OC 04.5 - Successful development of an electronic nose utilizing blood and urine metabolomics for the prediction of lower extremity deep venous thrombosis

Saturday, June 22, 2024

14:00 – 14:15 ICT

Room: Ballroom B4

Presenting Author(s)

Gang Wang, PhD

Professor
The Second Affiliated Hospital of Xi’an Jiaotong University, Xi’an, China.
Xi’an, Shaanxi, China (People's Republic)

Co-Author(s)

XX

Xi Xu

PhD student
The Second Affiliated Hospital of Xi’an Jiaotong University, Xi’an, China.
Xi’an, Shaanxi, China (People's Republic)
JL

Jiamei Li

Lecturer
The Second Affiliated Hospital of Xi’an Jiaotong University, Xi’an, China.
Xi’an, Shaanxi, China (People's Republic)
XJ

Xuting Jin

MD student
The Second Affiliated Hospital of Xi’an Jiaotong University, Xi’an, China.
Xi’an, Shaanxi, China (People's Republic)
YH

Yanli Hou

Attending Doctor
The Second Affiliated Hospital of Xi’an Jiaotong University, Xi’an, China.
Xi’an, Shaanxi, China (People's Republic)
JS

Jing Shi

Resident
The Second Affiliated Hospital of Xi’an Jiaotong University, Xi’an, China.
Xi’an, Shaanxi, China (People's Republic)
WG

Wenyi Gao

Resident
The Second Affiliated Hospital of Xi’an Jiaotong University, Xi’an, China.
Xi’an, Shaanxi, China (People's Republic)

Background: The incidence of lower extremity deep venous thrombosis (LEDVT) is particularly high in intensive care units (ICU), especially among critically ill patients with hypercoagulable states caused by tumors, infections, etc. Early identification of LEDVT individuals is crucial to prevent further development, even into pulmonary embolism and shock. However, current predictive methods such as pre-test probability scores and D-dimer tests lack specificity and sensitivity. Volatile organic compounds (VOCs) in blood and urine reflect disease-related metabolic changes, showing promise in LEDVT prediction. Nevertheless, the correlation between VOCs and LEDVT remains unexplored.

Aims: This study aims to identify the VOCs from high-risk patients and develop an electronic nose (E-Nose) for the early prediction of LEDVT.

Methods: VOCs in blood and urine of high-risk LEDVT patients in ICU were analyzed daily using gas chromatography-mass spectrometry. Ultrasound examinations were conducted to confirm LEDVT occurrence. VOCs analyses were terminated upon ultrasound confirmation, including a total of 78 LEDVT patients. Differential analysis was performed on VOCs collected within 24 hours of admission and 24 hours before LEDVT onset. Identified VOCs were utilized to develop an E-Nose and early LEDVT prediction model. Model validation was conducted using an external dataset of VOCs from 59 high-risk LEDVT patients.

Results: Three VOCs (octane, acetaldehyde, and ethylenediamine) in blood and two VOCs (nonanal and decanal) in urine were identified. The blood-based prediction model achieved an area under the receiver operating characteristic curves (AUC) of 0.98, with 90.56% accuracy, 90.27% sensitivity, and 84.44% specificity. The urine-based model achieved an AUC of 0.97, with 94.33% accuracy, 98.25% sensitivity, and 83.4% specificity. Both models achieved high prediction accuracy in external validation sets.

Conclusion(s): The E-Nose based on VOCs in blood and urine shows potential as a novel, rapid, and reliable tool for early LEDVT prediction.