Donate to Science & Enterprise

S&E on Mastodon

S&E on LinkedIn

S&E on Flipboard

Please share Science & Enterprise

AI Recreates Tumor Environment for Precision Cancer Care

Breast tumor microenvironment

Breast tumor microenvironment. (NIH, Flickr.

10 Aug. 2022. A bioinformatics company designed an algorithm that digitally recreates the support network for tumors from RNA sequences, to guide precise treatments for cancer patients. Researchers from Boston Gene Corp. in Waltham, Massachusetts describe the algorithm and demonstrate its utility for understanding therapy options for patients in Monday’s issue of the journal Cancer Cell.

The tumor microenvironment is an emerging area for research on cancer growth and treatments. The microenvironment is a system of immune cells, fibrous cells like those found in skin and connective tissue, blood vessels, and other cells that protect cancerous tumors and support their growth. The microenvironment provides nutrients for the tumor and in some cases a fibrous protective shell. Understanding the composition of the tumor microenvironment in each patient can help physicians prescribe a precise treatment strategy to overcome the barriers created by that protective system.

Boston Gene develops software for analyzing the detailed health status of cancer patients for decisions on the optimal treatments addressing their individual conditions. The company says it provides deep genetic profiling of patients, an understanding of their immune system status, risks from inherited diseases, and factors that may affect proposed treatments including immunotherapies. Part of that deep understanding of cancer patients, says Boston Gene, is a detailed analysis of the tumor microenvironment, using artificial intelligence algorithms.

Millions of RNA sequences to train the algorithm

In their paper, a team led by Nathan Fowler and Alexander Bagaev, chief medical officer and head of product development respectively for Boston Gene, describes development of their machine learning algorithm called Kassandra. The algorithm is constructed as a decision tree, based on some 9,400 RNA profiles from blood and tissue samples associated with tumor microenvironment and cancer cells. From these profiles, say the authors, they created millions of artificial transcriptomes or messenger RNA sequences for subsequently training the decision tree.

In addition, the team validated the algorithm on samples of 4,000 hematoxylin and eosin or H&E cell and tissue slides, a standard in pathology, matched to separate single-cell, antibody, and RNA sequence measurements. The algorithm’s decision tree provides the mechanism for digitally reconstructing a tumor microenvironment, a process called cellular deconvolution. With Kassandra, the researchers say they accurately predict 51 different collections of blood and tissue found in tumor microenvironments, showing their role in tumor development and related processes.

With these digitally created tumor microenvironments, the researchers report the presence of CD8 positive T-cells with PD-1 checkpoint inhibitors that block immune responses to tumors, and PD-L1 proteins, both biomarkers correlated with cancer immunotherapy targets. The authors conclude that the Kassandra algorithm shows promise for better understanding a patient’s tumor microenvironment for guiding decisions on precision cancer therapies.

“Our findings,” says Fowler in a Boston Gene statement, “demonstrate the importance of cellular deconvolution based on RNA-seq to accurately understand the composition and activity of the tumor and the microenvironment to improve treatment outcomes. We look forward to further developing the Kassandra algorithm and showing its clinical applicability in precision oncology.”

More from Science & Enterprise:

We designed Science & Enterprise for busy readers including investors, researchers, entrepreneurs, and students. Except for a narrow cookies and privacy strip for first-time visitors, we have no pop-ups blocking the entire page, nor distracting animated GIF graphics. If you want to subscribe for daily email alerts, you can do that here, or find the link in the upper left-hand corner of the desktop page. The site is free, with no paywall. But, of course, donations are gratefully accepted.

*     *     *


Comments are closed.