Melanoma’s Deadly Shift: DNA Methylation Rewrites the Tumor as It Spreads

When melanoma cells leave the skin and seed distant organs, they carry their DNA mutations with them — but according to a new study, what changes most dramatically is not the genetic code itself, but the chemical annotations written on top of it. An Italian research team, working as part of the Epigenetic Immune-Oncology Consortium AIRC (EPICA), has carried out one of the most direct comparisons yet of primary melanomas and the metastases that arise from them in the very same patients. Their findings, published in the Journal of Translational Medicine, suggest that the epigenetic landscape undergoes a coordinated reprogramming during metastatic progression, one that may operate largely independently of the driver mutations that oncologists traditionally use to classify the disease.

The study, led by Antonella Manca, Maria Cristina Sini, and Giuseppe Palmieri of the Institute of Genetic and Biomedical Research at Italy’s National Research Council in Sassari, together with colleagues in Florence, Perugia, and Genoa, examined 76 paired tumor samples — 38 primary melanomas and the first detected metastasis from each of the same 38 patients. All patients had advanced disease at the time of enrollment. The researchers profiled mutations across 517 genes spanning the key oncogenic pathways using next-generation sequencing, and in parallel performed reduced representation bisulfite sequencing (RRBS), a technique that maps methylation changes at CpG islands, the short stretches of DNA rich in cytosine-guanine dinucleotides where methylation most often regulates gene activity.

The genetic results were, in a sense, reassuringly boring. When the team compared primary tumors with their matched metastases, they found no statistically significant discrepancies in the mutational status of the main pathways that control cell proliferation and survival in melanoma: MAPK, PI3K/AKT, and INK4/ARF. Discordance between paired samples was relatively uncommon — 15.8 percent of patients for BRAF, 7.9 percent for NRAS, 21.1 percent for PI3K/AKT pathway genes, and 23.7 percent for INK4/ARF. In other words, the metastasis largely inherits the mutation profile of the primary tumor it came from, at least within the canonical drivers of melanomagenesis.

But one genetic metric told a different story. Tumor mutation burden, or TMB — the density of mutations per megabase of DNA, widely used as a proxy for how immunogenic a tumor should be — dropped significantly between primary tumors and metastases in a meaningful fraction of cases. Among the discordant pairs, seven patients had a high-TMB primary tumor but a low-TMB metastasis, while only one showed the reverse pattern, a statistically significant imbalance (p = 0.003). Overall, two-thirds of the patients (25 of 38) had primary melanomas with high TMB, defined as more than 10 mutations per megabase, yet in roughly a quarter of those cases (7 of 25), the paired metastasis had lost that high-mutation load. The authors propose that this loss of mutational burden during progression may reduce the immunogenicity of the tumor, helping metastases evade immune surveillance and contributing to the immunosuppressive microenvironment their group had previously described in the same cohort.

The methylation data, by contrast, revealed a striking and systematic shift. Principal component analysis of the RRBS profiles showed a clear separation between primary tumors and metastases, with methylation levels declining after progression to metastasis — a difference significant at p < 0.01. Notably, this epigenetic drift did not correlate with the anatomic site of the distant metastasis, nor with TMB, BRAF status, or alterations in the main melanomagenesis pathways. The implication is provocative: the epigenetic reprogramming that accompanies metastasis appears to unfold along its own trajectory, largely decoupled from the intracellular mutation landscape that is usually the focus of molecular tumor profiling.

Digging into the differentially methylated regions, the researchers found a coherent epigenetic program that reads like a blueprint for metastatic behavior. In samples from patients with advanced disease (AJCC stages IIIB to IV), two major patterns emerged. First, genes acting as tumor suppressors or inhibitors of cell growth showed hypermethylation — a modification typically associated with gene silencing — which would favor uncontrolled proliferation and the loss of differentiated characteristics, a process known as dedifferentiation. Second, genes involved in cytoskeletal remodeling, invasion, and neuronal signaling showed hypomethylation, consistent with the well-recognized neuron-like reprogramming of melanoma cells. Melanocytes, after all, share a developmental lineage with neural crest cells, and melanomas that reactivate this neural program tend to be more invasive and more capable of spreading.

When the team clustered primary tumors by their methylation profiles, three distinct subtypes emerged: a poorly methylated group, an intermediate group, and a group displaying the CpG island methylator phenotype, or CIMP, in which many CpG islands across the genome acquire methylation simultaneously. This classification, familiar from other cancers such as colorectal and glioblastoma, had not been clearly mapped against clinical outcomes in paired melanoma samples before. The researchers then asked the clinically crucial question: does the methylation class of the primary tumor predict how patients fare?

The answer, with the numbers available, was a cautious no — but with an intriguing hint. For overall survival, analyzed in 38 patients, the log-rank test found no significant differences among the three methylation groups (p = 0.36), and univariate Cox models confirmed the lack of significance, with a hazard ratio of 0.56 for the intermediate versus low group (p = 0.196) and 1.05 for CIMP versus low (p = 0.921). Progression-free survival, measured from primary diagnosis to first relapse in 36 patients, told the same story (p = 0.6). When the low and CIMP groups were merged into a single category, the trend toward a survival difference strengthened but still fell short of statistical significance (p = 0.15). Yet the raw numbers were suggestive: 84 percent of patients in the low and CIMP categories died during follow-up (21 of 25), compared with 54 percent in the intermediate group (7 of 13), yielding a hazard ratio of 1.81. The authors are candid about the limitation — the sample size of 38 patients is simply too small to detect moderate survival differences with confidence — but they argue that the consistent trend toward longer survival for patients whose primary tumors show intermediate methylation deserves confirmation in larger cohorts.

The broader significance of the study lies in how it reframes the relationship between genetics and epigenetics in cancer progression. Targeted therapies and immunotherapies in melanoma are largely built on the genetic framework: BRAF inhibitors for BRAF-mutant disease, and immune checkpoint inhibitors whose efficacy correlates with mutation burden. If, as this study suggests, metastases can shed their high-TMB character while acquiring a coordinated hypomethylation program that promotes invasion and neural-like plasticity, then the molecular portrait a clinician sees at metastasis may differ in clinically meaningful ways from the one painted at diagnosis — even when the driver mutations are identical. The finding that methylation changes proceed independently of mutation status also raises the possibility that epigenetic therapies, such as DNA methyltransferase inhibitors, could target vulnerabilities that genetic profiling alone would miss.

For now, the study’s authors, writing on behalf of the EPICA consortium and funded by the Fondazione AIRC, frame their work as a step toward integrating methylation profiling into the molecular workup of melanoma. The paired-sample design — comparing primary tumors and metastases within the same patient, rather than across unrelated cohorts — is a methodological strength that controls for the enormous inter-patient variability that has complicated earlier epigenetic studies. The consistent directional changes they observed, from global hypomethylation in metastases to the hypermethylation of tumor suppressor genes and the hypomethylation of invasion-related genes, paint a picture of metastasis as an epigenetic achievement as much as a genetic one. Whether the intermediate methylation class ultimately proves to be a prognostic marker will require larger studies, but the message of this work is already clear: to understand how melanoma kills, researchers may need to read not just the sequence of the genome, but the chemical script written upon it.

Subject of Research: DNA methylation and mutation changes during melanoma progression from primary tumors to matched metastases

Article Title: Impact of methylation and mutation modifications on melanoma progression through evaluation of paired primary and secondary lesions from same patients

Article References: Manca, A., Sini, M. C., Nuvoli, L., Simi, S., Costabile, S., Bellezza, G., De Giorgi, V., Guadagno, A., Ghiorzo, P., Mandalà, M., Massi, D., Palmieri, G., on behalf of the Epigenetic Immune-Oncology Consortium AIRC (EPICA), Maio, M., Anichini, A., Pfeffer, U., & Cossu, A. (2026). Impact of methylation and mutation modifications on melanoma progression through evaluation of paired primary and secondary lesions from same patients. Journal of Translational Medicine, 24(1), Article 1280. https://doi.org/10.1186/s12967-026-09082-1

Image Credits: AI Generated

DOI: 10.1186/s12967-026-09082-1

Keywords: melanoma, DNA methylation, epigenetics, tumor mutation burden, metastasis, CIMP, RRBS, BRAF, NRAS, tumor microenvironment, prognosis, paired tumor samples