Metabolic network as a progression biomarker of premanifest Huntington's disease

Abstract

Background: The evaluation of effective disease-modifying therapies for neurodegenerative disorders relies on objective and accurate measures of progression in at-risk individuals. Here we used a computational approach to identify a functional brain network associated with the progression of preclinical Huntington's disease (HD).

Methods: Twelve premanifest HD mutation carriers were scanned with [18F]-fluorodeoxyglucose PET to measure cerebral metabolic activity at baseline and again at 1.5, 4, and 7 years. At each time point, the subjects were also scanned with [11C]-raclopride PET and structural MRI to measure concurrent declines in caudate/putamen D2 neuroreceptor binding and tissue volume. The rate of metabolic network progression in this cohort was compared with the corresponding estimate obtained in a separate group of 21 premanifest HD carriers who were scanned twice over a 2-year period.

Results: In the original premanifest cohort, network analysis disclosed a significant spatial covariance pattern characterized by progressive changes in striato-thalamic and cortical metabolic activity. In these subjects, network activity increased linearly over 7 years and was not influenced by intercurrent phenoconversion. The rate of network progression was nearly identical when measured in the validation sample. Network activity progressed at approximately twice the rate of single region measurements from the same subjects.

Conclusion: Metabolic network measurements provide a sensitive means of quantitatively evaluating disease progression in premanifest individuals. This approach may be incorporated into clinical trials to assess disease-modifying agents.

Trial registration: Registration is not required for observational studies.

Funding: NIH (National Institute of Neurological Disorders and Stroke, National Institute of Biomedical Imaging and Bioengineering) and CHDI Foundation Inc.

Figures

Figure 1. Flow diagram illustrating the design of the study.

109 participants (47 HD gene carriers and 62 age-matched healthy control [HC] subjects) were enrolled. A metabolic progression pattern was identified from the FDG PET scans obtained in a longitudinal cohort (HD1) comprising 12 premanifest HD carriers. Pattern expression was prospectively validated in a crosssectional validation cohort (HD2) comprising 9 premanifest carriers (for the evaluation of test-retest reliability) and 5 early-stage symptomatic HD subjects. The rate of network progression was validated in an independent longitudinal cohort (HD3) comprising 21 premanifest carriers. Metabolic network values were computed in a control group (HC1) comprising 12 healthy subjects and were used to standardize pattern expression in each subject. A second control group (HC2) comprising 20 healthy subjects was used to prospectively evaluate pattern expression in gene-negative individuals. A volume-loss progression pattern was identified in the structural MRI scans obtained in the longitudinal HD1 subjects. Progression rates for this pattern were assessed in both the HD1 and HD3 longitudinal cohorts. Individual scores of the volume-loss pattern were standardized with reference to values computed from a control group (HC4) comprising 18 healthy subjects. In the longitudinal HD1 cohort, the rates of network progression were compared with analogous regional measurements obtained in the caudate and putamen with [11C]-raclopride (RAC) PET (D2 receptor binding) and structural MRI (tissue volume). These regional values were standardized with reference to the corresponding HC3 (comprising 12 healthy subjects) and HC4 control groups.

Figure 2. HD metabolic progression pattern.

(A) This spatial covariance pattern was characterized by areas of declining (blue) and increasing (red) metabolic activity over time. The pattern is displayed as a reliability map of voxel weights thresholded at z = 2.33, P < 0.01 (1-tailed), using a bootstrap resampling procedure (ICV range = –6.02, 5.63, P < 0.0001; 1,000 iterations). (B) All premanifest HD1 subjects exhibited a monotonic increase in pattern expression (P < 0.001; permutation test) across the first 3 time points. (C) In the HD1 longitudinal cohort, pattern expression increased linearly with disease progression (P < 0.0001; IGM) at an estimated progression rate of 0.21/year (95% CI = 0.15, 0.27). Values from the 5 early symptomatic members of the HD2 testing cohort (yellow triangles) are provided for reference. (D) In the HD3 longitudinal testing cohort, pattern expression exhibited a similar linear increase with advancing disease (P < 0.0001; IGM) at a nearly identical rate of 0.19/year (95% CI = 0.11, 0.26). The longitudinal data from each subject are connected by lines. Red lines denote the initially premanifest HD1 subjects who subsequently phenoconverted (i.e., were clinically diagnosed as HD); blue lines denote their counterparts who did not phenoconvert during the study. Values before and after phenoconversion are represented by open and filled symbols. The horizontal broken line represents the mean (equal to 0) of the HC1 healthy control group; the dotted lines represent 2 SD above and below the normal mean. In C and D, the solid lines represent the best model-fitted lines with a 95% CI (broken curves).

Figure 3. Measurements of the HD metabolic progression pattern expression: test-retest reproducibility.

Test-retest reproducibility was excellent (intraclass correlation coefficient [ICC] = 0.96, P < 0.001) for metabolic pattern expression values computed prospectively in the 9 premanifest HD2 subjects who underwent repeat FDG PET studies at 4 imaging sites as part of the PREDICT-HD consortium. The line of identity (dotted line) falls within the 95% CI (broken curves) of the test-retest regression line (solid line). Data from the 4 imaging laboratories that participated in the test-retest study are signified by color. Site 1: North Shore University Hospital (n = 1); site 2: Indiana University (n = 2); site 3: University of Iowa (n = 3); site 4: University of Toronto (n = 3).

Figure 4. HD metabolic progression pattern: regional changes.

In the premanifest HD1 cohort, progressive declines (black lines) in regional metabolic activity (P < 0.005, RMANOVA) were present in (A) the caudate nucleus and anterior putamen, (B) the mediodorsal thalamus, and (C) the prefrontal cortex. In these regions, metabolic activity at all time points was lower for the subjects who phenoconverted at a later point in the study (red) as compared with those who remained clinically premanifest at follow-up (blue). Regional metabolic activity concurrently increased in several regions (Table 1), including (D) the pons (P < 0.001), in which the subsequent phenoconverters exhibited higher values than their nonphenoconverting counterparts at each time point. For each region, MNI coordinates at the peak voxel are given in parentheses. The y axes denote regional metabolic activity ratio-normalized to the global mean. Error bars represent 1 SE above or below the mean of the HD subjects. The broken line represents mean metabolic activity for the HC1 healthy control group; the dotted lines represent 1 SE above and below the normal mean.

Figure 5. Longitudinal changes in preclinical HD progression markers: analysis of nonphenoconverters.

(A) In the nonphenoconverting HD1 subjects, pattern expression increased linearly with disease progression (P = 0.001; IGM) at an estimated rate of 0.20/year (95% CI = 0.09, 0.31). (B) Caudate D2 receptor–binding values measured in the [11C]-raclopride PET scans of the nonphenoconverters exhibited a linear decline with advancing disease (–2.1% of normal mean per year, P < 0.0001; IGM). (C) A similar decline was noted for concurrent MRI measurements of caudate tissue volume in the same subjects (–1.5% of the normal mean per year, P = 0.001; IGM). The solid line depicts the best fitted line with the corresponding 95% CI (broken curves). The dotted lines represent 2 SD above and below the normal mean (dotted horizontal line) for each measure. In B and C, the individual data are represented as percentage of the mean value for age-matched healthy controls. (D) The rate of increase in the expression of the HD metabolic pattern (red) was greater (P ≤ 0.001, IGM) than the rates of decline measured for caudate D2 receptor binding (light blue) and tissue volume (dark blue). To compare network progression (increasing time course) with the corresponding changes in the region-based measures (declining time course), the latter values were flipped (i.e., reflected above the x axes) and plotted as absolute values (dotted lines). The y axes represent the standard z-scale. The horizontal dotted line represents the normal mean (equal to 0) for each parameter. The vertical dotted line represents the estimated time of phenoconversion (i.e., when YTO was 0).

Figure 6. HD volume-loss progression pattern.

(A) This spatial covariance pattern was characterized by progressive loss of tissue volume (blue) in several brain regions (see text). The pattern is displayed as a reliability map thresholded at z = 2.33, P < 0.01 (1-tailed) using a bootstrap resampling procedure (ICV = –8.78, 6.81, P < 0.0001; 1,000 iterations). (B) All premanifest HD1 subjects exhibited a monotonic increase in pattern expression (P < 0.005; permutation test) across the first 3 time points. (C) In the HD1 longitudinal cohort, pattern expression increased linearly with disease progression (P < 0.0001; IGM) at a rate of 0.16/year (95% CI = 0.12, 0.21). (D) In the HD3 cohort, however, longitudinal changes in the expression of the volume-loss pattern were not significant (P = 0.40). Red lines denote the initially premanifest HD1 subjects who subsequently phenoconverted. Blue lines denote their counterparts who did not phenoconvert by the end of the study. Post-phenoconversion values are represented by filled symbols. The horizontal broken line represents the mean (equal to 0) for the healthy control group; the dotted lines represent 2 SD above and below the normal mean. In C and D, the solid line represents the best fit according to the model; the broken curves represent the 95% CI of the fitted line.

Figure 7. Longitudinal changes in striatal D2 receptor binding and tissue volume.

(A) Composite [11C]-raclopride PET images from healthy control subjects (left) and premanifest HD carriers (right) scanned at baseline, 1.5, 4, and 7 years. In each image, dopamine D2 receptor binding was computed voxel-wise as (voxel/occipital-1) and displayed in standard space. The display was thresholded from 1.0 to 3.5. (B) Caudate (left) and putamen (right) D2 receptor–binding values measured using [11C]-raclopride PET exhibited a linear decrease with advancing disease (–2.1% and –1.8% of normal mean per year, P < 0.005; IGM); the decline in the caudate was faster than for the putamen (P < 0.002). (C) Gray matter tissue probability maps from healthy control subjects (left) and the premanifest HD carriers described above (right). Each map represents the average of the gray matter–segmented MRI scans from each group/time point. The display was thresholded from 0.0 to 1.0. (D) Caudate (left) and putamen (right) MRI-based tissue volume measurements also declined linearly with advancing disease (–2.3% and –1.7% of the normal mean per year, P < 0.0001; IGM). Progression rates did not differ for the 2 regions (P = 0.27). In B and D, individual values are represented as percentage of the mean (broken line) for an age-matched healthy control group; the dotted lines represent 2 SD above and below the normal mean. The data from the phenoconverters and nonphenoconverters in the longitudinal HD1 cohort are presented by red and blue lines, respectively. Caudate and putamen values for the symptomatic HD2 subjects (yellow triangles) are provided for reference.

Figure 8. Time course of disease progression: network vs. regional biomarkers.

Linear trajectories (solid lines) of the network and regional imaging measures for the premanifest HD1 longitudinal cohort according to the best-fitting models. The rate of increase in the HD metabolic pattern expression (red) was greater than that for the volume-loss progression pattern (orange; P < 0.05, IGM) and the rates of decline measured for caudate D2 receptor binding (light blue; P < 0.0001) and tissue volume (dark blue; P < 0.0005) (Table 2). To allow for direct comparison of network progression (increasing time course) with corresponding changes in regional measures (decreasing time course), the values for caudate D2 receptor binding and tissue volume were flipped and analyzed as increasing mirror lines (dotted lines). The y axis represents the standard z scale. The horizontal dotted line represents the normal mean (equal to 0) for each parameter. The vertical dotted line represents the time of phenoconversion (i.e., YTO was zero). The estimated value for the metabolic progression pattern at phenoconversion (i.e., the y axis intercept) is signified by a red arrow. The estimated “start time” for the decline of caudate D2 receptor binding (i.e., the x axis intercept) is signified by a light blue arrow. Inset: Bubble plots depicting the estimated rates of disease progression and values at phenoconversion (see Methods) for the HD metabolic and volume-loss patterns (red and orange discs) and for caudate D2 receptor binding and tissue volume (light and blue discs).The diameter of each disc is proportional to the SE for each parameter estimate.