12 January 2011. Alzheimer’s disease has seen a disheartening string of failed clinical trials in recent years. While some experts see these failures as evidence of too little, too late, others argue that they have more to do with targets than with timing. Either way, with new fluid marker, imaging, and cognitive tools coming online, researchers are pushing diagnosis back earlier in the progression of the disease. So is it time to push clinical trials back, as well? The strategy has its challenges, not least being how to select patients and monitor their disease progression when they have no dementia. Last October, a task force comprising drug development specialists from academia, the pharmaceutical industry, regulatory agencies, and a not-for-profit foundation met to search for consensus on some of the key technical challenges posed by early AD clinical trials. Their report, published in the December 23, 2010 Neurology, outlines broad principles that could be useful in early trial design.
“The goal was to move the whole field, including regulators, away from traditional MCI [mild cognitive impairment] trial design toward something we think is more powerful and appropriate,” said lead author Paul Aisen, University of San Diego, California, in an interview with ARF.
Aisen immediately addressed one of the major challenges in early trial design, that is, what the outcome measure should be. The task force concluded that “analysis of continuous clinical and cognitive measures is likely to be more efficient than survival analysis.” The task force concluded that “it may be reasonable and acceptable to use a single primary outcome measure to establish both benefit on primary disease symptoms and clinical relevance.” Conventional AD trials are done using a cognitive and a global measure, which adds complexity to trials but establishes clinical relevance. One single measure that could be used is the Clinical Dementia Rating Sum of Boxes, suggested Aisen. The CDR-SB can be used to measure cognition and clinical progression, as determined by analysis of data from the Alzheimer's Disease Neuroimaging Initiative (see Petersen et al., 2010).
In the past, every MCI-based trial used a survival type of analysis, meaning how long until a patient met the diagnosis of AD, as a primary outcome measure. “The reason was that regulatory agencies viewed MCI as heterogeneous, and were not supportive of AD trials aimed at treatment of MCI,” Aisen said. The argument was that you had to follow subjects until they got AD to be sure you were looking at the right population. “But now that we believe we can quite accurately diagnose AD prior to onset of dementia, we think that concern is mitigated,” said Aisen.
On that note, another consensus point was that “the selection of AD cases can be extended to patients pre-dementia using amnestic MCI criteria plus one or more biomarkers.” The field is gradually moving in that direction, anyway, with the well-received Dubois criteria that were published in 2007 (see ARF related news story) and revised last year (see Dubois et al., 2010), and new diagnostic guidelines for researchers proffered at last year’s ICAD meeting in Hawai’i (see ARF related news story). Both suggest that fluid and imaging biomarker analysis be combined with cognitive measures to diagnose Alzheimer’s prior to the dementia stage.
The panel reports consensus on three other points:
- The optimal stage for efficacy trials of disease-modifying intervention may be prior to dementia onset.
- CSF Aβ42 or amyloid PET imaging may be optimal biomarkers for selecting subjects for anti-amyloid interventions.
- Validation of surrogate endpoints in symptomatic patients may eventually provide a mechanism for developing treatments at the symptomatic stage of AD.
Whether regulatory agencies will adjust guidance based on these consensus issues is unclear (the FDA did not reply to a request for comment), but Aisen noted that a regulatory guidance usually follows consensus reached by researchers. “This document can support efforts of sponsors to gain acceptance of trial designs that incorporate these principles,” said Aisen.
It remains to be seen if these criteria will gain wide acceptance. Lon Schneider, University of Southern California, who attended the meeting and is a coauthor on the paper, told ARF that most of the consensus points are reasonable, though he is concerned that researchers may end up putting the cart before the horse. Rather than come up with a standard trial design for early AD, he’d approach it case by case. “We should first consider the drug candidate, then derive the clinical study and the regulatory plan from the expectation of how the drug will work,” he said.
Schneider also questions some of the specifics, including the emphasis on biomarker data for diagnosis. “Biomarkers may simply be identifying people with more advanced MCI,” he suggested. In a recent publication, he concluded that incorporating CSF biomarker analysis does not make for a more efficient trial (Schneider et al., 2010). Using data from the Alzheimer's Disease Neuroimaging Initiative, he found that a low CSF Aβ42, or high CSF tau/Aβ42, did not improve the power of the trial because over one or two years, the outcome in the biomarker group varied much more than in the non-biomarker group. Schneider noted that those who are biomarker positive are likely to have more severe pathology and more rapid but more variable deterioration.
The crux of the matter is that still not enough is known about the disease, he said. “We should be careful about making strong recommendations when we only have partial knowledge.”—Tom Fagan.
Aisen PS, Andrieu S, Sampaio C, Carillo M, Khachaturian ZS, Dubois B, Feldman HH, Petersen RC, Siemers E, Doody RS, Hendrix SB, Grundman M, Schneider LS, Schindler RJ, Salmon E, Potter WZ, Thomas RG, Salmon D, Donohue M, Bednar MM, Touchon J, Vellas B. Report of the task force on designing clinical trials in early (predementia) AD. Neurology 2010 December 22. Abstract