[Skip to Content]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address 18.206.168.65. Please contact the publisher to request reinstatement.
[Skip to Content Landing]
Figure.
Kaplan-Meier Survival Curves (Stepped) and Cubic Spline Curves (Smooth) by Follow-up Year and Study for Different Survival End Points
Kaplan-Meier Survival Curves (Stepped) and Cubic Spline Curves (Smooth) by Follow-up Year and Study for Different Survival End Points

Curves are unadjusted for covariates. COHORT indicates Cooperative Huntington Observational Research Trial; DCL, diagnostic confidence level.

Table 1.  
Descriptive Statistics for Variables Measured at Study Entry (Baseline)
Descriptive Statistics for Variables Measured at Study Entry (Baseline)
Table 2.  
Clinically Meaningful Change Point Estimate for a Hypothetical 3-Year Triala
Clinically Meaningful Change Point Estimate for a Hypothetical 3-Year Triala
Table 3.  
Training and Test Study Survival Comparisons for Various End Pointsa
Training and Test Study Survival Comparisons for Various End Pointsa
Table 4.  
Required Total Sample Size for a 3-Year, 2-Arm Parallel Trial as a Function of Trial Condition and End Point
Required Total Sample Size for a 3-Year, 2-Arm Parallel Trial as a Function of Trial Condition and End Point
1.
Ross  CA, Tabrizi  SJ.  Huntington’s disease: from molecular pathogenesis to clinical treatment.  Lancet Neurol. 2011;10(1):83-98.PubMedGoogle ScholarCrossref
2.
Biglan  KM, Shoulson  I, Kieburtz  K,  et al; Huntington Study Group PHAROS Investigators.  Clinical-genetic associations in the Prospective Huntington at Risk Observational Study (PHAROS): implications for clinical trials.  JAMA Neurol. 2016;73(1):102-110.PubMedGoogle ScholarCrossref
3.
Ross  CA, Aylward  EH, Wild  EJ,  et al.  Huntington disease: natural history, biomarkers and prospects for therapeutics.  Nat Rev Neurol. 2014;10(4):204-216.PubMedGoogle ScholarCrossref
4.
Huntington Study Group.  Unified Huntington’s Disease Rating Scale: reliability and consistency.  Mov Disord. 1996;11(2):136-142.PubMedGoogle ScholarCrossref
5.
Booth  CM, Eisenhauer  EA.  Progression-free survival: meaningful or simply measurable?  J Clin Oncol. 2012;30(10):1030-1033.PubMedGoogle ScholarCrossref
6.
Saad  ED, Katz  A.  Progression-free survival and time to progression as primary end points in advanced breast cancer: often used, sometimes loosely defined.  Ann Oncol. 2009;20(3):460-464.PubMedGoogle ScholarCrossref
7.
Soria  JC, Massard  C, Le Chevalier  T.  Should progression-free survival be the primary measure of efficacy for advanced NSCLC therapy?  Ann Oncol. 2010;21(12):2324-2332.PubMedGoogle ScholarCrossref
8.
Tabrizi  SJ, Langbehn  DR, Leavitt  BR,  et al; TRACK-HD Investigators.  Biological and clinical manifestations of Huntington’s disease in the longitudinal TRACK-HD study: cross-sectional analysis of baseline data.  Lancet Neurol. 2009;8(9):791-801.PubMedGoogle ScholarCrossref
9.
Tabrizi  SJ, Scahill  RI, Durr  A,  et al; TRACK-HD Investigators.  Biological and clinical changes in premanifest and early stage Huntington’s disease in the TRACK-HD study: the 12-month longitudinal analysis.  Lancet Neurol. 2011;10(1):31-42.PubMedGoogle ScholarCrossref
10.
Tabrizi  SJ, Scahill  RI, Owen  G,  et al; TRACK-HD Investigators.  Predictors of phenotypic progression and disease onset in premanifest and early-stage Huntington’s disease in the TRACK-HD study: analysis of 36-month observational data.  Lancet Neurol. 2013;12(7):637-649.PubMedGoogle ScholarCrossref
11.
Dorsey  E; Huntington Study Group COHORT Investigators.  Characterization of a large group of individuals with Huntington disease and their relatives enrolled in the COHORT study.  PLoS One. 2012;7(2):e29522.PubMedGoogle ScholarCrossref
12.
World Medical Association.  World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects.  JAMA. 2013;310(20):2191-2194.PubMedGoogle ScholarCrossref
13.
Crosby  RD, Kolotkin  RL, Williams  GR.  Defining clinically meaningful change in health-related quality of life.  J Clin Epidemiol. 2003;56(5):395-407.PubMedGoogle ScholarCrossref
14.
Geraci  M, Bottai  M.  Linear quantile mixed models.  Stat Comput. 2014;24(3):461-479.Google ScholarCrossref
15.
Royston  P, Parmar  MKB.  Restricted mean survival time: an alternative to the hazard ratio for the design and analysis of randomized trials with a time-to-event outcome.  BMC Med Res Methodol. 2013;13:152-162.PubMedGoogle ScholarCrossref
16.
Cox  DR, Oakes  D.  Analysis of Survival Data. London, England: Chapman & Hall/CRC; 1984.
17.
Machin  D, Campbell  MJ, Tan  SB, Tan  SH.  Sample Size Tables for Clinical Studies. Hoboken, NJ: Wiley-Blackwell; 2009.
18.
Noether  GE.  Sample size determination for some common nonparametric tests.  J Am Stat Assoc. 1987;82(398):645-647. doi:10.1080/01621459.1987.10478478Google ScholarCrossref
19.
Rahardja  D, Zhao  YD.  Unified sample size calculations using the competing probability.  Stat Biopharm Res. 2009;1(3):323-327. doi:10.1198/sbr.2009.0022Google ScholarCrossref
20.
Acion  L, Peterson  JJ, Temple  S, Arndt  S.  Probabilistic index: an intuitive non-parametric approach to measuring the size of treatment effects.  Stat Med. 2006;25(4):591-602.PubMedGoogle ScholarCrossref
21.
US Dept of Health and Human Services; Food and Drug Administration, Center for Drug Evaluation and Research; Center for Biologics Evaluation and Research. Guidance for industry: clinical trial endpoints for the approval of cancer drugs and biologics. https://www.fda.gov/downloads/Drugs/Guidances/ucm071590.pdf. Published May 2007. Accessed February 1, 2017.
22.
Wason  JMS, Dentamaro  A, Eisen  TG.  The power of phase II end-points for different possible mechanisms of action of an experimental treatment.  Eur J Cancer. 2015;51(8):984-992.PubMedGoogle ScholarCrossref
23.
Tomlinson  G, Detsky  AS.  Composite end points in randomized trials: there is no free lunch.  JAMA. 2010;303(3):267-268.PubMedGoogle ScholarCrossref
24.
Eisenhauer  EA, Therasse  P, Bogaerts  J,  et al.  New Response Evaluation Criteria In Solid Tumours: revised RECIST guideline (version 1.1).  Eur J Cancer. 2009;45(2):228-247.PubMedGoogle ScholarCrossref
25.
Therasse  P, Arbuck  SG, Eisenhauer  EA,  et al.  New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada.  J Natl Cancer Inst. 2000;92(3):205-216.PubMedGoogle ScholarCrossref
26.
Shoulson  I, Young  AB.  Milestones in Huntington disease.  Mov Disord. 2011;26(6):1127-1133.PubMedGoogle ScholarCrossref
Original Investigation
November 2017

Survival End Points for Huntington Disease Trials Prior to a Motor Diagnosis

Author Affiliations
  • 1Department of Psychiatry, Carver College of Medicine, University of Iowa, Iowa City
  • 2Department of Biostatistics, Carver College of Medicine, University of Iowa, Iowa City
  • 3Centre for Molecular Medicine and Therapeutics, Department of Medical Genetics, University of British Columbia, Vancouver, British Columbia, Canada
  • 4Department of Genetics and Cytogenetics, and INSERM UMR S679, Institut du Cerveau et de la Moelle Epinière, Hôpital de la Salpêtrière, Paris, France
  • 5Department of Neurology, Leiden University Medical Centre, Leiden, Netherlands
  • 6School of Psychology and Psychiatry, Monash University, Melbourne, Victoria, Australia
  • 7Department of Neurology, University of Münster, Münster, Germany
  • 8Cure Huntington’s Disease Initiative Foundation, Department of Neurology, University of Ulm, Ulm, Germany
  • 9Huntington’s Disease Research Centre, Institute of Neurology, University College London, London, England
  • 10Institute of Neurology, University College London, London, England
  • 11Huntington's Disease Centre, Department of Neurodegenerative Disease, Institute of Neurology, University College London, Queen Square, London, England
JAMA Neurol. 2017;74(11):1352-1360. doi:10.1001/jamaneurol.2017.2107
Key Points

Question  Can survival end points for prediagnosis Huntington disease trials be developed to provide feasible sample sizes?

Findings  This study involving participants in 2 Huntington disease studies found that progression-free survival—a composite of a motor diagnosis or a progression event—yields much smaller sample sizes than a motor diagnosis event alone.

Meaning  The progression-free survival end point may provide feasible sample sizes for clinical trial planning for Huntington disease gene expansion mutation carriers who have not yet received a motor diagnosis.

Abstract

Importance  Predictive genetic testing in Huntington disease (HD) enables therapeutic trials in HTT gene expansion mutation carriers prior to a motor diagnosis. Progression-free survival (PFS) is the composite of a motor diagnosis or a progression event, whichever comes first.

Objective  To determine if PFS provides feasible sample sizes for trials with mutation carriers who have not yet received a motor diagnosis.

Design, Setting, and Participants  This study uses data from the 2-phase, longitudinal cohort studies called Track and from a longitudinal cohort study called the Cooperative Huntington Observational Research Trial (COHORT). Track had 167 prediagnosis mutation carriers and 156 noncarriers, whereas COHORT had 366 prediagnosis mutation carriers and noncarriers. Track studies were conducted at 4 sites in 4 countries (Canada, France, England, and the Netherlands) from which data were collected from January 17, 2008, through November 17, 2014. The COHORT was conducted at 38 sites in 3 countries (Australia, Canada, and the United States) from which data were collected from February 14, 2006, through December 31, 2009. Results from the Track data were externally validated with data from the COHORT. The required sample size was estimated for a 2-arm prediagnosis clinical trial. Data analysis took place from May 1, 2016, to June 10, 2017.

Main Outcomes and Measures  The primary end point is PFS. Huntington disease progression events are defined for the Unified Huntington's Disease Rating Scale total motor score, total functional capacity, symbol digit modalities test, and Stroop word test.

Results  Of Track’s 167 prediagnosis mutation carriers, 93 (55.6%) were women, and the mean (SD) age was 40.06 (8.92) years; of the 156 noncarriers, 87 (55.7%) were women, and the mean (SD) age was 45.58 (10.30) years. Of the 366 COHORT participants, 229 (62.5%) were women and the mean (SD) age was 42.21 (12.48) years. The PFS curves of the Track mutation carriers showed good external validity with the COHORT mutation carriers after adjusting for initial progression. For required sample size, PFS with a motor diagnosis or total motor score progression required about 4 times fewer participants than a motor diagnosis alone. Including additional cognitive progression events further reduced the number. For example, a 3-year trial with 10% attrition and a treatment effect of 50% requires a total of 661 with motor diagnosis as the survival end point but only 177 with a total motor score PFS.

Conclusions and Relevance  Reasonably sized prediagnosis Huntington disease trials can be planned with PFS, and there is evidence of generalizability of this approach.

Introduction

Huntington disease (HD) is a devastating neurodegenerative disorder caused by a cytosine-adenine-guanine expansion on the HTT (OMIM 613004) gene of chromosome 4.1 Huntington disease is autosomal dominant with reduced penetrance for 36 to 39 repeats and full penetrance for 40 repeats or more.2 Progression leads to a triad of signs and symptoms, including motor, cognitive, and behavioral features.3 A reliable predictive genetic test is available that can be used to establish whether an individual is an HTT gene expansion mutation carrier prior to the emergence of any signs or symptoms. Early identification of carrier status enables trials to examine whether a therapeutic intervention might prevent or delay the pathological processes that lead to disease onset.

A landmark event in HD is motor diagnosis, which is determined by the standard motor examination score on the Unified Huntington’s Disease Rating Scale (UHDRS).4 Motor diagnosis is the highest rating on the UHDRS Diagnostic Confidence Level (DCL), which indicates the rater is at least 99% confident that the motor abnormalities displayed by the patient are unequivocal signs of HD. Despite the prominent role of motor diagnosis in HD research, it is reluctantly used as an end point in clinical trials.2 The reason for this reluctance is HD is a slow progressing disease and only a small number of motor diagnosis events for cohorts are followed over the short periods typical of clinical trials.

In HD therapy development, preventive clinical trials for mutation carriers who have not yet been diagnosed are being planned. With this population, it is crucial to have better-defined, feasible, and cost-effective end points over time. The number of events can be increased by considering a secondary variable associated with the definitive event of a motor diagnosis but has a faster rate of change. This alternative approach is known as progression-free survival (PFS), which is widely used in oncology trials.5-7Progression-free survival is defined as the time elapsed from treatment initiation to the first event of HD progression or motor diagnosis, whichever comes first.

This study examines PFS using 7 years of data from the Track observational study8-10 and evaluates the extent of reproducibility using data from the Cooperative Huntington Observational Research Trial (COHORT),11 a separate independent study. Our hypothesis is that PFS will provide sufficient events for the planning of feasible prediagnosis clinical trials and show reasonable generalizability.

Methods
Standard Protocol Approvals, Registrations, and Patient Consents

Study activities were reviewed and approved by the local ethics committees for the Track study and the institutional review boards for the COHORT study (the University of Rochester and each site approved the protocol). Written patient informed consent, according to the Declaration of Helsinki,12 was obtained from each participant (if unable, the participant had an authorized representative provide consent on their behalf), including consent for the distribution of deidentified data for research purposes. Data analysis took place from May 1, 2016, to June 10, 2017.

Study Population

Primary analysis involved HTT gene expansion mutation carriers and noncarriers from 2 phases of the Track study: Track-HD and Track-On. Track-HD is a longitudinal cohort study comprising prediagnosis and postdiagnosis mutation carriers and healthy noncarriers and was conducted at 4 sites in 4 countries (Canada, France, England, and the Netherlands) from which data were collected for the period from January 17, 2008, through November 17, 2014.8-10 The inclusion criteria for Track-HD were 18 to 65 years of age, tolerance for magnetic resonance imaging scans and biosample collection, absence of major psychiatric disorder or history of significant head injury, not active in an experimental therapeutic trial, and no comorbid medical conditions preventing assessment. Noncarriers were selected from spouses or partners of carriers or noncarrier siblings. Noncarriers were matched by age and sex to the carriers.

Track-On is also a longitudinal cohort study of mutation carriers and noncarriers. One hundred twenty-three of the 167 prediagnosis carriers (74%) and 99 of the 156 noncarriers (64%) transitioned to Track-On from Track-HD. Participants who transitioned could have a maximum of 7 years of data, from which data were collected for the period from January 17, 2008, through November 17, 2014. Table 1 shows descriptive statistics for key variables at study entry (baseline) by carrier status.

The external validation for this study involved participants in COHORT, a longitudinal cohort study of 366 HTT gene expansion mutation carriers and noncarriers conducted at 38 sites in 3 countries (Australia, Canada, and the United States) and from which data were collected for the period February 14, 2006, through December 31, 2009. Enrollment in COHORT was open to people who had tested positive for the HTT gene expansion mutation (prediagnosis or postdiagnosis) or people who were untested but had a family history of HD. Noncarriers were family members verified by genetic testing to not have the expansion mutation. Only confirmed prediagnosis mutation carriers were considered for the validation analysis. Table 1 shows descriptive statistics for the COHORT sample along with statistical comparisons with the Track carrier group. Note that the COHORT sample was significantly more progressed at baseline, as indicated by the clinical measures.

Measures

Progression-free survival required a definitive end point, which was motor diagnosis. It also required at least 1 secondary variable associated with motor diagnosis but potentially had a faster rate of change. The secondary variables considered were the UHDRS total motor score (TMS), total functional capacity, the symbol digit modalities test, and the Stroop word test.

End Points

Two types of survival end points were examined. The end point for traditional survival analysis was motor diagnosis (DCL = 4), and the time to first occurrence was analyzed. Progression-free survival is a type of composite event triggered by either a progression event or a motor diagnosis, whichever comes first. Thus, the time to the composite event was used for the PFS analysis.

The progression event was a change from baseline of sufficient size to be deemed important, referred to as clinically meaningful change (CMC).13 Statistical methods for CMC estimation are described. A progression event occurs for an individual if the change on the secondary variable meets or exceeds the CMC. For example, consider a CMC of 3 for the TMS and an individual has 4 time points with a TMS of 8, 9, 12, or 13 at baseline (0) and 1, 2, and 3 years of follow-up. The TMS difference from baseline is 0, 1, 4, or 5, and we assume that a DCL of 4 does not occur. The third and fourth values meet the criterion for a progression event, and the time to the PFS event is 2 years because it is the first instance of the progression event (time points after the event are ignored). Progression-free survival requires the composite event to occur after baseline; if there is no composite event over the observed epoch, the individual is considered right censored, which means the composite event will happen sometime in the future.

Statistical Analysis
Clinically Meaningful Change

Track data were used to develop the CMC. For the CMC to accurately reflect disease effects as opposed to normal aging effects, we used an extreme score based on analysis with only the noncarriers. Change from baseline was computed for each follow-up time, which represented change due to natural aging (although there were practice effects for the symbol digit modalities test and the Stroop word test). Extremes of these changes were computed on the basis of quantile mixed models for follow-up after baseline (eMethods in the Supplement). The quantile mixed model accounted for the dependency resulting from repeated measures, but unlike traditional mixed models, a quantile was estimated as a function of time rather than the mean.14 Change from baseline was regressed onto time on study, age at baseline, and their interaction. The 99th quantile was estimated for TMS because an increase from baseline indicated greater HD progression. Conversely, the first quantile was estimated for total functional capacity, the symbol digit modalities test, and the Stroop word test because a decrease from baseline indicated greater progression.

External Validation

To assess the replicability (generalizability) of the end points, the CMC developed in Track was applied to the COHORT data. Mutation carriers from both studies were combined, study membership was coded, and then the survival profiles of the studies were compared using 2 statistical methods. The first method was the Wald test of study difference (z score) using the Cox proportional hazards regression model, and the second method was the likelihood ratio test of study difference based on smoothed cubic spline survival models.15 The null hypothesis for both tests was that the survival curves of the studies were equivalent, possibly adjusting for covariates. Thus, evidence of the reproducibility of the PFS curves on the basis of Track CMC would be provided if there was no statistically significant study difference; significance was defined as a 2-sided P < .05 for all results. Spline modeling provided smooth survival curves that were not unduly affected by final times being event times (Figure). Study differences were examined without and with adjustment for covariates. The covariates were all baseline variables that showed a significant study difference from Table 1, except for follow-up.

Required Sample Size

A popular test of the equivalence of survival curves among groups is the log-rank test.16 Sample size can be estimated from standard formulas when testing the null hypothesis of equivalent survival curves, under the assumptions of proportional hazards and exponentially distributed survival times (eMethods in the Supplement).17 Sample size estimates for the log-rank test required an estimate of the survival proportion at study end. Survival proportions were estimated on the basis of the cubic spline survival curves. Sample sizes were estimated using the conventional type I error rate of 5% (2-tailed test) and type II error rate of 20% (power = 80%). Estimates were for a 3-year, 2-arm parallel trial with equal group sample size. To allow for attrition, the total sample size, N, was adjusted by the dropout rate, w. The adjustment for dropout was Nw = N/(1 − w), where w = 0, 0.10, 0.20.

To provide a benchmark for judging the performance of the nonparametric log-rank test, sample size was also estimated using the 2-group Mann-Whitney-Wilcoxon (MWW) test of TMS at only the last time point. The MWW is a test of difference in group TMS medians when the group distributions are identical except for a location shift.18 When the assumptions of the MWW test hold, it is more efficient than the log-rank test and will yield a smaller sample size.19

The treatment effect size was defined as the hypothetical proportion reduction (π) in the treatment TMS mean (μT) relative to the placebo mean (μP) at the study terminus: π = |(μP − μT)|/μP. The Track prediagnosis mutation carriers were treated as a proxy for the placebo group, and the hypothetical improvement in the treatment TMS mean was computed as μT = μP(1 − π). The quantity π was related to Cohen d as d = μP(π)/σ, where σ2 was the common group variance. For the Track data, the 3-year visit mean and SD were estimated to be

Image description not available.

thus, a 50% TMS mean reduction produced d = 6.59(0.50)/5.86 = 0.56. Cohen d was not appropriate for the nonparametric log-rank test or MWW test. Therefore, Cohen d was transformed into the area under the curve (AUC) of the receiver operating characteristic curve in medical diagnostic testing.20

The AUC is a nonparametric effect size on the 0-1 scale and has a convenient probability interpretation for the log-rank test. The AUC is the probability that a randomly sampled patient from the treatment group will delay an HD progression event longer than will a randomly sampled patient from the placebo group.19 There was no treatment effect when AUC = 0.50 because there was an equal chance of longer delay for both groups; only when there was an AUC > 0.50 did we have longer delay for the treatment group. For the aforementioned 50% TMS mean reduction (with d = 0.56), the effect size was AUC = 0.65 (eMethods in the Supplement).

Results

Of Track’s 167 prediagnosis mutation carriers, 93 (55.6%) were women, and the mean (SD) age was 40.06 (8.92) years; of the 156 noncarriers, 87 (55.7%) were women, and the mean (SD) age was 45.58 (10.30) years. Of the 366 COHORT participants, 229 (62.5%) were women, and the mean (SD) age was 42.21 (12.48) years. Results of the CMC analysis using the Track noncarrier group are shown in Table 2, which shows the point estimates and 95% bootstrap CIs for select ages. Because the UHDRS variables take only integer values, the CMC point estimates can be rounded up to the next extreme integer. For example, the CMC for TMS is 3 for 30 to 35 years of age, 4 for 40 to 50 years of age, and so on. The CMC for total functional capacity is −1 (loss of 1) for 35 years of age or older.

Results of the study survival curve comparison are presented in Table 3. Kaplan-Meier curves and fitted spline curves without covariate adjustment are shown in the Figure. Kaplan-Meier probabilities descend to 0 for some curves because the final time is an event time (spline curves are unaffected by this occurrence). The Figure shows that individuals from the COHORT had a greater risk of an event than those from the Track study, with the TMS curves being most similar. Likewise, the upper portion of Table 3 indicates significant study survival curve differences without covariate adjustment for all end points except TMS. The bottom portion of Table 3 shows that, after adjusting for baseline variables, the study differences were no longer significant, except for the symbol digit modalities test, which was not considered further for this reason.

The estimated total sample size is shown in Table 4 for a hypothetical 3-year parallel-arm clinical trial with 10% attrition (see eTable in the Supplement for results based on other attrition rates). As expected, traditional survival analysis based on the DCL had the largest estimated sample size. Progression-free survival with TMS progression showed substantially lower sample sizes than the DCL alone, being almost 4 times smaller for most effect sizes. However, PFS based on TMS progression had sample sizes that were approximately 1.5 times larger than the smallest possible sizes of the MWW test. Combining TMS with Stroop word test progression lowered the sample sizes to the point of being only about 1.3 times larger than the sizes of the MWW test.

Discussion

The goal of this study was to define CMC and an HD progression event for use in preventive trials with HTT gene expansion mutation carriers prior to receiving a motor diagnosis. The Track study was favorable for CMC analysis because it was conducted as a clinical trial but without an active treatment group.8-10 Clinically meaningful change developed in Track was used to define PFS end points, and the survival curves were found mostly to be similar for the COHORT data, especially after adjusting for progression differences at study entry. Therefore, the CMC values that we developed appear to be reasonable general indexes for defining HD progression events in clinical trial planning.

The required sample size for a clinical trial can be greatly reduced when a TMS progression event is used in combination with a motor diagnosis, which is consistent with our explicit hypothesis. For example, a 3-year trial with 10% attrition and a treatment effect of 50% requires a total of 661 with motor diagnosis as the survival end point but only 177 with a total motor score PFS. A motor diagnosis based on the DCL is perhaps the closest the field has to a gold standard for a landmark progression event in HD. As expected, survival analysis with time to motor diagnosis yielded the largest required sample size, whereas the PFS end points offered substantial reduction. With the help of PFS, one can retain the definitive outcome of a motor diagnosis while providing enough HD progression events for a reasonably sized trial. Total motor score progression, for example, requires approximately 4 times fewer participants for the range of effect sizes considered. It is notable that PFS is approved by the US Food and Drug Administration as a surrogate end point for cancer trials.21 Such an endorsement is an encouraging sign that PFS might eventually be successful in pivotal trials of HD.

The potential advantage of PFS is that it incorporates motor diagnosis and produces enough events to increase the feasibility of prediagnosis trials. This is not to say that PFS necessarily yields smaller sample sizes than do traditional methods. It involves dichotomizing continually measured UHDRS variables, which reduces information and can lead to lower efficiency.22 Our results show that, although PFS can drastically lower the sample size compared with using motor diagnosis alone, the sample size is still larger than a method that compares the TMS among groups at the last visit. Therefore, if efficiency is the only criterion for choosing an end point, then PFS may be less attractive than the traditional methods.

The primary appeal of PFS is that it is anchored to the event of a motor diagnosis, which is a well-established landmark in the progression of HD. In addition, PFS may be desirable because its effect size can be expressed as a probabilistic statement of potential patient benefit. For instance, AUC = 0.60 means that a given person in the treatment group has a 60 to 40 chance of delaying an HD progression event compared with a person in the untreated group. Huntington disease prevention therapy (eg, repeated lumbar punctures) will likely be demanding for participants, and treatment may only be desirable to pursue if the chance of benefit is sufficient in the minds of both the participants and researchers. The AUC effect size provides a clear means of articulating a minimum potential benefit.

The question remains as to what minimum AUC is acceptable. Pilot data can provide an indication of effect sizes that are in reach, or HD stakeholders can decide on a minimum probability. We offer the opinion that AUC values smaller than 0.60 seem to get uncomfortably close to a 50-50 chance of delay, which is just a coin flip’s chance. Furthermore, our results indicate that AUC = 0.60 requires a total N = 364 for the TMS end point in a 2-arm study with 10% dropout. When AUC is less than 0.57, more than 500 participants total will be required. The feasibility of a particular sample size depends on many factors, but we believe planning for a study with no greater than 400 participants will increase the likelihood of a trial being conducted. The lower bound of AUC = 0.60 is consistent with this goal.

The PFS end point is a composite by definition. There is recent increased interest in composite end points because of the potential of smaller studies and lower costs.23 An advantage of PFS is that the composite is defined in a clear manner, which does not require weights for combining variables. Because key variables are collected at the standard UHDRS examination, it is feasible to use such composite end points at little cost. The caveat here, as with any composite end point, is a potential lack of clarity regarding the nature of the effects. For example, PFS will not distinguish an individual who has a TMS progression event from another who has a Stroop word test event. Assuming a treatment has a benefit, the benefit must be reported in terms of potentially delaying the package of HD progression events. Perhaps PFS using TMS is most clear because the composite focuses only on motor signs and a treatment benefit can be expressed broadly as a delay in HD motor progression.

A particular CMC threshold does not necessarily speak to the importance of the progression event in the experience of a mutation carrier. In clinical trials in which PFS is used, an expert panel typically convenes to confirm the appropriateness of the CMC as the basis of a progression event. For example, panels of oncology experts have been convened to determine the criteria for a solid-tumor increase to define a tumor progression event, resulting in the Response Evaluation Criteria in Solid Tumors (RECIST) guidelines for cancer trials.24,25 Our CMC results are not meant to substitute for an expert panel; they are offered as database results that might be informative to such a panel.

The focus of this study is on preventive trials involving confirmed mutation expansion carriers who have not yet received a motor diagnosis. A mutation carrier may not have a motor diagnosis at prescreening but may have a diagnosis at the first visit. This situation is problematic because such a person must be analyzed according to the intent-to-treat principle, even though the person has 0 event time that is uninformative. An approach to address this problem is to disregard DCL status in the analysis and only use time to the TMS progression event as the outcome. The TMS progression event is defined relative to study entry; thus, there will always be an analyzable event time. Analysis with this modified end point is called time-to-progression analysis.5 Practically speaking, in our analysis, there is no difference in the number of events between time to progression and PFS (results not presented) because every individual who is eventually assigned a motor diagnosis has an earlier TMS progression event. Thus, motor diagnosis does not contribute to the event status of PFS, and it is equivalent to time to progression. The drawback, however, is that the method does not have the definitive end point of a motor diagnosis (which is ignored).

Limitations

The finding that the mutation carriers from the COHORT were more advanced at study entry than those from the Track study is likely due to differences in recruitment strategy. Track-HD explicitly aimed to recruit prediagnosis carriers who were relatively far from a motor diagnosis, whereas COHORT did not. The external validation considered mutation carriers in both studies who were willing to undergo genetic testing. We provide evidence that the participants from the 2 studies are similar after adjusting for baseline progression. However, it is unknown if our results generalize to the broader HD population because most at-risk individuals do not undergo genetic testing.26

Conclusions

Minimum values are proposed for assessing clinically meaningful change over time for HTT gene expanded mutation carriers who have not yet received a motor diagnosis. The change values can be used to define progression events that are easy to combine, yield trials of reasonable size, and may apply across studies. This approach is especially appealing when a researcher wants to examine whether a treatment delays motor onset.

Back to top
Article Information

Corresponding Author: Jeffrey D. Long, PhD, Department of Psychiatry, Carver College of Medicine, University of Iowa, 500 Newton Rd, Iowa City, IA 52242-1000 (jeffrey-long@uiowa.edu).

Accepted for Publication: June 13, 2017.

Published Online: September 18, 2017. doi:10.1001/jamaneurol.2017.2107

Author Contributions: Dr Long and Mr Mills had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Long, Roos, Stout, Landwehrmeyer, Langbehn, Tabrizi.

Acquisition, analysis, or interpretation of data: Long, Mills, Leavitt, Durr, Stout, Reilmann, Gregory, Scahill, Langbehn, Tabrizi.

Drafting of the manuscript: Long, Gregory, Tabrizi.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Long, Mills, Langbehn, Tabrizi.

Obtained funding: Roos, Stout, Langbehn, Tabrizi.

Administrative, technical, or material support: Roos, Stout, Scahill.

Study supervision: Leavitt, Roos, Stout, Reilmann, Landwehrmeyer, Tabrizi.

Conflict of Interest Disclosures: Dr Long reported having a consulting agreement with NeuroPhage; being a paid consultant for Roche Pharma and Azevan Pharmaceuticals; and receiving funding from the Cure Huntington’s Disease Initiative (CHDI) Foundation, the Michael J. Fox Foundation, and the National Institutes of Health (NIH). Dr Leavitt reported receiving research grants from Teva, the CHDI Foundation, and Lifemax, as well as personal consulting fees from Roche, uniQure, Novartis, Lifemax, and Raptor. Dr Durr reported receiving research grants from the French Agency for Research, the French Ministry for Social Affairs and Health, Pfizer Inc, and Annapurna Therapeutics, as well as partly holding patent B 06291873.5 on anaplerotic therapy of Huntington disease and other polyglutamine diseases. Dr Roos reported receiving grants from the Gossweiler Foundation and Teva, both via the administration of the Leiden University Medical Center, and being an advisor for UniQure. Dr Stout reported having served on an advisory board for Roche; having consulted for Prana Biotechnology; being a treasurer and board member for the Huntington's Study Group, Inc; and conducting business implementing cognitive assessments at Stout Neuropsych Pty Ltd, with contracts from Teva, Vaccinex, Omeros, and Ionis. Dr Reilmann reported being the founding director and owner of the George-Huntington-Institut, a private research institute focused on clinical and preclinical research in Huntington disease, and QuantiMedis, a clinical research organization providing quantitative motor services in clinical trials and research, as well as holding appointments at the Department of Radiology of the University of Münster and at the Department of Neurodegenerative Diseases and the Hertie Institute for Clinical Brain Research of the University of Tübingen. Dr Reilmann also provided consulting services, advisory board functions, clinical trial services, quantitative motor analyses, and/or lectures for Teva, Pfizer, uniQure, Ipsen, Vaccinex, Novartis, Raptor, Omeros, Siena Biotech, Neurosearch Inc, Lundbeck, Medivation, Wyeth, ISIS Pharma, Link Medicine, Prana Biotechnology, MEDA Pharma, Temmler Pharma, Desitin, AOP Orphan, and the CHDI Foundation; he received grant support from the High Q Foundation, the CHDI Foundation, the Deutsche Forschungsgemeinschaft (DFG), the European Union EU-FP7 Program, the Bundesministerium fur Bildung und Forschung (BMBF), the Deutsches Zentrum fur Neurodegeneration und Entzundung, and the European Huntington’s Disease Network (EHDN). Dr Landwehrmeyer reported having provided consulting services, advisory board functions, and clinical trial services and/or lectures for Allergan, Alnylam, Aventis, Amarin, AOP Orphan Pharmaceuticals AG, Bayer Pharma AG, Desitin, Genzyme, GlaxoSmithKline, Hoffmann–La Roche, Ipsen, ISIS Pharma, Lundbeck, Neurosearch Inc, Medesis, Medivation, Medtronic, Novartis, Pfizer, Prana Biotechnology, Raptor Pharmaceuticals, Sangamo/Shire, Sanofi-Aventis, Siena Biotech, Temmler Pharma GmbH, Trophos, and Teva; he has received research grant support from the CHDI Foundation, the BMBF, the DFG, and the European Commission EU-FP7 Program, and his study site in Ulm, Germany, has received compensation in the context of the observational REGISTRY Study of the EHDN. Dr Langbehn reported receiving research funding from the CHDI Foundation, Neurology Institute of the University College of London, and the NIH/National Institute on Drug Abuse (grant 5R01AA021165-03), as well as being a paid statistical consultant for Roche and Voyager for Huntington disease clinical trial design. Dr Tabrizi reported having served on advisory boards or being a consultant for F. Hoffmann–La Roche Ltd, Ionis Pharmaceuticals, Ixico Technologies, Shire Human Genetic Therapies, Takeda Pharmaceuticals International, and Teva Pharmaceuticals; all honoraria paid for these consultancies and advisory boards go to University College London, Dr Tabrizi’s employer.

Funding/Support: The Track-HD group was supported in part by the CHDI Foundation, a not-for-profit organization dedicated to finding treatments for persons with Huntington disease.

Role of the Funder/Sponsor: Several members of the CHDI Foundation management sat on the executive committee of the Track study. The other funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Group Information: The Track-HD investigators are as follows: C. Campbell, M. Campbell, E. Frajman, I. Labuschagne, C. Milchman, A. O’Regan, and J. Stout (Monash University, Victoria, Australia); A. Coleman, R. Dar Santos, J. Decolongon, and A. Sturrock (University of British Columbia, Vancouver, Canada); E. Bardinet, C. Jauffret, D. Justo, S. Lehericy, C. Marelli, K. Nigaud, and R. Valabrègue (APHP, Hôpital Salpêtriere, Paris, France); N. Bechtel, S. Bohlen, and R. Reilmann (University of Münster, Münster, Germany); B. Landwehrmeyer (University of Ulm, Ulm, Germany); A. Hoffman and P. Kraus (University of Bochum, Bochum, Germany); S. J. A. van den Bogaard, E. M. Dumas, J. van der Grond, E. P. t’Hart, C. Jurgens, and M.-N. Witjes-Ane (Leiden University Medical Centre, Leiden, the Netherlands); N. Arran, J. Callaghan, D. Craufurd, and C. Stopford (St Mary’s Hospital, Manchester, England); C. Frost and R. Jones (London School of Hygiene and Tropical Medicine, London, England); H. Crawford, N. C. Fox, C. Gibbard, N. Hobbs, N. Lahiri, I. Malone, R. Ordidge, G. Owen, T. Pepple, J. Read, M. J. Say, and D. Whitehead (University College London, London, England); S. Keenan (Imperial College London, London, England); D. M. Cash (IXICO, London, England); C. Berna, S. Hicks, and C. Kennard (University of Oxford, Oxford, England); T. Acharya, E. Axelson, H. Johnson, and C. Wang (University of Iowa, Iowa City); B. Borowsky (CHDI Foundation, New York, NY); S. Lee and W. Monaco (Massachusetts General Hospital, Harvard, Boston); and C. Campbell, S. Queller, and K. Whitlock (Indiana University, Bloomington).

The Track-On investigators are as follows: I. Labuschagne (Monash University, Victoria, Australia); A. Coleman, J. Decolongon, M. Fan, and T. Petkau (University of British Columbia, Vancouver, Canada); C. Jauffret, D. Justo, S. Lehericy, K. Nigaud, and R. Valabrègue (ICM and APHP, Hôpital Salpêtriere, Paris, France); N. Weber (George-Huntington-Institut, Münster, Germany); I. Mayer and M. Orth (University of Ulm, Ulm, Germany); A. Schoonderbeek and E. P. t’Hart (Leiden University Medical Centre, Leiden, the Netherlands); D. Craufurd (St Mary’s Hospital, Manchester, England); A. Cassidy, C. Frost, and R. Keogh (London School of Hygiene and Tropical Medicine, London, England); C. Berna, H. Crawford, M. Desikan, R. Ghosh, D. Hensman Moss, E. Johnson, P. McColgan, G. Owen, M. Papoutsi, J. Read, A. Razi, and D. Mahaleskshmi (University College London, London, England); H. Johnson (University of Iowa, Iowa City); and B. Borowsky (CHDI Foundation, New York, NY).

Additional Contributions: We thank the Track-HD study participants and their families. We also acknowledge the support of the National Institute for Health Research University College London Hospitals Biomedical Research Centre and the Manchester Biomedical Research Centre. Samples and/or data from COHORT, which received support from HP Therapeutics, Inc., were used in this study. We thank the Huntington Study Group COHORT investigators and coordinators, who collected data and/or samples for this study, as well as all participants and their families, who made this work possible.

References
1.
Ross  CA, Tabrizi  SJ.  Huntington’s disease: from molecular pathogenesis to clinical treatment.  Lancet Neurol. 2011;10(1):83-98.PubMedGoogle ScholarCrossref
2.
Biglan  KM, Shoulson  I, Kieburtz  K,  et al; Huntington Study Group PHAROS Investigators.  Clinical-genetic associations in the Prospective Huntington at Risk Observational Study (PHAROS): implications for clinical trials.  JAMA Neurol. 2016;73(1):102-110.PubMedGoogle ScholarCrossref
3.
Ross  CA, Aylward  EH, Wild  EJ,  et al.  Huntington disease: natural history, biomarkers and prospects for therapeutics.  Nat Rev Neurol. 2014;10(4):204-216.PubMedGoogle ScholarCrossref
4.
Huntington Study Group.  Unified Huntington’s Disease Rating Scale: reliability and consistency.  Mov Disord. 1996;11(2):136-142.PubMedGoogle ScholarCrossref
5.
Booth  CM, Eisenhauer  EA.  Progression-free survival: meaningful or simply measurable?  J Clin Oncol. 2012;30(10):1030-1033.PubMedGoogle ScholarCrossref
6.
Saad  ED, Katz  A.  Progression-free survival and time to progression as primary end points in advanced breast cancer: often used, sometimes loosely defined.  Ann Oncol. 2009;20(3):460-464.PubMedGoogle ScholarCrossref
7.
Soria  JC, Massard  C, Le Chevalier  T.  Should progression-free survival be the primary measure of efficacy for advanced NSCLC therapy?  Ann Oncol. 2010;21(12):2324-2332.PubMedGoogle ScholarCrossref
8.
Tabrizi  SJ, Langbehn  DR, Leavitt  BR,  et al; TRACK-HD Investigators.  Biological and clinical manifestations of Huntington’s disease in the longitudinal TRACK-HD study: cross-sectional analysis of baseline data.  Lancet Neurol. 2009;8(9):791-801.PubMedGoogle ScholarCrossref
9.
Tabrizi  SJ, Scahill  RI, Durr  A,  et al; TRACK-HD Investigators.  Biological and clinical changes in premanifest and early stage Huntington’s disease in the TRACK-HD study: the 12-month longitudinal analysis.  Lancet Neurol. 2011;10(1):31-42.PubMedGoogle ScholarCrossref
10.
Tabrizi  SJ, Scahill  RI, Owen  G,  et al; TRACK-HD Investigators.  Predictors of phenotypic progression and disease onset in premanifest and early-stage Huntington’s disease in the TRACK-HD study: analysis of 36-month observational data.  Lancet Neurol. 2013;12(7):637-649.PubMedGoogle ScholarCrossref
11.
Dorsey  E; Huntington Study Group COHORT Investigators.  Characterization of a large group of individuals with Huntington disease and their relatives enrolled in the COHORT study.  PLoS One. 2012;7(2):e29522.PubMedGoogle ScholarCrossref
12.
World Medical Association.  World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects.  JAMA. 2013;310(20):2191-2194.PubMedGoogle ScholarCrossref
13.
Crosby  RD, Kolotkin  RL, Williams  GR.  Defining clinically meaningful change in health-related quality of life.  J Clin Epidemiol. 2003;56(5):395-407.PubMedGoogle ScholarCrossref
14.
Geraci  M, Bottai  M.  Linear quantile mixed models.  Stat Comput. 2014;24(3):461-479.Google ScholarCrossref
15.
Royston  P, Parmar  MKB.  Restricted mean survival time: an alternative to the hazard ratio for the design and analysis of randomized trials with a time-to-event outcome.  BMC Med Res Methodol. 2013;13:152-162.PubMedGoogle ScholarCrossref
16.
Cox  DR, Oakes  D.  Analysis of Survival Data. London, England: Chapman & Hall/CRC; 1984.
17.
Machin  D, Campbell  MJ, Tan  SB, Tan  SH.  Sample Size Tables for Clinical Studies. Hoboken, NJ: Wiley-Blackwell; 2009.
18.
Noether  GE.  Sample size determination for some common nonparametric tests.  J Am Stat Assoc. 1987;82(398):645-647. doi:10.1080/01621459.1987.10478478Google ScholarCrossref
19.
Rahardja  D, Zhao  YD.  Unified sample size calculations using the competing probability.  Stat Biopharm Res. 2009;1(3):323-327. doi:10.1198/sbr.2009.0022Google ScholarCrossref
20.
Acion  L, Peterson  JJ, Temple  S, Arndt  S.  Probabilistic index: an intuitive non-parametric approach to measuring the size of treatment effects.  Stat Med. 2006;25(4):591-602.PubMedGoogle ScholarCrossref
21.
US Dept of Health and Human Services; Food and Drug Administration, Center for Drug Evaluation and Research; Center for Biologics Evaluation and Research. Guidance for industry: clinical trial endpoints for the approval of cancer drugs and biologics. https://www.fda.gov/downloads/Drugs/Guidances/ucm071590.pdf. Published May 2007. Accessed February 1, 2017.
22.
Wason  JMS, Dentamaro  A, Eisen  TG.  The power of phase II end-points for different possible mechanisms of action of an experimental treatment.  Eur J Cancer. 2015;51(8):984-992.PubMedGoogle ScholarCrossref
23.
Tomlinson  G, Detsky  AS.  Composite end points in randomized trials: there is no free lunch.  JAMA. 2010;303(3):267-268.PubMedGoogle ScholarCrossref
24.
Eisenhauer  EA, Therasse  P, Bogaerts  J,  et al.  New Response Evaluation Criteria In Solid Tumours: revised RECIST guideline (version 1.1).  Eur J Cancer. 2009;45(2):228-247.PubMedGoogle ScholarCrossref
25.
Therasse  P, Arbuck  SG, Eisenhauer  EA,  et al.  New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada.  J Natl Cancer Inst. 2000;92(3):205-216.PubMedGoogle ScholarCrossref
26.
Shoulson  I, Young  AB.  Milestones in Huntington disease.  Mov Disord. 2011;26(6):1127-1133.PubMedGoogle ScholarCrossref
×