Statistics from Altmetric.com
Tensor based morphometry
The true and ultimate aim of research in Huntington’s disease (HD) is to arrest the progressive neurodegeneration and thus the debilitating clinical and functional deterioration—or so it seems to those involved in the clinical care of the unfortunate individuals affected by it. Two major clinical challenges confront us: to retard, or better, to abolish disease progression in symptomatic individuals; and to prevent phenoconversion in presymptomatic individuals.
Huntington’s disease is considered to be rare, but extrapolation of its estimated prevalence—in western countries 5 to 10 per 100 000—yields for the current 25 countries of the European Union (with 455 million people in 2004) a total number of between 22 500 and 45 000 clinically affected patients in various stages of the disease. The number of presymptomatic mutation carriers is much higher. Recent advances in our understanding of the molecular pathobiology have raised hopes that rational treatments may be near. Huntington’s disease is caused by an expanded CAG triplet repeat in exon 1 of the huntingtin gene, and this mutation is expressed and translated into an expanded polyglutamine sequence in about half of all huntingtin protein molecules that are produced in the body cells. The physiological functions of the protein are still unknown. But many details of huntingtin interactions, cleavage, conformational changes, aggregation, and proteasomal breakdown have been clarified in the past decade.1 Based upon these mechanistic models, various potential drugs have been proposed. In fact, efficacy of such proposals has been demonstrated in simple cellular or more complex animal models, such as transgenic flies or mice.2 Another approach has been the large scale screening of promising compounds in simple systems3; and even older hypotheses regarding the causes of neurodegeneration, such as excitotoxicity, have yielded proposals for neuroprotective treatments. In fact, most of the past and ongoing neuroprotection trials in HD have studied compounds derived from this hypothetic excitotoxicity.
But these trials have faced one major problem. All of them have used serial clinical or functional assessments of symptomatic patients as the primary outcome measure. Despite their demonstrated robustness, linearity, and relevance, the intrinsic variability of clinical outcome measures in treated cohorts has been large. As a result, phase II and III neuroprotection trials have typically included relatively large numbers of symptomatic patients who were followed for several years. Thus a North American trial which studied the effects of remacemide and coenzyme Q (CoQ) against placebo in a 2×2 factorial design enrolled 347 patients and followed them for 30 months. The study failed to show any benefits of the treatment, although a trend towards an effect of CoQ has been suggested.4 Based upon this study, calculations predicted that a sufficiently powered CoQ trial would require more than 1000 patients to be followed for at least two years. The European EHDI trial that studies the effects of riluzole on disease progression in a 1:2 randomisation design enrolled 560 patients who were followed for 37 months. Results of this trial are pending. Previous interventions which studied baclofen, idebenone, vitamin E, and lamotrigine included 100 patients or fewer, but all of them turned out to be clearly underpowered. The lesson: Huntington trials need large numbers of participating patients, to be followed for years.
If we were to conduct 10 such European protection trials simultaneously, about a quarter of all the patients in the 25 EU countries would have to be enrolled and followed for many years. If we were to conduct an intervention in presymptomatic patients, the difficulties would be compounded because of the very long follow up times required to demonstrate phenoconversion. The logistics issue then becomes: how can we persuade sufficient participants and generate resources for such trials? Shorter trials that require fewer patients would clearly be preferable. The statistical answer would be: let us reduce end point variability.
Enter the paper by Kipps et al5 (this issue, pp 650–5), about imaging the structural disease progression in preclinical disease. Using a novel approach to statistical imaging, called tensor based morphometry, they were able to demonstrate over a period of two years a progressive regional grey matter loss in 17 presymptomatic Huntington mutation carriers compared with 13 mutation negative controls. In contrast, clinical measurements failed to pick up any deterioration. If this result can be confirmed by others, imaging would become a powerful tool in neuroprotection trials. Imagine an intervention trial in which 40 patients are randomised and followed for two years, with, as the major measurements, an MRI scan at baseline and at the end of the trial. Although such measurements would have to be considered surrogate end points, they would nevertheless assume a crucial role in the selection of compounds to undergo the final test of clinical efficacy: a randomised controlled clinical trial with 1000 patients enrolled and followed for 30 months, with clinical outcome as the primary end point. We could certainly start planning a “primary prevention” trial aimed at postponing (or abolishing?) the onset of the disease.
Tensor based morphometry
Competing interests: none declared
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.