Cardiopulmonary bypass (CPB) is essential to cardiac surgery but is associated with organ injury from non-physiological perfusion. Goal-directed perfusion (GDP) has emerged as a physiology-based strategy to optimize intraoperative perfusion by targeting specific hemodynamic and metabolic parameters, particularly oxygen delivery (DO₂). This review critically appraises the evidence supporting GDP during CPB, emphasizing comparative strategies, clinical outcomes, and limitations.

A narrative review of the literature was conducted, incorporating evidence from randomized controlled trials, observational studies, meta-analyses, and international guidelines addressing perfusion strategies during CPB. This review included studies evaluating DO₂-guided, pressure-guided, and oxygen-metabolic-indicator-guided approaches, focusing on target definitions, outcome measures, and patient populations.

Randomized and observational studies have consistently demonstrated that maintaining adequate DO₂ levels during CPB is associated with a reduced incidence of cardiac surgery–associated acute kidney injury (CSA-AKI), leading to the guideline endorsement of DO₂-guided perfusion. In contrast, evidence on neurological outcomes, length of stay, and mortality remains limited and inconsistent. Pressure-guided and venous-oxygen-saturation– or oxygen-metabolic-indicator–guided strategies provide complementary physiologic information but have not independently demonstrated superiority in randomized trials. Substantial heterogeneity has been reported in perfusion targets, outcome definitions, patient risk profiles, and potential trade-offs, including increased transfusion exposure and implementation complexity.

Current evidence supports DO₂-guided perfusion as an effective strategy to reduce CSA-AKI during CPB, although its impact on broader clinical outcomes remains uncertain. Future studies should focus on individualized perfusion targets that account for patient-specific risk, metabolic demand, temperature, and autoregulatory capacity, rather than universal thresholds.