Utility of peripheral blood for cytogenetic and mutation analysis in myelodysplastic syndrome

The myelodysplastic syndromes (MDSs) are clonal disorders of hematopoiesis that occur predominantly in the elderly (median age 72 years) and are characterized by morphologic dysplasia, ineffective hematopoiesis, peripheral blood (PB) cytopenias, chromosomal aberrations, and propensity to myeloid leukemic transformation. The advent of high-throughput and high-resolution techniques for genetic analysis has shown that more than 80% of MDS patients harbor somatic mutations and/or genomic aberrations in their bone marrow (BM), which provide pathogenetic as well as diagnostic and prognostic insights into this disease.^1-4  Frequent BM aspirates may be required for morphological⁵  and genetic assessment, especially after BM transplant. In addition, in a significant patient proportion, the BM is hypocellular (10% to 15%)⁶  and/or fibrotic (17%),⁷  making the aspiration procedure painful and uncomfortable, especially in the elderly. In MDS the molecular analysis of copy number changes and genetic mutations has been done primarily on BM-derived DNA samples. Previous studies that used fluorescent in situ hybridization (FISH) and single nucleotide polymorphism arrays (SNP-As) to compare the karyotype concordance between BM and PB showed the usefulness of this approach. However, a comprehensive genetic analysis to compare the karyotype and mutation profile between BM and PB in MDS has not been performed.^8-11 

In this pilot study we investigated the presence of BM-derived genetic markers in both PB and serum from the same patients using high-resolution SNP-A karyotyping, 454 parallel sequencing (454-PS), and Sanger sequencing of 22 genes most frequently mutated in MDS and acute myeloid leukemia (AML).

Genomic DNA from PB and BM was extracted (Qiagen) from frozen cell pellets and 100 ng was whole genome amplified (WGA; Qiagen), both per manufacturer’s protocols. Serum DNA was purified from 200 µL of serum using a modified sodium iodide/Triton-based lysis followed by isopropanol precipitation as described.¹²  Affymetrix SNP 6.0 array (SNP-A) karyotyping and 454-PS of all exons of DNMT3a, RUNX1, CEBPα, TP53, EZH2, and ZRSR2 and mutation “hot spots” for NPM1, FLT3, ASXL1, IDH1, IDH2, MPL, JAK2, BRAF, cCBL, NRAS, KRAS, C-KIT, SF3B1, SRSF2, and U2AF35 were performed and analyzed as previously described.^13,14 TET2 was analyzed using Sanger sequencing. Independent validation for all mutations was performed using Sanger sequencing of unamplified genomic DNA. Polymerase chain reaction (PCR) conditions for serum were identical to those for PB; however, a second 10-cycle PCR reaction using nested primers (US1–GTAGTGCGATGGCCAGT, US2–CAGTGTGCAGCGATGAC) was required to provide adequate amplicon yield for Sanger sequencing. The study was approved by the local research ethics committee under project 0033 and conducted in accordance with the Declaration of Helsinki.

Karyotype aberrations were assessed using SNP-A on PB samples from 31 MDS patients, from whom metaphase cytogenetics (MC) and BM SNP-A karyotypes were available. These consisted of the following: normal karyotype (n = 9), del5q (n = 9), del7q/-7 (n = 5), trisomy 8 (n = 2), complex (n = 4), isodiXq13 (n = 1), and t(2:4)(q33;q27) (n = 1).

Concurrent BM and PB samples were available for 22 patients; nonconcurrent PB samples, median of 364 (14 to 1379) days after initial BM aspirate, were available for the remaining 9 patients. SNP-A karyotyping showed 100% concordance between BM and PB karyotype, except for 1 sample from a patient (USN1902) with a complex karyotype comprising 30 chromosomal aberrations in the BM, of which only 15 could be detected in the PB, suggesting the presence of multiple mosaic clones (Figure 1A; supplemental Table 1, available on the Blood web site).^15,16  Comparison of the PB SNP-A karyotypes with BM metaphase cytogenetics showed 100% concordance, except for the presence of 9 identical regions of uniparental disomy (UPD) in the PB and BM of 7 cases (UPD18q12.1-q21.31, UPD14q11.2-q32.33, UPD3p14.1-p13, UPD3q25.1-q26.31, UPDXq21.33-q23, UPD1p36.33-p13.3, UPD13q11-q34, UPD11p15.5-p13, and UPD15q15.1-q26.3). Translocations identified in 5 cases [t(7:21), der(1:7) (n = 2), t(1;6;8)(³²P;p21;q12), and t(2;4)(q33;q27)] by MC in the BM were not detected using SNP-A analysis either in the PB or BM (supplemental Table 1). Of note, in 8 cases with an isolated 5q deletion in the BM as assessed by MC, the 5q deletion could also be detected in the PB using SNP-A as well as FISH analysis (Figure 1B-C; supplemental Table 1).

Mononuclear cells from PB in general showed a lower clone size in comparison with those of BM, as indicated by the smaller mean copy number (CN) for chromosomal aberrations (PB vs BM): deletions (CN of 1.8 vs 1.6) and gains (CN of 2.2 vs 2.4; Figure 1A).

Mutational analysis of BM-derived genomic DNA from 21 MDS patients showed 38 mutations comprising TET2, SF3B1, and TP53 (n = 7 cases each); ASXL1 (n = 3); DNMT3a and FLT3 (n = 3 cases each); U2AF35 (n = 2); and NRAS, cCBL, JAK2, IDH2, and SRSF2 (n = 1 case each). Concurrent PB was available for 13 patients and nonconcurrent PB samples were available for the other 8 patients (median 409 [96 to 1073 days]). Mutational analysis of concurrent and nonconcurrent PB genomic DNA showed that 37 of the 38 (97%) mutations present in BM could also be detected in PB, with the exception of 1 NRAS mutation seen in a relatively small BM clone (size, 11%) that was undetectable in a concurrent PB sample (USN1533; Table 1). Interestingly, USN2233 (PB) was analyzed post azacitidine treatment and showed a mutation in TP53 (V157F, 1.5%) that was undetectable using Sanger sequencing, which was consistent with a normal PB SNP-A karyotype also observed in the respective BM (USN2232). This mutation was detected in the presentation sample (USN1894) at a clone size of 24% with concomitant genomic aberrations: on del5p15.33-p14.3, del5q11.2-q33.3, del17p13.3-p11.2, and gain19q12-q13.43 (supplemental Figure 1).

Similar to SNP-A karyotype analysis, the mutant clone size in PB was lower (median 25% [1.5% to 50%]) but was not significantly different (P = .4) from the BM clone size (median 33% [5% to 68%]), which is in agreement with previous karyotype studies in PB.¹⁰  In contrast to the results obtained using 454-PS that showed 97% concordance, Sanger sequencing resulted in only 65% concordance (smallest detected clone size was 21%) between BM and PB. These differences are attributable to the superior sensitivity of parallel sequencing technology to identify low-level clones. In addition, we compared the mutation profiles of unamplified and WGA PB DNA from 2 patients with mutations in TET2 and U2AF35 (Table 1). Contrary to a recent report, this analysis showed similar mutation profiles for both DNA samples.¹² 

To further study the utility of these procedures, we isolated DNA from serum of 12 patients with a total of 19 mutations in their BM. Analysis using electrophoresis and DNAOK! reagent (Web Scientific) showed highly fragmented serum-derived DNA unsuitable for SNP-A analysis. Sanger sequencing correctly identified only 8 mutations with no preference for any of the genes (42% concordance); 6 mutations were identified as wild type and 5 samples failed to amplify (Table 1).

In conclusion, our study showed excellent concordance both for SNP-A karyotype and mutational analyses between BM and PB, albeit with a lower clonal burden in PB, using FISH, SNP-A, and 454-PS that was independent of the differential PB profile of the patient cohort (supplemental Table 2). Patient serum is not recommended for mutation detection. The detection of regions of UPD in both PB and BM provides additional prognostic information because UPD on chromosome 7q has been associated with more aggressive clinical behavior in MDS and UPD on 5q may identify patients with potential response to lenalidomide.^3,17-19  Our results show the feasibility of using PB for initial genetic diagnostic screening of patients with MDS and sequential monitoring of disease clones following treatment, thus forming a prelude for validation in larger studies.

We thank Rajani Chelliah for assisting with sample processing and tissue separation.

We acknowledge Leukaemia Lymphoma Research (UK)/British Society for Haematology for supporting A.K. and King’s College London for funding the King’s College Haemato-Oncology Tissue Bank, from which all local samples were processed.

Contribution: G.J.M. conceived the study; A.M.M., H.A., A.K., S.S., S.M., F.M., A.E.S., and J.G. performed experiments; A.M.M., G.J.M., and J.G. contributed to design, analysis, and manuscript preparation; and A.K. and G.J.M. provided clinical details and analysis.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: G. J. Mufti, King’s College London, Department of Haematological Medicine, The Rayne Institute, 123 Coldharbour Lane, London, SE5 9NU, United Kingdom; e-mail: ghulam.mufti@kcl.ac.uk.