Advancing Alzheimer's Treatment Means Using The Best Biomarker(s) — But Which?
By Frederick Huie, MD, SYNAPS Dx advisory council member, and Emmanuel C. Gorospe, MD, MPH, CPI, FACP, chief medical officer, Great Lakes Research Network
A significant challenge in Alzheimer's disease (AD) clinical trials stems from the extended preclinical and prodromal phases lasting two to three decades before clinical symptoms emerge. While lifestyle changes and comorbidity management hold promise in altering the disease trajectory, urgent emphasis is needed.
This approach, endorsed by multiple studies and the Alzheimer's Association, doesn't diminish the crucial need for effective therapeutics. It does, however, raise a fundamental dilemma: addressing the chronic nature of AD within the confines of acute clinical trials. The question remains: How can a disease-modifying therapeutic (DMT) be developed for a condition characterized by prolonged preclinical phases and acute trial structures that must demonstrate valid clinical efficacy?
The disease is named after Dr. Alois Alzheimer, who first characterized it neuropathologically and identified what we know as amyloid plaques and tau neurofibrillary tangles in the post-mortem brain of his patient Frau Deter. Utilizing the latest advances in histology in the early 1900s, Dr. Alzheimer and colleagues used a new staining technique called the Bielschowsky technique to readily identify plaques and tangles under a light microscope. Research in this area hit a lull for several decades, with presenile dementia, as it was referred to, being largely considered a part of aging during that time.
In the mid-1980s and early 1990s, several discoveries were made, including the composition of plaques known as amyloid-beta (Aβ), a peptide cleaved from the amyloid precursor protein (APP). This cleavage occurred through a series of enzymes that were either non-amyloidogenic (alpha- and gamma-secretase) or amyloidogenic (gamma- and beta-secretase). Other key findings included the composition of intracellular neurofibrillary tangles, the isolation of APP to chromosome 21, which is tripled in Down’s syndrome, nearly all cases of which lead to AD, and the identification of three dominantly inherited mutations that increase the production of Aβ.
These discoveries not only provided therapeutic targets for AD but also offered potential biomarkers. However, the path forward becomes murky. Despite numerous clinical trials attempting to eliminate Aβ from the brain through vaccination, beta-secretase inhibitors, or, more commonly, monoclonal antibody (mAb) therapy, the outcomes have largely fallen short of primary and secondary measures. This trend is evident even in the Colombian cohort with autosomal dominant AD (ADAD), characterized by genetic mutations leading to 100% penetrance and disease onset.
Challenges In AD Clinical Trials
While many mAb therapeutics have effectively removed amyloid from the brain (note: Aβ is an FDA-approved surrogate endpoint), as indicated by PET and other biofluid measures, only LEQEMBI recently demonstrated a 27% average slowing of cognitive decline in an 18-month clinical trial. Notably, this effect varied between genders, with a 12% slowing of decline in women versus a more significant 43% in men.
The trial reported a -0.45 difference in Clinical Dementia Rating Sum of Boxes (CDR-SB) between the placebo and drug arms (negative values indicating a slowing of decline). The impact of a <0.5 CDR-SB point on the trial readout remains unclear, especially considering that 26.4% of patients experienced infusion-related reactions, and 12.6% exhibited amyloid-related imaging abnormalities (ARIA).
In the wake of 40+ failed AD clinical trials and the modest impact of LEQEMBI, a critical question arises: what's amiss? Historically, AD trial recruitment, based on clinical symptoms, lacked specificity, with a diagnostic accuracy of 50% at best. Incorporating costly amyloid PET scans for recruitment improved specificity but fell short, as amyloid, though necessary, proved insufficient for AD. Accumulation, linked to factors like sleep deprivation, trauma, aging, and infections, characterizes amyloid accumulation in late-onset AD (LOAD).
Klunk et al.'s amyloid PET demonstration revealed cerebral amyloid burden in AD and control groups. Subsequent studies found 20%-40% of elderly individuals amyloid-positive without apparent dementia signs. In individuals 80 or older, 60% presented AD neuropathology at autopsy without cognitive impairment. A study, supported by the Alzheimer's Association, highlighted a key limitation: "amyloid PET's high prevalence of positivity in normal older individuals."
Diagnosing AD
At the Alzheimer’s Association International Conference 2023, Clifford Jack shared a case challenging conventional AD diagnosis. A dementia patient with A+T-N+ biomarkers, per National Institute on Aging and Alzheimer’s Association (NIA-AA) guidelines, seemed to have AD. Yet, detailed analysis by Dr. Jack revealed advanced TDP-43 encephalopathy as the cause. This underscores that while one may have cerebral amyloid and cognitive impairment, the latter's cause may not be AD. Amyloid, though necessary, isn't sufficient for AD, making cerebrospinal fluid (CSF) and blood-based Aβ and tau measures mere surrogates for the neuropathology.
Based upon the available evidence, cerebral amyloid is a risk factor for AD, not a determinant. While the presence of amyloid increases the likelihood of developing AD, it is essential to note that most individuals with brain amyloid will not develop AD. Jack et al. acknowledged this in the 2018 NIA-AA guidelines stating “because of increasing death rates with age and the long preclinical period of AD, it cannot be known if that person would have developed T, (N), or (C) had they lived longer.”
Against this backdrop, the challenges associated with developing a DMT for AD come into sharp focus. Clinical trials often enroll older adults with cerebral amyloid, but a significant portion of them may not progress to AD. While amyloid serves a diagnostic purpose, it lacks prognostic utility. Recent Phase 3 trials, mandating the presence of both Aβ and tau for enrollment, may yield more robust cohorts.
Outcomes Measures
Another critical question arises: What should be the primary outcome measures? Successfully reducing Aβ levels has been accomplished multiple times, but achieving a substantial slowdown or cessation of cognitive decline remains elusive. This raises the crucial issue of determining the appropriate biomarker for clinical outcomes. Existing evidence from research studies and numerous clinical trials casts doubt on the prognostic utility of Aβ, and the definitive connection between Aβ and cognition is still undetermined. The evident disparity, exemplified by LEQEMBI's removal of 100% of cerebral amyloid resulting in only an average 27% slowing — rather than stopping — of decline, underscores the complexity of this issue.
Ultimately, cognition stands as the paramount concern for patients and their loved ones. Analogously, consider a mAb that diminishes the alpha-synuclein protein in Parkinson's disease patients without substantial improvement in motor function — would such an outcome be deemed acceptable and clinically meaningful? This prompts a searching reflection on what constitutes clinical significance.
In the context of a three-year randomized controlled clinical trial, the discernible impact between the drug and placebo arms should yield results that are unequivocally interpretable. To achieve this, it becomes imperative to precisely define what is considered clinically meaningful. Is it a reduction in dementia risk, and if so, by what degree? Alternatively, is it the halting or slowing of decline, and again, by what magnitude? The pursuit of clarity extends beyond biomarker changes, which demonstrate uncertain prognostic utility, and their translation into meaningful implications for patients remains unclear, especially in preclinical cohorts.
Equally disconcerting is the lack of scientific establishment surrounding cognitive outcome measures employed in clinical trials, such as the unverified PACC, while others like the well-established CDR rely on physician input, introducing inherent subjectivity. Notably, patients in AD clinical trials experiencing ARIA are unblinded due to treatment-related side effects. Upon their return to drug treatment in the trial, both patients and physicians become unblinded, leading to improved cognitive testing performance. Physicians, whether consciously or unconsciously, exhibit bias in scoring patients more favorably on the CDR.
The prevailing perspective in the field has predominantly embraced a reductionist approach to AD – pinpointing Aβ as the primary culprit leading to downstream tau aggregation. The simplistic notion that “fixing the Aβ issue” will inherently improve cognition may not hold true, as scientists challenge the linear relationship between Aβ and tau.
The disease's complexity surpasses the conventional understanding, as highlighted by the most recent draft of NIA-AA guidelines, which posits that amyloid alone constitutes the "biological definition of AD."
A study based on imaging and clinical trajectories even suggested the existence of four distinct "types" of AD. Notably, the majority of patients exhibit multiple pathologies, with pure AD being the exception rather than the rule. This complexity is further compounded by age-related comorbidities. Modelling indicates that "pathological Alzheimer’s disease only accounted for 50% of the cognitive loss," and recent research concludes that much of the pathological heterogeneity in neurologically unimpaired individuals is age-related.
Forward Progress
To progress meaningfully, our definitions and expectations must be clearly outlined to design randomized controlled trials (RCTs) with biomarkers that genuinely reflect the questions at hand. Amyloid alone proves insufficient as a recruitment biomarker due to its prevalence in the aging population and its poor performance as an efficacy marker for cognition (RE: LEQEMBI disparity between Aβ and cognition).
The suggestion of utilizing a combination of biomarkers, such as amyloid and tau, has been put forth, acknowledging that those with both pathologies and cognitive normalcy make up less than 10% of the population. Notably, the role of synapse loss in AD, strongly correlating with cognitive symptoms, has largely gone unrecognized.
Recent studies highlight that a key differentiator between individuals with AD neuropathology and dementia versus those who remain cognitively intact is a higher number of functioning synapses. This crucial aspect of AD was not discernible in the investigations conducted by Dr. Alzheimer and colleagues in the 1900s.
Shaping The Future
As the AD community grapples with the intricacies of biomarkers, trial outcomes, and the need for more precise diagnostic tools, the path forward requires a significant shift, redefining expectations and embracing a more comprehensive understanding of AD. Only by addressing these challenges can meaningful progress in AD research occur to benefit patients, caregivers, and the entire healthcare ecosystem.
About The Authors:
Frederick Huie, MD, SYNAPS Dx advisory council member, is a healthcare industry expert with 20+ years’ experience in clinical programs around risk, quality, and cost containment. He served 11 years with UnitedHealthcare’s Optum, clinical performance solutions division, designing clinical programs for early identification of chronic illness. More recently, Huie was with Aetna working on outcomes and cost containment.
He developed advanced algorithms around pharmacy, medical claims, labs, and natural language processing of chart records for predictive models geared to promote pre-clinical diagnosis of chronic illness and monitoring of active disease. Algorithms are actively used on a population of over 20 million Medicare and Medicaid lives across multiple healthcare payers.
Currently working in the public healthcare sector working for both Fire Department of New York, medical affairs emergency medical services (EMS), and department of sanitation New York (DSNY) occupational health, Huie is board certified in general surgery by American College of Surgeons.
Emmanuel C. Gorospe, MD, MPH, CPI, FACP chief medical officer at Great Lakes Research Network, is a Johns Hopkins and Mayo Clinic-trained, board-certified physician who has served as chief medical officer for a clinical research network. He is a certified principal investigator and has published as either primary author or co-author of more than 150 peer-reviewed articles, book chapters, and abstracts.
He started his career in aging research and has pursued other roles in drug and medical device development in different therapeutic areas. Most of his clinical research work is conducted in medically-underserved populations. He has worked with Alzheimer's advocacy groups at the community level to improve awareness and encourage participation in clinical trials. Currently, he is also a medical director at a clinical trials software company working on increasing efficiency in study enrollment and improving participation of underrepresented minority groups.