Biomarker-based clinical subphenotypes of cardiac surgery associated acute kidney injury and their association with in-hospital treatments and outcomes: a data-driven cluster analysis

• Published in: European heart journal Vol. 44; no. Supplement_2
• Main Authors: Huang, X H, Zeng, J T, Su, X T, Zheng, Z
• Format: Journal Article
• Language: English
• Published: 09-11-2023
• ISSN: 0195-668X; 1522-9645
• DOI: 10.1093/eurheartj/ehad655.2741

Abstract
Background
Cardiac Surgery-Associated Acute Kidney Injury (CSA-AKI) is a prevalent and detrimental complication after cardiac surgery. Its etiology, pathophysiology, and clinical features are often obscure and heterogenous, with varying prognosis and therapeutic requirements. Cluster analysis has been increasingly used to profile complex diseases, but no prior studies have applied it to the CSA-AKI population.
Purpose
We aim to employ an unsupervised, data-driven cluster analysis approach utilizing routinely available and previously reported perioperative biomarkers to identify CSA-AKI subphenotypes that exhibit varying treatment requirements and in-hospital outcomes.
Methods
Patients diagnosed with CSA-AKI in accordance with the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines, who underwent coronary bypass grafting surgery in a single institution and had comprehensive perioperative data, were included (n=7 796). Cluster analysis based on 12 routinely available perioperative biomarkers was performed to partition CSA-AKI subphenotypes. Treatment requirements for vasoactive supports, diuretics, and dialysis, as well as AKI duration and in-hospital mortality rates, were compared among the identified subphenotypes.
Results
Three distinct clusters of patients with CSA-AKI were identified with different clinical characteristics, treatment needs, and in-hospital outcomes. Cluster 1(n=2 452, 31.5%) was characterized by normal NT-proBNP level, less comorbidities, predominantly isolated surgery, and better prognosis. Cluster 3 (n=1 060, 13.6%) was distinguished by extremely inflammatory states (significantly higher hsCRP level), more comorbidities, more complex surgery, and worse prognosis. Cluster 2 (n=4 284, 55.0%) was considered the intermediate cluster. CSA-AKI patients in cluster 3 exhibited significantly higher in-hospital mortality rate, longer AKI duration and hospital stay, and increased requirements for vasoactive supports, diuretics, and dialysis compared with the other two clusters. Notably, according to the current KDIGO criteria, 87.4% of cluster 3 patients are classified as only stage 1.
Conclusions
Three clinical subphenotypes of CSA-AKI were identified based on routine perioperative biomarkers, which demonstrated varying treatment requirements and in-hospital outcomes. This cluster analysis augments the current severity-based staging system for general AKI, providing additional risk stratification and treatment implications that are more tailored to cardiac surgery patients.