Abstract
Animal models have been used to gain pathophysiologic insights into Parkinson's disease (PD) and aid in the translational efforts of interventions with therapeutic potential in human clinical trials. However, no disease-modifying therapy for PD has successfully emerged from model predictions. These translational disappointments warrant a reappraisal of the types of preclinical questions asked of animal models. Besides the limitations of experimental designs, the one-size convergence and oversimplification yielded by a model cannot recapitulate the molecular diversity within and between PD patients. Here, we compare the strengths and pitfalls of different models, review the discrepancies between animal and human data on similar pathologic and molecular mechanisms, assess the potential of organoids as novel modeling tools, and evaluate the types of questions for which models can guide and misguide. We propose that animal models may be of greatest utility in the evaluation of molecular mechanisms, neural pathways, drug toxicity, and safety but can be unreliable or misleading when used to generate pathophysiologic hypotheses or predict therapeutic efficacy for compounds with potential neuroprotective effects in humans. To enhance the translational disease-modification potential, the modeling must reflect the biology not of a diseased population but of subtypes of diseased humans to distinguish What data are relevant and to Whom.