*3.1. Sample*

The sample included 21 JOHD individuals (12 female). From this group, nine participants were seen once, nine were seen twice, two were seen three times, one was seen four times, and three were examined on five occasions, for a total of 49 observations.

There were 115 GNE individuals (64 female). From this group, 60 participants were seen once, 30 were seen twice, 17 were seen three times, and 8 were seen for 4 visits for a total of 203 observations. There were no significant differences in distribution of sex between JOHD and GNE individuals (χ<sup>2</sup> (1, *N* = 137) = 1.10 × <sup>10</sup>−31, *p* = > 0.99). Average elapsed time between follow-up visits was 1.3 years (SD = 1.8 years). All participants were seen between March 2006 and February 2020 with an average of 1.65 visits and median of one visit.

Mean age at evaluation was significantly different between groups: JOHD patients were 15.23 years old on average (SD = 5.55) and GNE individuals were 13.47 years old on average (SD = 3.87 years; t (126.66) = 2.03), *p* = 0.04). CAG repeats ranged from 15 to 34 in the GNE group (median = 19) and from 54 to 102 in the JOHD group (median = 76). Distribution of total motor impairment scores (sum of core UHDRS items) ranged from 19 to 103 in the JOHD group. Average disease duration (defined as age at time of assessment minus age at time of clinical diagnosis) for individuals with JOHD was 3.6 years (SD = 1.5 years), meaning that most JOHD participants were early in the course of the motor manifest stage of the disease. Full group statistics are shown in Table 1.



Note: GNE, Gene-Non-Expanded, i.e., participants with a family history of Huntington's Disease who did not inherit the mutant expansion; JOHD, participants with juvenile onset Huntington's Disease; CAG, CAG repeat expansion length; TMS, total motor score, calculated as cumulative Unified Huntington's Disease Rating (UHDRS) items.

#### *3.2. Behavioral Performance Group Di*ff*erences*

The JOHD group had statistically significantly higher scores than the GNE group on all subscales of the BRIEF (Emotional Control, Inhibit, Shift, Monitor, Plan/Organize, and Working Memory all FDR < 0.001; Organization of Materials FDR = 0.0015; see Figure 1 and Table 2).

**Figure 1.** Differences in Behavior Rating Inventory of Executive Function (BRIEF) subscale scores between JOHD (blue) and GNE (red) participants. The x-axis shows age- and sex-adjusted estimates from mixed linear effects models after controlling for random effects of repeated measures and family ID. The y-axis shows subscales of the Behavioral Rating Inventory of Executive Function (BRIEF). The larger circles represent the means and horizontal lines indicate 95% confidence limits.

The JOHD group had statistically significantly higher scores than the GNE group on the Aggression/Opposition and Hyperactivity/Inattention subscales of the PBS (FDR = 0.00017 and <0.0001 respectively; see Figure 2 and Table 2). In contrast, there were no significant group differences in parent reported measures of Depression/Anxiety and Physical Health subscales of the PBS (both FDR > 0.1). There was no significant main effect of sex for any measure.



Note: BRIEF and PBS raw scores are reported. Diff Means indicates the difference in group mean estimates. MargMean indicates the estimated marginal means for each group. Abbreviations: GNE, Gene-Non-Expanded group; JOHD, Juvenile-Onset Huntington's Disease group. \* indicates False Discovery Rate (FDR) *p*-adjusted < 0.0005.

**Figure 2.** Differences in Pediatric Behavior Scale (PBS) subscale scores between JOHD (blue) and GNE (red) participants. The x-axis shows age- and sex-adjusted estimates from mixed linear effects models after controlling for random effects of repeated measures and family ID. The y-axis shows subscales of the Pediatric Behavior Scale-short form (PBS). The larger circles represent the means and horizontal lines indicate 95% confidence limits.

#### *3.3. Genetic Expansion Correlations*

Within JOHD, all measures had a negative correlation with CAG repeat length, however this reached significance for BRIEF Inhibit (*p* = 0.048), Plan/Organize (*p* = 0.034), and Initiate (*p* = 0.013) subscales of the BRIEF and the Aggression/Opposition (*p* = 0.038) scale of the PBS (see Table 3). A negative association indicates that JOHD participants with the highest CAG repeat tended to have the lowest behavioral scores. All other BRIEF and PBS subscales were not significantly predicted by CAG repeat length.


**Table 3.** Genetic expansion effects in JOHD.

Note: Statistics based on measures that were significantly different between GNE and JOHD groups. JOHD, Juvenile-Onset Huntington's Disease group. Higher means indicate more behavioral/executive problems on the BRIEF and PBS. \* indicates *p*-value < 0.05.

#### *3.4. BRIEF-A Report Type Di*ff*erences*

In total, there were 35 observations for BRIEF-A (JOHD = 13, GNE = 22; see Supplementary Table S3). The difference score in the GNE group was generally close to 0, except for Inhibit (*p* = 0.003) and Working Memory (*p* = 0.002), where informants reported more problems than participants. In contrast, in the JOHD group the difference scores were consistently different from 0, with informants reporting more problems than JOHD patients (all FDR < 0.05; see Figure 3 and Table 4). Age and sex did not have significant effects on the difference between informant- and self-reported scores.

**Figure 3.** Differences in discrepancies between informant- and self-reported BRIEF-Adult (BRIEF-A) subscale scores between JOHD (blue) and GNE (red) participants. The x-axis shows the difference between informant and self-reported scores (calculated as informant—self) on the Behavioral Rating Inventory of Executive Function-Adult form (BRIEF-A); age- and sex-adjusted estimates are plotted from mixed linear effects models after controlling for random effects of family ID. The larger circles represent the means and horizontal lines indicate 95% confidence limits. The vertical black line at 0 marks no difference between informant and self-report scores.


**Table 4.** BRIEF-A model statistics and marginal means.

Note: Statistics based on group differences between the difference in BRIEF-Adult report type, as calculated by informant-report scores—self-report scores. Diff Means indicates the difference in group mean estimates. Marg Mean indicates the estimated marginal means for each group. GNE, gene-non-expanded group; JOHD, juvenile-onset Huntington's Disease group. Higher means indicate more behavioral/executive problems on the BRIEF-A. \* indicates FDR *p*-adjusted < 0.05.

## **4. Discussion**

The Kids-JOHD study is the first ever prospective, longitudinal study of this ultra-rare population. Therefore, this is the first analysis of behavioral symptoms of JOHD that measures behavior on a continuum, rather than using reports of behavioral issues from retrospective analyses of medical records or qualitatively by parents and caretakers of these patients [3,5,8,9,11,16]. Our findings provide quantitative support for the notion that behavioral dysfunction is prevalent among persons with JOHD [3,5,8,9,11,16]. Specific commonly reported symptoms in JOHD include aggressive and oppositional behavior, and difficulties with attention; consistent with this, the largest group differences we found were in parental reports of Hyperactivity/Inattention followed by Aggression/Opposition [3,5,8,10,16].

Importantly, JOHD patients did not exhibit significant anxiety and depression. Different from externalizing behaviors that are easy for others to see, such as aggression and impulsivity, internalizing behaviors associated with subjective feelings of mood and being nervous may be harder to rate objectively [17]. Regardless of the inherent issues in parent reports of internalizing symptoms, it is clear that these symptoms were no more frequent in the JOHD subjects compared to the GNE children, supporting the notion that internalizing behaviors are not significantly affected in JOHD.

In the present study, mutant *HTT* CAG repeat expansion length was negatively correlated with all measures, but reached statistical significance with Inhibit, Plan/Organize, Initiate, and Aggression/Opposition indicating that patients with longer repeat lengths (typically resulting in childhood-onset JOHD [18]) exhibit fewer problems in these domains, while patients with short repeats (typically resulting in adolescent-onset JOHD [18]) exhibit more problem behaviors. These findings align with current reports that older-onset JOHD patients exhibit more behavioral issues than younger-onse<sup>t</sup> JOHD patients [7]. In addition, opposition and aggression behavior normally peak in adolescence; therefore, an active brain disease during a time in which these behaviors normally peak may be one potential rationale for the adolescent onset having greater behavioral disturbance. However, in review of the reports from caregivers of those with childhood onset, problems with aggression and opposition were common early in the course of the disease, years prior to diagnosis. It may be that by the time of motor onset, the childhood onset patients have moved past a period of externalizing behavior [1,7].

Analyses evaluating di fferences between informant and self-reported measures of behavioral regulation and executive function indicated a possible lack of insight among JOHD patients in their behaviors. While the GNE group had similar scores between informant and self-reports, the JOHD participants consistently rated themselves as having fewer behavioral and executive functioning problems than what was reported by their informants. Limited insight into behavioral and cognitive di fficulties is a known feature in patients with AOHD [19]. These results sugges<sup>t</sup> that informant ratings are crucial when quantifying behavioral and executive dysfunction in JOHD.

Parents often report that cognitive and behavior issues are the first harbinger of change and can occur sometimes years prior to final motor diagnosis. Some families in the JOHD community have lobbied for using behavioral changes as a diagnostic criteria for disease [8]. This would be inappropriate for several reasons. Any symptom utilized for diagnosis has to be sensitive and specific to the disease. Although all of the subjects here are already motor manifest, it is important to point out that behavioral symptoms are not present in all patients, therefore these behavioral ratings are not sensitive to the presence of JOHD. Secondly elevated behavioral ratings are not specific to JOHD [6]. There were many children in the GNE group with elevated scores (in fact the two highest scores in the entire sample on hyperactivity and inattention from the PBS came from GNE participants). Although as a group, the JOHD sample had elevated scores compared to GNE, the presence of elevated scores in any one individual is not specific to JOHD. This underscores the notion that changes in behavior should not be utilized for diagnostic purposes for children at risk for AOHD or JOHD.

This study was not without limitations. First, with JOHD being an extremely rare disorder affecting only 1–10% of individuals with HD [1], our sample was limited to 21 individuals with JOHD; however, we leveraged an accelerated longitudinal design to increase the number of observations. Second, parent and informant ratings are objective measures of behavior. Since parents of JOHD are aware of their diagnosis, they may be biased in reporting their child as having greater symptoms, simply knowing that it is commonly known amongs<sup>t</sup> these families that behavioral disturbances occur in children with JOHD. Finally, since self-reported measures were not used in this study for individuals younger than 18, we relied on a small sample for our self vs. informant analyses.
