Development of Molecular Mutations during the Evolution of Therapy-Related MDS and Therapy-Related AML

Introduction: Therapy-related myelodysplastic syndrome (t-MDS) and acute myeloid leukemia (t-AML) develop after the application of chemotherapy for malignancies in a significant number of patients (pts). Mutations in TP53 have been described recently to be present even before chemotherapy for the prior malignancy and thus also before any sign of t-MDS or t-AML. Data suggested that chemotherapy selected the TP53mutated clone which evolved to t-MDS/t-AML. More comprehensive genetic analyses, however, have been lacking so far.

Aim: To identify molecular mutations by a comprehensive gene panel in pts at t-MDS/t-AML diagnosis and to backtrack them to prior time points.

Patients and Methods: We searched our database for pts diagnosed with t-MDS or t-AML for whom in addition ≥1 prior peripheral blood or bone marrow sample from assessment of a previously treated malignancy was stored. Diagnosis of t-MDS and t-AML was performed by cytomorphology, cytochemistry and cytogenetics according to WHO classification 2008 in all cases. A total of 11 pts were identified (3/8 females/males; median age at t-MDS/t-AML diagnosis 72 years, range 50-81 years). 8 pts had t-MDS and 3 had t-AML. All pts had received chemotherapy for CLL before. All pts underwent mutation analysis at t-MDS/t-AML diagnosis by a 26 gene panel targeting ASXL1, BCOR, BRAF, CBL, DNMT3A, ETV6, EZH2, FLT3-TKD, GATA1, GATA2, IDH1, IDH2, JAK2, KIT, KRAS, MPL, NPM1, NRAS, PHF6, RUNX1, SF3B1, SRSF2, TET2, TP53, U2AF1, and WT1. The library was generated with the ThunderStorm (RainDance Technologies, Billerica, MA) and sequenced on MiSeq instruments (Illumina, San Diego, CA). Specific mutations identified at t-MDS/t-AML diagnosis were selectively analyzed in prior samples of the respective patients. Mutations were considered for this analysis only if they were present at t-MDS/t-AML diagnosis at mutation loads clearly higher than residual CLL infiltration. Accordingly, mutations were excluded from this analysis if their load was in the range of residual CLL infiltration or lower. One not yet described genetic variant was also excluded.

Results: 13 mutations were identified at t-MDS/t-AML diagnosis in 8/11 pts. While in 3 pts no mutations were found, 5 pts had 1 mutation, 2 had 2, and 1 had 4 mutations. Mean number of mutations per pt was 1.6. TP53 was mutated most frequently (n=5), RUNX1 was mutated in 2 pts, and FLT3-TKD, IDH2, KRAS, NPM1, NRAS, and U2AF1 in 1 pt each. Mean mutation load was 27% (range 4-48%) while mean CLL infiltration at the same time point was 2% (range 0-4%). Thus, the attribution of the described mutations to t-MDS/t-AML is highly likely. We then analyzed a total of 13 samples (8 bone marrow, 5 peripheral blood) drawn prior to t-MDS/t-AML diagnosis from the 8 pts for the respective mutations identified at t-MDS/t-AML diagnosis. In 5/8 patients the respective specific mutations identified at t-MDS/t-AML diagnosis were found in at least one prior sample. Genes found mutated in the prior samples were TP53 in 2 cases and IDH2, KRAS, NPM1, RUNX1, and U2AF1 in 1 case each. Mutation loads in general were lower in prior samples as compared to samples at t-MDS/t-AML diagnosis (median 54-fold lower, range 1.5 to 205-fold), except for one sample with a similar load at both time points which both times was clearly higher than the residual CLL infiltration (50% and 42% vs. 9% and 4%). Specifically, in 3/4 patients with samples available from the time point of CLL diagnosis all of these mutations (n=4) were not detectable at a sensitivity level of 1% while in 1 patient 2 mutations were not detectable and a U2AF1mutation was identified with a 1.9% load. This further supports the concept of these mutations being related to a pre-malignant clone which in the majority of cases might have been present at undetectable levels at the time point of CLL diagnosis or which even developed only during chemotherapy and later evolved into t-MDS/t-AML. The mean interval from first detection of the respective mutations to t-MDS/t-AML diagnosis was 10 months (range 4-25 months).

Conclusions: Mutational screening applying a 26 gene panel identified molecular mutations in the majority of pts. These mutations were present up to 2 years before t-MDS/t-AML diagnosis. Further studies focusing on patients at risk of t-MDS/t-AML should clarify the role of early molecular screening helping to potentially improve diagnosis and management of t-MDS/t-AML.