Longitudinal Psychiatric Symptoms in Prodromal Huntington’s Disease: A Decade of Data

Huntington’s disease is an autosomal dominant neurodegenerative disease that typically manifests in adulthood with motor, cognitive, and psychiatric symptoms. The disease occurs as a result of an expanded number of cytosine-adenine-guanine (CAG) trinucleotide repeats in the protein-coding DNA sequence of the Huntingtin gene on chromosome 4. A formal diagnosis of Huntington’s disease is usually given based on the presence of significant motor symptoms, including chorea, rigidity, and bradykinesia. Huntington’s disease is progressive over several years, leading to functional decline and premature death (1, 2). Cognitive changes include progressive deficits in learning, executive and sensory functions, and attention, resulting in dementia (3–6). Psychiatric manifestations in Huntington’s disease include depression, irritability, apathy, perseverations, obsessions, and occasionally psychosis (7–14). With a few exceptions (7, 15), psychiatric changes are not typically related to indices of progression.

The neurodegeneration of striatal and cortical structures underlying the symptomatic presentation of Huntington’s disease has been known for some time, and while there is an association between the number of CAG repeats and age at onset, there is significant unexplained variability in the presentation and severity of symptoms, in particular their psychiatric manifestations. The ability to test for the Huntington’s disease mutation prior to onset of symptoms, known as the illness prodrome, has provided the opportunity to study the development of disease manifestations over time. The longitudinal Neurobiological Predictors of Huntington’s Disease study (PREDICT-HD) has identified motor, cognitive, psychiatric, and brain imaging changes that occur in individuals with the mutation (16–19). We previously reported an increase in psychiatric syndromes for individuals in the prodrome of Huntington’s disease at the time of study entry, and we have shown increases in number and severity of symptoms cross-sectionally as estimated proximity to motor diagnosis increases (4, 19–21).

Here, we report longitudinal psychiatric ratings from up to 10 years of PREDICT-HD follow-up in order to better characterize psychiatric symptoms that occur in the Huntington’s disease prodrome up to and through motor conversion. We have also developed statistical models to evaluate whether there are differences in the trajectories of psychiatric manifestations as reported by participants compared with companions, which has implications for psychiatric assessment in Huntington’s disease and other disorders in which changes in awareness are prominent (4, 22).

Method

Participants

Participants were from the PREDICT-HD study at 33 sites in six countries (the United States, Canada, Germany, Australia, Spain, and the United Kingdom). Eligible participants had to have independently undergone testing for the Huntington’s disease gene mutation and had to know their gene status prior to study participation. That is, independent from study enrollment, every participant was aware of his or her family history and gene mutation status for Huntington’s disease. The study did not enroll participants who were at risk but had not undergone predictive testing for the Huntington’s disease mutation. Recruitment efforts included talks at regional, national, and international Huntington’s disease lay meetings, flyer dissemination to genetic counseling and Huntington’s disease clinics, and development of a PREDICT-HD web site with links to and from other Huntington’s disease sites. Individuals with the CAG repeat expansion (CAG length ≥36) (23, 24) served as case participants, and those who tested negative for the gene expansion (CAG length <36) served as controls. All participants were at least 18 years old and had not been diagnosed with manifest Huntington’s disease at study entry. Exclusion criteria included unstable medical or psychiatric illness, active substance abuse, history of a significant developmental cognitive disorder, significant history of head trauma or CNS disease, presence of a pacemaker or other metallic implants, use of antipsychotic medication in the 6 months before enrollment, or use of a phenothiazine antiemetic medication in the 3 months before enrollment. The study protocol was approved by each site’s institutional review board. All participants provided written informed consent after receiving a complete description of the study.

All participants were seen at study sites annually. The analysis used data from 1,305 participants (6,112 observations) and 1,235 companions (5,365 observations). Companions were predominately a spouse or partner (74%), followed by a friend or neighbor (8%), a parent (7%), and a sibling (5%); approximately 75% of companions reported living with the participant. Companions reported having known the participants for a mean of 20.77 years (SD=13.59). The median number of follow-up visits was five (range, 1–10).

Progression Groups

Individuals entered PREDICT-HD with different genetic exposure according to their CAG repeat length and current age. To yield valid inferences, participants with the expanded gene sequence were classified into three groups based on their CAG-age product score (25), computed as age at baseline × (CAG length − 33.66). The CAG-age product formula was derived from a parametric accelerated failure time model predicting motor diagnosis from age at study entry, CAG length, and their interaction. The CAG-age product score is similar to the “disease burden” score of Penny et al. (26) and purports to index the cumulative toxicity of mutant Huntingtin at the time of study entry. Cutoffs were derived for the best-fitting subgroups based on an optimization algorithm using an earlier sample of PREDICT-HD participants (25). Based on gene status and the CAG-age product distribution, four groups were defined in this analysis: control (gene nonexpanded) and low, medium, and high probability of motor diagnosis within 5 years. The estimated time to motor diagnosis for each CAG-age product group is >12.8 years for the low-probability group, 7.6–12.8 years for the medium-probability group, and <7.6 years for the high-probability group.

Measures

The Symptom Checklist–90–Revised (SCL-90-R) is a 90-item respondent-report scale of psychiatric symptoms, with each item rated on a scale of 0, not at all, to 4, extremely, based on the degree of distress over the previous 7 days (27). Three global measures of psychiatric symptoms and scores for nine specific symptom domains are produced from the scale: a global severity index, a positive symptom total, a positive symptom distress index, and domain scores for somatization, obsessive-compulsive symptoms, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psychoticism. For all participants, the SCL-90-R was administered by a trained staff member. The PREDICT-HD protocol called for annual examinations including standardized cognitive, motor, functional, and psychiatric ratings as well as brain scans.

Statistical Analysis

Baseline characteristics were compared by group, using analysis of variance for continuous variables and chi-square tests for categorical variables. All tests were two-tailed.

Group comparison of baseline and longitudinal change.

To examine potential differences between the controls and each of the three groups with the CAG repeat expansion at baseline and over time, participant and companion ratings on the SCL-90-R were analyzed separately using linear mixed-effects regression (28). The time metric was duration, defined as time since study entry, expressed in years. All models included sex, years of education, and age at study entry as covariates. Three linear mixed-effects regression models were fitted for each outcome variable: no effect, baseline differences only, and baseline and longitudinal differences among groups. The models were evaluated using Akaike’s information criterion (AIC), corrected for small-sample bias (AICc). The effect sizes were the t-values of either the intercept (baseline difference) or slope difference (longitudinal change) among groups, each computed as the difference of the sample estimates divided by its standard error. Details of all statistical analyses are provided in the data supplement that accompanies the online edition of this article.

Participant and companion comparison.

To test whether there were longitudinal differences between participant and companion ratings on the SCL-90-R in each group, participant and companion ratings were modeled simultaneously using multiresponse linear mixed-effects regression. All models included sex, years of education, and age at study entry as covariates. The models were assessed by AICc values, as described above, to examine which groups had statistically reliable slope differences. The relative importance of the longitudinal change discrepancy of each group was assessed by the sum of the weights (wAICc) across all models with unequal group slopes, as described by Burnham and Anderson (29). Sums closer to 1 indicate higher importance. Model parameters were averaged over all the candidate models after multiplying the weight of the model and the estimated parameters for the given model. Using these model-averaged parameters over all models, fitted curves for participant and companion ratings were drawn as described by Burnham and Anderson (29).

Results

Baseline and Longitudinal Change in Self-Reported and Companion-Reported SCL-90-R Ratings

Participant characteristics at study entry are presented in Table 1. As expected, there were significant age differences between the prodromal Huntington’s disease groups, with increasing age associated with increasing proximity to estimated motor diagnosis.

Of the 24 psychiatric outcome variables examined (12 participant ratings and 12 companion ratings), 19 showed significant baseline and longitudinal differences between participants with prodromal Huntington’s disease and control participants (Table 2); eight of these were participant ratings (global severity index, positive symptom total, positive symptom distress index, obsessive-compulsive symptoms, interpersonal sensitivity, depression, hostility, and phobic anxiety) and 11 were companion ratings (global severity index, positive symptom total, positive symptom distress index, somatization, obsessive-compulsive symptoms, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, and psychoticism). Three participant ratings (anxiety, paranoid ideation, psychoticism) and one companion rating (paranoid ideation) showed only baseline differences between participants with prodromal Huntington’s disease and controls (i.e., no longitudinal differences).

Tables 3 and 4 show the comparison of each prodromal group (low, medium, and high probability of motor diagnosis within 5 years) with the controls on each baseline and longitudinal psychiatric outcome. As shown in Table 3, 11 of the 12 participant psychiatric variables and 11 of 12 of the companion psychiatric variables showed significant differences at baseline between the controls and the prodromal Huntington’s disease groups with medium and high probabilities of diagnosis. In the prodromal Huntington’s disease group with a low probability of motor diagnosis within 5 years, nine of the 12 participant-rated psychiatric variables and only one of the companion-rated psychiatric outcomes were significantly different from those for the controls. When the groups were compared longitudinally (Table 4), companion-rated psychiatric outcomes showed significant change for 11 of the 12 outcomes for the high-probability group, six of the 12 outcomes for the medium-probability group, and two of the 12 outcomes for the low-probability group. Participant-reported psychiatric symptoms in the prodromal Huntington’s disease participants showed significant change over time on seven of the 12 measures for the high-probability group. None of the participant-rated variables showed significant change over time for the medium- or low-probability groups.

Comparisons Between Participant and Companion Ratings

A multiresponse linear mixed-effects regression analysis was performed to examine whether the differences in longitudinal change among participants and companions were statistically reliable. When the importance of the longitudinal change discrepancy between participants and companions was compared across CAG-age product groups for each measure, as shown in the Table 5, differences (weight >0.7) were observed for seven measures: global severity index, positive symptom total, obsessive-compulsive symptoms, depression, anxiety, phobic anxiety, and paranoid ideation. For all seven measures, the high-probability group had relatively high importance, with weights of 0.75 and above. The medium-probability group had relatively high importance only for positive symptom total and obsessive-compulsive symptoms. The low-probability group did not have relatively high importance for any measure. For these seven measures, fitted curves are displayed in Figure 1 using model-averaged coefficients across all possible models. For the other five measures, which did not have groups with longitudinal change discrepancies (positive symptom distress index, somatization, interpersonal sensitivity, hostility and psychoticism), fitted curves are provided in the online data supplement.

Discussion

To our knowledge, this is the largest published longitudinal study of psychiatric manifestations in prodromal Huntington’s disease. Some previous research has suggested that severity of psychiatric and/or behavioral manifestations of Huntington’s disease is not associated with progression of the disease (2, 30). Contrary to this popularly held belief, however, 19 of 24 measures examined showed cross-sectional and longitudinal differences between participants with prodromal Huntington’s disease and gene-mutation-negative controls. Of the remaining five measures analyzed, three also showed elevated baseline scores, although change over time did not differ. Although previous research has consistently emphasized the clinical importance of the psychiatric symptoms associated with Huntington’s disease, and a wealth of studies have noted the higher prevalence of psychiatric symptoms and increased psychiatric distress in Huntington’s disease, few studies have documented longitudinal progression of the psychiatric disturbances in Huntington’s disease. These findings are consistent with the report of Craufurd et al. (7), which provided longitudinal data for 111 persons with a diagnosis of Huntington’s disease and found that all three subscales of the Problem Behaviors Assessment for Huntington’s Disease showed significant change over time. Furthermore, Craufurd et al. showed that although all three psychiatric subscales (apathy, irritability, depression) worsened over time in the earliest stages of Huntington’s disease (stages I and II), only apathy continued to progress with disease progression into stages II–IV. Comparisons with other research in prodromal Huntington’s disease, however, are more central to the present findings. Kirkwood et al. (31) followed 45 prodromal gene mutation carriers and reported a greater increase in psychiatric abnormalities over an interval of 3.7 years. The greatest changes were noted in irritability and hostility. More recently, Tabrizi et al. (32) followed 120 prodromal Huntington’s disease gene mutation carriers and reported worsening apathy over 36 months. The present study replicates and extends these findings and provides confidence that the psychiatric symptoms seen in Huntington’s disease progress with disease severity and thus are likely secondary to the neurodegenerative disease process.

The detection of neurodegenerative diseases has progressed considerably, as reflected in DSM-5, which describes the emergence of mild cognitive impairment as a precursor, or prodromal stage, evident in progressive brain disease. Mild cognitive impairment in Huntington’s disease has been documented in at least 40% of persons carrying the gene mutation (33, 34), and numerous studies have demonstrated that cognitive decline can be detected a decade before onset of motor symptoms (35). Given the plethora of publications demonstrating early cognitive decline in Huntington’s disease, in concert with the findings of the present study showing psychiatric symptom progression prior to onset of motor symptoms, the monosymptomatic diagnostic criteria seem out of date. Since mental health professionals may see prodromal Huntington’s disease in clinical practice a decade before movement disorder specialists do, an earlier diagnosis may be prudent for many individuals.

The present findings suggest that the source of measurement can significantly influence psychiatric measures in prodromal and diagnosed Huntington’s disease. All findings from this research were more robust when psychiatric symptoms were rated by companions (rather than self-reported by participants with the CAG repeat expansion) for those with prodromal Huntington’s disease. For instance, companion ratings worsened over time for 11 of 12 symptoms in the prodromal group with a high probability of imminent motor diagnosis, six of 12 symptoms in those with a medium probability of diagnosis, and two of 12 symptoms in those considered to be far from their estimated motor onset. Only seven of 12 participant ratings showed significant change over time, and only in the group in closest proximity to motor diagnosis. When analyses were conducted to examine the importance of discrepant ratings between companions and the prodromal participants, findings showed that ratings on seven of 12 psychiatric symptom scales were inconsistent between the raters. In participants closer to estimated motor onset, the companion ratings appear to be more valid, whereas in gene-mutation carriers who are furthest from estimated motor onset, self-reported psychiatric symptoms may be more appropriate.

These findings are consistent with cross-sectional observations that individuals with Huntington’s disease have decreased awareness of symptoms (36–39). It is important to note that even in carriers of the Huntington’s disease mutation who do not yet have a motor diagnosis, decreased awareness of both motor (22) and psychiatric (4) symptoms has been documented. The present study is the first to provide longitudinal data of reduced awareness in a prodromal Huntington’s disease cohort over such a long follow-up period. Our findings indicate that unawareness increases in concert with Huntington’s disease progression during the prodromal disease stages, as indicated by participant-companion differences arising mostly in the medium and high CAG-age product groups.

Difficulties with awareness or insight occur across the spectrum of neuropsychiatric disorders, including stroke, brain injury, dementia, psychosis, and mood disorders (40). The etiologies of these disparate conditions vary, and it is not known whether there are common mechanisms that underlie the development of unawareness across disorders. Several studies in different disorders have noted associations between decreased awareness of symptoms and pathology in the right hemisphere, dorsolateral frontal regions, and, less often, the parietotemporal region. However, the findings are not consistent enough to indicate shared mechanisms with certainty. The results presented here and in other studies of Huntington’s disease symptoms and brain pathology are consistent with these anatomic findings, but more research is needed to support this connection. As described previously by others (41), it is likely that neural networks connecting several regions are responsible for the phenomenon of decreased awareness. The presentation of lack of awareness also varies among disorders. In frontotemporal dementia, for example, insight is reduced early in the course of the illness, whereas in Alzheimer’s dementia this is a later manifestation (42). Unawareness is also irreversible in dementias, whereas it is often transient in other disorders, such as stroke, traumatic brain injury, mood disorders, and schizophrenia. The progressive brain degeneration that occurs in Huntington’s disease may indicate that reduced insight is permanent, but additional study is needed to understand its course and causes.

There are some limitations to consider in the interpretation of these results. Some issues are related to the nature of the assessments. With the SCL-90-R, respondents are asked to provide ratings for the previous 7 days. As these measures were obtained annually, they may have missed episodes of psychopathology that occurred between assessments. While these findings support changes in participant-companion reporting over time, more detailed assessments at more time points would be beneficial. Another issue is related to potential rater biases. Both participants and companions were aware of participants’ Huntington’s disease genetic status, which may have affected their perception of symptoms. Sample selection bias may also affect the generalizability of these results. Only a fraction of individuals at risk for Huntington’s disease obtain genetic testing, which is required prior to enrollment in the PREDICT-HD study. There may be factors unique to those within the larger Huntington’s disease population who obtain genetic testing and participate in research studies. It is interesting to observe, however, that the sample in a recent study of pre-manifest Huntington’s disease in which genetic testing was not a requirement had demographic characteristics similar to our sample, including education levels (43). Differences in awareness may also be present in our sample compared with the overall population. However, as previously described by Duff et al. (4), it is equally likely that individuals with lower or higher awareness would participate in the study.

Despite these limitations, the results reported here provide initial information regarding psychiatric symptoms that occur in individuals who will develop Huntington’s disease and the importance of obtaining assessments from companions. This is critical to consider in future studies that assess behavioral manifestations and psychiatric symptoms, including those that investigate their underlying pathophysiology in Huntington’s disease and in therapeutic trials, as well as clinical assessment and management of persons who will develop Huntington’s disease.