Title : Comparative prognostic performance of lactate–proenkephalin, SOFA–proenkephalin, and triple combination models for early mortality prediction in septic critically ill patients

Abstract:

Background: Early mortality prediction in sepsis remains challenging because of the complex interaction between tissue hypoperfusion, organ dysfunction, and systemic inflammation. Conventional prognostic tools such as serum lactate and Sequential Organ Failure Assessment (SOFA) score are widely used in critically ill patients, while proenkephalin (PENK) has recently emerged as a promising biomarker associated with renal dysfunction and adverse outcomes. Combining biomarkers and physiologic scoring systems reflecting different pathophysiologic mechanisms may improve early risk stratification in sepsis.

Objective: To compare the prognostic performance of lactate–proenkephalin, SOFA–proenkephalin, and lactate–SOFA–proenkephalin combination models for predicting 7-day mortality in septic critically ill patients. 

Methods: We conducted a single-center retrospective observational cohort study involving adult septic patients admitted to the intensive care unit (ICU) and high care unit (HCU) of a tertiary referral hospital in Indonesia. Baseline serum lactate, SOFA score, and proenkephalin (PENK H1) levels were obtained within the first 24 hours of admission. The primary outcome was all-cause 7-day mortality.

Receiver operating characteristic (ROC) curve analysis was performed to evaluate discriminative performance. Logistic regression models were constructed to generate combined prediction models. Comparative ROC analysis was used to compare the prognostic accuracy of individual biomarkers and combination models.

Results: A total of 98 septic critically ill patients were included, of whom 47 (48.0%) died within 7 days. Individually, lactate demonstrated the highest predictive performance with an area under the curve (AUC) of 0.673 (95% CI 0.566–0.780), followed by proenkephalin with an AUC of 0.653 (95% CI 0.545–0.761) and SOFA score with an AUC of 0.625 (95% CI 0.514–0.736).

The lactate–proenkephalin combination model demonstrated improved prognostic performance with an estimated AUC of 0.761 (95% CI 0.664–0.858), outperforming either biomarker alone. The SOFA–proenkephalin combination model also improved discriminative ability compared with SOFA score alone, with an estimated AUC of 0.724 (95% CI 0.621–0.827).

Among all evaluated models, the triple combination model integrating lactate, SOFA score, and proenkephalin demonstrated the best overall prognostic performance, with an estimated AUC of 0.812 (95% CI 0.724–0.900), indicating good discriminative ability for predicting 7-day mortality.

In multivariable logistic regression analysis, lactate (p=0.030) and proenkephalin (p=0.048) remained independently associated with mortality, whereas SOFA score was not independently associated after adjustment (p=0.193).

Conclusion: Combination models integrating biomarkers and physiologic scoring systems improved early mortality prediction in septic critically ill patients. The triple combination model consisting of lactate, SOFA score, and proenkephalin demonstrated the best overall prognostic performance, suggesting that integrating markers of tissue hypoperfusion, organ dysfunction, and renal injury may enhance early risk stratification in sepsis. These findings support the potential role of multimodal prognostic approaches, particularly in resource-limited critical care settings.

Keywords: Sepsis; Proenkephalin; Lactate; SOFA Score; Biomarker Combination; Mortality Prediction; Critical Care