Biomarkers for Huntington's disease: an update

Abstract

Huntington's disease (HD) is a devastating autosomal-dominant neurodegenerative condition caused by a CAG repeat expansion in the gene encoding huntingtin which is characterised by progressive motor impairment, cognitive decline and neuropsychiatric disturbances. There are currently no disease-modifying treatments available to patients, but a number of therapeutic strategies are currently being investigated, chief among them are nucleotide-based ‘gene silencing' approaches, modulation of huntingtin post-translation modification and enhancing clearance of the mutant protein. In 2008, the authors' review highlighted the need to develop and validate biomarkers and provided a systematic head-to-head comparison of such measures. They searched the PubMed database for publications, which covered each of the subheadings mentioned below. They identified from these list studies which had relevance to biomarker development, as defined in their previous review. Building on a tradition of collaborative research in HD, great advances have been made in the field since that time and a range of outcome measures are now being recommended in order to assess efficacy in future therapeutic trials.

Keywords::

• Huntington's
• disease
• biomarkers
• imaging
• biofluids

1. Biomarkers for clinical trials

1.1 Neuroimaging

It has long been established that neuroimaging can demonstrate changes within the brain in both early Huntington's disease (HD) subjects and premanifest gene carriers, with both regional and global measures sensitive to change from the very early stages. There is now increasing recognition of the role of white matter in the degenerative process, with widespread change detectable more than 15 years before expected disease onset [1,2]. Thanks to large-scale studies such as TRACK-HD [1] and PREDICT-HD [2], there is now evidence that these imaging metrics are able to show reproducible group differences across multiple sites and that change over time can be demonstrated robustly in this slowly progressive disease over periods as short as 12 months, a timeframe which is realistic for clinical trials [3]. The greatest effect sizes have been reported using rates of caudate loss, with global measures also demonstrating great power to detect disease-related change, providing sample sizes which could feasibly be recruited for therapeutic trials [4]. TRACK-HD has also demonstrated the primacy of cross-site standardisation and robust quality control measures to ensure consistency of these data.

More recently, imaging modalities other than structural magnetic resonance imaging (MRI) have been proposed as suitable candidates for clinical trials. Diffusion weighted imaging (DWI), which specifically investigates the structural integrity of white matter, has demonstrated change over time in early HD subjects [5], although this finding was not reproduced by a larger study [6]. Functional MRI (fMRI) has the potential to detect disease-related variation in brain activity, showing cross-sectional differences prior to disease onset [7]. However, a recent study failed to show change over a 2-year period [8]. While fMRI and DWI remain highly promising candidates for probing dysfunctions in neural networks that are the hallmark of premanifest HD, further work is required to investigate their sensitivity to longitudinal change as well as validate measures across multiple sites before the clinical utility of these modalities are established.

Other recent work has shown cross-sectional disease-related differences in motor cortex excitability using transcranial magnetic stimulation [9]. Altered metabolite profiles, suggestive of neuronal dysfunction have also been reported in early HD and treatments aiming to improve neuronal health might effectively be assessed using such a metric [10]. Again, further cross-site validation and follow-up data are required to assess the utility of these modalities.

1.2 Quantitative clinical measures

Clinical ratings scales remain the most widely used measures of progression in HD. However, due to their large inter-rater variability and floor and ceiling effects, recent work has focused on more objective techniques which may better capture the range of deficits in both early and premanifest stages of disease.

Quantitative motor assessments such as finger tapping, grip force and tongue protrusion tasks have shown significant cross-sectional effects across the disease spectrum [1]. These measures have also demonstrated robust change over time in early HD subjects [3,11]; effect sizes suggest such metrics may prove useful as outcome measures in clinical trials [4]. They are relatively insensitive over short time intervals in the premanifest stage showing small effect sizes longitudinally [11], however, over a 36-month interval these measures look more promising as outcome measures for trials in premanifest subjects (TRACK-HD, unpublished data). Similarly, oculomotor and neuropsychiatric assessments which show cross-sectional disease-related deficits [1], appear to be relatively insensitive to longitudinal change over periods of up to 24 months [4]. However, some neuropsychiatric measures show progression over 36 months (TRACK-HD, unpublished data).

Numerous studies have shown cross-sectional cognitive deficits, even many years before disease onset [1,2]. Longitudinal studies have shown change over time in early HD, although effect sizes vary considerably depending on the specific cognitive test employed [4,12]. Sensitivity to change in the premanifest stages is again very limited over 24 months [4], but after 36 months changes in cognition become apparent in this cohort (TRACK-HD, unpublished data).

1.3 Biofluids

Robust, pathologically relevant changes in accessible biofluids such as blood remain highly desirable. The pathways previously highlighted as suggesting potential biomarkers remain of interest, particularly oxidative, metabolic and inflammatory markers. These require testing in samples derived from multiple populations and over time, to establish their potential utility for future clinical trials. Newly identified biomarker candidates that require further investigation include cholesterol metabolites [13] and indirect markers of transcriptional dysregulation such as the histone component H2AFY [14]. Meanwhile clusterin, which the authors' work revealed as a possible HD plasma biomarker [15], has since been highlighted as a major genetic modifier and possible blood biomarker of Alzheimer's disease [16].

Accurately measuring levels of mutant (mHtt) and wild-type huntingtin in biosamples will be essential for testing therapeutics aimed at reducing mHtt levels, including promising nucleotide-based ‘gene silencing' treatments expected to enter clinical trials in the near future, as well as drugs aimed at altering the handling and removal of mHtt through post-translational modification and proteolysis. Existing Htt assays have suffered from low sensitivity and have been restricted to aggregated mHtt. A recently described agarose gel electrophoresis assay (AGERA) has lowered the threshold for aggregate detection and quantification [17], but the role of aggregates is debated and it is likely that soluble oligomeric mHtt is most toxic. Until recently, the detection of soluble Htt was limited by the intrinsic insensitivity of assays, and a lack of suitable antibodies. Novel antibodies and a fluorescence-enhanced antibody assay using time-resolved Förster resonance energy transfer (TR-FRET) has now yielded a reliable, highly sensitive method of quantifying both mutant and total huntingtin in blood samples [18]. This assay is likely to empower future clinical trials. Meanwhile, attempts are underway to develop assays that are even more sensitive and scalable.

Given its close biochemical and physical relationship with the brain, and utility in other neurodegenerative diseases as a source of biomarkers, cerebrospinal fluid (CSF) has natural appeal, despite being time-consuming and more invasive than phlebotomy. Efforts to identify CSF biomarkers for HD have been hampered by a shortage of samples and lack of consistency between existing small collections. In one major study, proteomic data generated from one sample set by five laboratories were integrated, revealing changes (mainly reductions) in many brain-specific proteins, but raising concerns about consistency between laboratories [19]. Pursuing a hypothesis-driven approach, another group has identified neurofilament light chain and tau as potential neuron-specific CSF markers [20]; but neither provides complete separation from controls or has been studied longitudinally. Multisite longitudinal CSF collection studies are in the planning stage.

2. Future directions

2.1 Neuroimaging

It is clear from studies such as TRACK-HD and PREDICT-HD that there are striking neuropathological changes up to 15 years prior to the onset of manifest disease [1,2] (Figure 1). Nevertheless, these subjects maintain their performance on functional measures. It would appear that there is a compensatory mechanism within the brain which finally breaks down at disease onset [7]; imaging modalities other than structural MRI will be required to elucidate this process. Track-On HD seeks to apply novel imaging techniques across multiple sites, following the premanifest cohort from the TRACK-HD study. Whether these imaging modalities are able to produce metrics which are robust and sensitive enough to be used in clinical trials is yet to be determined.

Figure 1. Parametric maps showing regions with statistically significant atrophy over 24 months in grey (top) and white matter (bottom) in far-from-onset premanifest subjects compared with controls. Results were adjusted for age, sex, study site and scan interval and are corrected for multiple comparisons with familywise error at the p < 0.05 level. Figure shown courtesy of the TRACK-HD study.

A means of quantifying wild-type and mutant huntingtin in the brain would be a valuable tool for demonstrating central nervous system (CNS) target engagement of huntingtin-lowering therapies in the main organ of interest, especially since early gene silencing therapeutics are likely to be delivered directly into the CNS rather than systemically. Work is underway to develop and test PET (positron emission tomography) ligands that will enable visualisation of huntingtin load, as a predictive tool and a means of assessing response to therapy.

2.2 Quantitative clinical measures

Further refinement and validation is required if objective quantitative measures of motor and cognitive function are to demonstrate sensitivity to premanifest disease, or be accepted as outcome measures in clinical trials. It is likely that clinical trials in premanifest subjects will need to be at least 36 months in duration if current functional and clinical outcome measures were to be used. A number of cognitive tasks show promise as biomarkers, particularly measures of visuospatial integration [3]. A cognitive ‘toolkit' for premanifest and manifest HD, which will build on the existing evidence, is nearing completion [21]. Neuropsychiatric measures are likely to be included in future clinical trials since assessments of patient well-being are of paramount importance, but these too require the development of more reproducible and sensitive methodology. In addition to these initiatives, ongoing longitudinal studies such as PREDICT-HD and Track-On HD seek to optimize measures in these different functional modalities.

2.3 Biofluids

Novel assays to measure soluble mutant huntingtin, and their refinement for high-throughput use, are likely to emerge as one of the most important developments in the field and will become a standard measure for clinical trials [18]. The use of CSF as the biofluid of choice for interrogating the CNS milieu is likely to occur in HD as it has in other neurodegenerative diseases, but standardised, multisite, longitudinal CSF natural history studies are needed [22]. Finally, the most promising biomarker candidates need to be studied ‘head-to-head' using the large longitudinal blood sample collections that studies such as TRACK-HD have generated [4].

Though a decade of research has not yielded robust biochemical biomarkers of HD, progress has been made. A consensus is emerging that the detection of neuron-specific proteins in blood may be an unrealistic aim, but since the mutant huntingtin protein is expressed ubiquitously, that does not mean that tracking change in peripheral tissues lacks merit. What is needed is scrupulous study of how any measure is involved in HD pathogenesis, and judicious interpretation of changes seen in the context of therapeutic trials.

3. Expert opinion

Considerable progress has been made towards identifying biomarkers which show promise as outcome measures in clinical trials. Multinational studies such as PREDICT-HD and TRACK-HD have proved invaluable in validating measures which are robust and reproducible across different sites and cultures. The most recent publication from the TRACK-HD study provides a direct comparison of the effect sizes for a range of modalities, and outlines a battery suitable for use in clinical trials in early HD [4]; this includes structural neuroimaging, quantitative motor, cognitive and neuropsychiatric assessments. However, as yet such a battery does not exist for a premanifest cohort as current functional measures appear to be relatively insensitive at this disease stage over a 24-month time period. The 36-month dataset from TRACK-HD is currently being analysed and it is anticipated that this may improve the battery for clinical trials in the premanifest cohort. Neuroimaging demonstrates the largest effect sizes, but further investigation of the relationship between such metrics and functional performance is required. There is clearly a need for robust functional assessments to be included in any trial and there are now a number of candidate biomarkers from different modalities which can feasibly be implemented. The individual elements of this battery selected for a particular trial will of course depend on the proposed mode of action of therapy, timeframe and cohort. It was previously predicted that ultimate evaluation of biomarker candidates would need to take place within the framework of clinical trials of disease-modifying treatments. This model for biomarker development is now deployed in HD clinical trials (e.g., mHtt and ILs are being measured as part of the PADDINGTON study of sirtuin-1 inhibition) and, though challenging, this approach is essential, since it is only when disease, biomarker and drug are studied together that the true interplay between them can emerge.

Declaration of interest

The research in SJ Tabrizi's laboratory is funded by MRC, BBSRC, High Q/CHDI Foundation, EU FP7 Framework, UCL/UCLH NIHR Biomedical Research Centre, UK HD Association, European HD Network and the UK Dementia and Neurodegenerative Diseases Network. EJ Wild is supported by the National Institutes for Health Research and the Academy of Medical Sciences, UK. UCLH/UCL receives a proportion of its funding from the Department of Health's NIHR Biomedical Research Centres funding scheme.