Synergy in Action

ISCTM 2012 Policy Forum Summary

ISCTM 8th Annual Meeting, Washington, DC
Tuesday, 21 February 2012
National Mental Health Research to Policy Forum
“Are Clinical Trials of New Treatments for Chronic CNS Diseases Early Enough and Long Enough?”
Co-Chairs: Reuven Ferziger, MD; Larry Ereshefsky, PharmD; Ron Manderscheid, PhD

Dr. Ereshefsky opened the session, stating that the goal was to review strategies for improving the design and conduct of long-term clinical trials in CNS diseases.  The focus of the session was on Alzheimer’s disease (AD), which was considered to be very timely in light of the recent challenge from President Obama to have effective treatments by 2025, and schizophrenia, a chronic disease with a long-term course and high healthcare costs.  Dr. Manderscheid noted that it was hoped that the ideas presented at this session could form the basis for policy paper with recommendations for changing the approach to research in schizophrenia and AD. 

Part 1: Framing the Issues
Dr. Goldman discussed the needs that policy makers have for data from long-term studies in CNS diseases such as AD and schizophrenia evaluating different treatment interventions.  The Mental Health Treatment Study (Social Security Administration), evaluating the effects of better access to treatment and employment services in patients with schizophrenia or affective disorders, and the Recovery After an Initial Schizophrenia Episode study (RAISE; NIMH), evaluating the effect of early intervention in schizophrenia were cited as examples of government sponsored long-term studies.  Issues associated with early treatment and long-term follow-up can impact the following policy areas: disability (when to re-evaluate), vocational (ability to return to work), labor, educational (ability to return to school), healthcare finance (cost/benefit), and housing. 
Dr. Manderscheid discussed the epidemiology and policy goals in the treatment of AD and schizophrenia.  Long-term follow-up is needed on patients who are released from psychiatric hospitals to collect information on treatment and outcomes.  Ideally this should be for 30-50 years; however, this is generally not feasible.  Unfortunately, case registries used for tracking these patients have been largely abandoned due to lack of funding.  
Dr. Heinssen gave the perspective on schizophrenia and emphasized the need to identify new targets in treating schizophrenia and the importance of early intervention.  In the U.S., initiation of treatment in schizophrenia is often delayed 1-3 years after the first episode, despite evidence that delaying treatment leads to a worse long-term outcome. The current model for managing chronic schizophrenia is inefficient, resulting in high rates of unemployment among patients with schizophrenia and annual economic costs in excess of $60 billion.  Long-term studies evaluating outcomes in patients who receive early intervention need to be conducted.  Initiatives from the NIMH to evaluate early intervention strategies include the North American Prodrome Longitudinal Study (NAPLS), looking at the prodromal state and factors that influence conversion to psychosis, and the RAISE study.  To date, very few prevention trials in prodromal schizophrenia have been performed, with only a half dozen randomized, controlled trials (RCTs) being conducted to evaluate medications, along with cognitive behavioral therapy, in reducing the transition to psychosis over a 12 month period. 
Dr. Anand provided the perspective on the need for earlier intervention and long-term trials in AD.  Currently available treatments for AD are approved for use during the mild to severe stages of the disease; however, behavioral symptoms can be identified 2-3 years before the diagnosis is made, and pathological changes in the brain may occur up to 10 years prior.  The worldwide prevalence of dementia is currently 25 million and is expected to reach 80 million by 2040 if no treatment is available to halt or delay the progression of the disease.  Current costs of AD in the US alone are $148 billion annually.  Despite this high cost and a steadily increasing mortality rate for AD, funding for AD research has significantly lagged that of other major diseases, such as HIV, cancer and heart disease.  The patients included in clinical trials of approved drugs for AD are thought to be too severely ill to gain much benefit from a disease modifying agent.  Unfortunately, trials done with approved AD medications in patients with Mild Cognitive Impairment (MCI), considered pre-AD, also failed to show an effect on disease progression, possibly because an intervention in these patients was already too late.  Going forward, strategies to identify at-risk patients even earlier in the pre-dementia phase are needed, as are novel therapeutic agents that will arrest or significantly delay the pathological process.  A delay of even one year in the onset of AD could result in substantial cost savings of approximately $1 trillion in the US alone.  Long-term therapeutic trials of 5-10 years’ duration, with treatment initiated very early in the disease process, are needed. 
Dr. Kraemer discussed the alignment of policy goals with research capabilities, and emphasized the need to do a better job of investigating the treatments that we have to ensure that effective treatments are not discarded and that a lot of money is not wasted on ineffective treatments.  Although long-term trials are needed, they are fraught with problems, including high cost, sporadic funding, high dropout rates, lack of compliance, and influence of other factors, such as concomitant treatments, on the outcome.  In addition, priorities change over time and new questions may develop that ongoing studies are not designed to address.  Also, in disease prevention studies, risk factors change as patients age.  Often, answers gained in research studies are not relevant to the questions that policy makers need addressed.  Going forward, the following changes in the way long-term studies are conducted should be considered: articulate specific research questions; focus on meaningful effect sizes, rather than p values; explore for moderators of treatment response, e.g. using sequential RCTs or looking at different outcomes in different stages of the trial; perform “accelerated lifetime studies” using overlapping cohorts of patients at different stages of the disease rather than following all patients for 5-10 years.  These strategies should mitigate the issues with long-term trials, such as high dropout rates, and should decrease sample sizes, increase effect sizes, and allow results to be available sooner. 
Part 2: Policy Shareholders Perspectives
Dr. Breier presented the industry perspective, and noted that CNS diseases are a very high risk area, with only 10% of promising drug candidates ever getting launched.  Disease modifying agents for schizophrenia or AD would certainly be of interest, but long development times (>8 years) are an impediment.  A clear regulatory path to registration is needed, including definition of the very mild/pre-symptomatic population to be included, and the requirements for a “disease prevention” or “delayed time to onset” claim.  An initiative similar to MATRICS for cognitive impairment associated with schizophrenia would be a good model for academic/NIH/FDA/industry collaboration on defining the design, primary outcome measure(s), selection criteria, etc. for these studies. Unique challenges related to these trials would be the need for biomarkers to identify subjects, determining sample size (i.e. estimating conversion rates), trial duration and retention of subjects.  Smaller proof-of-concept trials may not be feasible with disease prevention as the target, unless a surrogate marker is identified.  Although earlier treatment would be profitable, treating young patients in the schizophrenia prodrome raises ethical issues.  Development of new and novel treatments would be favored over expanded use of existing drugs.  A one-year, double-blind RCT comparing olanzapine and placebo in patients diagnosed with prodromal symptoms of psychosis was presented as an example of the challenges in doing these trials.  In this study, recruitment and retention of patients was very difficult, significant side effects were noted, and the trial was stopped early. 
Dr. Laughren gave the FDA perspective on long-term schizophrenia trials.  The challenges in performing trials in schizophrenia include the fact that it is a lifelong disease, with severe functional impairment, morbidity and mortality, and currently we have a very limited understanding of the disease.  There are many stakeholders involved, including patients and their families, advocacy groups, government and private healthcare providers, researchers, Pharma and FDA.  If treatments are to be evaluated in prodromal schizophrenia, it needs to be better defined.  Treatment goals could include delaying the onset of the first episode, rather than disease prevention.  Use of non-drug therapies, in addition to drugs, need to be considered.  As with all drugs, for a disease modifying agent to be approved, “substantial evidence of efficacy from well controlled trials”, must be provided.  Standard designs for acute schizophrenia trials would most likely not work for trials assessing delay in disease onset.  Regarding long-term treatment trials in schizophrenia, maintenance studies are currently not required for approval of an antipsychotic drug.  Long-term, randomized-withdrawal trials have been performed for most drugs, and generally come out positive, but don’t really assess effects of the drug in the long-term course of the disease.  Suggestions have been made for improving registration trials, including doing more efficient, sequential, short-term trials; however, this model would lead to longer development time, which does not fit the business model.  Regulatory acceptance of observational trials has also been suggested, but these would never be accepted to support efficacy.  FDA efforts to improve conduct of schizophrenia trials include exploratory analyses of summary data and patient level data, and establishment of data standards for trials.  The NEWMEDS consortium, which is an international collaboration of academia and industry, is also working on this.  
Dr. Katz provided the FDA opinion on long-term trials in AD, focusing on trials of disease modifying agents.  Trials of amyloid modifying agents are ongoing; however, there are a number of problems that still need to be addressed, including identifying appropriate patients for inclusion, use of appropriate measures to assess drug effects, and determining adequate duration for these studies.  Patients diagnosed with AD may already be too late in the disease process to benefit, but currently, for sporadic AD, there is no way to accurately identify with high precision which asymptomatic patients will progress to dementia.  MCI may also be too late, as these patients already have substantial amyloid burden.  From a regulatory perspective, a drug that could significantly delay or prevent dementia would certainly be approvable.  However, to provide evidence of efficacy, you would need to demonstrate an effect on a surrogate marker that predicts clinical benefit, as well as show an effect on at least one clinical outcome, e.g. time to diagnosis of AD, cognitive measure, or global outcome.  The duration of these trials should be as short as possible to demonstrate a clinically meaningful drug effect, but most likely would need to be several years in duration. 
In the ensuing discussion, Dan Begel (Psychiatrist) suggested that there is disconnect between what is evaluated by the clinician in the field vs. clinical trials, and that better measures need to be developed to truly assess what the patient is thinking and feeling.  Dr. Kraemer suggested using several measures and developing a composite outcome that would give a better overall picture.  In response to a question by Nina Schooler, an example of overlapping sequential cohorts was provided by Dr. Manderscheid, in which cohorts of patients with overlapping age ranges, e.g. 40-45, 45-50, 50-55, etc., would be run separately but simultaneously, and the data would be pieced together to create a long-term follow-up study.  Judy Kando noted that accelerated development using large, long-term trials has been done routinely in other areas, such as oncology and cardiology.  Dr. Kraemer commented that these trials have problems similar to those in CNS diseases.  Dr. Laughren acknowledged that cardiovascular diseases are better understood, making long-term prevention trials easier to conduct.  Dr. Anand added that attrition is also less of a problem in these areas than in CNS, where we often see a 30% dropout rate in a 6-week study; this may be related to the fact that patients with cardiovascular diseases may have no noticeable symptoms when treated with placebo.  
Dr. Ereshefsky suggested that a national screening effort, as was done for AIDS, should be considered for AD; however, since there are no available treatments to prevent/delay the disease onset, it was questioned whether this would be of value and whether or not patients would want to know.  Dr. Anand added that screening at this point would be impractical because of the lack of adequate biomarkers and the need to screen patients as early as 50 years of age, to capture the early stage of the disease.  Up to 100 subjects may need to be screened to find one confirmed pre-AD patient.  
Part 3: Research Stakeholder Perspectives
Dr. Breier described the current status of early intervention trials in schizophrenia.  These studies require an enriched sample, and there is a need for better diagnostic markers to identify patients in the prodromal phase.  An ADNI-like initiative was suggested to identify surrogate endpoints.  To aid in recruitment of subjects and standardization of procedures, a clinical trial network was proposed with each site using similar protocols for neuroimaging, e.g. measuring cortical gray matter changes.  Progressive brain changes have been noted in untreated first episode schizophrenic patients.  Other biological measures could include N-acetyl cysteine.  A 12-month study in which serial MRI/MRS is performed was proposed.  
Dr. Anand presented that status of long-term trials in AD. Ongoing treatment trials in both prodromal and mild/moderate AD are being conducted using amyloid-modifying agents, including secretase inhibitors and anti-Aβ antibodies, which it is hoped will arrest or slow the progression of the disease.  In addition, several large, multi-center, long-term, collaborative trials are ongoing or planned, including ADNI, AIBL, DIAN, A4 and ICTUS, which are designed to better understand the progression of AD, and include neuroimaging and assessment of biomarkers, in addition to cognitive testing, and in some studies, the evaluation of drug therapy (to be provided by Pharma).  Unfortunately, to date most amyloid-modifying therapies have failed to demonstrate a therapeutic effect or have been associated with significant toxicity.  One possible reason for the lack of effect may be that the patients included in these studies have been too far along in the disease process to benefit.  Also, amyloid-modifying agents impact only one facet of AD pathology, and do not affect other processes such as changes in tau.  A proposal was made that patients included in prodromal AD trials should have the following characteristics: memory complaints (subject or informant); presence of amyloid on PET; and lack of clinical symptoms, significant structural changes in the brain or high CSF tau.  These studies would require a large sample size (>300/group) and need to be conducted for up to 10 years.  To reduce attrition, no restrictions should be made on use of concomitant medications, and patients with concomitant illnesses should be allowed to continue.  The proposed primary efficacy endpoint would be a 30% decrease in the rate of cognitive decline on a composite measure at 3 years.  Other measures should include behavior, functioning, neuroimaging and genetic/bio-markers.  For regulatory purposes, the company would break the blind internally at the 3-year time point and submit these data for approval.  Patients would continue on blinded treatment during long-term follow-up, with an option to switch patients to the active treatment, if proven effective.  The long-term treatment phase of the study would require alternative funding, perhaps through a government/industry partnership.  
In the panel discussion, Dr. Kraemer noted that AD and schizophrenia are most likely not due to a defect in a single gene; therefore, a combination of genetic markers along with biomarkers will be needed to identify patients at risk.  Substantial data on patient characteristics could be obtained from large RCTs and risk studies that have already been conducted.  In addition, lessons learned from long-term risk studies in other areas, e.g. cardiology, may be helpful.  Dr. Marder noted that in schizophrenia we can’t predict the outcome after the initial episode; therefore, we can only treat the symptoms as they occur.  Currently, we do not have the ability to change the trajectory of the illness.  Dr. Ereshefsky raised the possibility that a surrogate illness approach might be useful, e.g. Down’s syndrome for AD, since similar amyloid pathology is observed.  This would be difficult in schizophrenia, although targeting gray matter changes (up to 4% decrease/year), rather than prodromal symptoms might be one approach.  
In the subsequent discussion, Janet Williams suggested that unblinded data from all clinical trials, including failed studies, should be made available for analysis.  It was noted that the Coalition Against Major Diseases, an academic/regulatory/industry collaboration, has more than 6000 AD patients in its database, including data from biomarkers and ADNI.  Adapting standards for data collection in clinical trials going forward would make it easier for pooling data in these types of large databases.  Dr. Kraemer cautioned that, when pooling data from large multi-center trials there is large variability among sites, and pooling of sites leads to regression to the mean; instead, each site should be looked at separately for effect size.