Spread the love

 

Cancer is one of the leading causes of death worldwide, and distant metastases cause more than 90% of cancer deaths. Due to the limited resolution of imaging technologies such as bioluminescence and MRI, scientists have not been able to comprehensively detect metastatic cells throughout the body, which has greatly limited understanding of the mechanisms of spread of different types of cancer and hindered the development of effective therapies.

 

A research team led by Dr. Ali Ertürk, director of the Institute of Tissue Engineering and Regenerative Medicine at Helmholtz Zentrum München, Germany, has been working to develop new technologies to overcome obstacles in the area of ​​“metastatic cancer detection”. Previously, they developed a method of tissue removal and fixation called vDISCO. This method turns the mouse’s body into a transparent state, allowing imaging of individual cells. Using a laser scanning microscope, researchers were able to detect the smallest metastases in a mouse, even individual cancer cells.

 

However, although this technology is very “cool”, it has a big limitation that manually analyzing such high-resolution imaging data manually is a very time-consuming process. How to improve the efficiency of analysis is a problem that Dr. Ertürk’s team has been thinking about and trying to solve.

 

Since 90% of cancer patients die from metastasis, successfully controlling and eradicating cancer depends heavily on our ability to track and target all spreading cancer cells and metastases. In the cover paper of this issue, Pan et al. Developed a deep learning technology called DeepMACT, which can detect the smallest metastases in the whole body of a mouse and how these metastases are targeted by therapeutic antibodies. The image above shows multiple huge metastases in the mouse lung, some of which were targeted by therapeutic antibodies, and some were not.
On December 12, the research team released the latest breakthrough: they developed a deep learning-based algorithm called DeepMACT. Using this algorithm, Dr. Ertürk and others can more efficiently detect and analyze cancer metastasis. Related results appeared on the cover of this issue of Cell.

 

“Now we can perform high-throughput transfer analysis. With just a few clicks, DeepMACT can complete several months of manual inspection in less than an hour,” said Oliver Schoppe, co-author of the paper.

Experimental design and schematic diagram of DeepMACT for cancer metastasis and antibody drug targeted analysis

 

The full name of DeepMACT is deep learning-enabled metastasis analysis in cleared tissue, and its application is mainly divided into three steps: the first step is to use vDISCO protocol to fix and process mice to enhance the fluorescence signal of cancer cells; the second step, Transparent mice are imaged from head to toe to reveal all metastases; and the obtained images are combined into a complete 3D imaging of the mouse; in the third step, a trained algorithm is used to analyze the 3D images to detect the whole body of the mouse Cancer metastasis as well as antibody-based drug targeting.

 

vDISCO visualizes systemic metastasis in mice

Studies have confirmed that DeepMACT’s performance in detecting transfer is comparable to human experts, but it is more than 300 times faster.

 

Deep learning-based detection enables quantitative analysis at a single transfer level

In addition, using DeepMACT, researchers have gained new insights into the unique metastatic characteristics of different tumor models. By analyzing the progression of breast cancer metastasis in mice, DeepMACT revealed that small metastases in mice increased significantly over time. These features cannot be detected by traditional bioluminescence imaging. DeepMACT enables the first quantitative analysis of metastatic processes throughout the body.

 

DeepMACT reliably detects metastasis in all organs of various tumor models

More importantly, using DeepMACT, scientists can also analyze the targeting of tumor antibody therapies in detail. In the study, Dr. Ertürk and others used DeepMACT to quantify the efficacy of a therapeutic antibody called 6A10, which has previously been shown to slow tumor growth. The analysis results based on DeepMACT show that 6A10 actually “does not work” for up to 23% of the transfers.

DeepMACT enables quantitative analysis of drug delivery effects at a single metastatic level

In conclusion, the above research results show that DeepMACT not only provides a powerful method for comprehensive analysis of cancer metastasis, but also provides a sensitive tool for preclinical evaluation of therapeutic drugs.

 

It is worth mentioning that DeepMACT is publicly available. Dr. Ertürk said, “Today, clinical trials in the field of oncology have a success rate of about 5%. We believe that DeepMACT technology can significantly improve the drug development process, help find more powerful drug candidates that can move into clinical trials, and ultimately provide cancer patients Provide more accurate and effective medicine. “

Leave a Reply

Your email address will not be published. Required fields are marked *