Genome-wide sequencing enables genetic distinction of MDS/MPN subtypes

Based on the World Health Organization (WHO) classification, myelodysplastic/myeloproliferative neoplasms (MDS/MPN) are a group of myeloid neoplasms with clinical and pathological characteristics that overlap with MPN and MDS. The group includes the following conditions in adult patients: chronic myelomonocytic leukemia (CMML), atypical chronic myeloid leukemia (aCML), MDS/MPN-unclassifiable (MDS/MPN-U), and MDS/MPN with ring sideroblasts and thrombocytosis (MDS/MPN-RS-T).¹ Despite the fact that a high proportion of MDS/MPN cases harbour myeloid-related somatic mutations, these are still not considered under the current diagnostic work-up.² To further investigate the mutational landscape between the various MDS/MPN subtypes, and thus, to enable their diagnostic genetic distinction, Laura Palomo et al. published an extensive genome-wide sequencing study in Blood.² We hereby provide a summary of the published study results. 

Study design

• All patients underwent cytomorphological as well as conventional chromosome banding analysis (CBA)
• Whole genome sequencing was performed in 349 patients and whole exome sequencing in the remaining 18 patients, and data were analysed with the Illumina’s BaseSpace Sequence Hub
• Variant allele frequency estimations were used to estimate clonal and subclonal variant associations 

Patient characteristics

• A total of 367 patients with MDS/MPN diagnosis based on the WHO classification¹ with the following subtypes:
  • CMML (n = 119)
  • aCML (n = 71)
  • MDS/MPN-RS-T (n = 71)
  • MDS/MPN-U (n = 106)
• Chromosomal abnormalities:
  • Abnormal karyotype as detected by CBA was observed in 29% of patients, with the most common chromosomal abnormalities being: +8 (12%), −7/del(7q) (5%), and −Y (3%)
  • They were not very common in patients with the MDS/MPN-RS-T and CMML subtypes, but were quite common in patients with the MDS/MPN-U and aCML subtypes (Table 1)
• The key patient baseline characteristics per disease subgroup are shown below in Table 1

Table 1. Key patient baseline characteristics²

Results

Thirty genes were found to be recurrently mutated in ≥ 3% of patients with all of them having been previously associated in myeloid neoplasms. The most frequently detected ones are shown below in Table 2 

Table 2. Most common genes recurrently mutated in ≥ 3% of MDS/MPN patients²

• The authors then assessed the existence of genotype-phenotype associations and potential correlations between MDS/MPN subtypes and molecular events. The strongest subtype-mutation associations identified are shown in Table 2
• The MDS/MPN-U subtype was strongly associated with the less common TP53 and U2AF1 mutations
• Epigenetic regulators and splicing factors had significantly higher variant allele fractions (VAFs) when compared with transcription factors or signalling genes, and were identified as primary driver mutations
• • The SF3B1 mutation was detected as a primary hit in 21% of patients, and in 12% of cases was considered as a single founder event
  • Mutations in the epigenetic regulators DNMT3A, TET2, ASXL1 (DTA genes) were identified in the ancestral clone in 67% of patients
• The mutational profile and clonal architecture varied amongst MDS/MPN subtypes with specific mutation combinations and clonal architectures correlating to different MDS/MPN subtypes. All the correlations are shown below in Table 3. In general, the MDS/MPN-RS-T subtype was the least heterogenous molecular profile while the MDS/MPN-U subtype displayed the highest genetic heterogeneity 

Table 3. Mutational and clonal profile of different MDS/MPN subtypes²

• The authors further performed a principal component analysis to further identify prognostic mutation combinations that could help with the differential diagnosis of the various MDS/MPN subtypes. The results of this analysis showed that
  • CMML was significantly represented by the TET2 mutation combination
  • aCML was significantly represented by combinations between ASXL1, EZH2, and SETBP1
  • SRSF2 mutations were linked to CMML when combined with TET2 mutations, but with aCML when found together with SETBP1 mutations
  • RUNX1 mutations were linked to CMML when combined with SRSF2, but with aCML when co-mutated with EZH2
  • MDS/MPN-RS-T was significantly represented by SF3B1 together with JAK2 or DNMT3A mutation combinations
• 74% of patients with previously diagnosed MDS/MPN-U could be re-classified based on their mutational profile:
  • The ’CMML-like’ (17%), ‘aCML-like’ (33%), or ‘MDS/MPN-RS-T-like’ (11%) had mutations and outcomes similar to the respective subtype they resembled
  • There were two additional categories, the ‘TP53’ subtype in 13% of the patients harbouring TP53 mutations which correlated with a bad outcome, and the ‘Other’ subtype (26%) which could not be further characterized
• The presence of cytogenetic abnormalities was significantly linked to a lower overall survival in all MDS/MPN subtypes, except for patients with aCML
  • ASXL1 was significantly and consistently linked to unfavourable patient outcomes for all MDS/MPN subtypes except for aCML 

Conclusion

The results of this study provide insights and suggest the use of molecular markers for the potential differential diagnosis of MDS/MPN subtypes. For all subtypes, specific genetic clusters were identified that can help in their clinical identification. Additionally, almost 75% of the clinically highly heterogenous MDS/MPN-U subtype could be associated with other MDS/MPN subtypes or a prognostically adverse ’TP53’ subtype based on their molecular signature. Despite the lack of paired germline controls in this study, the results did show that different genetic mutations could be used to characterize most of the MDS/MPN subtypes. Furthermore, co-occurring additional mutations were associated with different outcomes.  These results pave the way for the consideration of the mutational profiles in the current clinical work-up of patients with MDS/MPN.