S130 TRANSIENT SUBCLONES CARRYING NT5C2 MUTATIONS DEFINE A HIGH‐RISK PATIENT GROUP WITH POOR OUTCOME IN PEDIATRIC RELAPSED B‐CELL PRECURSOR ACUTE LYMPHOBLASTIC LEUKEMIA

Background: Acute lymphoblastic leukemia (ALL) is the most common cancer in children. Despite an overall favorable prognosis, 15–20% of patients suffer a relapse. In the treatment of relapsed ALL, resistance to chemotherapy is a major challenge. The purine nucleotide metabolizing cytosolic 5́‐nucleotidase II (NT5C2) harbors activating mutations in 20% of pediatric relapsed T‐ALL and in 3–10% of relapsed B‐cell precursor ALL cases, which are considered to drive relapse formation by causing resistance to purine analogue‐based maintenance treatment of ALL. However, the impact of NT5C2 mutations for second‐line treatment and long‐term outcome of relapsed patients is unknown. Aims: To elucidate the clinical significance of NT5C2 mutations in relapsed ALL, we studied the NT5C2 gene in a large cohort of 455 patients with relapsed B‐cell precursor ALL and assessed correlations with clinical presentation, genetics, response to treatment and outcome. Methods: We detected NT5C2 mutations by Sanger sequencing and amplicon‐based next generation sequencing. To detect subclonal NT5C2 mutations, we established allele‐specific quantitative real‐time PCR assays following minimal residual disease measurement guidelines. All relapsed patients were treated according to the German ALL‐REZ BFM 2002 protocol. Results: By sequencing, we found n = 36 NT5C2 mutations in n = 34 samples. About 30% of NT5C2 mutations appeared subclonal with an estimated allele frequency of less than 20%. The variants p.R39Q (8/36, 22%) and p.R367Q (14/36, 39%) were the most prevalent mutations in our cohort. Using allele‐specific quantitative real‐time PCR targeted to p.R39Q and p.R367Q, we identified additional 69 subclonal NT5C2 mutations. In total, NT5C2 mutations were present in 16% of patients with relapsed B‐cell precursor ALL (74/455), but the majority of cases (66%) showed subclonal mutations only. NT5C2 mutations most frequently occurred in relapses of so‐called B‐other‐ALL that lacks established cytogenetic changes ( P  < 0.001) and were correlated with CDKN2A/B loss. In addition, predominant, but not subclonal NT5C2 mutations were associated with NRAS mutations indicating that RAS signaling may be involved in NT5C2 ‐driven relapse formation. Both, subclonal and predominant NT5C2 mutations, were associated with inferior event‐free survival rates of relapsed patients compared to wild‐type NT5C2 (0.189±0.058 and 0.333 ± 0.096, respectively, versus 0.575 ± 0.026; P  < 0.001). Response to relapse treatment was particularly poor in relapsed patients with subclonal NT5C2 mutations only. This group showed an increased rate of nonresponse of more than 30% ( P  < 0.001). Moreover, patients of this group who responded to relapse treatment, showed high minimal residual disease levels at the end of relapse induction treatment in 80% of cases. To investigate whether outgrowth of NT5C2 mutant subclones during relapse treatment contributes to the poor response and outcome, we analyzed follow‐up samples and found that subclonal NT5C2 mutations never grew out to predominant mutations at the time of nonresponse or second relapse. In fact, analysis of samples taken after relapse induction treatment showed that NT5C2 mutant clones were frequently eradicated early during relapse treatment (14/20, 70%), whereas NT5C2 wild‐type leukemic cells persisted at minimal residual disease levels higher than 10E‐03 (16/20, 80%). Summary/Conclusion: Subclonal NT5C2 mutations define a patient group with poor response to treatment and outcome in relapsed B‐cell precursor ALL, but due to their transient character they may not play a driving role in this clinical phenotype.