Aberrant Striatal Value Representation in Huntington's Disease Gene Carriers 25 Years Before Onset

Background In this study, we asked whether differences in striatal activity during a reinforcement learning (RL) task with gain and loss domains could be one of the earliest functional imaging features associated with carrying the Huntington's disease (HD) gene. Based on previous work, we hypothesized that HD gene carriers would show either neural or behavioral asymmetry between gain and loss learning. Methods We recruited 35 HD gene carriers, expected to demonstrate onset of motor symptoms in an average of 26 years, and 35 well-matched gene-negative control subjects. Participants were placed in a functional magnetic resonance imaging scanner, where they completed an RL task in which they were required to learn to choose between abstract stimuli with the aim of gaining rewards and avoiding losses. Task behavior was modeled using an RL model, and variables from this model were used to probe functional magnetic resonance imaging data. Results In comparison with well-matched control subjects, gene carriers more than 25 years from motor onset showed exaggerated striatal responses to gain-predicting stimuli compared with loss-predicting stimuli (p = .002) in our RL task. Using computational analysis, we also found group differences in striatal representation of stimulus value (p = .0004). We found no group differences in behavior, cognitive scores, or caudate volumes. Conclusions Behaviorally, gene carriers 9 years from predicted onset have been shown to learn better from gains than from losses. Our data suggest that a window exists in which HD-related functional neural changes are detectable long before associated behavioral change and 25 years before predicted motor onset. These represent the earliest functional imaging differences between HD gene carriers and control subjects.

reported with McFadden's pseudo-R 2 calculated as 1 -log likelihood of the model divided by log likelihood of null model (in which choices are determined by chance). This model reinforcement learning model provided good model fits for both gains and losses (McFadden's pseudo-R 2 gain = 0.65 +/-0.27, pseudo-R 2 loss = 0.48 +/-0.25, mean ± STD) with no difference in fits between the groups for either gains or losses (Zgains = 0.81, p = 0.42, Zlosses = -0.83, p = 0.40). Furthermore, comparing this model to a three-parameter model (including a reward multiplier term) and models in which the initial q value (q0) was both 0 and treated as a free parameter, the model described above performed best in terms of model comparison based on summed BICs across participants and valences (Fig. S2) Go and No-Go response in gains and loss conditions were approximately 50% as expected and not significantly different between groups and was not considered further (Gain-Go Controls: 47.2% ± 0.07, HDGC: 46.9% ± 0.07, pgains = 0.86, Losses-Go: Controls: 45% ± 0.08, HDGC: 48% ± 0.08 plosses = 0.09, (mean ± STD)).

fMRI image acquisition:
Each volume contained 48 slices with a 3mm 3 resolution. Volume TR was 3.36 seconds with a slice tilt of -30 degrees, a Z-shim of -0.4 and ascending slice acquisition order. T1 weighted images were collected for structural alignment and volumetric analysis. The 3D T1-weighted sequence was an optimised MPRAGE protocol, with an echo time (TE) of 3.34ms and a repetition time (TR) of 2530ms.
The inversion time was 1100ms, and the flip angle was 7 degrees. The field of view was 256x256x176mm, with 1mm isotropic voxels. Parallel imaging acceleration (GeneRalized Autocalibrating Partial Parallel Acquisition, GRAPPA, acceleration factor (R)=2) was applied along with 3D distortion correction and pre-scan normalisation. Following the task, field maps were acquired for unwarping. Physiological monitoring of heartbeat and breathing were recorded for 31 of 35 participants. Excluding participants with missing physiological monitoring did not influence results Table S1). fMRI pre-processing:

(Supplementary
Images were processed using SPM12. Images were un-warped using acquired field maps, slice-time correction to the middle slice, corrected for motion, and then warped into Montreal Neurological Institute (MNI) template space and spatially smoothed with a Gaussian kernel of 6-mm. Following unwarping and warping to standard space images underwent manual quality check. No subjects were excluded for excessive head motion or quality assurance reasons.

Structural imaging processing:
All T1-weighted scans passed visual quality control checks for the absence of significant motion or other artefacts before processing. Bias correction was performed using the N3 procedure. An automated segmentation procedure, Multi-Atlas Label Propagation with Expectation-Maximisation based refinement (MALP-EM), was used to measure caudate volume (1). All settings were applied using default parameters, except for the inclusion of a brain mask for each participant based on a previously generated whole-brain region derived from semi-automated delineation. MALP-EM has been validated for use in HD.   Participants were presented initially with a fixation cross. This was replaced by two abstract symbols.
Participants saw three such abstract pairs -a 'gain pair', 'loss pair' and 'neutral pair'. Using a button box in their right-hand participants were instructed to press the button to choose the top symbol or withhold their response for 3 seconds to choose the lower symbol. Their choice was displayed with a red marker before the outcome was revealed. In the gain pair, one symbol was associated with reward