Skip to main content
Premium Trial:

Request an Annual Quote

Metastases May Form Early in Development of Colorectal Cancer

NEW YORK (GenomeWeb) – Many colorectal cancer metastases may have arisen and spread even before the primary tumor was large enough to be detected, according to a new study.

A Stanford University-led team of researchers used exome sequencing data from patients with metastatic colorectal cancer to examine how the cancer spread to distant sites. Generally, metastases have been thought to arise as a subset of cancer cells accumulates additional genomic alterations that enable them to spread.

However, as they reported in Nature Genetics today, the Stanford team found few genomic differences between the paired primary and metastatic tumors they studied, suggesting that metastatic driver genes are acquired early in tumor development. They further found that these metastases tend to spread while the initial tumor is quite small.

"In the majority of metastatic colorectal cancer patients analyzed in this study, the cancer cells had already spread and begun to grow long before the primary tumor was clinically detectable," senior author Christina Curtis, an assistant professor of medicine and genetics at Stanford, said in a statement. "This indicates that metastatic competence was attained very early after the birth of the cancer. This runs counter to the prevailing assumption that metastasis occurs late in advanced primary tumors."

She and her colleagues analyzed exome sequencing data from 23 patients with colorectal cancer who had distant metastases to the liver, a common metastatic site, or the brain, a less common, but often fatal metastatic site. Within this cohort, the researchers noted high concordance among putative driver genes. Mutations in KRAS, TP53, SMAD4, and others were similar in primary and metastatic tumor pairs. In addition, primary and metastatic tumor pairs were also likely to share somatic SNVs and small indels.

Through a phylogenetic analysis, the researchers traced the origins of these metastatic tumors. For all but one of the cases analyzed, they found the distant metastases corresponded to a monophyletic clade. Further, they noted most of these metastases actually diverged early in the emergence of the primary tumor. Most of the liver and all of the brain metastases harbored many private clonal somatic SNVs, but no subclonal ones. This indicated to the researchers that a single cell or a small group of genetically similar cells seed most metastases.

"The cells that formed the metastasis were more closely related to the ancestors of the primary tumor than its present-day relatives," Curtis said. "Moreover, the metastasis shared early drivers present in the 'trunk' of the evolutionary tree, but harbored few additional drivers. This suggested that these cancers acquired metastatic competence very early on during their growth."

She and her colleagues also built a computational model to examine the development and evolution of metastases. They found that in their dataset 83 percent of the primary metastatic tumor pairs from 17 of the 21 patients likely underwent metastatic dissemination when the primary tumor was below the limits of clinical detection — smaller than 0.01 cm3 in size. They further estimated using clinical data that metastatic dissemination likely occurred more than three years before patients underwent surgery.

At the same time, the researchers found in a separate cohort of 2,751 patients with colorectal cancer, including 938 patients with metastatic cancer, that most metastases harbored a set of core colorectal cancer driver genes, but also an additional candidate metastasis driver gene. In particular, they noted that the gene PTRT — a part of the STAT3 signaling pathway — appears to be a highly specific driver of metastasis.

This suggested to the researchers that early dissemination can occur in many colorectal cancer patients, underscoring the need for early detection, possibly through detecting cell-free tumor DNA as these small tumors fall at the limits of detection for imaging approaches.