Risks and benefits of unapproved disease-modifying treatments for neurodegenerative disease

Abstract

Objective

To determine whether patients randomized to unapproved, disease-modifying interventions in neurodegenerative disease trials have better outcomes than patients randomized to placebo by performing a systematic review and meta-analysis of risk and benefit experienced by patients in randomized placebo-controlled trials testing investigational treatments for Alzheimer disease, Parkinson disease, Huntington disease, or amyotrophic lateral sclerosis (ALS).

Methods

We searched MEDLINE, Embase, and ClinicalTrials.gov for results of randomized trials testing non–Food and Drug Administration–approved, putatively disease-modifying interventions from January 2005 to May 2018. Trial characteristics were double-extracted. Coprimary endpoints were the treatment advantage over placebo on efficacy (standardized mean difference in outcomes) and safety (risk ratios of serious adverse events and withdrawals due to adverse events), calculated with random effects meta-analyses. The study was registered on PROSPERO (CRD42018103798).

Results

We included 113 trials (n = 39,875 patients). There was no significant efficacy advantage associated with assignment to putatively disease-modifying interventions compared to placebo for Alzheimer disease (standardized mean difference [SMD] −0.03, 95% confidence interval [CI] −0.07 to 0.01), Parkinson disease (SMD −0.09, 95% CI −0.32 to 0.15), ALS (SMD 0.02, 95% CI −0.25 to 0.30), or Huntington disease (0.02, 95% CI −0.27 to 0.31). Patients with Alzheimer disease assigned to active treatment were at higher risk of experiencing serious adverse events (risk ratio [RR] 1.15, 95% CI 1.04–1.27) and withdrawals due to adverse events (RR 1.44, 95% CI 1.21–1.70).

Conclusions

Assignment to active treatment was not beneficial for any of the indications examined and may have been slightly disadvantageous for patients with Alzheimer disease. Our findings suggest that patients with neurodegenerative diseases are not, on the whole, harmed by assignment to placebo when participating in trials.

Neurodegenerative diseases follow an inexorable course and markedly compromise quality of life and longevity. With few validated treatments that meaningfully affect progression, patients with neurodegenerative diseases may view clinical trials as opportunities to access potentially life-extending new treatments. In recent years, patient advocacy groups¹ and libertarian thinktanks² have also pressed for policies that would facilitate access to investigational therapies outside of trials.

Little is known about how such policies would affect patient outcomes; neither is much known about whether accessing unapproved treatments for neurodegenerative disease within trials confers advantages compared with receiving placebo. Because failure rates in neurologic drug development are so high^3,4 and no treatments for neurodegenerative diseases have demonstrated disease modification in large randomized trials,⁵ unapproved treatments likely do not confer advantages beyond symptomatic relief.⁶ Moreover, some putatively disease-modifying treatments have presented safety issues.^7,8 Nevertheless, small but statistically insignificant benefit associated with treatment assignment, when aggregated across trials, might add up to an overall advantage.

To address whether access to putatively disease-modifying treatments confers a clinical benefit to patients with neurodegenerative diseases, we performed a meta-analysis of randomized placebo-controlled trials (RCTs) testing unapproved interventions. We focused on Alzheimer disease (AD), Parkinson disease (PD), amyotrophic lateral sclerosis (ALS), and Huntington disease (HD).^9,10 To date, only 2 disease-modifying drugs have been approved by the US Food and Drug Administration (FDA) for these indications (one by European Medicines Agency), both for ALS and both showing a marginal advantage.^11,12 We hypothesized that patients assigned to placebo could have a small net clinical advantage over patients assigned to treatment due to side effects from unapproved interventions counterbalancing limited efficacy.

Methods

Data sources

Our sample of trials for this systematic review and meta-analysis was generated in 3 sequential steps. First, we identified a sample of drugs and biologics under development for each disease by querying ClinicalTrials.gov for all registered interventional trials of drugs or biologics involving each indication using Drug Trials Visualiser beta version 0.17.¹³ We supplemented our list with drugs described on Alzforum,¹⁴ the Michael J. Fox Foundation website,¹⁵ the ALS Research Forum,¹⁶ and the Huntington's Disease Society of America website.¹⁷ Because dietary supplements, including vitamins and plant extracts, would be accessible outside trial participation, they were excluded from our sample. We defined supplements as any substance available without a prescription, either over the counter or through online order. Drugs and biologics were curated into their generic names, and synonyms were identified through searches of the disease-specific databases listed above.

In the second step, the drugs and biologics identified above were used as keywords in a search of MEDLINE and Embase for published RCTs testing each intervention in our sample in each of the 4 neurodegenerative diseases. The search strategy used for identifying RCTs is described elsewhere.¹⁸ Publication database searches were supplemented with a search of ClinicalTrials.gov for trial results that had not been published elsewhere.

In the third step, trial reports were screened for eligibility using the following inclusion criteria: (1) published between January 1, 2005, and May 23, 2018; (2) original, full-length publications, abstracts, or results postings on ClinicalTrials.gov; (3) English language; (4) randomized and placebo-controlled trial; (5) single or double blinded; (6) enrollment of patients with a diagnosis of AD, PD, ALS, or HD; and (7) reported ADAS-Cog, UPDRS, ALSFRS, or UHDRS score as measures of efficacy. Trials were further screened on the basis of whether they tested unapproved agents for disease-modifying activity. Interventions were deemed unapproved if they had not been approved by the FDA for any indication before the study enrollment start date and thus would not be available to patients through off-label prescription. Interventions were deemed to be disease modifying if they met the following criteria: the aim of the trial was to treat the whole disease or symptoms that are prevalent and correlated with disease progression, and the intervention was not being tested as an adjunctive or add-on therapy. We defined add-on or adjunctive therapy as a therapy aimed at either enhancing or optimizing the effect of an existing therapy.¹⁹ If a drug that had been tested as an adjunctive therapy was also tested in a separate monotherapy trial, this trial was still eligible for inclusion. Our screening process is illustrated in figure S1 (available from Figshare, https://doi.org/10.6084/m9.figshare.10052426.v2).

Data extraction

Trials were manually double-extracted with Numbat²⁰ to capture trial phase, study duration, patient enrollment, sponsor, and intervention. We extracted information on method of randomization, blinding, and patient withdrawals to calculate Jadad scores for risk of bias assessment.²¹ We also extracted the methods of missing data imputation in the trials to supplement this assessment.

To assess benefit associated with treatment assignment, change from baseline to endpoint was extracted for treatment and placebo arms with the Alzheimer's Disease Assessment Scale–Cognitive Subscale (ADAS-Cog),²² Unified Parkinson's Disease Rating Scale (UPDRS),²³ ALS Functional Rating Scale (ALSFRS),²⁴ or the Unified Huntington's Disease Rating Scale (UHDRS)–Total Motor Score²⁵ (the Total Motor Score subscale was chosen because several studies did not report the aggregated score or Total Functional Capacity subscale). We extracted the change in score from baseline to either the endpoint prespecified in the publication or the latest time point available if the final endpoint was not prespecified. For crossover, delayed start, and open-label extension trials, data were extracted from the last time point before switching or unblinding. For PD, because the UPDRS is composed of 4 subscales for which the total was not always available, we extracted in order of preference: the total UPDRS; the total of the Motor and Activities of Daily Living subscales (UPDRS II/III), or the Motor Subscale (UPDRS III). Furthermore, because UPDRS can be measured in the “on” or “off” state for patients experiencing motor fluctuations, we included only trials testing either previously untreated patients or treated patients experiencing motor fluctuations with scores reported in the “off” state. Trials of treated patients not yet experiencing motor fluctuations were excluded from the efficacy analysis.

To assess risk associated with treatment assignment, the reported proportions of patients experiencing serious adverse events (SAEs) and withdrawals due to adverse events (WAEs) were extracted for each arm. SAEs were defined as any adverse drug experience that is life-threatening or results in hospitalization, disability, or death.²⁶ SAEs of any cause were included because of the complexity of attributing causality in safety reporting.²⁷ If a trial reported a subset or multiple categories of SAEs and the total number of patients experiencing any SAE was not explicitly stated, we extracted the highest number without double-counting patients. WAEs were extracted as an additional measure of treatment risk. Total withdrawals used to calculate attrition rates included patients who withdrew either before or after receiving the study drug at any time before the prespecified final endpoint.

Missing data were sought from ClinicalTrials.gov or sponsor documents such as presentations or press releases. For publications in which efficacy data were presented graphically, we used graphical analysis software (ImageJ).²⁸ Investigators were contacted between August 27 and 29, 2018, if efficacy outcome data were incomplete.

Data synthesis

The primary analyses of safety and efficacy associated with treatment assignment were based on the subset of trials in our sample with >24 weeks of follow-up (long-duration trials). As a secondary analysis, we performed identical analyses in trials involving ≤24 weeks of follow-up (short-duration trials). Patient benefit was analyzed with a random-effects meta-analysis comparing the standardized mean difference (SMD)²⁹ in change from baseline between the treatment and placebo arms within each indication. All measures of variance were converted to SDs using the methods outlined in the Cochrane Handbook,³⁰ and multiple treatment arms in the same trial were pooled using inverse variance weighting.³¹ Patient risk was analyzed with random-effects meta-analyses to calculate risk ratios (RRs) of SAEs and WAEs between the treatment and placebo arms. To facilitate interpretation, we also computed and present the risk difference and number needed to be treated for 1 additional patient to be harmed (NNTH).³² A correction factor of 0.5 was added to SAE and WAE values for trials that reported zero event rates in both arms to allow these trials to contribute to the overall pooled effect size and confidence intervals (CIs).³³ Trials that did not report either SAEs or WAEs were not included in the respective analysis. Heterogeneity tests were performed with Higgins I².³⁴

We performed subgroup analyses to probe whether safety and efficacy associated with assignment to experimental treatment differed between phase 2 and phase 3 trials. The p values for subgroup comparisons represent the between-subgroup heterogeneity statistic Q, based on a random-effects model. We also performed meta-regressions of efficacy, SAEs, and WAEs with the covariate of trial duration. To ensure that our definition of disease modification did not significantly affect our results, we reanalyzed our primary outcomes (table 1) comparing 2 modified definitions to our primary definition as sensitivity analyses. In the intent-based analysis, we included trials within our primary sample that expressed intent to develop a disease-modifying therapy. In the mechanism-based analysis, we included trials testing unapproved interventions that are not members of a drug class containing a previously approved symptomatic therapy (e.g., dopamine agonists for PD). In addition, we compared treatment advantage in subgroups stratified by the method used to combine dose arms (pooled vs high dose), methods of imputation used in the trials (mixed model repeated measures vs last observation carried forward vs observed cases), and attrition rates (<15% vs ≥15%).

Table 1.

Summary of primary outcome measures

We defined p ≤ 0.05 as our threshold of statistical significance. Because of the exploratory nature of our analyses, we did not adjust for multiple outcomes and analyses.³⁵ All meta-analyses and statistical tests were performed with R version 3.4.2 with the meta package.³⁶ Analyses were prespecified before the outset of extraction, and the study was prospectively registered on PROSPERO (CRD42018103798).³⁷

Ethics

Our study does not involve human participants and thus was not submitted for ethics review.

Data availability

Supplementary figures, tables, and references are available from FigShare (figures S1–S6, tables S1–S6, and e-references available at https://doi.org/10.6084/m9.figshare.10052426.v2). Raw data will be made available to investigators on request to the corresponding author.

Results

Trial characteristics

We identified 113 trials that met eligibility (figure 1): 69 AD trials; 20 PD trials; 16 ALS trials; and 8 HD trials. Of these, 52 were long-duration trials included in our primary efficacy analysis; an additional 10 trials reported safety but not efficacy endpoints and were included only in our primary safety analyses. Studies included in our primary safety analysis enrolled 31,029 patients, with 18,565 assigned to experimental treatment, and tested interventions consisting of 45% small-molecule drugs, 34% large-molecule drugs, and 21% biologics. With the exception of 1 trial, all trials scored ≥3 for quality with the Jadad scale (figure S2 available from Figshare, https://doi.org/10.6084/m9.figshare.10052426.v2). Overall attrition exceeded 15% in 54% of trials (figure S3 available from Figshare). Characteristics of trials in our primary sample are shown in table 2; summarized data are available in table S1 available from Figshare.

Figure 1. PRISMA diagrams.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagrams for (A) Alzheimer disease, (B) Parkinson disease, (C) amyotrophic lateral sclerosis (ALS), and (D) Huntington disease. Diagrams include trial records in both the primary (long-duration) and secondary (short-duration) samples. ADAS-Cog = Alzheimer's Disease Assessment Scale–Cognitive Subscale; ALSFRS = ALS Functional Rating Scale; FDA = Food and Drug Administration; RCT = randomized placebo-controlled trial; SAE = serious adverse events; UHDRS = Unified Huntington's Disease Rating Scale; UPDRS = Unified Parkinson's Disease Rating Scale; WAE = withdrawal due to adverse events.

Table 2.

Characteristics of trials included in the primary safety sample

Benefit associated with treatment assignment

Assignment to disease-modifying experimental interventions did not demonstrate statistically significant efficacy compared to placebo assignment for long-duration trials of AD (SMD −0.03, 95% CI −0.07 to 0.01), PD (SMD −0.09, 95% CI −0.32 to 0.15), ALS (SMD 0.02, 95% CI −0.25 to 0.30), or HD (SMD 0.02, 95% CI −0.27 to 0.31) (figure 2).

Figure 2. Benefit associated with treatment assignment.

Standardized mean difference between treatment and control groups on disease-specific efficacy scales (Alzheimer disease: Alzheimer's Disease Assessment Scale–Cognitive Subscale; Parkinson disease: Unified Parkinson's Disease Rating Scale; amyotrophic lateral sclerosis [ALS]: ALS Functional Rating Scale; Huntington disease: Unified Huntington's Disease Rating Scale–Total Motor Score) in long-duration trials (>24 weeks). Note that, for clarity, all the scales are represented such that a positive mean change represents a worsening, regardless of the original directionality of the scale. Letters following the e-reference number are used to differentiate between either multiple studies within the same publication or multiple publications from the same year. CI = confidence interval; SMD = standardized mean difference. e-References are available on FigShare, https://doi.org/10.6084/m9.figshare.10052426.v2.

Risk associated with treatment assignment

Patients assigned to disease-modifying treatment were significantly more likely to experience SAEs in long-duration AD trials (RR 1.15, 95% CI 1.04–1.27). Patients assigned to treatment had a nonstatistically significant increased risk of SAEs in PD (RR 1.32, 95% CI 0.70–2.48), ALS (RR 1.22, 95% CI 0.82–1.80), and HD (RR 1.40, 95% CI 0.50–3.89) (figure 3).The risk of WAEs was significantly higher in treatment arms for long-duration AD trials (RR 1.44, 95% CI 1.21–1.70). The RR of WAEs was not statistically significant for patients assigned to treatment in trials for PD (RR 1.35, 95% CI 0.86–2.11), ALS (RR 0.88, 95% CI 0.57–1.35), or HD (RR 1.18, 95% CI 0.41–3.37) (figure 4).

Figure 3. Risk of SAEs associated with treatment assignment.

Risk ratio (RR) of serious adverse events (SAEs) in long-duration trials (>24 weeks) of Alzheimer disease, Parkinson disease, amyotrophic lateral sclerosis (ALS), and Huntington disease (n represents the number of SAEs; N represents the number of patients randomized). A correction factor of 0.5 was added for trials with no SAEs in either the treatment or placebo arm. Letters following the e-reference number are used to differentiate between either multiple studies within the same publication or multiple publications from the same year. CI = confidence interval. e-References are available on FigShare, https://doi.org/10.6084/m9.figshare.10052426.v2.

Figure 4. Risk of WAEs associated with treatment assignment.

Risk ratio (RR) of withdrawals due to adverse events (WAEs) in long-duration trials (>24 weeks) (n represents the number of WAEs; N represents the number of patients randomized). A correction factor of 0.5 was added for trials with no WAEs in either the treatment or placebo arm. Letters following the e-reference number are used to differentiate between either multiple studies within the same publication or multiple publications from the same year. ALS = amyotrophic lateral sclerosis; CI = confidence interval. e-References are available on FigShare, https://doi.org/10.6084/m9.figshare.10052426.v2.

Risk and benefit in short-duration trials

The secondary analysis of short-duration trials showed a significant efficacy advantage for patients assigned to disease-modifying experimental interventions in PD (SMD −0.53, 95% CI −0.97 to −0.08) and HD (SMD −0.31, 95% CI −0.59 to −0.03) trials but not in AD (SMD −0.05, 95% CI −0.11 to 0.00) or ALS (SMD −0.11, 95% CI −0.35 to 0.12) trials (figure S4 available from Figshare, https://doi.org/10.6084/m9.figshare.10052426.v2).

Assignment to treatment was not significantly associated with greater risk of SAEs in AD (RR 0.99, 95% CI 0.71–1.38), PD (RR 0.52, 95% CI 0.23–1.19), ALS (RR 1.15, 95% CI 0.78–1.69), or HD (RR 0.76, 95% CI 0.20–2.88) (figure S5 available from Figshare, https://doi.org/10.6084/m9.figshare.10052426.v2). The risk of WAEs for patients assigned to treatment was significantly higher in short-duration AD trials (RR 1.53, 95% CI 1.09–2.14) but was not significantly higher than the risk for patients receiving placebo in PD (RR 1.48, 95% CI 0.61–3.55), ALS (RR 1.22, 95% CI 0.60–2.49), or HD (RR 0.91, 95% CI 0.37–2.24) trials (figure S6 available from Figshare).

Effect of trial duration

Meta-regression of treatment efficacy advantage in all 4 indications with the covariate of trial duration revealed a significant interaction with a stronger treatment advantage in shorter trials compared to longer trials (p = 0.01, r² = 9.27%). For each indication individually, no significant correlations were found, but all 4 indications exhibited the same trend: treatment advantage decreased as trial duration increased (figure 5). Meta-regression showed no significant correlations between trial duration and the comparative risk of SAEs (p = 0.87) or WAEs (p = 0.80).

Figure 5. Correlation between efficacy and trial duration.

Meta-regression of efficacy with covariate of trial duration. Effect size represents the standardized mean difference (SMD) between the treatment and placebo arms; a positive SMD indicates a trial that favored the placebo arm. The p values correspond to the slope of the regression line. The r² values were 4.96%, 0.27%, 0.14%, and 19.40% for Alzheimer disease, Parkinson disease, amyotrophic lateral sclerosis (ALS), and Huntington disease, respectively.

Effect of trial phase

Subgroup analysis of phase 2 and phase 3 trials showed no significant differences in efficacy treatment advantage between phases in AD (p = 0.29), PD (p = 0.43), ALS (p = 0.57), or HD (p = 0.24). Significant differences in safety were found between phase 2 and 3 PD trials with a trend toward larger treatment disadvantages in phase 3 on both measures of SAEs (p < 0.01) and WAEs (p = 0.01). In ALS, assignment to treatment in phase 3 trials was associated with significantly lower risk of WAEs (p = 0.01). No significant advantages on safety measures were found between phase 2 and 3 trials in the other indications with respect to SAEs or WAEs (table S2 available from Figshare, https://doi.org/10.6084/m9.figshare.10052426.v2).

Sensitivity analyses

Samples derived from modified definitions of disease-modifying treatments showed no significant differences from the primary results in the intent-based or mechanism-based analyses (table S3 available from Figshare, https://doi.org/10.6084/m9.figshare.10052426.v2). Comparison of pooled and high-dose arms showed no significant differences from the primary analysis on efficacy, SAEs, or WAEs in any indication (table S4 available from Figshare). Comparing method of imputation (last observation carried forward vs mixed model repeated measures vs observed cases) in the AD sample revealed no significant differences; this analysis was not performed for the other 3 indications because of the limited number of trials available for analysis (table S5 available from Figshare). Stratification by attrition rate (<15% vs ≥ 15%) revealed a significant difference in SAE risk for AD, with high-attrition trials presenting a significantly higher risk for patients assigned to treatment. No other significant differences in treatment advantage were found between high- and low-attrition trials (table S6 available from Figshare).

Discussion

Our findings suggest that, on the whole, patients assigned to investigational treatment are no better off than patients assigned to placebo in RCTs testing unapproved, disease-modifying interventions. Across all 4 indications, patients assigned to investigational treatment did not experience better efficacy outcomes; no SMDs were statistically significant, and the largest treatment advantage of −0.09 in PD did not meet the threshold for a small effect size by Cohen criteria,³⁸ thus bringing into question any potential efficacy. Patients assigned to investigational treatment were more likely to experience SAEs in all 4 indications; however, statistical significance was seen only in AD, and even so, the safety advantages of placebo assignment were slight. The SAE and WAE risk differences correspond to NNTHs of 33 patients for SAEs (95% CI 20–100) and 33 patients for WAEs (95% CI 20–50). In comparison, a meta-analysis of RCTs testing approved cholinesterase inhibitors for AD found NNTHs of 12 for all adverse events and 16 for WAEs.³⁹

In short-duration trials, assignment to investigational treatment proved advantageous in PD and HD, with AD and ALS also trending toward a treatment advantage. The discrepancy in the observed benefit between short- and long-duration trials might have 2 explanations. First, it could reflect a regression to the mean from the selection of only those compounds that show large effects in the short term for longer-term testing. Second, it could reflect that compounds have symptomatic effects in the short term. Patient expectations may be heightened by large effect sizes in earlier-phase trials. Our data on the long-term effects caution against such heightened expectations.

What do these results mean for patients pursuing access to treatments through trial participation or expanded access and the clinicians guiding them through these processes? Patients with neurodegenerative diseases may be willing to endure high levels of risk to access unproven, putatively disease-modifying treatments through trials or expanded access.⁴⁰ Furthermore, clinicians may prescribe treatments with unfavorable risk/benefit profiles because of patient demand or a desire to help distressed patients, especially when the alternative is palliative care. Our findings do not rule out that some individual patients derive benefits from accessing investigational treatments, nor do they exclude the possibility that, at some point in the future, some patients will benefit by accessing an unapproved treatment. They do, however, provide evidentiary grounds for clinicians to temper patient expectations in informed consent discussions. If disease-modifying treatments had large benefits for a small number of patients or small benefits for a larger number of patients, one might expect the substantially greater statistical power afforded by this meta-analysis to detect an advantage for treatment assignment. Instead, we observed none.

Numerous recent initiatives have sought to lower barriers to accessing investigational treatments for patients. These include the US FDA's Expanded Access Program, which enables patients with serious or life-threatening conditions who are ineligible for trial participation to access unapproved therapies for which the potential benefit justifies the risk.⁴¹ Over the last 5 years, >9,000 applications have been approved.⁴² National right-to-try legislation, which attempts to bypass FDA oversight altogether, requires only that a drug have completed a phase 1 trial.⁴³ Our findings of a lack of advantage associated with treatment assignment for patients in phase 3 compared to phase 2 trials suggest the fallibility of inferring treatment benefit from early-phase trial evidence. Recent high-profile failures such as a string of negative late-phase trials of antiamyloid agents in AD reinforce the need for cautious interpretation of early-phase results and for balanced reporting in publications, academia and industry press releases, and the media.⁴⁴ Furthermore, our findings support the obvious point that regulatory approval standards have, on balance, prevented patients with debilitating illnesses from being further burdened by the side effects of ineffective treatments. The value of early access policies on improving outcomes for patients is uncertain.

Our study has limitations. First, we included only trials with published results. Many trials testing novel neurology drugs are never published.⁴⁵ An example is the drug latrepirdine: our sample included the positive phase 2 AD trial of this drug,⁴⁶ but the nonpositive phase 3 data were not available.⁴⁷ Publication biases would most likely lead to an overestimate of benefit and underestimate of risk associated with treatment assignment. A second limitation is the heterogeneity of trials. However, our study did not set out to estimate risk and benefit for a sample of patients exposed to the same treatment; our aggregate estimates should therefore be understood as providing a general description of the risk and benefit to patients in clinical trials for neurodegenerative disease. Finally, the sample of PD, ALS, and HD trials captured in our search was small and hence underpowered to detect modest or small advantages or disadvantages associated with treatment assignment. This limits our ability to draw firm conclusions for these conditions.

Our analysis may be useful to clinicians looking to provide reassurance to patients who fear missing out on therapeutic benefit through randomization to placebo, trial ineligibility, or lack of expanded access programs. Surveys show that patients are apprehensive of placebos in trials and perceive assignment to comparator arms as depriving them of clinically advantageous treatments.^48–51 Indeed, trials often attempt to overcome aversion to placebo arms by using 2:1 randomization ratios.⁵² Our findings indicate that this practice is unnecessary and potentially disadvantageous to patients. Those tasked with designing trials or advising patients in their choice of treatment should consider the clinical impact that experimental treatments for neurodegenerative disease have historically had on patients. Our meta-analysis suggests that, over the last 15 years of testing, neither patients assigned to placebo arms nor patients deprived of investigational treatments due to drug regulations have suffered medically by lack of access; if anything, they may have been slightly better off.

Glossary

AD

Alzheimer disease

ADAS-Cog

Alzheimer's Disease Assessment Scale–Cognitive Subscale

ALS

amyotrophic lateral sclerosis

ALSFRS

ALS Functional Rating Scale

CI

confidence interval

FDA

Food and Drug Administration

HD

Huntington disease

NNTH

number needed to be treated for 1 additional patient to be harmed

PD

Parkinson disease

RCT

randomized placebo-controlled trial

RR

risk ratio

SAE

serious adverse events

SMD

standardized mean difference

UHDRS

Unified Huntington's Disease Rating Scale

UPDRS

Unified Parkinson's Disease Rating Scale

WAE

withdrawal due to adverse events

Appendix. Authors

Study funding

Funded by the Canadian Institutes of Health Research (PJT 148726).

Disclosure

A.C. Feustel, A. MacPherson, D.A. Fergusson, and K. Kieburtz report no disclosures relevant to the manuscript. J. Kimmelman serves on a Data Safety Monitoring Board in a remunerative capacity for Ultragenyx Inc. Go to Neurology.org/N for full disclosures.