Can Brain Metabolism Predict Huntington’s Symptoms?

With several potential disease-modifying treatments for Huntington’s poised to enter clinical trials, scientists need ways to track the progression of the disease, particularly at the preclinical stage. In the September Journal of Clinical Investigation, researchers led by David Eidelberg at the Feinstein Institute for Medical Research, Manhasset, New York, describe a promising new marker. Using positron emission tomography with fluorodeoxyglucose (FDG PET, which measures glucose consumption), the authors were able to identify a pattern of altered brain metabolism in people carrying the mutant huntingtin gene. This brain abnormality develops as much as two decades before clinical symptoms, and steadily worsens throughout this long preclinical and early clinical period.

FDG PET appears to be a more sensitive indicator of disease progression than better-studied biomarkers such as shrinking brain volume and dopamine receptor binding, the authors found. The results suggest that glucose metabolism could track clinical trial outcomes, although that remains to be proven.

“I think this is an exciting and important new addition to the biomarker armamentarium in Huntington’s disease, and could strongly facilitate a future clinical trial,” said Raj Ratan at the Burke Medical Research Institute, White Plains, New York. Ratan was not involved in the work. Other researchers called the findings new and intriguing, but noted they need to be replicated by other groups. Ratan said that metabolic network activity represents a physiologic measure that could help researchers formulate hypotheses about how Huntington’s disrupts different parts of the brain, and how those disruptions relate to clinical symptoms. Metabolic networks are subsets of brain regions where changes in metabolic activity are synchronized.

Beginning in middle age, people who have inherited mutant huntingtin develop problems with movement, cognition, and emotion, and those symptoms progressively worsen over time. No treatments exist to slow the deterioration, although clinical trials are getting underway (see ARF related news story). Meanwhile, two large international collaborations, TRACK-HD and PREDICT-HD, are hunting for biomarkers that predict the onset of the disease and track its progression (see ARF related news story). Where many of these efforts examine cerebrospinal fluid biomarkers or measure brain volume changes by magnetic resonance imaging (MRI), Eidelberg chose to focus instead on brain metabolism. He first studied it in Parkinson’s disease, where he identified metabolic networks that distinguish different forms of the disorder (see Mure et al., 2011). Turning to Huntington’s, he found regional differences in metabolism between mutation carriers and healthy controls, including weakened activity in the thalamus and striatum, but at first saw no pattern of consistent changes over time (see Feigin et al., 2007).

In the new work, Eidelberg and colleagues looked for metabolic patterns that would predict progression. First author Chris Tang and colleagues analyzed longitudinal FDG PET data from 12 presymptomatic huntingtin (htt) mutation carriers using a sophisticated computational algorithm called Ordinal Trends Canonical Variance Analysis. The algorithm looks for networks whose activity varies together (see Habeck et al., 2005; Moeller and Habeck, 2006). Tang found that over seven years, metabolic activity progressively waned in several brain areas, such as the caudate, putamen, thalamus, insula, posterior cingulate gyrus, and prefrontal and medial occipital cortices, while perking up in a smaller set of regions, including the cerebellum, pons, hippocampus, and orbitofrontal cortex. To use these changes as a biomarker, the authors derived a single score from the data. They related metabolism in carriers to the mean metabolic activity in 12 age-matched, healthy controls. This score diverged each year by an additional 0.21 standard deviations from the control population. Notably, this rate remained constant over time, even in participants who became symptomatic.

Based on this constant, the authors projected back in time that HD causes brain metabolism to begin diverging from normal about 20 years before the expected age of symptom onset, reaching two standard deviations from normal about 10 years before disease diagnosis. By the time symptoms appear, metabolic activity in this network is about four standard deviations away from normal, the authors found. This jibes with recent findings for other neurodegenerative diseases, which are now known to start much earlier than previously believed. For example, amyloid accumulation begins up to 30 years before diagnosis of Alzheimer’s disease (see ARF related news story; ARF related news story).

To validate the biomarker findings, co-authors at the University Medical Center in Groningen, Netherlands, performed an independent two-year longitudinal study on 21 htt mutation carriers. They found a nearly identical metabolic decline of 0.19 SDs per year. The authors also reported that repeat scanning of individual study participants showed good reproducibility. Under a grant from PREDICT-HD, co-authors at four sites scanned a total of 14 HD carriers twice, with three weeks between the first and second scan, and obtained a 0.95 correlation.

How does the metabolic score stack up against other markers of HD? The authors compared the FDG PET pattern to commonly used HD preclinical biomarkers. Compared with MRI measures of brain volume and PET measures of dopamine binding, the metabolic network biomarker worsened 30 percent faster, and diverged further from population norms by the time symptoms began. Because metabolism appears more sensitive, it could facilitate smaller clinical trials, Eidelberg said. He estimated that a cohort of 120 mutation carriers would provide 90 percent power to detect a 30 percent slowing of disease progression over a two-year trial. This contrasts with perhaps 500 participants needed with current markers, he told Alzforum. Eidelberg noted that several pharmaceutical companies have expressed interest in testing this biomarker, although he declined to name them.

In the Alzheimer’s field, thus far no disease progression biomarker has proven itself a good outcome measure for clinical trials (see ARF related news story). However, Eidelberg believes that metabolic networks can better reflect cognitive change than other markers of pathology do. “Networks represent an amalgamation of synaptic activity in the brain. They better reflect aspects of pathology that mediate behaviors, and for that reason they may correlate better with clinical symptoms,” he said.

In ongoing work, Eidelberg conducts a validation study of the metabolic network biomarker at multiple sites as part of PREDICT-HD. In addition, the nonprofit foundation Cure Huntington’s Disease Initiative (CHDI), which has funded some of Eidelberg’s studies in the past, works with researchers led by Anissa Abi-Dargham at Columbia University in New York City to independently replicate the findings, said Cristina Sampaio, CHDI’s chief clinical officer. Co-author Jane Paulsen at the University of Iowa, who performed some of the test-retest studies, said that the methods used for measuring this metabolic network are improving rapidly to a point where more sophisticated techniques are now ready for widespread use. “The progress is exciting,” she wrote to Alzforum.

The current study did not examine whether the metabolic pattern the authors found is specific for HD or if it occurs in other neurodegenerative disorders. Specificity of the FDG-PET finding to a given disease is what has limited its use in AD research. However, Eidelberg told Alzforum that Ordinal Trends Canonical Variance Analysis turns up unique networks when performed on FDG PET data from other conditions, as well. For example, using data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), he has found a distinct metabolic pattern that changes with AD progression and correlates with cognitive scores. He speculated that if the finding holds up, this biomarker may show promise as an outcome measure in AD trials.—Madolyn Bowman Rogers.

Comments

No Available Comments

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. Study Finds Ways to Predict, Track Huntington’s Disease
  2. In Big Picture, Familial AD’s Biomarker Data Resemble LOAD
  3. From Natural History, A "Renaissance" for Amyloid Hypothesis
  4. CTAD: New Data on Sola, Bapi, Spark Theragnostics Debate

Paper Citations

  1. Mure H, Hirano S, Tang CC, Isaias IU, Antonini A, Ma Y, Dhawan V, Eidelberg D. Parkinson's disease tremor-related metabolic network: characterization, progression, and treatment effects. Neuroimage. 2011 Jan 15;54(2):1244-53. PubMed.
  2. Feigin A, Tang C, Ma Y, Mattis P, Zgaljardic D, Guttman M, Paulsen JS, Dhawan V, Eidelberg D. Thalamic metabolism and symptom onset in preclinical Huntington's disease. Brain. 2007 Nov;130(Pt 11):2858-67. PubMed.
  3. Moeller JR, Habeck CG. Reciprocal Benefits of Mass-Univariate and Multivariate Modeling in Brain Mapping: Applications to Event-Related Functional MRI, H(2) (15)O-, and FDG-PET. Int J Biomed Imaging. 2006;2006:79862. PubMed.

Other Citations

  1. ARF related news story

External Citations

  1. TRACK-HD
  2. PREDICT-HD
  3. Ordinal Trends Canonical Variance Analysis
  4. Cure Huntington’s Disease Initiative

Further Reading

Annotate

To make an annotation you must Login or Register.