Metastases are generally thought of as a later event that occurs some time after the primary cancer develops, but new research shows that metastases may be seeded years before cancer is even diagnosed by clones that were “born to be bad.” This tantalizing idea has experimental support, at least for colorectal cancer, from research conducted by Christina Curtis, PhD, and her lab at the Stanford University School of Medicine in California.
The team showed that metastatic ‘seeding’ (ie, when the first tumor cell colonizes the metastatic site) is a very early event and likely happens years before the primary cancer is even detected.
“In 80% of patients with metastatic CRC, we determined that seeding of the metastatic sites occurred exceedingly early, before the primary tumor was clinically detectable,” Curtis told Medscape Medical News.
“Our data suggest that some tumors are born to be bad, with their malignant and potentially metastatic potential specified early,” she added.
The Curtis lab is now in the process of determining if the paradigm of early metastatic seeding is also true of other cancers, including those of the breast and lung.
Implications for the Clinical Translation
Approached for comments about this new research, pathologist Anirban Maitra, MBBS, of the University of Texas MD Anderson Cancer Center in Houston, pointed out that metastases are the most common cause of cancer death in solid cancers.
“What this study is telling us is that metastases observed from an image on a CT scan is an imprint of what was seeded years ago, and profiling a primary tumor provides an opportunity to address metastases before they are actually detected,” he told Medscape Medical News.
These observations are of immediate clinical relevance.
“These observations are of immediate clinical relevance and the findings provide the opportunity to predict the biology of CRC tumors on the basis of genomic profiling of the primary tumor,” said Maitra, who was not involved with the current study.
Profiling primary tumors before obvious metastatic spread has occurred offers an opportunity to identify which tumors are indolent and which are aggressive, he explained. “Because metastases might be predicted based on the profiling of the primary tumor, therapy can be appropriately tailored so that patients with indolent tumors are not overtreated and those with tumors ‘born to be bad’ can be treated aggressively from early on,” he said.
Findings Upend Long-Held Dogma About Metastasis
For Curtis, who runs a cancer computational and systems biology lab at Stanford and codirects the Molecular Tumor Board, the journey to trace a tumor’s evolution started several years ago. In a study published in 2015, Curtis challenged the prevailing notion of ongoing sequential clonal evolution. Her lab proposed a “Big Bang” model of colorectal tumor growth.
According to this model, after the tumor is founded, it evolves in an effectively neutral fashion such that additional mutations are largely passenger events and do not result in subsequent clonal expansions. “These findings also provided the first quantitative evidence for the early origin of intra-tumor heterogeneity,” she told Medscape Medical News.
In a continuation of their quest, Curtis and her lab tried to get a handle on how and when tumors evolve to give rise to lethal metastases. Since not all patients with CRC will develop metastatic disease, Curtis and her lab started with those who did.
The team analyzed biopsies taken from 23 patients with metastatic CRC — some with liver metastases and others with brain metastases — using exome sequencing of multiple regions of the primary tumor and paired metastases. They found that the genomic divergence between the primary and metastatic sites is low.
Curtis and colleagues then devised a computer model to simulate the spatial growth of realistically sized “virtual” tumors composed of up to 109 cells (equivalent to a 10 cm3 tumor). They compared the virtual tumors simulated under known parameters with actual patient genomic data to infer how large the primary tumor was when the metastasis was seeded. In the majority of patients they studied, they found that metastatic seeding occurred before the primary tumor was clinically detectable (104 – 108 cells or .0001 – 1 cm3) and often only 105 cells.
“By determining the fraction of cells (ie, frequency) with particular mutations in the primary tumor and metastatic biopsy, it is possible to infer the relative timing of metastatic seeding in the life history of a tumor,” Curtis said.
“These results also suggest a long time frame between the ‘birth’ of the malignant clone and diagnosis of the primary tumor,” she said. “Perhaps on the order of 3 years,” she added when pressed, and implied that there is a window of opportunity during which the nascent tumor could be intercepted.
“These findings stand in contrast to the long-held dogma that metastases arise from a minor clone in a genetically advanced tumor that arose after multiple waves of clonal expansion over a prolonged time interval,” Curtis said.
“Most driver mutations found in the metastasis were acquired early, during initiation and early growth of the primary tumor,” she added.
Metastases Seeded From Clones That Are “Born to Be Bad”
A cell that starts its journey from the primary tumor site has to migrate out of the tumor, survive different obstacles, and set up shop at a foreign site, Curtis explained.
She suggested that this is possible only with clones that are ‘born to be bad.’ These are clones that have accumulated several driver mutations at the primary site. Some of these drivers include mutations in KRAS, TP53, SMAD4, PTPRT, AMER1, and TCF7L2.
Although metastatic biopsies had a larger number of high-frequency mutations, most of these are passenger mutations and not associated with tumor growth, she indicated. The reason for few drivers is that metastatic drivers overlap with those for tumor invasiveness and growth, Curtis explained, adding that specific a combination of drivers confers metastatic competence.
To validate this idea, Curtis and her lab analyzed tumors from over 2700 patients with CRC — one third with metastatic disease and two thirds with stage I-III disease — who were part of the MSK-Impact and GENIE studies.
When they examined the genetic mutations in these tumor samples, they identified PTPRT as a driver, which in combination with canonical drivers (APC, KRAS, and TP53), was observed almost exclusively in patients with metastatic disease. Other combinations of drivers associated with metastasis include TCF7L2 and AMER1.
“Not all colorectal cancers will metastasize. This prognostic information can aid the development of biomarkers to stratify patients with aggressive disease,” she said, so as to identify those patients who have tumors that are likely to metastasize.
Curtis added that circulating tumor DNA (ctDNA) can be used to non-invasively monitor residual disease and indicated that recent studies have shown the potential of serial ctDNA analyses to detect recurrence months earlier than standard radiologic imaging.
Also approached for comment, Anil K. Rustgi, MD, director of the Herbert Irving Comprehensive Cancer Center at Columbia University, New York City, suggested that the signature of metastases can be used to identify and more closely monitor patients who are likely to experience future metastases.
“The study pivots therapy into prevention,” he told Medscape Medical News, explaining that conceptually it is now possible to think of treatment strategies to prevent the onset of metastases. “This study hints that a biomarker for metastases in CRC may be possible.” Rustgi was not associated with the current study.
Curtis agreed. “Determining whether patients with detectable ctDNA postsurgery harbor specific combinations of alterations associated with metastatic disease may inform the approach to treatment stratification,” she said.
Identifying these patients before metastases sets in is important as they may benefit from adjuvant therapy to target micrometastases. “Patients with aggressive tumors based on profiling of primary tumors should be provided systemic therapy either in the adjuvant or neoadjuvant setting so that micrometastases can be cleansed,” commented MD Anderson’s Maitra. He added that, as a molecular pathologist, he can envisage that prognosticators of metastases could be added to the panel of mutations that are already being screened.
However, both Maitra and Curtis agree that these findings need to be prospectively validated in clinical trials.
The finding that systemic spread can occur early, noted Curtis, highlights the importance of detecting malignancy at the earliest possible stage.
However, she also stressed that further research is needed before population-based screening approaches are implemented. A reasonable starting point for such efforts could be individuals with a family history or high risk of CRC or other cancers. More generally, a better understanding of premalignancy is needed and should inform the approach to earlier detection, Curtis said.
The study was funded in part by the American Cancer Society, the Emerson Collective Cancer Research Fund, the Wunderglo Foundation, and the National Cancer Institute. Curtis is a scientific advisor to GRAIL and reports stock options as well as consulting for GRAIL and Genentech . Maitra receives royalties from Cosmos Wisdom Biotechnology for a biomarker assay related to pancreatic cancer early detection. Rustgi has disclosed no relevant financial relationships.
Nature Genetics. Published online June 17, 2019. Abstract