With the advent of high throughput and high resolution techniques, >80% of myelodysplastic syndrome (MDS) patients harbour somatic mutations and/or genomic aberrations, which provide diagnostic and prognostic utility; however, frequent bone marrow (BM) aspirates are required. In a significant minority, the BM is hypocellular and fibrotic with suboptimal in vitro growth and, additionally, the procedure causes discomfort particularly in the elderly. This led us to investigate the use of peripheral blood (PB) and serum to identify and monitor BM derived genetic markers using high resolution single nucleotide polymorphism array (SNP-A) karyotyping and 454 parallel sequencing (454-PS) of a 22 gene myeloid panel comprising of all the exons of DNMT3a, RUNX1, CEBPα, TP53, EZH2, TET2 and ZRSR2 and mutations ‘hotposts’ for NPM1, FLT3, ASXL1, IDH1, IDH2, MPL, JAK2, BRAF, cCBL, NRAS, KRAS, C-KIT, SF3B1, SRSF2, and U2AF1.
We selected 23 MDS patients with concurrent BM and PB samples and detected 45 mutations in TET2, SF3B1, ASXL1, TP53, DNMT3a, FLT3, U2AF1, NRAS, cCBL, JAK2 and IDH2 in BM and subsequently analysed their PB using 454-PS with independent validation performed by Sanger sequencing (SS). All the mutations identified in BM were detected in PB with the exception of a single NRAS (BM-11%). Nine patients had a single mutation in SF3B1, ASXL1, TP53, TET2, DNMT3a and U2AF1 with the remaining patients having multiple coexisting mutations. Overall there was no significant difference in the mutation burden between the PB (median 25%(1.5%-50%)) and the BM (median 33%(5%-68%)). Concurrent analysis of unamplified and whole genome amplified PB DNA from 3 patients with mutations in TET2, U2AF1, ASXL1 and NRAS showed no difference in their mutation profile. As expected, in three patients post therapy the mutation burden in the PB was lower than in the pre-treatment BM sample.
The lower mutation burden and sequence context in the PB contributed significantly to the quality of SS analysis. Prior knowledge of the mutation site resulted in 98% concordance (smallest clone size - 1.5%) between BM and PB, however, a blind approach decreased this to 84% (smallest clone size - 15%). In addition, serum was available from 14 patients (22 known mutations in BM) and SS of serum DNA identified 12 mutations correctly (U2AF1, FLT3, SF3B1, TET2, TP53, ASXL1, DNMT3a and IDH2, 4 as wildtype (TET2, cCBL, ASXL1 and SF3B1) and 6 failed to amplify (ASXL1, TP53 and TET2) without any preference for specific genes and the failure attributed to the quantity of serum DNA.
Karyotype aberrations in PB were assessed using Affymetrix SNP 6.0 arrays on 31 MDS patients; normal karyotype (n=11), del5q (n=9), del7q/-7 (n=5), trisomy 8 (n=2), complex (n=2), isodiXq13 (n=1) and t(2:4)(q33;q27). An overall karyotype concordance of 94% was observed in PB with the 2 discordant cases showing normal karyotype in PB and BM by SNP-A but having monosomy 7 (partial cytogenetic remission after 5-azacitidine) and t(2:4)(q33;q27) respectively in their BM by MC. The mean copy number (CN) in the PB was lower than BM (PB vs BM); deletions (CN of 1.8 vs 1.6) and gains(CN of 2.2 vs 2.4), implying a smaller abnormal clone in the PB. Concurrent SNP-A karyotype from BM from 9 patients was concordant with SNP-A karyotype from PB, although a patient with complex karyotype determined by SNP-A in the BM and having 30 aberrations had only 15 aberrations detected in the PB.
To determine if the 5q deletion was lineage restricted, we enriched PB for CD3+, CD19+ and CD3-CD19- populations from 4 patients. FISH and SNP-A karyotyping showed the presence of the 5q deletion using both techniques in all three fractions, however at a lower level in PB lymphocytes indicating the presence of a smaller clone.
In conclusion, our study showed an excellent concordance between BM and PB, both for karyotype and mutational analyses using high resolution SNP-A karyotyping and 454-PS suggesting its clinical utility as a surrogate for BM, thereby, avoiding the discomfort of repeated BM aspirates and help in monitoring response to therapy more frequently.
No relevant conflicts of interest to declare.
HAA and AMM contributed equally