Further Insights Into The Molecular Landscape Of De Novo Acute Myeloid Leukemia (AML) Investigating 1,291 Patients

The Cancer Genome Atlas Research Network (TCGA) published a hallmark sequencing study on molecular mutations in 200 fully characterized adult de novo AML (NEJM 2013). According to their data AML harbor in average 13 mutations in the coding region of the genome of which 5 are in genes known to be recurrently mutated in AML. Further, detailed data on co-occurrence and mutual exclusiveness of molecular mutations was presented. However, given the heterogeneity of AML a cohort of 200 AML might not be fully representative.

1. Compare the published mutation frequency to our cohort 2. Evaluate, whether the mutation frequencies vary with age. 3. Determine the number of additional mutations in genetically defined AML subgroups 4. Analyze the co-occurrence of molecular mutations.

1,291 adult de novo AML (700 m/591 f; median: 68 yrs; 18-100 yrs) were analyzed for mutations by different PCR assays and next-generation sequencing including the 11 most frequently mutated genes reported by TCGA (FLT3-ITD/-TKD, NPM1, DNMT3A, IDH1, IDH2, TET2, RUNX1, TP53, NRAS, CEPBA, WT1) and also ASXL1, KRAS, MLL-PTD (that had been found at lower frequencies by TCGA), and CBL. Cytogenetics was performed in all cases.

Mutations were found in NPM1: n=410/1,189 (34.5%), DNMT3A: n=105/340 (30.9%), TET2: n=104/349 (29.8%), FLT3-ITD: n=305/1,231 (24.8%), RUNX1: n=201/1,045 (19.2%), IDH2: n=154/938 (16.4%), ASXL1: n=157/1,000 (15.7%), TP53: n=97/743 (13.1%), NRAS: n=101/842 (12.0%), IDH1: n=93/1,053 (8.8%), FLT3-TKD: n=94/1,132 (8.3%), MLL-PTD: 98/1,181 (8.3%), CEPBA: n=84/1,105 (7.6%) (double-mut: n=38; single-mut: n=46), KRAS: n=38/552 (6.9%), WT1: n=58/918 (6.3%), and CBL: 8/352 (2.3%). These mutation frequencies are comparable to those reported by TCGA. Only ASXL1 mutations were less frequently observed by TCGA (2.5%).

The following mutations were more frequent in pts <60 yrs: FLT3-ITD (P=0.003), NPM1mut and WT1mut (P<0.001 for both). In contrast, ASXL1, RUNX1 (P<0.001, each) and TET2mut (P=0.005) were more frequent in pts ≥60yrs.

A total of 802 pts were investigated for at least 9 markers (ASXL1, FLT3-ITD, FLT3-TKD, CEBPA, MLL-PTD, IDH1, IDH2, NPM1, RUNX1): The median number of molecular mutations was 2 (range, 0-5; mean±SD, 1.6±0.9). The lowest number of additional mutations was observed in pts with RUNX1-RUNX1T1 (0.3±0.6) and reciprocal MLL rearrangements (mean±SD, 0.4±0.6) followed by CBFB-MYH11 (0.6±0.8), NPM1 (0.9±0.7), CEPBAmut (0.9±1.0), and MLL-PTD (1.2±0.7).

In concordance with TCGA results, a significant coincidence of ASXL1mut with IDH2mut and RUNX1mut was found. A total of 335 pts was screened for FLT3-ITD, DNMT3Amut, and NPM1mut in parallel and there was a high coincidence: 27/335 (8.1%) with all 3 mutations and further 63 (18.8%) with 2 out of 3; all combinations P<0.001, each). Beyond the observations within the TCGA study, we found additional positive correlations such as IDH1mut to DNMT3A (P=0.004) and as well to NPM1mut (P=<0.001), and FLT3-ITD to MLL-PTD (P=0.010) as well as to WT1mut (p=0.001).

Furthermore, according to the TCGA data, the following mutations were mutually exclusive: TP53mut to NPM1mut and to FLT3-ITD (P<0.001, each), and in addition RUNX1mut to NPM1mut (P<0.001). However, we could not confirm the mutual exclusiveness of RUNX1mut and FLT3-ITD as 21.0% of RUNX1mut AML also showed FLT3-ITD.

Beyond the TCGA data, we found the following mutations to show significant negative correlations:

1. 1)

   TP53 to ASXL1, CEBPA, IDH1, FLT3-TKD, and RUNX1;

2. 2)

   FLT3-ITD to ASXL1, FLT3-TKD, KRAS, and NRAS;

3. 3)

   NPM1 to ASXL1, CEBPA, and MLL-PTD;

4. 4)

   RUNX1 to DNMT3A, and FLT3-TKD;

5. 5)

   WT1 to ASXL1, IDH1, and IDH2;

6. 6)

   ASXL1 to FLT3-TKD.

MLL translocations were significantly negatively correlated with FLT3-ITD, NPM1, DNMT3A, IDH2, and RUNX1mut, as well were RUNX1-RUNX1T1 rearrangements with FLT3-ITD, NPM1, and IDH2mut, and CBFB-MYH11 rearrangements with FLT3-ITD and NPM1mut.

1) Investigation of a large cohort of de novo AML largely confirmed the mutation frequencies of the TCGA data, but revealed a higher frequency of ASXL1mut. 2) In addition, we depicted new patterns of positive and negative correlations of genetic alterations. 3) This further emphasizes the variety of pathways of leukemogenesis in de novo AML requiring additional analyses to delineate the prognostic impact of different marker combinations and their impact on treatment decisions.

Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Bacher:MLL Munich Leukemia Laboratory: Employment. Alpermann:MLL Munich Leukemia Laboratory: Employment. Kern:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Kohlmann:MLL Munich Leukemia Laboratory: Employment. Schnittger:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership.