222 Chapter 8 8.4.5 Circulating proteins The combination of protein biomarkers with nucleic acid signatures could also improve test sensitivity by leveraging their complementarity. An example of a blood-based multi-cancer detection test relying on the presence of both cancer-specific mutations and overexpressed protein levels is the CancerSEEK test (14). The urine proteome is considered to be less complex as compared to blood, due to glomerular reabsorption of most proteins. Urinary protein biomarkers have been explored for the detection of varying cancer types (106), including endometrial (107-109), lung (110), and ovarian (108, 111) cancer. One potential drawback is the dynamic nature of protein concentrations in urine, which may present comparable challenges to those encountered with urine cfDNA (106). 8.4.6 Circulating metabolites Changes in metabolites that reflect reprogramming of cellular metabolism, a core hallmark of cancer, are also attractive biomarkers beyond the (epi)genomic landscape (112). Glycosaminoglycans are involved in cancer development and represent a promising class of tumor metabolism biomarkers. Free glycosaminoglycan profiles can be accurately quantified from both plasma and urine for the detection of multiple cancer types (113, 114). 8.4.7 Exfoliated tumor cells Urine cytology offers an easily applicable and affordable method to detect cancer types that are known to exfoliate cancerous cells into the lower genital tract, such as bladder, cervical, or endometrial cancer (115). The routine collection and cytological examination of urine for bladder cancer detection is an established method in diagnostic laboratories. As extensively studied for circulating tumor cells in plasma, in-depth molecular characterization of exfoliated cancer cells in the urine using innovative single-cell sequencing technologies would offer a valuable approach for cancer diagnosis and clinical decision-making (116). Taken together, integrating different biomarker classes poses a promising approach to improving the performance of cancer detection methods. The detection of early-stage cancers or cancers that shed limited amounts of cfDNA could particularly be improved using biomarkers independent of the (epi)genetic landscape. Combining multiple biomarker signatures retrieved from large datasets requires complex computational methods, such as machine learning algorithms or neural networks. Large patient cohorts are required to accurately train and validate multi-dimensional molecular classifiers for cancer detection (44). Moreover, standardization of pre-analytics is crucial for accurate and reliable biomarker analysis and should be streamlined across different centers (117).
RkJQdWJsaXNoZXIy MjY0ODMw