MDS is characterized by life-threatening peripheral cytopenias and a tendency to progress to a subset of AML that is difficult to treat. Improvements in treatment approaches for MDS and this subset of AML may require clarification of the pathogenesis of MDS. A current pathogenetic model characterizes MDS as a combination of hyperproliferation of marrow progenitors with excessive apoptosis, leading to poor production of peripheral blood cells. An alternative model characterizes MDS as a mutator phenotype with DNA damage as the direct cause of both peripheral cytopenias due to impairment of cell cycling of hematopoietic progenitors and leukemic progression. To evaluate these possibilities, we analyzed marrow samples from MDS patients using multiparametric flow cytometry techniques that allowed analysis of the cell cycle, including mitotic activity, in individual hematopoietic precursor compartments, and additional analysis of DNA damage and apoptosis in both individual hematopoietic precursor compartments and cell cycle subsets. Of particular note, neither erythroid, myeloid, nor CD34+ cells in MDS patients exhibited significantly increased mitotic events versus controls, indicating that MDS marrow is not hyperproliferative. These results call into question previous interpretations of DNA content [23
] and in vivo
DNA labeling [26
] studies as indicative of increased proliferation of marrow cells in MDS. As our data show, a finding of increased cells in S or G2 by DNA content without simultaneous analysis of mitotic activity is not definitive evidence of proliferation. We also demonstrate unexpected heterogeneity in cell cycle fractions in different hematopoietic progenitor subsets in both controls and MDS samples, with erythroid progenitors showing a 1.5- to 2-fold increase in S and G2 events versus myeloid progenitors. Thus, failure to compensate for erythroid hyperplasia, a common finding in MDS, in interpreting DNA content and labeling data could result in misinterpretation of results as hyperproliferation, rather than normal proliferation with an expanded erythroid compartment. Other factors may also be contributory to variance of our results versus in vivo
DNA labeling studies. Incorporation of labeled nucleotides in DNA is not uniquely specific for DNA synthesis (e.g., it may occur with DNA repair, and our findings suggest increased DNA damage in MDS), and in vivo
DNA labeling studies can include no normal controls, as the agents are mutagenic.
Under normal physiologic control in mammalian cells, S, G2, and M should maintain constant ratios. Our demonstration of a significant increase in the G2 compartment in myeloid precursors in MDS patients versus controls, with no increase in mitotic events, indicates aberrant cell cycle progression with delayed G2 transit, not proliferation. Mean G2 cells were also increased above controls in erythroid precursors, but the difference was not significant. A possible explanation for delayed G2 transit is DNA damage. To evaluate this possibility, we analyzed MDS samples for histone H2A.X phosphorylation at serine 139 (p-H2A.X), this phosphorylation being mediated by ATM in response to double-strand DNA breaks [20
]. We observed a highly significant 14-fold elevation of p-H2A.X in myeloid precursors in MDS samples and a 5-fold (not significant) elevation in erythroid precursors. These findings suggest that hematopoietic precursors in MDS have high levels of DNA damage, a possible explanation for both the delayed cell cycle transit that our data suggest and the preleukemic phenotype of MDS. Horibe et al. have recently reported similar results, using immunohistochemistry to demonstrate activation of ATM and phosphorylation of its substrate H2A.X in marrow samples from MDS patients, with little activity in control marrows. In addition, their observation of activation of the checkpoint genes Chk2 and p53 in MDS is consistent with the delayed G2 transit that our data suggest [32
Using Annexin V reactivity we were unable to demonstrate the increased apoptosis reported in MDS marrow by others (reviewed in [33
]). An explanation for this disparity is uncertain. We did support marrow cell viability ex vivo
with RPMI, coupled with rapid processing and analysis. We found no correlation of p-H2A.X versus Annexin V in most settings. We did note this correlation in G0/G1 in myeloid progenitors. DNA degradation during late apoptosis may result in marked phosphorylation of H2A.X (which we did not detect) [20
], while DNA damage (with associated H2A.X phosphorylation) can trigger apoptosis. It is possible that the correlation we observed in this limited setting is due to the latter.
Our finding of normal mitotic rates in MDS samples is consistent with our Annexin V results, suggesting that our results reflect more accurately the in vivo physiologic state in MDS. It would be kinetically impossible to maintain the typical hypercellular marrow in MDS with increased apoptosis but no increase in mitotic activity. It should be noted that assays of apoptosis are performed ex vivo. Production of uric acid is an in vivo surrogate for cell death; the routine absence of hyperuricemia in MDS is inconsistent with increased in vivo apoptosis. Our finding of no apoptotic response in cells in G2 despite evidence of increased DNA damage suggests an alternative interpretation of data in MDS, namely that in MDS there may be a failure in vivo to trigger effective apoptosis in response to DNA damage. If so, cells with unrepaired DNA damage could accumulate in marrow, resulting in hypercellularity. With the additional stress of aspiration, these cells would proceed to apoptosis, resulting in a false impression of excessive in vivo apoptosis.
Multiparametric flow cytometry was crucial for performance of these studies, allowing simultaneous quantitative analysis of cell cycle and other parameters in electronically segregated subpopulations of marrow cells. Use of MPM-2, with direct quantitation of mitoses, gave a more complete assessment of the cell cycle than DNA content alone [12
]. Use of DRAQ5 for DNA content analysis allowed simultaneous analysis with other parameters in segregated hematopoietic subpopulations, which demonstrated unexpected heterogeneity of virtually all parameters tested among different hematopoietic subpopulations. Our results cast doubt on the interpretability of whole marrow analyses of these parameters in previously reported studies and allowed demonstration of abnormal cell cycle progression in one hematopoietic subset. We demonstrated differences in results for some parameters (p-H2A.X, Annexin V) in different cell cycle fractions and that these differences also varied in differing hematopoietic progenitor subpopulations. This complexity would be difficult to demonstrate with any other available analytical technology.