Summary: Researchers at Princeton have created a computational model that improves tracking of how cancer spreads through the body.
Source: Princeton University
Princeton researchers have introduced a new computational method that significantly improves the ability to reconstruct how cancer cells migrate from one site in the body to another.
Metastasis — the process by which cancer cells move from a primary tumor to distant organs — is responsible for roughly 90 percent of deaths from solid tumors, including cancers of the breast, prostate and colon. Better understanding of the routes and mechanisms of metastatic spread can point to new strategies to prevent and treat lethal disease.
“Are there specific genetic changes that enable tumor cells to leave a primary site and colonize distant organs?” asked Ben Raphael, professor of computer science at Princeton and senior author of the new study. “That question remains a central challenge in cancer research.”
In a study published in the May issue of Nature Genetics, Raphael and colleagues describe an algorithm named MACHINA — short for metastatic and clonal history integrative analysis — that reconstructs metastasis histories by combining DNA sequencing data with information about the anatomical locations where tumor samples were taken.
“MACHINA lets researchers infer the past migration events that led to a patient’s pattern of tumors using DNA data collected today,” Raphael explained.
By integrating both mutation patterns and anatomical origins of tumor samples, MACHINA produces clearer, more biologically plausible migration histories than many earlier approaches that relied on sequence data alone. Some prior analyses suggested complex, highly convoluted migration patterns that did not align well with existing biological knowledge of how tumors spread.
“Modern data sets are large and complex, but complex data do not necessarily require complex explanations,” Raphael said. Using a joint model of clonal evolution and migration, MACHINA often finds simpler, more parsimonious explanations for metastatic spread.
For example, reanalysis of a breast cancer case that had previously been interpreted as involving 14 separate migration events suggested a much simpler scenario: MACHINA found that a single lung metastasis likely seeded multiple additional sites in just five migration steps. The team also applied the algorithm to patient data from melanoma, ovarian and prostate cancers, revealing similar trends toward more parsimonious migration histories.
Key elements that improve MACHINA’s accuracy include an explicit model for comigration of genetically distinct cells — reflecting experimental evidence that tumor cells may travel in clusters — and an approach that accounts for uncertainty in sequencing data arising from mixtures of tumor and normal cells. These features help the algorithm handle the noisy, heterogeneous data typical of tumor DNA sequencing and reduce overinterpretation.
Andrea Sottoriva, the Chris Rokos Fellow in Evolution and Cancer at The Institute of Cancer Research in London, noted that MACHINA addresses important technical challenges. “This method will likely be widely adopted by the genomic community and should clarify the evolutionary processes that drive the most lethal phase of cancer,” he said.
Beyond refining histories for individual patients, MACHINA makes it feasible to analyze metastasis patterns across large cohorts. Such studies could identify recurrent mutations or pathways that drive spread in particular cancer types, guiding the development of targeted interventions.
Raphael and his team also plan to extend MACHINA by integrating additional data types, including tumor DNA found in circulating tumor cells and in blood plasma, as well as epigenetic information — reversible chemical modifications that affect gene regulation. Combining these data sources should sharpen the algorithm’s resolution and reveal further mechanisms of dissemination.
“Improving computational tools is like improving a microscope,” Raphael said. “Sharper methods reveal details that were previously hidden and help scientists see how cancer evolves and spreads.”
Other authors on the study include Mohammed El-Kebir, formerly a postdoctoral researcher in Raphael’s group and now an assistant professor at the University of Illinois at Urbana-Champaign, and Ph.D. student Gryte Satas.
Funding: This work was supported by the National Institutes of Health and the National Science Foundation.
Source: Molly Sharlach, Princeton University
Publisher: Organized by NeuroscienceNews.com
Image Source: NeuroscienceNews.com image is in the public domain.
Original Research: Abstract for “Inferring parsimonious migration histories for metastatic cancers” by Mohammed El-Kebir, Gryte Satas and Benjamin J. Raphael, Nature Genetics. Published April 26, 2018.
DOI: 10.1038/s41588-018-0106-z
Suggested citation examples have been provided by the original publisher for reference and archival use.
Abstract
Inferring parsimonious migration histories for metastatic cancers
Metastasis describes the spread of cancer cells from a primary tumor to other anatomical sites. Although metastasis was long considered to result primarily from single-cell seeding events, recent phylogenetic studies have reported complex migration patterns, including polyclonal seeding and reseeding. Determining migration histories from somatic mutation data is challenging due to intratumor heterogeneity and possible discordance between clonal lineages and actual cellular migration. MACHINA is a multi-objective optimization algorithm that jointly infers clonal lineages and parsimonious migration histories from DNA sequencing data. Analyses across multiple cancer data sets indicate that migration patterns are often underdetermined by sequencing alone and that previously reported complex migration scenarios may be less common than believed. MACHINA’s rigorous, integrative approach provides a clearer framework for studying the drivers of metastasis.