Exploring the diagnostic potential of chemometric fingerprinting from FFF-sorted biological samples to track and detect exosomal population in cancer patients’ serum

There is a growing interest raised by extracellular vesicles (EVs) both as diagnostic analytes and therapeutic vessels. Consequently, many techniques and approaches are exploited to collect EVs from various sources (cell media or human blood serum, saliva and other body fluids) in order to explore the potentialities of non-invasive liquid biopsies, improve diagnosis, and understand the biological function and synergistic action on EVs in cellular communication. The drawbacks of these approaches reside in the many steps required for EV isolation, the scarce throughput, and the complexity of EV characterization, by means of labelling, mass spectrometry and immunoassays (ELISA). Field flow fractionation (FFF), a soft separation technique able to work in physiological conditions and sort particles from complex samples according to hydrodynamic size, was previously employed to successfully isolate EVs from cell culture medium, though a pre-concentration step was required. More recent advancements of our FFF-based EV isolation protocol allowed to also process biological samples such as human blood serum without any pretreatment (e.g. ultrafiltration): this is a crucial advancement in EV research since it would cut down cost and limit sample manipulation. Preliminary results showed that it is possible, through FFF separation, to collect EV-enriched fractions to undergo ELISA characterization. This work, involving samples from patients with colorectal cancer (CRC), put the basis for the fast isolation and identification of diagnostic markers avoiding invasive procedures. The advantages of using an FFF platform to isolate EVs also include the chance of characterizing the samples by the means of online detection prior to fraction collection, obtaining multidimensional information which contains UV absorption and fluorescence signals relative to the species eluted during the process, together with mass/size characterization obtained by multiangle light scattering. With the aim of exploring the potential of such information, a chemometric approach based on visualization and classification techniques was carried out. The feasibility study tackled the complexity of the data sets obtained and pointed out the useful descriptors for the different EV. At the same time useful information to forecast and direct EV enrichment for maximizing information from the sample and reducing invasive medical procedures, were obtained.