Page 242 - 2021_02-Haematologica-web

M.M. Ouseph et al.
lymphomas/leukemias, often in the context of other chro- mosomal abnormalities.2,6-8 LOY is one of the most com- mon recurrent cytogenetic abnormalities seen in MDS, with an incidence of up to 30% in males.3,5,8-13 The inci- dence of isolated LOY in MDS is lower, ranging from 4 to 10%.8,14 Although there is no definitive association between LOY and clinical or biological characteristics in MDS, some groups have reported a higher incidence of ring sideroblasts, lower bone marrow myeloid to ery- throid ratio, blast proportion, and blood leukocyte count in MDS patients with LOY than in those without LOY.9 MDS patients with isolated LOY have longer overall and leukemia-free survival compared to MDS patients with normal cytogenetics; conversely, in AML, chronic myeloid leukemia and chronic lymphocytic leukemia, isolated LOY has been associated with a poorer prognosis.2,14-20 These observations suggest that MDS with LOY might be associated with a unique biology.
In individual cases, it can be difficult to distinguish whether LOY in MDS is a disease-associated alteration or just incidental aging-associated somatic mosaicism. In cur- rent World Health Organization (WHO) diagnostic crite- ria, LOY by itself cannot be used to define MDS, unlike other more strongly MDS-associated cytogenetic aberra- tions such as del(5q) or del(7q).21 Although the incidence of LOY increases with age, the percentage of metaphases with LOY in healthy people typically does not increase with age, so that a high proportion of metaphases may be more likely to represent a marker of disease rather than progressive clonal expansion of an inconsequential LOY cell population during aging.5,22 Some studies have suggest- ed that 75-100% of metaphases with LOY in bone mar- row are likely to be disease-associated rather than being age-associated.2,23,24
Recently, large genome-wide association studies enrolling more than 200,000 men in the UK Biobank demonstrated an association of germline single nucleotide polymorphisms of 156 autosomal loci as well as differen- tial methylation of various genes with LOY.4,25-27 Whether any of the LOY-associated single nucleotide polymor- phisms or epigenetic changes contribute to a risk of hema- tologic neoplasia is unclear.
While LOY involving a higher proportion of metaphases has been consistently associated with MDS, it is uncertain whether there is a corresponding increased incidence of pathogenic MDS-associated mutations, which typically affect genes involved in DNA methylation, DNA damage repair, chromatin modification, RNA processing, tran- scription and signal transduction.28 In addition, although karyotype evolution has been reported in patients with LOY over time, the rate at which patients with LOY accu- mulate additional mutations is unknown.23 The goal of this study was to evaluate the landscape of somatic muta- tions in patients with LOY.
Methods
The cytogenetic databases of Brigham and Women’s Hospital and Massachusetts General Hospital (Boston, MA, USA) were searched for patients with isolated LOY cytogenetic abnormality on conventional karyotyping of bone marrow cells (defined as ≥3 metaphases with isolated LOY) reported between 01/2005 and 08/2018. Patients with lymphoid or plasma cell malignancies involving bone marrow and those who had received chemother-
apy within the preceding 3 months were excluded, as were those without a matching marrow specimen for morphological evalua- tion.
Medical records were reviewed to identify age at presentation, clinical diagnosis, and clinical course including treatment. Marrow aspirate smears and biopsy cores were re-evaluated by two hematopathologists (OW and MO) and were classified according to the revised 2017 WHO classification of myeloid neoplasms.21 Marrow samples that did not fulfill WHO-defined diagnostic criteria were separated into cases with minimal dys- plasia (affecting <10% cells in any lineage) and no dysplasia.
Fresh marrow aspirates at the time of initial presentation of LOY or concurrent blood samples were subjected to next-gener- ation sequencing (NGS)-based molecular analysis for genes asso- ciated with hematologic neoplasms. For marrow samples that had not undergone NGS for clinical indications, NGS was per- formed on DNA extracted from the corresponding archived frozen cytogenetic cell pellets. NGS panels including 95 (Rapid Heme Panel of Brigham and Women’s Hospital) or 103 (Heme SnapShot Panel of Massachusetts General Hospital) genes of importance in myeloid neoplasia (hotspots in oncogenes and most of the coding regions of tumor suppressors) (Online Supplementary Table S1) were used for mutation analysis. The for- mer is an amplicon-based approach using an Illumina Truseq Custom Amplicon kit (San Diego, CA, USA), with an average depth of 1500x.29 The latter is an anchored multiplex polymerase chain reaction-based panel (Archer DX, Boulder, CO, USA) with an average demultiplexed coverage of 350x.30 After filtering for recurrent assay specific false positives, single nucleotide variants and small insertions/deletions at an allele fraction of ≥0.05 were filtered based on population databases to eliminate likely germline variants. Somatic variants were further classified based on information available in literature, in silico tools, ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/), Catalog of Somatic Mutations in Cancer (https://cancer.sanger.ac.uk/cosmic) and pub- lished guidelines for the interpretation and reporting of sequence variants in cancer.31 Only likely pathogenic and pathogenic vari- ants identified in patients’ samples were analyzed for this study.
The selection of patients and analysis of clinical/laboratory results for this study were approved by the Partners Healthcare Institutional Review Board (Protocol #: 2009P001369). Statistical analysis was performed using JMP®, version 14 (SAS Institute Inc., Cary, NC, USA). A P-value of <0.05 was considered statisti- cally significant. Individual P-values and tests performed are pro- vided in the Results section. Briefly, in addition to descriptive sta- tistics, analysis of distribution of categorical variables were per- formed using the Fisher exact test or Pearson χ2 test. Continuous variables were compared by one-way analysis of variance (ANOVA) or the Wilcoxon/Kruskal-Wallis test with χ2 approxi- mation. A log-rank test was used to compare progression among patients with long-term follow-up. Multivariate analysis with a row-wise method for estimation of correlation and logistic regression analysis were used to compare predictor variables; and generate correlation coefficients, confidence intervals and corre- lation probabilities.
Results
Identification of cases
We identified 106 LOY cases reported as an isolated cytogenetic abnormality from 88 patients (Online Supplementary Figure S1A). Of these 106 samples, 15 were subsequently excluded from analysis because of lack of molecular analysis data (i.e., they failed the quality control
556
haematologica | 2021; 106(2)