Summary:

Liquid biopsy is the term used to describe the use of peripheral blood as a source of cancer diagnosis. This minimally invasive technique targets biological samples as tumour signatures in the blood, such as circulating tumour cells (CTCs). Since CTCs are heterogeneous and rare, detecting them is a major challenge. As CTCs undergo mechanical stress during the process of metastasis, studying the response of the cell and its subcellular elements to different mechanical stimuli provides a reliable and practical method for early diagnosis, evaluation of therapy and monitoring of the disease. Interest in the link between the mechanical response of the cell at subcellular level and biological properties requires certain characteristics to be obtained from an appropriate measurement method. Atomic force microscopy (AFM), the reference technique, measures the deformation of an adherent cell with great sensitivity for elastic and viscoelastic properties. Its limitations for cells in suspension prevent its use with liquid biopsy products. Microfluidics-based approaches analyse the mechanical characteristics of cells in suspension. However, the absence of active compression prevents practical control of cell deformation for in-depth analysis. In summary, existing techniques show limitations when it comes to obtaining a practical and reliable means for all the required characteristics, e.g. mechanical characterisation of cells in suspension and controlled compression even with high deformation without compromising subcellular visualisation. I developed a method combining MEMS technology, artificial intelligence (AI), confocal imaging and finite element modelling to study the link between the mechanical characteristics and the metastatic potential of a cell. The first step was to use the biophysical signatures of cells to predict their metastatic potential. I chose six breast cancer cell lines, forming three groups according to their metastatic potential: high (SUM159PT & MDA-MB-231), low (MCF7 & T47D), and not (MCF10A & HTRET). The cell lines in each group did not differ significantly, but each group was significantly different. The application of AI algorithms resulted in a prediction rate of nearly 95% between cancerous and non-cancerous cells and over 80% between the three groups, demonstrating that biophysical properties are good candidates for predicting metastatic potential. I then adapted the method to integrate MEMS tweezers with confocal microscopy in order to visualise the subcellular elements of a cancer cell simultaneously with its mechanical characterisation. Finally, the real-time visualisation of subcellular elements combined with measurements of the mechanical characteristics of the cells was used to develop a cell-specific finite element model for further analysis of subcellular elements such as the nucleus. The method demonstrated makes it possible to obtain models representing exact cellular structures in order to predict their biological properties on the basis of mechanical properties. The link between the mechanical response of cells and subcellular elements eliminates time-consuming and costly biological steps. The specific models developed by the method demonstrated can lead to critical applications commonly used for early diagnosis, drug testing and therapy monitoring with minimally invasive sample handling, e.g. liquid biopsy.

Abstract:

Liquid biopsy is the term used to describe the use of peripheral blood as a source of cancer diagnosis. This minimally invasive technique targets biological samples as tumour signatures in the blood, such as circulating tumour cells (CTCs). Since CTCs are heterogeneous and rare, detecting them is a major challenge. As CTCs undergo mechanical stress during the process of metastasis, studying the response of the cell and its subcellular elements to different mechanical stimuli provides a reliable and practical method for early diagnosis, evaluation of therapy and monitoring of the disease. Interest in the link between the mechanical response of the cell at subcellular level and biological properties requires certain characteristics to be obtained from an appropriate measurement method. Atomic force microscopy (AFM), the reference technique, measures the deformation of an adherent cell with great sensitivity for elastic and viscoelastic properties. Its limitations for cells in suspension prevent its use with liquid biopsy products. Microfluidics-based approaches analyse the mechanical characteristics of cells in suspension. However, the absence of active compression prevents practical control of cell deformation for in-depth analysis. In summary, existing techniques show limitations when it comes to obtaining a practical and reliable means for all the required characteristics, e.g. mechanical characterisation of cells in suspension and controlled compression even with high deformation without compromising subcellular visualisation. I developed a method combining MEMS technology, artificial intelligence (AI), confocal imaging and finite element modelling to study the link between the mechanical characteristics and the metastatic potential of a cell. The first step was to use the biophysical signatures of cells to predict their metastatic potential. I chose six breast cancer cell lines, forming three groups according to their metastatic potential: high (SUM159PT & MDA-MB-231), low (MCF7 & T47D), and not (MCF10A & HTRET). The cell lines in each group did not differ significantly, but each group was significantly different. The application of AI algorithms resulted in a prediction rate of almost 95% between cancerous and non-cancerous cells and over 80% between the three groups, demonstrating that biophysical properties are good candidates for predicting metastatic potential. I then adapted the method to integrate MEMS tweezers with confocal microscopy in order to visualise the subcellular elements of a cancer cell simultaneously with its mechanical characterisation. Finally, the real-time visualisation of subcellular elements combined with measurements of the mechanical characteristics of the cells was used to develop a cell-specific finite element model for further analysis of subcellular elements such as the nucleus. The method demonstrated makes it possible to obtain models representing exact cellular structures in order to predict their biological properties on the basis of mechanical properties. The link between the mechanical response of cells and subcellular elements eliminates time-consuming and costly biological steps. The specific models developed by the method demonstrated can lead to critical applications commonly used for early diagnosis, drug testing and therapy monitoring with minimally invasive sample handling, e.g. liquid biopsy.