Objectives: To test the prognostic impact of arterial lactate concentration at the end-of-surgery (LCT-EOS) on postoperative outcome after elective liver-resections and to identify the predictors of an increase in LCT-EOS.
Background data: A recent systematic-review of risk-prediction-models for liver resections has evidenced their poor accuracy and a deficit in the evaluation of intraoperative events. LCT-EOS is a marker of impaired tissue oxygenation.
Methods: This prospectively-designed study was based on a training-cohort of 519 patients and a validation-cohort of 466 patients. For each of the endpoints (high comprehensive complication index (CCI) scores, 90-day-mortality and severe-morbidity), prognostic-models were built by logistic-regression using the training-cohort. These models were thereafter tested in the validation-cohort and their performance (discrimination, accuracy, calibration) assessed. Independent predictors of LCT-EOS were also identified.
Results: In the training-cohort, LCT-EOS cutoff best discriminating high-CCI, 90-day-mortality and severe-morbidity were 3, 3 and 2.8 mmol/L (and the corresponding AUROC 0.86, 0.87 and 0.76). LCT-EOS was an independent predictor of endpoints and adding LCT-EOS to the other predictors increased by 16.4%, 34.5% and 17.7% the accuracy of the models for high-CCI, 90-day-mortality and severe-morbidity, respectively. The models had high calibration and accuracy. Diabetes, repeat-hepatectomy, major-hepatectomy, synchronous-major-procedure, inflow-occlusion and blood-transfusion were independent predictors of LCT-EOS >3 mmol/L.
Conclusions: LCT-EOS >3 mmol/L is an early predictor of postoperative-outcome and should be used as a tool to determine patients requiring critical-care and as an endpoint in studies measuring the impact of perioperative interventions.