ѻý

Tool Predicts Kidney Injury Risk in Stenting

Last Updated May 30, 2013
MedpageToday

This article is a collaboration between ѻý and:

A technique using standard clinical factors shows promise for predicting kidney injury associated with contrast medium during coronary stenting, researchers found.

In the fully detailed model, which contains 46 variables, the predicted risk (low, intermediate, high) of contrast-induced nephropathy (CIN) was well matched with the actual rate: low -- 0.46%, intermediate -- 2.59%, and high -- 13.6%, reported Hitinder S. Gurm, MD, of the University of Michigan in Ann Arbor, and colleagues.

Action Points

  • New tools have been developed and validated for predicting risk of contrast-induced nephropathy (CIN) in patients undergoing contemporary percutaneous coronary intervention.
  • The tools appear to be easy to use and reliably estimate the risk of both CIN and new requirements for dialysis.

In the reduced model, which contains 15 of the most influential variables, the predicted risk again correlated well with the actual rate of CIN: low -- 0.51%, intermediate -- 2.83%, and high -- 12.99%, according to an online version of the study slated for the June 4 edition of the Journal of the American College of Cardiology.

Both the full and reduced models had excellent discrimination power for CIN, with areas under the receiver-operating curve (AUC) of 0.85 and 0.84, respectively.

The values for the risk categories in both models are:

  • Low risk -- less than 1%
  • Intermediate risk -- 1% to 7%
  • High risk -- greater than 7%

Both models also had excellent discrimination for identifying patients at risk for new renal dialysis, with an AUC of 0.88 for both.

Gurm and colleagues also tested the two models against the standard glomerular filtration rate (GFR)-based strategy for risk assessment.

Using the conventional cutoff of less than 60 ml/min/1.73 m2 in a cohort of low-to-intermediate-risk patients, the GFR-based strategy would identify only about two-thirds (62%) of those who would go on to develop CIN.

In the same cohort, the full and reduced models would identify 75% and 72%, respectively.

Although the full model resulted in slightly greater discrimination, the difference was not significant, researchers said. The net reclassification improvement for the full model relative to the reduced model was 2.92% (95% CI -0.14-6.03, P=0.062).

The investigators said they envision the fuller-detailed model to be used initially for "quality assessment and benchmarking across institutions and operators." The simpler model would be more appropriate initially for bedside risk calculation.

Gurm and colleagues also pointed out that their models demonstrated much higher discrimination power for predicting acute kidney injury (AKI) compared with other models.

To develop their models, the researchers used data from approximately 48,000 patients in the Blue Cross Blue Shield of Michigan Cardiovascular Consortium, a quality improvement collaborative that tracks the inpatient outcome of consecutive patients undergoing percutaneous coronary intervention (PCI).

Another 25,000 patients were used to validate the model.

Because the database reflects contemporary practice across multiple institutions and operations, the models should be generalizable to routine clinical practice, Gurm and colleagues said.

They also noted that the models were based on preprocedure variables, thus allowing interventionalists to use the tool for risk stratification before PCI -- giving ample time to devise alternative strategies.

The 15 most influential variables selected for the reduced model are:

  • Patient presentation: PCI indication, PCI status, coronary artery disease presentation, cardiogenic shock, heart failure within 2 weeks, and pre-PCI left ventricular ejection fraction
  • Clinical history: Diabetes and/or diabetes therapy
  • Patient characteristics: Age, weight, and height
  • Preprocedural laboratory assessments: Creatine kinase-MB, serum creatinine, hemoglobin, troponin I, and troponin T

The full model also includes eight preprocedural therapy variables -- none of which made the cut for the reduced model -- two more variables for patient presentation, 18 more for clinical history, and three more patient characteristic factors.

Limitations of the study include its observational nature, the potential for the cause of acute kidney injury to be multifactorial rather than solely based on contrast medium, limited data on serum creatinine, and lack of data on type and amount of hydration.

From the American Heart Association:

Disclosures

The BMC2 registry is funded by Blue Cross Blue Shield of Michigan. The sponsor had no role in study design or review or the decision to submit the work for publication.

Gurm receives research funding from Blue Cross Blue Shield of Michigan, the National Institutes of Health, and the Agency for Healthcare Research and Quality. One co-author is employed by Blue Cross Blue Shield of Michigan. All other co-authors have reported that they have no conflicts of interest.

Primary Source

Journal of the American College of Cardiology

Source Reference: Gurn HS, et al "A novel tool for reliable and accurate prediction of renal complications in patients undergoing percutaneous coronary intervention" J Am Coll Cardiol 2013; 61: 2242–2248.