Label-free chimeric antigen receptor T-cell expression analysis using neural networks and statistical distribution...

Area of Science:

• Biotechnology
• Cellular Biology
• Immunotherapy

Background:

• Chimeric antigen receptor T (CAR-T)-cell therapy is a key treatment for blood cancers.
• Accurate monitoring of CAR expression is vital for CAR-T cell therapy effectiveness and safety.
• Current methods like flow cytometry are labor-intensive and can affect cell properties.

Purpose of the Study:

• To develop a novel, label-free method for predicting CAR expression rates in T cells.
• To establish a practical solution for real-time monitoring in CAR-T cell manufacturing and research.
• To improve upon the accuracy and efficiency of current CAR expression assessment techniques.

Main Methods:

• Utilized bright-field microscopy combined with deep learning (convolutional neural networks).
• Integrated classification score distribution analysis with Gaussian fitting for rate estimation.
• Validated the method against standard antibody-based flow cytometry measurements.

Main Results:

• Achieved prediction accuracy within 5% of standard flow cytometry measurements across four donors.
• Reduced the maximum prediction error from 13% to 4.4%.
• Demonstrated the first successful label-free prediction of CAR expression rates in T cells.

Conclusions:

• The novel label-free method provides a practical and accurate approach for monitoring CAR expression.
• This technique simplifies CAR-T cell analysis, relying only on standard bright-field microscopy.
• The method is suitable for both research and manufacturing settings, enhancing CAR-T cell therapy development.