, Stockbridge N, Alonso-Garcia A, Ruilope LM, Martinez FA, Zannad F, Pitt B, Fiuzat M, O’Connor CM. Implications of geographical variation on clinical outcomes of cardiovascular trials. American Heart Journal. 2012;164:303-312. 23. Blair JEA, Zannad F, Konstam MA. Continental differences in clinical characteristics, management, and outcomes in patients hospitalized with worsening heart failure: results from the EVEREST (Efficacy of Vasopressin Antagonism in Heart Failure: Outcome Study with Tolvaptan) program. Journal of the American College of Cardiology 2008;52:1640-1648. 24. Stough WG, Zannad F, Pitt B, Goldstein S. Globalization of cardiovascular clinical research: the balance between meeting medical needs and maintaining scientific standards. American Heart Journal. 2007;154:232-238. 25. Grubb GS, Archer DF, Constantine GD. Differences between the United States and Europe in clinical trials of hormonal contraceptive efficacy. Obstetrics & Gynecology. 2008;111:63S-64S. 26. Lete I, Pérez de Arrilucea M, Rodriquez M, Belleo E. Efficacy, safety, and patient acceptability of the etonogestrel and ethinyl estradiol vaginal ring. Journal of Contraception. 2014;5:39-48. 27. FDA: General meeting on contraceptives. in Advisory committee briefing document FDA; 2007. 28. Welge JA, Keck JPE, Meinhold JM. Predictors of Response to Treatment of Acute Bipolar Manic Episodes With Divalproex Sodium or Placebo in 2 Randomized, Controlled, Parallel-Group Trials. Journal of Clinical Psychopharmacology. 2004;24:607-612. 29. Zarate CAJ, Narendran R, Tohen M, Graeney JJ, Berman A, Pike S, Madrid A. Clinical predictors of acute response with olanzapine in psychotic mood disorders. Journal of Clinical Psychiatry. 1998;59:24- 28. 30. Swann AC, Janicak PL, Calabrese JR, Bowden CL, Dilsaver SC, Morris DD, Petty F, Davis JM. Structure of mania: depressive, irritable, and psychotic clusters with different retrospectively-assessed course patterns of illness in randomized clinical trial participants. Journal of Affective Disorder. 2001;67:123-132.

PART 2 - CHAPTER 2 - Efﬁcacy of drug treatment for acute mania differs across geographic regions| 43

03 CHAPTER 3: PLACEBO RESPONSE IN ANTIPSYCHOTIC TRIALS OF PATIENTS WITH ACUTE MANIA

RESULTS OF AN INDIVIDUAL PATIENT DATA META-ANALYSIS

European Neuropsychopharmacology. 2015 Jul;25(7):1018-26.

We examined the role of placebo response in acute mania trials. Specifically, whether placebo response: (1) predicts treatment effect, (2) can be predicted by patient and study characteristics, and (3) can be predicted by a parsimonious model. We performed a meta-analysis of individual patient data from 10 registration studies (n=1,019) for the indication acute manic episode of bipolar disorder. We assessed the effect of 14 determinants on placebo response. Primary outcome measures were mean symptom change score (MCS) on the Young Mania Rating Scale (YMRS) and response rate (RR), defined as ≥50% YMRS symptom improvement from baseline to endpoint. The overall placebo response was 8.5 points improvement on the YMRS (=27.9%) with a RR of 32.8%. Placebo response was significantly associated with the overall treatment response. Five determinants significantly (p<0.05) predicted the placebo response. The multivariate prediction model, which consisted of baseline severity, psychotic features at baseline, number of geographic regions, and region, explained 10.4% and 5.5% of the variance in MSC and RR, respectively. Our findings showed that the placebo response in efficacy trials of antipsychotics for acute mania is substantial and an important determinant of treatment effect. Placebo response is influenced by patient characteristics (illness severity and presence of psychotic features) and by study characteristics (study year, number of geographic regions and region). However, the prediction model could only explain the placebo response to a limited extent. Therefore, limiting trials to certain patients in certain geographic regions seems not a viable strategy to improve assay sensitivity.

46 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success INTRODUCTION

Efﬁcacy trials of medicines for psychiatric disorders have a higher failure rate than trials of medicines for other medical disorders. (1-5) These failures mostly become manifest in phase III clinical trials, where drugs are tested for efﬁcacy and safety in the target group in doubled blinded randomised placebo-controlled studies. (1) Unfortunately, it diminishes the likelihood of new medicines becoming available for psychiatric disorders. One of the most frequently cited reasons for the high failure rate is the relatively high placebo response in trials of psychiatric medications. (2-8) This makes it important to understand the role of the placebo response in psychiatric treatment.

Numerous studies have reported on the placebo response in psychiatric trials, (6, 7, 9-16) and, contrary to clinical belief, it is even reported to be high in patients with schizophrenia and bipolar disorder (8, 17-20) However, fewer research has focused on determinants of the placebo response, especially not in trials involving patients with an acute manic episode of bipolar disorder. (8, 10, 18-20) In a recent meta-regression analysis of data from 38 placebo-controlled randomized controlled trials involving patients with an acute manic or mixed episode, Yildiz et al. (2011) unexpectedly found that symptomatic improvement was similar in the placebo arms of trials of effective and ineffective drugs, and that a higher placebo response was associated with patient (female, older age) and study (more study sites, more recent studies) characteristics. (8) However, as only a limited number of patient and study characteristics were included, and other, not studied, patient and study characteristics might be related to the magnitude of the placebo response, it is important to take these aspects into consideration when designing studies with a low placebo response and a high success potential. Moreover, meta-regression analysis has a relatively low power compared to individual patient data (IPD) meta-analysis, and only the latter allows calculation of the total amount of variance in the placebo response due to patient and study characteristics.

The aim of this study was to examine the magnitude of the placebo response, the effect of the placebo response on the treatment effect, and the role of patient and study characteristics in the prediction of the placebo response in placebo-controlled trials of antipsychotics in patients with acute mania using IPD meta-analysis. More speciﬁcally,

PART 2 - CHAPTER 3 - Placebo response in antipsychotic trials of patients with acute mania| 47 (1) we tested whether the magnitude of the placebo response predicts the treatment effect, (2) we investigated which patient and study characteristics predict the placebo response, and (3) we identiﬁed the most parsimonious model to predict the placebo response.

EXPERIMENTAL PROCEDURES

Selection of studies We included all short-term efﬁcacy studies that assessed antipsychotics and which had been submitted to the Dutch Medicines Evaluation Board (CBG-MEB) during an eleven-year period as part of market authorization application for the indication acute manic episode of bipolar disorder. All studies were double-blind, randomized, placebo-controlled trials involving patients diagnosed with DSM-IV bipolar disorder. Pharmaceutical companies provided their raw patient data, which enabled us to perform a meta-analysis of IPD.

The studies investigated ﬁve different types of antipsychotic medication. Active drugs were included and analysed as treatment. Nine studies studied antipsychotic mono- therapy, one study used an add-on, placebo-controlled design. In order to protect the company’s interests, the medications investigated are referred to as compounds A to E. We restricted the analyses to the data of patients given a proven effective dose of the medication according to the Summary of Product Characteristics (SmPC) if the drug was registered for the treatment of the acute manic episode; if the drug was not registered for this indication, then expert consensus established whether a dose was effective, based on the doses mentioned in SmPCs for related disorders.

Assessments We used the Young Mania Rating Scale (YMRS), an interview-based questionnaire, to assess the severity of the acute manic episode of bipolar disorder. The YMRS comprises 11 items: 7 items are scored on a 0-4 scale and 4 are scored on a 0-8 scale. Total scores ranges from 0 (no symptoms) to 60 (severe symptoms). (21)

Outcome measures We used two efﬁcacy outcomes: the standardized difference in mean change score on the YMRS from baseline to follow-up and the percentage responders, with response

48 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success deﬁned as a decrease ≥ 50% from baseline to follow-up on the YMRS.

The study endpoint was chosen as 3 weeks after baseline, since this is the duration recommended in the EMA Committee for Proprietary Medicinal Products (CPMP) guideline on the clinical investigation of medicinal products for the treatment and prevention of bipolar disorder. (22) If outcome data at week 3 were missing, we used week 4 data in studies that lasted longer than 3 weeks, and if data at week 4 were missing or not available, last observation carried forward (LOCF) was used.

Predictors Patient characteristics included age, (4, 6, 8, 15, 18) gender, (4, 6, 8, 15, 18) body mass index (BMI), ethnicity, (15) illness severity at baseline, (4, 6, 8, 23) and psychotic features at baseline. (8, 19) Ethnicity was classified as Caucasian, Black, Asian, and Other (Oriental, Native American, Hispanic, and Other), with Caucasian as reference category. The presence of psychotic features at baseline was defined as a score of 3 (‘flight of ideas; tangentially; difficult to follow; rhyming; echolalia’) or 4 (incoherent; communication impossible) on question 7 of the YMRS or a score of 6 (‘Grandiose or paranoid ideas; Ideas of reference’) or 8 (‘Delusions; Hallucinations’) on question 8 (‘Content’) of the YMRS questionnaire.(21)

Study characteristics included study year, (6, 8, 9, 15, 18) number of visits per protocol, (15, 23) number of study arms, (6) number of countries, (6, 8, 16, 19) region, (24) number of regions, mean change score on the YMRS in the treatment arm, (6, 8, 9, 15) and proportion of patients assigned to receive placebo. (4, 8, 16, 18, 24) Region was classiﬁed into three areas: Europe, USA, and Other. This classiﬁcation was based mainly on the country grouping used by the World Health Organization (WHO). (25) The number of visits from baseline to week 3 was established on the basis of the study protocols.

STATISTICAL ANALYSIS

First, we calculated the total score for the YMRS questionnaire at baseline and follow- up. Then we performed a multilevel mixed effect linear regression analyses with a random intercept for study for mean change score, and a multilevel mixed effect logistic

PART 2 - CHAPTER 3 - Placebo response in antipsychotic trials of patients with acute mania| 49 Table 1. Patient and study characteristics per study1,2 .

% patients placebo 49.6 52.2 34.3 21.3 50.0 33.6 31.9 51.7 30.8 48.9 40.1

Visits (weeks) 0, 1, 2, 3 0, 1, 2, 3 0, 0.3, 0.5, 1, 2, 3 0, 0.3, 0.5, 1, 2, 3 0, 0.3, 0.5, 1, 2, 3 0, 0.5, 1, 2, 3, 0, 0.5, 1, 2, 3, 0, 0.5, 1, 2, 3 0,1,2,3 0, 0.5, 1, 2, 3 Tot

Number countries 1 1 6 8 10 11 7 1 8 1 25

) Active comparator in study; 3

Region USA USA EUR, USA, Other EUR, USA, Other EUR, USA, Other Europe, Other EUR, Other Other Europe, Other USA Tot

YMRS baseline 27.74 (6.349) 29.88 (7.291) 28.74 (5.263) 26.55 (5.006) 26.70 (4.975) 33.02 (6.535) 33.95 (6.897) 37.12 (7.818) 31.35 (6.895) 29.58 (5.861) 30.45 (6.516)

Psychotic at BSL (N/Tot) 5/69 8/60 11/122 6/105 12/150 11/100 19/96 72/1019

72.0%, Asians

Ethnicity Caucasians 69.6%, Blacks 21.7%, Other 8.7% Caucasians 86.7%, Blacks 11.7%, Other 1.7% Caucasians 50.0%, Blacks 25.4%, Asians 23.8%, Other 0.8% Caucasians 68.6%, Blacks 20.0%, Asians 9.5%, Other 1.9% Caucasians 76.0%, Blacks 19.3%, Asians 4.0%, Other 0.7% Caucasians 22.0%, Other 6.0% Caucasians 49.0%, Asians 51.0% Asians 100.0% Caucasians 54.7%, Asians 2.8%, Other 42.5% Caucasians 71.7%, Blacks 14.1%, Other 14.1% Caucasians 57.9%, Blacks 11.4%, Asians 23.4%, Other 7.4%

Female 0.46 0.50 0.43 0.48 0.49 0.63 0.42 0.40 0.47 0.36 0.46

BMI 27.88 (6.30) 28.51 (7.53) 27.57 (6.67) 25.30 (4.62) 23.00 (5.33) 26.65 (6.52)

) Study characteristics based on total study group (treatment + (active comparator) + placebo). 2

8.77 Age 3 (10.31) 38.2 (10.18) 40.42 (11.22) 38.29 (9.95) 40.27 (11.17) 40.69 (12.35) 41.06 (13.65) 35.62 (11.69) 37.33 (12.60) 40.33 (11.95) 39.17 (11.69)

group only; 3 -

AC - - - C - E - - E - Tot

Studied compound A A B B B C C D D D Tot

Nr. patients 69 60 122 105 150 100 96 119 106 92 1019

isits per protocol per week; V

) Patient characteristics based on placebo )

1 4

Study 1 2 3 4 5 6 7 8 9 10 Tot

50 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success regression analyses with a random intercept for study for response rate, and performed a likelihood ratio test to test where the placebo response predicts the treatment effect. We used the regression coefﬁcient (β) for the continuous outcome (mean change score) and the odds ratio (OR) as the effect measure for the dichotomous outcome (responder).

Table 2. Effect of placebo response on treatment effect.

Response rate β SE (b) OR 95%-CI p-value Constant -2.244 0.342 0.106 0.054-0.2073 0.054 % Responders (placebo) 0.046 0.010 1.047 1.027-1.0683 <0.001 Study group (treatment) 1.903 0.400 6.704 3.063-14.6753 <0.001 % Responders (placebo) * Treatment1 -0.037 0.012 0.963 0.941-0.9863 0.002 Mean change score Constant 0.055 3.237 NA -6.291-6.4004 0.987 % Responders (placebo) -0.259 0.100 NA -0.455—0.0644 0.009 Study group (treatment) -11.472 2.052 NA -15.493—7.4504 <0.001 % Responders (placebo) * Treatment2 0.189 0.061 NA 0.069-0.3104 0.002 1) LRχ2(1)=9.46, prob> χ2=0.0021; 2) LRχ2(1)=10.01, prob> χ2=0.0016 ; 3) 95%-CI based on OR ; 4) 95%-CI based on β ; NA = Not Applicable

The ﬁrst analyses were performed with SPSS version 20 (SPSS20), and the multilevel mixed effect regression analyses were performed with STATA 12.

To find predictors of the placebo response, we performed for each potential predictor a separate mixed effects linear regression analysis with a random intercept for study for the continuous outcome (change score) and mixed effect logistic regression analysis with a random intercept for study for the responder outcome. To adjust for the fact that we tested for 14 predictors, a Bonferroni corrected significance threshold of p=0.004 (0.05/14) was used in all analyses. Because this correction seriously affects the power of each test we decided to refer to predictors with a p-value between 0.004 and 0.01 as borderline significant.

To identify the most parsimonious model to predict the placebo response, we conducted step-wise multi-level (logistic) regression analyses (backward elimination) for all predictors with a univariate p-value of less than 0.20, hereby taking into account that several predictors were correlated. Analyses were performed with SPSS20 and STATA 12.

PART 2 - CHAPTER 3 - Placebo response in antipsychotic trials of patients with acute mania| 51 RESULTS

Study characteristics Data from 10 studies involving in total 1019 patients on placebo were analysed. The studies examined the effect of ﬁve different antipsychotics. Deleting the add-on study from the analyses did not change the results. Table 1 presents the patient and study characteristics for each study.

Placebo response and treatment effect Overall, the placebo response was -8.5 points on the YMRS (27.9% improvement) and the response rate was 32.8%. The placebo response rate was significantly associated with the effect of antipsychotics in placebo-controlled trials involving patients with acute mania: a one percent higher placebo response rate was associated with a significantly smaller difference in response rate (β=-0.037, SE=0.012, OR=0.963, LRχ2(1)=9.46, prob>χ2 =0.0021,) and a significantly smaller difference in mean change scores (β=0.189, SE=0.061, LRχ2(1)=10.01, prob> χ2=0.0016) between the active drug and placebo. Placebo mean change score was non-significantly associated with treatment effect by response rate and mean change score (p>0.05). (Table 2)

Table 3. Determinants of the response rate.

Univariate regression Responders Dummy variables β SE(b)1 OR 95%-CI2 p-value Age -0.002 0.006 0.998 0.986-1.009 0.671 Female Ntot=546 0.375 0.136 1.455 1.115-1.898 0.006 BMI Ntot=573 -0.007 0.014 0.993 0.966-1.021 0.610 Ethnicity Overall 0.0523 Blacks vs Caucasian -0.122 0.361 0.886 0.437-1.796 0.736 Asian versus Caucasian 0.548 0.340 1.729 0.889-3.364 0.107 Other versus Caucasian -0.007 0.297 0.993 0.555-1.778 0.982 Severity -0.034 0.011 0.967 0.947-0.988 0.002 Psychotic features at -3.025 0.723 0.049 0.012-0.2003 <0.001 baseline Study year 0.082 0.028 1.085 1.028-1.146 0.003 Number of study arms 0.284 0.212 1.328 0.877-2.013 0.180 Number of countries 0.037 0.027 1.037 -0.016-0.089 0.169 Number of regions Overall 0.0024 Two versus one region -0.014 0.190 0.986 0.680-1.431 0.942 Three versus one region 0.525 0.175 1.691 1.200-2.383 0.003 Region Overall 0.0055 Europe versus USA 0.087 0.204 1.091 0.731-1.628 0.669 Other versus USA 0.639 0.217 1.895 1.239-2.899 0.003 Number of visits 0.302 0.107 1.353 1.097-1.669 0.005 MSC treatment group -0.041 0.033 0.960 0.900-1.024 0.215 %patients receiving -0.003 0.010 1.003 0.983-1.023 0.784 placebo 1) SE of regression coefficient (β) ; 2) 95%-CI of OR ; 3) Wald χ2(3) = 7.73 ; 4) Wald χ2(3)= 12.35 ; 5) Wald χ2(2)= 10.70

52 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success

Table 4. Determinants of the mean change score.

Univariate regression Mean change score Dummy variables β SE 95%-CI z-value p-value Age 0.039 0.029 -0.018-0.097 1.35 0.179 Female Ntot=546 -1.259 0.679 -2.589-0.071 -1.86 0.063 BMI Ntot=573 0.054 0.067 -0.077-0.185 0.80 0.421 Ethnicity Overall 0.4001 Blacks vs Caucasian 0.875 1.691 -2.440-4.189 0.52 0.605 Asian versus Caucasian -1.416 1.714 -4.775-1.942 -0.83 0.408 Other versus Caucasian 0.169 1.424 -2.623-2.961 0.12 0.906 Severity -0.224 0.054 -0.323—0.118 -4.13 <0.001 Psychotic features at 6.395 1.549 3.359-9.432 4.13 <0.001 baseline Study year -0.429 0.198 -0.818—0.041 -2.16 0.030 Number of study arms -2.803 1.201 -5.156-0.450 -2.33 0.020 Number of countries -0.342 0.155 -0.645—0.038 -2.21 0.027 Number of regions Overall 0.0432 Two versus one region -2.316 1.397 -5.053-0.422 -1.66 0.097 Three versus one region -3.236 1.325 -5.832-0.640 -2.44 0.015 Region Overall 0.0103 Europe versus USA -0.033 0.957 -1.909-1.843 -0.03 0.972 Other versus USA -2.732 1.030 -4.751—0.734 2.65 0.008 Number of visits -1.470 0.809 -3.056-0.116 -1.82 0.069 MSC treatment group 0.063 0.250 -0.428-0.552 0.25 0.805 %patients receiving -0.028 0.068 -0.106-0.161 -0.41 0.685 placebo 1) Wald χ2(3)= 2.94; 2) Wald χ2(2)= 6.30; 3) Wald χ2(2)= 9.18

Predictors of response rate With regard to patient characteristics, a higher placebo response rate was found in patients with less severe illness at baseline (OR=0.967, 95%-CI 0.947-0.988), and in patients without psychotic features at baseline (OR=0.049, 95%-CI 0.012-0.200). Furthermore, a higher placebo response was found in women than in men, although this was borderline signiﬁcant (OR=1.455, 95%-CI 1.115-1.898). The placebo response rate was not signiﬁcantly predicted by age (p=0.671), BMI (p=0.610), or ethnicity (p=0.052). (Table 3)

With regard to study characteristics, the placebo response rate was higher in more recent studies than in older studies (OR=1.085, 95%-CI 1.028-1.146), in studies with patients from three regions compared to one region (OR=1.691, 95%-CI 1.200-2.383), and in patients from the Other regions compared to patients from the USA or Europe (OR =1.895, 95%-CI 1.239-2.899). Furthermore, a higher number of visits per protocol predicted a higher placebo response, although this effect was borderline signiﬁcant (OR=1.353, 95%-CI 1.097-1.669). The placebo response rate was not signiﬁcantly predicted by the number of countries (p=0.169), the number of study arms (p=0.180),

PART 2 - CHAPTER 3 - Placebo response in antipsychotic trials of patients with acute mania| 53 the percentage patients assigned to the placebo condition (p=0.784), or the mean change score on the YMRS in the treatment arm (p=0.215). (Table 3)

Table 5. Prediction Model

Responders Mean change score Determinant Dummy variables β SE(b)1 OR 95%-CI2 p-value R3 β SE 95%-CI p-value R3 Severity -0.036 0.011 0.965 0.944-0.987 0.002 -0.254 0.054 -0.360--0.149 <0.001 Psychotic features at -3.002 0.726 0.050 0.0120-0.206 <0.001 7.051 1.355 4.391-9.710 <0.001 baseline Number of regions Overall <0.0014 <0.0016 Two versus one region -0.023 0.216 0.977 0.640-1.493 0.915 -1.988 1.077 -4.102-0.126 0.065 Three versus one region 0.622 0.185 1.862 1.296-2.676 0.001 -4.805 0.892 -6.557—3.054 <0.001 Region Overall <0.0015 0.0037 Europe versus USA 0.043 0.205 1.044 0.699-1.560 0.833 0.700 1.046 -1.352-2.752 0.503 Other versus USA 0.846 0.192 2.331 1.601-3.394 0.000 -2.372 0.947 -4.231—0.514 0.012 Total 0.055 0.104 1) SE of regression coefficient (β) ; 2) 95%-CI of OR ; 3) coefficient of determination ; 4) χ2(2)= 15.52 ; 5) χ2(2)= 24.13; 6) F(2, 1019)= 15.006; 7) F(2, 1019)= 5.735

Predictors of mean change score

With regard to patient characteristics, the placebo mean change score was higher in

patients with high versus low illness severity at baseline (β=-0.224, 95%-CI -0.323—

0.118), and in patients without psychotic features at baseline (β=6.395, 95%-CI 3.359-

9.432). It was not signiﬁcantly predicted by age (p=0.179), BMI (p=0.421), gender

(0.063), or ethnicity (p=0.400). (Table 4)

With regard to study characteristics, the placebo mean change score was higher in patients from regions other than the USA or Europe (β=-2.732, 95%-CI -4.751— 0.734). Study year (p=0.030), the number of study arms (p=0.020), the number of countries (p=0.027), the number of regions (p=0.043), the number of visits (p=0.069), the percentage patients assigned to receive placebo (p=0.685), and the mean change score on the YMRS in the treatment arm (p=0.805) did not signiﬁcantly predict the mean change score. (Table 4)

Prediction Model Analyses revealed that the most parsimonious model to predict the placebo response consisted of two patient characteristics and two study characteristics (response rate and mean change score, respectively). Of the patient characteristics, illness severity at baseline (OR=0.965, 95%-CI 0.944-0.987 and β=-0.254, 95%-CI 0.360--0.149, respectively) and the presence of psychotic features at baseline (OR=0.050, 95%-CI 0.012-0.206 and β=7.051, 95%-CI 4.391-9.710, respectively) were included in the model. Of the study characteristics, region (χ2(2)=24.13 and χ2(2)=10.475, respectively) and number of regions (χ2(2)=15.52 and χ2(2)=15.01, respectively) were included. Less

54 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success severe illness at baseline (for the response rate) and more severe illness at baseline (for the mean YMRS change score), the presence of psychotic features at baseline, patients from the USA and Europe versus the Other region, and the inclusion of three versus one or two regions were strongly predictive of a higher placebo response. The explained variance of the prediction model was 5.5% for the response rate (R2=0.055) and 10.4% for the mean change score (R2=0.104). (Table 5)

DISCUSSION

The magnitude, the effect, and the predictors of the placebo response were investigated by analysing IPD from 10 pivotal double-blinded placebo-controlled short-term efﬁcacy studies with antipsychotics for the indication acute manic episode in bipolar disorder. The overall placebo response rate was substantial, 32.8%, and the overall placebo mean change score on the YMRS from baseline to follow-up was 8.5 points (27.9% improvement). Consistent with previous studies, a higher placebo response rate was strongly associated with a smaller effect size. Predictors of a higher placebo response included absence of psychotic features at baseline, higher illness severity score at baseline (for YMRS mean change score), lower illness severity score at baseline (for response rate), recent study year (for response rate), inclusion of three geographic regions compared with one or two (for response rate), and patients from regions other than the USA and Europe. Borderline signiﬁcant predictors of a higher placebo response included female sex (for response rate) and more study visits described in the protocol (for response rate). In a parsimonious prediction model, only two patient and two study characteristics independently contributed to the prediction of the placebo response (baseline illness severity, psychotic features at baseline, region, and number of regions), together explaining ≤ 10% of the variance in the placebo response.

Our findings with regard to the overall magnitude of the placebo response is consistent with Yildiz et al., who found a placebo response rate of 30.8% and a mean change score of 6.92. Furthermore, our findings with regard to the predictors of the placebo response are also generally in line with those of Yildiz et al. 2011, who found female gender to predict a higher placebo response. However, in our study female gender was only borderline significant in response rate. In relation to Yildiz et al. (2011), there were also some contradictory findings; we found a more recent study and the absence

PART 2 - CHAPTER 3 - Placebo response in antipsychotic trials of patients with acute mania| 55 of psychotic features to increase the placebo response, whereas Yildiz et al. (2011) did not find an association with these variables, and in accordance to Sysko et al. (2007) we did not find age to significantly predict the placebo response. (8, 18) Unfortunately, the number of study sites, which appeared a strong predictor in Yildiz et al (2011), could not be assessed because we lacked the information. Even though region was not studied before, our findings are in line with those of Vieta et al. 2011, who showed a higher placebo response in patients from the USA versus patients from India and Russia. (26) The effect of illness severity at baseline on the placebo response has not been studied previously in patients experiencing acute manic episodes of bipolar disorder. Lastly, contrary to the findings of studies assessing other severe psychiatric disorders, the mean YMRS change score in the treatment arm and the percentage of patients in the placebo arm did not predict the placebo response in patients with acute mania. (4, 27, 28)

In our study, the different direction of the relation between illness severity and the two different outcomes (MCS vs. RR) is probably due to the definition of the two outcome measures; for the more severely ill at baseline there is more room for improvement in the YMRS total score (MCS), whereas for them it is harder to meet criteria for response (≥ 50% improvement from baseline to follow-up). With regard to the psychotic features, patients without psychotic features show more improvement and response on placebo than patients with psychotic features, indicating that the patients with acute mania without psychotic features are more likely to show placebo response. More recent studies had a higher placebo response (RR), which is in line with the reviews by Sysko and Walsh (2007) and Keck 2000. (18, 19) As also suggested by Keck et al., this could be attributed either to changes in study design over the years, or to an overall higher standard of care for patients with bipolar disorder. In our study, we found that study design features such as the geographic region and the number regions within a study significantly predicted placebo response, showing higher placebo response in the Other region and when three regions versus one region were included. Finally, more visits per protocol predicted a higher placebo response (RR), which could be explained by more structure, control and motivation obtained by more visits. Unfortunately, we were not able to test the effect of some other study characteristics due to unbalanced or limited data in our study, e.g. flexible versus fixed dosing (29) and number of study sites (8).

Compared with previous studies, our study had several strengths. We were able to

56 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success analyse IPD, which is a more powerful method than meta-regression and which has been recommended in previous studies. (6, 30) IPD also enabled us to analyse the effect of patient characteristics in addition to study characteristics, including a number of characteristics not previously investigated in patients with acute mania (BMI, severity at baseline, region, number of regions, mean YMRS change score treatment group). Furthermore, the face that we used all studies offered to the MEB for registration, i.e. irrespective of whether they had been published or not, reduced the possibility of publication bias. However, using only studies offered to the MEB for registration, and all studies were conducted by the pharmaceutic industry, increased the danger of selection bias. However, analysing our database on selection bias in the publication bias program of Comprehensive Meta-Analysis (CMA), Rosenthal’s fail safe N was 1649, meaning it would have taken 1649 missed studies with a mean effect size of ‘0’ for the p value of the summary effect to become non-significant. Furthermore, the point estimate would only change from 0.458 to 0.415 if adjusted for publication bias. Therefore, the role of selection bias is considered limited. Second, we did not have individual item scores for the YMRS questionnaire of three studies or the BMI data of five studies, which meant that we could not determine the impact of psychotic features and BMI on outcomes in these studies. Third, data were available for only a limited number of determinants, so that potential predictors of the placebo response could not be evaluated, such as duration and onset of illness, prior history of medication, and prior history of hospitalization. Also, since acute mania has a fluctuating course, it is important to study whether entry into a clinical trial before or after the peak of the manic episode predicts the placebo response. (20) These determinants should be investigated in future trials, in order to improve our understanding of the placebo response.

Our ﬁndings conﬁrm the hypothesis that the high failure rate of psychiatric trials is at least partly due to a high placebo response rate. Over the years, this high placebo response and the related negative trials have become a serious problem for the pharmaceutical industry. At the same time, it can be considered positive for patients participating in a trial: even in the placebo condition there is a chance of improvement and response. An ethical dilemma? No, clinicians should always strive for the best standard treatment including those that enhance to placebo response in pharmacological trials. The pharmaceutical industry has to accept that and choose designs that increase the probability to obtain positive outcomes such as reliable

PART 2 - CHAPTER 3 - Placebo response in antipsychotic trials of patients with acute mania| 57 (dimensional) outcome measures, adequate statistical power, good compliance, and adequate training of trial personal especially in studies with multiple study sides. One may also restrict the number of study sites, but this may jeopardize the generalizability of the study. We identified two patient and two study characteristics that independently predicted the placebo response, but the practical relevance of this finding is limited given the small percentage of variance explained. Restricting future studies to patients with psychotic features and recruited only in the USA or Europe would seriously limit the generalizability of the findings without significantly reducing the placebo response and would not improve assay sensitivity, i.e. it would not increase the probability of showing statistically significant and clinically relevant effects of the medication under study. Generalizability would even be further jeopardized if studies were also restricted to male patients and to studies using a limited number of visits per protocol (the borderline significant predictors). The placebo response in trials assessing the efficacy of medication in patients with an acute manic episode of bipolar disorder remains a great challenge.

58 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success REFERENCES

1. Nutt D, Goodwin G. ECNP Summit on the future of CNS drug research in Europe 2011: report prepared for ECNP by David Nutt and Guy Goodwin. European Neuropsychopharmacology 2011;21:495-499. 2. Walsh BT, Seidman SN, Sysko R, Gould M. Placebo response in studies of major depression: variable, substantial, and growing. . JAMA. 2002;287:1840-1847. 3. Vieta E, Cruz N. Increasing rates of placebo response over time in mania studies. Journal of Clinical Psychiatry. 2008;69:681-682. 4. Kemp AS, Schooler NR, Kalali AH, al. e. What is causing the reduced drug–placebo difference in recent schizophrenia clinical trials and what can be done about it? Schizophrenia Bulletin. 2010;36:504- 509. 5. Khan A, Detke M, Khan SR, Mallinckrodt C. Placebo response and anti-depressant clinical trial outcome. Journal of Nervous and Mental Disease. 2003;191:211-218. 6. Agid O, Siu CO, Potkin SG, Kapur S, Watsky EJ, Vanderburg D, Zipursky RB, Remington G. Meta- Regression Analysis of Placebo Response in Antipsychotic Trials, 1970–2010. American Journal of Psychiatry. 2013;170:1335-1344. 7. Kinon BJ, Potts AJ, Watson SB. Placebo response in clinical trials with schizophrenia patients. Current Opinion in Psychiatry. 2011;24:107-113. 8. Yildiz A, Vieta E, Tohen M, Baldessarini RJ. Factors modifying drug and placebo responses in randomized trials for bipolar mania. International Journal of Neuropsychopharmacology. 2011;14:863- 875. 9. Gispen-de Wied C, Stoyanova V, Yu Y, Isaac M, Pani L, de Andres-Trelles F. The placebo arm in clinical studies for treatment of psychiatric disorders: a regulatory dilemma. European Neuropsychopharmacology. 2012;22:804-811. 10. Vieta E, Carné X. The Use of Placebo in Clinical Trials on Bipolar Disorder: A new Approach for an Old Debate. Psychotherapy and Psychosomatics. 2005;74:10-16. 11. Chen JK, Wang SJ, Khin NA, Hung HMJ, Laughren TP. Trial design issues and treatment effect modeling in multi-regional schizophrenia trials. Pharmaceutical Statistics. 2010;0:217-229. 12. Khin NA, Chen YF, Yang Y, Yang P, Laughren TP. Exploratory analyses of efﬁcacy data from major depressive disorder trials submitted to the U.S. Food and Drug Administration in support of new drug applications. Journal of Clinical Psychiatry. 2011;72:464-472. 13. Post RM, Denicoff KD, Leverich S. Special Issues in Trial Design and Use of Placebo in Bipolar Illness. Biological Psychiatry. 2000;47:727-732. 14. Sun W, Laughren TP, Zhu H, Hochhaus G, Wang Y. Development of a placebo effect model combined with a dropout model for bipolar disorder. Journal of Pharmacokinetics and Pharmacodynamics. 2013;40:359-368. 15. Cohen D, Consoli A, Bodeau N, Purper-Ouakil D, Deniau E, Guile JM, C. D. Predictors of Placebo Response in Randomized Controlled Trials of Psychotropic Drugs for Children and Adolescents with Internalizing Disorders. . Journal of Child and Adolescent Psychopharmacology. 2010;20:39-47. 16. Mallinckrodt CH, Tamura RN, Tanaka Y. Recent developments in improving signal detection and reducing placebo response in psychiatric clinical trials. Journal of Psychiatric Research 2011;45:1202- 1207. 17. Storosum JG, Wohlfarth T, Schene A, Elferink A, van Zwieten BJ, van den Brink W. Magnitude of effect of lithium in short-term efﬁcacy studies of moderate to severe manic episode. Bipolar Disorders.

PART 2 - CHAPTER 3 - Placebo response in antipsychotic trials of patients with acute mania| 59 2007;9:793-798. 18. Sysko R, Walsh BT. Systematic review of placebo response in studies of bipolar mania. Journal of Clinical Psychiatry. 2007;68:1213–1217. 19. Keck JPE, Welge JA, McElroy SL, Arnold LM, al. e. Placebo effect in randomized, controlled studies of acute bipolar mania and depression. Biological Psychiatry. 2000;47:748-755. 20. Chengappa KNR, Tohen M, Levine J, Jacobs T, Thase ME, Sanger TM, Kupfer DJ. Response to placebo among bipolar I disorder patients experiencing their first manic episode. Bipolar Disorders. 2000;2:332–335. 21. Young RC, Biggs JT, Ziegler VE, Meyer DA. A rating scale for mania: reliability, validity and sensitivity. British Journal of Psychiatry. 1978;133:429-435. 22. CPMP: Note for guidance on clinical investigation of medicinal product of the treatment and prevention of bipolar disorder. Edited by products Cfpm2001. 23. Montgomery SA. The failure of placebo-controlled studies European Neuropsychopharmacology. 1999;9:271-276. 24. Mallinckrodt CH, Zhang L, Prucka WR, Millen BA. Signal detection and placebo response in schizophrenia: parallels with depression. . Psychopharmacology Bulletin. 2010;43:53-72. 25. WHO: Http://www.who.int/quantifying_ehimpacts/global/ebdcountgroup/en/ 26. Vieta E, Pappadopulos E, Mandel FS, Lombardo I. Impact of geographical and cultural factors on clinical trials in acute mania: lessons from a ziprasidone and haloperidol placebo-controlled study. International Journal of Neuropsychopharmacology. 2011;14:1017-1027. 27. Papakostas GI, Fava M. Does the probability of receiving placebo influence clinical trial outcome? A meta-regression of double-blind, randomized clinical trials in MDD. European Neuropsychopharmacology. 2009;19:34-40. 28. Zimmerman M, Posternak M: Placebo response in antidepressant efficacy trials: Relationship to number of active treatment groups. in 156t Annu Meet Am Psychiatr Assoc. San Francisco2003. pp. 17- 22. 29. Khan A, Khan SR, Walens G, Kolts R, Giller EL. Frequency of Positive Studies Among Fixed and Flexible Dose Antidepressant Clinical Trials: An Analysis of the Food and Drug Administraton Summary Basis of Approval Reports. Neuropsychopharmacology 2003;2003:552-557. 30. Fournier JC, DeRubeis RJ, Dimidjian S, Amsterdam JD, Shelton RC, Fawcett J. Antidepressant drug effects and depression severity: a patient level meta-analysis. JAMA. 2010;303:47-55.

60 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success PART 2 - CHAPTER 3 - Placebo response in antipsychotic trials of patients with acute mania| 61

04 CHAPTER 4: NET GAIN ANALYSIS, AN ADDITION TO RESPONDER ANALYSIS

THE CASE OF ANTIPSYCHOTIC TREATMENT OF ACUTE MANIA

Regulatory Toxicology and Pharmacology. 2015 Jul 8;73(1):227-231.

Net Gain Analysis (NGA) is proposed as an alternative to Responders Analysis (RA) as a more comprehensive method to tap clinical relevance of the effect of treatment. NGA is the group difference in responders minus the group difference in deteriorators; while RA is the group difference in responders. We examined the performance of these two methods in a dataset consisting of individual patient data from 10 randomized controlled trials (N=2666) of ﬁve different antipsychotics in patients with acute mania by comparing the rank ordering of the ﬁve compounds according to both systems (NGA and RA). The rank order did not differ between the 2 systems but the inferiority of one compound was revealed more evidently by the NGA in comparison to the RA.

64 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success INTRODUCTION

To grant a licence for registration, a drug is tested on efficacy and safety in phase III clinical trials. The drug should be able to demonstrate not only a statistically significant improvement on the primary efficacy endpoint compared to the placebo or active comparator, but the magnitude of effect must also be clinically relevant (1).

To translate statistical significant efficacy to clinical relevance, both the US Food and Drug Administration (FDA) and the European Medicines Agency (EMA) specifically endorse a responder analyses in clinical trials. (2, 3). For this reason, the responder analysis (RA) is incorporated in most regulatory guidelines. Responder analysis requires an a priori defined minimum improvement relative to baseline that marks a clinically relevant effect. (4-10) This way, it allows the definition of success to depend on baseline illness severity, resulting in a more clinically relevant assessment of the actual effect of the treatment drug.

In short-term efficacy studies in patients with an acute manic episode the mean change is used to assess significant improvement and clinical relevance is most often defined as at least 50% improvement compared to baseline and a minimal predefined difference in the percentage of responders (those with a reduction of at least 50% compared to baseline) between the active compound and the placebo or active comparator. (11)

One may question, however, whether the difference in the percentage responders between the treatment conditions is a complete operationalization of clinical relevance. Responder analysis focuses only on the differential probability of a successful treatment and does not take the differential probability of deterioration into account. In almost all diseases, deterioration could be caused by non-response to treatment in combination with the natural course of the disease, or by a speciﬁc mechanism of action of treatment or be due to a particular phenotype of the patient. Although several alternatives for responder analysis have been proposed, including the Lehmann method, the shifted responder, and the relative effect, they all disregard the differential probability of deterioration. (7, 12, 13) This is of particularly importance for diseases like an acute manic episode where the probability of becoming a responder is limited (approximately 50%) (14) and deterioration in the acute phase may cause substantial mental, physical

PART 2 - CHAPTER 4 - Net gain analysis, an addition to responder analysis| 65 and social harm. In these situations, net gain analysis (NGA) might be a useful alternative for the assessment of clinical relevance. Net gain is deﬁned as the difference in the percentage responders between the treatment and placebo group minus the difference in percentage deteriorators between these groups.

This article describes this new theoretical framework the ‘net gain analysis’ and compares its results with those of the traditional responder analysis in the case of antipsychotic treatment of patients with acute mania using data from 10 randomized controlled trials using ﬁve different antipsychotic compounds.

METHODS

Selection of studies We pooled the individual patient data of all short-term efﬁcacy studies that assessed the antipsychotic treatment of acute mania and had been submitted to the Dutch Medicines Evaluation Board (CBG-MEB) in an eleven-years period as part of market authorization application for the indication acute manic episode of bipolar disorder. All studies were double-blind, randomized, placebo-controlled trials including patients diagnosed with DSM-IV acute manic episode of bipolar disorder. The pharmaceutical companies conducting these studies provided their raw patient data.

The studies investigated a total of ﬁve different antipsychotic compounds. Active comparators were included and analysed as treatment. In order to protect the company’s interests, the medications investigated are referred to as compounds A to E. We restricted the analyses to the patients who were prescribed the medication in an effective dose according to the Summary of Product Characteristics (SmPC) if the drug was granted a licence for the treatment of the acute manic episode; if the drug was not granted a licence for this indication, then expert consensus established whether a dose was effective, based on the doses mentioned in SmPCs for related disorders. The individual patient data per compound were pooled for analyses.

Assessments We used the Young Mania Rating Scale (YMRS), an interview-based questionnaire, to assess the severity of the acute manic episode of bipolar disorder. The YMRS comprises

66 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success 11 items: 7 items are scored on a 0-4 scale and 4 are scored on a 0-8 scale. Total scores ranged from 0 (no symptoms) to 60 (severe symptoms).(15)

Outcome measures Response was defined as a decrease in YMRS total score between baseline and follow-up of at least 50%. Deterioration was defined as an increase in YMRS total score between baseline and follow-up of at least 10%. Net gain was defined as the difference in percentage responders between the treatment and placebo group minus the difference in percentage deteriorators between the treatment and placebo group.

The 50% cut-off point for responders was based on earlier literature, since there was no consensus in the EMA Committee for Proprietary Medicinal Products (CPMP) guideline on the clinical investigation of medicinal products for the treatment and prevention of bipolar disorder. The 10% individual cut-off point for deterioration was based on the literature and expert opinion. (11)

The study endpoint was 3 weeks after baseline measurement, since this is the duration recommended in the EMA guideline on the clinical investigation of bipolar disorder for short-term efﬁcacy studies in the acute manic episode. (11) When outcome data at week 3 were missing, we used week 4 data in studies that lasted longer than 3 weeks, and when data at week 4 were missing too or not available, last observation carried forward (LOCF) was used.

In the three studies that compared the effect of two active medications with placebo (three-way studies) we had to use the placebo group twice to be able to compute the gain for each active drug separately.

STATISTICAL ANALYSIS

Net gain of the active treatment over the placebo condition is defined as (%-Responderstreatment - %-Respondersplacebo) – (%-Deterioratorstreatment - %-Deterioratorsplacebo). Response is a priori deﬁned as a decrease of the YMRS total score of at least 50% compared to baseline, deterioration as an increase of the YMRS total score of at least 10% compared to baseline.

PART 2 - CHAPTER 4 - Net gain analysis, an addition to responder analysis| 67 Table 1. Patient and study characteristics per study1.

Visits (weeks) 0, 1, 2, 3 0, 1, 2, 3 0, 0.3, 0.5, 1, 2, 3 0, 0.3, 0.5, 1, 2, 3 0, 0.3, 0.5, 1, 2, 3 0, 0.5, 1, 2, 3, 0, 0.5, 1, 2, 3, 0, 0.5, 1, 2, 3 0,1,2,3 0, 0.5, 1, 2, 3 Tot

Region USA USA EUR, USA, Other EUR, USA, Other EUR, USA, Other Europe, Other EUR, Other Other Europe, Other USA Tot

YMRS baseline 28.20 (6.53) 29.30 (7.00) 28.29 (5.30) 27.24 (5.00) 26.68 (5.25) 33.01 (6.23) 33.28 (6.68) 36.74 (7.79) 31.57 (6.84) 29.27 (5.48) 30.08 (6.79)

Blacks %, Other

% %, Other %, Other % 7 %, Blacks %, Blacks %, Blacks %, Blacks %, Asians %, Asians %, Other %, Blacks 1.2

3 24.4 3.7 6.1 15.4% 0.0 4.2 9.6 % ) Visits per protocol per week 72.7%, 48.6 68.4 51.7 69.1 3

Other 7.2% 100.0

%, %, Other %, Asians %, Asians 9. %, Other % % %, Other 1 % 0 0.1 3.9 0.4 9.7 Ethnicity Caucasians 2 Caucasians 8 1 Caucasians 26.1 Caucasians 20.9 Caucasians 76. 19.3%, Asians 0.7% Caucasians 7 10.4 Caucasians 48.3 Asians Caucasians 5 4 Caucasians 1 Caucasians 59.8%, Blacks 12.6%, Asians 18.9%, Other 8.7% 0.8% 1.

Female 0.48 0.50 0.46 0.43 0.46 0.63 0.43 0.36 0.45 0.44 0.46

BMI 28.21 (6.73) 27.90 (6.61) 27.55 (6.30) 25.63 (4.58) 23.42 (5.54) 29.94 (6.33) ) Active comparator (AC) in study;

Age 39.48 (10.93) 38.63 (10.36) 39.30 (11.15) 39.27 (10.91) 40.24 (11.01) 42.90 (12.84) 39.44 (13.08) 35.36 (11.90) 39.04 (12.56) 40.17 (11.66) 39.48 (11.80)

AC - - - C - E - - E - Tot

C C B B B A A D D D Tot Studied compound

105 106

100

122 150 96 119 69 60 92 /Placebo 2 reatment/ Nr. Patient T AC 70/ 55/ 234/ 195/193/ 150/ 100/98/ 107/ 111/ 127/111/ 96/ 2666

) Patient characteristics based on treatment group only Study 1 2 3 4 5 6 7 8 9 10 Tot 1

68 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success Between-compound comparisons were examined with a difference of proportion test. In the case of net gain, we took into account the fact that the percentage responder and percentage deteriorators in a study are correlated, in the calculation of the standard error. To assess whether the net gain analysis affected the rank order of the five different antipsychotic compounds we ranked the compounds by efficacy in terms of the percentages of responders and the percentages of net gain. Ranking was based on the pattern of the between-compound p-values, the inferiority/superiority pattern among the different compounds, and the magnitude of effect in the two analyses. For the overall effect of compound in terms of responder analysis was assessed with multilevel logistic regression with a random intercept for study. The overall effect of compound in terms of net gain was assessed with nominal multilevel logistic regression with a random intercept for study. The random intercepts take the dependence of the data, due to the fact that we pooled different studies, into account. Fisher’s combined test was used to test the significance of the overall effect of medication type on net gain score. We used a correction factor of α(k+1)/2k to correct for dependencies of the k tests. (16)

All analyses were performed with SPSS version 20 (SPSS20) and Excel 2010.

RESULTS

Table 2. Results responder, deterioration and net gain analysis by compound type.

Compound A B C D E p-value %-Respondertreatment 53.6 47.8 46.0 50.6 46.9 %-Respondersplacebo 27.1 40.3 29.6 29.0 29.1 ∆ %-Responders 26.5 7.5 16.4 21.6 17.8 0.0122 p-value1 <0.001 0.045 <0.001 <0.001 0.001 Ranking (RA) 1 5 3 2 4 % Deterioratorstreatment 13.6 3.1 6.5 6.4 4.3 % Deterioratorsplacebo 22.5 8.8 15.9 17.7 16.0 ∆ %-Deteriorators -8.9 -5.7 -9.4 -11.3 -13.7 0.5642 p-value1 0.068 <0.001 0.002 <0.001 <0.001 Ranking (DA) 5 3 4 2 1 ∆ %-Responders 26.5 7.5 16.4 21.6 17.8 ∆ %-Deteriorators -8.9 -5.7 -9.4 -11.3 -13.7 Net Gain (%) 35.4 13.2 25.8 33.0 29.5 95%-CI [0.229;0.479] [0.087;0.177] [0.198;0.318] [0.265;0.395] [0.217;0.373] p-value <0.001 <0.001 <0.001 <0.001 <0.001 <0.0013 Ranking (NGA) 1 5 4 2 3 1) Based on multilevel multinomial logistic regression; 2) based on multilevel multinomial logistic regression; 3) based on Fisher’s combined test; χ 2(20)=88.8

PART 2 - CHAPTER 4 - Net gain analysis, an addition to responder analysis| 69

Table 3A. Efﬁcacy by responder analysis (95%-CI between different compounds).

A B C D A B [0.058 ; 0.232]* C [-0.035 ; 0.236] [-0.185 ; 0.007] D [-0.088 ; 0.185] [-0.238 ; -0.044]* [-0.154 ; 0.049] E [-0.061 ; 0.235] [-0.216 ; 0.010] [-0.130 ; 0.10] [-0.079 ; 0.155] *) = Significant difference (p<0.05)

Table 3B. Efﬁcacy by net gain analysis (95%-CI between different compounds),

A B C D A B [0.110 ; 0.334]* C [-0.022 ; 0.214] [-0.214 ; -0.038]* D [-0.106 ; 0.154] [-0.294 ; -0.102]* [-0.170 ; 0.026] E [-0.087 ; 0.205] [-0.263 ; -0.063]* [-0.140 ; 0.066] [-0.079 ; 0.149] *) = Significant difference (p<0.05)

Figure 1. Error bar graph responder analysis and net gain analysis by compound.

60 Responder Analysis Net Gain Analysis

10 Percentage Responders/Net Gain

0 A B C D E A B C D E Compound

70 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success Study characteristics Data of 10 studies involving 2,666 patients were analysed. Table 1 presents the patient and study characteristics for each study.

Responder analyses The entitlement responders in the active drug groups varied between compounds, ranging from 46% to 54%. We also found a substantial variation in placebo response between the studies (27% - 40%). Using responder analysis as the measure of clinical relevance, we found the overall effect of all compounds to be signiﬁcant. However, not all compounds were equally effective and some compounds differed statistically signiﬁcant from others (p=0.012) (Table 2). Between-compound comparisons showed that compound B (responder difference between active drug and placebo: 7.5%) was less good than compounds A and D with 26.5% and 21.6%, respectively. (Table 2, Table 3a, and Fig. 1)

GAIN ANALYSES

Deterioration The percentage deteriorators in the active drug groups varied between compounds, ranging from 3% to 14%. In the placebo groups this percentage ranged from 9% to 23%. (Table 3) All compounds but compound A (p=0.068) signiﬁcantly reduced the risk for deterioration (p<0.01) and the reduction was comparable for all ﬁve compounds (p=0.564). (Table 2)

Net Gain Net gain of the compounds varied from 13.2% for compound B to 35.4% for compound A. Using net gain analysis as the measure of clinical relevance, we found the overall effect of all compounds to be signiﬁcant. However, not all compounds were equally effective and some compounds differed statistically signiﬁcant from others (p<0.001) (Table 2). Between-compound comparisons showed that the net gain of compound B (net gain between active drug and placebo: 13.2%) was lower than of all other compounds. (Table 2, Table 3b, and Fig. 1)

In terms of the responder analyses, all ﬁve antipsychotic compounds were signiﬁcantly

PART 2 - CHAPTER 4 - Net gain analysis, an addition to responder analysis| 71 more effective than placebo (p<0.05). Of the ﬁve different compounds, only compound B was signiﬁcantly less effective than compounds A and D (p=0.005, p=0.004, respectively). Therefore, compound B was ranked last. The remaining compounds were ranked by the magnitude of the responder effect. This resulted in the following rank order: A (26.5%), D (21.6%), E (17.8%), C (16.4%), and B (7.5%).

The ranking in terms of net gain analysis, showed that compound B was signiﬁcantly inferior to all other compounds while the differences between the other four compounds were not statistical signiﬁcant. Therefore, compound B was ranked last. The remaining compounds were ranked by the magnitude of their net gain. This resulted in the following rank order: A (35.4%), E (33.0%), D (29.5%), C (25.8%), B (13.2%).

Sensitivity analysis The cut-off point for deterioration was arbitrary. To check whether a different cut-off point would have changed the conclusions of our net gain analysis, we performed a sensitivity analysis using two alternative cut-off points for deterioration; individual deterioration ≥0% and individual deterioration ≥20% compared to baseline. Although, as expected, the net gain per compound slightly changed, the alternative cut-off points did not affect the signiﬁcance of the net gain and the ranking of the compounds in terms of net gain. Results of the net gain analyses using these alternative deterioration cut-off points are presented in Supplement S.1 and S.2.

DISCUSSION

The aim of this study was to examine a new theoretical framework with regard to the assessment of clinical relevance. We examined whether a more comprehensive measure of clinical relevance using the newly developed net gain analysis is better than the traditionally applied responder analysis in the case of antipsychotic treatment of patients with acute mania.

Although, in terms of responder analysis compounds A to E were all significant more effective than placebo, compound B was significantly less effective than compounds A and D. In terms of net gain analysis, all compounds were also significantly more effective than placebo, however, compound B was now significantly less effective than

72 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success all other compounds. This suggests that the net gain analysis is a more precise efﬁcacy measure, describing the differences in clinical relevance of the different compounds more comprehensively.

The ranking of the compounds in terms of clinical relevance was very similar for the responder analysis and the net gain analysis, but the magnitude of the effects of the compounds was more comparable when using the net gain analysis compared to the responder analysis. This is in line with literature, were the comparability in efﬁcacy of different antipsychotics is broadly discussed and frequently described to be interchangeable (17-20) though the opposite has also been stated (21). Guidelines usually do not rank order the effectiveness of the different antipsychotics. (18, 22, 23) Our additional approach for the assessment of clinical relevance, might contribute to solving this issue in the future.

The use of net gain analysis might be an interesting additional approach to determine clinical relevance in the registration and clinical prescription of drugs. However, we are not the ﬁrst to question the validity and usefulness of the current operationalization of clinical relevance in terms of responder analysis. Some have criticized the arbitrary choice of the cut-off point of response (7, 24, 25), whereas others have mentioned that dichotomization results in the loss of information and statistical power, and an increase in the probability of false positive results. (26) Furthermore, a number of alternative approaches like the Lehmann Alternative, Shifted Responders and the Relative Effect have been proposed. (7, 12, 13) However, none of these alternatives have taken the risk on differential (compound speciﬁc) deterioration into account.

Net gain analysis might only be relevant for disorders and medications with relatively low success rates, and in disorders with a type of deterioration that causes great mental, physical and/or social harm, such as in the acute treatment of mania. Furthermore, net gain analysis could also be an interesting approach when assessing the clinical relevance of interventions in terms of quality of life, where the probability of deterioration due to differences in side effects between the active drug and the placebo can easily result in an improvement in one domain (e.g. physical health) at the expense of a deterioration in another domain (e.g. mental health).

PART 2 - CHAPTER 4 - Net gain analysis, an addition to responder analysis| 73 Our study has several strengths. We were able to compare the clinical relevance of five different antipsychotics compared to placebo. This also enabled us to compare the different antipsychotics among themselves using individual patient data. However, our study also has some limitations. First, our study was based on post-hoc analyses. Second, both the cut-off points for defining response and deterioration were arbitrarily chosen. Especially the cut-off point of deterioration was based on a very small exiting literature and on expert opinion. We chose for the 10% cut-off point taking into account the natural course of the acute mania and that an increase of 10% or more in the acute phase can, according to experts, be considered as serious deterioration, hereby excluding random variation, measurement error and subclinical deterioration. However, in a sensitivity analysis using to other cut-off points (≥0% increase and ≥20% increase: Baker et al. (27), very similar results were obtained. (See supplementary material) Another limitation is that net gain is only interpretable at study-level and not at the individual patient level, i.e. there is no such thing as a number needed to gain that is comparable to the number needed to treat (NNT). Therefore, the net gain analysis should preferably be used by the regulatory authorities, to assess medication on efficacy, and by expert-groups for the development of treatment guidelines.

CONCLUSION

The newly developed net gain analysis is a more comprehensive approach than the traditional responder analysis to establish clinical relevance in medications that are statistically signiﬁcantly more effective than placebo. However, in the case of antipsychotic treatment of acute mania, the results in terms of net gain analysis were only marginally different from the results in terms of responder analysis. This does not have to be the case, however, for other medications, other outcomes, and other diseases.

RECOMMENDATION

It is recommended to investigate in future studies whether the net gain analysis is an interesting addition to responder analysis. These studies should not be restricted acute mania, but to all acute psychiatric and somatic disorders with low to moderate success rates and with a substantial risk of deterioration with serious harm. Special attention should be given to the new analysis when assessing the quality of life, where a more comprehensive model as the net gain analysis aligns with the treatment goal.

74 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success REFERENCES

1. Uryniak T, Chan ISF, Fedorov VV, Jiang Q, Oppenheimer L, Snapinn SM, Teng C, Zhang J. Responder Analyses—A PhRMA Position paper American Statistical Association Statistics in Biopharmaceutical. 2011;3. 2. ICH: Harmonized Tripartite Guideline Statistical Principles for Clinical Trials: E9 in International Conference on Harmonisation1998. 3. CHMP: Guideline on the choice of the non-inferiority margin. Edited by Agency EM2005. 4. FDA: Guidance for Industry Patient-Reported Outcome Measures: Use in Medical Product Development to Support Labeling Claims. Edited by Services USDoHaH2009. 5. CHMP: Guideline on clinical investigation of medicinal products indicated for the treatment of social anxiety disorder (SAD). Committee for Medicinal Products for Human Use 2006. 6. Broich K, CHMP. Committee for Medicinal Products for Human Use (CHMP) assessment on efficacy of antidepressants. European Neuropsychopharmacology. 2009;19:305-308. 7. Kieser M, Friede T, Gondan M. Assessment of statistical significance and clinical relevance. Statistics in Medicine. 2013;32:1707-1719. 8. Senn S, Julious S. Measurement in clinical trials: A neglected issue for statisticians? Statistics in Medicine. 2009;28:3189-3209. 9. Snapinn SM, Jiang Q. Responder analyses and the assessment of a clinically relevant treatment effect. Trials. 2007;8:31. 10. Kieser M, Rohmel J, Friede T. Power and sample size determination when assessing the clinical relevance of trial results by ‘responder analyses’. Statistics in Medicine. 2004;23:3287–3305. 11. CHMP: Note for guidance on clinical investigation of medicinal product of the treatment and prevention of bipolar disorder. . Edited by Use CfMPfH2001. 12. Jones PW. Interpreting thresholds for a clinically significant change in health status in asthma and COPD. . European Respiratory Journal. 2002;19:396-404. 13. Victor N. Clinically relevant differences and shifted null hypotheses. Methods of Information in Medicine 1987;26:109-116. 14. Yildiz A, Vieta E, Tohen M, Baldessarini RJ. Factors modifying drug and placebo responses in randomized trials for bipolar mania. International Journal of Neuropsychopharmacology. 2011;14:863- 875. 15. Young RC, Biggs JT, Ziegler VE, Meyer DA. A rating scale for mania: reliability, validity and sensitivity. British Journal of Psychiatry. 1978;133:429-435. 16. Wolf FM: Meta-Analysis - Quantitatieve Methods for Research Synthesis. Newbury Parkc, SAGE Publications, Inc 1986. 17. Perlis RH, Welge JA, Vornik LA, Hirschfeld RM, Keck PEJ. Atypical antipsychotics in the treatment of mania: a meta-analysis of randomized, placebo-controlled trials. Journal of Clinical Psychiatry. 2006;67:509-516. 18. Scherk H, Pajonk FG, Leucht S. Second-generation antipsychotic agents in the treatment of acute mania: a systematic review and meta-analysis of randomized controlled trials. Archives of General Psychiatry. 2007;64:442-455. 19. Smith LA, Cornelius V, Warnock A, Tacchic MJ, Taylor D. Pharmacological interventions for acute bipolar mania: a systematic review of randomized placebo-controlled trials. Bipolar Disorders. 2007;9:551-560. 20. Yildiz A, Nikodem M, Vieta E, Correll CU, Baldessarini RJ. A network meta-analysis on comparative

PART 2 - CHAPTER 4 - Net gain analysis, an addition to responder analysis| 75 efficacy and all-cause discontinuation of antimanic treatments in acute bipolar mania. Psychological Medicine. 2014;18:1-19. 21. Cipriani A, Barbui C, Salanti G, Rendell J, Brown R, Stockton S, Purgato M, Spineli LM, Goodwin GM, Geddes JR. Comparative efficacy and acceptability of antimanic drugs in acute mania: a multiple- treatments meta-analysis. Lancet. 2011;378:1306-1315. 22. Goodwin GM. Evidence-based guidelines for treating bipolar disorder: revised second edition - recommendations from the British Association for Psychopharmacology. Journal of Psychopharmacology. 2009;23. 23. Tamayo JM, Zarate CAJ, Vieta E, Vázquez G, Tohen M. Level of response and safety of pharmacological monotherapy in the treatment of acute bipolar I disorder phases: a systematic review and meta-analysis. International Journal of Neuropsychopharmacology. 2010;13:813-832. 24. Royston P, Altman DG, Sauerbrei W. Dichotomizing continuous predictors in multiple regression: a bad idea. Statistics in Medicine. 2006;25:127-141. 25. Senn S. Disappointing Dichotomies. Pharmaceutical Statistics 2003;2:239-240. 26. Austin PC, Brunner LJ. Inflation of the type I error rate when a continuous confounding variable is categorized in logistic regression analyses. Statistics in Medicine. 2004;23:1159-1178. 27. Baker RW, Milton DR, Stauffer VL, Gelenberg A, Tohen M. Placebo-controlled trials do not find association of olanzapine with exacerbation of bipolar mania. Journal of Affective Disorders. 2003;73:147-153.

76 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success Table S1 Deterioration (≥0%).

c b b

12 value - 0.001 0.0 0.593 p

4 1 3 E 5.7 15.2 15.2 46.9 29.1 17.8 20.9 17.8 33.1 - - 0.001 <0.001 <0.001 0.250 ; 0.412

2 2 2 D 9.0 14.7 14.7 50.6 29.0 21.6 23.7 21.6 36.3 - - <0.001 <0.001 <0.001 0.293 ;0.433

3 3 4 C 9.0 13.3 13.3 46.0 29.6 16.4 22.3 16.4 29.7 - - <0.001 <0.001 <0.001 0.234 ; 0.360

. 5 4 5 B 6.8 6.8 7.5 5.4 7.5 - - 47.8 40.3 12.2 14.3 0.045 <0.001 <0.001 0.095 ; 0.191

(20)=102.3 2

;

1 5 1 A 15.7 15.7 53.6 27.1 26.5 20.0 35.7 26.5 42.2 0.001 - - 0.006 <0.001 < Compound 0.284 ; 0.560

multinomial logistic regression.

treatment placebo

multilevel logistic regression.

a a CI Responders Deteriorators Responders Deteriorators - - - - - ased on multilevel ased on Fisher’s combined test Responders Responders - - value value value % % % %

- - - ) B ) Based on ) B Det≥0% % % ∆ p Ranking (RA) % Deteriorators % Deteriorators ∆ p Ranking (DA) ∆ ∆ Net Gain (%) 95% p Ranking (NGA) a b c

PART 2 - CHAPTER 4 - Net gain analysis, an addition to responder analysis| 77 Table S2 Deterioration (≥20%).

c b b

12 value - 0.0 0.835 p <0.001

4 1 3 E 9.8 9.8 3.3 - - 46.9 29.1 17.8 13.1 17.8 27.7 0.001 0.001 <0.001 0.201 ; 0.353

2 3 2 D 7.2 7.2 4.2 - - 50.6 29.0 21.6 11.4 21.6 28.8 0.001 0.001 < <0.001 0.226 ; 0.350

3 4 4 C 6.8 6.8 4.5 - - 46.0 29.6 16.4 11.3 16.4 23.2 0.001 <0.001 <0.001 0.175 ; 0.289

. 5 2 5 B 4.0 4.0 7.5 1.6 5.6 7.5 - - 47.8 40.3 11.5 0.045 0.001 0.002 0.072 ; 0.158

(20)=119.4 2

;

1 5 1 A 8.3 8.3 7.2 - - 53.6 27.1 26.5 15.5 26.5 34.8 0.042 <0.001 <0.001 0.238 ; 0.458 Compound ultinomial logistic regression.

treatment placebo

multilevel logistic regression.

a a CI Responders Deteriorators Responders Deteriorators - - - - - ased on multilevel m ased on Fisher’s combined test Responders Responders - - value value value % % % %

- - - ) B ) Based on ) B Det≥20% % % ∆ p Ranking (RA) % Deteriorators % Deteriorators ∆ p Ranking (DA) ∆ ∆ Net Gain (%) 95% p Ranking (NGA) a b c

78 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success Box S.1 Formulas for net gain analysis.

PART 2 - CHAPTER 4 - Net gain analysis, an addition to responder analysis| 79

PARTquestions 3Clinical

05 CHAPTER 5: DOES INSIGHT AFFECT THE EFFICACY OF ANTIPSYCHOTICS IN ACUTE MANIA?

AN INDIVIDUAL PATIENT DATA REGRESSION META-ANALYSIS

Accepted for Journal of Clinical Psychopharmacology

Patients having an acute manic episode of bipolar disorder often lack insight into their condition. Since little is known about the possible effect of insight on treatment efficacy, we examined whether insight at the start of treatment affects the efficacy of antipsychotic treatment in patients with acute mania. We used individual patient data from seven randomized, double-blind, placebo-controlled registration studies of four antipsychotics in patients with acute mania (N=1904). Insight was measured with item 11 of the Young Mania Rating Scale (YMRS) at baseline and study endpoint 3 weeks later. Treatment outcome was defined by (a) mean change score (MCS), (b) response defined as ≥50% improvement on YMRS, and (c) remission defined as YMRS score <8 at study endpoint. We used multilevel mixed effect linear (or logistic) regression analyses of individual patient data (IPD) to assess the interaction between baseline insight and treatment outcomes. At treatment initiation, 1207 (63.5%) patients had impaired or no insight into their condition. Level of insight significantly modified the efficacy of treatment by MCS (p=0.039), response rate (p=0.033), and remission rate (p=0.043), with greater improvement in patients with more impaired insight. We therefore recommend that patients experiencing acute mania should be treated immediately and not be delayed until patients regain insight.

84 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success INTRODUCTION

Insight into one’s illness is an important, but complex, phenomenon in patients with a psychiatric disorder (1-3). Despite many changes in the concept over the last decades (4), insight is now generally deﬁned as (a) recognition that one has a mental disorder, (b) ability to recognize symptoms, (c) and understanding of the need for treatment (2, 5). Insight is often impaired in patients with severe mental illnesses, such as bipolar disorder and schizophrenia (1, 6-8). In patients with bipolar disorder, insight is most impaired during an acute manic episode (9). This could be a problem because acute mania can cause serious harm to patients and their environment, especially when not treated adequately (10). Insight might have an impact on treatment acceptance and adherence during the acute phase of the illness, as well as on adherence to prophylactic treatment and recognition of early warning signs of relapse during periods of remission (11).

In patients with bipolar disorder in remission, lack of insight is associated with low medication adherence (8) and poor clinical outcomes (7, 12, 13). In patients with an acute manic episode with psychotic features, Smith et al. (2014) recently found that there were no signiﬁcant differences in psychosocial and clinical outcomes after 18 months between patients with full, partial, or no insight (14). However, the short-term effect of insight in patients having an acute manic episode has been studied less, and the effect of insight on the treatment of patients with acute mania has never been studied.

Therefore, the aim of this study was to investigate whether insight at treatment initiation inﬂuences the efﬁcacy of antipsychotic treatment in patients with acute mania, evaluated 3 weeks after treatment initiation.

METHODS

Selection of studies We pooled the individual patient data of all short-term efﬁcacy studies that assessed the antipsychotic treatment of acute mania and that had been submitted to the Dutch Medicines Evaluation Board (CBG-MEB) during an 11-year period as part of market authorization application for the indication acute manic episode of bipolar disorder. All studies were double blind, randomized, placebo-controlled trials including patients diagnosed with and Diagnostic and Statistical Manual of Mental Disorders IV (DSM-IV)

PART 3 - CHAPTER 5 - Does insight affect the efﬁcacy of antipsychotics in acute mania? | 85 acute manic episode of bipolar disorder. The pharmaceutical companies conducting these studies provided their raw patient data. When individual item scores were not available, the study was excluded.

Assessments The seven studies investigated four antipsychotic compounds. Active antipsychotic comparators were also included and analysed as active treatment. Six studies examined the effect of antipsychotic mono-therapy and one study used an add-on, placebo- controlled design. All studies were conducted in hospitalized inpatients (at least for the ﬁrst 7 days). We restricted the analyses to the patients who were prescribed the medication in an effective dose according to the Summary of Product Characteristics (SmPC) if the drug had been granted a licence for the treatment of the acute manic episode; if the drug had not been granted a licence for this indication, then expert consensus established whether a dose was effective, based on the doses mentioned in SmPCs for related disorders.

We used the Young Mania Rating Scale (YMRS), an interview-based questionnaire, to assess insight into and the severity of the acute manic episode of bipolar disorder. The YMRS comprises 11 items: 7 items are scored on a 0-4 scale and 4 are scored on a 0-8 scale. Total scores range from 0 (no symptoms) to 60 (severe symptoms). We used item 11 ‘Insight’ of the YMRS questionnaire to study insight, describing insight on a 0-4 scale. A score of 0 is described as ‘excellent insight’ (patient admits illness, agrees with need for treatment), a score of 1 as ‘good insight’ (patient admits to being possibly ill), a score of 2 as ‘moderate insight’ (patient admits behaviour change, but denies illness), a score of 3 as ‘poor insight’ (patient admits possible change in behaviour, but denies illness), and a score of 4 as ‘no insight’ (patient denies any behaviour change) (15).

Outcome measures In order to prevent item overlap between the predictor and the outcome measure, we used the following three efficacy outcomes based on the YMRS total scores at baseline and follow-up after removal of YMRS item about insight: (1) the standardized mean change score on the YMRS from baseline to study endpoint, (2) the percentage of responders, with response being defined as a decrease ≥ 50% on the YMRS from baseline to study endpoint, and (3) the percentage of patients in remission, defined

86 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success as a YMRS score <8 at study endpoint. The cut-off scores for response and remission were based on the Task Force Report of the International Society for Bipolar Disorders (ISBD), since there is no consensus in the EMA Committee for Proprietary Medicinal Products (CPMP) guideline on the clinical investigation of medicinal products for the treatment and prevention of bipolar disorder (16, 17).

The study endpoint was 3 weeks after the baseline measurement, since this is the time point recommended in the EMA guideline on the clinical investigation of bipolar disorder for short-term efﬁcacy studies in the acute manic episode (17). If outcome data at week 3 were missing, we used week 4 data in studies that lasted longer than 3 weeks, and if data at week 4 were not available, the last observation carried forward (LOCF) procedure was used.

STATISTICAL ANALYSIS

To assess whether insight modiﬁes the effect of antipsychotics compared with placebo on the mean change in YMRS scores from baseline to endpoint, we performed a multilevel mixed effect linear regression analyses with a random intercept for study. To assess whether insight modiﬁes the effect of treatment on response and remission rates, we performed two multilevel mixed effect logistic regression analyses with a random intercept for study, using a likelihood ratio test to investigate the interaction of treatment group and level of insight.

For all analyses we used ‘excellent insight’ (score 0 on item 11) and placebo arm as reference groups. We adjusted all ﬁndings for age, body mass index (BMI), gender, and illness severity at baseline. Illness severity and insight scores at weeks 1 and 2 were calculated using SPSS version 20 (SPSS20). All multilevel mixed effect linear regression analyses for mean change score were performed with SPSS20, whereas all multilevel mixed effect logistic regression analyses for response and remission were performed with the xtmixed and xtmemixed programs of STATA 12.

PART 3 - CHAPTER 5 - Does insight affect the efﬁcacy of antipsychotics in acute mania? | 87 Table 1. Patient and study characteristics per study (N=1904).

Trial Name Nr. Patient Studied Active Age BMI Female Ethnicity YMRS baseline Treatment/ compound Comparator Mean (SD) Mean (SD) (Fraction) Mean (SD) AC2/Placebo R076477-BIM-3001 234/122 Paliperidone ER - 39.30 (11.15) 28.21(6.73) 0.46 Caucasians 48.6%, Blacks 26.1%, 28.29 (5.30) Asians 24.4%, Other 0.8% R076477-BIM-3002 195/193/105 Paliperidone ER Quetiapine 39.27 (10.91) 27.90 (6.61) 0.43 Caucasians 68.4%, Blacks 20.9%, 27.24 (5.00) Asians 9.7%, Other 1.0% R076477-BIM-3003 150/150 Paliperidone ER - 40.24 (11.01) 27.55 (6.30) 0.46 Caucasians 76.3%, Blacks 19.3%, 26.68 (5.25) Asians 3.7%, Other 0.7% F1D-MC-HGGW 70/69 Olanzapine - 39.48 (10.93) 0.48 Caucasians 72.7%, Blacks 20.1%, 28.20 (6.53) Other 7.2% F1D-MC-HGEH 55/60 Olanzapine - 38.63 (10.36) 0.50 Caucasians 80.0%, Blacks 13.9%, 29.30 (7.00) Other 6.1% 5077IL/0104 100/98/100 Quetiapine Haloperidol 42.90 (12.84) 25.63 (4.58) 0.63 Caucasians 74.2%, Asians 10.4%, 33.01 (6.23) Other 15.4% 5077IL/0105 107/96 Quetiapine - 39.44 (13.08) 23.42 (5.54) 0.43 Caucasians 51.7%, Asians 48.3% 33.28 (6.68)

Tot 1202/702 =1904 NA NA 39.99 (11.56) 26.94 (6.33) 0.48 Caucasians 66.1%, Blacks 15.7%, 29.09 (6.25) Asians 14.4%, Other 3.8% 1) Active comparator in study; NA = Not Applicable

RESULTS

Study characteristics

The data of 1904 patients from seven studies were analysed: 1202 patients received antipsychotic treatment and 702 patients received placebo. Excluding the add-on study from the analyses did not change the results. At baseline, the treatment group had a mean YMRS score of 29.1, a mean age of 40.0 years, and consisted of 47.8% females. A total of 696 patients (36.5%) had ‘excellent insight’, 359 (18.9%) were ‘good insight’, 333 (17.5%) had ‘moderate insight’, 316 (16.6%) had ‘poor insight’, and in 200 (10.5%) patients ‘no insight’. Thus insight was impaired or absent in 63.5% of the patients. Table 1 presents the patient and study characteristics for each study.

Potential confounders We adjusted all findings for age, BMI, gender, and mania severity at baseline in the multilevel mixed model linear and logistic regression analyses. Age (p=0.341) and gender (p=0.178) were not significantly associated with insight at baseline, whereas BMI and illness severity at baseline were significantly associated with insight at baseline (<0.001): patients with impaired or no insight had a lower BMI and more severe disease than did patients with excellent insight. None of the above variables modified the effect of baseline insight on treatment outcomes (all p>0.05).

EFFECT OF INSIGHT AT BASELINE ON THE EFFECT OF ANTIPSYCHOTIC TREATMENT OF MANIC SYMPTOMS

Mean change score Baseline insight signiﬁcantly modiﬁed the effect of antipsychotic treatment on the mean

88 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success Table 2. Effect of insight (4 levels) at baseline on the mean change score, response, and remission rate in symptom severity, adjusted for age, BMI, gender, and severity at baseline (N=1904).

Parameter b1 SE2 OR3 95%-CI p-value Mean change score Intercept -2.637 2.013 -6.592; 1.3194 0.191 Insight 1 at baseline5 0.062 1.184 -2.261; 2.3854 0.958 Insight 2 at baseline5 2.460 1.191 0.123; 4.7974 0.039 Insight 3 at baseline5 2.257 1.172 -0.042; 4.5564 0.054 Insight 4 at baseline5 2.234 1.379 -0.472; 4.9404 0.106 Study group (treatment)6 -2.097 0.861 -3.786; -0.4084 0.015 Insight 1 * treatment -1.446 1.440 -4.271; 1.3804 0.316 Insight 2 * treatment -3.083 1.430 -5.888; -0.2774 0.031 Insight 3 * treatment -3.992 1.437 -6.810; -1.1734 0.006 Insight 4 * treatment -3.108 1.658 -6.359; -0.1444 0.061 Age7 0.004 0.020 -0.036; 0.0434 0.857 BMI7 0.082 0.040 0.004; 0.1604 0.040 Gender (female)7 -0.551 0.478 -1.489; 0.3864 0.249 Severity at baseline7 -0.360 0.045 -0.448; -0.2734 <0.001 Overall interaction of insight score at baseline and study group on mean change score; p=0.039, F=2.534 Response rate Intercept 1.077 0.419 2.934 1.292; 6.667 0.010 Insight 1 at baseline5 -0.178 0.254 0.837 0.509; 1.377 0.484 Insight 2 at baseline5 -0.268 0.257 0.765 0.463; 1.265 0.297 Insight 3 at baseline5 -0.207 0.252 0.813 0.496; 1.333 0.411 Insight 4 at baseline5 -0.351 0.303 0.704 0.389; 1.276 0.247 Study group (treatment)6 0.054 0.182 1.055 0.738; 1.509 0.768 Insight 1 at baseline * treatment 0.631 0.308 1.880 1.029; 3.435 0.040 Insight 2 at baseline * treatment 0.573 0.309 1.774 0.968; 3.251 0.064 Insight 3 at baseline * treatment 0.806 0.312 2.238 1.215; 4.124 0.010 Insight 4 at baseline * treatment 0.867 0.369 2.380 1.155; 4.903 0.019 Age7 0.002 0.004 1.002 0.994; 1.011 0.639 BMI7 -0.017 0.009 0.983 0.967; 1.000 0.049 Gender (female)7 0.164 0.102 1.178 0.964; 1.439 0.109 Severity at baseline7 -0.040 0.009 0.961 0.944; 0.979 <0.001 Overall interaction of insight score at baseline and study group on response rate; p=0.033, LRχ2 =10.49 Remission rate Intercept 2.257 0.516 9.555 3.476; 26.269 <0.001 Insight 1 at baseline5 -0.308 0.320 0.735 0.392; 1.377 0.0336 Insight 2 at baseline5 -0.280 0.318 0.756 0.405; 1.411 0.379 Insight 3 at baseline5 0.037 0.310 1.038 0.565; 1.905 0.905 Insight 4 at baseline5 -0.074 0.397 0.929 0.437; 2.021 0.852 Study group (treatment)6 0.077 0.230 1.080 0.688; 1.696 0.737 Insight 1 at baseline * treatment 0.890 0.378 2.435 1.162; 5.103 0.018 Insight 2 at baseline * treatment 0.578 0.375 1.783 0.855; 3.717 0.123 Insight 3 at baseline * treatment 0.588 0.374 1.800 0.866; 3.744 0.115 Insight 4 at baseline * treatment 1.226 0.464 3.407 1.372; 8.455 0.008 Age7 <-0.001 0.005 1.000 0.990; 1.010 0.980 BMI7 -0.023 0.010 0.977 0.958; 0.997 0.027 Gender (female)7 0.188 0.120 1.207 0.953; 1.528 0.118 Severity at baseline7 -0.111 0.012 0.895 0.874; 0.917 <0.001 Overall interaction of insight score at baseline and study group on response rate; p=0.043, LRχ2 =9.83 1) b = regression coefficient of multilevel mixed effect linear regression analyses 2) SE = Standard error of b 3) OR = Odds ratio of multilevel mixed effect logistic regression analyses 4) 95%-CI of b 5) Reference group = excellent insight (score 0); insight 1 (score 1) etc at baseline 6) Reference group = placebo group 7) Correction factors 8) 95%-CI of OR

PART 3 - CHAPTER 5 - Does insight affect the efﬁcacy of antipsychotics in acute mania? | 89 change score (p=0.039, F=2.534). The mean change scores of patients with excellent insight or impaired or no insight were (borderline) signiﬁcantly different (p=0.031, p=0.006, p=0.061), with patients with impaired or no insight showing a greater improvement from baseline than patients with excellent insight. Compared with excellent insight, the adjusted effect (b) (SE, 95%-CI) of baseline insight on the effect of treatment on the mean change score was –3.108 (SE=1.658, 95%-CI -6.359; -0.144) for no insight (score 4), -3.992 (SE=1.437, 95%-CI -6.810; -1.173) for poor insight (score 3), -3.083 (SE=1.430, 95%-CI -5.888; -0.277) for moderate insight (score 2), and -1.446 (SE=1.440, 95%-CI -4.271; 1.380) for good insight (score 1) at baseline (Table 2).

Response In the logistic regression analysis, baseline insight significantly modified the effect of antipsychotic treatment on the response rate (p=0.033, LR χ²(4)=10.49). The response rate of patients with good, impaired, or no insight were all significantly higher than in patients with excellent insight (p=0.040, p=0.064, p=0.010 p=0.019, respectively). Compared with excellent insight, the adjusted odds ratio (OR) (95%-CI) of baseline insight on the effect of treatment on response rate was 2.380 (95%-CI 1.155; 4.903) for no insight (score 4), 2.238 (95%-CI 1.215; 4.124) for poor insight (score 3), 1.774 (95%-CI 1.215; 4.124) for moderate insight (score 2), and 1.880 (95%-CI 1.029; 3.435) for good insight (score 1) (Table 2).

Remission In the logistic regression analysis for remission, baseline insight significantly modified the effect of antipsychotic treatment on the remission rate (p=0.043, LR χ²(4)=9.83). The remission rate was significantly (p<0.05) higher in patients with good or no insight versus patients with excellent insight. Compared with excellent insight at baseline, the adjusted effect (OR; 95-CI) for baseline insight on the effect of treatment by remission rate was 3.407 (95%-CI 1.372; 8.455) for patients with no insight (score 4) and 2.435 (95%-CI 1.162; 5.103) for patients with good insight (score 1) (Table 2).

SENSITIVITY ANALYSIS

To test whether the type of antipsychotic inﬂuenced results, we used multilevel mixed effect (logistic) regression analyses with a random intercept for study and assessed the

90 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success three-way interaction of treatment by antipsychotic compound by insight at baseline on study outcomes. Type of antipsychotic had signiﬁcantly moderated the effect on the inﬂuence of insight on mean change score (p=0.0424, LRχ2 (12)=21.59) and response rate (p=0.008, LRχ2 (12)=27.10) but not on the effect of insight on remission (p=0.075, LRχ2 (12) =19.63).

As there are substantial differences across geographic regions in the efficacy of antipsychotics in patients with acute mania (29), we performed stratified analyses for patients from different regions (Europe, USA, and the rest of the world) to examine whether region modified the effect of baseline insight on the effect of antipsychotic treatment. After this stratification by region, type of antipsychotic no longer significantly (p>0.05) influenced the effect of treatment on MCS, response, or remission.

DISCUSSION

We investigated whether insight at the start of treatment affects the efficacy of antipsychotics in patients with acute mania. Unexpectedly, we found that antipsychotics were more effective in patients with impaired or no insight than in patients with excellent insight. To our knowledge, this is the first study to investigate the impact of insight on the efficacy of antipsychotics in patients with an acute manic episode. Earlier studies of insight in patients with bipolar disorder focused on patients with bipolar disorder in remission, and in these patients impaired insight was found to result in a poorer clinical outcome – reduced medication adherence (8), neurocognitive dysfunction (18, 19), and increased the risk of hospitalization, emergency room visits, and suicidal and violent events (13). None of these studies focused on the effect of insight on the efficacy of treatment. We found that impaired insight was significantly associated with a greater reduction in symptoms (YMRS), response, and remission in patients treated with antipsychotics. Smith et al. (2014) studied patients with a first episode of mania with psychotic features and investigated whether poor insight during a first episode was predictive of poor psychosocial and clinical outcomes after 18 months. They stratified patients into three groups based on the insight item from the YMRS questionnaire: full insight (score 0-1), impaired insight (score 2-3), and absent insight (score 4). Even though they found significant differences between the insight groups at baseline, there were no significant differences in psychosocial and clinical outcomes between the insight groups

PART 3 - CHAPTER 5 - Does insight affect the efficacy of antipsychotics in acute mania? | 91 at 18 months (14). These findings and our findings suggest that the level of insight in the acute episode is an important determinant of the short-term treatment response but may be less relevant for the long-term clinical outcome.

In the past, impaired insight was thought to be related to psychosis. Several studies have reported that insight remains impaired when psychosis improves (20, 21) and that insight is impaired in psychotic and non-psychotic manic patients (22). Over the years, the theory evolved that insight is, at least partially, an independent phenomenon (18) and considered state dependent in patients with bipolar disorder. Our ﬁndings support this theory (22, 23): some patients still showed impaired or no insight even though they responded to treatment (8.9% in patients with ‘moderate or poor’ insight and 1.5% ‘no insight’) or achieved remission (3.3% in patients with ‘moderate or poor insight’ and 0.3% ‘no insight’). This is in line with the ﬁndings of Peralta and Cuesta (1998), who found residual impairments of insight in some patients with mania at hospital discharge (24).

As insight is partially state dependent in patients with a bipolar disorder (22, 23), insight during an acute manic episode might be different from insight during remission. This may result in different treatment recommendations. For instance, in the treatment of patients with bipolar disorder in remission, studies recommend special treatment strategies for patients with impaired insight (psycho-education and psychotherapy) because improvement of insight could be a promising target to improve clinical outcomes. However, in an acute manic episode, rapid improvement of symptoms is required and therefore psycho-education and psychotherapy may not be the ﬁrst treatment option, and especially not in patients with impaired or no insight. In these patients, treatment should primarily rely on medication (25). Nevertheless, since some patients show impaired or no insight at hospital discharge (24), psychosocial interventions may be particularly useful after an acute manic episode (9, 24).

We did not find age to be significantly associated with insight. Information in the literature is inconsistent for patients with bipolar disorder. Smith et al. (2014) and Cassidy et al. (2010) found no correlation between age and lack of insight in patients in an acute manic episode (11, 14), whereas Braw et al. (2012) found that age was significantly related with the level of insight in schizophrenic and bipolar patients in

92 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success remission (7). In relation to these studies, our ﬁndings suggest that the association between age and insight might be different in patients with bipolar disorder during an acute manic episode and during remission.

This study had some limitations. First, the type of antipsychotic used influenced the results. Since we could not find a logical explanation for this finding, we investigated whether this effect was influenced by geographic differences in the efficacy of antipsychotics (26). After stratification for geographic region, type of antipsychotic no longer modified the effect of insight on treatment outcomes. Second, our study was a secondary analysis of data originally gathered to answer a different research question. As a consequence, we were restricted in our analyses by the variables that were available in all studies and were not able to examine the potential confounding effect of treatment adherence on outcome. This resulted in a third limitation,namely, that insight was assessed with just one question of the YMRS (15). Unfortunately, we could not use a more comprehensive assessment scale, such as the Scale to Assess Unawareness of Mental Disorders (SUMD) or the Schedule for Assessment of Insight (SAI), to provide a more complete picture of insight (20, 21). Nevertheless, the YMRS questionnaire is a commonly used to assess acute mania in clinical practice, whereas the SUMD or the SAI assessment scales are typically used for research purposes. Therefore, with regard to clinical applicability, the YMRS questionnaire used in this study could be considered more clinically relevant. Lastly, when interpreting our findings, it must be remembered that we studied insight in patients who agreed to participate in a clinical trial. We found several patients to lack insight, which is surprising because these patients agreed to participate in the trial, showed up at their appointment with the clinician, and may have taken their medication as prescribed. One must therefore question whether these patients are a valid reflection of a patient population in routine clinical practice.

Our study had several strengths. We were able to assess the individual patient data for 1904 patients from seven randomized, double-blind, placebo-controlled studies. To our knowledge, this is a uniquely large sample for assessing insight in patients with bipolar disorder. Patients were treated with four antipsychotics and analyses showed that our findings could not be attributed to differences in the antipsychotic used. Therefore, our findings should be interpreted as an antipsychotic class-effect rather than specific compound effect. Lastly, as previously described, most earlier studies have focused on

PART 3 - CHAPTER 5 - Does insight affect the efﬁcacy of antipsychotics in acute mania? | 93 insight in patients with bipolar disorder in remission, and this is the study of the effect of insight on the outcome of antipsychotic treatment in patients in an acute manic episode.

In conclusion, insight into one’s condition at treatment initiation significantly modifies the efficacy of antipsychotic treatment in patients with acute mania, with treatment being more effective in patients with impaired or no insight than in patients with excellent insight. We therefore recommend that adequate antipsychotic treatment be given to patients with impaired or no insight during an acute manic episode and that treatment initiation should not be delayed until patients gain insight. However, because there are no comparable studies, future prospective studies assessing the effect of insight on the treatment of acute mania should use more comprehensive questionnaires in order to confirm our findings and to support the implications of these findings.

94 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success REFERENCES

1. Trevisi M, Talamo A, Bandinelli PL, Ducci G, Kotzalidis GD, Santucci C, Manfredi G, Girardi N, Tatarelli R. Insight and Awareness as Related to Psychopathology and Cognition. Psychopathology. 2012;45:235-243. 2. Amador XF, Flaum M, Andreasen NC, Strauss DH, Yale SA, Clark SC, Gorman JM. Awareness of illness in schizophrenia and schizoaffective and mood disorders. Archives of General Psychiatry. 1994;51:826-836. 3. Bressi C, Porcellana M, Marinaccio PM, Nocito EP, Ciabatti M, Magri L, Altamura AC. The association between insight and symptoms in bipolar inpatients: An Italian prospective study. European Psychiatry. 2012;27:619-624. 4. Markova IS, Berrios GE. The meaning of insight in clinical psychiatry. Britisch Journal of Psychiatry. 1992;160:850-860. 5. David AS. Insight psychosis. Britisch Journal of Psychiatry. 1990;156:798-808. 6. Aspiazu S, Mosquera F, Ibañez B, Vega P, Barbeito S, López P, Ruiz de Azúa S, Ugarte A, Vieta E, González-Pinto A. Manic and depressive symptoms and insight in ﬁrst episode psychosis. Psychiatric Research. 2010;178:480-486. 7. Braw Y, Sitman R, Sela T, Erez G, Bloch Y, Levkovitz Y. Comparison of insight among schizophrenia and bipolar disorder patients in remission of affective and positive symptoms: Analysis and critique. European Psychiatry. 2012;27:612-618. 8. Yen C-F, Chen S-F, Ko C-H, Yeh M-L, Yang S-J, Yen J-J, Huang C-F, Wu C-C. Relationships between insight and medication adherence in outpatients with schizophrenia and bipolar disorder: Prospective study. Psychiatry and Clinical Neurosciences. 2005;59:403-409. 9. Depp CA, Harmell AL, Savla GN, Mausbach BT, Jeste DV, Palmer BW. A prospective study of the trajectories of clinical insight, affective symptoms, and cognitive ability in bipolar disorder. Journal ofAffectiveDisorders. 2014;152-154:250-255. 10. Andrade C. The risk of harm in mania and the very early time course of improvement: important but neglected variables in treatment research. Bipolar Disorders 2004;6:446-447. 11. Cassidy F. Insight in bipolar disorder: relationship to episode subtypes and symptom dimensions. Neuropsychiatric Disease and Treatment. 2010;6:627-631. 12. Yen C-F, Chen C-S, Yeh M-L, Ker J-H, Yang S-J, Yen J-Y. Correlates of insight among patients with bipolar I disorder in remission. Journal of Affective Disorders. 2004;78:57-60. 13. Yen C-F, Chen C-S, Yen J-Y, Ko C-H. The predictive effect of insight on adverse clinical outcomes in bipolar I disorder: A two-year prospective study. Journal of Affective Disorders. 2008;108:121-127. 14. Smith LT, Shelton CL, Berk M, Hasty MK, Cotton SM, Henry L, Daglas R, Gentle E, McGorry PD, Macneil CA, Conus P. The impact of insight in a ﬁrst-episode mania with psychosis population on outcome at 18 months. Journal of Affective Disorders. 2014;167:74-79. 15. Young RC, Biggs JT, Ziegler VE, Meyer DA. A rating scale for mania: reliability, validity and sensitivity. British Journal of Psychiatry. 1978;133:429-435. 16. Tohen M, Frank E, Bowden CL, Colom F, Ghaemi SN, Yatham LN, Malhi GS, Calabrese JR, Nolen WA, Vieta E, Kapczinski F, Goodwin GM, Suppes T, Sachs GS, Chengappa KNR, Grunze H, Mitchell PB, Kanba S, Berk M. The International Society for Bipolar Disorders (ISBD) Task Force report on the nomenclature of course and outcome in bipolar disorders. Bipolar Disorders 2009;11:453-473. 17. Note for guidance on clinical investigation of medicinal product of the treatment and prevention of bipolar disorder. Committee for Medicinal Products for Human Use 2001.

PART 3 - CHAPTER 5 - Does insight affect the efﬁcacy of antipsychotics in acute mania? | 95 18. Varga M, Magnusson A, Flekkøy K, Rønneberg U, Opjordsmoen S. Insight, symptoms and neurocognition in bipolar I patients. Journal of Affective Disorder. 2006;91:1-9. 19. Dias VV, Brissos S, Carita AI. Clinical and neurocognitive correlates of insight in patients with bipolar I disorder in remission. Acta Psychiatrica Scandinavica. 2008;117:28-34. 20. Amador XF, Strauss DH, Yale SA, Flaum MM, Endicott JE, Gorman JM. Assessment of insight in psychosis. American Journal of Psychiatry. 1993;150:873-879. 21. David A, Buchanan A, Reed A, Almeida O. The assessment of insight in psychosis. Britisch Journal of Psychiatry. 1992;161:599-602. 22. Ghaemi SN, Rosenquist KJ. Is Insight in Mania State-Dependent? A Meta-Analysis. Journal of Nervous and Mental Disease. 2004;192:771-775. 23. Yen C-F, Chen C-S, Ko C-H, Yen J-Y, Huang C-F. Changes in insight among patients with bipolar I disorder: a 2-year prospective study. . Bipolar Disorders. 2007;9:238-242. 24. Peralta V, Cuesta MJ. Lack of insight in mood disorders. Journal of Affective Disorder. 1998;49:55- 58. 25. Hirschfeld RMA. Guideline Watch (November 2005): Practice Guideline for the Treatment of Patients with Bipolar Disorder, 2nd Edition. The journal of lifelong learning in psychiatry. 2007;V. 26. Welten CCM, Koeter MWJ, Wohlfarth TD, van den Brink W, Storosum JG, Gispen-de Wied CC, Leufkens HGM, Denys DAJP. Efﬁcacy of drug treatment for acute mania differs across geographic regions: An individual patient data meta-analysis of placebo-controlled studies. Journal of Psychopharmacology.

2015;29:923-932.

96 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success PART 3 - CHAPTER 5 - Does insight affect the efﬁcacy of antipsychotics in acute mania? | 97

06 CHAPTER 6: EARLY NON-RESPONSE IN THE ANTIPSYCHOTIC TREATMENT OF ACUTE MANIA; A CRITERION FOR RECONSIDERING TREATMENT?

RESULTS FROM AN INDIVIDUAL PATIENT DATA META-ANALYSIS

Accepted for Journal of Clinical Psychiatry

Carlijn C.M. Welten, M.D.1,2*, Maarten W.J. Koeter, Ph.D1 , Tamar D. Wohlfarth, Ph.D2, Jitschak G. Storosum, M.D., Ph.D1, Wim van den Brink, M.D., Ph.D1, Christine C. Gispen-de Wied, M.D., Ph.D2, Hubert G.M. Leufkens, Ph.D2, Damiaan A.J.P. Denys, M.D., Ph.D1,3 1Dept. of Psychiatry, Academic Medical Centre, University of Amsterdam, Amsterdam, the Netherlands; 2Medicines Evaluation Board, Utrecht, the Netherlands; 3Netherlands Institute for Neuroscience, Royal Netherlands Academy of Arts and Sciences, Amsterdam, the Netherlands ABSTRACT

OBJECTIVE: To investigated whether early non-response to antipsychotic treatment predicts treatment failure and, if so, to establish the best deﬁnition or criterion of an early non-response.

DATA SOURCES: Registration studies submitted to the Dutch Medicines Evaluation Board (CBG-MEB) as part of market authorization application of their medicine for the indication acute manic episode of bipolar disorder and of which the pharmaceutical companies agreed to provide us the raw individual patient data of all their studies submitted to the CBG-MEB.

STUDY SELECTION: All double blind, randomized, placebo-controlled trials assessing the efﬁcacy of antipsychotics were included (N=10).

DATA EXTRACTION: All patients with data available for completers analysis (N=1243), symptom severity scores on the Young Mania Rating Scale (YMRS) at week 0, 1 and 2 and on study endpoint (week 3 or 4).

RESULTS: The a priori chances of non-response and non-remission at study endpoint were 40.9% (95%-CI 38.2-43.6) and 65.3% (95% CI 62.0-68.6), respectively. Early non-response in weeks 1 and 2, defined by cut-off scores, ranging from less than 10% to less than 50% reduction in symptoms compared to baseline on the YMRS, significantly predicted non-response (less than a 50% symptom reduction) and non-remission (YMRS score higher than 8) in week 3. The predictive value of early non-response (PVnr_se) at week 1 for both non-response and non-remission at study endpoint declined linearly with increasing cut-off score of early non-response; non-response: 76.0 (95%CI 69.7=82.3) for less than 10% response to 48.7 (95%CI 45.5-51.9) for less than 50% response; non remission: 92.2 (95%CI 88.3-96.1) for less than 10% response to 76.8 (95%CI 74.1-79.5) for less than 50% response. A similar linear decline was observed for increasing cut-off scores of early non-response at week 2 for non-response, but not for non-remission at endpoint: non-response 90.3 (95%CI 84.6-96.0) for less than 10% response to 65.0 (95%CI 61.4-68.6) for less than 50% response; non remission: 94.2 (95%CI 89.7-98.7) for less than 10% response and 93.2 (95%CI 93.1-95.1) for less than 50% response. Specific antipsychotic characteristics did not modify these findings at either time point (p=0.127, p=0.213).

CONCLUSION: When patients fail to respond early (1–2 weeks) after the initiation of antipsychotic treatment for acute mania, clinicians should reconsider their treatment choice using a two-stage strategy.

100 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success INTRODUCTION

Acute manic episodes of bipolar disorder cause patients and their surroundings severe emotional turmoil. These episodes may result in excess morbidity and even mortality (1-4). The aim of acute treatment is to rapidly reduce symptom severity in an attempt to counteract these risks. Contrary to what was earlier believed, first- (5-7) and second- generation (8-10) antipsychotics have an anti-manic effect soon after treatment initiation (11). This makes it important to establish whether patients who do not have an early response ultimately respond to treatment. This issue has been investigated in small studies of one (12) or two antipsychotic agents (13, 14). In a pooled post hoc analysis, Szegedi et al. (2013) state that lack of improvement in the first week of treatment was significantly associated with a lack of improvement at study endpoint (14). Kemp et al. (2011) also found that manic patients treated with an antipsychotic without sufficient improvement (< 25% symptom reduction) at week 1 were less likely to reach response or remission by week 3 (12). Similar findings have been reported in patients treated with an anti-depressant for bipolar depression or major depressive disorder (MDD). With regard to bipolar depression, Kemp et al. (2010) found that the absence of early improvement was a highly reliable predictor of eventual non-response. For MDD, Nierenberg and colleagues already found in 1995 that absence of early improvement after 3 weeks of treatment reliably predicted non-response and non-remission at study endpoint (15). This result was, recently confirmed by Kudlow et al. (2014), who found a high negative predictive value of non-response at week 2-4 for treatment success at endpoint (16). Due to limited knowledge about the early response (or lack thereof) to antipsychotics, current guidelines do not provide consensus about how and when to determine whether treatment for acute mania is effective or not (17). Most guidelines propose a period of 2 weeks in which to decide whether to interrupt, switch, or continue current medication (17, 18), although a 1-week period has also been found to be long enough to assess whether a patient responds to treatment (12, 14). Surprisingly, the NICE Guideline on Bipolar Disorders (2014) does not propose a time period at all (19). To increase our knowledge on how to predict whether patients with acute mania will respond to treatment, the aims of this study were to investigate whether an early non-response to antipsychotic treatment is predictive of a later lack of response or remission and, if so, which criterion of an early non-response is the best predictor of later treatment failure.

PART 3 - CHAPTER 6 - Early non-response in the antipsychotic treatment of acute mania| 101 METHODS

Selection of studies We included all short-term efﬁcacy studies assessing antipsychotics that were submitted to the Dutch Medicines Evaluation Board (CBG-MEB) during an 11-year period as part of market authorization application for the indication acute manic episode of bipolar disorder. All studies were double-blind, randomized, placebo-controlled trials involving patients with a DSM-IV acute manic episode of bipolar disorder. Pharmaceutical companies provided their raw patient data, which enabled us to perform an individual patient data meta-analysis.

The studies investigated ﬁve different antipsychotics; active antipsychotic comparators were included and analysed as treatment. In order to protect the interests of participating companies, no drug names are mentioned. We restricted the analyses to the data of patients who were prescribed medication in an effective dose according to the Summary of Product Characteristics (SmPC) if the drug had been granted a licence for the treatment of an acute manic episode; if the drug had not been granted a licence for this indication, expert consensus was established on the effective dose, mainly based on the doses mentioned in SmPCs for related disorders.

Assessments The Young Mania Rating Scale (YMRS), an interview-based questionnaire, was used to assess the severity of the acute manic episode of bipolar disorder. The YMRS comprises 11 items: 7 items are scored on a 0-4 scale and 4 items are scored on a 0-8 scale. Total scores range from 0 (no symptoms) to 60 (most severe symptoms) (20).

To identify the best definition of early non-response and when to assess it, we used different cut-off scores for non-response (≤10%, 15%, 20%, 25%, 30%, 35%, 40%, 45%, and 50% reduction in the YMRS score compared to baseline) measured at weeks 1 and 2 after treatment initiation and calculated whether these were predictive of subsequent treatment non-response (defined as ≤50% reduction in YMRS score at study endpoint compared with baseline) or non-remission (YMRS score > 8 at endpoint). The definitions of response and remission are based on the Task Force Report of the International Society for Bipolar Disorders (ISBD), because there is no consensus about

102 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success this in the EMA Committee for Proprietary Medicinal Products (CPMP) guideline on the clinical investigation of medicinal products for the treatment and prevention of bipolar disorder (21, 22). The study endpoint was 3 weeks after the baseline measurement. This is the time point recommended by the CPMP guideline for investigating the short- term efﬁcacy of drugs for use in the acute manic episodes of bipolar disorders (21). If outcome data at week 3 were missing, we used data at week 4, if available.

STATISTICAL ANALYSIS

Analyses were restricted to the treatment groups, since our study objective was to determine the predictive value of early non-response to the antipsychotic treatment of acute mania. Results of similar analyses from the placebo groups will be presented brieﬂy in the supplementary eTable 1. Two different analyses were considered: (1) an intention-to-treat (ITT) analysis including all patients with data available at least for week 0 and study endpoint (N=1264); and (2) a completers analysis including only patients with data available for week 0, week 1, week 2, and study endpoint (N=1243). Because the difference between these two study groups was only 21 patients, we decided to provide data for the completers analysis (N=1243).

To answer the ﬁrst research question, we used individual patient data to calculate treatment response at 1 and 2 weeks. To identify the most adequate criterion of early non-response, we calculated the predictive value for treatment failure (PVnr_se) of each cut-off score for early non-response. The PVnr_se was deﬁned as the probability that early non-responders become non-responders or non-remitters at study endpoint, i.e. Pr(non-response or non-remission at endpoint|early non-response).

To assess the potential effect of differences in antipsychotics on response and remission rates, we performed two multilevel mixed effect logistic regression analyses with a random intercept for study, and performed a likelihood ratio test to investigate the effect of speciﬁc antipsychotics on response and remission.

RESULTS

Study characteristics Data were analysed from 10 studies involving 2666 patients, of which 1908 patients

PART 3 - CHAPTER 6 - Early non-response in the antipsychotic treatment of acute mania| 103 Table 1. Patient and study characteristics of completers per studya.

Visits (weeks) 0, 0.5, 1, 2, 3, 0, 0.5, 1, 2, 3, 0, 0.5, 1, 2, 3 0,1,2,3 0, 0.5, 1, 2, 3 0, 0.3, 0.5, 1, 2, 3 0, 0.3, 0.5, 1, 2, 3 0, 0.3, 0.5, 1, 2, 3 0, 1, 2, 3 0, 1, 2, 3 Tot

Region Europe, Other EUR, Other Other Europe, Other USA EUR, USA, Other EUR, USA, Other EUR, USA, Other USA USA Tot

YMRS baseline 32.70 (6.08) 33.50 (6.76) 36.76 (8.02) 31.69 (6.81) 27.79 (4.58) 27.76 (4.92) 26.82 (4.67) 26.61 (5.41) 27.21 (6.22) 28.74 (6.11) 30.09 (6.87)

11.0

African

Ethnicity (%) Caucasian 74.7, Asian 9.0, Other 16.3 Caucasian 51.1, Asian 48.9 African American 0.5, Asian 95.5 Caucasian 60.8, Asian 36.9, Other 2.4 Caucasian 67.1, African American 22.0, Other Caucasian 47.7, African American 21.8, Asian 30.0, Other 0.5 Caucasian 71.0, American 19.3, Asian 9.1, Other 0.6 Caucasian 79.4, African American 16.7, Asian 3.5, Other 0.4 Caucasian 80.3, African American 15.2, Other 4.5 Caucasian 81.2, African American 11.6, Other 7.2 Caucasian 59.9, African American 10.0, Asian 11.2, Other 18.9

) Visits per protocol per week c

Female 0.64 0.42 0.36 0.46 0.49 0.49 0.45 0.50 0.52 0.48 0.48

)

BMI Mean (SE 25.73 (4.45) 23.51 (5.62) 28.23 (6.96) 27.78 (6.52) 27.61 (6.46) 28.81 (6.32)

)

) Active comparator (AC) in study; b Age Mean (SE 43.48 (12.95) 39.76 (13.11) 35.06 (11.80) 39.86 (12.79) 38.49 (11.56) 39.40 (11.51) 40.35 (10.86) 40.29 (11.10) 36.94 (11.05) 39.23 (11.32) 39.73 (12.05)

AC E - - E - - A - - - Tot

Studied compound A A B B B C C C D D Tot

Nr. Patient Treatment/ Placebo 161/72 101/77 101/86 204/89 50/32 147/73 290/62 107/121 42/24 40/29 1,908

) Patient characteristics based on treatment group only Study 1 2 3 4 5 6 7 8 9 10 Tot a

104 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success met criteria for the per the completers analyses. Of these patients, 1243 were treated with antipsychotics. Their mean (SE) age was 39.79 (12.11), mean body mass index (BMI) was 26.81 (6.08), 47% of patients were women, and mean severity score at baseline was 30.02 (6.62). The included patients consisted of 61.6% Caucasians, 10.1% African Americans, 11.0% Asians, and 17.3% patients with another ethnic background (Table 1).

Effect of early non-response cut-off scores on treatment failure The a priori chances of non-response and non-remission were 40.9% (95% CI 38.2- 43.6) and 65.3% (95% CI 62.0-68.6), respectively. Table 2 and Figures 1A and 1B show that the predictive value of early non-response at weeks 1 and 2 for treatment non-response at endpoint (PVnr_se) decreased linearly with increasing cut-off score for early non-response at week 1 and week 2. Table 3 and Figures 1A and 1B show that a similar linear decreasing pattern was observed for the prediction of non-remission by increasing early non-response cut-offs at week 1, whereas early non-response cut-offs at week 2 were not related to non-remission at endpoint due to the overall very high non-remission rates (ceiling effect). More patients at week 1 than at week 2 met the definition of non-response/non-remission at endpoint, regardless of the cut-off score used, whereas the predictive value of non-response at week 1 for treatment failure was lower than that at week 2 (Tables 2 and 3). For example, in week 1 early non-response defined as a ≤10% reduction in YMRS score resulted in a PVnr_se of 76.0% (95%-CI 69.7-82.3, N=179) for non-response, whereas early non-response defined as a ≤25% reduction in YMRS score resulted in a PVnr_se of 64.9% (95%-CI 60.6-69.2, N=470), and non-response defined as a ≤50% reduction in YMRS score resulted in a PVnr_se of 48.7% (95%-CI 45.5-51.9, N=912) (Table 2). Similarly, for non-remission at endpoint, week 1 early non-response defined as a ≤10% reduction in YMRS score resulted in a PVnr_se of 92.2.0% (95%-CI 88.3-96.1, N=179), whereas early non-response defined as a ≤25% reduction in YMRS score resulted in a PVnr_se of 87.4% (95%-CI 84.4-90.4, N=470), and non-response defined as a ≤50% reduction in YMRS score resulted in a PVnr_se of 76.8 (95%-CI 74.4-79.5, N=912). Contrary, in week 2 the PVnr_se was 94.2% (95%-CI 89.7-98.7, N=103) for early non- response defined as ≤10% reduction in YMRS score, 93.4 (95%-CI 90.3-96.5, N=243) for early non-response defined as a ≤25% reduction in YMRS score, and 93.2% (95%-CI 91.3-95.1, N=660) (Table 3). The positive predictive values for early response at week

PART 3 - CHAPTER 6 - Early non-response in the antipsychotic treatment of acute mania| 105

Table 2. Effect of early non-response on response outcome of antipsychotic treatment (Ntot=1243). Non-response

Criteria for non-response N Cumulative PVnr_se (%)b 95%-CIc (% reduction in symptoms) %a Week 1 ≤10% 179 14.4 76.0 69.7-82.3 ≤15% 270 21.7 74.8 69.6-80.0 ≤20% 359 28.9 70.5 65.8-75.2 ≤25% 470 37.8 64.9 60.6-69.2 ≤30% 574 46.2 59.8 55.8-63.8 ≤35% 694 55.8 56.8 53.1-60.5 ≤40% 785 63.2 53.4 49.9-56.9 ≤45% 854 68.7 51.1 47.7-54.5 ≤50% 912 73.4 48.7 45.5-51.9 Week 2 ≤10% 103 8.3 90.3 84.6-96.0 ≤15% 142 11.4 86.6 81.0-92.2 ≤20% 183 14.7 85.8 80.7-90.9 ≤25% 243 19.5 85.2 80.7-89.7 ≤30% 307 24.7 82.7 78.5-86.9 ≤35% 390 31.4 80.3 76.4-84.2 ≤40% 464 37.3 76.5 72.6-80.4 ≤45% 577 46.4 70.0 66.3-73.7 ≤50% 688 55.3 65.0 61.4-68.6 a ) Cumulative % of N/Ntot b) PVnr_se (%) = predictive value of early non response on non-response at study endpoint c) 95%-CI of PVnr_se (%) for non-response

Table 3. Effect of early non-response on remission of antipsychotic treatment (Ntot=1243).

Non-remission b c Criteria for non-response N Cumulative PVnr_se (%) 95%-CI (% reduction in symptoms) %a Week 1 ≤10% 179 14.4 92.2 88.3-96.1 ≤15% 270 21.7 91.9 88.6-95.2 ≤20% 350 28.2 90.5 87.4-93.6 ≤25% 470 37.8 87.4 84.4-90.4 ≤30% 574 46.2 84.7 81.8-87.6 ≤35% 694 55.8 84.3 81.6-87.0 ≤40% 786 63.2 81.0 78.3-83.7 ≤45% 854 68.7 78.9 76.2-81.6 ≤50% 912 73.4 76.8 74.1-79.5 Week 2 ≤10% 103 8.3 94.2 89.7-98.7

≤15% 142 11.4 91.5 86.9-96.1 ≤20% 183 14.7 91.8 87.8-95.8 ≤25% 243 19.5 93.4 90.3-96.5 ≤30% 307 24.7 93.2 90.4-96.0

≤35% 390 31.4 93.6 91.2-96.0

≤40% 464 37.3 93.3 91.0-95.6

≤45% 577 46.4 93.1 91.0-95.2

≤50% 660 53.1 93.2 91.3-95.1

a) Cumulative % of N/Ntot

b) PVnr_se (%) = predictive value of early non response on non-remission at study endpoint c ) 95%-CI of PVnr_se (%) for non-remission

106 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success

1 and week 2 and the positive and negative predictive values for the placebo group are presented in the supplementary eTables 1 and 2, respectively.

Optimal combination of cut-off score for early non-response and time point The most adequate criterion for early non-response could not be determined because of the linear relationship between cut-off score and PVnr_se and the linear relationship between the cut-off score and the number of patients who met criteria for non-response or non-remission (Figures 1A and 1B). To illustrate the potential effect of a change in treatment strategy based on early non-response, we provide a brief example, deﬁning early non-response as a ≤25% reduction in YMRS score in week 1 and a ≤50% reduction in YMRS score in week 2 for non-response at study-endpoint. These cut-off scores have been used in previous studies. (12-14)

The a priori chance of non-response at study endpoint was 40.9%. At week 1, a non- responder (≤25% reduction in YMRS score) had a 64.9% chance of treatment failure, resulting in an increase in the PVnr_se of 24.0%. A clinician can now decide to change the existing treatment, e.g. dose elevation, adjuvant medication, or switch to another medication. If the patient is switched to another antipsychotic, he or she again has an a priori chance of treatment failure of 40.9%. Overall, when switched to an alternative antipsychotic treatment, this patient now “only” has a 40.9*40.9=16.7% chance of treatment failure. Patients who do not meet the deﬁnition for non-response should be re-evaluated in week 2. If the criterion for non-response (≤50%) is met in week 2, there again is a 64.9% chance of non-response at study endpoint.

Effect of type of antipsychotic Potential differences in the type of antipsychotic used did not affect our ﬁndings. When non-response in week 1 (deﬁned as ≤25% reduction in YMRS score) was entered in a multilevel mixed effect logistic regression analyses with a random intercept for study, the likelihood ratio test was (LRχ2(8))=12.60, p=0.127. For non-response in week 2 (≤50% reduction in YMRS score) LRχ2(8)=10.80, p=0.213.

DISCUSSION

The aim of this study was to investigate whether a lack of response to antipsychotic

PART 3 - CHAPTER 6 - Early non-response in the antipsychotic treatment of acute mania| 107 Figure 1A. Predictive value of early non-response for non-response and non-remission at study-endpoint (PVnr_se) at week 1.

Figure 1B. Predictive value of early non-response for non-response and non-remission at study-endpoint (PVnr_se) at week 2.

108 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success treatment after 1 or 2 weeks predicts future treatment failure, and to determine the optimal criteria (cut-off score to define non-response, measured at 1 or 2 weeks) for prompting re-evaluation of treatment choices. We found early non-response at weeks 1 and 2, defined as ≤10% to ≤50% reduction in symptom scores from baseline, respectively, to be a significant and clinically relevant predictor of treatment failure in terms of both non-response and non-remission. At both 1 and 2 weeks, the predictive value for treatment failure (PVnr_se) and the number of patients meeting criteria for non-response were linearly related to the cut-off score. A higher cut-off score (e.g. ≤50% versus ≤10% reduction in symptom score) resulted in a lower PVnr_se and an increased number of patients meeting criteria for non-response (week 1 and 2) and non-remission (week 1). Because of this linear relationship, we could not determine the best combination of criteria (cut-off score and evaluation in week 1 or 2) for prompting treatment reconsideration. However, our data do suggest that if patients fail to improve after 1 or 2 weeks of antipsychotic treatment for acute manic episodes, clinicians should reconsider a change in treatment strategy.

Although this may not seem a surprising finding, it may have important clinical implications and could be helpful for clinicians when deciding on treatment strategy: waiting for a treatment response might not be the best strategy, especially not in outpatients and in patients with serious self-defeating behaviours (e.g. money spending, promiscuity) or a lack of support from their social network. However, the question remains how long one should wait to change treatment strategy? Two considerations are important: (1) a very poor response in week 1 (e.g. ≤10% symptom reduction) increases the chance of non-response at endpoint from 41% to 76% and may call for an immediate change in treatment strategy, whereas a weak response in week 1 (≤50%) only increases the chance of non-response at endpoint from 41% to 49% and thus an immediate change in treatment strategy is not needed, (2) potential non-responders can still be identiﬁed in week 2 with very high predictive power. Therefore, a two-stage follow-up strategy, with monitoring of illness severity in weeks 1 and 2 after treatment initiation is recommended in patients with acute mania. For remission, these data look rather different: a very poor response in week 1 increases the chance of non-remission from 65% to 92%, whereas a weak response increases the chance of non-remission to 77%. One may either conclude that any non-response calls for immediate change in treatment strategy or that remission on such a short-term is not a realistic treatment

PART 3 - CHAPTER 6 - Early non-response in the antipsychotic treatment of acute mania| 109 goal. We feel that the latter conclusion is clinically more accurate. Our findings are in line with those of earlier studies on early non-response and the prediction of outcome in acute mania. For example, Ketter et al. (2010) found a significant early treatment response of ziprasidone at day 4 after treatment initiation (13). Subsequently, Kemp et al. (2011) and Szegedi et al. (2013) studied two different antipsychotics and found that at week 1 (≤25% symptom reduction) the negative predictive values for non-response were 75% in Kemp et al. (2011) and 67% and 80% in Szegedi et al (2011), whereas the negative predictive values for non-remission at week 2 were 95% and 85% and 76%, respectively (12, 14). These findings are very similar to our findings of the negative predictive values for early non-response (≤ 25% symptoms reduction) for non-response (65%) and for non-remission (87%).

One of the strengths of our study is that, in addition to earlier studies (12-14), we conducted an individual patient data meta-analysis of five different antipsychotic drugs and were able to include 1243 patients for complete case analyses. We now know that the type of drug does not modify the effect of early non-response on treatment success or failure and thus findings can be generalized to all antipsychotics. Furthermore, contrary to Szegedi and colleagues (2012), we were also able to assess early non-response rates in week 2 and therefore could assess the most parsimonious time frame (1 or 2 weeks) to reconsider antipsychotic treatment (14). Considering these strengths, our findings could help to promote agreement in the psychiatric field about the recommendation of the most adequate time frame to reconsider antipsychotic treatment in patients with acute mania (17). Most guidelines (e.g. the Guideline of The World Federation of Societies of Biological Psychiatry (WFSBP) and the Dutch Guideline for Bipolar Disorder) now recommend a time frame of 2 weeks (23, 24), although one recent guideline does not make any recommendation at all (NICE 2014) (19). In contrast, most of the empirical literature recommends 1 week (12-14). We recommend a two-stage strategy, taking into account the level of non-response in weeks 1 and 2 and the need to achieve a rapid improvement in patients’ symptoms, given the nature of the manic behaviour and the presence or absence of a supporting network.

This study had some limitations. The ﬁrst limitation is that our study endpoint at 3 or 4 weeks could have led to an overestimation of non-response and non-remission rates, since it may take longer to achieve symptom reduction or remission in some patients

110 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success (25). With a priori chances of non-response and non-remission at study endpoint of 40.9% and 65.3% respectively, one could argue that treatment for 3 weeks may be too short to assess efficacy and that longer study periods are needed to test the efficacy of treatments for an acute manic episode. However, our endpoint is in accordance to the EMA Committee for Proprietary Medicinal Products (CPMP) guideline on the clinical investigation of medicinal products for the treatment and prevention of bipolar disorder (21). Nevertheless, we think that future studies are needed to predict non-response and non-remission at week 6, 8 or even later. Second, our study was a post-hoc analysis of existing RCTs suggesting that early non-response can be used to predict non-response and non-remission at endpoint and that early non-response could be used for timely switches in treatment strategy. However, future prospective studies using the switching recommendations are needed to test the predictive validity of these recommendations in clinical practice. Third, it has to be emphasized that our findings were only based on efficacy results and did not take into account tolerability, patient compliance, and/ or patient preference (26). Unfortunately, we had no data on these variables. However, we do not think that these potential reasons for early non-response will change our conclusions with regard to the right time to consider switching to another treatment strategy.

Lastly, if patients do not respond to antipsychotic treatment, clinicians should consider changing the drug used. The next questions are which changes are likely to be most effective in these patients and how should the change be implemented? There is little evidence with regard to resistance to treatment for episodes of acute mania, and although switching antipsychotic medication in patients with acute mania is common clinical practice, little is known about the best switching strategy (e.g., abrupt switch, cross-taper switch, and plateau cross-taper switch) (27-29).

In conclusion, early non-response to antipsychotic treatment is a clinically relevant predictor of treatment outcome and could help clinicians to decide whether their pharmacological treatment strategy for acute mania should be changed. If a patient fails to improve in the ﬁrst 2 weeks of treatment, waiting for treatment to become effective may not be the right option. We advise reconsidering treatment options before week 3, using a two-stage strategy taking into account the level of non-response in weeks 1 and 2 and the need to achieve a rapid improvement in symptoms, given the

PART 3 - CHAPTER 6 - Early non-response in the antipsychotic treatment of acute mania| 111 nature of manic behaviour and the presence or absence of a supporting network. However, future prospective studies using these switching recommendations are needed to test the predictive validity of these recommendations in clinical practice.

CLINICAL POINTS

• Although antipsychotics are the ﬁrst line treatment for an acute manic episode of bipolar disorder, success rates are not very high. • Non-response to antipsychotic treatment after one or two weeks is a strong predictor of non-response and non-remission after three to four weeks.. • Therefore, existing antipsychotic treatments of acute mania should be seriously reconsidered in the case of non-response after one or two weeks.

112 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success REFERENCES

1. Black DW, Winokur G, Hulbert J, Nasrallab A. Predictors of Immediate Response in the Treatment of Mania: The Importance of Comorbidity. Biological Psychiatry. 1988;24:191-198. 2. Andrade C. The risk of harm in mania and the very early time course of improvement: important but neglected variables in treatment research. Bipolar Disorders 2004;6:446-447. 3. Clements C, Morriss R, Jones S, Peters S, Roberts C, Kapur N. Suicide in bipolar disorder in a national English sample, 1996–2009: frequency, trends and characteristics Psychological Medicine. 2013;43:2593-2602. 4. Khan A, Faucett J, Morrison S, Brown WA. Comparative Mortality Risk in Adult Patients With Schizophrenia, Depression, Bipolar Disorder, Anxiety Disorders, and Attention-Deficit/Hyperactivity Disorder Participating in Psychopharmacology Clinical Trials. JAMA Psychiatry. 2013;70:1091-1099. 5. Garfinkel PE, Stancer HC, Persad EA. A comparison of haloperidol, lithium carbonate and their combination in the treatment of mania. Journal of Affective Disorders 1980;2:279-288. 6. Brown D, Silverstone T, Cookson J. Carbamazepine compared to haloperidol in acute mania. International Clinical Psychopharmacology. 1989;4:229-238. 7. Segal J, Berk M, Brook S. Risperidone compared with both lithium and haloperidol in mania, a double-blind randomized controlled trial. Clinical Neuropharmacology. 1998;21:176-180. 8. Tohen M, Sanger TM, McElroy SL, Tollefson GD, Chengappa KN, Daniel DG, al. e. Olanzapine versus placebo in the treatment of acute mania. Olanzapine HGEH Study Group. . American Journal of Psychiatry. 1999;156:702-709. 9. Potkin SG, Keck JPE, Segal S, Ice K, English P. Ziprasidone in acute bipolar mania, a 21-day randomized, double-blind, placebo-controlled replication trial. Journal of Clinical Psychopharmacology 2005;25:301-310. 10. Keck JPE, Versiani M, Potkin S, West SA, Giller E, Ice K. Ziprasidone in the treatment of acute bipolar mania, a three-week, placebo-controlled, double-blind, randomized trial. American Journal of Psychiatry 2003;160:741-748. 11. Tohen M, Jacobs TG, Feldman PD. Onset of action of antipsychotics in the treatment of mania. Bipolar Disorders. 2000;2:261-268. 12. Kemp DE, Johnson E, Wang WV, Tohen M, Calabrse JR. Clinical Utility of Early Improvement to Predict Response or Remission in Acute Mania: Focus on Olanzapine and Risperidone. Journal of Clinical Psychiatry. 2011;72:1236-1241. 13. Ketter TA, Agid O, Kapur S, Loebel A, Siu CO, Romano SJ. Rapid antipsychotic response with ziprasidone predicts subsequent acute manic/mixed episode remission. Journal of Psychiatric Research 2010;44:8-14. 14. Szegedi A, Zhao J, McIntyre RS. Early improvement as a predictor of acute treatment outcome in manic or mixed episodes in bipolar-1 disorder: A pooled, post hoc analysis from the asenapine development program. Journal of Affective Disorders. 2012;150:745-752. 15. Nierenberg AA, McLean NE, Alpert JE, Worthington JJ, Rosenbaum JF, Fava M. Early nonresponse to fluoxetine as a predictor of poor 8-week outcome. Am J Psychiatry. 1995 152:1500-1503. 16. Kudlow PA, McIntyre RS, Lam RW. Early Switching Strategies in Antidepressant Non-Responders: Current Evidence and Future Research Directions. CNS Drugs. 2014;28:601-609. 17. Nivoli MA, Murru A, Goikolea JM, Crespo JM, Montes JM, González-Pinto A, García-Portilla P, Bobes J, Sáiz-Ruiz J, vieta E. New treatment guidelines for acute bipolar mania: a critical review. Journal of Affective Disorders. 2012;140:125-141.

PART 3 - CHAPTER 6 - Early non-response in the antipsychotic treatment of acute mania| 113 18. Nolen WA, Kupka RW, Schulte PFJ, Knoppert-van der Klein EAM, Honig A, Reichart CG, Goossens PJJ, Daemen P, Ravelli DP: Richtlijn bipolaire stoornissen - Tweede, herziene versie, 2008. Utrecht, Nederlandse Vereniging voor Psychiatrie; 2008. 19. NICE: Bipolar disorder: the assessment and management of bipolar disorder in adults, children and young people in primary and secondary care in NICE clinical guideline 185, National Collaborating Centre for Mental Health; 2014. 20. Young RC, Biggs JT, Ziegler VE, Meyer DA. A rating scale for mania: reliability, validity and sensitivity. British Journal of Psychiatry. 1978;133:429-435. 21. CHMP: Note for guidance on clinical investigation of medicinal product of the treatment and prevention of bipolar disorder. Committee for Medicinal Products for Human Use 2001. 22. Tohen M, Frank E, Bowden CL, Colom F, Ghaemi SN, Yatham LN, Malhi GS, Calabrese JR, Nolen WA, Vieta E, Kapczinski F, Goodwin GM, Suppes T, Sachs GS, Chengappa KNR, Grunze H, Mitchell PB, Kanba S, Berk M. The International Society for Bipolar Disorders (ISBD) Task Force report on the nomenclature of course and outcome in bipolar disorders. Bipolar Disorders 2009;11:453-473. 23. Nolen WA, Kupka RW, Schulte PFJ, Knoppert-van der Klein EAM, Honig A, Reichart CG, Goossens PJJ, Daemen P, Ravelli DP: Richtlijn bipolaire stoornissen; tweede, herziene versie, 2008. De Tijdstroom, Utrecht; 2008. 24. Grunze H, Vieta E, Goodwin GM, Bowden C, Licht RW, Möller H-J, Kasper S, disorders WTfotgfb. The World Federation of Societies of Biological Psychiatry (WFSBP) Guidelines for the Biological Treatment of Bipolar Disorders: Update 2009 on the Treatment of Acute Mania. The World Journal of Biological Psychiatry. 2009;10:85-116. 25. Houston JP, Ketter TA, Case M, Bowden C, Degenhardt EK, Jamal HH, Tohen M. Early symptom change and prediction of subsequent remission with olanzapine augmentation in divalproex-resistant bipolar mixed episodes. Journal of Psychiatric Research 2011;45:169-173. 26. Bernardo M, Vieta E, Saiz Ruiz J, Rico-Villadermoros F, Álamo C, Bobes J, Group R. Recommendations for switching antipsychotics. A position statement of the Spanish Society of Psychiatry and the Spanish Society of Biological Psychiatry. Revista de Psiquiatría y Salud Mental (Barc). 2011;4:150-168. 27. Grande I, Bernando M, Bobes J, Saiz0Ruiz J, Álamo C, Vieta E. Antipsychotic switching in bipolar disorder: a systematic review. International Journal of Neuropsychopharmacology. 2014;17:497-507. 28. Pae CU, Lee KU, Kim JJ, Lee CU, Bahk WM, Lee SJ, Lee C, Paik IH. Switching to quetiapine in patients with acute mania who were intolerant to risperidone. Human Psychopharmacology Bulletin. 2004;19:47-51. 29. Lee H-B, Yoon B-H, Kwon Y-J, Woo YS, Lee J-G, M-D. K, Bahk W-M. The Efﬁcacy and Safety of Switching to Ziprasidone from Olanzapine in Patients with Bipolar I Disorder: An 8-Week, Multicenter, Open-Label Study. Clinical Drug Investigation. 2013;33:743-753. Supplementary eTable 1. Positive predictive values (PPV, 95%-CI) of early response on response at study endpoint in antipsychotic (Ntot=1243) and placebo group (Ntot=665).

c CI - 59.3 62.2 65.9 69.5 73.3 76.6 80.7 86.2 88.1 59.2 61.9 65.0 69.2 72.8 76.6 80.4 85.8 88.5 ------95% 50.7 53.2 56.5 59.5 62.7 65.4 68.9 74.2 75.3 50.8 53.5 56.4 60.4 63.8 67.4 71.2 76.8 79.3

55.0 57.7 61.2 64.5 68.0 71.0 74.8 80.2 81.7 55.0 57.7 60.7 64.8 68.3 72.0 75.8 81.3 83.9 PPV (%)

Placebo Group 77.1 69.0 61.2 52.5 44.2 38.3 31.6 25.1 21.4 82.9 78.5 73.8 68.4 62.6 55.9 49.6 42.6 37.3 Cumulative %

N 513 459 407 349 294 255 210 167 142 551 522 491 455 416 372 330 283 248

CI -

513 459 407 349 294 255 210 167 142 551 522 491 455 416 372 330 283 248 95%

65.9 69.5 72.2 74.9 76.7 80.9 82.5 83.8 83.4 64.4 65.8 67.7 70.8 73.8 78.2 81.5 85.7 90.6 PPV (%)

Antipsychotic Group 85.6 78.3 71.1 59.0 53.8 44.2 36.8 31.3 26.6 91.7 88.6 85.3 80.5 75.3 68.6 62.7 53.6 44.7 Cumulative %

N 973 884 733 669 549 458 389 331 936 853 779 666 555 1064 1140 1101 1060 1000

tot

>10% >15% >20% >25% >30% >35% >40% >45% >50% >10% >15% >20% >25% >30% >35% >40% >45% >50%

CI of PPV (%) - Criteria for response ek 1 ) PPV (%) = positive predictive value of early response on response at study endpoint ) Cumulative % of N/N ) 95% (% reduction in symptoms) We Week 2 a b c

PART 3 - CHAPTER 6 - Early non-response in the antipsychotic treatment of acute mania| 115 Supplementary eTable 2. Results early non-response on non-response and non-remission treatment outcome in placebo group (Ntot=665).

c CI - 99.9 98.7 97.9 96.6 95.8 95.8 94.1 93.2 91.8 98.2 97.7 97.6 96.7 100.8 100.5 100.4 100.3 100.0 ------95% 94.9 93.5 92.7 91.4 90.8 91.0 89.1 88.0 86.4 93.6 93.3 93.4 92.3 97.4 96.7 97.4 97.7 96.8

(%)

97.4 96.1 95.3 94.0 93.3 93.4 91.6 90.6 89.1 99.1 98.6 98.9 99.0 98.4 95.9 95.5 95.5 94.5 nr_se PV remission -

a Non

22.9 31.0 38.8 47.5 55.8 61.7 68.4 74.9 78.6 17.1 21.5 26.2 31.6 37.4 44.1 50.4 57.4 62.7 Cumulative %

N 152 206 258 316 371 410 455 498 523 114 143 174 210 249 293 335 382 417

CI - 89.5 85.1 82.6 78.9 75.8 73.8 71.3 69.5 67.6 99.9 99.1 98.1 97.4 94.4 90.6 87.0 83.8 80.4 ------95% 77.7 74.1 72.4 69.3 66.6 64.8 62.7 61.1 59.4 93.1 92.5 91.5 91.2 87.2 82.8 79.0 75.8 72.2

(%)

83.6 79.6 77.5 74.1 71.2 69.3 67.0 65.3 63.5 96.5 95.8 94.8 94.3 90.8 86.7 83.0 79.8 76.3 nr_se PV remission at study endpoint response at study endpoint - - response

- a Non

22.9 31.0 38.8 47.5 55.8 61.7 68.4 74.9 78.6 17.1 21.5 26.2 31.6 37.4 44.1 50.4 57.4 62.7 Cumulative %

N 152 206 258 316 371 410 455 498 523 114 143 174 210 249 293 335 382 417

tot response (%)

- ≤10% ≤15% ≤20% ≤25% ≤30% ≤35% ≤40% ≤45% ≤50% ≤10% ≤15% ≤20% ≤25% ≤30% ≤35% ≤40% ≤45% ≤50% nr_se

(%) = predictive value of early non response on non (%) = predictive value of early non response on non

CI of PV - nr_se nr_se reduction in symptoms) Cumulative % of N/N Criteria for non ) PV ) PV ) ) 95% (% Week 1 Week 2 a b c d

116 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success PART 3 - CHAPTER 6 - Early non-response in the antipsychotic treatment of acute mania| 117

PARTdiscussion and 4Summary

07 CHAPTER 7: SUMMARY AND DISCUSSION

SUMMARY AND DISCUSSION

An acute manic episode of bipolar disorder is treated with medication to rapidly reduce manic symptom severity. Before patients can have access to a medication, it needs to be proven safe and effective. The balance between efficacy and safety is assessed by regulatory authorities such as the Medicines Evaluation Board (MEB) for the Netherlands or the European Medicines Agency (EMA) for Europe. This assessment can be challenging and the outcomes are often debated. To better understand and improve these clinical trials for drug registration, the MEB facilitated me access to a unique database, including the raw individual patient data (IPD) of twelve double blind, randomized, placebo-controlled trials of drugs for patients diagnosed with a DSM-IV acute manic episode of bipolar disorder. This enabled me to examine whether the results of these clinical trials could be extrapolated across geographic regions, to study the sources for the high failure rate of psychiatric clinical trials and the high placebo responses, and whether placebo response can be predicted. In addition, it enabled me to examine whether we are using the best measure of a clinically relevant effect. Since regulatory questions are closely related to clinical questions and I had access to the individual patient data, I also studied some important clinical questions. To improve treatment efficiency, I examined whether the level of insight in one’s illness influences the efficacy of antipsychotic treatment and whether final treatment failure at study endpoint can be predicted by early non-response.

SUMMARY OF FINDINGS

REGULATORY QUESTIONS

In the ﬁrst part of my thesis I addressed three regulatory questions about the short-term efﬁcacy of drug treatment in patients with acute mania.

Geographic differences in efficacy In chapter 2, I investigated whether there are differences across geographic regions (USA, Europe, and ‘Other’ regions) in the efficacy of medication for the treatment of an acute manic episode and found significant and substantial differences in the effect size across regions, with a substantial larger effect size in patients in Europe and the Other region compared to patients in the USA. Regional differences in baseline

PART 4 - CHAPTER 7 - Summary and discussion| 123 characteristics (age, gender, ethnicity, BMI, and initial severity), differences in placebo response and discontinuation rate between patients in the USA and Other regions could not explain the regional differences in mean change score from baseline to follow-up. However, differences in initial severity did explain some of the differences in response rate between US and European patients; patients in the USA were less severely ill at baseline than patients in Europe and this partly explained the lower response rate in the US patients. Similar geographic differences in effect have been found in studies about schizophrenia, contraception, and cardiovascular problems and, therefore, my ﬁndings indicate the presence of a more general problem and may affect the requirements of regulatory authorities with regard to regional extrapolation of study results.

Placebo response Next, in chapter 3, I examined whether the magnitude of the placebo response predicts treatment efficacy of antipsychotics in patients with acute mania and investigated which patient and study characteristics predict placebo response. The findings showed that the placebo effect three weeks after treatment initiation was substantial; 8.5 points improvement on the YMRS (=27.9%) and a response rate of 32.8%. Consistent with previous studies, a higher placebo response rate was strongly associated with a smaller effect size. There were five significant predictors (study and patient characteristics) of the placebo effect; higher illness severity at baseline for the YMRS mean change score (MCS), lower illness severity at baseline for the response rate (RR), absence of psychotic features for MCS and RR, more recent study year for MCS and RR, three versus one regions for MCS and RR, and patients from the Other region versus patients from the USA or European region for MCS and RR. However, the final prediction model explained only 5-10% of the variance in the placebo response. Therefore, I conclude that restricting trial participation to certain patients in certain geographic regions is not an effective strategy to improve assay sensitivity in placebo-controlled trials with antipsychotics in patients with acute mania.

Net gain analysis Finally, in chapter 4, I propose a new approach to measure outcome in RCTs: ‘net gain analysis’ with net gain deﬁned as the difference in the percentage responders between the treatment and placebo group minus the difference in the percentage deteriorators between these groups. I investigated whether the net gain analysis is

124 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success better or more comprehensive than the traditionally applied responder analysis in the case of antipsychotic treatment of patients with acute mania. The ranking of the compounds in terms of clinical relevance was very similar for the responder analysis and the net gain analysis, but the magnitude of the effects of the various compounds were more comparable when using the net gain analysis compared to the responder analysis. These ﬁndings suggested that the net gain analysis is a more precise efﬁcacy measure, describing the differences in clinical relevance of the different compounds more comprehensively.

CLINICAL QUESTIONS

In the second part of this thesis I addressed two questions regarding the clinical practice of antipsychotic treatment in patients with acute mania.

Insight In chapter 5, I examined whether the level of insight in one’s illness at the start of treatment affects the efficacy of antipsychotics in patients with acute mania. Unexpectedly, the findings showed that antipsychotics are more effective in patients with impaired or no insight than in patients with excellent insight. These findings suggest that patients with acute mania should immediately receive adequate treatment, including antipsychotics, independent on their level of insight and not to wait with medication until adequate insight has been established.

Early non-response In chapter 6, I examined whether early non-response to antipsychotic treatment predicts non-response and non-remission at study-endpoint and concluded that early non- response at week one and week two, defined as ≤10% to ≤50% symptom reduction compared to baseline, is a significant and clinical relevant predictor for treatment failure at week 3-4. Given the linear relation between cut-off score and the predictive value for non-response (at week 1 and 2) and non-remission (week 1) at study endpoint, no optimal combination of early non-response definition and time-point for treatment reconsideration could be determined. Based on these findings, I recommended reconsidering treatment options before week three using a two-stage strategy that takes into account the level of non-response at week one and week two, the need

PART 4 - CHAPTER 7 - Summary and discussion| 125 for a rapid improvement given the nature of the manic behavior, and the presence or absence of a supporting network.

General discussion and future directions For this thesis I had access to a unique database containing individual patient data (IPD) of twelve randomized controlled trials (RCTs). These RCTs were initially conducted for the purpose of registration of a compound for the indication acute mania and are normally used for this one purpose only. However, the regulatory field is closely connected to the clinical field since they both focus on clinically relevant effects and improvement of patient outcomes. Due to the exceptional availability of IPD in the database, I had the unique opportunity to also examine the data for clinical questions. Though closely related, the focus of both fields also shows differences. Where regulatory authorities decide on the clinically relevant effect for a certain patient population, the clinician decides on the clinical relevant effect for an individual patient. This difference generates different hurdles, dilemma’s, and questions while both focusing on that common goal; helping the patient. So, even though both fields are overlapping, their hurdles and questions are colored in a regulatory or clinical way. Therefore I decided to make a distinction between regulatory and clinical issues in the following discussion.

REGULATORY ISSUES

Geographic differences Over the years, clinical trials are conducted increasingly across the globe (1). Between 2005 and 2011, 61.9% of studies submitted to the European Medicines Agency (EMA) were from non-European countries; 34.1% from North America (of which 29.6% from the USA) and 27.8% from the rest of the world. Over the years, the share of studies conducted in the rest of the world has increased whereas North America’s share has decreased (2). Furthermore, as stated in the International Workshop on Ethical and GCP aspects of the acceptance of clinical trials submitted in Marketing Authorisation Applications to EMA, there is an increasing number of international multicenter trials (3). Herefore, regulatory bodies as the EMA, FDA, State Food Administration China, and others bodies collaborated to set international standards for the acceptance of clinical trials conducted outside their jurisdiction. This would enhance ethical and methodological aspects of clinical trials but would also diminish the doubling of

126 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success scientiﬁc efforts by scientists and study participants. Among many others, one point of discussion was on applicability of data from non-European countries to the European population (4). Therefore, the generalizability of trial results across geographic regions was questioned in this thesis (chapter 2). I found signiﬁcant and substantial differences in the magnitude of effect of medications for patients with acute mania in the USA and patients in the rest of the world (Europe and Other region), showing substantial smaller effect sizes in patients in the USA.

The impact of geographic and cultural differences on the efficacy of drug treatments for patients with acute mania was earlier studied by Vieta et al. (2011) (5). They found a significant smaller effect size for mean improvement from baseline to follow-up in patients in the USA compared to patients in Russia and India. Generalizing the results of these countries to regions as used in my study (USA, Europe, and Other region), my results are comparable to the findings of Vieta et al.; significant and substantial smaller effects of drug treatment in US patients versus patients in the rest of the world. Geographic variation in efficacy of drug treatment has also been studied in patients with schizophrenia. Mattila et al. (2014) also found, although not significant, a substantial smaller effect size of atypical antipsychotic treatments in patients in North America versus patients in Europe and the rest of the world (6). Interestingly, geographic variation seems not to be restricted to psychiatric disorders (5, 6). For example, over the years, efficacy of contraceptives frequently showed lower efficacy in clinical trials conducted in US patients versus trials conducted in European patients (7, 8). This issue was even addressed by the Food and Drugs Administration (FDA) in the FDA Advisory Committee Briefing Document – Advisory Committee for Reproductive Health Drug (2007) by asking the question ‘Should a certain minimum percentage of the subjects in phase 3 studies be studied at US sites?’ (9).

I tried to explain the geographic differences in efficacy in several ways. First, differences in compound distribution between the regions could not explain our findings. Second, of all differences in baseline characteristics, only the inclusion of less severely ill patients in the USA could explain some of the difference in response rate between patients in Europe and patients in the USA. The inclusion of significantly less severely ill US patients in clinical trials is an interesting and important element in my study. Unfortunately based on my data, I was not able to explain this phenomenon. However, I want to highlight

PART 4 - CHAPTER 7 - Summary and discussion| 127 one hypothesis. As raised by Cipriani et al. (2011), who conducted a meta-analysis on the efficacy of antimanic drugs, severely ill manic patients are hard to include in a clinical trial, due to the legal provision to provide informant consent to participate in a trial (10). As shown in this thesis (chapter 5), patients in an acute manic episode often have impaired or no insight in their illness (11, 12) and severity is correlated with the level of insight, showing more impaired insight in more severely affected manic patients. Therefore, requiring informed consent from severely ill manic patients could be a major hurdle for participation in a clinical trial. Therefore, I re-analyzed my data to identify the distribution of the level of insight across the three regions and found that a substantial higher percentage of patients in the USA (48.2%) showed excellent insight compared to patients from Europe (5.0%) and the other Region (3.7%). Only 1.7% of the patients in the USA showed no insight, versus 6.8% of the patients in Europe and 11.7% in the Other region. Since in general, EU countries and the USA have comparable civil rights, these findings may implicate that in the USA informed consent is applied more stringently than in European countries. However, re-analyzing my data, l found that the level of insight at baseline did not modify the efficacy of the medication (in MCS and RR) in the three regions. Therefore, it seems that regional differences in insight do not explain the differences in effect size between the geographic regions.

I also examined whether differences in placebo response across regions could explain the regional differences in effect. Interestingly, placebo response was highest in patients from the Other region, as was shown in my study on the placebo response (chapter 3). Therefore, regional differences in placebo response are more likely to have resulted in an underestimation of the regional difference in effect rather than an explanation for the regional difference of effect. Furthermore, I analyzed whether differences in discontinuation rates between the three regions could explain the regional differences in efficacy. Even though significant more patients discontinued treatment in the US (40%) than in Europe (21%) and the Other regions (21%), this did not explain the geographic differences in effect. Finally, study year was not an explanation of my findings. Unfortunately, I was limited in the number and range of variables to empirically explain my findings and therefore I would like to address other potential explanatory factors.

In schizophrenia and contraceptive trials, similar differences in effect were found across

128 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success geographic regions and therefore, biological factors may be less relevant than structural geographic differences in conducting a clinical trial. This assumption is supported by the fact that the US population is a recent mix from European, African and Native American ancestors, and therefore genetically comparable to Europe and the Other region (13). Moreover, if biological factors would be of great importance, geographic differences would have been most prominent between Europe and the Other region. Unfortunately, this hypothesis could not be tested in my study, since my data consisted of RCTs primarily performed to prove drug efficacy and therefore did not contain information about genotypes needed for pharmacogenetic analyses. Trussell and Portman (2013) have stated that, among other factors, higher failure rates of contraceptives in US trials possible reflect differences in health care systems (14). Different health care systems and possibly more important cultural differences, could affect the reason for participation in a clinical trial and thereby modify the study population. In general, the majority of people believes that clinical trials are important (15) and that the use of humans is favored in clinical trials (16), however only a minority of people participates in a clinical trial (16). For participants, the most common primary motivation is gaining personal medical benefit (1, 17-21), and only the secondary motivator is altruistic (1, 20, 21). In contrast, in studies of patients in developing countries, financial benefit was found to be an important factor for study participation. Nappo et al. (2013) found that in patients in Brazil, financial reward was as important as personal medical benefit (1). Therefore, in future studies it should be investigated whether differences in health care system, solidarity versus non-solidarity based health-care systems, social welfare, and cultural factors affect patient participation in clinical trials, thereby influencing the study population, and indirectly the study result.

In this thesis I had to conclude that based on my data, geographic differences in the magnitude of effect of drug treatment of patients with acute mania could only partly be explained. However, since the medications showed a significant and relevant effect in all three regions, the extrapolation of the efficacy results across geographic regions seems justified. However, this may be different for other disorders and other treatments. Therefore, the external validity of the findings and the consistency of study results across geographic regions remain complex. It is therefore strongly recommended to study the geographic differences in future studies to find an explanation for this phenomenon. This could result in conclusions with regard to the US study population; they might present

PART 4 - CHAPTER 7 - Summary and discussion| 129 a specific subgroup of acute manic patients with a higher level of insight and/or a less severe symptom pattern. However, the US study population might also prove to be a reasonable reflection of the US manic patients and clinical practice. Overall, future studies could help to judge whether the FDA can and will accept European or other data and vice versa. It must be addressed that the extrapolation of results from patients in Europe to patients in the Other regions, and the other way around, seems justified. Of note, it is not aspired to restrain globalization in clinical trials and extrapolating results across geographic regions is even stimulated. Therefore, a valid method for extrapolating results across regions can hopefully be developed in future studies, and used by regulatory authorities.

Placebo response Although differences in placebo response could not explain geographic differences in efficacy of medication in patients with an acute manic episode, it is frequently stated that high placebo responses are the main reason for the high failure rates in psychiatric trials (e.g. (22-24)), resulting in a withdrawal of the pharmaceutical industry from psychiatry. In order to better understand this phenomenon, I studied the placebo response in chapter 3 and found that the placebo response was indeed substantial and that a high placebo response negatively affected the efficacy of antipsychotics in the treatment of patients with acute mania. In addition, I found five significant predictors for a higher placebo response; a more recent study year, inclusion of patients from three versus one region, patients from the Other region versus patients from the USA, absence of psychotic features, and a higher (MCS) and lower (RR) illness severity at baseline. Studies conducted in a more recent study year showed a higher placebo response, as was also found in schizophrenia trials (24). This could be attributed to differences in trial design (e.g. providing rescue medication, quality of clinical trial, multi-center trials), differences in patient population, and maybe an improvement of the standard of care for bipolar patients over the years (24-26). Unfortunately I was not able to study whether the number of study sites, as found in Yildiz et al. (2011), was a predictor for placebo response in my study (23). Therefore I analyzed the number of countries as a proxy for the number of study site, but this variable was not a significant predictor of placebo response.

Subsequently, I tried to identify the most parsimonious prediction model for placebo

130 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success response. However, the model only explained 5-10% of the variance in the placebo response. Therefore, based on my findings, it is not recommended to restrict future studies to certain patients in certain regions. I understand that this is a disappointing finding for the pharmaceutical industries. Since the placebo response increases over time, the future prospect on the use of the placebo arm in registration studies could be argued. On the one hand, due to the high placebo response, regulatory authorities may compromise in a certain way with the pharmaceutical industries to stimulate further research in the psychopharmacologic field. On the other hand, if the placebo response continuous to increase over time, the question rises if psychopharmacologic treatment of patients with acute mania remains necessary or only recommended for a certain subgroup of acute manic patients. At this moment, the results can be considered positive for the patients participating in a clinical trial (23, 26). For many years, the use of a placebo arm in psychiatric trials has been discussed. Opponents stated that the use of placebo in psychiatric patients is unethical (27), especially in disorders were patients are at risk and successful treatment is available (25), as in an acute manic episode of bipolar disorder. However, until now, the use of placebo remains required to show efficacy in registration studies for the indication bipolar mania. (28) Finally, the high placebo response in patients with acute mania is interesting for clinical practice; it is known from previous studies that the placebo response represents a true neurophysiological effect (29) and thus a high placebo response in this patient group could have a positive impact on treatment outcome in clinical practice. More research is needed, to study the placebo response. It is recommended to study the determinants in this study but also focus on other determinants not analyzed in our study. Furthermore, future studies on neuroimaging and the neurochemical process of the placebo response could contribute to understand the placebo effect.

Net gain analysis An important question has always been how to translate the results from clinical trials to clinical practice. The overall goal of drug development and drug registration is an effective and adequate treatment for patients in clinical practice. It is a well-known fact that the results of clinical trials, characterized by an extensive range of in- and exclusion criteria, perfectly trained staff, and great efforts to keep patients under follow- up, cannot be achieved in daily practice because of several hurdles and difﬁculties (30). I examined the clinical relevance of the well-established responder analyses

PART 4 - CHAPTER 7 - Summary and discussion| 131 by comparing it with a new approach: ‘net gain analysis’ (chapter 4). The results showed that for the antipsychotic treatment of acute mania, the responder analysis is an adequate outcome measure. However, net gain analysis is a more precise and more comprehensive efﬁcacy measure.

I would like to note that this new approach of measuring effect should not be restricted to studies on the effect of antipsychotics in patients with an acute manic episode and that it can be applied to all drugs (somatic and psychiatric) with a limited effect size and a relatively high risk of deterioration and/or adverse events. For example, in this rapidly developing world, life expectancy is steadily increasing (31). However, a longer life expectancy also comes with an increased risk of chronic diseases (32, 33). For chronic diseases, known to frequently affect quality of life, treatment is most often not curative but based on symptomatic improvement and a better quality of life. In these chronic diseases, net gain analysis could be of great value, because treatment-related symptomatic improvements may go hand in hand with side effects reducing the quality of life and symptomatic improvements may not outweigh the side effects and thus the treatment may not result in a net gain in quality of life. In these cases one may doubt whether the efﬁcacy of the drug is clinically relevant.

In the treatment of physical disorders, the concept of net gain analysis has already been applied, e.g. carbasalate calcium (Ascal®), for the primary prevention of a myocardial infarction in patients with a coronary heart disease event risk of 0.5%/ year. Carbasalate calcium shows a number needed to treat (NNT) of 256 to prevent myocardial infarction but also has a number needed to harm (NNH) of 500 for non- minor bleeding complications. The net gain of carbasalate calcium in this patient group is not sufficient and therefore carbasalate calcium is not recommended in this patient group (34). The concept of net gain has also been applied previously in the treatment of an acute manic episode. After years and years of first choice treatment with haloperidol in patients with acute mania, it is now recommended to prescribe atypical antipsychotics (35). Haloperidol still shows good response rates but atypical antipsychotics have the advantage of a different adverse event profile (10), including a smaller risk of tardive dyskinesia and a higher risk of metabolic side effects (10, 36). The final decision can be considered as an interpretation of the results in terms of net gain between haloperidol and atypical antipsychotics.

132 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success These examples show that the concept of net gain has already been applied (implicitly and often without formal calculation) in daily clinical practice, based on the characteristics, wishes, perspective, and prognosis of the individual patient. However, it must be emphasized that the net gain analysis as deﬁned in my study, can only be used for the assessment of groups, since a patient cannot be a responder and a deteriorator at the same time. It is recommended to study the theoretical framework of the net gain analysis in several other (chronic somatic and psychiatric) disorders, to investigate whether the concept of net gain is a valid and relevant approach for the assessment of clinical efﬁcacy and clinical relevance by regulatory authorities.

CLINICAL QUESTIONS

Insight In chapter 5, I examined whether insight in one’s illness affects the efficacy of antipsychotic treatment in patients with an acute manic episode. Since impaired insight was associated with lower medication adherence (37) and poor clinical outcome (38-40) in patients with bipolar disorder in remission, I hypothesized that impaired or no insight in patients with acute mania would be associated with a smaller treatment effect. However, antipsychotic treatment in patients with impaired or no insight was more effective than in patients with excellent insight.

Even though baseline mania severity and BMI were associated with the level of insight, the relationship between insight and symptomatic outcome was not modified by illness severity at baseline, BMI, age or gender. However, the type of antipsychotic compounds seemed to modify the relation; a finding that could be related to the unequal distribution of different antipsychotics across the different geographic regions (chapter 2). I therefore analyzed whether this modifying effect of compound could be attributed to differences in efficacy between patients from the three regions. I stratified the patients into the three regions and within these strata the type of antipsychotic compound no longer modified the relation between the level of insight and symptomatic outcome.

The question now arises how patients having poor or no insight in their illness, could have participated in a clinical trial. From personal experiences as a resident at the crisis department of Amsterdam, I know the struggle to achieve treatment adherence in patients with acute mania and impaired or no illness insight. This was even shown

PART 4 - CHAPTER 7 - Summary and discussion| 133 in patients with bipolar disorder in remission (40). Furthermore, it can be a major challenge to obtain informed consent for trial participation from patients having poor or no illness insight. As expound before, this may explain why less US patients had poor or no insight. It would be interesting to study this question in future studies, thereby increasing the understanding of clinical trials.

In conclusion, my findings show that antipsychotic treatment is more effective in patients having impaired or no insight than in patients with excellent insight. Taking into account that the level of insight is correlated with illness severity and thus the risk of personal and social harm, I recommend that patients experiencing an acute manic episode should be treated immediately and that treatment should not be delayed until patients gain insight into their condition. An important question is whether and how this finding should influence our policy regarding forced antipsychotic treatment in severely manic patients with impaired or no insight, who are not willing to take medication. It seems that these patient really need medication and that they recover even faster from their manic symptoms, which in turn reduces their risk of personal and social harm. However, the friction between practical solutions and self-determination cannot be solved by just one empirical study showing good outcome even in those patients with no insight. Furthermore, the findings show the importance of subgroup analysis in patients with acute mania. For regulatory authorities, asking for subgroup analysis could be of great additional value to the assessment of efficacy. This could eventually also help the clinician to improve and apply personalized medicine based on individual patient characteristics.

Early non-response Unfortunately, in clinical practice, antipsychotics are not equally effective in all patients and personalized medicine is still in its infancy. Therefore the trial-and-error approach remains daily practice. This makes it important to detect non-response to antipsychotic treatment in an early stage and interrupt or reconsider unsuccessful treatment as soon as possible. I studied whether early non-response predicts antipsychotic treatment failure at study endpoint and tried to determine the most adequate criterion for early non- response to predict treatment failure, described by non-response and non-remission at study-endpoint (chapter 6). The current guidelines are inconsistent and often unclear in the recommendation of a

134 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success time frame to reconsider drug treatment in patients with acute mania. Moreover, there has been a notable omission of a time frame recommendation in the NICE guideline (2014) and in the APA guideline on bipolar disorders (35, 41). The Guideline of The World Federation of Societies of Biological Psychiatry (WFSBP) and the Dutch NVvP Guideline for Bipolar Disorders, however, do recommend a time frame of two weeks to reconsider medication (42, 43). This is remarkable, because most of the empirical literature recommends a time frame of only one week (44-46). In line with Ketter et al. (2010), Kemp et al. (2011), and Szegedi et al. (2012), I found early non-response at week one to be a signiﬁcant and clinical relevant predictor for treatment failure at endpoint (44-46). In my study I additionally studied early non-response at week two and found week two also to be a signiﬁcant and relevant predictor of treatment failure at endpoint.

At both points in time, the predictive value of non-response at study endpoint and the number of patients meeting criteria for non-response were linearly related to the level of the cut-off score. Based on these findings, it is understandable that the NICE and APA guidelines do not give any recommendation. It also justifies the recommendations of Ketter (2010), Kemp (2011), and Szegedi (2012) to reconsider medication after one week (44-46). Based on our findings, I recommend a two-stage strategy in which the clinician assesses the level of non-response at week one and week two and takes into account the need for a rapid improvement given the nature of the manic behavior and the presence or absence of a supporting network. For example, reconsidering medication already after one week has greater urgency in a severely manic patient having no insight in her illness, showing severe self-destructive behavior without anyone in her network to provide help and to support her, than in a patient having good insight in her manic symptoms, with an experienced and strong social network helping and guiding her.

Of note, the NICE, NVvP, and the WFSBP guidelines consistently recommend switching to an alternative antipsychotic compound when antipsychotic treatment shows lack of effect (35, 43, 47). However, little is known about which changes are likely to be most effective and how these changes should be implemented. Therefore, future prospective studies are needed to answer these questions in order to effectively and efﬁciently switch antipsychotic treatment of patients with acute mania.

PART 4 - CHAPTER 7 - Summary and discussion| 135 METHODOLOGICAL CONSIDERATIONS The unique opportunity to study the individual patient data (IPD) of 12 randomized controlled trials of medication versus placebo for the indication acute mania resulted in several strengths in our studies. First, due to the availability of IPD, the registration studies could be used to study a range of regulatory and clinical questions for which the data were originally not gathered. Also, potential explanatory variables could be analyzed and subgroup analyses could be performed. Additionally, the individual patient data enabled me to analyze confounding and modifying effects at the patient level and not only on a aggregated (group or study) level. This strengthens the clinical implications of our findings, providing more precise and accurate results. The opportunity of using (raw) data from RCTs to enable scientific research has been discussed and supported by deputies of several European regulatory agencies (48) and other academic parties (49, 50) As discussed by Eichler et al. (2012), bundling raw data and computing secondary analyses can create new research, increase medical knowledge and, as in our thesis, enables prediction analyses (48, 51). However, to make bundled secondary analyses general practice, several factors (e.g. patient confidentiality, financial conflicts of interest, guarantee of quality, ownership of data) must first be agreed on (48-50).

The second strength of this thesis is that I was able to study 3,207 patients with acute mania. This high number of patients gave power to my results and enhances the value of my ﬁndings. Furthermore, due to the high number of patients I had the opportunity to study subgroups. A broad range of subgroup analyses was possible, and I only encountered power limitations in the analyses of ethnic differences across geographic regions.

The final strength is that the type of compound did not modify the findings and therefore, I could derive conclusions for the group ‘drugs for the treatment of acute mania’ in the study regarding geographic differences and for the group ‘antipsychotics’ in the studies regarding net gain analysis, insight and early non-response. This way, the findings can be considered as a class-effect of antipsychotics in net gain analysis, insight and early non-response and could therefore help in the development of future guidelines.

Of course, the ﬁndings and conclusions of my thesis must also be interpreted in the light of several limitations. First, the sample of RCTs used for these studies could have

136 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success resulted in selection bias. However, I was able to include all studies that were offered to the MEB, regardless of the study design or the acceptance of registration of the compounds for the indication acute mania. The selection procedure in our study may also have resulted in a strength; publication bias was avoided. This is important since publication bias is an important bias in medical research, showing mainly positive studies published in medical journals.

Second, for confidentiality reasons regarding the regulatory dossiers, we were not able to report on specific information regarding the individual compounds that were included in the study (e.g. product name, study year, and trial name,). Therefore, compounds were referred to as compound A, compound B etc. This lack of transparency makes it hard to check or replicate my findings. Fortunately, the approach of the MEB and of the pharmaceutical companies on this matter has transitioned from confidentiality and blinding to clarity and transparency of data. As a consequence, the compounds and the studies could be disclosed in the last submitted article on insight (chapter 5).

A third limitation is that I conducted secondary analyses, which have the disadvantage of being post-hoc analyses. Post-hoc analyses are frequently criticized for squeezing out results, which could lead to statistical coincidence without clinical meaning (52, 53). However, as stated by Srinicas et al. (2015), a post hoc analysis could also be viewed as an “analysis with unique limitations and strengths that often raise new questions to be addressed in further trials” (53). Or, as proposed by Curran-Everett and Milgrom (2013), viewed “as good use of loads of potential clinical important information, originally gathered for other purposes, which otherwise would be lost” (52). Of course, secondary analyses can lead to the well-known ﬁshing effect by multiple comparisons, but in our study we used explicit hypotheses and Bonferroni corrections to counteract this potential problem (54).

Using registration studies for secondary analyses has resulted in the fourth limitation; I was limited in the variables that were available for analyses to explain the ﬁndings. Several other patient characteristics and study characteristics may have contributed to my ﬁndings. As for patient characteristics, e.g. duration of illness, time of inclusion, history of medication, history of hospitalization, reasons for participation, and social environment could have substantially contributed to my thesis.

PART 4 - CHAPTER 7 - Summary and discussion| 137 A ﬁnal limitation is that the unique opportunity to use the IPD of randomized controlled trials for studying clinical questions may have introduced selection bias; the results are based on a population of eligible patients that decided to participate in a clinical trial. Therefore, the study population may only represent a subgroup of patients that are highly motivated and rarely seen in clinical practice. This may have affected the extrapolation to clinical practice. As stated before, more research is needed to correctly translate the study population of randomized controlled trials to the patients encountered in regular clinical practice.

CONCLUSION

The findings in my thesis show that more knowledge is needed to adequately interpret the findings from clinical trials for drug registration. Drug treatment starts with the assessment of a clinical trial of a (new) drug for a particular indication. If we understand and improve the assessment of randomized controlled trials, the patient will eventually benefit. Moreover, as shown in my thesis, clinical trials for drug registration should not only be used to obtain market authorization, but they can and should also be used for regulatory science purposes. Bundled individual patient data of randomized controlled trials are of great value to medical research, and can provide powerful information. As shown in two of my studies, subgroup analysis could retrospectively provide important information on personalized medicine and improve efficiency in drug treatment. Therefore, the ambition expressed by many regulators, including the MEB, to facilitate full access to trial data after a regulatory decision has been made is fully endorsed.

138 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success REFERENCES

1. Nappo SA, Lafrate GB, Sanchez ZM. Motives for participating in a clinical research trial: a pilot study in Brazil. BioMed Central Public Health. 2013;13. 2. Agency EM: Clinical trials submitted in marketing-authorisation applications to the European Medicines Agency. 2013. 3. Agency EM: International Workshop on ethical and GCP aspects of the acceptance of clinical trials submitted in Marketing Authorisation Applications to EMA. 2010. 4. Trials TEMAWGoTCC: Reflection paper on ethical and GCP aspects of clinical trials of medicinal products for human use conducted in third countries and submitted in marketing authorisation applications to the EMA. . 2010. 5. Vieta E, Pappadopulos E, Mandel FS, Lombardo I. Impact of geographical and cultural factors on clinical trials in acute mania:lessons from a ziprasidone and haloperidol placebo-controlled study. International Journal of Neuropsychopharmacology. 2011;14:1017-1027. 6. Mattila T, Wohlfarth T, Koeter M, Storosum J, van den Brink W, de Haan L, Leufkens H, Denys D. Geographic variation in efficacy of atypical antipsychotics for the acute treatment of schizophrenia – An individual patient data meta-analysis. European Neuropsychopharmacology. 2014;24:1067-1077. 7. Grubb GS, Archer DF, Constantine GD. Differences between the United States and Europe in clinical trials of hormonal contraceptive efficacy. Obstetrics & Gynecology. 2008;111:63S-64S. 8. Lete I, Pérez de Arrilucea M, Rodriquez M, Belleo E. Efficacy, safety, and patient acceptability of the etonogestrel and ethinyl estradiol vaginal ring. Journal of Contraception. 2014;5:39-48. 9. FDA: General meeting on contraceptives. in Advisory committee briefing document FDA; 2007. 10. Cipriani A, Barbui C, Salanti G, Rendell J, Brown R, Stockton S, Purgato M, Spineli LM, Goodwin GM, Geddes JR. Comparative efficacy and acceptability of antimanic drugs in acute mania: a multiple- treatments meta-analysis. Lancet. 2011;378:1306-1315. 11. Depp CA, Harmell AL, Savla GN, Mausbach BT, Jeste DV, Palmer BW. A prospective study of the trajectories of clinical insight, affective symptoms, and cognitive ability in bipolar disorder. Journal ofAffectiveDisorders. 2014;152-154:250-255. 12. Aspiazu S, Mosquera F, Ibañez B, Vega P, Barbeito S, López P, Ruiz de Azúa S, Ugarte A, Vieta E, González-Pinto A. Manic and depressive symptoms and insight in first episode psychosis. Psychiatric Research. 2010;178:480-486. 13. Bryc K, Durand EY, Macpherson JM, Reich D, Mountain JL. The Genetic Ancestry of African Americans, Latinos, and European Americans across the United States. The American Journal of Human Genetics 2015;96:37-53. 14. Trussel J, Portman D. The creeping pearl: why has the rate of contraceptive failure increased in clinical trials of combined hormonal contraceptive pills? Contraception. 2013;88:604-610. 15. Ohmann C, Deimling A. Attitude towards clinical trials: Results of a survey of persons interested in research. Inflammation Research. 2004;53. 16. Burns KEA, Magyarody N, Jiang D, Wald R. Attitudes and views of the general public towards research participation. Internal Medicine Journal. 2013;43:531-540. 17. Carroll R, Antiqua J, Taichman D, Palevsky H, Forfia P, Kawut S, Halpern SD. Motivations of patients with pulmonary arterial hypertension to participate in randomized clinical trials. Clinical Trials. 2012;9:348-357. 18. Castillo AG, Jandorf L, Thélémaque LD, King S, Duhamel K. Reported Benefits of Participation in a Research Study. Journal of Community Health. 2012;37:59-64.

PART 4 - CHAPTER 7 - Summary and discussion| 139 19. Townsend A, Cox SM. Accessing health services through the back door: a qualitative interview study investigating reasons why people participate in health research in Canada. BioMed Central Medical Ethics. 2013;14. 20. Wendler D, Krohmal B, Emanuel EJ, Grady C. Why patients continue to participate in clinical research. Archives of Internal Medicine. 2008;168:1294-1299. 21. Chong S-A, Ong Y-Y, Subramaniam M, Abdin E, Marx CE, Campbell AV. An assessment of the understanding and motivations of patients with schizophrenia about participating in a clinical trial. Contemporary Clinical Trials. 2009;30:446-450. 22. Vieta E, Cruz N. Increasing rates of placebo response over time in mania studies. Journal of Clinical Psychiatry. 2008;69:681-682. 23. Yildiz A, Vieta E, Tohen M, Baldessarini RJ. Factors modifying drug and placebo responses in randomized trials for bipolar mania. International Journal of Neuropsychopharmacology. 2011;14:863- 875. 24. Agid O, Siu CO, Potkin SG, Kapur S, Watsky EJ, Vanderburg D, Zipursky RB, Remington G. Meta- Regression Analysis of Placebo Response in Antipsychotic Trials, 1970–2010. American Journal of Psychiatry. 2013;170:1335-1344. 25. Montgomery SA. The failure of placebo-controlled studies European Neuropsychopharmacology. 1999;9:271-276. 26. Keck JPE, Welge JA, McElroy SL, Arnold LM, al. e. Placebo effect in randomized, controlled studies of acute bipolar mania and depression. Biological Psychiatry. 2000;47:748-755. 27. Rothman KJ, Michels KN. The continuing unethical use of placebo controls. New England Journal of Medicine. 1994;331:394-398. 28. EMEA: Note for guidance on clinical invesitgation of medicinal products for the treatment and prevention of bipolar disorder. Edited by (CPMP) Cfpmp. London2001. 29. Hyland ME. Using the placebo response in clinical practice. Journal of Clinical Medicine. 2003;3:347-350. 30. Eichler H-G, Abadie E, Breckenridge A, Flamion B, Gustafsson LL, Leufkens H, Rowland M, Schneider CK, Bloechl-Daum B. Bridging the efﬁcacy–effectiveness gap: a regulator’s perspective on addressing variability of drug response. Nature Reviews Drug Discorvery. 2011;10:495-506. 31. WHO WB, UNESCO, CIA and individual country databases for global health and causes of death: http://ec.europa.eu/health/reports/european/health_glance_2014_en.htm. 2014. 32. RIVM: http://www.rivm.nl/Documenten_en_publicaties/Algemeen_Actueel/Nieuwsberichten/2014/ Toekomstverkenning_RIVM_Een_gezonder_Nederland_met_meer_chronisch_zieken. 2014. 33. Ward BW, Schiller JS, Goodman RA. Multiple Chronic Conditions Among US Adults: A 2012 Update. Preventing Chronic Disease. 2014;11:130389. 34. Sanmuganathan PS, Ghahramani P, Jackson PR, Wallis EJ, Ramsay LE. Aspirin for primary prevention of coronary heart disease: safety and absolute beneﬁt related to coronary risk derived from meta-analysis of randomised trials. Heart. 2001;85:265-271. 35. NICE: Bipolar disorder: the assessment and management of bipolar disorder in adults, children and young people in primary and secondary care in NICE clinical guideline 185, National Collaborating Centre for Mental Health; 2014. 36. Newcomer JW. Second-generation (atypical) antipsychotics and metabolic effects: a comprehensive literature review. CNS Drugs 2005;19:1-93. 37. Yen C-F, Chen S-F, Ko C-H, Yeh M-L, Yang S-J, Yen J-J, Huang C-F, Wu C-C. Relationships between

140 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success insight and medication adherence in outpatients with schizophrenia and bipolar disorder: Prospective study. Psychiatry and Clinical Neurosciences. 2005;59:403-409. 38. Braw Y, Sitman R, Sela T, Erez G, Bloch Y, Levkovitz Y. Comparison of insight among schizophrenia and bipolar disorder patients in remission of affective and positive symptoms: Analysis and critique. European Psychiatry. 2012;27:612-618. 39. Yen C-F, Chen C-S, Yeh M-L, Ker J-H, Yang S-J, Yen J-Y. Correlates of insight among patients with bipolar I disorder in remission. Journal of Affective Disorders. 2004;78:57-60. 40. Yen C-F, Chen C-S, Yen J-Y, Ko C-H. The predictive effect of insight on adverse clinical outcomes in bipolar I disorder: A two-year prospective study. Journal of Affective Disorders. 2008;108:121-127. 41. Hirschfeld RMA: Guideline Watch (November 2005): Practice Guideline for the Treatment of Patients With Bipolar Disorder, 2nd Edition. in The journal of lifelong learning in psychiatry. Arlington, VA, American Psychiatric Association.; 2007. 42. Nolen WA, Kupka RW, Schulte PFJ, Knoppert-van der Klein EAM, Honig A, Reichart CG, Goossens PJJ, Daemen P, Ravelli DP: Richtlijn bipolaire stoornissen; tweede, herziene versie, 2008. De Tijdstroom, Utrecht; 2008. 43. Grunze H, Vieta E, Goodwin GM, Bowden C, Licht RW, Möller H-J, Kasper S, disorders WTfotgfb. The World Federation of Societies of Biological Psychiatry (WFSBP) Guidelines for the Biological Treatment of Bipolar Disorders: Update 2009 on the Treatment of Acute Mania. The World Journal of Biological Psychiatry. 2009;10:85-116. 44. Ketter TA, Agid O, Kapur S, Loebel A, Siu CO, Romano SJ. Rapid antipsychotic response with ziprasidone predicts subsequent acute manic/mixed episode remission. Journal of Psychiatric Research 2010;44:8-14. 45. Szegedi A, Zhao J, McIntyre RS. Early improvement as a predictor of acute treatment outcome in manic or mixed episodes in bipolar-1 disorder: A pooled, post hoc analysis from the asenapine development program. Journal of Affective Disorders. 2012;150:745-752. 46. Kemp DE, Johnson E, Wang WV, Tohen M, Calabrse JR. Clinical Utility of Early Improvement to Predict Response or Remission in Acute Mania: Focus on Olanzapine and Risperidone. Journal of Clinical Psychiatry. 2011;72:1236-1241. 47. Nolen WA, Kupka RW, Schulte PFJ, Knoppert-van der Klein EAM, Honig A, Reichart CG, Goossens PJJ, Daemen P, Ravelli DP: Richtlijn bipolaire stoornissen - Tweede, herziene versie, 2008. Utrecht, Nederlandse Vereniging voor Psychiatrie; 2008. 48. Eichler H-G, Abadie E, Breckenridge A, Leufkens HGM, Rasi G. Open Clinical Trial Data for All? A View from Regulators PLOS Medicine. 2012;9. 49. Dosh iP, Jefferson T, Del Mar C. The imperative to share clinical study reports. PLOS Medicine. 2012;9. 50. Hrynaszkiewicz I, Altman DG. Towards agreement on best practice for publishing raw clinical trial data. Trials. 2009;10. 51. Selker HP, Ruthazer R, Terrin N, Grifﬁth JL, Concannon T, Kent DM. Random treatment assignment using mathematical equipoise for comparative effectiveness trials. Clinical and Translation Science. 2012;4:10-16. 52. Curran-Everett D, Milgrom H. Post-hoc data analysis: beneﬁts and limitations. Current Opinion in Clinical Immunology. 2013;13:223-224. 53. Srinivas TR, Ho B, Kang J, Kaplan B. Post Hoc Analyses - After the Facts. Transplantation. 2015;99:17-20. 54. Borenstein M, Hedges LV, Higgins JPT, Rothstein H: Introduction to Meta-Analysis, Wiley; 2009.

PART 4 - CHAPTER 7 - Summary and discussion| 141

PART 5Appendix

NEDERLANDSE SAMENVATTING

DE MEDICAMENTEUZE BEHANDELING VAN PATIENTEN MET ACUTE MANIE

HET BEGRIJPEN VAN KLINISCHE STUDIES EN EEN SUCCESVOLLE BEHANDELING

Nederlandse samenvatting

Dit proefschrift gaat over de medicinale behandeling van een acuut manische episode. Acute manie is een onderdeel van de bipolaire stoornis type 1 die wordt gekarakteriseerd door periodes van depressie, acute manie en periodes van remissie.

De acute manie is een ernstige en potentieel zeer schadelijke periode voor de patiënt en zijn omgeving. Patiënten in een manische episode voelen zich vaak buitengewoon goed, alsof ze de hele wereld aan kunnen. Vaak hebben ze weinig tot geen behoefte aan slaap, praten snel en aaneengesloten en zijn hierin moeilijk te onderbreken. Patiënten beschrijven vaak dat de gedachten door hun hoofd racen, ze snel zijn afgeleid en bevlogen met sociale, werkgerelateerde of seksuele activiteiten. Helaas zijn ze ook vaak excessief bezig met activiteiten die potentieel veel schade kunnen berokkenen, zoals sociale conﬂicten, het excessief uitgeven van geld wat tot ernstige schulden kan leiden en seksueel ongeremd gedrag. Een manische episode kan enkele weken duren. Wanneer de patiënt hiervan herstelt wordt hij of zij vaak geconfronteerd met de desastreuze gevolgen van zijn of haar gedrag tijdens de acute manie.

De bipolaire stoornis is een levenslange aandoening, die helaas niet kan worden genezen. Het doel van de behandeling is dan ook het voorkómen van een nieuwe manische of depressieve episode. Patiënten worden behandeld met geneesmiddelen zoals stemmingsstabilisatoren, antipsychotica en antidepressiva. Daarnaast krijgen ze frequent psycho-educatie of psychotherapie. Desondanks kan bij de meeste patiënten met een bipolaire stoornis een terugval naar een manische of depressieve episode niet worden voorkomen. Eenmaal in een manische fase is het zaak de patiënt hier zo snel mogelijk uit te krijgen om schade te voorkomen of te beperken. Hiervoor krijgt de patiënt medicijnen. Het eerste keus middel is een antipsychoticum. Wanneer dit niet werkt wordt overgestapt naar een ander antipsychoticum of wordt een ander medicijn aan het oorspronkelijk antipsychoticum toegevoegd.

Idealiter krijgt een acuut manische patiënt alleen geneesmiddelen voorgeschreven die zijn geregistreerd voor de indicatie acute manie. Om deze registratie te verkrijgen moet in een fase III klinische studie worden aangetoond dat het geneesmiddel

PART 5 - Nederlandse samenvatting| 147 effectief en veilig is en kwalitatief gezien aan alle eisen voldoet. Dit wil zeggen dat het geneesmiddel een signiﬁcant en klinisch relevant effect moet laten zien en dat dit effect moet opwegen tegen de risico’s die gebruik van het middel met zich meebrengt (bijwerkingen). Er moet sprake zijn van een positieve voordelen/nadelen balans. De beoordeling c.q. registratie van geneesmiddelen wordt uitgevoerd door regulatoire autoriteiten. In Nederland is dit het College ter Beoordeling van Geneesmiddelen (CBG). De beoordeling van een klinische studie op effectiviteit en veiligheid roept vaak veel discussie op binnen de autoriteit en bij Europese aanvragen tussen de verschillende autoriteiten; waren de patiënten die aan de studie meededen representatief voor de dagelijkse praktijk, waren de uitkomst maten waarmee effectiviteit werd aangetoond adequaat gekozen, was het gevonden effect klinisch relevant en wegen de effecten van het geneesmiddel op tegen de risico’s op bijwerkingen?

Dit proefschrift komt voort uit een samenwerking tussen het CBG en het Academisch Medisch Centrum (AMC). Via het CBG kreek ik toegang gekregen tot een unieke database met individuele patiënten data (IPD). De database bevatte 12 dubbelblind, gerandomiseerde, placebo-gecontroleerde klinische studies met patiënten gediagnosticeerd met een manische episode (N=3207). De ernst van de acute manie werd vastgesteld met de Young Mania Rating Scale (YMRS) en de Mania Rating Scale from the Schedule for Affective Disorders and Schizophrenia – Change Version (MRS van de SADS-C), waarbij een hogere score ernstigere manische symptomen beschrijft.

Deze database gaf mij de kans om dieper in te gaan op enkele onopgeloste regulatoire vragen rondom de beoordeling van klinische studies voor patiënten met een acute manie, om zo een beter inzicht in deze registratie studies te krijgen en de resultaten van deze studies beter te kunnen interpreteren. Omdat het regulatoire veld sterk overlapt met de klinische praktijk gaf de database mij tevens de mogelijkheid dieper in te gaan op enkele klinische vraagstukken zoals, is de effectiviteit van antipsychotica voor de behandeling van acute manie afhankelijk van het zieke inzicht van de patiënt en is vroegtijdig niet aanslaan van antipsychotica een goede voorspeller voor het uiteindelijk falen van behandeling met antipsychotica?

148 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success REGULATOIRE VRAGEN

De 21e eeuw wordt gekenmerkt door globalisering op alle vlakken van de maatschappij. Zo ook bij de ontwikkeling van nieuwe medicijnen. Dit heeft als gevolg dat regulatoire organen als het CBG in Nederland en de Food and Drug Administration (FDA) in de Verenigde Staten (VS), nu vaak de resultaten van klinische studies moeten beoordelen die zijn uitgevoerd in een ander deel van de wereld, met patiënten die mogelijk sterk verschillen van de patiënten uit hun eigen contreien. Echter, er zijn mondiale verschillen in patiënten eigenschappen, gezondheidssystemen, omgevingsfactoren en culturen. Het is daarom maar de vraag of resultaten van een studie uit de ene regio ook gelden voor een andere regio. Dit is vooral een belangrijke vraag in de psychiatrie, waar omgevings- en culturele factoren van grote invloed kunnen zijn. Daarom is in hoofdstuk 2 onderzocht of er geograﬁsche verschillen tussen drie regio’s (Europa, VS en Overig) zijn in de effectiviteit van medicatie voor patiënten met een acute manie en of deze verschillen kunnen worden verklaard door verschillen in patiëntkenmerken tussen patiënten van de verschillende regio’s of door verschillen in placebo response binnen de drie regio’s.

Er waren signiﬁcante verschillen in de effectiviteit van geneesmiddelen tussen de regio’s waarbij onderzoeken bij patiënten uit Europa en de regio Overig een substantieel groter effect lieten zien dan onderzoeken bij patiënten uit de VS. Dit gold zowel voor effect in termen van gemiddelde verbetering op de ernstschaal als succes in termen van percentage ‘responders’ waarbij iemand ‘responder’ is als zijn/haar ernstscore met ten minste 50% is gedaald tussen begin en eind van de studie. Echter, ondanks dat er signiﬁcante verschillen waren tussen de drie regio’s op basis van patiënten kenmerken (leeftijd, geslacht, BMI, etniciteit, ernst van de ziekte), placebo response en het percentage patiënten die voortijdig stopte met de studie, konden de resultaten slechts deels worden verklaard; de patiënten in de VS waren minder ernstig ziek dan patiënten in Europa en dit zou het verschil in succesvol behandelde patiënten (responders) tussen deze twee regio’s kunnen verklaren.

Voor de medicatie voor acute manie hebben onze bevindingen weinig directe gevolgen. De medicijnen zijn, ondanks de grote verschillen, in alle regio’s effectief. Onze bevindingen zijn echter in lijn met bevindingen in studies bij schizofrene patiënten,

PART 5 - Nederlandse samenvatting| 149 anticonceptie en cardiovasculaire aandoeningen, waarbij ook een lagere effectiviteit van geneesmiddelen werd gevonden bij patiënten uit de VS. Voor medicijnen geïndiceerd voor minder ernstige ziektes, medicijnen met lagere effectiviteit en hogere kans op bijwerkingen, of voor medicijnen voor chronisch gebruik, kunnen onze bevindingen echter wél gevolgen hebben. Hier zouden lagere effect groottes in de VS misschien voor een negatieve beneﬁt/risk balans kunnen zorgen. Daarom zullen het CBG en de FDA altijd opnieuw moeten inschatten of bevindingen kunnen worden geëxtrapoleerd over regio’s of dat evidentie uit de eigen regio vereist is. Meer onderzoek naar verklaringen voor deze regionale verschillen zou de inschatting kunnen onderbouwen.

Naast regionale verschillen in effectiviteit van medicijnen is bij klinische studies in de psychiatrie ook sprake van effect verschillen tussen studies en grote verschillen in effectiviteit van interventies tussen patiënten. Dit laatste resulteert in een hoog percentage studies die falen de effectiviteit van een geneesmiddel aan te tonen, hoger dan in de meeste andere gebieden van de geneeskunde. Aangezien de ontwikkeling van een medicijn zeer hoge kosten met zich meebrengt heeft het mislukken van een fase III studie grote ﬁnanciële consequenties. De lage slagingskans heeft tot gevolg dat er minder medicatie ontwikkelingsonderzoek wordt gedaan binnen de psychiatrie, hetgeen uiteindelijk ten koste gaat van de psychiatrische patiënt.

Het lage succespercentage van fase III studies in de psychiatrie wordt vaak geweten aan de relatief hoge placebo response. Het effect van een geneesmiddel wordt namelijk vaak bepaald ten opzichte van het effect van placebo. Daarom is in hoofdstuk 3 onderzocht of de hoogte van de placebo response samenhangt met de effectgrootte in klinische studies bij patiënten met acute manie. Welke patiënt- en studie kenmerken hangen samen met de hoogte van de placebo response en kunnen we op basis van deze kenmerken de placebo response voorspellen. De resultaten bevestigden de hypothese; een hoge placebo response was sterk geassocieerd met een kleiner medicatie effect in een studie. Vijf factoren (studie- en patiënten kenmerken) hingen samen met een hoge placebo response: ernst van manische symptomen bij het begin van de studie, afwezigheid van psychotische symptomen, recentere studies, inclusie van patiënten uit drie versus één regio en inclusie van patiënten uit de regio Overig versus patiënten uit de VS of Europa. Toch kon het meest krachtige voorspellingsmodel slechts 5 tot 10% van de placebo response voorspellen. Dit betekent dan ook dat het

150 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success beperken van studies tot bepaalde regio’s en bepaalde typen patiënten, in studies voor acute manie, geen efﬁciënte strategie is om de effectiviteit van een geneesmiddel beter aan te kunnen tonen.

In de psychiatrie zouden de relatief grote verschillen tussen studies in het effect van een geneesmiddel ook het gevolg kunnen zijn van methodologisch aspecten. In kortdurende studies naar de medicamenteuze behandeling van acute manie wordt effectiviteit gedeﬁnieerd in termen van verschillen in YMRS of MRS scores tussen baseline en het einde van de studie (mean change score, MCS), en in termen van het percentage succesvol behandelde patiënten (response rate, RR), waarbij succesvol wordt gedeﬁnieerd als ten minste 50% verbetering ten opzicht van het begin van de studie. Echter, van medicijnen voor acute manie is bekend dat ze minder dan 60% kans op een succesvolle behandeling geven. Er is dus een grote groep (≥40%) die onvoldoende verbeterd op de medicatie, überhaupt niet op de medicijnen reageert of zelfs verslechtert. Misschien is het zinvol om de negatieve gevolgen van medicatiegebruik ook in de effectiviteitsmaat te betrekken, immers, het uitblijven van effect op een geneesmiddel in een acuut manische episode leidt tot continuatie van schadelijk gedrag. Effectiviteit van medicijnen voor acute manie zou bijvoorbeeld kunnen worden uitgedrukt in termen van de kans op ‘netto winst’. Dat wil zeggen de kans op een succesvolle behandeling minus de kans dat een patiënt überhaupt niet reageert op het medicijn of verslechterd. In hoofdstuk 4 heb ik de uitkomsten van de effectiviteits-analyse gebaseerd op de ‘netto winst’ maat in patiënten met acute manie en vergeleken met de uitkomsten op de gebruikelijke effectmaat, gebaseerd op de responder analyse. Voor acute manie blijkt de ‘netto winst’ maat een iets preciezere en veelomvattendere effect maat dan de gebruikelijke effectmaat. Het concept van netto winst zou echter niet moeten worden beperkt tot de acute manie maar worden getoetst bij andere psychiatrische ziektebeelden en somatische ziektes, waarbij de kans op verslechtering of bijwerkingen moet worden meegenomen in de bepaling of een medicijn effectief is.

KLINISCHE VRAGEN

Wanneer een antipsychoticum wordt geregistreerd voor de indicatie acute manie, is het klinisch effectief en weegt de kans op bijwerkingen op tegen de effecten van

PART 5 - Nederlandse samenvatting| 151 het medicijn. De arts kan het antipsychoticum nu aan zijn of haar acuut manische patiënt voorschrijven. Bij de meeste patiënten met acute manie is het ziekte-inzicht verminderd, beperkt of geheel afwezig. Aangezien een manische episode veelal met antipsychotica wordt behandeld is het belangrijk te weten of het niveau van ziekte- inzicht de effectiviteit van het antipsychoticum beïnvloed. Als beperkt ziekte-inzicht de effectiviteit van antipsychotica negatief zou beïnvloeden, zou de arts bij binnenkomst van de patiënt kunnen overwegen om naast het antipsychoticum adjuvante medicatie te starten en/of psycho-educatie of psychotherapie in te zetten. In hoofdstuk 5 van dit proefschrift is daarom onderzocht of het niveau van ziekte-inzicht de effectiviteit van antipsychotica beïnvloed. Dit bleek inderdaad het geval. Echter, tegen de verwachting in vonden we dat de antipsychotica effectiever waren bij patiënten met beperkt of geen ziekte-inzicht dan bij patiënten met perfect ziekte-inzicht. Deze resultaten suggereren dat patiënten met acute manie en een beperkt of afwezig inzicht in hun ziekte meteen antipsychotica moeten krijgen en dat er niet moet worden gewacht tot adequaat ziekte- inzicht is bereikt. Deze bevindingen zouden toekomstige richtlijnen kunnen helpen in hun beleid voor deze patiëntengroep.

Door de bescheiden kans op een succesvolle behandeling door het eerste middel van keus, een antipsychoticum, en de sterk verschillende effecten tussen patiënten onderling, blijft de behandeling van de acute manie een kwestie van ‘trial and error’; sommige patiënten zullen succesvol reageren, anderen niet. Daarom is het belangrijk te weten welke patiënten niet genoeg op het antipsychoticum zullen reageren, zodat er adequaat maatregelen kunnen worden genomen om de patiënt zo veel mogelijk kans op snelle verbetering te geven. In hoofdstuk 6 heb ik onderzocht of het niet-reageren op antipsychotica na 1 en 2 weken, met niet-reageren gedefinieerd als ≤10% tot ≤50% verbetering van symptomen ten opzicht van baseline, een voorspeller is voor het falen van de behandeling aan het einde van de studie. Tevens heb ik gekeken of ik een optimaal criterium kon definiëren (combinatie van tijdsframe en afkappunt voor niet- reageren) voor heroverweging van de behandeling. Niet-reageren op week 1 en week 2 bleek een klinisch significante voorspeller voor falen van behandeling (gedefinieerd als niet-reageren of niet-in-remissie) op het einde van de studie. Echter, door een lineair verband tussen afkappunt voor niet-reageren, week 1 of week 2 (niet-reageren) en week 1 (niet-in-remissie) en de kans op het falen van behandeling, kon geen optimaal criterium worden bepaald. Op basis van onze bevindingen adviseren we bij niet-

152 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success reageren op week 1 of week 2 de behandeling vóór de derde week te heroverwegen op basis van een twee-staps-beleid, waarbij het niveau van niet-reageren op week 1 en week 2, de noodzaak voor snelle verbetering op basis van inhoud van het manische gedrag, en de aan- of afwezigheid van een adequaat ondersteunend netwerk, moet worden meegenomen in de overweging.

CONCLUSIE

Dit proefschrift laat zien dat er voor de interpretatie en beoordeling van klinische studies voor het registreren van geneesmiddelen meer inzicht nodig is. De medicamenteuze behandeling van de patiënt begint met de beoordeling van de uitkomsten van de fase III studie voor registratie. Als we ons inzicht in deze studies kunnen vergroten en de beoordeling kunnen verbeteren, zal de patiënt uiteindelijk proﬁteren. Nog belangrijker, zoals ik hebben laten zien in dit proefschrift, zouden klinische studies voor de registratie van een geneesmiddel niet alleen voor het verkrijgen van marktbevoegdheid moeten worden gebruikt, maar ook voor onderzoeksdoeleinden. Wanneer de individuele patiënten data van de klinische studies worden gebundeld kunnen ze een unieke database vormen die door de hogere power veel waarde kan toevoegen aan de medische wetenschap. Het zou zonde zijn deze data na verkrijgen van marktbevoegdheden verloren te laten gaan. Zoals aangetoond in dit proefschrift kunnen subgroep analyses retrospectief belangrijke informatie voor ‘personalized medicine’ opleveren en kan medicatie op basis hiervan effectiever worden ingezet. Farmaceutische bedrijven zouden moeten worden gestimuleerd om hun ruwe data (al dan niet anoniem) beschikbaar te stellen voor wetenschappers en samenwerken om het gebruik en de effectiviteit van geneesmiddelen in de klinische praktijk te verbeteren.

PART 5 - Nederlandse samenvatting| 153

DANKWOORD

Dankwoord

Dit proefschrift had nooit tot stand kunnen komen zonder de hulp van velen. Het waren twee intensieve, leerzame, spannende maar ook turbulente jaren. Ik combineerde mijn promotieonderzoek eerst met een parttime baan als klinisch beoordelaar bij het College ter Beoordeling van Geneesmiddelen en later met een opleidingsplek tot psychiater in het AMC. Graag zou ik van deze gelegenheid gebruik willen maken om iedereen die mij zo gesteund heeft, te bedanken.

Om te beginnen, mijn promotores prof. dr. D.A.J.P. Denys en prof. dr. H.G.M. Leufkens. Beste Damiaan, als mijn promotor van het AMC wist u mij telkens weer op scherp te stellen door het vanzelfsprekende te bekritiseren, en de gouden standaard niet zomaar voor lief te nemen. Uw prachtige manier van spreken over wetenschap en de psychiatrie zijn inspirerend en maken mij elke keer weer zo ongelooﬂijk trots op het vak. Heel veel dank voor uw begeleiding en vertrouwen. Beste Bert, u was mijn promotor van het CBG maar ook de voorzitter van het College. U gaf mij veel vrijheid tijdens het schrijven van mijn proefschrift maar was altijd bereikbaar voor vragen, gaf uw input en kritiek op de belangrijke momenten. Tijdens de College vergaderingen gaf u mij telkens weer met een grote glimlach het woord voor de verdediging van mijn rapport, waarbij de ernst van de vergadering en de luchtigheid van uw persoonlijkheid een combinatie vormen die ik niet snel zal vergeten. Ik waardeer uw aandacht voor de persoon achter de promovendi ontzettend.

Beste copromotores dr. M.W.J. Koeter en dr. T.D. Wohlfarth. Beste Maarten, jouw foto had op de cover van mijn proefschrift moeten staan. Als copromotor vanuit het AMC was jij mijn leermeester in de statistiek maar evenzo mijn leermeester in de wiskundige logica van het leven; aardappelen uit de magnetron zijn gezonder dan wanneer je ze kookt. Samen hebben we uren achtereen aan SPSS gezeten, jij met extra slappe kofﬁe en ik met extra sterke kofﬁe. Als ik je kamer binnenstormde, namen we eerst de tijd om op eenzelfde energielevel te komen waarna we met veel lol bergen werk verzetten. Dank voor jouw privé colleges statistiek, maar vooral voor het plezier waarmee we samen zo intensief samen hebben gewerkt. Herr Koeter, zonder jou was dit proefschrift er nooit geweest en daar ben ik je voor altijd dankbaar voor.

PART 5 - Dankwoord| 157 Beste Tamar, als copromotor vanuit het CBG, was jij voor mij de moeder van de begeleidingsgroep, mijn steunpunt op vele vlakken; door jouw gevoel voor taal, onze gesprekken in de glazen huisjes van het CBG, over de telefoon en bij jullie thuis was jij een steunpilaar in moeilijke periodes binnen het promotie-traject en in mijn privéleven. Heel veel dank daarvoor.

Beste prof. dr. W. van den Brink, beste Wim. Jouw onuitputtelijke energie en passie, van ’s morgens vroeg tot diep in de nacht, jouw onwaarschijnlijk snelle en uitgebreide commentaren op mijn stukken, jouw treffende opmerkingen en vooral warme persoonlijkheid waren ongekend en daar ben ik je ontzettend dankbaar voor. Jij en Tamar waren een ware steun tijdens de turbulente laatste fase van mijn proefschrift.

Beste dr. J.G. Storosum, beste Jitschak. Als opleider van de psychiatrie in het AMC wist jij mij bij onze eerste kennismaking zó uit te dagen dat ik besloot aan dit promotieonderzoek te beginnen. Jij leerde mij al snel minder gebruik te maken van superlatieven ‘want dat in combinatie met het feit dat ik een hockeymeisje was kwam mij niet ten goede’ (helaas zie je daar in dit dankwoord weinig van terug). Ik dank je voor jouw ervaren klinische maar zeker ook regulatoire blik die zorgde voor treffende vraagstellingen maar zeker ook voor jouw grote steun in de afrondende fase van mijn proefschrift. Ik verheug me erop de komende jaren nog veel van je te leren.

Beste dr. C.C. Gispen - de Wied, beste Christine. Als hoofd van de afdeling Wetenschap van het CBG was jij meer op de achtergrond betrokken maar altijd bereid om mee te denken en mijn stukken van commentaar te voorzien. Dank daarvoor.

Beste collega’s van de AIAR gang, beste Judy, Jochem, Masha, Kim, Suzanne, Tim, Renee, Ruth, Minnie, Anne-Marije, Mieke en Maaike wat hebben we lopen keten. Als vreemde eend op jullie gang voelde ik me ongelooﬂijk thuis en ik kijk met een grote glimlach terug op dat intensieve jaar.

Beste collega’s van het CBG en in het bijzonder beste Cristel, Tamar, Lieke, Maartje, Liesbeth, André, Violetta, Gilbert, Marco, Joost en Sipko, de overgang van de kliniek naar het CBG was groot maar jullie hebben mij hier met veel geduld in begeleid. Ik heb ontzettend veel van jullie geleerd.

158 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success Beste dames van het secretariaat Renske en Marianne, al moeten jullie soms echt wel eens gek van mij zijn geworden, jullie waren altijd bereid om te helpen. Veel dank daarvoor.

Beste AIOS collega’s van het AMC, tijdens het afronden van mijn proefschrift heb ik jullie leren kennen als warme, ambitieuze en inspirerende collega’s en ik ben dankbaar dat ik samen met jullie het vak mag leren.

Beste Debbie, de vormgeving van dit boekje had niemand mooier en enthousiaster kunnen doen. Je bent een groot talent en vooral een heerlijk persoon. Veel dank daarvoor.

Beste paranimf Karlijn, 24 jaar geleden werden we Karlijn en Carlijn en dat zal altijd zo blijven. Vanuit Heerlen gingen we samen naar Utrecht en daar kozen we allebei voor de studie Geneeskunde. Hoewel we verder verschillende paden bewandelen blijf jij altijd mijn weg naar huis. Voor jouw steun tijdens de afgelopen twee intensieve jaren ben ik je heel dankbaar en daarom ben ik blij en trots dat jij als mijn paranimf naast mij zult staan. Beste paranimf Ewoud, eens mijn oudste broer altijd mijn oudste broer. Door jouw warme karakter, jouw open blik en gevoel voor humor is het een feest om bij jou te zijn. Ik weet dat je er altijd onvoorwaardelijk voor mij zult zijn en daarom ben ik trots dat jij tijdens mijn verdediging naast mij staat.

Beste Elsemiek, als er een derde paranimf zou mogen zijn dan was jij dat zonder twijfel geworden. Als er de afgelopen twee jaar geklaagd moest worden wilde jij altijd luisteren. Vanaf de dag dat ik jou bij Nederlands Jeugd ontmoette, draag ik jou op handen en ben ik ongelooﬂijk trots op onze bijzondere vriendschap. Ik weet zeker dat ik de rest van mijn leven, dag in dag uit, met jou van het leven een feest zal maken.

Beste vriendinnen van jaarclub Tjop en in het bijzonder lieve Ninah, Veer, Floor, Roos en Maartje, dat jullie als leken mijn onderzoek op de voet volgden, is mij ontzettend dierbaar en een bevestiging van onze vriendschap.

Beste huisgenoten van de Koningslaan, al waren jullie tijdens het schrijven van dit proefschrift meer op de achtergrond betrokken, onze vriendschap is een prachtig goed

PART 5 - Dankwoord| 159 en ik weet zeker dat deze voor altijd zal blijven bestaan.

Beste Linde, jouw oprechte blijdschap bij elke behaalde mijlpaal van dit proefschrift was aanstekelijk en gaf altijd veel energie, ik ben ontzettend blij met onze vriendschap.

Beste teamgenoten van Hurley dames 3 en in het bijzonder Julie, Annebet, Wieke, en Marjolein, na een dag schrijven gaat er niets boven een avond sporten en grappen maken met vriendinnen. Jullie zijn een mooi stel, ongelooﬂijk ambitieus, ongelooﬂijk sportief en dat alles met heel veel plezier, een prachtige combinatie waar ik me ontzettend bij thuis voel.

Lieve familie, lieve mama, papa, Ewoud, Johanna en Valérie, Gijs, Heleen, Maes & Johanna, Pieter en Freija. Lieve mama, er is nog geen dag voorbij gegaan zonder jouw onvoorwaardelijke steun en support. Jij was degene die mij steunde en aanmoedigde om voor de psychiatrie te kiezen en daar ben ik je ontzettend dankbaar voor. Ik geniet ervan dat ik mijn liefde en fascinatie voor dit vak met jou kan delen. Ik ben ongelooﬂijk trots op jou, je bent een voorbeeld voor de moeder die ik ooit hoop te worden. “Een proefschrift schrijven is doorzetten” zei jij ooit papa en dat heb ik elke dag van de afgelopen twee jaar geweten. Mijn enorme daadkracht is een karaktereigenschap die ik ook altijd zo sterk bij jou terug zie. Lieve Ewoud, Johanna en jullie kleine Valérie, op zondagmiddag even snel langs ﬁetsen en je weet zeker dat je pas een uur later weg gaat. Jullie deur staat altijd open voor een glas wijn en een goed gesprek. Jullie zijn familie mensen in hart en nieren en combineren dat op een prachtige manier met jullie bijzondere ambitie. Lieve Gijs, Heleen en jullie Maes en Johanna, met twee artsen aan het roer weten jullie op bewonderingswaardige wijze werk, vrienden en het familie leven in balans te houden. Gijs, jij was mijn voorbeeld als een van de jongste promovendi ooit en ik weet nog hoe trots ik op jouw verdediging naar jou keek. Nu kijk ik trots naar hoe jij van je gezin houdt en ben jij opnieuw een voorbeeld voor mij. Lieve Pieter en Freija, tijdens ons dinertje met spaghetti, bier en wijn stelden jullie mij de meest logische vragen over mijn proefschrift waardoor ik later kon gaan zitten en dit boekje kon gaan schrijven. Pieter, jij bent de lijm van ons vieren, een familie mens met een ontzettend groot hart waar je altijd bij terecht kan. Doordat onze levens altijd zo op elkaar leken door hockey bij Nederlands jeugd, Utrecht, en gemeenschappelijke vrienden heb ik je van zo veel kanten zo goed leren kennen. Ik ben trots op jouw drive

160 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success en prachtige carrière maar niets verslaat je onnavolgbare dance moves. Lieve familie van Thomas, lieve Agnes, Bous, Dorine, Ivo en Anne-marie, heel veel dank voor het warme nest waarin ik al jaren zo graag kom. Jullie natuurlijke rust was soms het perfecte en noodzakelijke intermezzo in deze twee intensieve jaren.

Lieve Thomas, jouw grote en onbevooroordeelde hart leert mij elke dag dat het niet gaat om wat je bereikt maar om hoe en met wie je het bereikt. Ik kan je dan ook met trots vertellen dat de mensen met wie ik samenwerkte dit boek tot een prachtboek hebben gemaakt. Zelfs in deze twee jaren van keihard werken liet jij mij elke dag weer zien waar het in het leven om gaat; liefde, vriendschap, en ultiem van elkaar genieten. Ik hou ongelooﬂijk veel van jou, jij bent de motor van mijn veel te overdadige energie en ik droom er elke dag van samen oud te worden en met onze oude, gerimpelde blote billen over het strand te lopen. When you smile at the world, the world smiles at you.

PART 5 - Dankwoord| 161

PORTFOLIO

Portfolio

Name Carlijn C.M. Welten Institutions Medicines Evaluation Board (MEB) Academic Medical Center (AMC) PhD period November 2013 – October 2015 Promotores Prof. D.A.J.P. Denys and prof. H.G.M. Leufkens Co-promotores Dr. M.W.J. Koeter and dr. T.D. Wohlfarth Other members Prof. W. van den Brink, dr. J.G. Storosum, dr. C.C. Gispen-de Wied

PUBLICATIONS Accepted/published

Carlijn CM Welten, MWJ Koeter, TD Wohlfarth, JG Storosum, W van den Brink, CC Gispen-de Wied, HGM Leufkens, DAJP Denys. Efﬁcacy of drug treatment for acute mania differs across geographic regions: An individual patient data meta-analysis of placebo-controlled studies. Journal of Psychopharmacology. 2015; 1-10.

Carlijn CM Welten, MWJ Koeter, TD Wohlfarth, JG Storosum, W van den Brink, CC Gispen-de Wied, HGM Leufkens, DAJP Denys. Placebo response in antipsychotic trials of patients with acute mania: Results of an individual patient data meta-analysis. Eur Neuropsychopharmacol. 2015 Jul;25(7):1018-26.

Carlijn CM Welten, MWJ Koeter, TD Wohlfarth, JG Storosum, W van den Brink, CC Gispen-de Wied, HGM Leufkens, DAJP Denys. Net gain analysis, an addition to responder analysis - The case of antipsychotic treatment of acute mania. Regul Toxicol Pharmacol. 2015 Jul 8;73(1):227-231

Carlijn CM Welten, MWJ Koeter, TD Wohlfarth, JG Storosum, W van den Brink, CC Gispen-de Wied, HGM Leufkens, DAJP Denys. Does insight affect the efﬁcacy of antipsychotics in acute mania? Accepted for Journal of Clinical Psychopharmacology.

Carlijn CM Welten, MWJ Koeter, TD Wohlfarth, JG Storosum, W van den Brink, CC Gispen-de Wied, HGM Leufkens, DAJP Denys. Early non-response in the antipsychotic treatment of acute mania; a criterion for reconsidering treatment? Accepted for Journal of Clinical Psychiatry.

PRESENTATIONS

Oral presentations Karakteristieken van klinische trials in patiënten met een bipolaire stoornis - February 2015 Wetenschapsdag, Medicines Evaluation Board, Utrecht Geographic differences and the placebo response in clinical trials for November 2014 patients with acute mania - Medicines Evaluation Board, Utrecht

Poster presentation Efﬁcacy of drug treatment for patients with acute mania differs across September 2014 geographic regions - Figon Dutch Medicines Day, Ede

PART 5 - Portfolio| 165 WORK

Resident psychiatry (AIOS) April 2015 - Current Academic Medical Center (AMC), Amsterdam Clinical assessor, department of psychiatry April 2014 - April 2015 Medicines Evaluation Board, Utrecht Resident psychiatry (ANIOS), psychiatric crisis department of Amsterdam October 2013 - April 2014 Arkin, Spoedeisende Psychiatrie Amsterdam, Amsterdam

(Inter)national Conferences European College for Neuropsychopharmacology (ECNP) Congress October 2014 Berlin, Germany Najaarssymposium Bipolaire Stoornissen September 2014 Utrecht

166 | DRUG TREATMENT FOR PATIENTS WITH ACUTE MANIA Understanding clinical trials and treatment success PART 5 - Portfolio| 167

ABOUT THE AUTHOR

About the author

Carlijn Welten was born in Utrecht on the 30th of October 1987. She grew up in Heerlen and in 2005 she graduated from the Gymnasium at the Bernardinus College. Carlijn studied Medicine at the Utrecht University (UU). During secondary school and her studies Carlijn played for the Dutch Field Hockey Team under 16, under 18, and under 21. In 2006 Carlijn was elected to play for the Dutch field hockey team. She won four medals and played her last tournament for the Dutch Team in the summer of 2011. Carlijn finished Medicine after her final elective clinical rotation at the closed department for 12-18 year olds (department Panama) of the Bascule in Amsterdam. She graduated as a medical doctor at the Utrecht University in the summer of 2013. After graduation she went to New Zealand to play the National Hockey League. In October 2013 she started as a resident not in training at the psychiatric crisis department of Amsterdam (Arkin, Spoedeisende Psychiatrie). Meanwhile, Carlijn started her PhD project on the pharmacotherapy of the acute manic episode of bipolar disorder. In April 2014 she started a dual job as a clinical assessor at the Medicines Evaluation Board (MEB) in Utrecht and PhD student at the Academic Medical Centre (AMC), University of Amsterdam (UvA). Her research project was under supervision of prof. dr. H.G.M. Leufkens (MEB) and prof. dr. D.A.J.P. Denys (AMC, UvA). In April 2015 she started her psychiatry residency program at the AMC under supervision of prof. dr. Damiaan Denys.

PART 5 - About the author| 171