Human study design and population

In total, 40 subjects who underwent electrophysiological interventional therapy at our hospital between June 2018 and June 2019 were enrolled. The exclusion criteria were as follows: (1) age <18 years; (2) systemic complications, such as malignant tumour, autoimmune disease, endocrine disease or end-stage renal dysfunction; and (3) blood-borne infectious diseases, such as hepatitis B, hepatitis C and human immunodeficiency virus/acquired immunodeficiency syndrome. Clinical status was used to classify patients into two groups: (1) patients with HF-associated symptoms and an LV ejection fraction (LVEF) ≤40% (HF group) and (2) patients diagnosed with AF or supraventricular tachycardia undergoing catheter ablation, with no evidence of HF (control group). Clinical parameters, including biochemistry, and echocardiographic parameters were obtained from the electronic health records. The levels of BNP, galectin 3 and TIMP-1 in the artery blood samples of healthy volunteers (samples were obtained from our previous research cohort¹⁴) were measured and compared with those of our HF patients and the controls.

All participants provided written informed consent. This study conformed to the principles of the Declaration of Helsinki and was approved by the Ethics Committee of the hospital (2013-496).

Sample collection and preparation

Blood samples (3–4 mL) from the femoral artery (FA) and coronary sinus (CS) were simultaneously collected using EDTA VACUETTE tubes (Greiner Bio-One, Thailand). The samples were immediately centrifuged at 2000 g for 10 min at 4°C. The plasma (supernatant) was immediately separated and stored at −80°C until use.¹⁵

Biomarker measurement

Plasma levels of the six biomarkers were measured using enzyme-linked immunosorbent assay (ELISA) kits according to the manufacturer's instructions, as follows: Human BNP ELISA Kit (Abcam, ab193694), Human Galectin-3 Quantikine ELISA Kit (RD, DGAL30), Human GDF-15 Quantikine ELISA Kit (RD, DGD150), Myeloperoxidase Human Instant ELISA Kit (Invitrogen, BMS2038INST), Human ST2/IL-33R Quantikine ELISA Kit (RD, DST200) and Human TIMP-1 ELISA Kit (Abcam, ab100651). In this study, transcoronary changes were calculated as the difference between the FA and the CS.

HF model establishment

Seventeen male beagles (12 months old; 11–13 kg) were randomly assigned to the sham (N = 5), HF (N = 6) or therapy group (N = 6). One week after settling, all beagles underwent echocardiography and electrocardiography (ECG). Beagles in the HF and therapy groups underwent cardiac pacemaker implantation, while those in the control group underwent a sham operation.

Animals were anaesthetized using intravenous pentobarbital (3%; 30 mg/kg). Under fluoroscopy, the atrial lead was implanted into the right auricle, a ventricular passive lead (7231, Shanghai Lepu Medical Instrument Co., Ltd.) was implanted in the apex of the right ventricle and a ventricular active lead (7211, Shanghai Lepu Medical Instrument Co., Ltd.) was implanted in the septum. The leads were connected to a pacemaker placed subcutaneously on the right thoracic wall. After the operation, pacemakers were turned on using a programming instrument (QM2352A, dual-chamber pacing system analyser, Shanghai Lepu Medical Instrument Co., Ltd.). Pacing was continued for 6 weeks (pacing mode, AOO; pacing frequency, 250 times/min; output, 3.6 V; and pulse width, 0.4 ms). To prevent infection, beagles were injected with 80 U penicillin intramuscularly twice a day for 7 days and fed with a high-protein diet.

After 6 weeks of pacing, follow-up echocardiography and ECG were performed. Subsequently, in the therapy group, pacemakers were replaced with a cardiac contractility modulation (CCM) device (Optimizer II; Impulse Dynamics, USA), which enhances contractility and is beneficial for HF treatment. The parameters were set as follows: pulse width, 2 ms; output voltage, 7 V; R-wave delay, 30 ms; and stimulation, 6–8 h every day. After 1 week, echocardiography and ECG were repeated.

Animal study

The experimental animals were euthanized, hearts were flushed with phosphate-buffered saline (PBS) before excision and cardiac tissue samples were harvested. Tissue samples were washed with cold PBS to remove the blood and were quickly snap-frozen in liquid nitrogen and stored at −80°C until use. All animal experiments complied with the ARRIVE guidelines and were performed in accordance with the National Institutes of Health's (NIH) Guide for the Care and Use of Laboratory Animals (NIH Publications No. 8023, revised 1978). The study protocol was approved by the Ethics Committee of our hospital (FW-2022-0051).

Pathology and molecular biology methods are described in the supporting information.

Statistical analysis

For the clinical characteristics, continuous variables are shown as mean [standard deviation (SD)] or median [with inter-quartile range (IQR)], as appropriate, while categorical variables are shown as number (percentage). Between-group comparisons were performed using Student's t test for normally distributed and the Kruskal–Wallis test for non-normally distributed continuous variables. Categorical variables were compared using Pearson's χ² test. Spearman's test was used for correlation calculation. Data were analysed using STATA 15.0. Plots were generated through Prism 7.0 (GraphPad Software, Inc., La Jolla, CA, USA).

Clinical characteristics in HF patients

In this study, a cohort of 30 patients with HF and 10 healthy controls were enrolled. The clinical characteristics are detailed in Table 1. HF patients were stratified according to the New York Heart Association (NYHA) functional classification, with Classes II–IV, while the control group comprised individuals with normal cardiac function. There were significant differences between HF patients and controls in left atrial diameter [LAD, 42.0 (38.0–45.0) vs. 34.0 (31.0–36.0) mm, P < 0.01, data shown as the median with IQR] and LV end-diastolic dimension (LVEDD, 67.3 ± 9.5 vs. 45.7 ± 3.2 mm, P < 0.01). Moreover, the LVEF was markedly lower in the HF group, with a mean value of 30.5% (IQR: 26.0–38.0) compared with 62.5% (IQR: 62.0–66.0) in the controls (P < 0.01). Additionally, the plasma levels of NT-proBNP were significantly elevated in HF patients, with a median value of 941.5 (IQR: 374.9–2114.0) pg/mL versus 29.4 (IQR: 14.8–82.7) pg/mL in the control group (P < 0.01). No significant differences were noted between the two groups in the remaining evaluated parameters.

Plasma biomarker levels and transcoronary changes of HF biomarkers

There were no significant differences in the biomarker between the control and healthy populations. This finding underscores the representativeness of our control group (Figure S1). In comparison with the control group, the HF group exhibited significantly elevated levels of BNP in both the FA and CS samples (Figure 1A; FA: 841.5 ± 727.2 vs. 67.9 ± 21.3 ng/mL, P < 0.0001; CS: 1132.0 ± 959.1 vs. 90.40 ± 67.5 ng/mL, P < 0.0001, respectively). Notably, the control group did not exhibit a transcoronary gradient for BNP, whereas the CS levels were significantly higher than the FA levels in the HF group (Figure 1B, 1132.0 ± 959.1 vs. 841.5 ± 727.2 ng/mL, P = 0.005). As well, Gal-3 and TIMP-1 levels were markedly increased in the HF group compared with the control group across both FA and CS measurements (Figure 1C, Gal-3: FA: 9.5 ± 3.0 vs. 6.2 ± 3.1 ng/mL, P = 0.004; CS: 19.7 ± 16.4 vs. 4.6 ± 1.6 ng/mL, P < 0.0001; Figure 1E, TIMP-1: FA: 286.7 ± 68.9 vs. 223.1 ± 43.0 ng/mL, P = 0.0028; CS: 377.3 ± 108.9 vs. 237.7 ± 75.5 ng/mL, P = 0.0007, respectively). And transcoronary gradients for Gal-3 and TIMP-1 were also observed only in the HF group (Figure 1D, Gal-3: FA: 9.5 ± 3.0 vs. CS: 19.7 ± 16.4 ng/mL, P = 0.002; Figure 1F, TIMP-1: 286.7 ± 68.9 vs. 377.3 ± 108.9 ng/mL, P = 0.001). These results suggest that cardiac tissue could produce BNP, Gal-3 and TIMP-1, which are released into the circulation in HF settings. Although the HF group had significantly higher FA and CS levels of GDF-15 (Figure 1G,H, 782.7 ± 232.5 vs. 2338.0 ± 2373.0 pg/mL, P < 0.0001; 837.0 ± 326.8 vs. 2302.0 ± 1838.0 pg/mL, P < 0.0001), MPO (Figure 1I,J, 39.5 ± 27.5 vs. 152.2 ± 95.3 ng/mL, P < 0.0001; 36.3 ± 26.3 vs. 183.7 ± 169.8 ng/mL, P < 0.0001) and sST2 (Figure 1K,L, 17 637.0 ± 9341.0 vs. 28 938.0 ± 15 507.0 pg/mL, P = 0.0137; 16 219 ± 6733 vs. 27 263 ± 11 663 pg/mL, P = 0.0068; respectively) than the control group, transcoronary gradients were not observed in either group, indicating no significant production or absorption of these biomarkers in the heart.

Correlation between biomarker gradients and clinical parameters

The correlation analysis unveiled a significant correlation between BNP levels and HF-associated clinical features, such as LVEF and LVEDD (Figure S2). Transcoronary BNP gradients also exhibited strong correlations with these clinical features (LVEF, r = −0.68, P < 0.0001; LVEDD, r = 0.57, P = 0.002) (Figure 2A,B). Notably, BNP levels did not correlate with blood urea nitrogen, creatinine, hepatorenal function indices, including alanine aminotransferase and aspartate aminotransferase, nor with inflammatory markers such as high-sensitivity C-reactive protein and cardiac troponin T (Figure S2). Also, significant correlations were observed between LVEDD and the transcoronary gradients of TIMP-1 (Figure 2C, r = 0.37, P = 0.04). Beyond these, no other significant correlations were found between other biomarker gradients and clinical characteristics (Figures S2 and S3). The findings revealed that BNP, as a biomarker for HF, is of great advantage for clinical applications.

Intracardiac production of HF biomarkers in HF model

An illustrative schematic of the model development is depicted in Figure 3A. Dogs subjected to HF and CCM groups developed characteristic HF symptoms, such as diminished appetite, tachypnoea and reduced physical activity, at 6 weeks after induction. In contrast, the sham group remained asymptomatic throughout the study period. Baseline echocardiographic assessments revealed no significant intergroup differences. However, after 7 weeks, the HF group exhibited a marked decline in cardiac architecture and function, as evidenced by increased LVEDD, decreased LVEF, LV fractional shortening (LVFS) and an elevated E/A ratio compared with the control group (LVEDD: 41.5 ± 3.7 vs. 24.0 ± 8.3 mm, P < 0.0001; LVEF: 36.8 ± 2.3% vs. 62.4 ± 3.8%, P < 0.0001; LVFS: 27.7 ± 5.0% vs. 39.2 ± 4.3%, P = 0.001; and E/A ratio: 1.9 ± 0.18 vs. 1.3 ± 0.25, P = 0.003, Figure 3A–E). The HF dogs exhibited a partial restoration of cardiac mechanical function in response to CCM therapy (LVEDD: 31.7 ± 4.2 vs. 41.5 ± 3.7 mm, P = 0.01; LVEF: 53.6 ± 6.0% vs. 36.8 ± 2.3%, P < 0.0001; LVFS: 34.5 ± 6.7% vs. 27.7 ± 5.0%, P = 0.06; and E/A ratio: 1.7 ± 0.3 vs. 1.9 ± 0.18; P = 0.17; Figure 3B–E). As evidenced by histological staining, the HF group had an increased deposition of blue-stained collagen fibres in LV, whereas CCM therapy partially reversed the pathological changes (Figure 3F,G). All the above results confirmed the successful construction of the HF model, which provided a solid material basis for subsequent research.

We employed real-time quantitative PCR (qPCR) to quantify mRNA levels in distinct chambers of the heart. For concordance with the protein levels in plasmas as quantitated by ELISA, we found that the mRNA expression levels of BNP, Gal-3 and TIMP-1 were significantly elevated in the cardiac tissues of the HF group compared with those in other groups (Figure 4A–C). In the HF group, BNP expression was clearly elevated throughout the heart tissues (Figure 4A). Gal-3 expression was predominant in the LV, whereas the expression of TIMP-1 is predominantly localized to the right heart (Figure 4B,C). The gene expression of ST2, GDF-15 and MPO in the myocardium did not differ statistically significantly between the three groups (Figure 4D–F). These results were in agreement with the transcoronary changes observed in the ELISA study.

We performed western blot (WB) to assess the protein levels in myocardium in order to confirm the elevated expression of BNP, Gal-3 and TIMP-1 in HF cardiac tissues. As illustrated in Figure 5, the myocardium (all atrial and ventricular tissues) showed increased BNP expression in the HF group when compared with the control group (Figure 5A). Gal-3 protein expression was significantly higher in ventricular tissue of HF hearts, especially LV (Figure 5B). Same as reverse transcription PCR (RT-PCR) analysis of cardia tissues indicating, protein TIMP-1 expression is mainly elevated in the right heart (Figure 5C). And the gene expression signature of BNP, Gal-3 and TIMP-1 in HF cardiac tissues could at least partially reverse by administration of therapy for HF (CCM therapy) (Figure 5D–F).