Targeted Sequencing of 647 Patients with Myeloid Malignancies, Marrow Failures and Cytopenia: A Prospective Analysis

The diagnosis of myelodysplastic syndrome, especially in the absence of ring sideroblasts, excess blasts or cytogenetic abnormalities, is challenging in daily clinical practise and is biased by poor inter-observer concordance. The array of genetic abnormalities identified in MDS by targeted sequencing might aid in diagnosis, especially in cases with mild cytopenias and subtle dysplastic features. However, the presence of such somatic mutations in 'normal' elderly population leading to clonal haematopoiesis of indeterminate potential (CHIP) further confounds use of somatic mutations to diagnose MDS.

We set out to prospectively evaluate 647 unselected, consecutive patients with variable degrees of cytopenia referred to our molecular diagnostics laboratory at Kings College Hospital with a 24 gene targeted mutation panel initially using Roche 454 amplicon sequencing panel and subsequently an Illumina MiSeq platform. The average read depth was 200 x and all known pathogenetic variants >10% variant allele frequency (VAF) were reported if present in the Cosmic database..

The median age was 57 years (range 16-92) and sequencing was performed on peripheral blood DNA in 58% (374/647) and bone marrow in the remainder. One third (36%, 234/647) harboured a somatic abnormality, with 51% (120/235) having more than one mutation. 443 mutations were detected in 235 patients, with 1.9 mutations/patient. The median VAF was 38% (9-99%) and the most frequently mutated genes were TET2 (N=67, 12%), ASXL1 (N=63, 10%) and SRSF2 (N=45, 7%). All 24 genes, except KDM6A, was mutated at least in one patient.

Comprehensive clinical information was available for patients seen in our institution (n=422) and revealed a similar overall frequency of mutation detection (30%, 128/422). Mutations were extremely rare in patients (8%, N=102) presenting with single cytopenia (especially neutropenia) and a non-diagnostic bone marrow with no morphological features of MDS. None of patients with aplastic anaemia (23%, n=99) at diagnosis had mutations detected by our panel, although 3 acquired abnormalities before or at the time of MDS transformation. Of the 186 patients with MDS, 56% (N=105) harboured mutations, with a relatively low frequency in cases with RCMD/RA (26%, 26/98) compared to published literature. Contrarily mutation frequency was higher in patients with RARS, RAEB and AML with somatic aberrations detected in 100%, 70%, 90% cases respectively. No cases of hypoplastic MDS had detectable mutations.

Further analysis, by including variants with VAF 5-10% (N=9) and variants of unknown significance (VUS) (N=31) increased the mutation frequency, i.e. mutations were present in 40% (39/98) of RCMD/RA patients and 17% (17/102) of cytopenic patients with no evidence of morphological MDS.

Overall the ability to detect mutations only in a third of RCMD/RA patients shows the considerable drawback in using mutations alone for diagnosis in RCMD which is currently based purely on morphology which is subjective and with poor concordance between morphologists.

The cautious/conservative reporting of only pathogenetic variants with VAF >10% in our routine clinical practice led to decreased frequency of detectable mutations in MDS compared to published literature. This also avoided the reporting of variants of unknown clinical significance in the context of isolated or mild cytopenia in the absence of overt dysplasia in our cohort of unselected cytopenias. The reported incidence of mutations in MDS is influenced by the chosen VAF cut-off, dynamic evolution of the COSMIC database, inclusion of mutations that have not yet been reported in the literature as pathogenic and inclusion of stop or missense mutations predicted to be pathogenic. This appears to differ in this real-life referral population compared with well-defined and curated MDS cohorts.

Our current diagnostic panel has been expanded to include a repertoire of 35 genes and we are planning to combine this with SNP-A karyotyping to better characterise patients with cytopenia, early MDS and cases with subtle dysplasia to improve certainty of diagnosis.