Molecular Alterations in Refractory Acute Myeloid Leukemia and Persistent Preleukemic Lesions Are Disclosed Using Mesenchymal Stromal Cells As Germline Control

Introduction:

Over the last years, genome and exome sequencing approaches have increased our knowledge of molecular alterations in acute myeloid leukemia (AML). However, some important limitations still need to be addressed. First, insights into the spectrum of molecular alterations of patients with refractory AML are rare, partly due to the lack of remission samples as germline control. As these patients have a dismal prognosis, there remains an unmet need to improve therapeutic options and to identify druggable molecular lesions. Secondly, in AML patients achieving a complete remission (CR), preleukemic alterations may persist in CR and are underestimated in frequency and relevance. In this work, we investigated mesenchymal stromal cells (MSC) as germline control to decipher the spectrum of molecular alterations in refractory patients with induction failure and to disclose preleukemic hits in patients achieving CR.

Patients and methods:

Bone marrow (BM) aspirates at initial diagnosis (ID) were obtained from 18 AML patients (9 pts with subsequent induction failure and 9 pts that achieved CR after first induction). MSC were expanded to passage 4 and defined as CD73⁺/CD105⁺/CD271^+/low/CD45^-/CD33^- plastic-adherent cells. For all patients, BM hematopoietic cells (BM-HC; n=18) as well as MSC (n=18) were analysed at the time of first diagnosis. All samples (n=45) were analysed by exome sequencing on a HiSeq2500 (100 bp paired end) with four samples per lane. For variant calling, MSC were used as germline control. We demanded a variant allele frequency (VAF) of >20%, coverage of >30 reads and translational consequences. In germline samples, the VAF had to be < 5%. For patients achieving CR, BM-HC at CR were also studied (n=9). We repeated the analysis with CR BM-HC as germline control and compared the two results. For all patients, clinical as well as molecular characteristics were available.

Results:

We obtained an average coverage of 96 reads per base for the protein coding regions. 96% of the target region was covered at least 10-fold.

The use of MSC as germline control allowed us to detect somatic mutations at initial diagnosis of refractory AML. In 9 refractory AML samples, we found 90 single nucleotide variants (SNV) and indels, which resulted in a median of 11 alterations per sample (range: 3-17). The spectrum of mutations showed an unexpectedly high rate of mutations in the spliceosome gene SRSF2 (3/9). Other recurrent mutations affected TET2 (2/9) and WT1 (2/9). Genes frequently mutated in non-selected AML were only present in one refractory patient (DNMT3A, RUNX1, IDH2, ASXL1, TP53, NRAS) or not found mutated (IDH1, KRAS).

To uncover preleukemic alterations in AML patients achieving CR (n=9), we compared MSC and BM-HC at CR as germline controls. Using MSC as germline, we called 97 SNVs and indels (median: 11 per sample; range: 4-18) in the leukemic cells at ID. Thirty-three additional SNVs were called in the leukemic BM by using MSC as germline, whereas these would have been missed using BM-HC at CR as germline (median: 3 SNVs per sample, range: 0-7). These represent preleukemic hits persistent in CR with a VAF between 5% (lower bound) and 75%. Recurrently mutated genes included genes recently associated with clonal haematopoiesis in the elderly population: DNMT3A (3/9; VAF: 18%, 24%, 75%) and TET2 (2/9; VAF: 13%, 23%). In addition, mutations in ASXL1 (VAF: 14%), SRSF2 (VAF: 15%), and RUNX1 (VAF: 5%) persisted in at least one patient in CR. This unbiased approach also allowed us to identify lesions, which have not yet been associated with AML, but account for clonal events in remission. Candidates included genes linked to cancer like PROX1 (VAF: 5%), or ERBB2 (VAF: 35%), but also genes involved in NF-kB activation such as CARD8 (VAF: 30%), or NLRC3 (VAF: 10%).

Conclusion:

The use of MSC allows to unravel molecular lesions in refractory AML by exome sequencing. Refractory AML patients showed a high rate of mutations in the spliceosome gene SRSF2 that needs further investigations as potential therapeutic target for patients with treatment failure. Moreover, the comparison of two different germline controls (MSC and BM-HC in CR) allowed detecting persistent preleukemic alterations. In addition to known hits like in DNMT3A, TET2, or ASXL1, we systematically identified a broader spectrum of premalignant events that indicate clonal hematopoietic expansion and thereby may provide insights into leukemic transformation.