Main finding

In this study, we utilized a Lasso-Cox regression to select predictors from a pool of 20 biomarkers representative of metabolism, malnutrition, and inflammation. Subsequently, we developed a risk assessment tool termed MIS, comprising ALB, RDW-SD, Lymphocyte, hs-CRP, UA, and T3. The MIS demonstrated efficacy in the prognostic prediction for HFpEF patients and also exhibited consistent predictive performance across cross-validation iterations. These findings underscore the utility of the MIS in identifying high-risk individuals and facilitating targeted therapeutic interventions.

The role of metabolic inflammation in HFpEF

Systemic proinflammatory states, mostly induced by obesity and metabolic stress, have increasingly been identified as a predominant determinant of HFpEF pathophysiology by recent studies. Metabolic disorders often coincide with disruptions in the immune system. Indeed, there has been a bidirectional effect between metabolic activities and the regulation of immune cells. This reciprocal interaction is increasingly recognized as a pivotal factor in the development of cardiometabolic disorders like HFpEF [3]. In this study, we identified the prognostic role of biomarkers involved in uric acid (UA) and thyroid hormone metabolism, independent of glucose and lipid metabolism. Serum UA is the final product of purine metabolism. Beyond its diagnostic utility for identifying gout, UA levels have been linked to metabolic syndrome and cardiovascular disease [14, 15]. In patients with chronic HF, serum UA has shown significant associations with outcomes across the whole EF phenotypes [11]. While systemic inflammation and endothelial dysfunction, potentially linked to UA, are suggested as consequences of HFrEF, they are speculated to act as contributors to HFpEF. Recent research also suggests that elevated serum UA levels correlate with increased cytokine levels and heightened inflammatory responses, which may play a more substantial role in HFpEF than HFrEF [16, 17].

A thyroid hormone directly affects the myocardium, the conduction system, and the peripheral vasculature. Hypothyroidism is associated with hyperlipidemia and ventricular arrhythmias, hyperthyroidism is associated with atrial arrhythmias, and both are associated with hypertension and HF [18]. Based on our findings, the level of triiodothyronine (T3) was negatively correlated with HFpEF prognosis. According to previous studies, isolated low T3 is associated with more severe HF and an over 2-fold risk of adverse outcomes [19]. In the cardiomyocyte of failure heart, hypoxia and inflammation reduce deiodinase activity. This results in reduced plasma T3 levels and decreased intracellular bioavailability of T3. The effects of thyroid hormones on myocardium include upregulation of myosin heavy chain-α and downregulation of myosin heavy chain-β, regulation of calcium cycling through SERCA2a, and enhancement of adrenergic responsiveness [12]. The activation of the SERCA2a could enhance both systolic and diastolic function, and the latter is an important therapeutic target in HFpEF.

Malnutrition-inflammation complex in HFpEF and related biomarkers

Numerous studies indicate that some of cardiovascular risk factors are associated with elevated risk of adverse outcomes in HF patients, such as a lower BMI, blood pressure and serum cholesterol concentration. These observations are in contrast to that in the general population, which have been referred to as “reverse epidemiology” [4]. The occurrence of the “malnutrition-inflammation complex syndrome” in HF patients offers a potential explanation for the existence of “reverse epidemiology”. A reduction in lipoprotein can compromise their endotoxin-scavenging function, making HF patients with malnutrition susceptible to inflammatory endotoxemia [4]. In this study, the Lasso-Cox model finally selected 4 biomarkers, including the RDW-SD, ALB, Lymphocyte and hs-CRP, which can reflect the severity of malnutrition and inflammation within HFpEF patients.

RDW-SD is a metric that quantifies the diversity in the size of circulating red blood cells. A study have proposed that is better to use RDW-SD to eliminate the confounding influence of mean corpuscular volume (MCV) on RDW [20]. Current evidence suggests that RDW is recognized as an indicator of chronic inflammation, exhibiting a notable correlation with inflammatory markers. Additionally, disturbances in iron metabolism, renal dysfunction, and malnutrition have been implicated in the mechanism of elevated RDW levels among HF patients [21]. Our prior studies have established that RDW had an independent association with mortality across diverse ejection fraction categories and etiologies among HF patients [7, 22]. Furthermore, ALB serves as another biomarker reflecting nutritional and inflammatory status. Notably, hypoalbuminemia, defined as an ALB level below 3.4 mg/dl, is observed in approximately 28% of patients with HFpEF [23]. According to Frank-Starling’s law, hypoalbuminemia-induced decreases in plasma oncotic pressure facilitate fluid shifts from the blood vessels into the tissues, thereby contributing to cardiogenic pulmonary edema and exacerbating the severity of condition in patients [24].

Prior research has established several nutritional and inflammatory risk scores, incorporating some of the biomarkers investigated in our study and validating their effectiveness. For instance, the geriatric nutritional risk index (GNRI) has been proven to be a reliable screening tool for malnutrition in the elderly, utilizing objective measures such as height, weight, and serum ALB. This index has been linked to the risk of cardiovascular and all-cause mortality in patients with HFpEF [5]. Another study identified CRP, RDW and neutrophil-to-lymphocyte ratio (NLR) as components of an inflammatory prognostic score among 538 acute HF patients [25]. Additionally, the Pan-Immune-Inflammation Value, calculated using biomarkers of complete blood cell counts, has demonstrated superior predictive ability in patients with ST-segment elevation myocardial infarctions [26]. However, these studies have not comprehensively evaluated the discrimination, calibration, and reclassification performance of these scores. Further studies are required to construct a comprehensive index that incorporates biomarkers related to metabolism, malnutrition, and inflammation. Therefore, we specifically focus on HFpEF patients, incorporating 20 biomarkers to develop a risk score that accurately reflects the underlying pathophysiology of HFpEF.

Targeted therapies for metabolic malnutrition and inflammation in HFpEF

Our study has established the prognostic significance of serum UA and hs-CRP, and incorporated them into a comprehensive risk-scoring system. Prior research analyzed the data in the SOCRATES-REDUCED study to explore the influence of vericiguat on hs-CRP and serum UA levels in HFrEF patients. Notably, this study revealed that 12-week treatment with vericiguat was associated with a notable reduction in both hs-CRP and UA levels [27]. These findings suggest a potential anti-inflammatory benefit of vericiguat in HFrEF patients. However, the VITALITY-HFpEF study showed that vericiguat failed to improve the KCCQ compared with placebo [28]. A post-hoc analysis of PARAGON-HF also found that Sacubitril–valsartan can reduce the serum UA level and the initiation of UA-related therapy. While the effect of Sacubitril–valsartan was not significantly modified by serum UA levels, its beneficial effects were more pronounced in terms of renal outcomes [29]. To sum up, these data suggest that UA may be a relevant therapeutic target in HFpEF.

Sodium-glucose cotransporter 2 (SGLT2) inhibitors have emerged as a promising therapeutic option for patients with both HFrEF and HFpEF [30]. While the precise mechanisms underlying the cardiovascular benefits of SGLT2 inhibition remain unclear, there is speculation that these inhibitors may target metabolic inflammatory pathways. A recent study revealed that patients treated with SGLT2 inhibitors exhibited lower levels of circulating IL-6, serum UA, and fasting insulin compared to those receiving other glucose-lowering drugs [31]. Furthermore, vitro models suggest that SGLT2 inhibitors possess a tangible anti-inflammatory activity, potentially mediated by their ability to reduce UA and insulin concentrations. This effect complements other proposed mechanisms that explain the observed benefits of this drug on cardiovascular and renal endpoints [31]. Given that the components of the Metabolism-malnutrition-inflammation risk score (MIS) have demonstrated potential as therapeutic targets in HFpEF patients, this score may serve as a useful tool in guiding individualized treatment strategies.

Limitations

Firstly, its retrospective nature may have led to selection bias and overlooked potential confounding factors, limiting the comprehensiveness of our analysis. Secondly, while the study focused on 20 easily accessible biomarkers commonly used in clinical settings, it did not include more specialized biomarkers, such as those derived from metabolomics or proteomics. Finally, while the prognostic risk score model established in this study has undergone internal cross-validation, it has not yet been validated by an external cohort. Therefore, we must emphasize the need for further external-validation.