2019 American Diabetes Association Presentation for HART AKI

A Clinical, Proteomics and Artificial Intelligence-Driven Model to Predict Acute Kidney Injury in Diabetic Patients Undergoing Coronary Angiography


American Diabetes Association (ADA) Scientific Sessions, San Francisco, CA

June 10, 2019

Highlights & Key Findings

  • Acute kidney injury (AKI) following cardiac procedures has substantial impact on prognosis; patients with diabetes mellitus (DM) are at high risk. We previously developed a clinical/proteomics panel for predicting procedural AKI and now examine its performance in those with DM.

  • An estimated 23.4 million, or 9.1 percent, of American adults are diagnosed with diabetes, with an estimated 7.6 million, or 3.1 percent, of American adults with undiagnosed diabetes. Additionally, about 81.6 million, or 33.9 percent, of American adults have prediabetes or abnormally high blood sugar with or without symptoms.

  • Procedural AKI was defined as an abrupt reduction in kidney function with an absolute increase in serum creatinine of more than or equal to 0.3 mg/dL, a percentage increase in serum creatinine of ≥50%, or a reduction in urine output (documented oliguria of <0.5 mL/kg per hour for >6 hours), within 7 days after contrast exposure.

  • Besides DM, five (5) predictors were in the final model. Three (3) predictors had a positive or direct association with AKI development: blood urea nitrogen to creatinine ratio, C-reactive protein and osteopontin). Two (2) predictors had an inverse association with AKI risk: CD5 antigen-like and Factor VII.

  • Among 217 patients with DM, 18 (8.3%) developed AKI

  • The final model had an Area Under the Receiver Operating Characteristic Curve (AUC) of 0.87 for predicting procedural AKI (P<0.001)

  • The optimal score cut-off had 100% sensitivity and a negative predictive value (NPV) of 100% for procedural AKI


  • In diabetic patients undergoing coronary angiography for various acute and non-acute indications, we describe a clinical and proteomics-supported biomarker model with high accuracy for predicting procedural AKI in patients undergoing coronary angiography. Earlier detection of AKI risk may allow us to mitigate risk by altering management approach in high risk patients.

Scroll Down for ADA 2019 Presentation for HART AKI

The ADA 2019 Presentation for HART AKI