Targeted Mutation Analysis of Acute Myeloid Leukemia at Diagnosis and Relapse Shows Different Clonal Evolution Patterns Involving Different Functional Classes of Genes

AML relapse is often associated with a clonal evolution at the cytogenetic and molecular level and therefore represents a challenge for the treatment. Targeted sequencing is now usually done at diagnosis in AML, as only a small core group of genes is frequently mutated in AML and MDS. This approach, contrary to Whole Genome Sequencing, is cheaper and allows the detection of variant allele fractions as low as 2% with a rapid turnover and high sequencing coverage and depths.

In this study we analyzed the karyotype and the mutations occuring in a panel of genes in 31 AML patients at diagnosis and at the relapse.

The Illumina TruSight Myeloid Sequencing Panel® was used that covers all exons of 15 genes and hotspots of 39 additionnal genes involved in myeloid malignancies. Three different pipelines were used for data analysis including GATK, Picard, Samtools, VarScan reference tools. We tried to assess the clonality and the clonal evolution patterns at relapse. Five different clonal evolution patterns including cytogenetic and molecular analyses were observed: (1) Stability, defined by absence of clonal change, (2) Gain, strictly defined by acquisition of additional variations (mutations or cytogenetic alterations), (3) Loss, strictly defined by loss of variants or regression, (4) Gain and Loss, indicating the combination of both "Gain" and "Loss" patterns, (5) Emergence, defined by the emergence of alterations that is unrelated to those found at diagnosis.

The cohort included 20 men and 11 women with a median age of 59 years old for which frozen DNA was available both at diagnosis and relapse that were referred in a single center. The elapsed time between the first remission and relapse (TTR) was 1.6 to 113.8 months. Almost all patients relapsed prematurely (less than 12 months). The overall survival and the survival after relapse were respectively 23.7 months and 9.1 months.

The cytogenetic evolution patterns observed at relapse were the following: Stability (40%), Gain (27%), Loss (31%), Gain and Loss (4%), Emergence (<1%). The mutation evolution patterns observed at relapse were the following: Stability (26%), Gain (23%), Loss (32%), Gain and Loss (16%), Emergence (3%) . No correlation between cytogenetic and mutation pattern was observed.

Patients without change at relapse ("Stability" pattern) had less mutated genes at diagnosis than in the other patterns (p=0.009). Whereas patients with the "Gain and Loss" pattern had more mutated genes at diagnosis than the other patients (≥4 mutated genes, p=0.001). Thirty five percent of the alterations acquired at relapse ("Gain" and "Gain and Loss" patterns) were found in tumor suppressor genes: TP53 was mutated in 60% of these patients, but mutations were also observed in signaling genes (25% of the patients). Thirty eight percent of the alterations lost at relapse ("Loss" and "Gain and Loss" patterns) are found in signaling genes (i.e FLT3, KIT, KRAS, NRAS ) and 23% in myeloid transcription factors (i.e CEBPA, RUNX1, GATA2 ). It was also been observed that 47% of mutations thatwere observed in each group at diagnosis and relapse were localized in DNA methylation and chromatin modification genes (i.e DNMT3A, IDH1/2, TET2, ASXL1, EZH2 ).

Interestingly, at diagnosis, mutations of DNA methylation and spliceosome complex genes were preferentially observed in the "Gain" pattern group (p=0.040). In contrast, mutations of signaling genes were preferentially mutated at diagnosis in the "Loss" group (p=0.020).

Even if not statistically significant, we observed that evolution and prognosis seemed to be different among the different patterns. The "Gain" pattern tends to have shorter overall survival (median 14.7 months vs 22.9 months in Stability group) whereas in the "Loss" group, overall survival appeared longer (median 34.1 months). Moreover, patients in the "Gain" group were significantly older (p=0.017), and were classified as AML with myelodysplasia-related changes in most of the cases.

To our knowledge, our work is the first reporting multiplex targeted NGS sequencing in AML at diagnosis and relapse. The comparison of clonal landscape between diagnosis and relapse allowed us to define 5 evolution patterns that differed in terms of mutation landscape and evolution.

Even if these data must be confirmed on a larger cohort, our study shows the feasibility of clonal analysis using targeted NGS at diagnosis and relapse in AML and its potential prognosis interest in clinical routine.