Leukemia (2015), 1–11 © 2015 Macmillan Publishers Limited All rights reserved 0887-6924/15 www.nature.com/leu
ORIGINAL ARTICLE
High concordance of genomic and cytogenetic aberrations between peripheral blood and bone marrow in myelodysplastic syndrome (MDS) AM Mohamedali1,2,3, J Gäken1,3, M Ahmed1,2, F Malik1, AE Smith1,2, S Best1,2, S Mian1, T Gaymes1,2, R Ireland2, AG Kulasekararaj1,2 and GJ Mufti1,2 Bone marrow (BM) genetic abnormalities in myelodysplastic syndrome (MDS) have provided important biological and prognostic information; however, frequent BM sampling in older patients has been associated with significant morbidity. Utilizing singlenucleotide polymorphism array (SNP-A) and targeted gene sequencing (TGS) of 24 frequently mutated genes in MDS, we assessed the concordance of genetic abnormalities in BM and peripheral blood (PB) samples concurrently from 201 MDS patients. SNP-A karyotype in BM was abnormal in 108 (54%) and normal in 93 (46%) patients, with 95% (190/201) having an identical PB karyotype. The median copy number (CN) for deletions was significantly lower in BM (CN:1.4 (1–1.9)) than in PB (CN:1.5 (1–1.95), P o 0.001). Using TGS, 71% (130/183) patients had BM somatic mutations with 95% (124/130) having identical mutations in PB. The mutant allele burden was lower in PB (median 27% (1–96%)) compared with BM (median 29% (1–100%); P = 0.14) with no significant difference in the number, types of mutations or World Health Organization subtype. In all patients with discordant SNP (n = 11) and mutation (n = 6) profiles between BM and PB, shared abnormalities were always present irrespective of treatment status. Overall, 86% of patients had a clonal aberration with 95% having identical SNP-A karyotype and mutations in PB, thus enabling frequent assessment of response to treatment and disease evolution especially in patients with fibrotic or hypocellular marrows. Leukemia advance online publication, 5 June 2015; doi:10.1038/leu.2015.110
INTRODUCTION The heterogeneity of myelodysplastic syndrome (MDS) manifests in morphology, cytogenetic abnormalities and depth of cytopenia, all affect the natural history and rate of progression of the disease.1,2 The morphological assessment of peripheral blood (PB) and bone marrow (BM) dysplasia, although critical, is subject to wide inter-observer variation, especially in the absence of objectively defined abnormal cell populations, for example, ring sideroblasts or myeloblasts.1 Cytogenetic abnormalities detected by metaphase cytogenetics (MC) are fundamental to disease assessment and prognosis3,4 and are also extremely important as surrogate markers of clonality. However, in approximately 50% of cases, MC is normal and therefore not informative as a marker of clonality.2 Additionally, poor quality of BM sample, usually due to hypocellularity or fibrosis (disease or treatment associated) or failure of successful culture/metaphase yield can preclude an accurate morphological and cytogenetic analysis in about 10–15% of cases. These situations are regularly encountered in clinical practice and pose diagnostic challenges. Furthermore, although BM provides accurate and vital information on diagnosis, progression and response to therapy, it is an invasive procedure that can cause pain and discomfort, especially in the elderly, preventing regular/frequent sampling to monitor disease status. The development of novel disease-modifying agents for MDS again necessitates the need for frequent BM sampling to assess response to targeted therapies.
Recent molecular techniques such as single-nucleotide polymorphism array (SNP-A) karyotyping have complemented MC by their ability to detect cryptic unbalanced genomic aberrations as well as acquired copy number neutral loss of heterozygosity (CN-LOH) in myeloid disorders.5,6 In addition, high-throughput targeted sequencing has identified a number of somatic mutations in up to 80% of MDS patients.7–13 There is emerging data that mutations affecting specific gene(s) have the potential to add to the diagnosis and prognosis especially in early/transforming stages or following treatment. Patient cohort studies have indeed provided prognostic models based on genetic data derived from sequencing of large numbers of samples from MDS patients.7,14,15 Importantly, all the genetic and cytogenetic data are obtained from BM as the source of DNA. Easier collection of PB provides a clear advantage for acquiring patient samples making it feasible to follow disease progression and response to disease-modifying therapies enabling more frequent sampling. Conventional MC karyotyping from PB is often unsuccessful due to leukopenia, presumed lack of circulating clonal population and impaired growth in cultures. Use of an extended panel of fluorescent in situ hybridization probes on PB mononuclear cells enriched for CD34+ although helpful but is limited to identifying only predefined lesions.16 Our previously published pilot study has shown the utility of molecular assays such as targeted gene sequencing and SNP-A karyotyping, using PB as a surrogate for BM.17 Following on from this preliminary work, we extended the cohort (n = 201) and
1 Department of Haematological Medicine, The Rayne Institute, King’s College London School of Medicine, London, UK and 2Department of Haematology, King’s College Hospital, Department of Haematology, London, UK. Correspondence: Professor GJ Mufti, Department of Haematological Medicine, The Rayne Institute, King’s College London, 123 Coldharbour Lane, London SE5 9NU, UK. Email: [email protected] 3 These authors contributed equally to this work. Received 18 December 2014; revised 24 March 2015; accepted 16 April 2015; accepted article preview online 6 May 2015
Genomic aberrations in myelodysplasia AM Mohamedali et al
2 have optimized a sequencing gene panel targeting the 24 frequently mutated genes in MDS and a high-resolution SNP array for karyotyping to determine whether genetic abnormalities in the BM are reflected in the PB thus enabling easy assessment of response to treatment and/or disease progression. METHODS Patients MDS patients were selected for SNP-A karyotyping (n = 201) and targeted gene mutation analysis (n = 183) using concurrent blood and BM samples (Table 1). These patients were seen at King’s College Hospital from June 2005 to March 2014 and enrolled in this study. All patients provided written informed consent in accordance with National Research Ethics Protocol.
SNP array Genomic DNA from 201 patients was extracted from 200 μl of concurrent samples of PB and BM using the DNA Extraction Kit (Qiagen, Manchester, UK) and processed for SNP array analysis using the Affymetrix CytoHD and Cyto750 K platforms (Affymetrix, Woodburn Green, UK), all as per the manufacturer’s protocol. SNP array data was quality checked by the following parameters; Waviness SD o0.12, SNPQC415 and MAPDo0.25. Array data not satisfying these criteria were visually inspected to determine quality and repeated if necessary. Affymetrix CEL files were processed using Affymetrix Genotyping Console 3.0 and analysed using Chromosome Analysis Suite (ChAS v2.0, Affymetrix). CN aberrations having 450% overlap with variants present in the database of genomic variants and a reference collated from 1000 healthy subjects (Affymetrix) were excluded from further analysis. Regions of CN-LOH were excluded if they were o20 Mb and located interstitial in the chromosome. Regions of CN-LOH that included the telomeric end were included for analysis, irrespective of size.6
Mutation detection High-throughput sequencing assay targeting either the entire coding region or known hotspots for 24 genes: ASXL1, CBL, CEBPA, DNMT3A, ETV6, EZH2, FLT3, GATA2, IDH1, IDH2, JAK2, KDM6A, KIT, KRAS, NPM1, NRAS, RUNX1, SF3B1, SRSF2, STAG2, TET2, TP53, U2AF1 and ZRSR2, was developed (Supplementary Table S1) and performed using the MiSeq Instrument and V3 Sequencing Kits (Illumina, San Diego, CA, USA), as per the manufacture's protocol. A total of 389 amplicons were designed with a median length of 206 bp (range 150–240 bp). The technical reproducibility of the panel was assessed by sequencing 17 samples (BM = 9, PB = 8) and comparing the results to the TruSight Myeloid Sequencing Panel (Illumina) and showed no difference in the sensitivity of mutation detection (Supplementary Table S2). The sequencing library was constructed from 300 ng genomic DNA per patient resulting in a median of 7.6 million paired-end reads per run that was processed using the GATK pipeline package (Broad Institute, Boston, MA, USA). Synonymous/non-synonymous variants and SNPs present at 41% in dbSNP 137 and 40.1% in ESP6500 databases were excluded. The remaining variants had a median depth of 839 independent reads (25–13 559) and were included for further analysis if they had an allele burden of 410%, present in COSMIC (Catalogue of Somatic Mutations in Cancer) or had been previously reported. Novel missense variants were included if the read depth was 4300 in both BM and PB and at a frequency of o0.001% in dbSNP (143). Variants with reads o 100 were manually verified using the Broad Institute Integrative Genomic Browser.
Statistical analysis Statistical calculations were performed using SPSS version 22 (SPSS Inc., Portsmouth, UK). A P-value of ⩽ 0.05 was considered statistically significant.
RESULTS Clinical variables of the patient cohort The demographic and clinical characteristics of the 201 patients that were included in this study are detailed in Table 1. All patients were risk stratified at diagnosis according to the International Leukemia (2015), 1 – 11
Table 1. Demographic details of the 201 patients with myelodysplastic syndromes Total cohort (n = 201), N (%) Sex, n (%) Male Female Age at sampling, years Median Range WHO classification RA RARS 5q- syndrome RCMD RAEB-1 RAEB-2 MDS-U AML tMDS/AML CMML MPD/MDS
106 (53) 95 (47) 63 (17–88)
8 12 26 75 24 17 12 10 6 7 4
(4) (6) (13) (37) (12) (8) (6) (5) (3) (4) (2)
BM metaphase cytogenetics Normal Abnormal Failed
113 (56) 65 (32) 23 (12)
IPSS cytogenetic risk groups Good Intermediate Poor Failed
139 14 24 23
Cytopenias Hb (g/dl), median Hb range ANC (×103/μl), median ANC range PLT (×103/μl), median PLT range Bone marrow blasts o5% 5–10% 11–19% 420%
(69) (7) (12) (12)
10.6 6.1–16 1.69 0–41 126 6–776 140 30 17 14
(70) (15) (8) (7)
Therapy BSC Lenalidomide 5-Azacytidine AML-type chemotherapy Allogeneic HSCT Others
60 28 35 25 42 11
(30) (14) (17) (12) (21) (6)
Follow-up (months) Median Range
21 0.3–106
Molecular genotyping (PB and BM) SNP-A karyotyping Targeted mutational analysis
201 (100) 183 (91)
Abbreviations: AML, acute myeloid leukaemia; ANC, absolute neutrophil count; BM, bone marrow; BSC, best supportive care; CMML, chronic myelomonocytic leukaemia; Hb, haemoglobin; HSCT, haemopoietic stem cell transplantation; IPSS, International Prognostic Scoring System; MDS-U, myelodysplastic syndrome undefined; MPD, myeloproliferative disorders; PB, peripheral blood; PLT, platelet; RA, refractory anaemia; RAEB, refractory anaemia with excess blasts; RARS, refractory anaemia with ringed sideroblasts; RCMD, refractory anaemia with multilineage dysplasia; SNPA, single-nucleotide polymorphism array; tMDS/AML, therapy-related myeloid neoplasms; WHO, World Health Organization.
© 2015 Macmillan Publishers Limited
Genomic aberrations in myelodysplasia AM Mohamedali et al
3 Prognostic Scoring System (IPSS) and clinical variables, including French–American–British type and World Health Organization (WHO) subtype, were ascertained both at the presentations and at the time of sample collection. The cohort was followed up to March 2014 for recording survival and other disease characteristics. The survival data for patients who underwent allogeneic hematopoietic stem cell transplantation (n = 42, 21%) was censored on the day of the transplant and other treatments received is annotated in Table 1. The median age of the cohort was 63 years (17–88) and was followed up for a median of 21 months (0.3–106 months). Sixty (30%) patients received supportive care only while others were treated with lenalidomide 28 (14%), 5-azacytidine 35 (17%) and acute myeloid leukaemia chemotherapy 25 (12%). The WHO subtypes at time of sampling were: refractory anaemia (n = 8), 5q- syndrome (n = 26), refractory anaemia with ringed sideroblasts (n = 12), refractory cytopenia with multilineage dysplasia (RCMD, n = 75, including 11 cases of hypoplastic MDS), refractory anaemia with excess blasts (RAEB, n = 41), acute myeloid leukaemia (n = 10), therapy-related myeloid neoplasms (n = 6), MDS-undefined (n = 12) and myeloproliferative/MDS overlap (n = 11). MC analysis of BM identified 113/201 (56%) with a
a SNP-A NK - 9 SNP-A AK -14
SNP-A NK- 7 SNP-A AK- 58
b
normal karyotype (NK-MC), 65 (32%) with an abnormal karyotype (AK-MC) and 23 (12%) where MC failed. SNP-A complements current routine techniques SNP-A was informative for all 201 patients (for both PB and BM samples), normal (NK-SNP) in 93 (46%) and abnormal (AK-SNP) karyotype in 108 (54%) (Figure 1a). SNP-A identified cryptic chromosomal abnormalities in 36 (32%) and in 58 (88%) patients with NK-MC and AK-MC, respectively. Of the 23 patients where MC failed, 14 (61%) had SNP defects (single abnormality (n = 6), complex karyotype (n = 8)). Balanced translocations were not detected by SNP-A in 8 patients; however, additional concurrent abnormalities detected by BM MC were detectable by SNP-A in 5/8 patients from both BM and PB. SNP-A abnormalities detected in the BM and PB are highly concordant Overall, 95% (190/201) patients had an identical SNP-A profile in the BM and PB. Of the 108 patients with SNP-A abnormalities, 90% (97/108) had an exactly identical SNP karyogram between the BM and PB samples. BM SNP-A identified 396 aberrations and 356
Metaphase Cytogenetics
Single Nucleotide Polymorphism Array
NI 23(12%)
Abnormal 65(32%)
Normal 113(56%)
SNP-A NK - 77 SNP-A AK - 36
Abnormal 108(54%)
Normal 93(46%)
Copy Number Ch.10
BM
P
ORIGINAL ARTICLE
High concordance of genomic and cytogenetic aberrations between peripheral blood and bone marrow in myelodysplastic syndrome (MDS) AM Mohamedali1,2,3, J Gäken1,3, M Ahmed1,2, F Malik1, AE Smith1,2, S Best1,2, S Mian1, T Gaymes1,2, R Ireland2, AG Kulasekararaj1,2 and GJ Mufti1,2 Bone marrow (BM) genetic abnormalities in myelodysplastic syndrome (MDS) have provided important biological and prognostic information; however, frequent BM sampling in older patients has been associated with significant morbidity. Utilizing singlenucleotide polymorphism array (SNP-A) and targeted gene sequencing (TGS) of 24 frequently mutated genes in MDS, we assessed the concordance of genetic abnormalities in BM and peripheral blood (PB) samples concurrently from 201 MDS patients. SNP-A karyotype in BM was abnormal in 108 (54%) and normal in 93 (46%) patients, with 95% (190/201) having an identical PB karyotype. The median copy number (CN) for deletions was significantly lower in BM (CN:1.4 (1–1.9)) than in PB (CN:1.5 (1–1.95), P o 0.001). Using TGS, 71% (130/183) patients had BM somatic mutations with 95% (124/130) having identical mutations in PB. The mutant allele burden was lower in PB (median 27% (1–96%)) compared with BM (median 29% (1–100%); P = 0.14) with no significant difference in the number, types of mutations or World Health Organization subtype. In all patients with discordant SNP (n = 11) and mutation (n = 6) profiles between BM and PB, shared abnormalities were always present irrespective of treatment status. Overall, 86% of patients had a clonal aberration with 95% having identical SNP-A karyotype and mutations in PB, thus enabling frequent assessment of response to treatment and disease evolution especially in patients with fibrotic or hypocellular marrows. Leukemia advance online publication, 5 June 2015; doi:10.1038/leu.2015.110
INTRODUCTION The heterogeneity of myelodysplastic syndrome (MDS) manifests in morphology, cytogenetic abnormalities and depth of cytopenia, all affect the natural history and rate of progression of the disease.1,2 The morphological assessment of peripheral blood (PB) and bone marrow (BM) dysplasia, although critical, is subject to wide inter-observer variation, especially in the absence of objectively defined abnormal cell populations, for example, ring sideroblasts or myeloblasts.1 Cytogenetic abnormalities detected by metaphase cytogenetics (MC) are fundamental to disease assessment and prognosis3,4 and are also extremely important as surrogate markers of clonality. However, in approximately 50% of cases, MC is normal and therefore not informative as a marker of clonality.2 Additionally, poor quality of BM sample, usually due to hypocellularity or fibrosis (disease or treatment associated) or failure of successful culture/metaphase yield can preclude an accurate morphological and cytogenetic analysis in about 10–15% of cases. These situations are regularly encountered in clinical practice and pose diagnostic challenges. Furthermore, although BM provides accurate and vital information on diagnosis, progression and response to therapy, it is an invasive procedure that can cause pain and discomfort, especially in the elderly, preventing regular/frequent sampling to monitor disease status. The development of novel disease-modifying agents for MDS again necessitates the need for frequent BM sampling to assess response to targeted therapies.
Recent molecular techniques such as single-nucleotide polymorphism array (SNP-A) karyotyping have complemented MC by their ability to detect cryptic unbalanced genomic aberrations as well as acquired copy number neutral loss of heterozygosity (CN-LOH) in myeloid disorders.5,6 In addition, high-throughput targeted sequencing has identified a number of somatic mutations in up to 80% of MDS patients.7–13 There is emerging data that mutations affecting specific gene(s) have the potential to add to the diagnosis and prognosis especially in early/transforming stages or following treatment. Patient cohort studies have indeed provided prognostic models based on genetic data derived from sequencing of large numbers of samples from MDS patients.7,14,15 Importantly, all the genetic and cytogenetic data are obtained from BM as the source of DNA. Easier collection of PB provides a clear advantage for acquiring patient samples making it feasible to follow disease progression and response to disease-modifying therapies enabling more frequent sampling. Conventional MC karyotyping from PB is often unsuccessful due to leukopenia, presumed lack of circulating clonal population and impaired growth in cultures. Use of an extended panel of fluorescent in situ hybridization probes on PB mononuclear cells enriched for CD34+ although helpful but is limited to identifying only predefined lesions.16 Our previously published pilot study has shown the utility of molecular assays such as targeted gene sequencing and SNP-A karyotyping, using PB as a surrogate for BM.17 Following on from this preliminary work, we extended the cohort (n = 201) and
1 Department of Haematological Medicine, The Rayne Institute, King’s College London School of Medicine, London, UK and 2Department of Haematology, King’s College Hospital, Department of Haematology, London, UK. Correspondence: Professor GJ Mufti, Department of Haematological Medicine, The Rayne Institute, King’s College London, 123 Coldharbour Lane, London SE5 9NU, UK. Email: [email protected] 3 These authors contributed equally to this work. Received 18 December 2014; revised 24 March 2015; accepted 16 April 2015; accepted article preview online 6 May 2015
Genomic aberrations in myelodysplasia AM Mohamedali et al
2 have optimized a sequencing gene panel targeting the 24 frequently mutated genes in MDS and a high-resolution SNP array for karyotyping to determine whether genetic abnormalities in the BM are reflected in the PB thus enabling easy assessment of response to treatment and/or disease progression. METHODS Patients MDS patients were selected for SNP-A karyotyping (n = 201) and targeted gene mutation analysis (n = 183) using concurrent blood and BM samples (Table 1). These patients were seen at King’s College Hospital from June 2005 to March 2014 and enrolled in this study. All patients provided written informed consent in accordance with National Research Ethics Protocol.
SNP array Genomic DNA from 201 patients was extracted from 200 μl of concurrent samples of PB and BM using the DNA Extraction Kit (Qiagen, Manchester, UK) and processed for SNP array analysis using the Affymetrix CytoHD and Cyto750 K platforms (Affymetrix, Woodburn Green, UK), all as per the manufacturer’s protocol. SNP array data was quality checked by the following parameters; Waviness SD o0.12, SNPQC415 and MAPDo0.25. Array data not satisfying these criteria were visually inspected to determine quality and repeated if necessary. Affymetrix CEL files were processed using Affymetrix Genotyping Console 3.0 and analysed using Chromosome Analysis Suite (ChAS v2.0, Affymetrix). CN aberrations having 450% overlap with variants present in the database of genomic variants and a reference collated from 1000 healthy subjects (Affymetrix) were excluded from further analysis. Regions of CN-LOH were excluded if they were o20 Mb and located interstitial in the chromosome. Regions of CN-LOH that included the telomeric end were included for analysis, irrespective of size.6
Mutation detection High-throughput sequencing assay targeting either the entire coding region or known hotspots for 24 genes: ASXL1, CBL, CEBPA, DNMT3A, ETV6, EZH2, FLT3, GATA2, IDH1, IDH2, JAK2, KDM6A, KIT, KRAS, NPM1, NRAS, RUNX1, SF3B1, SRSF2, STAG2, TET2, TP53, U2AF1 and ZRSR2, was developed (Supplementary Table S1) and performed using the MiSeq Instrument and V3 Sequencing Kits (Illumina, San Diego, CA, USA), as per the manufacture's protocol. A total of 389 amplicons were designed with a median length of 206 bp (range 150–240 bp). The technical reproducibility of the panel was assessed by sequencing 17 samples (BM = 9, PB = 8) and comparing the results to the TruSight Myeloid Sequencing Panel (Illumina) and showed no difference in the sensitivity of mutation detection (Supplementary Table S2). The sequencing library was constructed from 300 ng genomic DNA per patient resulting in a median of 7.6 million paired-end reads per run that was processed using the GATK pipeline package (Broad Institute, Boston, MA, USA). Synonymous/non-synonymous variants and SNPs present at 41% in dbSNP 137 and 40.1% in ESP6500 databases were excluded. The remaining variants had a median depth of 839 independent reads (25–13 559) and were included for further analysis if they had an allele burden of 410%, present in COSMIC (Catalogue of Somatic Mutations in Cancer) or had been previously reported. Novel missense variants were included if the read depth was 4300 in both BM and PB and at a frequency of o0.001% in dbSNP (143). Variants with reads o 100 were manually verified using the Broad Institute Integrative Genomic Browser.
Statistical analysis Statistical calculations were performed using SPSS version 22 (SPSS Inc., Portsmouth, UK). A P-value of ⩽ 0.05 was considered statistically significant.
RESULTS Clinical variables of the patient cohort The demographic and clinical characteristics of the 201 patients that were included in this study are detailed in Table 1. All patients were risk stratified at diagnosis according to the International Leukemia (2015), 1 – 11
Table 1. Demographic details of the 201 patients with myelodysplastic syndromes Total cohort (n = 201), N (%) Sex, n (%) Male Female Age at sampling, years Median Range WHO classification RA RARS 5q- syndrome RCMD RAEB-1 RAEB-2 MDS-U AML tMDS/AML CMML MPD/MDS
106 (53) 95 (47) 63 (17–88)
8 12 26 75 24 17 12 10 6 7 4
(4) (6) (13) (37) (12) (8) (6) (5) (3) (4) (2)
BM metaphase cytogenetics Normal Abnormal Failed
113 (56) 65 (32) 23 (12)
IPSS cytogenetic risk groups Good Intermediate Poor Failed
139 14 24 23
Cytopenias Hb (g/dl), median Hb range ANC (×103/μl), median ANC range PLT (×103/μl), median PLT range Bone marrow blasts o5% 5–10% 11–19% 420%
(69) (7) (12) (12)
10.6 6.1–16 1.69 0–41 126 6–776 140 30 17 14
(70) (15) (8) (7)
Therapy BSC Lenalidomide 5-Azacytidine AML-type chemotherapy Allogeneic HSCT Others
60 28 35 25 42 11
(30) (14) (17) (12) (21) (6)
Follow-up (months) Median Range
21 0.3–106
Molecular genotyping (PB and BM) SNP-A karyotyping Targeted mutational analysis
201 (100) 183 (91)
Abbreviations: AML, acute myeloid leukaemia; ANC, absolute neutrophil count; BM, bone marrow; BSC, best supportive care; CMML, chronic myelomonocytic leukaemia; Hb, haemoglobin; HSCT, haemopoietic stem cell transplantation; IPSS, International Prognostic Scoring System; MDS-U, myelodysplastic syndrome undefined; MPD, myeloproliferative disorders; PB, peripheral blood; PLT, platelet; RA, refractory anaemia; RAEB, refractory anaemia with excess blasts; RARS, refractory anaemia with ringed sideroblasts; RCMD, refractory anaemia with multilineage dysplasia; SNPA, single-nucleotide polymorphism array; tMDS/AML, therapy-related myeloid neoplasms; WHO, World Health Organization.
© 2015 Macmillan Publishers Limited
Genomic aberrations in myelodysplasia AM Mohamedali et al
3 Prognostic Scoring System (IPSS) and clinical variables, including French–American–British type and World Health Organization (WHO) subtype, were ascertained both at the presentations and at the time of sample collection. The cohort was followed up to March 2014 for recording survival and other disease characteristics. The survival data for patients who underwent allogeneic hematopoietic stem cell transplantation (n = 42, 21%) was censored on the day of the transplant and other treatments received is annotated in Table 1. The median age of the cohort was 63 years (17–88) and was followed up for a median of 21 months (0.3–106 months). Sixty (30%) patients received supportive care only while others were treated with lenalidomide 28 (14%), 5-azacytidine 35 (17%) and acute myeloid leukaemia chemotherapy 25 (12%). The WHO subtypes at time of sampling were: refractory anaemia (n = 8), 5q- syndrome (n = 26), refractory anaemia with ringed sideroblasts (n = 12), refractory cytopenia with multilineage dysplasia (RCMD, n = 75, including 11 cases of hypoplastic MDS), refractory anaemia with excess blasts (RAEB, n = 41), acute myeloid leukaemia (n = 10), therapy-related myeloid neoplasms (n = 6), MDS-undefined (n = 12) and myeloproliferative/MDS overlap (n = 11). MC analysis of BM identified 113/201 (56%) with a
a SNP-A NK - 9 SNP-A AK -14
SNP-A NK- 7 SNP-A AK- 58
b
normal karyotype (NK-MC), 65 (32%) with an abnormal karyotype (AK-MC) and 23 (12%) where MC failed. SNP-A complements current routine techniques SNP-A was informative for all 201 patients (for both PB and BM samples), normal (NK-SNP) in 93 (46%) and abnormal (AK-SNP) karyotype in 108 (54%) (Figure 1a). SNP-A identified cryptic chromosomal abnormalities in 36 (32%) and in 58 (88%) patients with NK-MC and AK-MC, respectively. Of the 23 patients where MC failed, 14 (61%) had SNP defects (single abnormality (n = 6), complex karyotype (n = 8)). Balanced translocations were not detected by SNP-A in 8 patients; however, additional concurrent abnormalities detected by BM MC were detectable by SNP-A in 5/8 patients from both BM and PB. SNP-A abnormalities detected in the BM and PB are highly concordant Overall, 95% (190/201) patients had an identical SNP-A profile in the BM and PB. Of the 108 patients with SNP-A abnormalities, 90% (97/108) had an exactly identical SNP karyogram between the BM and PB samples. BM SNP-A identified 396 aberrations and 356
Metaphase Cytogenetics
Single Nucleotide Polymorphism Array
NI 23(12%)
Abnormal 65(32%)
Normal 113(56%)
SNP-A NK - 77 SNP-A AK - 36
Abnormal 108(54%)
Normal 93(46%)
Copy Number Ch.10
BM
P
