Development of a prediction model for acute kidney injury in critically ill patients with advanced colorectal cancer based on white blood cell-related indicators

Background Currently, most prediction models for acute kidney injury (AKI) in patients with colorectal cancer (CRC) are only applicable to postoperative situations, or rely on complex machine learning algorithms. There is currently no simple and feasible AKI prediction model for severely ill patients with advanced CRC. Objective This study endeavors to devise and validate a novel clinical risk prediction model utilizing white blood cell (WBC)-related indicators to forecast AKI in critically ill patients with advanced CRC. Methods A retrospective cohort study design was employed. The training cohort was derived from the eICU Collaborative Research Database, and the external validation cohort came from the Intensive Care Unit (ICU) of Wuhan Union Hospital. To screen for independent predictors of AKI and develop a new risk score model using multivariable logistic regression analysis. The assessment of model performance was conducted using receiver operating characteristic (ROC) curve analysis and compared with other scoring systems. The associations between different risk groups and survival outcomes as well as WBC counts were visualized using a heatmap. Kaplan-Meier survival analysis was used to compare in-hospital survival rates between the two groups. ROC curve analysis and decision curve analysis (DCA) served to evaluate the model’s predictive ability for in-hospital mortality. Subgroup analysis was applied to assess the consistency of this model across different critically ill cohorts with advanced CRC. Results A total of 981 critically ill patients with advanced CRC were included, of whom 185 (24.2%) from the eICU database and 59 (27.2%) from the Wuhan Union Hospital ICU developed AKI. A new risk model was successfully established: <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline" id="im1"> <mml:mrow> <mml:mn>1.271</mml:mn> <mml:mo>×</mml:mo> <mml:mi>W</mml:mi> <mml:mi>N</mml:mi> <mml:mi>R</mml:mi> <mml:mo>+</mml:mo> <mml:mn>0.709</mml:mn> <mml:mo>×</mml:mo> <mml:mi>L</mml:mi> <mml:mi>g</mml:mi> <mml:mrow> <mml:mo stretchy="false">(</mml:mo> <mml:mrow> <mml:mi>W</mml:mi> <mml:mi>E</mml:mi> <mml:mi>R</mml:mi> </mml:mrow> <mml:mo stretchy="false">)</mml:mo> </mml:mrow> <mml:mo>+</mml:mo> <mml:mn>0.085</mml:mn> <mml:mo>×</mml:mo> <mml:mtext>MLR</mml:mtext> <mml:mo>+</mml:mo> <mml:mn>1.225</mml:mn> <mml:mo>×</mml:mo> <mml:mtext>Lg</mml:mtext> <mml:mrow> <mml:mo stretchy="false">(</mml:mo> <mml:mrow> <mml:mtext>SII</mml:mtext> </mml:mrow> <mml:mo stretchy="false">)</mml:mo> </mml:mrow> </mml:mrow> </mml:math> . The model showed acceptable discrimination for predicting AKI, with area under the curve (AUC) of 0.746 (95% CI, 0.704-0.788) in the training set and 0.716 (95% CI, 0.633-0.798) in the validation set, and provided good clinical net benefit. For predicting in-hospital mortality, the AUCs were 0.788 (95% CI, 0.736-0.840) and 0.693 (95% CI, 0.559-0.827) in the training and validation sets, respectively. Conclusion This study developed a novel prediction model for AKI in critically ill patients with advanced CRC using WBC-related indicators. This model offers a simple, readily implementable tool for AKI risk stratification in critically ill patients with advanced CRC, supporting early clinical intervention and prognostic assessment.