Analysis of Participant Withdrawal in Huntington Disease Clinical Trials

INTRODUCTION

Huntington disease (HD) is a progressive neurodegenerative disease consisting of motor, cognitive, and psychiatric symptoms. There is currently no treatment known to slow worsening of disease progression [1]. However, several potential disease modifying or symptomatic therapies for HD are currently under development [2, 3]. Retaining enrolled participants is critical for all clinical studies, especially those for rare diseases such as HD, where fewer individuals may be available to participate. Participant retention reduces the problems of bias and decreased power associated with missing data in the analysis of clinical trials for prospective HD therapies [4].

Most clinical trials in HD use specific numerical scores of one or more components of the Unified Huntington’s Disease Rating Scale (UHDRS) to determine eligibility to participate. The UHDRS is a clinical rating scale developed by investigators of the Huntington Study Group and later updated in 1999 (UHDRS 99), consisting of 4 components: motor, behavioral, cognitive, and functional assessments [5]. A change in the score of a specific UHDRS assessment, such as total functional capacity (TFC), often serves as the primary outcome variable for HD trials. All three studies included in this analysis used change in TFC score and/or death as the primary outcome variable [6–8].

To our knowledge, investigation of factors that may be associated with withdrawal from HD clinical trials has not been previously reported. To better understand what factors may be associated with increased probability of withdrawal, we performed analyses of HD participant retention in three clinical trials including; coenzyme Q₁₀ and remacemide in HD (CARE-HD), a pilot study of minocycline in HD (DOMINO), and coenzyme Q₁₀ in HD (2CARE) (Table 1) [6–8].

MATERIALS AND METHODS

RESULTS

Reasons for withdrawal

Probabilities of withdrawal at 12 months varied across the three clinical trials analyzed: CARE-HD (2.9%), DOMINO (10.5%), and 2CARE (5.9%) (Fig. 1). The reasons for withdrawal varied slightly among the three studies. Mean percentages for each category across studies revealed that loss to follow-up was the most common reason for withdrawal (23.2%), followed by death (15.9%), loss of interest/desire to participate (15.2%), adverse events (11.3%), interest in other clinical trials/medications (7.1%), disease progression (5.0%), and difficulty traveling (4.0%) (Fig. 2). The category of “other” reasons was especially common in the CARE-HD trial (31.6%).

OPEN IN VIEWER

Fig.1

Kaplan-Meier curves showing the estimated probabilities of participant withdrawal in three Huntington disease clinical trials.

OPEN IN VIEWER

Fig.2

Reasons for withdrawal in three Huntington disease clinical trials.

DISCUSSION

Early withdrawal of participants has an impact on the statistical power and interpretability of a study, so understanding the reasons for its occurrence is important. Screening potential trial participants for certain baseline characteristics that are associated with higher withdrawal rates, and potentially excluding those individuals from enrollment in a trial, may preserve statistical power. However, this consideration should be balanced with the ability to recruit trial participants. Identification of time-dependent covariates that are associated with earlier time to withdrawal may also permit identification of participants at greater risk of withdrawal. These participants could then benefit from more targeted efforts for retention.

Analyses of three clinical trials assessing pharmacologic therapies in individuals with HD revealed some of the common reasons for study withdrawal for HD participants. These reasons included loss to follow-up, death, loss of interest/desire to participate, adverse events, interest in other clinical trials/medications, disease progression, and difficulty traveling. The most common reason for withdrawal varied slightly among the three trials, which may have been due to specific eligibility criteria, treatment safety profile, and duration of follow-up. It should be noted that the reasons for withdrawal reported in the original trial publications may differ from those reported here due to the more detailed process for categorizing reasons for withdrawal in the current investigation.

Loss to follow-up was the most common reason for withdrawal when examining the mean percentage across the three trials, but the percentage of withdrawals due to this reason varied among the trials: 21.1% of withdrawals in CARE-HD, 21.1% in DOMINO, and 27.6% in 2CARE. Longer duration of follow-up in 2CARE (60 months), compared to CARE-HD (30 months) and DOMINO (18 months), may partly account for these findings. In addition, follow-up visits in 2CARE were scheduled at 6-month intervals, which may have influenced the rate of loss to follow-up. Interim phone calls to participants/caregivers may contribute to maintaining good relationships between enrolling sites and participants. The 2nd most common reason for withdrawal was death (15.9%). The percentage of withdrawals attributed to death was highest in 2CARE (24.1%), which may also be due to the longer duration of follow-up in this study.

Loss of interest/desire to participate was the 3rd most common reason for study withdrawal, but was least frequently seen in DOMINO. This may be due to the shorter duration of follow-up in this study. Adverse events were the 4th most common reason for study withdrawal with the highest percentage being in DOMINO. This may be secondary to greater intolerance of minocycline compared to coenzyme Q₁₀. In DOMINO, there were 4 withdrawals due to adverse events in the minocycline group and none in the placebo group. Future studies that allow dosage reductions in the setting of adverse events, assess more well-tolerated drugs, or allow follow-up evaluations off of study drug may allow greater retention in a study. Of the other three categories used, disease progression is unfortunately inevitable; however, withdrawal for this reason may be reduced by enrolling individuals with higher baseline TFC scores or having a shorter duration of follow-up. In future trials, withdrawal due to difficulty traveling may be reduced by several methods including: requiring the participation of a caregiver who will accompany the subject to study visits, providing paid transportation to study sites for participants, or allowing telemedicine study visits in place of some clinical site visits. A participant’s interest in other trials cannot be easily anticipated so it would be more difficult to address this.

The average probability of withdrawal at 12 months for these trials was 6.4%. The numbers of participants who withdrew from the DOMINO (n = 19) and CARE-HD (n = 38) trials were low, causing decreased power to detect significant associations of predictors with time to withdrawal in these studies. However, a considerably larger number of withdrawal events was observed in 2CARE, likely due to the larger sample size and longer duration of follow-up. In 2CARE, worse baseline total motor score and worse functional assessment and independence scale scores were associated with shorter time to withdrawal. Time-dependent variables in 2CARE that were associated with a shorter time to withdrawal included these same variables in addition to worse scores on the symbol digit modalities test, behavioral assessment, and TFC. Evaluation of all 3 studies indicated that gender, marital status, education, ethnicity, and age were not associated with time to withdrawal.

One of the limitations of this study is the use of subjective categorization of withdrawal reasons. The categorization process was strengthened by having two independent evaluators (both neurologists) analyze cases of withdrawal, followed by common consensus of categorization between these two individuals and an additional statistician and neurologist. Another limitation is the performance of multiple statistical tests. To mitigate this issue, we used a relatively stringent significance level of 0.01 (two-tailed). Nevertheless, we consider this work preliminary and hypothesis generating rather than confirmatory. In addition, we could not analyze other factors that may have affected withdrawal rate that were not available for analysis such as socioeconomic status, health of caregiver, child care issues, owing a car, and the ability to drive independently.

There are several aspects of the selected studies that should be considered when attempting to generalize our results to future HD clinical trials. The trials considered here were designed and executed by an academic research consortium with investigators having subspecialty expertise and being devoted to HD research and clinical care. Also, understanding of the disease process continues to evolve among patients and caregivers in the HD community, as does an appreciation of the importance of clinical trials. Advances in technology such as wearable sensors, e-diaries, and telemedicine visits may also have a profound effect on retention in future HD clinical trials.

Although not examined in this study, another design feature that may increase retention in HD clinical trials is the requirement of a caregiver or individual that could give consent for future study related activities if a participant lost capacity for decision making during the course of a prolonged trial. Since this was a requirement in the 2CARE trial we were unable to examine the effect of the presence or absence of this factor in our analysis. The 2CARE study also added the option of follow-up via telephone for some visits later in the trial if a person had difficulty getting to a clinical trial site. Future clinical trials may also address difficulty with traveling by including the option of completing some visits via telemedicine. However, clinical trial visits using telemedicine technology produce their own challenges, including feasibility and reliability of study assessments and investigator licensing regulations that may vary by state and country [9].

Ultimately, if the objective of a clinical trial in HD is to assess the neuroprotective effect of an intervention, eligibility criteria requiring higher baseline functional status and better behavioral and motor scores on the UHDRS assessments would likely be advantageous. Previous studies have shown that strategies to improve protocol adherence and participant retention may include enrolling people at early disease stages such as in amyotrophic lateral sclerosis clinical trials [10]. In line with this hypothesis, higher executive control and visual memory performance was shown to be associated with treatment completion in borderline personality disorder clinical trials [11]. On the other hand, restricting eligibility criteria may have implications on recruitment rate, which both add to the duration and cost of the trial. Additional studies are needed to examine the impact of eligibility criteria on enrollment, recruitment, and statistical power.

CONFLICTS OF INTEREST

Merit Cudkowicz has provided consultation for Biohaven, Mitsubishi Tanabe Pharma, Biogen, Immunity Pharma, Karyopharm, Cytokinetics, and Denali Therapeutics. Nazem Atassi has provided consultation for Biogen and Mitsubishi Tanabe Pharma. Other authors have no conflicts of interest to report.