Quantitative Analysis of Patient-Specific Lesions in Primary and Xenografted Myelodysplastic Syndromes Reveals Complex Hierarchies and Subclonal Diversity That Evolve during Disease Progression

Background

Myelodysplastic Syndromes (MDS) likely arise from an evolutionary process involving accumulation of somatic mutations and selection that occurs at the level of hematopoietic stem cells. Using next generation sequencing (NGS), several groups found recurrent somatic mutations to be associated with MDS. However, the history of stepwise molecular progression is still poorly defined. Likewise, little is known about patient-specific subclonal compositions, which may well contribute to the tremendous heterogeneity observed both in terms of clinical manifestations and response to treatment. Therefore, we sought to reconstruct patient-specific clonal hierarchies and decipher their dynamic evolution during long-term disease monitoring in order to better understand MDS pathogenesis and aid in adapting targeted therapeutic options in the future.

Methods

Bone marrow (BM) or CD34+ cells from patients with MDS were subjected to mutational screening by whole exome sequencing (WES, n=44) or targeted NGS (n=28) interrogating up to 17 recurrently mutated genes. Mesenchymal stromal cells (MSCs) were used as germline control. Allelic burdens were quantified with custom pyrosequencing assays and interphase-FISH in FACS purified myeloid, erythroid, lymphoid and stem cells isolated from both patients’ BM and corresponding xenografts in NSG mice. For unbiased follow up analysis, BM or CD34+ cells from two distant time points (median: 3 years, range 0.5-4.0 years) were analyzed for 13 patients using WES and these data were used to design patient-specific mutational panels. These panels contained amplicons representing mutational events uniquely present at one time point as well as others present at both time points. The panels were then subjected to ultra-deep-sequencing (UDS) to accurately quantify mutational burdens in all follow up (FU) samples.

Results

Integrative mutational data allowed us to reconstruct patient-specific mutational hierarchies in 35 cases, which revealed both linear and branching evolution in MDS. The data is also in support of the notion that potential founder lesions are highly enriched in genes involved in RNA splicing and epigenetic regulators, which we further show to be frequently detected in primary and/or xenografted lymphocytes. In contrast, we clearly demonstrate that large scale cytogenetic lesions (e.g. monosomy 7, trisomy 8, del(5q)) occur as late mutational events in at least 85% of the cases analyzed (n=17/20, 3 unresolved cases). Moreover, we could readily demonstrate the existence of subclonal heterogeneity in the patients’ BM, with variable contribution to different lineages and also cases showing lineage restriction of specific sub-clones. Most importantly, UDS analysis of FU samples from 13 independent patients (median FU 3.3 years, range FU 0.5-11.8 years and median of 5 samples per patient) with detailed clinical data revealed a highly dynamic clonal evolution during the course of the disease and dramatic shifts in the composition of mutational (sub-)clusters during therapy. Treatment often resulted in complete disappearance of specific sub-clones and most importantly, simultaneous outgrowth of previously undetectable subclones that subsequently dominated the marrow.

Conclusion

By reconstructing patient-specific mutational hierarchies we gained invaluable insights into the dynamic clonal composition and the molecular progression of human MDS. Our study shows that acquisition of large scale cytogenetic lesions appears to be a rather late event, which likely indicates that such lesions might not be tolerated by healthy stem cells. Most importantly, WES and subsequent deep sequencing analysis of FU samples demonstrated that patient-specific clonal diversity is highly dynamic and modulated during the course of the disease. This is especially true upon response to treatment, where concomitant disappearance and emergence of new pathogenic subclones was observed. Our work unravels two major paths of mutational acquisition by which MDS cells evade therapeutic pressure: (1) linear evolution of the previously sensitive subclone or (2) by branching evolution of an earlier independent founder clone. Our findings therefore strongly emphasize the importance of a genetically unbiased disease monitoring and the development of novel therapeutic strategies aiming at targeting founder lesions that are present in all ensuing subclones.