GC-MS Urine Analysis Detects Bladder Cancer

Bladder cancer is the 11th most common cancer in the United Kingdom, with around 10,500 new cases each year. Doctors usually diagnose and monitor it using cystoscopy, a procedure that has costly, slow, and unpleasant for patients. These drawbacks have pushed scientists to look for easier ways to spot the disease, especially through urine, since it is in constant contact with tumor tissue and might carry telltale chemical clues. To explore this, UK researchers took urine samples, extracted compounds from them using solvents, and then ran the samples through gas chromatography-mass spectrometry (GC-MS), essentially a way of sniffing out and identifying different chemicals in a sample. Their goal was to find specific substances that could act as warning signs for bladder cancer. The findings were published in the journal BJC Reports.

Why Might Traditional Methods for Finding Disease Biomarkers Fall Short Compared to Newer Approaches?

Over the past several years, scientists have become more interested in using naturally occurring chemical byproducts in the body (volatile organic compounds [VOCs]) to help diagnose disease, an approach referred to as volatilomics.^2-5 Many of these compounds are simply leftovers from the body's everyday metabolism, and they show up in bodily fluids like urine in different amounts. So far, researchers have identified 444 different VOCs in the urine of healthy people, showing just how many compounds might potentially be tied to different health conditions.⁶

Traditionally, scientists have looked for disease markers by comparing one compound at a time between groups (say, cancer patients versus healthy people) and flagging the ones that show a big difference. But this approach assumes that a disease will show up as one or two compounds behaving in a dramatically different way, which is not necessarily how things actually work in the body.^7,8 “This approach, the authors of the article point out,¹ “underpins most prior VOC studies, which report specific biomarkers with differing concentrations in cancer patients. Machine learning methods can model complex non-linear interactions between multiple metabolites, potentially revealing coordinated metabolic patterns that are invisible to univariate analysis.”

“Many prior studies, the authors continue,¹ “employed solid phase microextraction (SPME) and univariate or simple multivariate statistical methods to identify individual biomarkers. However, analytical challenges associated with SPME may limit compound detection, whilst univariate statistical approaches cannot detect diagnostic signals residing in coordinated patterns across multiple metabolites. None of these prior studies used liquid-liquid extraction (LLE) for VOC extraction or comprehensively compared classical statistical approaches with machine learning methods capable of modelling metabolite interactions.”

Can Urine-Based Chemical Testing Combined with Machine Learning Accurately Detect Bladder Cancer?

For this study, researchers collected urine from 100 people (50 with bladder cancer and 50 without) and tested each sample using GC-MS to look for compounds that might signal cancer. They then analyzed the results in two ways: with traditional statistics, and with machine learning, where a computer program learns to spot patterns in the data on its own.¹

They tried five different machine learning approaches and used a method that narrows things down to the most useful combination of compounds. The best-performing approach, the algorithm XGBoost, was noticeably better at telling cancer patients from healthy people than the traditional statistical method (essentially a stronger, more reliable score for how well it could tell the two groups apart).¹

Using a set of eight specific compounds, the test correctly identified cancer or non-cancer cases about 85% of the time, striking a balance between catching real cases and avoiding false alarms. If the settings were adjusted to prioritize catching as many cancer cases as possible (better suited for a broad screening test) it could correctly flag 95% of actual cancer cases, though at the cost of more false alarms in healthy people (correctly clearing 70% of them).¹

“The findings,” write the authors of the paper,¹ “indicate that solvent extraction of urine shows promise for isolating putative biomarkers of bladder cancer. Employing machine learning achieved diagnostic accuracy potentially suitable for clinical deployment as a non-invasive bladder cancer detection tool.”

Read More on Similar Topics
Detecting Estrogen Signatures in MCN Tumors via Targeted LC-MS/MS
FDA Grants Breakthrough Status to AI-Enhanced GC-MS Device for Bladder Cancer Detection

References

1. Deza, I.; de Lacy Costello, B.; Drabińska-Fois, N. et al. Urinary Volatilomics Using Liquid-Liquid Extraction and Gas Chromatography-Mass Spectrometry (GC-MS) Combined with Machine Learning Algorithms as a Tool for Diagnosis and Surveillance of Urothelial Bladder Cancer. BJC Rep. 2026, 4 (1), 42. DOI: 10.1038/s44276-026-00244-8
2. Opitz, P.; Herbarth, O. The Volatilome – Investigation of Volatile Organic Metabolites (VOM) as Potential Tumor Markers in Patients with Head and Neck Squamous Cell Carcinoma (HNSCC). J Otolaryngol Head Neck Surg. 2018, 47. DOI: 10.1186/s40463-018-0288-5
3. Taunk, K.; Taware, R.; More, T. H. et al. A Non-Invasive Approach to Explore the Discriminatory Potential of the Urinary Volatilome of Invasive Ductal Carcinoma of the Breast. RSC Adv. 2018, 8 (44), 25040-25050. DOI: 10.1039/c8ra02083c
4. Broza, Y. Y.; Mochalski, P.; Ruzsanyi, V. et al. Hybrid Volatolomics and Disease Detection. Angew Chem Int Ed. 2015, 54, 11036–11048. DOI: 10.1002/anie.201500153
5. Barash, O.; Zhang, W.; Halpern, J. M. et al. Differentiation Between Genetic Mutations of Breast Cancer by Breath Volatolomics. Oncotarget 2015, 6 (42), 44864-44876. DOI: 10.18632/oncotarget.6269
6. Drabińska, N.; Flynn, C.; Ratcliffe, N. et al. A Literature Survey of All Volatiles from Healthy Human Breath and Bodily Fluids: The Human Volatilome. J Breath Res. 2021, 15, 34001. DOI: 10.1088/1752-7163/abf1d0
7. Lett, L.; George, M.; Slater, R. et al. Investigation of Urinary Volatile Organic Compounds as Novel Diagnostic and Surveillance Biomarkers of Bladder Cancer. Br J Cancer 2022. DOI: 10.1038/s41416-022-01785-8
8. Khalid, T.; White, P.; De Lacy Costello, B. et al. A Pilot Study Combining a GC-Sensor Device with a Statistical Model for the Identification of Bladder Cancer from Urine Headspace. PLoS One 2013, 8. DOI: 10.1371/journal.pone.0069602