Precision medicine and patient-centered outcomes: Learning from APOE for prevention clinical trials in older adults

Abstract

For conditions that often entail significant impairment, such as Alzheimer's disease or stroke, there is an established standard in clinical trials to adopt outcome measures that integrate function in both cognitive and physical domains, such as the Clinical Dementia Rating Scale¹ and the modified Rankin Scale.² What is perhaps a newer direction is adopting a similar paradigm, blending the brain and the body, within the context of prevention. Several recent trials including ASPREE (Aspirin in Reducing Events in the Elderly),³ STAREE (Statins in Reducing Events in the Elderly),⁴ and PREVENTABLE (Pragmatic Evaluation of Events and Benefits of Lipid-Lowering in Older Adults)⁵ have adopted a primary composite outcome of survival free of dementia and disability, integrating the incidence of all-cause dementia, persistent disability in activities of daily living, and all-cause mortality. The appeal of this outcome is that it is inherently patient-centric, focused on the foundational goal of maintaining functional independence in aging populations. However, as with anything new, there are still a myriad of questions that need to be explored before adopting survival free of dementia and disability as the default outcome choice for prevention trials in older adults. Does it conform, in theory and in practice, to recommendations for adopting composite outcomes?⁶ What interventions might be most likely to prevent both incident disability and cognitive impairment? What risk factors should one consider in trying to optimize inclusion and exclusion criteria, as well as event rates? What role does precision medicine have with such an endpoint?

In this issue of the Journal of the American Geriatrics Society, Clocchiatti-Tuozzo et al. begin to examine the latter two questions through the lens of genetics, evaluating the association of apolipoprotein E (APOE) genotypes with the incidence of dementia, disability, and all-cause mortality in the Health and Retirement Study.⁷ Among 14,527 older adults (median age 55 years, 58% female, >70% of European ancestry, 25% carriers of at least one APOE ε4 allele), without dementia or disability at baseline followed for a median of 18 years, carriers of the APOE ε4 allele had an increased incidence of the composite outcome of incident dementia, disability, or death (adjusted hazard Ratio (aHR) = 1.15, 95% CI: 1.09–1.21), while being a carrier of the APOE ε2 genotype was associated with a non-statistically significant decrease in risk for that outcome (aHR = 0.94, 95% CI: 0.87–1.01). In looking at the individual components of the composite outcome, carrying the APOE ε4 allele was associated with increased risk of dementia and mortality, but not disability (aHR = 0.98, 95% CI: 0.89–1.08). In secondary analyses that focused on the subgroup of participants that could be eligible for PREVENTABLE depending on prevalent statin use (age ≥75 years without cardiovascular disease at baseline), participants carrying at least one APOE ε4 allele had a 24% higher adjusted risk of incident dementia, disability, or death (aHR = 1.24, 95% CI: 1.10–1.41).

A natural question to ask, given the well-established association between APOE, dementia, and mortality, is why these results warrant discussion? This study adds another layer of evidence that makes a compelling case that APOE may be a critical consideration in interpreting the risks and benefits of lipid-lowering therapy being evaluated in both STAREE and PREVENTABLE. Even though it did not specifically consider prevalent or incident use of lipid-lowering therapy, recent analyses from the Chicago Health and Aging Project show that statin initiation was associated with a lower risk of incident Alzheimer's disease dementia in APOE ε4 carriers, but not in noncarriers.⁸ Another important consideration is the complex relationship between cardiovascular disease, dementia, and mortality in aging populations.⁹ A recent combined analysis of the Whitehall II and Three-City cohorts showed that dementia explains much, though not all, of the elevated mortality risk observed in APOE ε4 homozygotes, whereas in APOE ε4 heterozygotes, the association with mortality seems more related to the incidence of cardiovascular disease.¹⁰ Critically, both STAREE and PREVENTABLE are also following participants for incident major cardiovascular events and so provide the opportunity to evaluate the interplay of APOE genotype, cardiovascular disease, and dementia in contributing to the effectiveness of lipid-lowering for maintaining functional independence.

While this study shines a light on APOE and the role of genetics, there are two lingering questions that require further research. First, within the specific context of lipid-lowering therapy, it remains to be tested whether genetics is the most useful lens for thinking about risk stratification. Biomarkers of subclinical disease, such as coronary calcium based on computed tomography imaging, are likely to have higher utility for clinical decision-making with respect to cardiovascular disease.^{11, 12} Whether this expectation also holds with respect to dementia and disability is an important hypothesis to be tested. Second, for future clinical trials that consider survival free of dementia and disability as an outcome, is it sufficient to primarily focus on APOE, or should one, as an example, consider broader polygenic risk for Alzheimer's disease, cerebral small vessel disease, and/or cardiovascular disease? A key consideration will be whether genetics offers only a lens to quantify baseline risk, which would make it likely a modest contributor to trial efficiency, versus a more integral indicator of treatment effect heterogeneity.

Finally, the result that the APOE ε4 variant associates with dementia and all-cause mortality, but not disability, provides a subtle reminder of the complexity of physiology and how it can potentially undermine even well-designed composite outcomes. On the surface, dementia and physical disability satisfy many of the requisite conditions for being combined as a trial outcome. Both are clinically meaningful endpoints of high importance to patients. There is significant overlap between the physiologic processes associated with both conditions, and so it is a reasonable expectation that interventions, pharmacologic or otherwise, could offer benefits for preventing both dementia and disability.¹³ And yet, past experience with clinical trials in geriatrics has shown that mixed or null results are the norm, with only one example I am aware of where an intervention offered benefit on both cognitive function or dementia AND physical function or disability (Table 1). This experience certainly prompts me to wonder whether survival free of dementia and disability may be an overly ambitious goal, however appealing it may be from a patient perspective. With that said, most randomized trials to date have not indicated contrasting effects, that is, where a therapy reduces incident persistent disability but increases dementia risk. If the situation is commonly benefit for one domain, and no effect on the other, the primary risk is dilution of the treatment effect and insufficient statistical power. However, this scenario could be anticipated as part of designing future trials, and may represent an acceptable trade-off if there is not a clear a priori expectation about whether an intervention may be more likely to prevent dementia versus disability. Another caveat are the results from the Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability trial and the current shift toward testing multi-domain interventions. While these results are contrasted by the results of the Muitidomain Alzheimer Preventive Trial, I expect a summary of randomized trial evidence may start to look different as the field evaluates additional multi-domain interventions, perhaps blending diet and exercise with pharmacological agents such as metformin, sodium-glucose cotransporter-2 inhibitors, glucagon-like peptide-1 agonists, or the monoclonal antibody therapies for Alzheimer's disease.

In conclusion, the work of Clocchiatti-Tuozzo et al. provides additional support for evaluating the role of APOE in trials like STAREE and PREVENTABLE. Beyond that, it largely produces more questions than answers. However, it provides a well-timed opportunity to think toward the future of prevention trials for older adults, and how genetics and precision medicine may play a role in fine-tuning the balance of risk and benefit for therapies aimed at maintaining functional independence.

Dr. Pajewski wrote, reviewed, and prepared this editorial.

Dr. Pajewski is supported by grant numbers U19AG065188, R01HL155396, R01AG071807, and R01AG081287 from the National Institutes of Health related to the PREVENTABLE trial.

Dr. Pajewski is an investigator on the PREVENTABLE trial.

The funders had no role in the preparation or decision to publish this editorial.