Abstract

1. Introduction

Biomarkers of Alzheimer’s disease (AD) are present years before a person has overt cognitive impairment (Bateman, et al., 2011, Lim, et al., 2014, Morris, et al., 2009, Pietrzak, et al., 2014, Price, et al., 2009), supporting the hypothesis that interventions initiated at these early “preclinical” stages, when neurodegeneration is minimal, may have the greatest likelihood of altering the natural history of AD (Sperling, et al., 2011b). To facilitate testing this hypothesis, a working group sponsored by the National Institute on Aging and the Alzheimer’s Association proposed research diagnostic criteria for a preclinical stage of AD (Sperling, et al., 2011a). In this stage, cognition remains normal or only subtly impaired, but biomarker evidence of AD is present.

Preclinical AD trials must implement one of two designs: blinded or transparent enrollment (Kim, et al., 2015). Blinded designs do not disclose biomarker results to participants. They enroll a proportion of participants who do not demonstrate AD biomarkers so that enrollment is not a de facto disclosure of biomarker status. These participants are non-randomly assigned to placebo, undergo all study procedures, and are followed for the duration of the study. With transparent enrollment, only those who demonstrate biomarker criteria are enrolled and randomized. Biomarker results are disclosed when an investigator informs a person whether he or she is eligible. Both designs have unique risks. For example, blinded enrollment trials may inadvertently disclose biomarker status to participants that do not wish to learn it (Hooper, et al., 2013, Kim, et al., 2015), while transparent enrollment trials bring unique challenges related to confidentiality and the social and psychological impact of learning biomarker results (Arias and Karlawish, 2014).

Which of these designs should researchers use? The answer to this question engages several considerations, including the use of limited resources, the need for timely progress, study feasibility, and the ethical implications of trial designs. The purpose of this study is to empirically inform one specific consideration: the impact on participant recruitment. It is unknown how these two designs will impact recruitment timelines. Also unknown are the factors that might explain why one design is more appealing than the other. Participants’ views cannot entirely settle the competing ethical, clinical and resource considerations, but they do provide an important perspective on how, on a person-by-person basis, they settle these issues. Absent empirical data, trialists, institutional review boards and funders can only speculate over how blinded versus transparent designs impact enrollment, or simply implement these designs and learn from the efforts.

Transparent enrollment preclinical AD trials require people to learn risk information for a disease for which no treatment exists, or may ever exist. But compared to blinded enrollment trials, these trials require fewer participants and closely approximate how clinicians will diagnose and treat sporadic preclinical AD (Burns and Klunk, 2012). If transparent enrollment trials suffer from slow recruitment, then, despite smaller overall sample sizes, these trials may be less efficient than those using blinded designs. Here, we test the hypothesis that a preclinical AD trial with transparent enrollment will have poorer recruitment than one with blinded enrollment. A secondary aim was to identify clinical and trial factors associated with willingness to enroll. We chose to study persons with interest in AD research because they approximate the kinds of persons who would be recruited for a preclinical AD trial.

2. Methods

3. Results

3.3. Importance of trial factors

In both randomized groups and when combining data across groups, Friedman tests showed that participant ratings of importance differed among the seven trial factors (p<0.0001). In the transparent design, the ratings of study risks and the requirement of a study partner were significantly higher than every other factor (Wilcoxon signed rank test, p≤0.02). In the blinded design, study risks were rated higher than every other factor (p≤0.01); the requirement of study partner was rated higher than visit frequency and study length; and required procedures were rated higher than visit frequency, study length and visit location. When data were combined across groups, participants rated the study risks and the requirement of a study partner as more important than the remaining five factors (p<0.05 for all pairwise comparisons; Figure 1). Among the remaining factors, study procedures was deemed more important than study length and visit frequency (p<0.05).

Open in a separate window

Figure 1

Participant ratings of the importance of trial factors in the decision whether to enroll. The proportions participants providing each likert response are illustrated for each factor. *p<0.05 vs stated study risks; ^#p<0.05 vs requirement of a study partner; ^@p<0.05 vs study procedures.

3.4. Incentives for participation

Within each group and in combined analyses, ratings differed among the six potential incentives (Friedman tests, p<0.0001). In both the transparent and blinded designs, receiving cognitive and genetic testing results were rated higher than blood test results and financial compensation (Wilcoxon signed rank test, p<0.05 for all comparisons). In the blinded design, receiving cognitive testing results was also rated higher than receiving overall study results (p=0.02). In both groups, receiving personal AD risk estimates was rated higher than blood test results or financial compensation (p<0.05 for all comparisons) and financial incentive was rated lower than receiving overall study results. In the blinded design, financial incentive also differed from blood test results (p=0.03).

When data were combined across groups, receiving cognitive testing results was rated higher than receiving overall study results, blood test results, and financial compensation (p<0.05 for all comparisons; Figure 2). Participants rated genetic testing results as greater incentive than blood test results and financial compensation (p<0.0001). Financial compensation was rated significantly lower than every other incentive assessed.

Open in a separate window

Figure 2

Participant ratings of potential incentives for participation. The proportions of participants providing each likert response are illustrated for each potential incentive. *p<0.05 vs cognitive testing results; ^#p<0.05 vs genetic test results; ^@p<0.05 vs personal AD risk estimates; ^Ωp<0.05 vs overall study results; ^Ψp<0.05 vs personal blood test results.

4. Discussion

This study found no difference in the willingness to participate in AD prevention trials that use blinded versus transparent enrollment. Our sample represents the kind of persons researchers will attempt to recruit to preclinical AD trials, as it was composed primarily of community members who demonstrated interest in AD research. Many had attended public lectures on AD and, in our study, they demonstrated good knowledge of AD and favorable attitudes toward research.

Though some participants did not wish to learn their amyloid status, the frequency of these individuals (<5%) was not sufficient to lower the overall recruitment rate for the transparent enrollment trial. In fact, willingness was numerically higher in the transparent enrollment design. These results suggest that transparent enrollment trials, in addition to offering reduced sample sizes and cost, may not suffer from slower recruitment than blinded enrollment trials. They may even recruit faster.

Other studies examining the desire of healthy individuals to learn their AD risk have had similar findings and may be pertinent to understanding the feasibility of transparent preclinical AD trials. In a telephone survey of Western European and American adults, 67% of respondents reported that they were likely to have a presymptomatic diagnostic test for AD, were one to become available. Desire to undergo testing was highest among those concerned about getting AD (Wikler, et al., 2013). Biomarkers do not offer the certainty of a diagnostic test, however. Neumann and colleagues found similar (70%) rates for desire to undergo predictive genetic testing for AD, but rates fell to 45% for a partially predictive test (Neumann, et al., 2001). Alternatively, these authors found that >70% of respondents in another study were interested in even an imperfect predictive blood test for AD (Neumann, et al., 2012) and a survey of Health and Retirement Study participants found that 60% were interested in learning their AD risk (Roberts, et al., 2014). Collectively, these results suggest that a sizeable proportion of the population may be interested in learning AD risk information and these individuals, especially those concerned about AD, may be most likely to participate in preclinical AD trials and facilitate transparent enrollment.

Our results add to the understanding of barriers to preclinical AD trial recruitment (Grill, et al., 2013). The most important barriers to enrollment were the risk of the study therapy and the requirement of identifying and securing a study partner, a factor deemed more important than the length of the study, number of visits, length of visits, or the procedures involved. Given that preclinical AD trials will recruit cognitively normal, functionally independent community-dwelling individuals, investigators may need to consider reducing study partner requirements by implementing or developing technologies to secure information typically provided through in-person informant-based scales (Mundt, et al., 2007, Sano, et al., 2013).

Measuring research attitudes may identify those likely to enroll and returning research results may incentivize participants. Participants valued receiving cognitive and genetic testing results, more so than blood test or overall study results. A majority of participants responded that financial compensation would not impact their decision whether to enroll (Figure 2), suggesting that the cost of financial incentives for participation may be better spent on teleconferences or in-person events to share study data (Chakradhar, 2015).

To be shared, genetic testing must be performed by a CLIA-certified laboratory and disclosure should be performed in partnership with a genetic counselor (2014). When these requirements are fulfilled, disclosure of genetic information may be safe (Green, et al., 2009) but also has the potential to bias study outcomes (Lineweaver, et al., 2013). Similarly, participants’ perceptions of their expected performance on cognitive tests, which might be impacted by the return of test results, could bias subsequent outcomes (Hess, et al., 2003, Steele and Aronson, 1995). Full consideration of these issues will require further study, but may provide important data to instruct trial design and conduct.

Our study has limitations. The extent to which hypothetical responses predict actual behaviors is not clear and the observed rates of willingness to participate (>60%) may exceed those of a real trial. This is in part due to the hypothetical trial description, which was designed to produce a 50% willingness rate to power examination of differences between the groups. Initial preclinical AD trials are testing infused amyloid-lowering therapies, which carry risks such as amyloid-related imaging abnormalities (Sperling, et al., 2012). The ecological validity of our results may be limited since our hypothetical trial investigated an oral therapy that lacked these risks. Since many trials face protracted recruitment, frequently requiring additional interventions to ultimately achieve study samples, our results may be most instructive to initial enrollment of highly motivated participants in preclinical AD trials. Finally, our data leave several questions unanswered. Transparent enrollment will require more elaborate disclosure processes than was implemented here (Harkins, et al., 2015), and these processes may impact willingness to participate. Furthermore, both blinded and transparent designs are associated with unique risks (see Introduction) and this study did not explore how those risks will impact willingness to participate.

In conclusion, these results suggest that preclinical AD trials implementing transparent enrollment will not suffer slower recruitment because of the requirement of biomarker disclosure. It is possible that some participants will view learning their amyloid PET status as an incentive to enrollment, though this will require further study. This study also suggests that other trial factors, in particular the risks of the intervention and the requirement of a study partner, may have greater impact than biomarker disclosure on recruitment and to overcome these barriers investigators should consider incentives such as providing research results. Although transparent designs are neither coercive nor do they offer undue influence (Kim, et al., 2015), these results may have important implications to the ethics of preclinical AD trials, given that they support the feasibility of this design. Additional research, including assessment of the ongoing preclinical AD trials, will be critical to instruct future trial designs and methods, especially as it relates to biomarker disclosure.

Acknowledgments

Footnotes

References