Low Levels of Adropin Predicted New Incidents of Atrial Fibrillation in Patients with Heart Failure with Preserved Ejection Fraction

1. Introduction

Atrial fibrillation (AF) is one of the high prevalent arrhythmia in patients with heart failure (HF) that links adverse cardiac remodeling with an increased risk of all-cause and cardiovascular (CV) death, stroke, peripheral embolism and HF decompensation [1]. The pooled prevalence of AF in the inpatients with acutely decompensated HF with reduced (HFrEF) and preserved (HFpEF) ejection fraction was 34.4% and 42.8%, respectively [2]. Along with it, AF was detected in about 20% outpatients with HFrEF and 33% those with HFpEF [2]. To note, the prevalence rate of AF in patients with HFpEF may reach 60% [3].

Moreover, AF and HF are not only coexist, but also can be plausible triggers for progressing electrical and structural atrial remodeling acting through the overlap of several pre-disposing factors, such as conventional CV risk factors (hypertension, smoking, dyslipidemia, older age), established cardiac diseases (coronary artery disease, cardiomyopathies, valvular heart disease) and signature of comorbidities including type 2 diabetes mellitus, chronic kidney disease, overweigh / obesity [3,4]. Despite the nature of atrial remodeling in AF patients with HFpEF differs from that of HFrEF, left atrial myopathy is closely associated with cardiac fibrosis, oxidative stress, inflammation, neurohormonal activation and leads to progression of both of these conditions [4]. It has been suggested that a disproportionate prevalence of myocardial ischemia, loss of cardiac myocytes, exaggerated accumulation of the extracellular matrix in atrial tissue in HFrEF when compared with HFpEF potentiates a more rapid progression of left atrial (LA) myopathy than left ventricular (LV) dysfunction [5]. In contrast, hypertension, metabolic comorbidities (overweight/obesity, diabetes), age-associated disorders may intervene in left atrial structure and function before an occurrence of clinical signs / symptoms of HFpEF [6]. In fact, AF is considered a powerful trigger for progression of adverse cardiac remodeling and clinical manifestation of HFpEF regardless of the presentation of other risk factors and concomitant diseases. To best of our knowledge, guideline-directed medical therapy (GDMT) can prevent new cases of AF and the progression of atrial myopathy in patients with HF, however, AF can interfere with the efficacy of some components of GDMT for HFpEF [7,8,9]. As a result, identifying modifiable risk factors for AF may play a significant role in suitable preventive approach to detecting vulnerable patients with HFpEF [10].

Myocardium in HFpEF presents decreased respiratory function and increased reactive oxygen species production, which is considered as mitochondrial dysfunction [11], in close relation with exerkine, hepatokine and adipocytokine dysfunction, mitophagy abnormality, progenitor cell alteration that directly participate in dynamic nature of metabolic adaptation to disease progression [12,13,14]. Moreover, several phenotypes of all myocardial cells (cardiac myocytes, cardiac progenitor cells, endothelial cells, macrophages, fibroblasts, etc.), which are responsible for microvascular inflammation, abnormal vasoreactivity, cardiac hypertrophy and extracellular cardiac matrix accumulation/degradation, are closely regulated not only auto/paracrine factors, but also a large spectrum of cytokines and chemokines produced by skeletal myocytes, adipocytes, hepatocytes and kidney [15,16].

Adropin is a secreted multifunctional peptide that is responsible for metabolic flexibility and optimizing substrate utilization patterns in numerous physiological and pathological conditions [17]. Adropin expression is markedly regulated by various factors, including fat-reached diet, liver X receptor alpha, estrogen receptor alpha, retinoic acid receptor-related orphan nuclear receptor, regulator of reprogramming, and peroxisome proliferator-activated receptor gamma coactivator-1 expression by inhibiting sirtuin-1 and STAT3. Adropin was found to be expressed in various tissues, including the myocardium, skeletal muscles, adipose tissue, kidney, gastrointestinal tract and brain [17,18,19].

Adropin is involved in regulating glucose and lipid homeostasis, insulin sensitivity, adipogenesis, the inflammatory response, and immune reactions by modulating multiple signaling pathways (e.g. c-Jun N-terminal kinase [JNK], cAMP activated protein kinase A, activation of the glucose transporter protein and peroxisome proliferator-activated receptor gamma, Nrf2/HO-1, Vascular endothelial growth factor receptor [VEGFR]-2/phosphatidylinositol 3-kinase/AKT and VEGFR/ c-Src/ERK1/2 signaling, expressing antioxidant enzymes and Bcl-2 / Bax proteins). It also alleviates endoplasmic reticulum stress responses by reducing the phosphorylation of the inositol trisphosphate receptor in target cells [20,21]. Moreover, adropin acting through JNK may inhibit transforming growth factor-beta (TGF-beta)-induced fibroblast activation and fibrotic tissue remodeling and promotes tissue protection [22]. Other cardiovascular effects of adropin include the regulation of arterial stiffness and vasodilation via endothelium-derived nitric oxide bioavailability, reducing cardiac hypertrophy and fibrosis, enhancing diastolic function, anti-ischemic, anti-proliferative, anti-inflammatory and anti-apoptotic effects, mediating cardiac cells survival [23,24,25,26]. On contrary, a deficiency of adropin can disrupt the function of immune cells and inflammatory pathways, thereby impairing the regulatory capacity of the immune system and promoting systemic, microvascular and adipose tissue inflammation.

Previous clinical studies have revealed that circulating adropin levels were lower in patients with cardiovascular (atherosclerosis, acute myocardial infarction, acute and chronic HF, cardiac cachexia, stroke) and non-cardiovascular (metabolic dysfunction-associated steatotic liver disease, acute and chronic kidney diseases, overweight / obesity, type 2 diabetes mellitus) diseases [27,28,29,30,31,32]. For patients with HF regardless of its phenotypes, circulating levels of adropin were inversely correlated with HF severity [33,34]. Along with it, low levels of adropin are likely to be an independent predictor for occurrence of HF, kidney injury and diabetes mellitus, as well as to be the biomarker of poor clinical outcomes including cardiovascular death, HF-related hospitalization and diabetes-induced kidney disease [33,34,35,36]. However, the discriminative potency of adropin for AF in individuals with HFpEF remains unclear and requires face-to-face comparison with conventionally used and promising biomarkers, such as natriuretic peptides, cardiac troponins, soluble suppression of tumorigenicity-2 and galectin-3. The aim of the study is a) to investigation the possible discriminative value of adropin for new onset AF in patients with HFpEF who are being treated in accordance with conventional guideline and b) to compare it with predictive potencies of conventionally used predictors.

2. Materials and Methods

2.1. Study Population

We pre-screened 1253 adult patients with HFpEF in our local database according to the conventional criteria: left ventricular (LV) ejection fraction (EF) >50%; abnormal LV global longitudinal strain (GLS) < -16%, LV myocardial mass index (MMI) ≥95 g/m² in women or ≥115 g/m² in men, left atrial volume index (LAVI) >34 mL/m², and abnormal diastolic index E/e’ ≥ 13 units, N-terminal natriuretic pro-peptide (NT-proBNP) ≥ 300 pmol/mL [37]. Finally, between October 2020 and August 2022 were prospectively enrolled 953 outpatients with hemodynamically stable HFpEF and a presence of sinus rhythm on ECG and without a previous history of AF in the following medical centers: the private hospital Vita-Centre (Zaporozhye, Ukraine), EliteMedService (Zaporozhye, Ukraine), City Hospital #7 (Zaporozhye, Ukraine) and the private hospital “MIRUM clinic” (Kyiv, Ukraine). All patients continued to receive optimal GDMT and provided voluntary written informed consent to participate in the study (Figure 1). The major exclusion criteria were acute HF, HFrEF, HF with mildly reduced ejection fraction (HFmrEF), acute myocardial infarction, unstable angina, recent stroke / transient ischemic attack, acute viral and bacterial infection, known malignancy, active chemotherapy, with severe comorbidities, including end stage renal disease (ESRD), cognitive dysfunction/dementia, pregnancy/gestation. We followed patients for 3 years and divided them into two cohorts depending on the presence of AF: 172 patients exhibited AF, whereas 781 individuals preserved sinus rhythm on ECG.

2.2. Determination of AF and Follow-Up

We determined 3-year cases of any forms of AF according to 2020 ESC Guidelines for the diagnosis and management of AF [38]. To detect AF we utilized ECG at rest, continuous 72-h ECG monitoring (when needed), direct interview with patients and /or their relatives, contact with general practitioners, review of databases, and discharge reports. The following had to be present for a case of AF to be verified: an absence of distinct repeating P waves on an ECG; irregular atrial activations; and irregular R-R intervals. The patient-centered management of AF involved individualizing treatment strategies according to the conventional approach [39]. CHA2DS2-VASc and HAS-BLED scores were utilized to assess the risk of thromboembolism and bleeding and provide anticoagulation in patients with HFpEF after AF verification. Follow-up data were collected via clinic visits at baseline and during 3 years after the baseline with 3-month intervals.

2.3. Echocardiography Examination

In the study, all patients underwent a routine transthoracic B-mode and Doppler ultrasound examination. This was provided by an experienced echocardiographer using a GE Healthcare Vivid E95 scanner in apical 2- and 4-chamber views (General Electric Company, Horton, Norway). The conventional hemodynamic parameters included LVEF by using Simpson’s method, the LV end-diastolic (LVEDV) and end-systolic (LVESV) volumes, the left atrial volume index (LAVI), early diastolic blood filling (E), and the mean longitudinal strain ratio (e‘) were evaluated according to 2018 Guideline of the American Society of Echocardiography [40]. The estimated E/e' ratio was expressed as the ratio of the E-wave velocity to the average of the medial and lateral e' velocities. After acquisition of high-quality echocardiographic data during at least three cardiac cycles, LV GLS was obtained by 2D speckle-tracking image analysis. We stored the data in the DICOM format for subsequent analysis. Left ventricular hypertrophy was defined as an LVMMI of ≥95 g/m² in women and ≥115 g/m² in men [40].

2.4. Blood Sampling

Blood samples were obtained from all participants in fasting condition and collected in BD Vacutainer Serum Plus Tube. The samples were stored for 30 minutes at room temperature to clot. After clotting, the samples were centrifuged at 3000 rpm for 15 min at room temperature. Samples with hemolysis were not used for further evaluation. The supernatant was collected and stored at -70°C until analysis at certified laboratory of the Vita Centre (Zaporozhye, Ukraine).

2.5. Biomarkers Assessment

The local Vita Centre laboratory in Zaporozhye, Ukraine, used a Roche P800 analyzer from Basel, Switzerland, to determine conventional hematological and biochemical parameters. The following data were recorded: blood routine indices, electrolytes, liver enzymes, glucose, serum uric acid, serum creatinine and lipid profile (total cholesterol, triglycerides, high-density lipoprotein [HDL-C] and low-density lipoprotein [LDL-C] cholesterol). Concentrations of circulating biomarkers (N-terminal natriuretic pro-peptide [NT-proBNP], high-sensitivity cardiac troponin T [hs-TrT], soluble suppression of tumorigenicity-2 [sST2], galectin-3, tumor necrosis factor-alpha [TNF-α], high-sensitivity C-reactive protein [hs-CRP], and IL-6) were measured in serum using ELISA kits (Elabscience, Houston, Texas, USA) in accordance with the manufacturer's instructions. Adropin levels were measured with ELISA kit produced by Antibodies.com (Stockholm, Sweden). The data obtained from the ELISA analysis were subjected to standard curve-based evaluation. Each sample was analyzed in duplicate and the mean value was used for the final analysis. The intra- and inter-assay coefficients of variation for each marker were both below 10%.

2.7. Glomerular Filtration Rate Estimation

The CKD-EPI formula was utilized to estimate the glomerular filtration rate (eGFR) [41].

2.8. Determination of Insulin Resistance

The Homeostatic Assessment Model of Insulin Resistance (HOMA-IR) was performed to assess insulin resistance [42].

2.9. Statistics

All statistical analyses were conducted using SPSS 11.0 for Windows and Graph Pad Prism, version 9 (Graph Pad Software, San Diego, CA, USA). The data were tested for normality and homogeneity of variance using the Kolmogorov-Smirnov test. All continues variables were expressed as mean ± standard deviation [SD] median and interquartile range [IQR] depending on whether the data were normally distributed, whereas categorical variables were presented as number (n) and percentage (%). Spearman’s correlation coefficient (r) was utilized for correlations between hemodynamic parameters, comorbidities and biomarkers including the levels of adropin. We utilised the Benjamini-Hochberg procedure to calculate the false discovery rate-adjusted P-value for each pair of variables. The chi-squared test was used to compare categorical variables. To compare the data between the two groups, an independent samples t-test was used if the variances were homogeneous and a Mann–Whitney U test was used if they were not. Receiver operating characteristic (ROC) curves were used to calculate the cutoff values for variables in predicting AF. Youden's index (sensitivity + specificity - 1) was used to determine the optimal cut-off value for each possible predictor. Univariate and multivariate Cox proportional hazard models were employed to investigate the prognostic factors for AF. The odds ratio (OR) and 95% confidence interval (CI) were reported for each predictor variable. We compared the incremental prognostic capacity of models on using a binary prediction methodology based on the estimation of integrated discrimination indices (IDI) and net reclassification improvement (NRI). A p-value of less than 0.05 was considered to suggest a statistically significant difference.

3. Results

3.1. Baseline Clinical Characteristics

A total of 953 patients with HFpEF, New York Heart Association class II (n = 385) / III (n = 568) and were enrolled in the study and divided into two cohorts depending on a presence of new onset AF (n=172) or sinus rhythm on ECG (n=781) during 3 years. The clinical characteristics of patients outlined in Table 1.

The entire group of patients had a mean age of 69 years and 52.2% of them were male. The profile of comorbidity conditions included dyslipidemia (62.2%), hypertension (87.9%), stable coronary artery disease (36.0%), smoking (35.2%), abdominal obesity (29.2%), type 2 diabetes mellitus (37.4%), left ventricular hypertrophy (70.9%), and chronic kidney disease 1-3 stages (26.3%). Therefore, all patients were hemodynamically stable, had a mean systolic / diastolic blood pressure of 143 / 87 mm Hg, a mean LVEF of 54%, an average of LAVI of 38 mL/m², a mean of GLS of -15.9%. The therapy of HFpEF were personally optimized and included antagonists of renin-angiotensin-aldosterone system (ACE inhibitors, or angiotensin-II receptor blockers or angiotensin receptor-neprilysin inhibitors), beta-blockers, diuretics, sodium–glucose cotransporter-2 inhibitor and other concomitant medications depending on the presence of comorbidities and diseases.

Patients from AF cohort were older, more likely to have type 2 diabetes mellitus and CKD stages 1-3, higher LAVI, concentrations of hs-CRP, NT-proBNP, sST2 and lower adropin levels than those with sinus rhythm on ECG. The patients with HFpEF from the cohort with sinus rhythm on ECG tended to treat frequently with angiotensin-II receptor blockers, beta-blockers, and SGLT2 inhibitors when compared with those who had AF.

We did not find significant differences between the two patients cohorts with respect to gender, body mass index, anthropometric parameters, the presence of dyslipidemia, hypertension, stable coronary artery disease, smoking, abdominal obesity, left ventricular hypertrophy, New York Heart Association classes, systolic and diastolic blood pressure, left ventricular end-diastolic and end-systolic volumes, LVEF, LVMMI, E/e`, GLS, eGFR, lipid profile, glucose levels, HOMA-IR, HbA1c, and concentrations of TNF-alpha, galectin-3, IL-6, and hs-TrT.

3.2. Spearman’s Correlations Between Hemodynamic Parameters, Comorbidities and Biomarkers

The heat map provides a visual representation of the Spearmen correlation coefficients between each pair of variables including age, gender, hemodynamic parameters, comorbidities and circulating biomarkers (Figure 2). The study has shown that the levels of serum adropin positively correlated with LVEF (r = 0.33; P = 0.001), GLS (r = 0.33; P = 0.001), HDL-C levels (r=0.30; p=0.001), creatinine (r=0.26; p=0.002) and inversely correlated with LAVI (r = -0.36; P = 0.001), CKD (r = -0.34; P = 0.001), T2DM (r = -0.39; P = 0.001), BMI (r = -0.31; P = 0.001), triglyceride levels (r = -0.30; P = 0.001), fasting glucose (r = -0.30; P = 0.001), HOMA-IR (r = -0.30; P = 0.012), LDL-C (r = -0.24; P = 0.001) and NT-proBNP(r = -0.21; P = 0.012).

3.3. Predictive Factors of AF: The Results of the Receiver Operating Characteristic Curve Analysis

ROC curve analysis was performed to determine the optimal cutoff for possible predictors of new onset AF in patients with HFpEF (Table 2). Age ≥75 years showed the area under curve (AUC) for AF of 0.697 (95% confidence interval [CI] = 0.712 – 0.740, P = 0.001) with a sensitivity of 70.3% and a specificity of 61.6%.

The well-balanced predictive cutoff point for LAVI was 40 mL/m² (AUC = 0.841, 95% CI = 0.735 – 0.932) with a sensitivity of 72.1% and a specificity of 79.3%. The AUC for NT-proBNP was 0.843 (95% CI = 0.751 – 0.932) and predictive cutoff point was 1440 pmol/mL (sensitivity = 84.2%; specificity = 77.8%). Optimal cutoff points for hs-CRP and sST2 were 5.4 mg/L (AUC = 0.753, 95% CI = 0.646 – 0.860, sensitivity = 61.9%; specificity = 67.7%) and 15.5 ng/mL (AUC = 0.839, 95% CI = 0.766 - 0.912; sensitivity = 77.9%; specificity = 71.1%), respectively. The AUC for adropin was 0.893 (95% CI = 0.827 – 0.959) and predictive cutoff point was 2.95 ng/mL (sensitivity = 89.4%; specificity = 73.0%). ROC curves for these variables are presented by Figure 3.

3.4. Predictive Factors for New Onset AF: Unadjusted and Adjusted for Multivariate Cox Proportional Hazard Models

Cox proportional hazards model was constructed for assessment of independent predictors of AF (Table 3). Significant variables (P<0.05) comparing cohorts with and without AF were entered into univariate Cox regression analysis, and those retaining statistical significance (P<0.05) were entered into multivariate Cox proportional hazards model.

In unadjusted rough Cox regression model, age ≥ 75 years, type 2 diabetes mellitus, CKD stages 1–3, LAVI ≥40 mL/m², NT-proBNP ≥1440 pmol/mL, hs-CRP ≥5.40 mg/L, adropin ≤2.95 ng/mL, sST2 ≥15.5 ng/mL were identified as the predictors for new onset AF in HFpEF patients. After adjusting for age ≥ 75 years (Model 2), the presence of type 2 diabetes mellitus, CKD stages 1–3, LAVI ≥40 mL/m², NT-proBNP ≥1440 pmol/mL, adropin ≤2.95 ng/mL had discriminative values for AF. Model 3 has revealed that the only the levels of NT-proBNP ≥1440 pmol/mL and adropin ≤2.95 ng/mL were independent predictors of AF.

3.5. Comparison of the Predictive Models

We found that discriminative value of adropin was superior to NT-proBNP, while adding adropin to NT-proBNP did not improve predictive information of adropin alone.

Table 4. The comparisons of predictive models for AF.

Models	AUC		NRI		IDI
	M (95% CI)	p value	M (95% CI)	p value	M (95% CI)	p value
NT-proBNP ≥1440 pmol/mL	0.843 (0.751–0.932)	-	Reference		Reference
Adropin ≤2.95 ng/mL	0.893 (0.827–0.959)	0.040	0.25 (0.21–0.30)	0.046	0.41 (0.32–0.52)	0.044
NT-proBNP ≥1440 pmol/mL + Adropin ≤2.95 ng/mL	0.929 (0.837–0.974)	0.048	0.29 (0.24–0.35)	0.040	0.48 (0.40–0.56)	0.042

Abbreviations: AUC, area under curve; CI, confidence interval; IDI, integrated discrimination indices; M, mean value; NRI, net reclassification improvement; NT-proBNP, N-terminal brain natriuretic pro-peptide.

Table 4. The comparisons of predictive models for AF.

Models	AUC		NRI		IDI
	M (95% CI)	p value	M (95% CI)	p value	M (95% CI)	p value
NT-proBNP ≥1440 pmol/mL	0.843 (0.751–0.932)	-	Reference		Reference
Adropin ≤2.95 ng/mL	0.893 (0.827–0.959)	0.040	0.25 (0.21–0.30)	0.046	0.41 (0.32–0.52)	0.044
NT-proBNP ≥1440 pmol/mL + Adropin ≤2.95 ng/mL	0.929 (0.837–0.974)	0.048	0.29 (0.24–0.35)	0.040	0.48 (0.40–0.56)	0.042

Abbreviations: AUC, area under curve; CI, confidence interval; IDI, integrated discrimination indices; M, mean value; NRI, net reclassification improvement; NT-proBNP, N-terminal brain natriuretic pro-peptide.

3.6. Reproducibility of Adropin Versus NT-proBNP

The assessment of the reproducibility regarding adropin compared with NT-proBNP has shown that the intra-class correlation coefficients for inter-observer reproducibility of NT-proBNP and adropin were 0.75 (95% CI = 0.69-0.82) and 0.86 (95% CI = 0.80-0.91), respectively. We did not find significant changes in intra-observer reproducibility of adropin in relation to age and gender in eligible patients (p = 0.316 and p = 0.412).

4. Discussion

In the 3-year longitudinal multicenter study, we found that age ≥ 75 years, type 2 diabetes mellitus, CKD stages 1–3, LAVI ≥40 mL/m², NT-proBNP ≥1440 pmol/mL, hs-CRP ≥5.40 mg/L, adropin ≤2.95 ng/mL, sST2 ≥15.5 ng/mL significantly predicted incident AF. After adjusting for age ≥ 75 years, the presence of type 2 diabetes mellitus and CKD stages 1–3 adropin and NT-proBNP remained independent predictors for AF in this patient population. Moreover, we have first identified a higher predictive value of low circulating adropin when compared with elevated NT-proBNP for new episodes of AF in symptomatic patients with HFpEF.

Despite a wide implementation of international guideline for AF in the routine clinical praxis, identifying patients at high risk of new AF remains challenging. Previous clinical studies and meta-analysis have shown that AF episodes and their progression from paroxysmal to non-paroxysmal forms depend on combinations of several conditions and parameters, including older age, a history of angina pectoris, myocardial infarction, hypertension, diabetes, thyroid dysfunction, LVH, cardiac surgery, cardiomyopathies, LA asynchrony and LA remodeling, heart rhythm disturbance, LV diastolic and systolic dysfunction, low hematocrit and hemoglobin, elevated levels of HbA1c, natriuretic peptides, galectin-3, sST2, and hs-CRP [43,44,45,46,47]. There is little evidence for HFpEF patients without a previous history of AF who were treated with optimal GDMT. Although elevated level of NT-proBNP was a predictor of incident AF during the first 2 years of long-term observation, it remains unclear, whether the discriminative value of natriuretic peptide(s) remains accurate in individuals with HFpEF and coexisting overweight, obesity or type 2 diabetes mellitus [48,49]. Indeed, glycosylation and obesity may affect the prognostic value of NT-proBNP regarding AF [50,51], while NT-pro-BNP significantly improved the predictive ability of conventional cardiovascular risk factors and the novel CHARGE-AF risk score for AF [52]. On the other hand, the biomarkers, which relate to cardiac fibrosis (galectin-3, sST2), inflammation (ha-CRP, IL-6, TNF-alpha) and myocardial injury (cardiac troponins), did not provide the superiority over NT-proBNP in predicting AF in individuals without previous incident of AF [52,53]. In this context, a discovery of new non-invasive, easily accessible biomarkers that could stratify HFpEF patients depending on the risk of new onset AF appears to be promising.

In our study, we have identified low levels of adropin as promising biomarker of higher risk of AF in HFpEF patients who were optimally treated with GDMT and had target / near target levels of NT-proBNP (<1000 pmol/mL). These findings are likely to be practically useful, because they allow providing a clear approach to risk stratification beyond respectively time-consuming analysis of LA remodeling. Because in recent clinical studies has been proven a linear correlation between serum adropin concentrations and LA diameter, LAVI, LA strain in patients with AF [54,55], it could suggest that a decrease in the concentration of this peptide may precede the manifestation of AF and LA volume overload.

Being not only an independent risk factor for heart disease, but also a metabolic regulator of the cardiovascular system's adaptation to the progression of adverse cardiac remodeling and HF, adropin is likely to be a promising prognostic biomarker for these conditions. Unfortunately, there is little rigorous clinical evidence of the superiority of adropin over natriuretic peptides and other biomarkers of biomechanical stress, inflammation, cardiac fibrosis, and myocardial damage. Most of them were obtained in cohort studies and require further validation [28,30,33,35,55]. Nevertheless, our findings suggest that adropin may play a protective role in cardiovascular remodeling and that its deficiency represents a metabolic maladaptation that potentiates accelerated HF progression. Indeed, extensive researches have explored the pivotal role of adropin, particularly in mechanisms related to inflammation, immune response and oxidative stress through suppression of VEGFR2/phosphatidylinositol 3-kinase/AKT and VEGFR/c-Src/ERK1/2 signaling [56]. Moreover, adropin demonstrated direct cardiac and vascular protective effects via modulating brain-derived neurotrophic factor and inhibiting SIRT1 and TNF-alpha-depending apoptosis [57].

Another aspect of an association of low levels of adropin with the risk of AF is their possible link with endothelial dysfunction, accelerating atherosclerosis and lipid toxicity. Perhaps the loss of the suppressive effect of adropin on ROR-dependent regulation of lipid oxidation may significantly mediate its negative effect on the lipid core formation in atherosclerotic plaques, sub-intimal accumulation of oxidized lipoproteins, disruption of integral functions of the vascular wall, and regulation of vasodilation [56,58]. Therefore, adropin downregulated lipogenic proteins SEBP-1 and ADRP and thereby potentiates lipid toxicity, ROS production and mitochondrial dysfunction [59]. Overall, all these pathogenetic pathways are likely to be crucial for AF development.

On the other hand, in our study we did not find any significant differences in glucose homeostasis parameters, lipid profile between HFpEF patients with AF and with sinus rhythms, whereas adropin levels were sufficient different. However, the associations between adropin levels and the parameters mentioned above were found. We hypothesized that adropin playing an important role in the overall regulation of glucose and lipid metabolism is not the plausible factor that mediates variability of these parameters. On the contrary, liver perfusion and hepatocyte metabolism are likely to be in the loop of the regulation of adropin levels in circulating blood. Along with it, adipose tissue inflammation and concomitant diabetes may intervene in expression of adropin in remote tissue and thereby limit organ protective capability of the peptide.

Finally, as we had hypothesized, adropin is a promising indicator of a higher risk of new onset AF in individuals with HFpEF, whose predictive potency exceeds that of NT-proBNP. These findings open up a new perspective for a large-scale clinical study to develop a new risk stratification model for these patients.

5. Study Limitations

The study has several limitations. The first limitation affects the lack of data regarding a trajectory of biological markers during the observational period in connection with new cases of AF. Additionally, we did not investigate nutritional status among the patients and its association with the changes of adropin levels. Finally, we did not compare the discriminative value of adropin with previously validate risk scores, such as Cohorts for Heart and Aging Research in Genomic Epidemiology model for atrial fibrillation (CHARGE-AF). However, we do not believe that these limitations will have an impact on the interpretation of the results.

6. Conclusions

We found that after adjusting for age ≥ 75 years, a presence of type 2 diabetes mellitus and CKD stages 1–3, the levels of NT-proBNP ≥1440 pmol/mL and adropin ≤2.95 ng/mL were independent predictors of new onset AF in patients HFpEF. Moreover, adropin ≤2.95 ng/mL presented more predictive information than NT-proBNP ≥1440 pmol/mL alone, whereas the combination of both biomarkers did not improve the predictive ability of adropin alone.