Longitudinal Changes in Peritoneal Transport and Their Impact on Dialysis Outcomes: A Machine Learning Approach...

Area of Science:

• Nephrology
• Biomarkers
• Machine Learning

Background:

• Peritoneal transport patterns in peritoneal dialysis (PD) and the predictive value of dialysate biomarkers are not well understood.
• Assessing longitudinal changes in peritoneal equilibration test (PET) trajectories and their clinical impact is crucial.

Purpose of the Study:

• To evaluate the impact of PET trajectory changes on clinical outcomes in PD patients.
• To explore the contribution of dialysate biomarkers in predicting transport changes.
• To develop a machine learning model for predicting peritoneal transport transitions.

Main Methods:

• Prospective study of 132 PD patients (2016-2020).
• Patients classified into four PET trajectory groups (HH, HL, LH, LL).
• Dialysate biomarkers (MMP2, PAI-1) quantified; Support Vector Machine (SVM) model developed.

Main Results:

• The persistent high (HH) and low to high (LH) groups (reclassified as future high) faced significantly higher risks of adverse outcomes.
• Elevated Matrix Metalloproteinase 2 (MMP2) and Plasminogen Activator Inhibitor 1 (PAI-1) appearance rates (ARs) were linked to high transporter status.
• The SVM model integrating clinical and biomarker data achieved higher predictive accuracy (AUC 0.87) than clinical data alone (AUC 0.71).

Conclusions:

• Sustained or transitioning to high transporter status increases adverse outcomes in PD patients.
• Higher MMP2 AR and PAI-1 AR levels enhance risk stratification for adverse PET trajectory changes.
• Biomarker-integrated predictive models improve prognostic accuracy for early intervention in high-risk PD patients.