Fig. 4: Gene expression and survival features of NPM1 class I and II. | Nature Communications

Fig. 4: Gene expression and survival features of NPM1 class I and II.

From: The AML cellular state space unveils NPM1 immune evasion subtypes with distinct clinical outcomes

Fig. 4

a Hierarchical clustering analysis based on the average gene expression profile of AML immature cells. The clustering and heatmap are based on the 361 most variable genes (threshold set at 0.365 of the highest standard deviation) between the AML immature expression profiles from 38 AML samples. The proportion of AML immature cells for each sample is indicated below the heatmap together with the most common mutations, fusion genes and the genetic subtype of each sample. Lines denote separated clusters (C1-C4) and dashed lines indicate subclusters (denoted a, b, or c), based on the dendrogram above the heatmap. b UMAP representation of 48,656 cells from eight normal bone marrow (NBM) samples (gray) and 28,937 AML immature cells from eleven NPM1-mutated AML samples. NPM1-mutated samples from subcluster C3a (NPM1 class I) are indicated in different shades of blue while NPM1-mutated samples from subcluster C3b (NPM1 class II) are indicated in different shades of red. c Overall survival for NPM1 class I (blue) and NPM1 class II (red) for patients with hematopoietic stem cell transplantation (SCT, dark color) and without hematopoietic stem cell transplantation (No SCT, light color). Includes classifiable AMLs from TCGA, Beat-AML 2.0, and ClinSeq together with the current cohort (n = 206, two-sided Logrank test, P = 0.0047). The survival for NPM1 class I with SCT (n = 30) is significantly better than NPM1 class I without SCT (n = 67, two-sided Logrank test adjusted for multiple testing, P = 0.0033), whereas the survival for NPM1 class II with SCT is not significantly different from NPM1 class II without SCT (two-sided Logrank test adjusted for multiple testing, P = 0.43). Source data are provided as a Source Data file.

Back to article page