Aaron Newman PhD

Analyzing cells' behaviors in social networks

October 6, 2021

By Christopher Vaughan

A personal assistant, a tech CEO, an auto mechanic, and a stay-at-home dad might have very distinct personalities and behaviors in their day-to-day lives, but when they all get together at a weekly gathering of their French club, the way they behave and the way they interact with others might be completely different. A CEO who is gregarious and confident at her job might be unsure of her French and hang back. The stay-at-home dad, who spent a year in France in high school, might leave behind his usual introversion and take on a leadership role. 

Cells are much the same way. One type of cell can behave very differently when it’s environment changes. The changes both influence, and are influenced by, the cells of various types that surround it. These changes in behavior and interaction have deep implications for the development of normal and diseased tissues.

Researchers at the Stanford Institute for Stem Cell Biology and Regenerative Medicine have devised a powerful method of combining computer analysis with other analytical methods to examine how cells behave and interact in various environments and have used the method create a new understanding of how cancer develops and can be treated. 

“Now we can look at the building blocks of tissue, the way that the whole cellular ecosystem is structured, rather than just looking at the types of cells present,” said Aaron Newman, PhD, and assistant professor of biomedical data science and a member of the Stanford Institute for Stem Cell Biology and Regenerative Medicine. “It’s a much more powerful way of looking at tissue organization.”

Their method, called EcoTyper, combines new computer algorithms with those previously developed by the researchers to analyze cell types, single-cell RNA sequencing, cell sorting, and existing databases. One of the advantages of EcoTyper is that researchers can use vast reserves of stored tissue and huge public databases to run “virtual clinical trials” and analyze thousands of cancer cases on the computer in a highly cost-effective manner, Newman said. 

Their work was published September 30 in the journal Cell. Newman, along with assistant professor of biomedical data science Andrew Gentles, PhD, are co-senior authors on the article, which showcase EcoTyper’s capabilities with an analysis of the tissue architecture across different types of solid cancer tumors. In a companion research article also published September 30 in the journal Cancer Cell, Newman, along with professor Ash Alizadeh, MD PhD and colleagues, showcase EcoTyper’s capabilities on lymphomas. 

“Ecotyping” carcinoma, the most common cause of cancer-related death

While lung cancer might look very different from bladder cancer or other types of cancerous tumors under the microscope, the researchers found 10 clinically distinct subtypes of multicellular communities, dubbed “ecotypes”, that existed across all different types of tumors. Furthermore, the presence or absence of certain ecotypes in a tumor was highly predictive for outcomes and often indicated what kinds of treatment would work best, even for very different types of cancer, the researchers say.

“We found one ecotype that was predictive of a good response to a particular immunotherapy,” said Bogdan Luca, PhD, a postdoctoral researcher and co-first author of the paper, who led the work with Chloé Steen, PhD. “In fact, it was even a better predictor than other candidate biomarkers that we tested, even ones that were specifically sought out to be predictive of response,” Luca said. In addition, with the use of EcoTyper, the researchers were able to predict whether a premalignant lesion in the lungs would spontaneously regress or develop into lung cancer. 

“EcoTyper can provide a platform for future therapies, because you have a better idea of the bad cells in a tumor you want to attack and spend less effort on the ones that are less bad,” said Andrew Gentles, PhD, assistant professor of medicine. This focus on interacting cell populations in a tumor is different than current approaches, which usually target “driver mutations” or genes along a certain pathway. “Many cancer therapies are focused on a given cell type or gene, but there are always other cells contributing to the cancer or cells that don’t have that gene mutation.”

Ecotyping the most common blood cancer

In the research that was published in the Cancer Cell article, the researchers wanted to know if there really were two different subtypes of a certain kind of lymphoma, as has generally been accepted in the field. Using EcoTyper, they analyzed the microenvironment found amongst and around diffuse large B-cell lymphoma cells. When viewed in this way, the researchers were able to identify not two but nine different subtypes of this lymphoma.

Because the researchers were working from tissue samples from previous lymphoma cases, they also had a record of how these cancer cases turned out, and could correlate these statistics. “We found that not only were there many more subtypes of this B-cell lymphoma than previously recognized, but also were able to show that knowing which subtype people had gave us an improved ability to make predictions about how the cancer would likely progress,” said Chloé Steen, PhD, a postdoctoral scholar and co-first author of the study. 

One of the most striking features of this research is that the investigators were able to use their new understanding on a clinical trial that had seemed to be a failure. 

Clinical trials for new therapies are vast undertakings, often costing a billion dollars or more to get move a drug from discovery to FDA approval. So, when a promising drug shows no statistically significant advantage in a clinical trial, it is not only a huge disappointment for doctors and patients, it is also a major financial hit for pharmaceutical companies. 

Usually, clinical trials data is closely guarded by pharmaceutical companies, but in one case where a drug for lymphoma failed its clinical trial, the company felt that there was no economic value in that data anymore and put it on a public database. Newman, along with institute member Ash Alizadeh, MD, PhD, and their colleagues took that data and, in effect, reran the clinical trial, this time on the computer and including their new understanding of how many more types of this B-Cell lymphoma there were.

“What we saw was that there was in fact a specific lymphoma subtype that did respond to the therapy,” Alizadeh said.  “But in the original trial, they couldn’t identify these other subtypes, and so this promising sign of efficacy was lost among the negative results for all the other lymphoma subtypes.”

 “Being able to find the right drug and craft effective cancer treatments based on the particular subtypes of cancer a patient is the epitome of precision health and personalized medicine,” Alizadeh said. “Ecotyper helps us do that.”