BACKGROUND: Tissue oxygenation measured by near-infrared spectroscopy can guide hemodynamic management, yet the joint association of mean arterial pressure (MAP) and cardiac index (CI) with tissue oxygenation, and the conditional association of each at fixed levels of the other, remain incompletely defined. METHODS: We analyzed intraoperative 1-Hz signals from patients undergoing off-pump coronary artery bypass surgery. Synchronized MAP, CI, and forehead and forearm tissue oxygenation (SctO2/SftO2) were modeled using generalized additive models. Two-dimensional smoothing over MAP-CI estimated surfaces and two-dimensional slices characterized conditional relationships. Covariates were prespecified. Bootstrap quantified uncertainty and 95% confidence intervals. We emphasized estimation and visualization. RESULTS: Analytic cohorts included 1,864 patients for left SctO2, 1,863 for right SctO2, and 1,860 for SftO2, each with more than 20 million eligible observations. Adjusted surfaces characterized mean tissue oxygenation across combinations of MAP and CI. Two-dimensional slices showed that SctO2 depended on both MAP and CI. The MAP-SctO2 relationship showed a substantial reduction in slope over a MAP range of 70 to 100 mmHg, mimicking cerebral pressure autoregulation. SctO2 increased as CI rose, reached its highest level at approximately 4.2 L/min/m2, and then plateaued or slightly decreased at higher CI. SftO2 was sensitive to CI, not MAP; it increased most steeply at CI 2.8 L/min/m2 or less, showed a reduced slope over the CI range of 2.8 to 4.2 L/min/m2, and continued to increase at higher CI. These patterns were more prominent in younger patients, females, and those with a baseline blood pressure below 140/90 mmHg. CONCLUSIONS: During off-pump coronary artery bypass surgery, cerebral oxygenation was sensitive to both pressure and flow, whereas forearm oxygenation primarily tracked flow. Reductions in slope across specific pressure and flow ranges may reflect physiologic regulation, but should not be interpreted as evidence for treatment targets or practice-changing interventions.
Wang et al. (Mon,) studied this question.