Extended Data Fig. 5: The Primitive versus GMP axis governs AML prognosis. | Nature Medicine

Extended Data Fig. 5: The Primitive versus GMP axis governs AML prognosis.

From: A cellular hierarchy framework for understanding heterogeneity and predicting drug response in acute myeloid leukemia

Extended Data Fig. 5

A) 495 adult AML patients from GSE6891, projected by hierarchy composition and classified based on the reference cohorts. B) Overall survival outcomes of 495 AML patients from GSE6891, stratified by hierarchy subtype. C) 287 pediatric AML patients from the TARGET-AML cohort, projected by hierarchy composition and classified based on the reference cohorts. D) Overall survival outcomes 287 pediatric AML patients from TARGET-AML, stratified by hierarchy subtype. E) Correlation between a prognostic score trained by regularized cox regression using leukemic cell type abundances with the Primitive versus GMP axis (PC1) within the TCGA and BEAT-AML cohorts (combined n = 454). F) Overall survival outcomes of patients stratified by PC1 within the TCGA (n = 173), BEAT-AML (n=281), and GSE6891 (n = 495) cohorts. G) Association between cell-type abundance and induction failure in four independent studies: Bolouri et al 2018 (pediatric AML; n = 257 remission, n = 30 induction failure), Chiu et al 2019 (adult AML; n = 18 remission, n = 18 induction failure), Herold et al 2018 (adult AML; n = 164 remission, n = 86 induction failure), and Tyner et al 2018 (adult AML; n = 140 remission, n = 63 induction failure), represented through the test statistic from a two-sided Wilcoxon rank sum test. Green denotes higher relative abundance in induction failure patients compare to patients who achieved complete remission, while purple denotes lower relative abundance in induction failure patients. Differences with an uncorrected P < 0.10 are noted with an asterisk. H) Correlation between four prognostic AML scores with the relative abundance of each leukemic cell type across the TCGA, BEAT-AML, and Leucegene cohorts (combined n = 864). I) Relative abundance of Quiescent LSPC and GMP-like blasts among 864 AML patients split into high and low risk categories by median split for four prognostic scores. Significance was evaluated through two-sided Wilcoxon rank-sum tests. Box plots indicate the range of the central 50% of the data, with the central line marking the median. Whiskers extend from each box to 1.5*(interquartile range). J) GSEA analysis with gene signatures derived from the Primitive versus GMP axis in normal and malignant hematopoiesis, performed on genes ranked by univariate associations with overall survival within the TCGA and BEAT-AML cohorts.

Back to article page