BCOR and BCORL1 mutations in pediatric acute myeloid leukemia

Abstract

Acute myeloid leukemia (AML) is a heterogeneous disease characterized by the presence of different collaborating cytogenetic and molecular aberrations that are associated with treatment response.1–3 Approximately 20% of pediatric AML patients are classified as cytogenetically normal AML (CN-AML).4 In the past decade, new prognostic relevant aberrations have been identified in CN-AML, such as NPM1, WT1, FLT3-ITD and CEBPA double mutations (CEBPAdm), which may improve future risk-group stratification.5–8 In addition to these mutations, we recently described recurrent cryptic NUP98/NSD1 and NUP98/KDM5A translocations in pediatric CN-AML.9,10 Still, some CN-AML samples have not been fully characterized. Recently, Grossmann et al. discovered a somatic mutation in the BCL6 co-repressor (BCOR) gene in an adult CN-AML patient using whole-exome sequencing.11 BCOR mutations were subsequently identified in 10 of 262 (approx. 4%) adult CN-AML patients, unselected for cytogenetic abnormalities, and for whom a poor outcome had been predicted. In addition, Li et al. discovered mutations in the BCL6 corepressor-like 1 (BCORL1) gene in 2 of 8 patients with secondary adult AML.12 Further exploration identified somatic BCORL1 mutations in 10 of 173 (6%) of adult AML patients, of which 5 cases were diagnosed with secondary or treatment-related AML. In 8 of 10 patients these included nonsense, splice site, and frame-shift mutations that were predicted to result in truncation of the protein, suggesting that BCORL1 is a tumor suppressor gene that may be inactivated by mutations.13 In pediatric AML, the role of BCOR and BCORL1 is unknown. Therefore, we explored the frequency and impact of BCOR and BCORL1 mutations in pediatric AML in a molecularly well documented cohort of 230 pediatric AML patients. Genomic DNA was PCR amplified using the primers described in Online Supplementary Table S1. Purified PCR products were directly sequenced. Sequence data were analyzed using CLCWorkbench (CLC Bio, Aarhus, Denmark). BCOR variations were determined in comparison to the coding DNA sequence ENS0342274, and BCORL1 variations to ENS0540052. A mutation was defined as a nucleotide change not reported in the dbSNP database. SNPs were defined as a nucleotide change as described in the dbSNP database. Characteristics of the 230 pediatric AML patients included in this mutational screening are listed in Table 1. Table 1. Clinical characteristics of pediatric acute myeloid leukemia patients included in mutation screening. The complete coding sequence of BCOR and BCORL1 was screened in a nested cohort of 83 de novo pediatric AML patients, and 17 AML cell lines. This cohort represented all relevant cytogenetic subgroups in pediatric AML, with an enrichment for CN-AML (n=48). All CN-AML patients were screened for NPM1, CEBPAdm, FLT3-ITD, IDH1/2, WT1, cKIT, N/K-RAS, DNMT3A, ASXL1 and RUNX1 mutations, and MLL-PTD, NUP98/NSD1, NUP98/KDM5A, and MLL-rearrangements (Online Supplementary Table S2). None of the cell lines showed a mutation. In 4 of 83 patients a BCOR mutation (detailed in Table 2) was identified; 3 in exon 4 and 1 in exon 12. Three of these cases were CN-AML (3 of 48, 6.3%), and the fourth mutation was seen in an MNX1/ETV6 translocated AML. In only one of 83 patients a BCORL1 mutation was identified which resulted in a stop codon (Table 2). This CN-AML patient had an underlying xeroderma pigmentosum. Table 2. Characteristics of BCOR and BCORL1 aberrations detected in childhood cytogenetically normal-acute myeloid leukemia patients. An additional 147 pediatric AML patients were screened for mutations in BCOR and BCORL1 exon 4, based on the occurrence of mutations in this exon; 136 cases were de novo AML, 11 cases were secondary AML, of which 6 cases were MDS which progressed to AML and 5 were secondary to earlier therapy because of another malignancy. None of the additional cases were CN-AML. No additional mutations were found. In 2 of 230 cases a missense SNP was found in BCOR (p.V679I, rs144722432) and in 35 of 230 (15%) a missense SNP in BCORL1 (p.G209S, rs5932715). Both SNPs are predicted as tolerated by SIFT analysis, and in line with the predicted prevalence in the normal population.14 In one case missense SNP rs139887979 (p.D94N) was identified in BCORL1, predicted as damaging by SIFT analysis.14 We were unable to confirm this in germline or remission because such material was not available for these patients. Altogether we identified a BCOR mutation in 4 of 230 (1.7%) cases. The frequency of mutations in pediatric CN-AML patients (6.3%; 95%CI: 2.1–16.8) was comparable to that in adult CN-AML (3.8%; 95%CI: 2.1–6.9; P=ns). The frequency of BCORL1 mutations (1 of 83, 1.2%; 95%CI: 0.2–6.5) was significantly lower than in adults (10 of 173, 5.8%; 95%CI: 3.2–10.3; P<0.05) in patients of whom we screened the complete exon, and also lower when only analyzing exon 4 of BCORL1 [1 of 230, 0.4% (0.1–2.4) vs. 7 of 173, 4.0% (2.0–8.1); P<0.05]. This may be due to the enrichment for secondary AML and treatment-related AML in the adult cohort, 73 of 173 (42%) in the cohort of Li et al. versus 12 of 230 (5%) in this cohort.11,12,15 Apart from mutation analyses, we studied BCOR and BCORL1 gene expression levels using RT-qPCR in 65 patients of the initial cohort, including all mutated cases. Expression levels were determined using a SYBRgreen (Finnzymes Inc, Woburn, MA, USA) Taqman assay and average cycle threshold (CT) values were compared to the reference gene GAPDH, using the comparative cycle time method. Expression levels of 2 BCOR-mutated cases seemed lower than the other pediatric AML cases, but the 4 mutated cases did not significantly deviate from the non-mutated pediatric AML patients (Online Supplementary Figure S1). This is in accordance with findings in adult AML patients where both decreased and normal BCOR levels were observed in cases with BCOR mutations.11 In adult AML, BCORL1 mutations did not affect mRNA levels, similar to the finding of our single case [0.297% expression relative to GAPDH, mean in group (n=65) 0.275%, range 0.049–1.358].13,15 In the adult CN-AML cohort, the presence of BCOR mutations conferred a poor outcome. In our cohort, the low frequency of mutations did not allow survival analysis to be performed. Clinical outcome of the mutated cases is shown in Table 2. The presence of the non-synonymous SNP in BCORL1, rs5932715, did not influence clinical outcome (5-year probability of overall survival 57±9% vs. 64±4%, P=0.4; 5-year probability of event-free survival 49±9% vs. 45±4, P=0.9). In conclusion, in pediatric AML, BCOR and BCORL1 mutations rarely occur. Consequently, the clinical relevance is difficult to determine.