Drug development is fundamentally a test of a hypothesis. A therapy is designed to affect a molecule, a pathway, or a biological process believed to contribute to that disease. The clinical trial then asks whether changing that target changes outcomes that matter to patients. In neurodegenerative disease, that logic is sound, but applying it is unusually difficult.
Disease labels such as Alzheimer’s and Parkinson’s disease are clinically useful, yet do not necessarily define biologically uniform populations. Patients with the same diagnosis can have different combinations of protein pathology, patterns of brain involvement, and rates of progression and similar symptoms can arise for different biological reasons.
For drug developers, that distinction matters. If a therapy targets a specific molecular process, researchers need to understand which relevant biology is present, how it relates to functional impairment, and whether it changes after treatment. No single biomarker or clinical assessment can provide all of those answers. The future of drug development for neurodegenerative diseases will therefore depend on combining complementary measurements to make trials more informative.
Why one measure does not tell the whole story
The ability to detect disease-related proteins using biomarkers and imaging has changed the field, enabling researchers to study biological processes in living patients rather than relying solely on clinical symptoms or postmortem findings. But identifying one pathological feature does not necessarily explain the entire disease process in an individual patient.
Consider the layered nature of neurodegeneration. Where and how pathology progresses can influence whether disease manifests through memory loss, impaired movement, behavioral changes, sleep disturbances, or some combination. For therapeutic development then, trial design must account for both the target biology and the patient’s broader disease state.
A multimodal approach does not mean every patient must receive every available test. It means selecting measures that provide distinct and decision-relevant information.
Molecular biomarkers can establish whether a disease-related process or therapeutic target is present and potentially provide evidence of target engagement when measured over time. Imaging adds information about regional tissue patterns or pathological processes, while clinical assessments capture what patients and families experience with changes in cognition, daily function, movement, behavior, and quality of life. Digital measures of speech, movement, cognition, behavior, or sleep collected through video, voice, surveys, or web-based tools may allow investigators to capture data more frequently and in some cases, outside the clinic.
These modalities are not interchangeable. A biomarker can indicate biology without fully describing functional impact, a cognitive score can show impairment without identifying its cause, and digital data can quantify change over time but require clinical and biological context to be interpretable. Their value lies in how they all fit together.
From disease labels to biologically informed stratification
For a target-specific therapy, patient selection should begin with understanding if the relevant target biology is present. A developer may choose to enroll only participants with a defined molecular profile, creating a more homogeneous population to test a strong target hypothesis. Alternatively, a study may enroll a broader population while prospectively characterizing biological subtypes. That approach can require more patients, more measurements, and greater operational complexity, but it may reveal that a therapy works differently across distinct molecular or clinical profiles.
Neither strategy is universally correct. The appropriate design depends on the maturity of the target hypothesis, the available evidence, the prevalence of relevant biology, and the goal of the program. Multimodal characterization can then help developers ask more precise questions in clinical trials. Answering these questions may not make trials easier, but it may make them more informative and more capable of generating interpretable, valuable evidence.
Better endpoints to link target engagement to meaningful change
Neurodegenerative disease trials often face a familiar challenge of demonstrating that a biological change translates into a clinically meaningful benefit. For example, some biomarkers are particularly useful as threshold measures, however, binary results alone may not fully capture the full course of disease or explain whether a change in pathology affects day-to-day function.
This is why trial endpoints should connect multiple levels of evidence. A molecular measure can show whether a therapy engages its intended target, imaging can provide a related view of regional or pathological change, and cognitive, motor, behavioral, and functional measures can assess whether that biological effect is associated with changes that matter to patients. Digital measures may also help detect subtle or gradual changes that are difficult to capture during intermittent clinic visits.
Longitudinal digital measure may add another layer. Comparing a patient with a population average has value, but comparing that patient with their own baseline may be equally as informative. A change in speech cadence, gait, memory, sleep, or daily behavior may become meaningful when evaluated across repeated measurements and interpreted alongside biological data for a richer picture of therapeutic response.
The goal is fewer inconclusive trials, not an end to negative results
Multimodal assessment will not guarantee positive results. Drug development requires taking hypotheses seriously enough to test them, and well-run negative studies can still move the field forward. A clear negative result may show that a target, dose, population, or intervention strategy is not sufficient and that knowledge is extremely valuable.
The greater failure is an inconclusive trial. When investigators cannot determine whether participants had the intended biology, whether a therapy engaged its target, or whether the endpoint adequately reflected meaningful change, a study may generate more uncertainty than insight. That outcome costs time, resources, and the effort of patients and clinicians who chose to participate.
Integrating molecular, imaging, clinical, and digital measures can help reduce this ambiguity. A multimodal trial can clarify whether an overall negative result reflects a failed biological hypothesis, inadequate target engagement, an inappropriate endpoint, insufficient follow-up, or meaningful benefit in a subgroup that would otherwise have been obscured.
Translating this model to practice
Scientific promise alone does not guarantee adoption. Moving toward multimodal assessment, developers and clinicians will have to meet standards for validation and regulation while navigating reimbursement, clinical workflows, and existing infrastructure. More scalable approaches, including blood-based biomarkers and remotely collected digital measures, may broaden participation, but they do not eliminate the need for rigorous clinical interpretation.
The objective should not be to replace one test with a larger panel of disconnected tests. It should be to build an evidence-generation framework that links disease biology with disease manifestation and therapeutic response. For neurodegenerative drug development, the question goes beyond whether one measure can define disease. It is now whether we can combine the right measurements to learn, with confidence, which biology matters for which patient and whether a therapy is changing it.
