Association Between Microbial Tyrosine Decarboxylase Gene and Levodopa Responsiveness in Patients With Parkinson Disease

Background and Objectives

Interindividual variability in levodopa efficacy is a challenge for the personalized treatment of Parkinson disease (PD). Gut microbiota might represent a new approach for personalized medicine. Recently, a novel microbial levodopa metabolism pathway was identified, which is mediated by tyrosine decarboxylase mainly encoded by tyrosine decarboxylase gene (tyrDC) in Enterococcus faecalis. In this study, we aimed to identify whether the abundance of microbial tyrDC gene and E faecalis is associated with levodopa responsiveness and could predict the drug response.

Methods

This cross-sectional study enrolled patients with PD between December 2019 and January 2022 and evaluated levodopa responsiveness using a levodopa challenge test. Patients were stratified into moderate and good responders based on levodopa responsiveness. The tyrDC gene and E. faecalis abundance in fecal samples were measured using quantitative real-time PCR. Plasma levodopa concentrations were measured using liquid chromatography-tandem mass spectrometry analysis. The predictive models for levodopa responsiveness were constructed and verified through cross-validation and external validation.

Results

A total of 101 patients with PD were enrolled in the primary cohort and 43 were enrolled in the external validation cohort. Moderate responders had higher abundances of the tyrDC gene (3.6 [3.1–4.3] vs 2.6 [2.1–2.9], p < 0.001) and E faecalis (3.2 [2.5–4.4] vs 2.6 [2.1–3.6], p = 0.010) than good responders. The tyrDC gene abundance was independently associated with levodopa responsiveness (OR: 5.848; 95% CI: 2.664–12.838; p < 0.001). Notably, tyrDC gene abundance showed certain discriminative power for levodopa responsiveness in primary cohort (sensitivity: 80.0%; specificity: 84.3%; area under the curve [AUC]: 0.85; 95% CI: 0.77–0.93; p < 0.001) and external validation cohort (sensitivity: 85.0%; specificity: 95.7%; AUC: 0.95; 95% CI: 0.89–1.02; p < 0.001). The prediction of levodopa responsiveness based on tyrDC gene abundance had good calibration and discrimination in cross-validation (C-index in training and test sets: 0.856 and 0.851, respectively) and external validation (C-index: 0.952).

Discussion

The microbial tyrDC gene abundance could serve as a potential biomarker of levodopa responsiveness. Novel strategies targeting the tyrDC gene may provide new approaches for personalized levodopa treatment.