**5. Conclusions**

Surface-enhanced Raman spectroscopy can be used as an ultrasensitive (at the level of single cell), non-invasive, rapid, and label-free method that provides valuable structural and biochemical analysis of circulating tumor cells. Moreover, Raman spectroscopy coupled with multivariate techniques gives the statistical diagnostic approach for efficient screening of cancer cells.

Our work clearly demonstrates the potential of such a PCA-based SERS method for the direct detection and identification of prostate cancer (PC3) and cervical carcinoma (HeLa) cell lines in blood samples with excellent specificity and sensitivity. We offer a single technology that is optimal for: (i) isolation of CTCs from blood samples, and their (ii) enrichment, (iii) detection, and (iv) molecular analysis.

The developed strategy is based on the tailor-made membrane-based SERS platforms, prepared according to a novel procedure, which permits at that same time the separation, immobilization/enrichment, and enhancement of the week Raman signals of CTCs.

The diagnostic sensitivity of 98% can be achieved for differentiation of PC3, HeLa, and normal cells.

The presented SERS-based strategy for circulating cell detection offers, besides such unique advantages as an ability for rapid and label-free recognition of CTCs with excellent sensitivity and selectivity, also simple sample preparation and cost-effective measurements. Our results indicate that a SERS-based cancer sensor has a grea<sup>t</sup> potential to be introduced in a variety of studies conducted on different types of cancer cells, especially from clinical samples.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2079-4991/9/3/366/s1. Figure S1: Reference SERS spectra of (a) HeLa, (b) PC3, and (c) leucocytes placed directly on SERS substrate, Figure S2: SERS spectra of (a) leucocytes, (b) HeLa, and (c) PC3 cells recorded from different spots (ca. 100) within the same sample. The excitation wavelength was at 785 nm, laser power was 5 mW, and the acquisition time was 60 seconds, Figure S3: (a) First three PC (PC-1, PC-2, and PC-3) scores plot of 3D-PCA calculated for narrow range (700–750 cm<sup>−</sup>1) and (b) loadings plot of the first principal component showing the most prominent marker band at 723 cm<sup>−</sup>1,Table S1: The sizes of blood components, based on data from Handin et al., Table S2: The RSD of the selected intensities of SERS signals of leucocytes, HeLa, and PC3 cells recorded from 1000 different spots within the same sample.

**Author Contributions:** Conceptualization, A.K. and T.S.; Methodology, K.N.; Software, K.N.; Validation, J.T.-D. and A.G.; Investigation, E.W., E.K.-G., W.S.; Data Curation, K.N.; Writing—Original Draft Preparation, A.K.; ´ Writing—Review & Editing, A.K., T.S.; Funding Acquisition, A.K.

**Funding:** This research was funded by National Science Centre under gran<sup>t</sup> UMO 2017/25/B/ST4/01109.

**Conflicts of Interest:** The authors declare no conflicts of interest.
