Ultra-dense electrochemical chip, microfluidics, and machine learning for Cancer drug screening

• Grant number: 24/14758-7
• Support Opportunities: Regular Research Grants
• Start date: February 01, 2025
• End date: January 31, 2028
• Field of knowledge: Physical Sciences and Mathematics - Chemistry - Analytical Chemistry
• Principal Investigator: Renato Sousa Lima
• Grantee: Renato Sousa Lima
• Host Institution: Centro Nacional de Pesquisa em Energia e Materiais (CNPEM). Ministério da Ciência, Tecnologia e Inovação (Brasil). Campinas , SP, Brazil
• Associated researchers: FLAVIO MAKOTO SHIMIZU ; Gabriela Furlan Giordano ; Iris Renata Sousa Ribeiro ; Lucas Felipe de Lima ; William Reis de Araujo

Abstract

The preclinical phase is critical for the success of drug development, lasting 5 to 7 years and over which the viability of cell cultures (number of viable cells) is determined to assess the in vivo efficacy of drugs. In these drug susceptibility tests, viability analysis methods are usually laborious, time-consuming, and invasive. In this sense, electrochemical devices are a promising alternative to hasten up the study and, therefore, the selection of new drugs. However, providing these susceptibility tests with (1) high throughput, (2) simplicity, and (3) accuracy (for quantifying cell viability) remains a challenge. Therefore, we propose here the amalgamation of ultradense chips with microfluidics and machine learning (ML) to ensure susceptibility testing carrying the aforementioned attributes, whether from end-point or real-time assays toward pharmacokinetic assays. Another goal is to avoid the need for sensor calibrations for new determinations of viability and other important pharmacological parameter, i.e., the half-lethal concentration of the drug (LC50).Susceptibility tests of different 2D tumor cells to several off-the-shelf drugs will be performed on a glass chip (70 mm × 35 mm) with a high density of sensors (33 to 870), which are vertically arranged into a 3D grid pattern. Recently developed in our group (Adv. Healthcare Mater. 2024, 13, 2303509, ACS Appl. Mater. Interfaces 2024, DOI: 10.1021/acsami.4c01159, and ACS Sens. 2024, DOI: 10.1021/acssensors.4c01026), this sensor array has a low number of conductive lines that allows integrating multiple sensors on a single compact chip. Therefore, this device merges high resolution and reproducibility (merits of microfabrication) with low cost due to the large number of sensors per area. Additionally, due to its compact size, the chip is compatible with the use of microfluidics, a tool that drops reagent consumption and boosts reproducibility, among other advantages. In this project, microfluidic chips will be obtained by bonding channels in polydimethylsiloxane (PDMS) substrates. The devices will present 9 channels, with 5 sensors in each of them (there will be 45 sensors). Hence, up to 9 samples can be analyzed on a single chip, with measurements in quintuplicate for each case.The chip allows serial analyses by simply changing the contact of an electrode. Based on this property, rapid analyses of the Ru(NH3)63+ ion by square wave voltammetry (SWV) can be performed in series using a portable single-channel potentiostat, thus allowing drug susceptibility tests with high throughput and in a simple way. Since each SWV measurement will last 7 s, the analyses of all 45 sensors on the chip will be completed within just 315 s. SWV will play a decisive role not only in obtaining rapid individual analyses but also in providing a multivariate voltammogram. This voltammogram will be exploited to ensure accurate cell viability determination through ML. It is worth noting that the predicted cell viability data will be used to calculate the LC50.After application to different tumor cells and drugs, ML will also be adopted to obtain a single equation for predicting cell viability. With this, we aim to enable determination of viability through a universal equation, eliminating the need to calibrate the chip for each new cell or drug to be scrutinized. Ensuring the use of the method without this requirement will be essential for its applicability. In short, the solutions addressed in this project may help to accelerate the transition of effective and safe drugs from in silico studies to their use by patients. (AU)