Somatic Mutations and Clonal Hematopoiesis in Aplastic Anemia

Abstract

Background: In patients with acquired aplastic anemia, destruction of hematopoietic cells by the immune system leads to pancytopenia. Patients have a response to immunosuppressive therapy, but myelodysplastic syndromes and acute myeloid leukemia develop in about 15% of the patients, usually many months to years after the diagnosis of aplastic anemia.

Methods: We performed next-generation sequencing and array-based karyotyping using 668 blood samples obtained from 439 patients with aplastic anemia. We analyzed serial samples obtained from 82 patients.

Results: Somatic mutations in myeloid cancer candidate genes were present in one third of the patients, in a limited number of genes and at low initial variant allele frequency. Clonal hematopoiesis was detected in 47% of the patients, most frequently as acquired mutations. The prevalence of the mutations increased with age, and mutations had an age-related signature. DNMT3A-mutated and ASXL1-mutated clones tended to increase in size over time; the size of BCOR- and BCORL1-mutated and PIGA-mutated clones decreased or remained stable. Mutations in PIGA and BCOR and BCORL1 correlated with a better response to immunosuppressive therapy and longer and a higher rate of overall and progression-free survival; mutations in a subgroup of genes that included DNMT3A and ASXL1 were associated with worse outcomes. However, clonal dynamics were highly variable and might not necessarily have predicted the response to therapy and long-term survival among individual patients.

Conclusions: Clonal hematopoiesis was prevalent in aplastic anemia. Some mutations were related to clinical outcomes. A highly biased set of mutations is evidence of Darwinian selection in the failed bone marrow environment. The pattern of somatic clones in individual patients over time was variable and frequently unpredictable. (Funded by Grant-in-Aid for Scientific Research and others.).

Figures

Figure 1.. Somatic Mutations Identified by Targeted Sequencing

(A) Frequency of mutated genes and type of mutations in each gene identified in US (NIH + Cleveland-CLV) and Japanese (JPN) cohorts. (B) Frequency (upper panel) and mean number (lower panel) of mutations in different age groups are presented for indicated sets of mutations with corresponding regression lines. (C) Variant allele frequency of mutations detected at time of diagnosis and 6 months post-therapy in the NIH cohort. Mutations with higher allele frequency at diagnosis than at 6 months post-IST are shown with red lines, whereas the rest are depicted in grey lines.

Figure 2.. Clinical Correlations with Somatic Mutations

Gene set enrichment analysis using a PVS algorithm identified sets of genes that associate with good and poor response to immunosuppressive therapy (panel A), OS (panel B), and PFS (panel C) in the NIH cohort. (A) Inferior response to immunosuppressive therapy in a group of patients with “unfavorable” mutations (DNMT3A, ASXL1, TP53, RUNX1, JAK2, JAK3, or CSMD1) and a superior response with “favorable” mutations (PIGA or BCOR/BCORL1) compared to an “unmutated” group (P=0.029 by chi-square test). The width of each column represents the number of patients in each group and the response to immunosuppressive therapy is color-coded, with non-response shown in red, partial response in green, and complete response in blue. (B) Kaplan-Meier overall survival is compared among three groups: “favorable” mutations in PIGA or BCOR/BCORL1 (blue), “unmutated” (in green), and patients with “unfavorable” mutations in DNMT3A ASXL1, TP53, RUNX1, or CSMD1 (red). (C) Kaplan-Meier progression-free survival compares patients with “favourable” mutations in PIGA, or BCOR/BCORL1 (in green), “unmutated” cases (blue), and patients with “unfavorable” DNMT3A, ASXL1, RUNX1, JAK2, or JAK3 mutations (red). (D) Kaplan-Meier overall survival in patients younger than 60 years is compared among three groups: “favorable” mutations in PIGA or BCOR/BCORL1 (blue), “unmutated” (in green), and patients with “unfavorable” mutations in DNMT3A ASXL1, TP53, RUNX1, or CSMD1 (red). Statistical comparisons among groups used log-rank tests and p values are shown on the graphs. The “unmutated” group of patients included patients with other candidate gene mutations that did not cluster in gene set enrichment analysis with either favourable or unfavorable groups (results were similar when unmutated patients—no candidate gene mutations detected—were used as the reference group). Thirteen patients with “mixed” mutations were excluded from the gene set enrichment analysis.

Figure 3.. Temporal Profile of Mutations Detected by WES

(A-C) Representative secular analysis of somatic mutations and the relationship with peripheral blood counts and response to immunosuppressive therapy. Each case represents blood counts (upper panel) and variant allele frequency of mutations identified by whole exome sequencing (lower panel). Treatment received is shown at the bottom of each panel (CsA = cyclosporine and h/rATG = horse/rabbit antithymocyte globulin). Asterisks indicate the samples not subjected to whole exome sequencing. (D) Chronological history of clonal evolution from the onset to the last follow-up in NIH075. Each mutated gene is depicted in the representative cells. Vertical axis indicates the absolute volume of the clones. Representative mutations in each clone are also presented.

Figure 4.. Temporal Profile of Common Mutations in the NIH Cohort

Variant allele frequency of mutations in frequently affected genes such as (A) BCOR/BCORL1, (B) PIGA, (C) DNMT3A, (D) ASXL1, are plotted for all relevant mutations. Each line indicates the time course of VAFs for an individual mutation and response at 6 months was used to identify complete responders (CR-blue), partial responders (PR-green), and non-responders (NR-red) to immunosuppressive therapy.