Clinical and Molecular Determinants of Clonal Evolution in Aplastic Anemia and Paroxysmal Nocturnal Hemoglobinuria

Abstract

Purpose: Secondary myeloid neoplasms (sMNs) remain the most serious long-term complications in patients with aplastic anemia (AA) and paroxysmal nocturnal hemoglobinuria (PNH). However, sMNs lack specific predictors, dedicated surveillance measures, and early therapeutic interventions.

Patients and methods: We studied a multicenter, retrospective cohort of 1,008 patients (median follow-up 8.6 years) with AA and PNH to assess clinical and molecular determinants of clonal evolution.

Results: Although none of the patients transplanted upfront (n = 117) developed clonal complications (either sMN or secondary PNH), the 10-year cumulative incidence of sMN in nontransplanted cases was 11.6%. In severe AA, older age at presentation and lack of response to immunosuppressive therapy were independently associated with increased risk of sMN, whereas untreated patients had the highest risk among nonsevere cases. The elapsed time from AA to sMN was 4.5 years. sMN developed in 94 patients. The 5-year overall survival reached 40% and was independently associated with bone marrow blasts at sMN onset. Myelodysplastic syndrome with high-risk phenotypes, del7/7q, and ASXL1, SETBP1, RUNX1, and RAS pathway gene mutations were the most frequent characteristics. Cross-sectional studies of clonal dynamics from baseline to evolution revealed that PIGA/human leukocyte antigen lesions decreased over time, being replaced by clones with myeloid hits. PIGA and BCOR/L1 mutation carriers had a lower risk of sMN progression, whereas myeloid driver lesions marked the group with a higher risk.

Conclusion: The risk of sMN in AA is associated with disease severity, lack of response to treatment, and patients' age. sMNs display high-risk morphological, karyotypic, and molecular features. The landscape of acquired somatic mutations is complex and incompletely understood and should be considered with caution in medical management.

FIG 1.

Study design. (A) CONSORT diagram of patient inclusion with details of sequenced patients for (B) cross-sectional and longitudinal analyses. ^aLandmark: 6 months after diagnosis. ^bLandmark: 6 months after first-line (or diagnosis if no treatment). AA, aplastic anemia; HSCT, hematopoietic stem-cell transplant; MN, myeloid neoplasia; nSAA, nonsevere aplastic anemia; PNH, paroxysmal nocturnal hemoglobinuria; SAA, severe aplastic anemia.

FIG 2.

Incidence and study of factors influencing malignant progression. (A) The cumulative incidence of malignant progression according to disease phenotypes. At 10 years from diagnosis, the risk of progression to secondary myeloid neoplasms was 12.8% (10.0-15.9) for AA, 13.1% (6.0-22.9) for AA/PNH overlap syndrome, and 3.4% (0.9-9.0) for PNH (P = .13 globally, P = .47 in AA v AA/PNH, P = .06 in AA v PNH, and AA/PNH v PNH by Gray's test). According to disease severity, (B) the 10-year cumulative incidence of malignant progression was 12.4% (9.4-15.8) for severe and 13.6% (8.6-19.6) for nonsevere cases (P = .86 by Gray's test). Cumulative incidence curves for myeloid progression according to (C) treatment response and (D) age at diagnosis. At 10 years, the cumulative incidence of progression was 15.7% (11.5-20.6) in patients with partial or no response to treatment versus 8.5% (4.4-14.2) in cases achieving complete response (P = .02). At 10 years, the cumulative incidence of progression was 6.6% (3.7-10.8) in younger cases (age ≤ 35 years) versus 20.0% (14.3-26.2) in older patients (P < .001). Numbers at risk are color-coded and indicated below each curve. AA, aplastic anemia; CR, complete response; nSAA, nonsevere aplastic anemia; PNH, paroxysmal nocturnal hemoglobinuria; SAA, severe aplastic anemia.

FIG 3.

Features of malignant progression. (A) The 5-year OS according to bone marrow blast percentage at secondary myeloid neoplasia onset (58.2% [41.1-71.9] in patients with < 5% v 20.2% [8.6-35.3] in those with ≥ 5%). The Forest plot in (B) illustrates multivariate analysis for OS using a multivariate Cox cause–specific model. In (C), a waterfall plot depicts the genomic landscape of progressors (n = 82 patients with available molecular information). Disease subtype, cytogenetic risk groups according to R-IPSS and R-IPSS risk group (higher > 3.5; lower ≤ 3.5), and age categories (age < 60/≥ 60 years) are color-coded on top, whereas on the right side, frequencies of mutations for each gene are shown. (D) The survival curves of main cytogenetic risk groups (numbers at risk for each subgroup are color-coded and provided below the curves). In (E), a circle plot illustrates the relationship between mutations and cytogenetics. Three cases failed the cytogenetic testing and thereby were removed from the panels including cytogenetics features. AML, acute myeloid leukemia; BMF, bone marrow failure; CK, complex karyotype; HSCT, hematopoietic stem-cell transplant; MDS, myelodysplastic syndrome; MPN, myeloproliferative neoplasm; NK, normal karyotype; ns, not significant; OS, overall survival.

FIG 4.

Clonal dynamics in AA. In (A), the pie chart shows the frequency of mutants for PIGA and myeloid mutations with a breakdown of cases according to characteristics at onset and treatment response (n = 302). Bar charts in (B) illustrate the difference in molecular spectra between baseline and myeloid progression with frequencies estimated on the total of mutants (n = 302 at baseline; n = 82 at evolution). (C) The dynamics of changes in molecular lesions (HLA, PIGA, and myeloid) in patients sequenced at baseline (time point D, diagnosis), follow-up, (time point F), and evolution (time point E, see the Data Supplement for details). Frequencies are shown as solid lines on the left, and clonal burdens (median VAFs at each time point) are shown as dashed lines on the right with each color representing a different set of mutations. In (D), a stratified Cox proportional hazard function adjusting for age was fit to estimate baseline progression hazard rates and generate PFS curves of patients with AA and AA/PNH (n = 265). Patients were classified on the basis of features (myeloid driver mutations only = group 1; PIGA/PNH clone presence and BCOR/L1 = group 2, and no mutation = group 3) previously hypothesized to be associated with myeloid evolution (also see the Data Supplement; numbers at risk are indicated below the curves). Particularly, n = 7 patients (one evolved to secondary myeloid neoplasia) with concomitant PIGA/PNH clone and myeloid mutations were considered as belonging to group 2. Pairwise comparisons: group 1 versus group 2, P < .001; group 1 versus group 3, P = .026; and group 2 versus group 3, P = .025, log-rank test. The bar chart on the lower right illustrates the difference in age at onset in an analysis of patients with pure AA phenotype (n = 58). AA, aplastic anemia; CR, complete response; HLA, human leukocyte antigen; IST, immunosuppressive therapy; NR, no response; ns, not significant; PFS, progression-free survival; PNH, paroxysmal nocturnal hemoglobinuria; PR, partial response; VAF, variant allele frequency.