Plasma PINK1/Parkin ratio and ATF5 as potential biomarkers associated with dyskinesia in Parkinson's disease

Background: Motor complications (MCs), particularly dyskinesia (DYS), affect many PD patients within years of levodopa therapy, yet reliable biomarkers for MC subtypes remain elusive. Mitochondrial quality control (MQC) pathways, including mitophagy and the mitochondrial unfolded protein response (UPR mt ), are central to PD pathogenesis; however, their association with MCs remains to be clarified. Methods: In this cross-sectional study, a total of 247 PD patients were enrolled and analyzed in a mitophagy discovery set ( n  = 174), a UPR mt discovery set ( n  = 179), and an overlap subset ( n  = 106) with complete data for both biomarker panels, which was used for internal consistency analyses. Plasma levels of PINK1, Parkin, PGAM5, ATF4, ATF5, and CHOP were measured by ELISA. Patients were classified as MC, further subtyped as wearing-off (WO) or DYS, or NMC. Univariate analysis identified candidate biomarkers; multivariable logistic regression and LASSO regression evaluated associations after covariate adjustment; and ROC analysis exploratorily assessed discriminative accuracy, with bootstrap resampling used to evaluate model stability. Results: Neither mitophagy nor UPR mt markers distinguished MC from NMC. However, after batch correction and covariate adjustment, the PINK1/Parkin ratio (OR = 2.06, p  = 0.044) and ATF5 (OR = 2.27, p  = 0.010) remained associated with DYS. A three-biomarker panel comprising PINK1, the PINK1/Parkin ratio, and ATF5 achieved an AUC of 0.82 (95% CI: 0.68–0.93) in the overlap subset. Nonlinear associations with DYS status were observed, following a DYS > dyskinesia with wearing-off (DYSWO) > WO pattern. Conclusions: Circulating MQC biomarkers, particularly the PINK1/Parkin ratio and ATF5, are associated with prevalent DYS in PD, suggesting that MQC-related pathways may be involved in DYS pathophysiology.