Back

A Versatile Microfluidic Device for High-throughput Combinatorial Drug Screening

Jaiswal, S.; Reddy, M. P.; Paul, D.; Gandhi, P.; Majumder, A.

2026-09-10 cancer biology
10.64898/2026.09.09.750366 bioRxiv
Show abstract

Combination therapies can improve anticancer efficacy, but identifying effective drug pairs and dose combinations requires systematic exploration of multidimensional concentration spaces. Here, we extend a diffusion-driven, flowless microfluidic concentration gradient generator (CGG) to enable quantitative combinatorial drug screening without external pumps or continuous flow. The platform comprises a 5 x 5 array of interconnected culture nodes coupled to four peripheral reservoirs, in which overlapping diffusion fields generate spatially defined single- and multidrug exposures. Computational modelling was used to assign local drug concentrations to individual nodes, enabling direct correlation of the predicted exposure landscape with cellular response. Using MCF-7 breast cancer cells and 5-fluorouracil (5-FU) and doxorubicin (DOX) as model therapeutics, the platform resolved concentration-dependent single-agent responses, yielding IC50 values of 3.46 M for 5-FU and 1.22 M for DOX. Combinatorial loading generated 25 spatially defined 5-FU-DOX concentration pairs within a single device, which were resolved into two-dimensional concentration-response landscapes. Bliss independence analysis revealed concentration-specific drug interactions, with synergy predominating at low-to-intermediate concentrations and a transition towards additive and antagonistic responses at higher exposures. These findings establish a pump-free microfluidic strategy that integrates computational concentration mapping with spatially resolved pharmacological analysis to identify effective drug-combination windows within a single platform.

Matching journals

The top 2 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.