A pair of updated scoring models for estimating a patient’s risk for contrast-associated acute kidney injury during and immediately after percutaneous coronary intervention worked better than a widely used prior version in initial validation testing using data collected at a single U.S. tertiary-care hospital.
While the two new risk scores looked promising, they need further, external validation with additional, diverse patient cohorts, Roxana Mehran, MD, cautioned at the American Heart Association scientific sessions.
“Don’t change anything until we externally validate this,” urged Dr. Mehran, professor and director of the Center for Interventional Cardiovascular Research and Clinical Trials at Icahn School of Medicine at Mount Sinai in New York. External validation of the two new risk scores is in progress with planned reporting of the results in 2022, she said in an interview.
One of the two new algorithms, which both predict a patient’s risk for developing acute kidney injury (AKI) as a result of receiving iodinated contrast media within 48 hours of a percutaneous coronary intervention (PCI), relies on eight easily available variables taken from a patient’s medical record just prior to undergoing PCI: age, type of coronary disease (ranging from asymptomatic or stable angina to ST-segment elevation MI), estimated glomerular filtration rate, left ventricular ejection fraction, diabetes, hemoglobin, basal glucose, and heart failure.
This risk score accounted for 72% (a C-statistic of 0.72) of the observed AKI episodes during the derivation phase, which used data from 14,616 consecutive Mount Sinai patients who underwent PCI during January 2012–December 2017.
Internal validation using data from 5,606 consecutive Mount Sinai patients who underwent PCI during January 2018–December 2020 showed that the eight-item formula accounted for 84% of all incident AKI events occurring during or within 48 hours of a PCI procedure.
Accounting for periprocedural variables
A second risk score included the eight preprocedural variables plus four additional periprocedural variables: complex PCI anatomy, contrast volume during the procedure, development of a periprocedural bleed, and having slow or no reflow into affected coronaries (less than TIMI grade 3 flow) immediately after the procedure. The second model produced a C-statistic of 0.74 during derivation and accounted for 86% of incident AKI events in the validation analysis.
The data Dr. Mehran reported appeared in The Lancet .
She and her coauthors designed these two new algorithms to replace a “widely used” and externally validated risk score that Dr. Mehran and associates introduced in 2004. Despite its merits, the 17-year-old scoring formula has limitations including “low discrimination” with a C-statistic of 0.67, derivation from data that’s now 20 years old, and exclusion of patients with ST-elevation MIs, the authors said in the new report.
Dr. Mehran encouraged interventional cardiologists to use both new risk scores (once externally validated) when possible.
The eight-item preprocedural model “gives clinicians an idea about a patient’s risk [for incident AKI] before they go into the catheterization laboratory,” and then they can further refine the risk assessment during the procedure based on the four periprocedural risk factors, she explained. The goal is to target “tailored preventive strategies” to patients identified by the scoring algorithms as being at high risk for AKI.