IARC 60th Anniversary - 19-21 May 2026
Session : 19/05/26 - Posters
Urinary Metabolite Differences Assessed by Gas Chromatography-Mass Spectrometry for The Detection of Clinically Significant Prostate Cancer
LADUKAS A. 1,2, DEZA I. 3, DRABINSKA N. 5, COSTELLO B. 4, REYNOLDS J. 6, SWIFT J. 6, BAHL A. 7, PERSAD R. 8, SUZIEDELIS K. 9,11, KRISTIJONAS V. 10,11, SABALIAUSKAITE R. 12, SMAILYTE G. 1,2, PATASIUS A. 1,2, RATCLIFFE N. 4
1 Department of Public Health, Institute of Health Sciences, Faculty of Medicine, Vilnius University, Vilnius, Lithuania; 2 Laboratory of Cancer Epidemiology, National Cancer Institute, Vilnius, Lithuania; 3 Department of Computer Science and Creative Technologies, University of the West of England, Bristol, United Kingdom; 4 Department of Applied Sciences, University of the West of England, Bristol, United Kingdom; 5 Food Volatilomics and Sensomics Group, Department of Food Technology of Plant Origin, Faculty of Food Science and Nutrition, Poznan University of Life Sciences, Poznan, Poland; 6 Department of Chemistry, Loughborough University, Loughborough, United Kingdom; 7 Bristol Haematology and Oncology Centre, University Hospitals Bristol and Weston NHS Trust, Bristol, United Kingdom; 8 North Bristol Hospital Trust, Bristol, United Kingdom; 9 Life Sciences Centre, Institute of Biosciences, Department of Biochemistry and Molecular Biology, Vilnius University, Vilnius, Lithuania; 10 Life Sciences Centre, Institute of Biotechnology, Department of Biological DNA Modification, Vilnius University, Vilnius, Lithuania; 11 Laboratory of Molecular Oncology, National Cancer Institute, Vilnius, Lithuania; 12 Laboratory of Genetic Diagnostics, National Cancer Institute, Vilnius, Lithuania
Background
Use of prostate specific antigen (PSA) for the early prostate cancer detection has led to an increased number of prostate biopsies and potentially overdiagnosis and overtreatment of indolent prostate cancer. The lack of sensitive and specific biomarkers for the detection of clinically significant prostate cancer (csPC) is an important public health and clinical challenge.
The ability of canines to diagnose cancer by smelling volatile organic compounds (VOCs) in urine has motivated the application of spectroscopic methods for prostate cancer detection by identifying urinary metabolites.
Objectives
To evaluate the ability of spectroscopic methods to detect prostate cancer and clinically significant prostate cancer among individuals undergoing prostate biopsy due to elevated PSA levels, by detection of urinary VOCs.
Methods
184 men undergoing transrectal systematic prostate biopsy due to elevated PSA levels or suspicious digital rectal examination were enrolled to the study in one specialized cancer centre. Urine samples were collected before biopsy procedure and stored in cryotubes for further analysis.
We used solvent extraction, a little used method for VOC analyses, coupled with Gas Chromatography-Mass Spectrometry (GCMS) and acid treatment of urine to enhance biomarker discovery.
We assessed GCMS as a tool to look for differences in the metabolite fingerprint, using data processing/machine learning. The methodology we have developed—systematic preprocessing evaluation combined with rigorous feature selection and multi-model validation—provides a robust template for developing metabolomics classifiers for cancer types and disease contexts.
Results
A total of 184 patients underwent a prostate biopsy. Median PSA value was 7.22 ng/ml (range 1.24 – 4602 ng/ml). Benign pathology was reported in 72 (39.13%) cases while cancer was found in 112 (60.87%) cases. ISUP 1, ISUP 2, ISUP 3, ISUP 4 and ISUP 5 cancer was reported in 49 (26.6%), 34 (18.5%), 15 (8.2%), 3 (1.6%) and 11 (5.9%) cases, respectively. When defined as ISUP2 or higher, clinically significant prostate cancer was found in 63 (34.2%) individuals.
Our model, combining discriminative features of GCMS spectrum and PSA, achieved diagnostic accuracy of AUC = 0.842 (95% CI: 0.825 – 0.859) to differentiate between csPCa and the rest (benign histology + low grade (ISUP1) prostate cancer), representing performance that could translate into meaningful clinical impact.
Conclusion
The broad distribution of discriminative features across the GCMS spectrum, coupled with the requirement for high spectral resolution, suggests that cancer detection relies on subtle, coordinated changes in metabolite environments rather than dramatic concentration changes in individual biomarkers.
Inspired by the findings reported here, we are continuing research and development to strengthen the results in other, non-high-risk population.