PHILADELPHIA—In a challenge to the current dogma of melanoma staging and management, new evidence points to metastatic dissemination early during progression of the primary lesion, in which seeding of distant metastases can occur without a regionally confined intermediate stage. These recent observations argue for early systemic therapy in high-risk cases, some experts are now saying.
“Clinicians have held the view that melanoma progresses from local disease to locoregional metastases to distant metastases and then death, but the alternative, which many favor as well, is that it disseminates widely from the get-go,” said Boris Bastian, MD, PhD, Professor of Dermatology and Pathology at the Helen Diller Family Comprehensive Cancer Center of the University of California, San Francisco, speaking here at the International Meeting of the Society for Melanoma Research.
The apparent stepwise dissemination may merely be because the regional lymph node “gets a head start and grows faster than the distant metastases and gives rise to this pattern,” he explained.
In the study he discussed, to investigate the sequence and mechanisms of spread of the disease, his team used deep genome sequencing of primary tumors and two or more metastases as well as normal DNA from eight patients. “We made some surprising observations,” he said: One patient had a primary melanoma on the forehead, followed by a parotid metastasis 27 months later, a skin metastasis on the forehead at 29 months, and a skin metastasis at the angle of the jaw at 34 months. As expected, the primary and metastatic tumors shared several hundred single nucleotide variants (SNVs) with the “ancestral” or normal DNA. But the primary and other lesions had more than a hundred SNVs that were only partially or not at all shared with the normal DNA.
“Importantly, some of these SNVs differed among the primary and metastatic lesions—many only present in one tumor, called private SNVs.”
He said the partially shared mutations are useful to determine the ancestry of each tumor and assess how closely it is related to another tumor. Having done this analysis for each patient, the researchers found that 90 percent of the somatic variants were shared among all tumors, and 45 percent of the copy number variants were shared, but some variants were only partially shared in some tumors and some of them were not.
For the metastases, parts of their genetic makeups went their own ways early on, indicating that the metastases were seeded early in the disease process, a sort of divergent evolution from the primary tumor and from the other metastases, he said.
Bastian showed several examples of chromosomal analysis tracing the evolution of metastases. Some were related to locoregional spread, and others could not have arisen in sequence from that nodal basin. In other cases, partially shared mutations indicated a closer evolutionary lineage.
An unanticipated finding was seeing a mixed population of cells in a single metastasis. He said his first thought was that a metastasis would be founded by a single cell splitting off from a tumor, seeding a new lesion, followed by clonal expansion. “However, a pattern we actually saw repeatedly in our patients, was a metastatic lesion with mixed lineage.”
An example was a patient with a primary tumor on her heel, followed by multiple locoregional metastases on the foot and ankle with many shared ancestral mutations on chromosome 14 but also some deletions shared among all four metastases that were not present in the primary tumor. Bastian was also able to determine how metastases developed from other ones, including more than one cell lineage contributing to the founding of a single metastasis or one metastasis contributing cells to multiple other ones.
Interestingly, he said, a (presumably inguinal) lymph node metastasis shared some characteristics with the primary tumor and the foot/ankle metastases but lacked certain characteristics of those metastases, indicating that it arose from the primary tumor and was not a descendant of the local metastases. He also showed evidence of metastases recolonizing the primary tumor.
In conclusion, he said, metastatic seeding occurs early during progression of primary melanoma and need not occur in a stepwise fashion of local to locoregional to distant spread. This finding in a significant number of cases implies that “the lymph nodes are imperfect filters of tumor cells and that cells easily pass through them,” giving rise to early seeding of metastases beyond the lymph node.
In addition, recolonization of established tumors can occur by cells from metastases, he noted, citing the work of Joan Massagué, PhD. and Larry Norton, MD, at Memorial Sloan-Kettering Cancer Center that demonstrated this phenomenon in an experimental model of breast cancer.
Given the experimental evidence of early seeding of metastases and the likely reseeding of lesions by metastases, the findings “may actually militate for early deployment of systemic therapies rather than surgical” in high-risk cases.
After the presentation, an audience member asked if the researchers ever saw a distant lesion that had mutations that were completely different from the primary tumor, suggesting that the distant lesion could be another primary. Dr. Bastian replied that he always found that about 90 percent of the genetic variants were shared between distant lesions and the index lesion, indicating that the distant lesions did arise from the identified primary tumor.