# From proteomics to colloidal gold tests for urinary thrombomodulin: a prospective cohort study on accurate sepsis screening

## TL;DR

Researchers developed a non-invasive urine test using colloidal gold strips to detect thrombomodulin for accurate sepsis screening.

## Contribution

A novel colloidal gold test strip for sepsis screening based on urinary thrombomodulin levels was developed and validated.

## Key findings

- Urinary thrombomodulin (TM) levels decrease with increasing sepsis severity, while blood TM levels increase.
- A urinary TM threshold of 15.46 TU/mL achieved 57% sensitivity and 88% specificity for sepsis diagnosis.
- The colloidal gold test strip demonstrated 81.5% overall accuracy in sepsis diagnosis, comparable to immunofluorescence assays.

## Abstract

To develop a new non-invasive screening method for sepsis by detecting urine samples.

A prospective study was conducted to collect urine samples from a cohort of 22 individuals diagnosed with sepsis and admitted to the Intensive Care Unit (ICU) of a university-affiliated teaching hospital in China. Utilizing proteomic and bioinformatics analyses, we sought to identify potential biomarkers indicative of sepsis. These biomarkers were subsequently validated using serum and urine samples from 31 patients with septic shock, 83 patients with sepsis, and 50 healthy controls. Receiver operating characteristic (ROC) curves were employed to determine the optimal cutoff values for these biomarkers. Based on the diagnostic thresholds derived from ROC analysis, colloidal gold test strips were developed and applied to screen a cohort of 92 ICU patients. The diagnostic accuracy of these test strips was rigorously assessed by comparing their results with those from immunofluorescence assays.

Data-independent acquisition (DIA) proteomics analysis of urine samples identified 2,846 proteins, with stringent filtration criteria (fold change > 2 or < 0.5, P-value < 0.05) yielding 178 differentially expressed proteins (DEPs). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis revealed significant enrichment of DEPs in pathways associated with “cell adhesion molecules,” “lysosomes,” and metabolic processes. The Boruta algorithm, integrating Random Forest and Support Vector Machine (SVM) analysis, identified urinary thrombomodulin (TM) as a key candidate molecule. Immunofluorescence analysis for validation trial showed rising trend in blood TM levels across disease severities: 7.55 (6.58-8.72) TU/mL in healthy controls, 10.08 (8.00-14.15) TU/mL in general sepsis, and 12.30 (7.54-18.68) TU/mL in septic shock. Conversely, urinary TM levels decreased: 23.65 (18.08-31.06) TU/mL, 17.70 (13.80-28.80) TU/mL, and 5.84 (4.00-11.59) TU/mL, respectively. At a urinary TM threshold of 15.46 TU/mL, the ROC AUC for sepsis diagnosis is 0.72, with 57% sensitivity and 88% specificity (P<0.05), showing no significant difference comparable to blood TM (P>0.05). For septic shock diagnosis and 28-day mortality prediction, a urinary TM threshold of 11.85 TU/mL yields an ROC AUC of 0.92, with 93% sensitivity and 81% specificity, outperforming blood TM (P<0.05). A urinary TM colloidal gold test strip, which turns red at TM levels above 15.46 TU/mL, was developed and validated on urine samples from 43 sepsis and 49 non-sepsis patients, achieving 86.1% sensitivity, 77.6% specificity and an overall accuracy of 81.5% for sepsis diagnosis. The Kappa test validated the concordance of the colloidal gold strip test with Sepsis 3.0 diagnostic criteria, while the McNemar test indicated no significant difference in sepsis diagnosis efficacy between the strip test and chemiluminescent immunofluorescence (p=0.228).

The utilization of urine test strips for the detection of TM offers a precise, convenient, and practical method for the screening of sepsis.