Abstract: PO0247
Preoperative Biomarkers and Mortality Risk After Cardiac Surgery
Session Information
- AKI: Clinical, Outcomes, and Trials
November 04, 2021 | Location: On-Demand, Virtual Only
Abstract Time: 10:00 AM - 12:00 PM
Category: Acute Kidney Injury
- 102 AKI: Clinical, Outcomes, and Trials
Authors
- Liu, Caroline, Icahn School of Medicine at Mount Sinai, New York, New York, United States
- Menez, Steven, Johns Hopkins Medicine, Baltimore, Maryland, United States
- Moledina, Dennis G., Yale University School of Medicine, New Haven, Connecticut, United States
- Thiessen Philbrook, Heather, Johns Hopkins Medicine, Baltimore, Maryland, United States
- McArthur, Eric, Institute for Clinical Evaluative Sciences, Toronto, Ontario, Canada
- Obeid, Wassim, Johns Hopkins Medicine, Baltimore, Maryland, United States
- Mansour, Sherry, Yale University School of Medicine, New Haven, Connecticut, United States
- Garg, Amit X., Institute for Clinical Evaluative Sciences, Toronto, Ontario, Canada
- Parikh, Chirag R., Johns Hopkins Medicine, Baltimore, Maryland, United States
- Coca, Steven G., Icahn School of Medicine at Mount Sinai, New York, New York, United States
Group or Team Name
- TRIBE-AKI Consortium
Background
Cardiac surgery patients are at an increased risk for developing adverse outcomes. Preoperative blood and urine biomarkers may help stratify cardiac surgery patients at high risk for mortality.
Methods
The TRIBE-AKI study enrolled 1526 patients undergoing cardiac surgery in the USA and Canada from 2007-2010 and was randomly split into a training and test dataset (70:30). A total of 32 plasma and 17 urine biomarkers were measured preoperatively. The primary outcome was 3-year mortality. Random forest (RF) and LASSO logistic regression models were used to identify top biomarkers. Logistic regression models with the highest performing biomarkers and the Society of Thoracic Surgeons (STS) risk calculator were evaluated and the discriminatory ability was assessed in the test dataset.
Results
Death by 3 years occurred in 163 of the 1526 (10.7%) patients. LASSO logistic regression models retained the STS score and 6 plasma biomarkers (Troponin, IL-6, KIM1, NT-proBNP, TNFR1, YKL-40). The top 6 biomarkers identified by random forest were plasma KIM-1, TNFR1, eGFR, TNF-R2, hsTNT, and urine IL-8. In logistic regression models, the AUC in the test dataset for the STS clinical model was 0.68 (0.61, 0.76) and increased to 0.72 (0.65, 0.79) with the addition of 8 plasma and 2 urine biomarkers (plasma Troponin, IL-6, KIM-1, NT-proBNP, TNFR1, YKL-40, hFABP, TNFR2, and urine IL-8 and albumin; p=0.24).
Conclusion
The addition of biomarkers improved discrimination for 3-year mortality prediction minimally beyond clinical characteristics alone. The clinical utility of measurement of biomarkers pre-operatively prior to cardiac surgery is suspect.
Funding
- NIDDK Support