QSP Modeling for Parkinson’s Disease: Solving the Synthetic Control Challenge in Clinical Trials

Parkinson’s disease (PD) is the fastest-growing neurodegenerative disease, second only to dementia in prevalence. Its motor symptoms can be readily managed with standard-of-care dopaminergic therapy, such as L-Dopa or dopamine agonists. Yet the search for disease-modifying treatments has met with mixed success, and the reasons say a great deal about why QSP modeling for Parkinson’s disease has become such a valuable tool for designing, measuring, and interpreting trials.

Recent programs illustrate the point. Roche’s prasinezumab, an anti–alpha-synuclein antibody, did not reach statistical significance on its primary endpoint, but it did demonstrate a robust clinical signal in a subset of patients on specific comedications (Pagano et al., 2024); a new trial is now underway based on those findings. In parallel, the ongoing discussion around uniQure’s AMT-130 gene therapy for Huntington’s disease (Mwape et al., 2026), another neurodegenerative disease with a strong motor component, underscores the challenges companies face when they must rely on single-arm trials.

These programs share a common difficulty: in PD, demonstrating the benefit of a disease-modifying therapy is exceptionally hard. Below, we examine four challenges that complicate PD clinical development, and how quantitative systems pharmacology (QSP) modeling, delivered through a mechanism-based Virtual Twin® approach, can help sponsors detect a true functional signal and make better go/no-go decisions.

Challenge 1: Optimized standard-of-care makes clinical signals hard to detect

The gold standard of care for PD is L-Dopa together with various dopamine agonists. Because these regimens can be optimized to deliver maximal symptom relief, detecting an incremental functional signal from a disease-modifying agent in an augmentation trial becomes a tall order. In the absence of a validated surrogate biomarker for PD pathology, sponsors must rely on clinical readouts alone to prove benefit, a high bar to clear.

Physicians also retain the prerogative to adjust standard-of-care during a trial to accommodate changes in their patients’ symptoms. Although these modifications can be well documented, they add considerable variability that further complicates the interpretation of clinical benefit.

Challenge 2: A strong placebo response confounds the readout

A defining feature of PD clinical trials is the strong placebo response, driven by dopamine’s tight association with the brain’s reward circuitry. This response is even greater for more invasive interventions, such as gene and stem cell therapies, because patients are more deeply invested in the outcome.

These studies often face the added difficulty of enrolling a placebo-control arm. In small, single-arm Phase 2a studies, active-treatment outcomes are frequently compared against a synthetic control built from a historical observational study. But such longitudinal control lacks the strong placebo response that comes with the knowledge of being enrolled in a clinical trial—biasing the comparison.

Challenge 3: There is no validated biomarker for alpha-synuclein pathology

Unlike Alzheimer’s disease, where well-validated fluid and imaging biomarkers exist for amyloid and tau pathology, PD currently lacks reliable surrogate biomarkers. Efforts are underway to validate seed amplification assays (SAA) and to develop imaging biomarkers for alpha-synuclein, the core deposit in Lewy bodies, but functional clinical outcomes such as the Unified Parkinson’s Disease Rating Scale (UPDRS) remain the gold standard.

Regulatory agencies are increasingly open to more patient-centric readouts—for example, daily Good ON-time, the duration over which symptoms are well controlled, potentially captured through wearables (Rose et al., 2022). However, ON-time is largely driven by the pharmacokinetics and pharmacodynamics of L-Dopa formulations and dopamine agonists, introducing high variability between individual patient readouts.

Challenge 4: Motor symptoms range from rigidity to dyskinesia

Long-term L-Dopa treatment can trigger the onset of troublesome dyskinesia—the jittery, involuntary movement known as L-Dopa–induced dyskinesia (LID)—which affects both Good ON-time and the patient experience. LID is often managed with the NMDA modulator amantadine, which can further complicate the pharmacodynamic interaction with an investigational therapeutic.

The solution: QSP modeling for Parkinson’s disease with individualized Virtual Twin® controls

Certara offers a mechanism-based Virtual Twin® Quantitative Systems Pharmacology (QSP) model that serves as an individualized synthetic control for each real patient on their actual treatment. The model is built on the anatomical and functional connections of the cortico-striatal-thalamocortical motor circuit (Roberts et al., 2016). Informed by insights from deep-brain recording studies, its readout is extensively calibrated against clinical data for both UPDRS and Good ON-time.

Each Virtual Twin® carries the same baseline standard-of-care comedications and disease state as its real-world counterpart in the trial. Its clinical trajectory on motor symptoms, whether UPDRS or OFF-time, is then simulated in an untreated condition, with the placebo effect included. This enables an evaluation of clinical effect on a per-patient basis, significantly reducing the variability inherent in group averages. In effect, the approach delivers the statistical power of a crossover study, where each patient serves as their own placebo control, within the practicality of a traditional parallel-group design.

The result: sponsors can identify a real functional clinical signal in single-arm trials and small proof-of-concept Phase 2 studies for disease-modifying therapies—building confidence in the therapeutic molecule and supporting optimal clinical trial design.