Acquired spherocytosis due to somatic ANK1 mutations as a manifestation of clonal hematopoiesis in elderly patients.

Abstract

To the Editor: In western countries, the main cause of acquired hemolytic anemia is autoimmune hemolytic anemia (AIHA). A few cases of acquired hemolytic membrane disorders, red blood cell (RBC) enzymatic deficiencies or thalassemia-like acquired syndromes have been described so far. Most of them occurred in patients with myelodysplatic syndrome (MDS) or myeloproliferative neoplasm (MPN).1, 2 The most frequent acquired membrane disorder is elliptocytosis associated with del20q in patients with MDS. One case of acquired spherocytosis due to a large deletion encompassing the SPTB gene has been described in a 73-year-old patient with MDS, bone marrow karyotype showing complex rearrangements on Chromosomes 5 and 14. Somatic mosaicism for a G6PD mutation was found in a 65-year-old patient with acquired G6PD deficiency. Acquired alpha-thalassemia (Hemoglobin H disease) has been described in MDS and MPN and is caused by somatic mutations in the ATRX gene. We identified “pseudo-constitutional” somatic mutations of the ANK1 gene in blood and bone marrow of two unrelated patients who developed hemolytic anemia with spherocytosis at an old age. They were initially considered as having AHAI but had negative direct antiglobulin test (DAT) and did not respond to corticosteroids or immunomodulator treatment. Patient 1 was a 91-year-old man with no personal or family history of anemia (Figure 1A). He had a history of hypertension and chronic kidney disease. At 85 years of age, he developed regenerative hemolytic anemia with splenomegaly. White blood cells and platelet counts were normal. Blood smear showed numerous spherocytes and no schizocytes (Figure 1B). DAT (with screening for monospecific anti-IgA and IgM antibodies) was negative. Paroxysmal nocturnal hemoglobinuria (PNH) was ruled out by flow cytometry analysis. Eosin-5′-maleimide (EMA) binding test showed a significant decrease of fluorescence (22%), suggestive of hereditary spherocytosis (HS). The patient was first treated with corticosteroids and then rituximab for a possible DAT-negative AIHA, with no obvious effect on hemoglobin level. Bone marrow aspirate showed an isolated hyperplasia of the erythroid lineage but no features of dysplasia. Bone marrow karyotype was normal. Targeted next-generation sequencing (NGS) of 37 genes involved in MDS or MPN (Table S1) identified a mutation of DNMT3A (NM_022552): c.2312G>A, p.(Arg771Gln), with a variant allele frequency (VAF) of 48% in blood and 41% in bone marrow. Overall these abnormalities did not lead to the diagnosis of MDS or MPN according to the WHO 2016 classification but rather to a diagnosis of clonal hematopoiesis of indeterminate potential (CHIP).3, 4 His risk score for developing myeloid neoplasms as defined by Rossi et al.5 was 3 (intermediate risk). Patient 2 was a 86-year-old man with no family history of hemolytic anemia (Figure 1A). He had an isolated, unexplained macrocytosis discovered at age 65. He developed lung adenocarcinoma at age 69 and was successfully treated by surgical lobectomy and chemotherapy (cisplatin, vinorelbine, gemcitabine and paclitaxel). At age 85, he developed exertional dyspnea, which revealed a regenerative hemolytic anemia. DAT (with screening for anti-IgA and IgM antibodies) was negative, blood smear showed numerous spherocytes and no schizocytes (Figure 1B). There was no PNH clone. Ektacytometry was suggestive of HS but EMA binding test was inconclusive with a fluorescence decrease of only 2%. DAT-negative AIHA was suspected and he received corticosteroids, with no effect. Bone marrow aspirate showed isolated hyperplasia of the erythroid lineage. Targeted NGS found no pathogenic or likely pathogenic variation in 45 genes involved to MDS and MNP (Table S1). Bone marrow karyotype showed loss of Y chromosome and trisomy 15 in all examined mitoses (n = 19), leading to the diagnosis of age-related clonal hematopoiesis (CH). Both patients gave written informed consent for genetic analysis. For the first-line analysis of genes involved in congenital anemia, DNA was extracted from whole blood and NGS analysis was performed using a targeted capture of genes involved in RBC membrane disorders and congenital dyserythropoietic anemia. Genomic DNA from nonhematopoietic tissues was obtained from a skin biopsy in Patient 1 and from hair bulbs in Patient 2. Mutation analysis on non-hematopoietic tissues was performed using NGS and Sanger sequencing (see supplementary methods in Data S1 for extensive description of genetic analyses). In Patient 1, genetic analysis of blood DNA showed the presence of a new splicing variation in the ANK1 gene (NM_020476): c.2098-1G>T (depth 439X, VAF 36%). This genotype was compatible with HS. Due to the very late onset of the disease and the absence of family history, we checked for the presence of the variation in another tissue and found that it was absent from DNA extracted from skin, confirming the acquired status of this variation. Exome sequencing on bone marrow DNA confirmed the DNMT3A mutation and did not disclose any other mutation in genes involved in CHIP/MDS/MPN. In bone marrow, VAF was 32% for the ANK1 mutation (depth 173X) and 40% for the DNMT3A mutation (depth 169X). This suggests that the clone harboring the DNMT3A mutation represents 80% of the hematopoietic cells (assuming that the DNMT3A mutation is present at the heterozygous state in all clonal cells, as previously reported in most patients6), and that a large majority of cells in this clone harbor the ANK1 mutation (32/40 = 80%). The fact that the ANK1 mutation has a lower VAF than the DNMT3A mutation suggests that the ANK1 “passenger” mutation occurred later during clone development. In Patient 2, genetic analysis of blood DNA showed the presence of a new nonsense variation in the ANK1 gene (NM_020476): c.4098C>A; p.(Cys1366*) (depth 405X, VAF 41%). Again, we checked for the presence of the mutation on another tissue and found that it was absent from DNA extracted from hair bulbs, confirming that it was a somatic mutation. In bone marrow, VAF was 48% for the ANK1 mutation (depth 107X). Since the patient clearly had CH but no driver mutation identified by targeted NGS, we extended the genetic testing and performed comparative exome sequencing on bone marrow and hair bulb DNA. We found two acquired somatic variations in bone marrow: c.9023G>A p.(Arg3008His) in ATM (NM_000051, VAF: 31%, depth 175X) and c.5102C>G, p.(Thr1701Ser) in ANKRD11 (NM_013275, VAF 42%, depth 62X). Both variations are candidate driver mutations of CH in this patient, but with uncertainty about their causal role. ATM has been involved in therapy-related CH in patients with nonhematologic cancers.7 In addition, the c.9023G>A p.(Arg3008His) variation is known to be pathogenic. However, with a VAF of 31%, it is present in only 31/48 = 65% of clonal cells harboring the ANK1 mutation. ANKRD11 is a co-activator of p538 and considered as an actionable cancer target but it has never been involved in CH pathogenesis.7 In elderly patients, Zink et al. found that no more than 40% of cases of age-related CH (identified by whole genome sequencing) could be accounted for by a candidate driver mutation. They suggested that age-related CH could also result from epigenetic changes or even from neutral drift.9 Genetic analysis of DNA from different tissues allowed us to discover acquired ANK1 mutations in bone marrow and blood. The VAFs of these mutations in blood were high and compatible with a hereditary disorder (36% and 41%) because both patients had an age-related CH in which the clonal cell population had become predominant. The presence of an ANK1 mutation in most of the clonal cells led to the production of high levels of spherocytes, with hemolytic anemia as a consequence. Our findings demonstrate that hematopoietic clones can harbor additional “passenger” mutations with detrimental effects, leading to the phenocopy of a hereditary disorder like HS when the clone becomes predominant. Warm AIHA is a classical cause of spherocytic anemia. DAT-negative AIHA was first suspected in both patients but both failed to respond to corticosteroids (and rituximab for Patient 1). DAT is reported to be negative in approximately 5% of patients with AIHA10 but making the diagnosis of DAT-negative AIHA requires to have ruled out other causes of hemolytic anemia (such as PNH, microangiopathy or drug-induced HA and so on). RBC membrane disorders due to somatic mutations in genes encoding membrane proteins are new diagnoses to consider in elderly people with unexplained, DAT-negative spherocytic anemia. The patients have given their consent to genetic explorations and research use of their samples. Lamisse Mansour-Hendili, Frédéric Galactéros, and Benoît Funalot conceived the study, analyzed data, and wrote the manuscript. Edouard Flamarion, Marc Michel, Caroline Morbieu, Bertrand Godeau, Marion Camard, and Loïc Garçon provided patient care, generated clinical data, and revised the manuscript. Christine Gameiro performed NGS experiments and analysis. Ivan Sloma, Sihem Tarfi, Chloé Friedrich, and Olivier Kosmider performed NGS analysis. Bouchra Badaoui, Luc Darnige, and Véronique Picard performed hematological experiments and analysis. Ariane Lunati-Rozie revised the manuscript. Abdelrazak Aissat performed bioinformatic analysis. Nasséra Abermil, Sophie Kaltenbach, and Isabelle Radford-Weiss performed and validated cytogenetic analysis. Pascale Fanen and Pablo Bartolucci revised the manuscript. All authors approved the final version of the manuscript. The authors declare no conflict of interest. Data available on request from the authors Data S1 Supplementary methods Table S1. Genes analyzed in the NGS panels Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.