Lactate Dehydrogenase, Interleukin-6, and 24 h Intra-Abdominal Pressure Exhibit Early Predictive Value in...

Hypertriglyceridemia-induced acute pancreatitis (HTG-AP) is increasingly recognized as a major cause of acute pancreatitis in China. Acute respiratory distress syndrome (ARDS), a severe and early complication, significantly worsens prognosis and increases intensive care unit (ICU) burden. This study aimed to identify early predictive markers of ARDS in patients with HTG-AP, enabling timely risk stratification and clinical intervention to improve outcomes.

A retrospective analysis was conducted on 111 patients who were diagnosed with HTG-AP and admitted to the emergency ICU (EICU) between August 2022 and December 2024. Patients were categorized into ARDS (n = 61) and non-ARDS (n = 50) groups according to the complications associated with ARDS throughout the progression of the illness. Clinical variables were compared between groups. The Boruta technique was used to choose predictors, and then multivariate logistic regression was used. To assess prediction ability, receiver operating characteristic (ROC) curve analysis was carried out. We performed internal validation using five-fold cross-validation and bootstrap resampling (1000 iterations) to evaluate the model's capacity for generalization and stability. The clinical utility of the model was assessed using decision curve analysis (DCA).

Patients in the ARDS group had lower serum calcium (Ca²⁺) levels (p < 0.05) and longer ICU stays, as well as greater 24 h intra-abdominal pressure (IAP), interleukin-6 (IL-6), lactate dehydrogenase (LDH), and lactate (Lac) levels (p < 0.05) compared to the non-ARDS group. NEU, Lac, Cr, 24 h IAP, IL-6, and LDH are important possible risk factors, according to the Boruta algorithm. Multivariate logistic regression analysis identified 24 h IAP, IL-6, and LDH as independent predictors of concurrent ARDS (p < 0.05). The areas under the ROC curves (AUC) for 24 h IAP, IL-6, and LDH were 0.796 (95% CI: 0.715-0.877), 0.770 (95% CI: 0.683-0.858), and 0.868 (95% CI: 0.880-0.936), respectively (p < 0.05). When combined, the predictive model yielded an AUC of 0.907 (95% CI: 0.849-0.957), with a specificity of 90.00% and a sensitivity of 78.70%, and demonstrated noninferior predictive performance to the existing scoring systems. With an average AUROC score of 0.890 from five-fold cross-validation and a bootstrap-corrected C-index of 0.894, the model shows good discriminatory power. The model's clinical net benefit was verified by the DCA. Additionally, we have created a free online risk calculator for the general public (https://ardsprediction.streamlit.app/).

LDH, IL-6, and 24 h IAP are valuable early indicators of ARDS development in patients with HTG-AP. The integrated predictive model demonstrates excellent diagnostic performance and can support early clinical decision-making.