Distinct Gene Expression Profiles and Mutations Associate with Outcome in Younger Adults with De Novo Cytogenetically Normal Acute Myeloid Leukemia (CN-AML) (Alliance)

Approximately 20% of CN-AML patients (pts) younger than 60 years treated with current standard intensive combination chemotherapy regimens have primary refractory disease, and among pts who achieve a complete remission (CR), over half relapse. Thus, early detection of pts with low likelihood of achieving a CR and those with a high likelihood of relapsing is of crucial importance.

To identify molecular profiles associated with achieving and maintaining CR, we performed total transcriptome RNA sequencing and targeted DNA sequencing of 82 cancer- and leukemia-associated genes on pretreatment blood or bone marrow samples from 341 CN-AML pts less than 60 years old, similarly treated with intensive chemotherapy on Cancer and Leukemia Group B / Alliance for Clinical Trials in Oncology therapeutic trials. Pts who died within 30 days of diagnosis were excluded.

Pts were classified into 3 mutually exclusive outcome groups: primary refractory (n=55, 16% of pts), pts who achieved a CR but later relapsed (n=165, 48%) and pts who remained in CR for the duration of this study (n=121, 35%). Gene-level differential expression analysis between the 3 groups was performed after removing low-expressed genes, and a set of 385 genes met significance criteria (adjusted P < 0.05 and at least 1.5-fold increase or decrease between 2 groups). Hierarchical clustering pts using these 385 genes revealed 5 distinct pt clusters (Figure A). There were significant differences between the 5 clusters in the proportions of pts in the 3 outcome groups (P<0.001, Figure B).

Pts in cluster 1 (n=47 pts) almost all had bi-allelic CEBPA mutations (96%), and comparatively good outcome (all but one pt achieved a CR, and 49% achieved and maintained CR). In contrast, 94% of cluster 3 (n=69) pts harbored FLT3-internal tandem duplications (ITDs), and most pts either failed to achieve a CR (36%), or relapsed (51%).

Cluster 2 (n=138) and cluster 4 (n=39) had intermediate outcomes compared to the other clusters (11% and 8% primary refractory pts, respectively; 40% and 51% pts who relapsed after CR, respectively). Notably, within cluster 2, there was a sub-cluster of 24 poor-risk pts who were less like to achieve CR and maintain CR compared to the rest of the cluster 2 pts (P<0.001). Within this sub-cluster 25% of pts were primary refractory, 63% relapsed after CR, and only 13% maintained CR. The sub-cluster was defined by upregulation of 47 genes, only 6 of which were protein coding (BCO2, CXCL3, HILPDA, SDR42E2, SNAI1, and ZAR1L), and 35 were long-non coding RNAs (lncRNAs). To our knowledge none of these lncRNAs have been previously associated with AML relapse. As there were no significant differences in mutations among the 82 sequenced genes between this poor risk sub-cluster and the rest of the pts in cluster 2, expression of these genes might be useful for defining a poor-risk group, independent of mutations.

Finally, pts in cluster 5 (n=35) had a less distinct mutation pattern compared to other clusters. Most pts in cluster 5 did poorly. 23% were primary refractory and 67% achieved a CR then relapsed. Because no obvious mutations were detected that would drive this cluster, we performed gene set enrichment analysis (GSEA) for oncogenic signatures using the Broad Institute's GSEA and MSigDB software. Compared to the rest of the pts, those in cluster 5 were significantly enriched in genes shown to be downregulated by oncogenic KRAS expression (false discovery rate q<0.01). However, only one pt in cluster 5 and only 2% of pts in this study had KRAS mutations, and there was no difference in KRAS expression between cluster 5 pts and the rest of the cohort. Thus the observed GSEA differences in KRAS signaling are not due to KRAS mutations or expression changes, but are likely indirect effects of other genetic and epigenetic differences in the cluster 5 patients.

Our study shows that expression analysis can identify different outcome groups in younger CN-AML pts. We identified clusters of AML pts with poor CR rates and high incidences of relapse not associated with recurrent gene mutations, who were characterized by upregulation of lncRNAs, or enriched in KRAS signaling gene sets. These may represent new poor-risk subsets of AML pts requiring alternative treatment strategies.

Support: UG1CA23333801, U10CA180821, U10CA180882, U10CA180861, U24CA196171

ClinicalTrials.gov identifiers: NCT00048958 (8461), NCT00900224 (20202)

https://acknowledgments.alliancefound.org

Figure

View largeDownload slide

Figure

View largeDownload slide

Close modal