A unique clinical case Despite the success of immune checkpoint blockade (ICB) in melanoma, most patients eventually develop resistance. This study investigated a single metastatic melanoma patient who achieved a complete clinical response to ICB, then experienced delayed recurrence and ultimately died - offering a rare window into the full arc of resistance evolution over nine years.
Unprecedented longitudinal dataset The researchers collected and analyzed 37 tumor samples spanning the patient's entire clinical course: primary tumor, metastatic recurrence, pre-treatment lesions, on-treatment samples, post-progression tumors, and rapid autopsy specimens. Analyses included whole-exome sequencing, RNA sequencing, and highly multiplexed immunofluorescence (t-CyCIF).
Seven co-evolving tumor lineages Phylogenetic analysis revealed not one but seven distinct tumor lineages co-existing at the start of treatment, each with different genetic features. All recurrent tumors eventually emerged from a single lineage characterized by loss of chromosome 15q, with post-treatment clones accumulating additional resistance-driving alterations.
ICB transforms melanoma treatment Anti-PD-1 checkpoint blockade produces responses in 40-45% of patients with metastatic melanoma, representing a revolutionary advance. However, the majority of patients either fail to respond initially or develop resistance after an initial response, eventually succumbing to disease.
Known resistance mechanisms Multiple resistance pathways have been identified: beta-catenin activation, PTEN loss, impaired antigen presentation machinery (B2M loss), defects in interferon-gamma signaling, genome instability, and phenotypic plasticity with de-differentiation toward neural crest-like states. How these mechanisms interact and emerge within individual patients remains poorly understood.
Study design advantage Previous longitudinal studies had limited modalities or short timeframes. This study is the largest and most in-depth analysis of a single patient across all treatment phases - from initial diagnosis through complete response, recurrence, and death - providing an unprecedented map of resistance evolution.
Common tumor ancestor All 37 tumors shared a common ancestor with UV-damage mutation signature, 618 shared mutations, and driver mutations in IDH1, MAP2K1 (MEK1), CTNNB1 (beta-catenin), and ARID2. Notably absent were BRAF, NF1, or RAS mutations - suggesting a distinct mutational landscape.
Shared chromosomal losses All tumors shared loss of heterozygosity in chromosomes 3q, 6q, 9, 10, and 20 - spanning tumor suppressors CDKN2A/B and PTEN, interferon-gamma pathway genes IFNGR1 and JAK2, and chromatin remodeler ARID1B. These shared losses set the stage for later resistance evolution.
PTEN loss as near-universal event Homozygous PTEN deletion was found in 33 of 37 tumors, arising through a combination of shared chromosome 10 loss-of-heterozygosity and focal deletion. All 19 tumors examined by immunohistochemistry were PTEN-negative, suggesting multiple routes to complete PTEN loss across lineages.
Resistance lineage defined by 15q deletion All recurrent tumors arose from Lineage 3, defined by loss of chromosome 15q - a region containing B2M, required for MHC-I antigen presentation. Four independent whole-genome duplication events across lineages were associated with increased chromosomal instability and immunotherapy resistance.
Immune composition differences between lineages Bulk RNA-seq deconvolution (CIBERSORT) revealed significant differences in immune cell composition among tumor lineages. Lineages that responded to ICB showed enrichment for cytotoxic T cells and effector immune signatures, while resistant lineages showed relative immune exclusion.
Vasculogenic mimicry phenotype Highly multiplexed t-CyCIF imaging revealed a striking vasculogenic mimicry phenotype in NGFR-high tumor cells - neural crest-like cells that form channel-like structures mimicking blood vessels. These cells showed high PD-L1 expression and were found in close proximity to immune cells, suggesting active immune suppression.
Site-specific NGFR patterns Rapid autopsy samples revealed two distinct spatial patterns of NGFR-high cells: high polarity with proximity to immune cells in subcutaneous tumors versus a diffuse pattern in lung tumors. This site-specific difference suggests that the neural crest program plays different roles in different metastatic environments.
NGFR as a resistance marker Nerve growth factor receptor (NGFR) is a marker of neural crest-like de-differentiation in melanoma. NGFR-high tumor cells were enriched in resistant and post-treatment lesions, suggesting that phenotypic plasticity toward a neural crest state plays a functional role in immunotherapy resistance.
PD-L1 expression and immune proximity NGFR-high cells expressed high levels of PD-L1 and were spatially concentrated near immune cell clusters. This combination - expressing the immune checkpoint ligand while being located at the immune interface - suggests a mechanism by which these de-differentiated cells actively suppress anti-tumor immunity.
Convergent evolution of neural crest state The neural crest program was observed across multiple independent lineages, suggesting convergent positive selection for this state under immune pressure. This finding connects tumor evolutionary dynamics to phenotypic plasticity as dual mechanisms of ICB resistance.
Resistance is polyclonal and parallel A major clinical implication is that ICB resistance in melanoma is not driven by a single mutation or pathway but by multiple convergent alterations across parallel tumor lineages. Targeting any single resistance mechanism is unlikely to prevent relapse when multiple independent lineages can independently acquire resistance.
Monitoring implications The study demonstrates that resistance lineages were already present at the time of treatment initiation, not necessarily arising de novo after treatment. This suggests that early tumor molecular profiling - potentially including single-cell or multi-region sequencing - could identify high-risk lineages before clinical resistance becomes apparent.
NGFR as a therapeutic target The identification of NGFR-high cells as a convergent resistance population suggests this population could be a therapeutic target. Strategies targeting the neural crest state or the PD-L1 expression on NGFR-high cells might complement current checkpoint inhibitors and prevent or delay resistance.
Single patient case study The fundamental limitation is that all findings come from a single patient with an unusual and extended clinical course. While the depth of analysis is unprecedented, generalizability to other melanoma patients and other cancer types requires validation in larger cohorts.
Technical limitations Bulk RNA-seq deconvolution and spatial imaging provide different resolutions of tumor-immune interactions. Single-cell sequencing was performed only at specific timepoints, and comprehensive spatial proteomic mapping across all 37 samples was not feasible.
Future directions The study establishes a framework for understanding resistance evolution that can be applied to larger patient cohorts, particularly focusing on: (1) whether NGFR-high cells emerge early as a pre-existing state, (2) the functional role of B2M loss versus other antigen presentation defects, and (3) whether combining checkpoint blockade with NGFR pathway inhibitors prevents resistance emergence.