Main finding

In our study, the findings revealed that RDW-SD emerged as an independent predictor of the composite outcome of death and heart transplantation not only in the overall DCM patient population but also among those with pre-DM and DM. However, its independent prognostic significance was not observed in patients with DCM with NG after adjusting for confounding factors. These results underscore the potential of RDW-SD in identifying high-risk individuals within the patients with DCM.

The characteristics of different RDW parameters

RBCs represent the predominant blood cell type, primarily responsible for transporting oxygen throughout the circulatory system, from the lungs to peripheral tissues within the human body.19 In certain conditions, RBCs may undergo significant increases or decreases in their typical volume. Consequently, the SD of RBC volumes, referred to as RDW-SD, serves as a metric to quantify the extent of heterogeneity in RBC volume. Additionally, another parameter, RDW or RDW-CV, is calculated by dividing RDW-SD by the mean corpuscular volume (MCV) multiplied by 100 (RDW-SD/MCV×100). This parameter is widely employed to characterize the variability in the size of RBCs.2 20

It is noteworthy that previous studies have established a correlation between increasing age in the general population and elevated levels of both RDW-SD and MCV. This age-related increase in MCV tends to mitigate the age effect observed in the traditional RDW. Consequently, this study suggests that future investigations into the prognostic value of RDW should prioritize RDW-SD (the ‘absolute’ RDW), excluding the age-dependent effect of MCV from the calculation.21 Moreover, a study involving 1,257 hospitalized patients who underwent coronary CT angiography revealed through multivariate logistic regression analysis that RDW-SD independently predicted severe coronary artery calcification, whereas RDW did not exhibit the same predictive capability.22 In our previous research, despite considering RDW, RAR, and RDW-SD as candidate prognostic biomarkers, the machine learning process ultimately identified RDW-SD as the definitive predictive factor for patients with non-ischemic HF.9 Considering these factors, we have chosen to prioritize RDW-SD as the primary focus for prognostic association research in this study.

Potential mechanisms of the association between RDW and HF

Numerous factors contribute to RDW variation, including age, sex, genetic factors, anemia, renal function, and dyslipidemia.3 Previous studies have established RDW as an independent predictor of mortality in patients with HF across various clinical subtypes.23 Additionally, it has been shown to be independently associated with impaired exercise capacity in patients with HF.24 Furthermore, research indicates that elevated RDW levels at discharge are linked to an unfavorable long-term outcome for patients with HF, irrespective of hemoglobin levels and anemia status.25 RDW may serve as an indicator of nutritional deficiency, bone marrow depression, or chronic inflammation. Studies have highlighted a correlation between RDW and well-established inflammatory markers such as interleukin-6 and CRP.26 27 These conditions are frequently present in patients with HF, correlate with the disease’s severity, and are associated with an adverse prognosis.

Numerous studies have convincingly demonstrated that cell death serves as a prominent initiator of inflammation, contributing significantly to ischemic HF subsequent to myocardial infarction (MI).28 29 However, the inflammatory processes observed in non-ischemic DCM are not initially associated with cell death. In contrast to MI, emerging research suggests that cardiomyocytes are the primary loci where genes related to inflammation are expressed in response to non-ischemic stressors. The activation of Ca2+/calmodulin-regulated kinase is identified as the underlying mechanism triggering cardiac inflammation in response to non-ischemic stimuli.30 31 Therefore, the primary focus of this study was to explore the prognostic implications of a novel inflammatory marker, RDW-SD, in patients with non-ischemic DCM, with the aim of unraveling clues about the unique activation of inflammation in this specific condition.

The association between RDW and glycemia status and its implication for DCM

In a retrospective study involving 2,688 subjects without DM at baseline, the multivariate relative risk of developing DM was notably increased in the highest quartile of RDW compared with the lowest quartile, even after adjusting for HbA1c and hs-CRP.32 Another study revealed that newly diagnosed patients with DM with lower baseline RDWs were more likely to maintain 1 year euglycemia remission after short-term continuous subcutaneous insulin infusion.33 A recent genome-wide association study investigated inherited genetic variations associated with RDW in UK Biobank human volunteers. This study identified individual RDW variants that are correlated with BMI, lipids, HbA1c, metabolic syndrome, autoimmune disorders, and inflammatory diseases.34 Our study found that patients with pre-DM or DM had significantly higher levels of RDW-SD and hs-CRP, but not significantly higher levels of NT-ProBNP, which suggests that inflammation is more dominant than ventricular mechanical stretch in diabetes-related mechanisms.

Besides, inflammation is one of the important mechanisms in diabetes-related complications. In our study, the rate of renal dysfunction was high in patients with DM (24.9%), and the RDW-SD was shown to correlate with the level of eGFR. Importantly, we found that the prognostic value of RDW-SD is still independent when adjusting the renal function as a covariate. However, only 6 patients were diagnosed with diabetic foot among 375 patients with DM in our cohort. This prevalence is lower than the rate in a large diabetes cohort, which included 11,806 patients and 25% of patients had diabetic foot at any time point from 2010–2020.35 The lower rate of diabetic foot in our study might be because the diabetic complication was only diagnosed at baseline but not longitudinally observed in follow-up. Another reason might be because our cohort enrolled patients with non-ischemic DCM in a HF care unit, so the diagnosis rate of early-stage diabetic foot might be relatively lower. A previous study has found that people with DM plus foot and renal complications have a higher mortality risk.35 36 Therefore, early identification and intervention of diabetic complications in patients is very meaningful.

While there is currently no conclusive evidence regarding the prognostic role of RDW in patients with DCM, a prior study by Xanthopoulos et al enrolled 218 patients presenting with acute HF, without restrictions on etiology. This study found that RDW at admission was associated with a higher rate of worse prognosis in patients with HF, both with and without DM. The longitudinal changes of RDW in patients with DM significantly differ from those with non-DM.37 However, Xanthopoulos et al did not distinguish patients with pre-DM in their study population. A study by Huang et al explored the relationship between the RAR and carotid plaque in patients with coronary heart disease across different glucose metabolic states. In varying glucose metabolic conditions, RAR exhibited the strongest correlation with the risk of carotid plaques in patients with DM.6 This heightened correlation in patients with DM may be attributed to the chronic inflammation and oxidative stress associated with DM, impacting the deformability and mechanical properties of RBCs, leading to increased adhesion and heightened osmotic vulnerability.

In a different study assessing the prognostic value of a novel inflammatory biomarker, the fibrinogen-to-albumin ratio (FAR), in patients with acute decompensated HF (ADHF) across different glycemic metabolic states, results indicated that elevated FAR was independently associated with a poor prognosis in patients with ADHF with DM, but not in those with NG and pre-DM.38 This finding indicated that patients with pre-DM need specific prognostic markers. Our study’s results further support the notion that RDW-SD can serve as a prognostic biomarker in patients with pre-DM and DM, and this association is independent of factors including demographics, comorbidities, echocardiography and laboratory tests, and medical therapy, emphasizing its potential as an inflammatory biomarker to identify high-risk individuals in this population. Notably, the effect size of RDW-SD in the prognosis was compatible between the pre-DM and DM, which suggests a non-negligible effect of chronic inflammation in patients with pre-DM.

The sodium-glucose cotransporter 2 (SGLT2) inhibitor empagliflozin has demonstrated the ability to reverse LV remodeling in patients with heart failure with reduced ejection fraction and DM or pre-DM.39 Significantly, it has also shown improvement in cardiovascular and renal outcomes, regardless of baseline glycemic status.40 Proposed mechanisms of action for SGLT2 inhibitors include the reduction of cardiac oxidative stress and inflammation by promoting the actions of sirtuin-1 and upregulating hypoxia-inducible factors signaling.41 In the context of this study, where patients with DCM coexisting with DM or pre-DM and increased RDW-SD may indicate a higher degree of inflammatory activation, abnormal cardiac energy metabolism, and LV ventricular remodeling, inflammatory markers such as RDW-SD could potentially aid in identifying individuals who may benefit from SGLT2 inhibitors.

Limitations

This study comes with several limitations. First, due to its retrospective nature, there is a potential for selection bias. Second, the limited number of patients undergoing dynamic biomarker monitoring during follow-up hindered the analysis of any potential associations between changes in markers and patient prognosis. Third, the study exclusively enrolled patients with non-ischemic DCM, resulting in a relatively constrained sample size for each glycemia status subgroup. Therefore, caution is advised in generalizing the results to subgroups defined by glycemia status.