Biomarker-Based Risk Stratification in Pediatric Sepsis From a Low-Middle Income Country

Abstract

Objectives: Most biomarker studies of sepsis originate from high-income countries, whereas mortality risk is higher in low- and middle-income countries. The second version of the Pediatric Sepsis Biomarker Risk Model (PERSEVERE-II) has been validated in multiple North American PICUs for prognosis. Given differences in epidemiology, we assessed the performance of PERSEVERE-II in septic children from Pakistan, a low-middle income country. Due to uncertainty regarding how well PERSEVERE-II would perform, we also assessed the utility of other select biomarkers reflecting endotheliopathy, coagulopathy, and lung injury.

Design: Prospective cohort study.

Setting: PICU in Aga Khan University Hospital in Karachi, Pakistan.

Patients: Children (< 18 yr old) meeting pediatric modifications of adult Sepsis-3 criteria between November 2020 and February 2022 were eligible.

Interventions: None.

Measurements and main results: Plasma was collected within 24 hours of admission and biomarkers quantified. The area under the receiver operating characteristic curve for PERSEVERE-II to discriminate 28-day mortality was determined. Additional biomarkers were compared between survivors and nonsurvivors and between subjects with and without acute respiratory distress syndrome. In 86 subjects (20 nonsurvivors, 23%), PERSEVERE-II discriminated mortality (area under the receiver operating characteristic curve, 0.83; 95% CI, 0.72-0.94) and stratified the cohort into low-, medium-, and high-risk of mortality. Biomarkers reflecting endotheliopathy (angiopoietin 2, intracellular adhesion molecule 1) increased across worsening risk strata. Angiopoietin 2, soluble thrombomodulin, and plasminogen activator inhibitor 1 were higher in nonsurvivors, and soluble receptor for advanced glycation end-products and surfactant protein D were higher in children meeting acute respiratory distress syndrome criteria.

Conclusions: PERSEVERE-II performs well in septic children from Aga Khan University Hospital, representing the first validation of PERSEVERE-II in a low-middle income country. Patients possessed a biomarker profile comparable to that of sepsis from high-income countries, suggesting that biomarker-based enrichment strategies may be effective in this setting.

Figure 1.

Patient flowchart. ARDS = acute respiratory distress syndrome.

Figure 2.

Classification and regression tree-based Pediatric Sepsis Biomarker Risk Model (PERSEVERE)-II model stratifying septic children into one of seven terminal nodes (TNs). All subjects start at the root node at the top and subsequently stratified according to biomarker levels into TNs. TNs 1, 2, and 5 (green) are low-risk of 28-d mortality. TNs 4 and 6 (blue) are medium-risk. TNs 3 and 7 (red) are high-risk). For comparison, the color-coded low-, medium-, and high-risk TNs are identical to those identified using the PERSEVERE-II model in children from the United States, albeit with lower mortality rates in the original cohort (low-risk < 2%, medium-risk 15–20%, high-risk > 40% predicted mortality risk). TN7 in this cohort from Aga Khan University Hospital is pruned from the original PERSEVERE-II due to only having two subjects. CCL3 = C-C motif chemokine ligand 3, GI = gastrointestinal, HSPA1B = heat shock protein 72, IL8 = interleukin-8, MMP8 = matrix metalloproteinase 8.

Figure 3.

Kaplan-Meier survival curves for subjects stratified into low- (green), medium- (blue), and high-risk (red) Pediatric Sepsis Biomarker Risk Model-II strata. Overall log-rank p < 0.001. Pairwise comparisons (low-risk vs medium-risk p = 0.003; medium-risk vs high-risk p = 0.008; low-risk vs high-risk p < 0.001) are also significant.