Genotype-Resultant Morphology of Myelodysplastic Syndromes (MDS)

Morphology has dominated the diagnosis and classification of MDS for decades. With the advent of NGS, morphology is used as a gold standard to assess associations of specific disease sub-entities with individual mutations or their combinations. An example is the association between SF3B1 mutations and ringed sideroblasts, or with disease-defining lesions, such as t(8:21), 11q23 translocations, and EVI1 mutations. Beyond these discrete examples, assignments of diverse morphologic features to combinations of specific genotypes has been more challenging.

To identify potentially novel paradigms of associations between molecular hits and morphologies we examined 835 patients with MDS. Common somatic mutations, chromosomal abnormalities and morphologic reviews were subjected to blinded, objective standard diagnostic criteria. Abnormalities included individual lineage dysplasia and cytopenias, monocytosis, fibrosis, and increased megakaryocytes, all classified using standard criteria. To correlate the presence of mutations to morphologic criteria, only mutations with a clonal burden >10% were used. In total, 794 out of 835 patients (95%) harbored at least one lineage dysplastic feature. Myeloid, erythroid and megakaryocytic dysplasia occurred in 56% (470/835), 71% (590/835) and, 70% (587/835) of patients, respectively. Bi- and tri-lineage dysplasia were also identified in 36% (297/835) and 33% (278/835). In 90% (753/835) of patients there was cytopenia: 48% (402/835) had neutropenia, 61% (510/835) anemia, and 61% (510/835) thrombocytopenia. More than 50% of patients had multiple-cytopenias; 23% (194/835) had pancytopenia. Monocytosis was present in 20% (168/835) of patients and 31% (256/835) had elevated numbers of megakaryocytes. For analysis all possible combinations (25 pairs in total) were queried and those present in >80 patients were evaluated. Based on NGS among 36 genes, TET2 (27%), SF3B1 (23%), ASXL1 (19%), del(5q) (16%), SRSF2 (14%), DNMT3A (11%) and del(7q) (10%),, were frequently mutated genes/CNAs, each present in >10% of patients. A number of relationships between phenotypic features and molecular context were uncovered. For example, patients with thrombocytopenia also commonly had tri-lineage dysplasia [38% (198/510) vs. 25%, (80/325), p<.001, q<.001] or more myeloid dysplasia [63% (321/510) vs. 46% (149/325), p<.0001, q<.0001] but no megakaryocyte dysplasia [72% (366/510) vs. 68% (221/325), p=.027]. Moreover lineage specific dysplasia was not associated with lineage specific cytopenias. In fact, myeloid or erythroid dysplasia had no significant association with neutropenia or anemia (p=.14 or p=.1, respectively).

When we performed statistical analyses that sought correlations between clinical features and genetic alterations, 78 significant associations were identified. For instance, patients with erythroid dysplasia more frequently had SF3B1 mutations [in 15% of erythroid dysplasia patients, 90/590] than patients without it [6%, 15/245] (p=.0002, q=.001). In contrast, these patients were less likely to have EZH2 mutations [4.2% (25/590) vs. 11.4% (28/245), p<.001, q=.002]. Compared to patients with no myeloid dysplasia, patients with myeloid dysplasia were more likely to have TET2, STAG2, and SRSF2 mutations, and less likely to have SF3B1 mutations (p<.02). Compared to patients without megakaryocytic dysplasia, those with it more commonly had ASXL1, mutation, but less commonly SF3B1 and BCORL1 mutations. Patients with STAG2 mutations [8%, 66/835] were more likely to have neutropenia [62% (41/66) vs. 47% (361/769), p=.02, q=.08]. TP53 and ASXL1 mutations occurred frequently in patients with thrombocytopenia. Monocytosis co-existed more with 5 genetic mutations, including SRSF2 and RAS pathway genes.

Using various conventional and more advanced strategies, we present the first comprehensive genotype-phenotype association study that linked somatic molecular events to morphologic and clinical features. We demonstrate that the complexities of how molecular context governs morphologic changes can be deciphered. Certain rigorously defined morphologic configurations show specific associations with individual mutations or their combinations.