Skip to main content
. 2020 Feb 5;18:56. doi: 10.1186/s12967-020-02235-w

Fig. 1.

Fig. 1

Metastatic melanomas can be distinguished according to their immune-related tumor microenvironment. a Hierarchical clustering of metastatic melanoma samples according to their predicted infiltrated immune cell populations. Samples were labeled according to their molecular subtypes, anatomical location and mutation status of the main driver genes. The dotted red line represents the cutoff for group assignment indicated in the label (G1—pink, G2—green, G3—orange) or no group (white) and values represent the mean clusterwise Jaccard bootstrap values calculated to confirm cluster stability, with 0.75 or higher pointing to stable clusters. b Kaplan–Meier curves for 5-year overall survival rates and Cox’s proportional hazards ratios of melanoma patients according to their TME-based classification. The log-rank test was used to analyze the difference in survival curves between the groups. Global p = 0.038. **p = 0.01 (Log-rank test for G2 versus G3). c Forest plot representing the survival hazard ratio of relative fractions, divided in quartiles, for each immune cell type using all samples of the three groups. d T-cell receptor (TCR) clone count (in log scale) per TCR chain. e Inverse Simpson diversity index of combined TCR chains. f B-cell receptor (BCR) representation (in log scale) considering immunoglobulins (Ig) isotypes. BCR representation was calculated by multiplying clonotype count by the correspondent read counts. g Inverse Simpson diversity index of combined BCR chains. MW test was used for mean comparisons: *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. h Circus plot showing differentially expressed chemokines between G2 and G3 linked to their respective target TME population. Chemokines upregulated in G3 are represented in orange and those upregulated in G2 are in green