Pancytopenia accompanied by a severe decrease in bone marrow (BM) cellularity in children can be due to a broad variety of underlying disorders. Appropriate classification of bone marrow failure syndrome in children is challenging, particularly in relation to the morphological distinction between aplastic anemia (AA), refractory cytopenia of childhood (RCC), and refractory cytopenia with multilineage dysplasia (RCMD). The goal of this study was to characterize the molecular pathogenesis of these conditions by identifying the full spectrum of gene mutations in 29 patients with these disorders through the use of exome sequencing.

Patient and Methods

Diagnosis of AA, RCC, or RCMD was made on basis of the 2008 World Health Organization (WHO) classification criteria. AA patients exhibited no morphologically dysplastic changes in any of their hematopoietic cell lineages, while RCC patients had<10% dysplastic changes in two or more cell lineages or >10% in one cell lineage. Patients classified as RCMD exhibited >10% of the dysplastic changes in two or more cell lineages. Blood and BM samples were obtained from 29 children (16 boys and 13 girls) with AA (n = 8), RCC (n = 11), or RCMD (n = 10). The median age at diagnosis was 11 years (range, 2–15 years).

Exome capture from paired DNA (non-T cells/CD3+ lymphocyte) was performed using SureSelect® Human All Exon V3 (Agilent Technologies, Santa Clara, CA) covering 50 Mb of the coding exons, followed by massive parallel sequencing using HiSeq 2000 (Illumina, San Diego, CA) according to the manufacturer’s protocol. Candidate somatic mutations were detected through our pipeline for whole exome sequencing (genomon: http://genomon.hgc.jp/exome/index.html). All candidate somatic nucleotide changes were validated by Sanger sequencing.


Exome sequencing pipeline identified a total of 193 non-synonymous somatic mutations or indels candidates among the 29 patients (range, 2–15 per patient). After validation by Sanger sequencing, one nonsense, 11 missense, and two frame-shift mutations were confirmed as non-silent somatic mutations. The average numbers of mutations per sample were not significantly different when comparing morphological diagnostic groups (0.50 in AA, 0.36 in RCC, 0.60 in RCMD). Of these validated genes, BCOR (n = 2) and CSK (n = 2) mutations were recurrent genetic events. BCOR is a frequent mutational target in myelodysplastic syndrome, whereas CSK somatic mutations were not reported in human cancers. BCOR mutations were found both in AA (c.472delA:p.S158fs; patient 13) and in RCMD (c.G3856T:p.E1286X; patient 39). Both patients with CSK mutations were classified as RCC (c.G994A:p.D332N; patient 23 and 27).

When comparing the clinical outcomes of patients with somatic mutations (n = 7) versus those without somatic mutations (n = 22), response rate to immunosuppressive therapy at 6 months (50% vs. 50%), 5-year clonal evolution rate (95% confidential interval) [0% (0% - 0%) vs. 6% (0% - 26%)], and the 5-year overall survival rate (95% confidential interval) [100% (100% - 100%) vs. 95% (70% - 99%)] were not significantly different.


Whole exome sequencing analysis was used for gene mutational profiling of patients with idiopathic bone marrow failure syndromes; i.e., AA, RCC, and RCMD. Although BCOR and CSK somatic mutations were recurrently identified, idiopathic bone marrow failure syndromes in children are characterized by a paucity of gene mutations, irrespective of morphological diagnosis. These findings suggest that morphological diagnosis based on WHO classification system does not discriminate the mutational profile and pathogenesis of bone marrow failure in children.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.