Risk factors for cardiovascular disease mortality in patients with myelodysplastic syndromes: A nationwide, registry‐based cohort study

Abstract Cardiovascular disease (CVD) emerges as a major cause of death in patients with myelodysplastic syndrome (MDS), but predictors of fatal CVD and the effect of MDS‐specific treatments on CVD mortality remain largely unknown. In an analysis involving 831 patients with MDS with known causes of death, we noted an independent association of lower risk MDS, age >70 years, pre‐existing CVD, and treatment with erythropoiesis‐stimulating agents with a higher risk of death from CVD. If externally validated, these simple risk factors could increase clinicians’ awareness toward CVD complications and guide early introduction of intensive monitoring and preventive interventions in MDS patients.


INTRODUCTION
Increasing evidence exists of an overlap in the risk factors between the development of cardiovascular disease (CVD) and cancer [1], and many individuals diagnosed with cancer have underlying cardiovascular disease risk factors, such as advancing age, hypertension, tobacco use, and diabetes. Myelodysplastic syndromes (MDS) are a heterogeneous group of clonal stem cell-derived disorders characterized by cytopenias, ineffective hemopoiesis, and a substantial risk of progression to acute myeloid leukemia. MDS typically affect elderly individuals with several comorbidities, which may influence treatment decisions and clinical outcomes, independently of the International Prognostic Scoring System (IPSS) [2].
Recent data argue for an intrinsic tendency of MDS toward the development of CVD. In 2017, Jaiswal and coworkers described nearly a doubling in the risk of coronary heart disease in persons with clonal hematopoiesis of indeterminate potential (CHIP), a precursor of MDS [3]. In the same year, Brunner and colleagues found that the burden of cardiovascular deaths was especially noteworthy in lower risk MDS, as compared with higher risk disease [4]. However, findings from the above studies seem inconsistent with the study by Adrianzen Herrera and coworkers who reported increased risk of incident CVD in patients with higher risk and transfusion-dependent MDS [5], whereas no study has investigated the effect of MDS-specific treatments on CVD morbidity and mortality.
The underlying pathophysiology of the increased prevalence of CVD in CHIP and MDS is not clear, but systemic inflammation and recurrent somatic mutations may contribute to accelerated atherogenesis [6,7]. However, the biggest challenges at the moment are to identify simple, MDS-specific predictors for CVD morbidity and mortality and understand the role of MDS treatments within this complex picture.
The latter is undoubtedly a key health issue. In this study, we aimed to detect commonly used clinical and laboratory parameters that would help hematologists to recognize patients with MDS at increased risk of death from CVD.  Table 1 lists patients' baseline characteristics with pairwise comparisons for all the prognostic variables considered. CVD as a COD was Univariate analysis at diagnosis (Table 1 and Figure 1A) found no differences between the two groups in performance status, liver, pulmonary and renal comorbidity, gender, and red blood cell (RBC) trans- biochemical parameters, the neutrophil count, hemoglobin concentration, serum ferritin, and β2-microglobulin levels were similar between the two groups, whereas lower platelet count, higher peripheral blood and bone marrow blast percentages, and higher LDH levels were found in the non-CVD death group. We noted significantly higher EASIX values in the non-CVD death group but, because the calculation of EASIX depends of LDH and platelet count, these results may simply reflect the higher risk disease status of patients in this group.

RESULTS
Of note, more patients in the CVD death group were treated with ESAs with or without granulocyte colony-stimulating factor (G-CSF) TA B L E 1 Demographic, clinical, laboratory, and histologic findings and treatments of patients with myelodysplastic syndromes whose death was attributed to cardiovascular disease (CVD mortality) versus those who died from a noncardiovascular disease cause (non-CVD mortality) than the non-CVD death group (P < .001), although responses to ESAs ± G-CSF were comparable between the two groups (P = .594). No difference in other first-line regimens was observed between the two groups (P = .501), including chelating agents (P = .787).

Parameters
In multivariate analysis, only four parameters correlated independently with CVD death: lower risk MDS, namely, an IPSS-R score ≤3.5 The results confirmed that pre-existing CVD (P = .007), ESA use (P = .008), age >70 years (P = .011), and IPSS-R ≥3.5 (P = .015) presented a statistically significant between-subject effect and, thus, were all independently associated with CVD death.

DISCUSSION
In this large, nationwide cohort of MDS patients with known outcomes, we evaluated the effect of numerous clinical and laboratory variables on CVD mortality. Our findings seemingly challenge some of the basic results in the study by Adrianzen Herrera and co-workers who reported an increased risk for CVD events in higher risk, transfusiondependent patients with a Charlson Comorbidity Index (CCI) >1 [5].
We found no association of cardiovascular death with transfusion dependence and a strong association with lower risk disease according to IPSS-R, whereas neither single comorbidity nor the MDS-CI was independently associated with cardiovascular mortality in our analysis.
This is likely to be attributable to notable differences in methodolog-

ACKNOWLEDGMENTS
We would like to thank all local investigators and team members for their contribution in data collection for the Hellenic National Registry of Myelodysplastic and Hypoplastic Syndromes.

AUTHOR CONTRIBUTIONS
IK conceived the idea for this study and designed the study.