Distinct Pattern of Genomic Changes Associated with Hypoplastic Compared to Hyper/Normoblastic Myelodysplastic Syndromes (MDS)

While the bone marrows of most MDS patients (pts) show normal or increased cellularity (hyper-MDS), a small but significant fraction of pts present with a hypocellular bone marrow (hypo-MDS), which may compromise morphologic evaluation and karyotypic analyses. Identification of somatic mutations or their combination may help distinguish MDS from other bone marrow failure states, and illuminate the distinct molecular pathogenesis of these MDS subtypes.

We performed genomic analyses by using next-generation targeted deep sequencing (NGS) of 62 significantly affected genes selected based on the frequency observed in a separate cohort of MDS patients analyzed by whole exome sequencing (WES). Mutations were considered individually and grouped into several functional pathways which were hypothesized to characterize MDS pathogenesis. Hypo-MDS was defined using a standard definition of bone marrow cellularity </= 25% regardless of the age. Variant allele frequencies (VAFs) adjusted by zygosity were used to define clonal architecture of dominant clones.

Of 237 MDS pts seen between 1/2000-12/2012 with available clinical and mutational data, 32 (14%) had hypo-MDS. Compared to hyper-MDS, hypo-MDS pts presented with a lower white blood count (median 2.4 vs 3.7 X10⁹/L, p= .002), lower absolute neutrophil count (median, 1.07 vs 1.67 X10⁹/L, p=.009), and a higher bone marrow blast % (6% vs 2 %, p=.006). No difference in median age (68 vs 68 years, p = .88), median hemoglobin (9.5 vs 10.4 g/dL, p= .36), or IPSS-R scores (p= .68) was observed between the 2 groups. Cytogenetic analysis showed insignificant differences in: normal karyotype (38% in hyper-MDS vs 42% in hypo-MDS), +8 (6% vs 9%), -7/del-(7q) (5% vs 6%), complex >/=3 (17% vs 9%). With median follow up of 33.6 months (mo) (range, 0.4-128.5), the median overall survival for hyper-MDS was similar to hypo-MDS (23.4 vs 29.4 mo, p= .9). Overall, 76% of pts had at least one of the 62 screened mutations, the most common being: SF3B1 (15%), ASXL1 (14%), STAG2 (11%), TET2 (10%), RUNX1 (10%), U2AF1/2 (9%), and DNMT3A (8%). Compared to hyper-MDS, pts with hypo-MDS had a lower average number of somatic mutations (mean, 1.18 vs 1.80, p = .03) and fewer had 2 or more mutations (28% vs 49%, p=.02). Mutations in SF3B1 (3% vs 17%, p= .02), and IDH1/2 (0% vs 9%, p=.06) were more common in hyper-MDS, mutations in DDX41/DDX54 (9% vs 2%, p=.08) occurred more frequently in hypo-MDS. Within functional groups, mutations in splicing machinery genes were more common in hyper-MDS compared to hypo-MDS (34% vs 9%, p = .03). We then evaluated the VAFs of the dominant clones in both groups. The average size of the dominant clone was higher in hyper-MDS compared to hypo-MDS (61.9% vs 46.8%, p = .02). Further, clonal architecture analysis indicates different dominant clones between the two groups. Mutations in SF3B1, SUZ12, ZRSR2, U2AF1, TET2, STAG2, ASXL1, BCOR, and BCORL1 were more frequently observed as dominant clones in pts with hyper-MDS compared to hypo-MDS.

In conclusion, pts with hypo-MDS have smaller sized dominant clones with fewer acquired mutations compared to hyper-MDS. One possible explanation for this is that the immune system in pts with hypo-MDS may suppress the driver clone, inhibiting its growth and genetic evolution, thus limiting the acquisition of downstream lesions. Some founding clones, such as SF3B1, ASXL1, and TET2 may overcome this effect.