Predicting Therapy Outcomes In a Shorter Testing Window
High-throughput screening presents both opportunity and challenge. It generates a wealth of data, but it's also resource-intensive. Often the limitation comes from the assay itself. Cells, growth nutrients, and assay components all impose their own requirements.
Nan Qin is always considering these limitations in her lab at the University of Düsseldorf. She is currently developing a high-throughput screening platform for antibody-based immunotherapies and advanced targeted therapies against hematologic cancers using 96- and 384-well plates.
“Due to nutrient depletion, cell overconfluence, and drug instability over time, the incubation period is usually limited to a maximum of 72 hours,” Qin explains.
That cutoff might work for fast-acting drugs. But what about the ones that need more time to show efficacy?
When we last spoke with Qin back in 2023, she was working on matching pediatric brain tumor patients to the right drug. To facilitate personalized testing on patient samples, she implemented live-cell analysis to monitor cellular responses over extended, undisturbed periods. That approach is different from endpoint assays, which capture a single snapshot and may miss important data when the biology of interest needs more time to reveal itself.
Qin observed that certain drug candidates didn't show their anti-cancer effect until the 120-hour mark, a few days longer than a typical assay endpoint.
But what if the assay can't run that long? Most drug screening uses a 48- to 72-hour window and making it longer can create real challenges for labs. Is it possible to predict the results of a longer incubation without running the full test? Qin set out to answer this by using targeted therapy-based assays for blood cancers.
To solve this problem, Qin partnered with Gunnar Klau's team to create VUScope, a mathematical model that predicts how a drug will act over a 120-hour incubation using only the first 72 hours of data. That information lets them flag which drugs are likely worth the extra incubation time.
The published paper behind VUScope, out in Bioinformatics earlier this year, describes how the model was trained. The team used images captured every three hours during a full 120-hour incubation on the Incucyte® SX5 Live-Cell Analysis System. They tested 18 HDAC inhibitors in four cell lines. Generating that much training data wouldn't be possible without a platform designed for genuinely high-throughput, continuous imaging, capable of handling six 384-well plates at maximum throughput.
When the model used only the first 72 hours of data to predict results at 120 hours, it performed almost as well as when using the full data set.
“Careful evaluation of the predicted values allows us to determine whether an extended incubation time is needed for certain drugs to better assess their anticancer effects,” Qin says. “This definitely saved both costs and labor,” she adds.
This kind of modeling is especially useful for drug screening campaigns where traditional metrics like IC50 and EC50 don't tell the whole story on their own. As the paper notes, VUScope is also built on a large amount of experimental, or “wet lab,” data, with the results independently confirmed using a separate viability assay. Next, Qin's team wants to move beyond simple cell counts, building combined parameters that track how T cells and tumor cells change together over time for a clearer read on treatment efficacy.
For more on Qin's earlier work, read our 2023 conversation with her. To discover the benefits of real-time, live-cell analysis in your cell culture workflows, including 3D cell model applications, check out our Incucyte® systems.
