**1. Introduction**

Circulating tumor cells (CTCs) are shed from the cancerous mass of the primary tumor cells and may ge<sup>t</sup> into the peripheral blood and then transfer to distant tissues causing expansion of the cancer in the metastasis form [1]. CTCs are a crucial source of genetic material for clinical analysis, e.g., tumor diagnostics, and selection and monitoring of cancer therapy [2–6]. Tissue biopsies are invasive and expensive, therefore the detection and characterization of tumors based on CTC analysis in a sample of peripheral blood, known as 'liquid biopsy', is the subject of special interest. A variety of techniques have been developed for isolation, capturing, and analysis of circulating tumor cells. The most widely used approaches include polymerase chain reaction (PCR) and reverse transcription PCR

(RT-PCR) [7], immunofluorescence [8–10], fluorescence-based cytometry [11], fluorescence scanning microscopy [12–14], and label-free biochemical separation [15]. Recently, the microfluidic devices based on miniaturized nanomaterials and microfluidic reactions improved the sensitivity of detection and enabled continuous single-cell analysis [16,17]. The current 'gold standard method' of CTC detection and enumeration is the CellSearch system (Veridex LLC) for the analysis of CTCs in metastatic breast [18], prostate [19], and colon cancer patients [20]. In this method ferromagnetic beads are coated with antibodies against the epithelial cellular adhesion molecule (EpCAM), then immunostained with fluorescently labeled anti-cytokeratin (CK, an epithelial intermediate filament), anti-CD45 (a membrane antigen expressed by leucocytes) antibodies, then stained with DAPI (4,6-diamidino-2-phenylindole, a nuclear stain), and finally counted by automated cell image capture and analysis even from a 7.5 mL blood sample [21].

However, it should also be highlighted that all the above described techniques require the use of specific molecular markers for CTCs detection and are time-consuming and/or expensive, which limits these methods for routine clinical analysis. For example, RT-PCR enables only the examination of a limited number of genes at the same time and does not permit the morphological analysis of cells in subsequent tests [22]. Moreover, the ability of this technique to detect multiple cancer markers might by hindered by the lack of appropriate tumor marker expression. In all the above mentioned methods, the outcomes have to be confirmed by histopathological examination of the tissue collected during biopsy. In order to reduce high medical costs of biopsy, we attempted to develop a new approach to cancer diagnosis based on the Raman technique.

Surface-enhanced Raman spectroscopy (SERS) is a highly sensitive and specific method that allows for the detection and characterization of various compounds through their capability to generate specific molecular fingerprint signals. SERS spectroscopy reveals the huge enhancement of Raman scattering signals of molecules adsorbed on specially prepared metallic nanostructures, usually made of silver, gold, or copper [23]. The two main mechanisms, electromagnetic (EM) and chemical (CT), can increase Raman peaks intensities by 9–14 orders of magnitude relative to normal Raman spectroscopy, which gives the possibility of single molecule detection [24]. The huge enhancement factor (huge sensitivity), high selectivity, possibility of label-free, rapid, and non-destructive analysis leads to an increase in the practical applications of this technique, especially in biomedical and analytical studies.

Recently, various studies have shown the capability of SERS in tumor cells identification [25–27]. Zhang et al. presented a novel strategy based on the nitrocellulose membrane and SERS imaging method, which can be used for both CTC enrichment and detection [28]. Jun et al. [29] developed silica-encapsulated magnetic nanoparticles (MNPs) with unique properties for cancer cell targeting and identification. Wang et al. used SERS-active nanoparticles modified with epidermal growth factor (EGF) peptide as a targeting ligand for efficient CTC detection in blood plasma [30]. Wen et al. successfully developed a method which has an ability to quickly respond and can be used for CTC capturing and detection. The CTC capturing and detection efficiency was also proved via real blood samples from clinical subjects using magnetic nanospheres [31]. Shi et al. [32] reported detection of cervical carcinoma (HeLa) cells using the designed folate-conjugated SERS-active nanoparticles and the magnetic tapping strategy. Krafft et al. [33] presented a microfluidic chip combined with optical tweezers to collect the normal Raman spectra of circulating cells.

Most of the presented SERS-based methods of CTC analysis require an enrichment step of a few CTCs from the blood and an extra labelling strategy for preparation, e.g., Raman reporter or peptides encoded NPs and/or tumor cells pre-labelled with NPs [29], which increases the cost and time of analysis.

Currently, various methods have been explored to optimize the properties of SERS platforms in terms of size, shape, and composition of used plasmonic nanostructures [34–37]. Even though many efficient SERS supports have been developed [38,39], there are some disadvantages associated with them, e.g., low stability over time, and inability to perform in situ measurements and sample mapping. To avoid these problems, new production strategies of SERS platforms should be developed in order to

introduce SERS techniques to standard biomedical and analytical applications. We have demonstrated that polymer mats covered via the PVD (physical vapor deposition) technique with gold or gold–silver alloy may work as a very efficient SERS platform for the identification of bacteria from environmental, food, and human body fluid samples [40,41].

In our approach we offer a simple method that is optimal for CTC isolation, enrichment, detection, and molecular analysis. We have elaborated the procedure of membrane-based SERS platforms (MBSP) with appropriate pore sizes that permit the separation and enrichment of the prostate cancer cell line (PC3), cervical carcinoma cell line (HeLa), and leucocytes—an example of healthy cells in the blood samples. By covering tailor-made SERS substrates with a thin layer of Ag–Au alloy, the high enhancement of Raman signals of all studied CTCs was achieved.

Additionally, to improve the efficiency of discrimination between cancer and blood cells, principal component analysis (PCA) was adopted. PCA is one of the most commonly used methods in the classification of SERS data, which enables the identification of the spectral differences among the studied samples, extracting the characteristic marker bands and biochemical information from SERS spectra. The presented strategy may lead to development of new tools in CTC therapy.
