How Alzheimer’s Disease Studies Need to Evolve in the Targeted Therapy Era

While disease-targeted anti-amyloid monoclonal antibodies (mAbs) have advanced the treatment of Alzheimer’s disease (AD), their advent has also complicated the research landscape for future disease-modifying therapies for AD. That’s the premise behind a new perspective article published in Alzheimer’s & Dementia by an international group of expert AD trialists.

The authors go on to outline an operational framework to guide the next phase of AD research and move the field forward by addressing new research questions and priorities that have emerged with the use of anti-amyloid therapies.

“Future therapeutic progress requires moving beyond broad, one-size-fits-all trial designs toward biologically stratified, mechanism-driven designs,” says lead author Jagan Pillai, MD, PhD, Director of Cleveland Clinic’s Center for Brain Health in Cleveland. “Anti-amyloid monoclonal antibodies are a significant early step, but progress will require better-harmonized longitudinal cohorts, deeper integration of biomarkers, broader patient representation and robust use of combination trials tailored to the biological variability of Alzheimer’s disease.”

Accounting for biological heterogeneity

Phase 3 trials of mAbs such as lecanemab and donanemab have confirmed substantial clearance of amyloid-beta plaques, but the corresponding reductions in cognitive decline have been modest and variable among patients. “This divergence underscores both the uncertain causal links between biomarker changes and clinical benefits and the extent to which Alzheimer’s is a multifactorial condition driven by complex biological heterogeneity,” Dr. Pillai observes.

His group’s new article emphasizes that AD variability is one of the leading barriers to progress. Variations in clinical phenotype (e.g., amnestic vs. atypical), genetic predisposition (APOE ԑ4 status), baseline tau pathology burden, and concomitant vascular or nonamyloid protein pathologies can strongly influence individual treatment response. Because of this heterogeneity, the authors argue, future trials should stratify participants more carefully and use multimodal biomarkers (proteomic, immunologic, imaging and digital) instead of relying solely on broad diagnostic categories.

Biomarkers receive considerable discussion in the article. The authors support the use of biomarkers for patient selection and mechanistic monitoring — particularly amyloid, tau and measures of neurodegeneration — but warn against their overinterpretation as surrogate endpoints for clinical benefit. Biomarker trajectories vary across individuals and phenotypes, and plasma markers can be affected by systemic illnesses such as renal disease. The authors highlight the potential of scalable blood-based biomarkers like p-tau217, p-tau231, eMTBR-tau243, glial fibrillary acidic protein and neurofilament light chain, but they underscore the need to standardize and validate assays across populations.

Improving observational research

The article strongly advocates for better longitudinal observational cohorts in the mAb era. Existing resources such as Alzheimer’s Disease Research Centers, the National Alzheimer’s Coordinating Center and other large databases provide a foundation, but the authors say these systems need more detailed and harmonized capture of treatment exposure, including exact start and stop dates, interruptions, dose, route of administration and safety events. Most notably, amyloid-related imaging abnormalities (ARIAs) should not be treated only as adverse events, as they may also offer insight into vascular vulnerability, blood-brain barrier integrity and the biology of treatment response.

Long-term follow-up should be a priority — including after treatment cessation — to understand durability, washout effects and delayed consequences.

Study populations need to reflect real-world practice

AD trial data to date have had limited generalizability, the article notes. Study participants have often been disproportionately white, highly educated, urban and medically healthier than real-world patients. This creates blind spots in understanding how therapies perform in rural populations, low-resource settings, and low- and middle-income countries. Social drivers of health, caregiving burden, transportation barriers, insurance coverage and access to specialty care all influence who gets diagnosed, enrolled, treated and retained in studies. The authors call for community-engaged trial recruitment, broader international registries and context-specific study designs that better reflect global populations.

Rethinking trial design in the mAb era

The article points out a notable irony: AD clinical trials could be a victim of the anti-amyloid therapies’ relative success. As reimbursement and adoption of these agents expands, recruiting treatment-naïve control cohorts will increasingly pose logistical and ethical challenges.

In response, the authors call on trialists to evaluate future candidate therapies as add-on or combination regimens that augment established mAb foundations rather than testing them against simple monotherapy placebos.

They note that combination therapy approaches, modeled partly on cancer and HIV treatment, may eventually target amyloid plus tau, inflammation, vascular dysfunction, metabolism or resilience pathways. Trial endpoints should include both early mechanistic measures and later clinical outcomes, and adaptive or platform trial designs may improve study efficiency.

To evaluate the safety of secondary targeted therapies without confounding side-effect profiles, the authors highlight the need for sequential add-on trial designs to evaluate how they interact with existing mAbs. “Starting an investigational therapy concurrently with a monoclonal antibody risks obscuring adverse event signals, particularly ARIAs,” Dr. Pillai notes.

Because nearly all ARIAs manifest in the initial months of mAb administration, the authors recommend delaying introduction of an investigational add-on agent until six to 12 months after the monoclonal antibody is started, to ensure clearer safety attribution and cleaner evaluation of the add-on therapy’s efficacy.

Targeting partial responders and co-pathologies

The article also offers recommendations on participant selection and subgroup enrichment. Pivotal trials of anti-amyloid therapies found that approximately 15% to 19% of treated patients failed to achieve amyloid levels below established PET thresholds. These suboptimal or partial responders show persistent cognitive decline despite full-dose therapy, representing a critical patient cohort for secondary disease-modifying interventions.

At the same time, trial designs must also explicitly account for nonamyloid co-pathologies. Vascular disease, alpha-synuclein pathology or limbic-predominant age-related TDP-43 encephalopathy (LATE) frequently co-occur with AD pathology. The authors note that integrating emerging blood-based assays and validated vascular MRI markers into study protocols will allow investigators to isolate primary treatment response from cognitive decline driven by co-pathologies. They also call for more autopsy studies to clarify how mixed pathologies affect outcomes in the setting of anti-amyloid therapy.

More data transparency needed

Finally, the authors argue that the field needs greater transparency and data sharing. Patient-level trial data, including information on immunogenicity and anti-drug antibodies, should be made more available for independent analysis. Real-world electronic medical record data, artificial intelligence-based phenotyping, digital biomarkers and harmonized registries could all help identify responder and non-responder subgroups and improve trial design.

The full open-access perspective article is available here.