Treatments are likely to be most effective in the early stages of Alzheimer’s disease
Alzheimer’s disease begins as early as mid-life and only manifests as Alzheimer’s dementia once the disease is at an advanced stage [8, 9]. Therefore, Alzheimer’s disease is generally far advanced by the time Alzheimer’s dementia is diagnosed, even at an “early” stage of dementia. There is an emerging consensus that optimal disease modification would be best achieved at earlier stages of the disease before dementia develops. Multiple failed phase III trials (most recently solanezumab and verubecestat [10]) seem to support this. Although stage of disease may have mediated trial failure, other possibilities include insensitive outcome measures, poor target engagement, heterogeneity of sample across multiple sites and languages, insufficient phase II data informing phase III confirmation, and the possibility that the specific targeting of the amyloid pathway is inadequate to generate clinical benefit.
Thus, future treatments will be initially offered to people with mid-stage Alzheimer’s disease but without symptoms sufficiently severe to merit a diagnosis of dementia, i.e. prodromal Alzheimer’s disease (Fig. 1). Eventually, treatments might be offered to people at an even earlier stage: people with biomarker evidence of Alzheimer’s disease but minimal or no symptoms, i.e. preclinical Alzheimer’s disease. There are several preclinical trials ongoing and this is the primary focus of the European Prevention of Alzheimer’s Dementia (EPAD) Consortium (http://ep-ad.org/) [11].
Markers of cortical amyloidosis are fundamental
Treatment of Alzheimer’s dementia is currently ‘one size fits all’ with anyone with the clinical syndrome being considered for a cholinesterase inhibitor or memantine [12]. However, since early disease-modifying treatments for Alzheimer’s disease are likely to target amyloid aggregation—although tau-focused therapies are also in development—markers of cortical pathology such as amyloidosis will become central in assessing eligibility for new treatments. This drive towards molecular-based therapeutics will inevitably lead to molecular-based diagnostics. This would move the field towards precision medicine, allowing us to offer targeted treatments to individuals with cortical amyloidosis with the prospect of even greater precision in the future.
Therefore, markers of cortical pathology such as amyloidosis will become central in diagnosing Alzheimer’s disease—particularly at a preclinical stage—and in assessing eligibility for the new treatments. All current trials enrich the clinical sample recruited for the presence of cortical amyloidosis and it is highly likely that these selection criteria will be used to access the treatment in clinical practice.
Cerebral amyloidosis can be measured directly using amyloid PET imaging or indirectly in CSF. There is high concordance between these diagnostic approaches [13], so local availability, cultural factors, and staff experience will determine the choice between the two.
The use of biomarkers for personalising intervention and for gauging response
In a research setting, PET-Amyloid Imaging, CSF Aβ42, tau, and p-tau are useful biomarkers to differentiate people with Alzheimer’s dementia from controls [14]. However, the current utility of these biomarkers in a clinical context, given the low specificity of these tests (e.g. CSF Aβ) for Alzheimer’s dementia in the elderly and the inconsistent quality of the evidence base is less clear [15, 16]. Eligibility criteria for new disease-modifying treatments would initially need to echo the entry criteria for the phase III trials for the treatment, all of which now include a marker of amyloid burden. Further enrichment of the clinical population using APOE status should only be recommended in clinical practice where there was a demonstrable pharmacogenetic effect noted in the treatment’s licensing trial (either for efficacy or safety) unless new evidence emerges for its utility in risk prediction models.
Early disease-modifying treatments will probably slow, rather than reverse, neurodegeneration and so on-going treatment may be required to maintain this effect. Treatment might be lifelong and last several decades. However, there are currently no reliable and validated patient-related outcome instruments for the early stages of the disease. To ensure ongoing clinical and cost-effectiveness, explicit and early stopping criteria would be important, which will require a surrogate biomarker response (Fig. 2). However, no biomarker currently studied is known to be suitable for this purpose and current trials (except DIAN-TU [17]) use intermediary biomarker phenotypes such as PET-Amyloid to determine if a treatment is futile. Furthermore, since clinical trials assess this at a group level and not for individuals, one would only be able to derive a probability of a lack of long-term success for an individual in clinical practice. Therefore, there is currently no reliable way to decide whether an individual should stop or continue other than on the grounds of safety. Embedding this outcome into phase II and III trials would facilitate greater appropriate access to therapies by minimizing inappropriate, ineffective use, but the limited duration of these trials (up to two years) would be a challenge.
Accessing the vast amount of observational data being gathered through programmes like the European Medical Information Framework (EMIF) and Dementias Platform UK (DPUK) will give an indication of the expected rate of changes in sub-populations—though direct comparison of these data with randomised, controlled trial data would be problematic from a statistical perspective. Phase IV testing—as advocated below—would address some of these issues as would the use of readiness cohorts (e.g. EPAD) that can use run-in data from the same population as the trial to help guide expected placebo declines in the trial population.
The importance of phase IV testing
Substantial post-marketing research must be carried out before disease-modifying therapies are widely used and this should be co-produced by the public sector, charities, and industry alongside agreements on pricing and reimbursement that reflect optimum value for the NHS and patients. This would encompass accumulation of safety data as well as real world experience on mid- to long-term patient outcomes (including patient-related outcome measures), compliance, service impact, health economics, and longitudinal biomarker analysis to assist the establishment of a response biomarker. Generic phase IV protocols must be developed before these drugs are marketed so they can be safely and effectively introduced as widely as possible. This will involve close co-operation with bodies such as the National Institute for Health and Care Excellence (NICE), which approves the use of new treatments in the English and Welsh National Health Service and has major influence elsewhere. The establishment of better systems for gathering information from clinical practice would ease this phase IV testing if routinely collected electronic data contained phenotypic, clinical, and resource utilisation variables relevant to brain health in order to undertake large scale, standing phase IV work.