Comparison of Peripheral Blood and Bone Marrow Molecular Profiling in Primary Myelodysplastic Syndromes (MDS)

AMM and MA – Joint first authors

Diagnosis of myelodysplastic syndromes (MDS) relies on demonstrating peripheral blood (PB) / bone marrow (BM) dysplasia and cytogenetic abnormalities, forming the backbone of the revised International Prognostic Scoring System (IPSS-R). However, assessment of dysplasia is operator dependent and metaphase cytogenetic analysis (MC) is often normal or uninformative creating diagnostic challenges, particularly in patients with early or low risk MDS. Impressive advances have taken place in the last decade in the identification and chronicling of the genetic lesions leading to phenotypic diversity of MDS. We sought to evaluate the value of single nucleotide polymorphism array (SNP-A) based cytogenetic assessment and high throughput sequencing of the 24 genes most frequently mutated in MDS to determine if genetic abnormalities in the BM are reflected in the PB thus enabling easy assessment of response to treatment and/or disease progression.

A MiSeq based gene panel comprising of 24 frequently mutated MDS genes: ASXL1, CBL, CEBPA, DNMT3A, ETV6, EZH2, FLT3, GATA2, IDH1, IDH2, JAK2, KDM6, KIT, KRAS, NPM1, NRAS, RUNX1, SF3B1, SRSF2, STAG2, TET2, TP53, U2AF1, ZRSR2 and CytoHD/750K SNP-A karyotyping was applied to both PB and BM concurrent samples. Genomic aberrations and non-synonymous variant calls were filtered using public databases to exclude polymorphisms.

PB and BM from 201 MDS patients [median 62 years (17-88)] followed up for median 21 months (0.3-171) was analysed. Sixty (30%) patients received supportive care only whilst others were treated with MDS/AML directed therapies. The WHO subtypes were: 5q- syndrome (n=26), refractory cytopenia/s (RA/RCMD, n=53), refractory anaemia with excess blasts/acute myeloid leukaemia (RAEB/AML, n=51), refractory anaemia with ringed sideroblasts (RARS/RCMD-RS, n=20) and other subtypes (including myeloproliferative and hypoplastic MDS, n=51). Based on the IPSS-R risk groups 62% had low risk disease: classified as very low (n=35), low (n=89), intermediate (n=42), high (n=16) and very high (n=19) risk groups. Metaphase cytogenetic analysis was normal (NK-MC) in 113(56%), abnormal (AK-MC) in 65(32%) patients and in 23(12%) MC failed. SNP-A was informative in all patients identifying a normal (NK-SNP) in 93(46%) and abnormal (AK-SNP) in 108(54%) patients, respectively. A comparison of BM and PB by SNP-A, showed 190 patients having an identical karyotype (95% concordance). BM SNP-A identified 36 (32%) patients with SNP-A abnormalities not detected by MC, changing the IPSS-R in 49 (24%) patients overall. Inclusion of SNP-A abnormalities changed the IPSS-R risk group in 28 (25%) of NK-MC patients; from good to intermediate (n=21), poor (n=5) and very poor (n=2) groups, respectively. In 9/11 patients with discordant SNP-A karyotypes between BM and PB, the IPSS-R remained unchanged. We found no difference in the clonal burden between PB and BM for gains and CN-LOH. However, for deletions, the clonal size was significantly lower in BM [median 1.4 (1 – 1.9)], than PB [median 1.5 (1 – 1.95), p<0.001].

Mutational profiling from 183 patients revealed 248 and 240 mutations in BM and PB, respectively. BM and PB mutation profiles were identical in 171 patients (93% concordance) (Fig 1). In 11/12 patients with discordant mutation profiles, additional identical mutations and SNP-A abnormalities were noted in both BM and PB. Similarly, clone size was lower (p= 0.53) in the PB [median 27%, (1%-96%)] compared to BM [median 30%, (1%-100%)] and, in addition, there was no difference in the overall number or the type of mutation between the BM and PB. Interestingly, analysis of known poor risk MDS gene mutations (ASXL1, EZH2, RUNX1, P53 and U2AF1) identified 85 mutations in BM compared to 80 in the PB. In 5 patients where these mutations were discordant, additional identical mutations were detected in both BM and PB. Patients with ASXL1 and TP53 mutations had significantly lower clonal burden in PB (p<0.001) compared to BM in contrast to RUNX1 and EZH2 where there was no difference in the clone size. Using PB we were able to show that both SNP-A defects and gene mutations were efficiently detected when compared with BM.

This study confirms that PB can reliably substitute for BM for all molecular assays in MDS; which is highly relevant for the clinical assessment of this relatively elderly group of patients.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal