Microfluidics guided by deep learning for cancer immunotherapy screening.

Area of Science:

• Immunology
• Oncology
• Biotechnology

Background:

• T cell infiltration and cytotoxicity are crucial for inflammation and cancer immunotherapy.
• Current screening methods fail to assess T cell penetration into tumor stroma, hindering solid tumor treatment development.

Purpose of the Study:

• To develop an automated high-throughput microfluidic platform for simultaneous tracking of T cell infiltration and cytotoxicity in 3D tumor cultures.
• To evaluate treatment efficacy using a deep learning-based clinical tumor-infiltrating lymphocyte (TIL) score analyzer.
• To discover novel immuno- and combination therapies for solid tumors.

Main Methods:

• An automated microfluidic platform was developed for 3D tumor cultures with tunable stromal composition.
• Simultaneous tracking of T cell infiltration dynamics and cytotoxicity was performed.
• A clinical data-driven deep learning method (TIL score analyzer) was employed to score T cell infiltration patterns.
• A drug library was screened using this platform.

Main Results:

• The platform successfully evaluated T cell infiltration and cytotoxicity in 3D tumor models.
• Screening identified an epigenetic drug (lysine-specific histone demethylase 1 inhibitor, LSD1i) that significantly promoted T cell tumor infiltration.
• LSD1i demonstrated enhanced therapeutic efficacy when combined with an immune checkpoint inhibitor (anti-PD1) in vivo.

Conclusions:

• The developed automated system and strategy enable effective screening of immunocyte-solid tumor interactions.
• This approach facilitates the discovery of potent immuno- and combination therapies for solid tumors.
• The study highlights the potential of targeting T cell infiltration for improved cancer immunotherapy outcomes.