Session Summary: 2010 6th Annual Scientific Meeting – Clinical Trials in the Prodromal Phase of Alzheimer’s Disease – Methodological Challenges and Clinical Outcomes

This session was chaired by George Garibaldi from F. Hoffmann LaRoche and Michael Davidson from the Department of Psychiatry, Sheba Medical Center, Israel.

The focus of this session was to characterize both the validity and utility of clinical trials designed to assess treatments for prodromal Alzheimer’s disease (AD).  It is widely held that early intervention in neurodegenerative diseases prior to the emergence of overt clinical symptoms is desirable and should result in a greater impact on disease progression.  However, assessing the therapeutic index in a prodromal phase of these diseases is often beset by a number of methodological challenges.  One set of difficulties rests in sensitively and specifically identifying subjects at risk for these conditions in order for clinical study evaluation.  Also, because these neurodegenerative diseases are often slow to progress to their ultimate outcome, biomarkers must be identified that predict long term results.  Using AD as a model, this session focused on providing recent approaches to identifying prodromal AD patients and the use of biomarkers to track those changes in a clinical trial setting. Clinical, industry and regulatory input was provided regarding tactical and procedural issues regarding the implementation of these approaches in future clinical trials.

Michael Davidson, MD adeptly set the stage for the discussion by describing the challenges faced when identifying and providing prodromal disease treatments.  Dr. Davidson began by gave several examples of both short and long interval prodromes in CNS that preceded the onset of syndromic events by days, weeks, months or even years as a means to defining the salient features of a prodrome.  Whether the most important feature is the time lag between the prodrome and event it heralds; the inevitably of the event; the sensitivity/specificity of the prodrome; the symptomatic resemblance to the full blown illness; or the pathophsyiological relationship (causal path between the prodrome and the disease) appears to be vary widely according to the definition being applied.  Dr. Davison went on to define the utility of non-surrogate and surrogate biomarkers in prevention citing well established examples from cardiovascular medicine as well as their use and misuse.  Distinguishing AD prodrome from biomarkers is equally subjective.  Interestingly, inferences made about prodromes can be made from interpreting the failure of observation trial in prospective settings.   Noting that almost every disease is gradual what is really desired is to find a particular population that can benefit from intervention and any arguments on semantic may be irrelevant.

The keynote address was delivered by John C. Morris, MD who provided an overview of the AD prodrome and implications for patient selection and study design. Dr, Morris posited that prodromal AD typically is distinguished from DAT by a “failure to meet the threshold for dementia”, implying a dichotomous condition of “no dementia” vs. “dementia”.  However, AD is marked by continuous neuronal deterioration and applying a threshold to determine clinical status is therefore arbitrary, ambiguous, and variably applied.  As dementia requires functional impairment caused by cognitive loss (assessed by intra-individual decline, using one’s previously attained level of function as the control) this impairment” depends upon adequacy of information and the clinician’s “threshold”.  However, impairment in MCI/CIND is typically determined by inter-individual comparison of cognitive test performance (often for episodic memory) rather than a decline from previously attained levels leading to both false positive and negative inclusion of patients.  The focus on memory performance ignores early manifestations of AD as it has been shown that initial change is in visuospatial abilities (speeded performance) may predate memory difficulties in some persons.  Further, etiology is not fully considered as MCI/CIND encompasses non-AD dementias, reversible causes of cognitive impairment, and even some low-performing normal individuals.  Normative control groups may contain some patients with preclinical AD and relative performance of prodromal samples is misleading stemming from cutoff levels that are too lenient. Thus, the inclusion of people in the preclinical stage of AD in supposedly “normal” samples overestimates the decline in cognitive ability that is attributed solely to age.  Dr. Morris also noted that preclinical AD is characterized by cognitive decline marked by inflection points two to three years prior to diagnosis and this sharp decline is often in included in normative data.

Dr. Morris suggested that in an effort to reduce this heterogeneity informant-based clinical assessments are needed to capture intraindividual decline resulting in functional impairment. Additionally, using a cognitive cut off to define “impairment” and also incorporate biomarkers to establish likelihood that AD is causative (such as that by Dubois) can also be helpful.  Numerous examples of biomarkers in cognitively normal people were provided and in general conclusions from this data suggested that: cerebral Aβ42 deposition is the pathobiological phenotype of APOE4 and increases as a function of age in preclinical AD; Aβ42 changes characterize preclinical AD; while tau abnormalities occur later to mark symptomatic stages; A reduction of CSF Aβ42, and increased CSF tau identify very early symptomatic AD, in some instances prior to sufficient fibrillar Aβ for PIB detection; and nonfibrillar cerebral Aβ deposits (diffuse SPs) and preclinical AD are not benign and both CSF and amyloid imaging Aβ markers in normal elders predict symptomatic AD.

Dr. Morris supported the inclusion of such biomarkers in trials to both support accuracy of AD diagnosis, to enrich the sample, and to confirm the presence of the therapeutic target.  Obviously, using biomarkers as surrogates world require validation.  However, trials using agents that alter CSF production /clearance may benefit from their use nonetheless.  Finally, Dr. Morris proposed a paradigm shift in which AD researchers would move from notions of cure to prevention.  In this manner the detection of cognitively normal individuals with preclinical AD would be paramount and intervention to prevent neurodegeneration and symptomatic AD would be the goal.  In this shift, biomarkers such as CSF levels of Aβ42 and tau as well as amyloid imaging and are essential to detect preclinical AD (as once dementia occurs Aβ may no longer be the primary driver) .

Following the keynote address Harry Haroutunian, MD offered clues to identifying the prodrome for AD by presenting clinical correlations with various postmortem findings with a strong focus on the effects of age.  Dr.  Haroutunian emphasized the notion that although researchers have long recognized that age is the most important risk factor for developing AD, we have typically thought about age uniformly, viewed it as a confounding factor in analysis and have only recently begun to conduct systematic evaluations using specific older age cohorts.  This is very important as data suggests that the 85+ population is the fastest growing segment of our society and that dementia in the oldest old (>85 or >95) is a stronger predictor of morality than cardiovascular disease or cancer.  Further data suggesting the association between age and neuritic plaques and neurofibrillary tangles and their role in cognition was presented by Dr. Haroutunian.  In an attempt to determine transcriptional vulnerability Dr. Haroutunian conducted an assessment of RNA expression using Gene-Chips in patients stratified by dementia severity groups and concluded that amount of abnormal RNA expression varied in different brain regions. When evaluating the gene expression profile across fifteen different brain ranges in the young old versus the old-old Dr. Haroutunian reported that only 30% of the abnormally expressed genes are common to both groups suggesting a unique biological basis for dementia for each cohort.  This effect was not accounted for by cognitive reserve as defined by educational level. Thus, phenotype appears to matter significantly to the neurobiology of the disease and therefore to therapeutic response.   Medical illness, such as obesity, diabetes and hypertension and their treatment are just a few of the phenotypic variables that appear to be affected by age.  For example these factors appear to shift from a traditional risk to more protective factors from mid life to late life and those older patients with these risk factors actually have less cerebral pathology (plaques and tangles) than their younger cohorts.  In summary, Dr. Haroutunian noted that both age and phenotype are crucial to dementia neurobiology and potentially in therapeutic outcome.

Rachel Schindler, MD presented a summary of the recent developments in biomarkers and how these have changed clinical trials design. Dr. Schindler reviewed the well established challenges to typical to AD clinical trials such as including relatively large numbers of requisite patients, duration, progressive samples, and subjective assessments. The utility of various biomarkers was discussed in a comprehensive review including cutoffs for Aβ42 (at 190 pg/ml) and p-Tau in order to maximize clinical trial screening and enrichment of potential subjects.  The utility of combing biomarkers to differentiate AD form FTD, CJD and to predict MCI conversion to AD was also reviewed. Volumetric MRI as a biomarker was also examined in terms of disease state with regional changes appearing to have the best utility in early progression while whole brain changes are somewhat better for progression after the onset to AD, suggesting that volumetric imaging is one of the best biomarker for disease progression.  Similarly, hypometabolism in the posterior cingulate cortex was identified as the earliest and most sensitive functional marker for predicting conversion form MCI to AD. Dr. Schindler posited that there is a correlation between cerebral structural/functional imaging and biochemical makers with the degree of cognitive impairment and that many of these changes occur well before the diagnosis of dementia.  Therefore, the use of biomarkers aids the conduct of trials in early AD when many therapeutic agents may have their greatest effect.  The use of biomarkers in such trials could potentially reduce the sample size needed through identification of subjects, reduce trial duration, demonstrate a mechanism of action, monitor the effect of the intervention and possibly serve as the primary outcome measure in addition to or in place of cognitive and functional measures.

In a continuation of this theme Mark Cantillon, MD presented data specifically on neuroimaging as a biomarker for prodromal treatment providing evidence for both its utility as well as limitations. Dr. Cantillon posited that the five best biomarkers in AD research included both markers of neurodegeneration (CSF tau, FDG-PET and structural MRI), as well as markers of brain abeta deposition (CSF Aβ42 , and PET Aβ imaging). Dr. Cantillon presented evidence on various biomarker performance in the context of clinical trials assessing precision for detecting change, the clinical relevance of findings, and the possible impact on trial design concluding that structural MRI was the best potential candidate for trials of prodromal AD.  In the assessment of clinical relevance Dr. Cantillon proposed examining baseline makers’ ability to predict change in cognition and examining how changes in biomarker relate to changes in cognition.  Baseline biomarkers may also help to predict conversion form MCI to AD.  However, Dr. Cantillon cautioned that multivariate models of time to conversion generally do not provide evidence for biomarker significance when clinical variables are included in the model.  When clinical variables are omitted from such models the p values for biomarkers are significant.  For example, using FDG PET average metabolic levels on specific regions of interest can be helpful but FDG-PET has shown the greatest utility when examining average metabolic level across all regions of interest.  Dr. Cantillon presented evidence for surrogate markers that can potentially replace or be used as secondary endpoints in trials with the most promising markers having relatively better precision than standard performance measures.  Although the correlation with cognitive change is moderate at best and there is not enough data to capture treatment effects or to convince health care authorities at this time real value remains when using these biomarkers even when not as surrogate markers.  For example, these biomarkers have shown the greatest utility in clinical trials when used for entry criteria to confirm diagnosis, as stratification tools and as covariates with brain atrophy being the most promising.

The final session was conducted by Ravi Anand, MD who provided an industry perspective on clinical trials in prodromal AD reviewing several past trials.  Dr. Anand began by presenting various definitions of MCI and prodromal AD emphasizing the pivotal role of a deficit in recall that does not correct itself with cueing.  This general definition was used across multiple past industry trials of MCI and may be a key to future definitions of prodromal AD.  Dr. Anand acknowledged that the clinical construct of MCI in past trial was not very useful and suggested that the failure of past industry trials was not a function of the rate of conversion (estimated to be anywhere between 5 and 15% depending upon the trial) but more a matter of inappropriate treatment for the symptomatology, an inappropriate set of outcome measures and the enrollment of inappropriate patients who had disease severity that was not virtually unchangeable at the point of intervention. The fact that approximately half of the sample of one such trial meet autopsy criteria and 70% had hippocampal atrophy suggests that the patients had much greater disease severity than originally thought.  Other evidence from the PAQUID trial suggested that deficits in the Isaacs Set test, a measure of semantic memory and fluency, were evident more than a decade before the earliest signs of AD while deficits in visual memory and MMSE were not seen until alter. Thus, it may be more useful to think of prodromal AD as the transition between healthy aging and MCI with performance deficits being in a single cognitive domain. It is possible that the only way to reliable elicit such cognitive deficits may be under challenge condition such as high stress. Supporting this diagnosis would be a lack of, rather than a demand for, functional and behavioral deficits as well as biomarkers that would be used mainly to exclude patients (only minimal MRI atrophy would be permitted). Many of these notions were generally supported by panel members who stressed the need for repeated assessments beginning relatively early in a model that relies on within subject ratings to increase sensitivity.  It was recommended that slope analyses be used to detect the earliest point of deflection in prodromal AD.  Panel and audience members also suggested several technologies to aid serial cognitive assessments and suggested that brain health could be viewed in the same way the blood pressure with every potential patients knowing and tracking their biomarker numbers.  Given the changing population trends and accompanying financial burdens, the panel supported the notion of focusing on treatments that would delay the onset of AD even for 1, 3 or 5 years.  Similarities between statins and AD biomarkers were noted and there was much discussion concerning the heterogeneity of populations as well as the generalizability of biomarkers from those patients who have familial AD to those who have sporadic AD.  Finally, much like cardiovascular health the panel opined that it is likely that some type of treatment “cocktail” will most likely provide the greatest benefit to patients. The panel also agreed that more research would be needed in the identification of research methodologies appropriate for treatment trials at this earlier point of intervention.