Search for a command to run...
We read with interest the prospective longitudinal study by van Liere et al. on urinary volatile organic compounds (VOCs) for detecting colorectal neoplasia in Lynch syndrome (LS).1 This investigation addresses a critical unmet need, that is, optimizing surveillance in a population with up to 70% lifetime colorectal cancer (CRC) risk, for whom biannual colonoscopy imposes both clinical and psychological burdens.2 The study's methodological rigor, that is, multi-platform VOC profiling [gas chromatography-ion mobility spectrometry (GC-IMS), field asymmetric ion mobility spectrometry (FAIMS), gas chromatography–time-of-flight mass spectrometry (GC-TOF-MS)], blinded analyses, and machine learning models (MLM), sets an important benchmark. The observation that urinary VOC signatures normalize following polypectomy (AUC 0.84 for GC-IMS) is compelling, indicating potential application in personalized, risk-stratified follow-up and, more broadly, in monitoring intervention efficacy. The identification of decanoic acid as a discriminatory metabolite is concordant with accumulating evidence that dysregulated lipid metabolism underpins the earliest stages of CRC. Importantly, the open release of R code for GC-IMS feature selection directly addresses reproducibility, a persistent barrier in biomarker discovery, and provides a transparent framework for validation across independent cohorts. Nonetheless, the diagnostic performance remains modest (65% sensitivity, 70% specificity), markedly lower than faecal VOCs3 reported in the same cohort (95% sensitivity). This discrepancy may reflect issues of dilution and metabolite stability in urine, as well as confounders such as diet, age, and smoking status. Moreover, the limited number of advanced lesions constrains extrapolation to clinically meaningful endpoints such as advanced adenoma or CRC. These limitations typify the translational “valley of death” that continues to challenge the development of single-platform biomarkers. Within the broader biomarker landscape, Potievskaya et al. recently underscored that no single class, whether circulating tumour DNA, exosomes, VOCs, or metabolomics, has yet demonstrated the reproducibility and diagnostic power required for stand-alone CRC detection.4 In parallel, Seum et al.5 emphasized the promise of pre-diagnostic metabolomics, particularly lipid-based panels, which have achieved AUCs of up to 0.95 for CRC precursors in prospective cohorts. Most recently, Abu Bakar et al. systematically reviewed serum metabolite biomarkers for colorectal adenomas (CRA), identifying discriminatory candidates such as benzoic acid, acetate, and lactate (CRA vs. normal), and adenosine, pentothenate, and linoleic acid (CRA vs. CRC), with reported AUCs ranging from 0.7 to >0.9,6 Table 1. Nonetheless, even these serum markers were limited by poor reproducibility and methodological heterogeneity, reflecting recurrent challenges across the biomarker discovery field. We believe that urinary VOCs may find their optimal role not as a solitary test7 but as part of integrated, patient-friendly biomarker panels. Combining VOCs with serum metabolites, polygenic risk scores, metabolomic lipid panels, or faecal immunochemical testing (FIT) could enhance specificity without sacrificing accessibility.8 Moreover, embedding such assays within prospective, multicenter Lynch syndrome registries would allow validation across diverse diets, lifestyles, and geographies, all critical for translation. In sum, van Liere et al. advance the field of volatilomics by demonstrating feasibility, biological plausibility, and transparency in LS surveillance. Their study complements ongoing work in serum and metabolomic biomarkers, reinforcing the view that progress towards precision prevention will require harmonized, multi-omic strategies. Anastasios Koulaouzidis: Conceptualization; supervision; writing – original draft; writing – review and editing. Wojciech Marlicz: Supervision. Ramesh P. Arasaradnam: Conceptualization; writing – review and editing. Artificial intelligence tools (ChatGPT5, OpenAI, San Francisco, CA, USA) were used solely to assist with language editing and style refinement during manuscript preparation. No AI-generated data or figures were included. All scientific content, interpretations, and conclusions were developed, verified, and approved entirely by the authors. Wojciech Marlicz is shareholder and co-founder of Sanprobi sp. z o.o. spk and Endoklinika sp. z o.o. Anastasios Koulaouzidis is a co-founder and shareholder of AJM Med-i-caps Ltd.; consultant for Jinshan Science & Technology Ltd.; recipient of research grants from Given Imaging Ltd. (via ESGE) and IntroMedic and a former co-founder and director of iCERV Ltd. Anastasios Koulaouzidis also received lecture honoraria from Covidien/Medtronic, Jinshan, Dr. Falk Pharma UK, Ferring, Aquilant, Almiral and has advisory roles for Tillots, Ankon, and Dr. Falk Pharma UK. Ramesh P. Arasaradnam declares no conflict of interest.