Hepatocyte Growth Factor: A Marker of Cardiac Function, Mortality, and Disease Subtype in Cardiac Amyloidosis

Abstract

Background: It is important to reduce diagnostic delays for patients with cardiac amyloidosis (CA). Plasma biomarkers could streamline the diagnostic process and enhance prognostic accuracy.

Objectives: The authors aimed to identify circulating biomarkers capable of differentiating patients with CA from patients with heart failure (HF) and no amyloidosis. Additionally, we assessed whether these markers were associated with patient outcomes.

Methods: We performed focused protein screening in 12 patients with transthyretin CA, 5 patients with HF, and 16 healthy controls (HCs). To validate the findings, we used immunoassays to measure levels of differentially regulated proteins in a larger sample of 86 patients with transthyretin CA, 15 patients with light-chain CA, 16 patients with HF, and HCs. We compared protein levels between groups using multivariable general linear models. Associations between protein levels and all-cause mortality were assessed by receiver operating characteristic analysis.

Results: We identified 99 candidate proteins by proteomic screening. In the validation sample, 4 of these markers were higher in CA than in HCs. Levels of C-X-C motif chemokine ligand 9 and hepatocyte growth factor (HGF) were also higher in CA than in HF. HGF correlated with measures of cardiac function in patients with transthyretin and light chain CA. HGF had a good discriminatory ability for predicting all-cause mortality (area under the curve = 0.80, P < 0.001), similar to those of N-terminal pro-B-type natriuretic peptide and troponin T.

Conclusions: Plasma HGF is a promising screening tool for CA. Higher levels of HGF are associated with more severe HF and worse prognosis in patients with CA.

Keywords: ATTR-CM; amyloid light chain; biomarker; cardiac amyloidosis; hepatocyte growth factor.

Graphical abstract

Figure 1

Proteomic Analysis of Plasma in Patients With Transthyretin Amyloid Cardiomyopathy (A) Heatmap showing proteins dysregulated in ATTR-CM compared to patients with HF caused by other medical conditions than ATTR (HF) and HCs. The proteins were quantified using 3 Olink panels; (B) Tukey plots of normalized protein expression levels in ATTR-CM, HF, and HC. NPX, CXCL9, HGF, SERPINA12, DCN, TNFRSF13B, DPP4, TGFβR3, APOM. ∗P < 0.01, ∗∗P < 0.01, ∗∗∗P < 0.001 between indicated groups. Black dots indicate outliers. APOM = apolipoprotein M; ATTR-CM = transthyretin amyloid cardiomyopathy; CXCL9 = C-X-C motif chemokine ligand 9/monokine induced by gamma interferon; DCN = decorin; DPP4 = dipeptidyl peptidase 4; HC = healthy control; HF = heart failure; HGF = hepatocyte growth factor; NPX = normalized protein expression; SERPINA12 = serpin family A member 12; TGFβR3 = transforming growth factor beta receptor 3; TNFRSF13B = tumor necrosis factor receptor superfamily member 13B.

Figure 2

Validation of Dysregulated Proteins in Patients With Transthyretin Amyloid Cardiomyopathy Box-and-whisker plots of proteins assessed in a larger sample of patients with ATTR-CM (n = 86), HCs (n = 23), other HF (n = 16), and AL amyloidosis (n = 15). The proteins were assessed with enzyme immunoassays. CXCL9, HGF, DCN, TNFRSF13B, DPP4, TGFβR3, APOM. The top left P value reflects the overall group between-subjects effect from the GLM using age, sex, and BMI as covariates. Post hoc comparisons in the GLM were Sidak-adjusted. ∗P < 0.01, ∗∗P < 0.01, ∗∗∗P < 0.001 vs HC; ^†P < 0.05, ^††P < 0.05, ^†††P < 0.001 vs HF. P values comparing ATTR-CM and AL are indicated directly in graphs. AL = amyloid light chain; BMI = body mass index; GLM = general linear model; other abbreviations as in Figure 1.

Figure 3

Association Between Dysregulated Proteins and Cardiac Measures in Transthyretin Amyloid Cardiomyopathy and Amyloid Light Chain Amyloidosis (A) Box-and-whisker plots of plasma CXCL9 and HGF in relation to NYHA functional class. ∗P < 0.05, ∗∗P < 0.01 NYHA functional class III (n = 15) vs NYHA functional class I/II (n = 54) as analyzed by multivariable GLM with age, sex, and BMI as covariables. (B) Heatmap showing correlations (Spearman) between plasma proteins and echocardiographic, hemodynamic, and serologic indices of cardiac function. ∗P < 0.05, ∗∗P < 0.01. (C) HGF levels according to clinical cutoffs for left ventricular hypertrophy (ie, LVMI above sex-adjusted cutoff²¹); EF below 50%; pulmonary hypertension (PHTN, ie, mPAP >20 mm Hg) ∗P < 0.05. (D) Correlation plots showing association between HGF and LVEF (left panel) and mPAP (right panel) in ATTR-CM and AL amyloidosis. ∗∗∗P < 0.001. Correlation coefficients (rho) are given in the graphs in ATTR-CM (green) and AL amyloidosis (green). Global longitudinal strain was reported as negative values. Therefore, a positive correlation between the biomarkers and GLS shows that increasing marker levels were associated with poorer left ventricular function. AL = amyloid light chain; ATTR-CM = transthyretin amyloid cardiomyopathy; CO = cardiac output; cTnt = cardiac troponin T; CXCL9 = C-X-C motif chemokine ligand 9/MIG; EF = ejection fraction; GLS = global longitudinal strain; HGF = hepatocyte growth factor; IVSD = interventricular septum diastolic diameter; LVEF = left ventricular ejection fraction; LVMI = left ventricular mass index; mPAP = mean pulmonary artery pressure; NT-proBNP = N-terminal pro-brain natriuretic peptide; PHTN = pulmonary hypertension; PVR = pulmonary vascular resistance; RAP = right atrial pressure.

Figure 4

Association Between Dysregulated Proteins and All-Cause Mortality in Transthyretin Amyloid Cardiomyopathy and Amyloid Light Chain Amyloidosis (A) Association between dysregulated proteins (ie, CXCL9, HGF, TNFRSF13B, Galectin-9), cardiac markers (NT-proBNP, cTnt), CRP, and all-cause mortality (n = 20) as assessed by the area under the ROC. Numbers indicate the AUC for different markers. ∗P < 0.001. (B) Kaplan-Meier analysis of all-cause mortality according to dichotomized HGF levels (cutoff determined by Youden index). The log-rank P value is shown. (C) Cox regression of HGF and all-cause mortality with different levels of adjustment. Uni indicated the unadjusted association between HGF, NT-proBNP, cTnt, and all-cause mortality, while HGF+ indicates models with HGF + one-by-one adjustment with different confounders (PS1 is a propensity score with age, sex, and BMI; PS2 is a propensity score including all confounders: age, sex, BMI, eGFR, NT-proBNP, cTnt, EF50, and NYHA functional class III). ∗P < 0.01; ∗∗P < 0.001. AUC = area under the curve; BMI = body mass index; CRP = C-reactive protein; cTnt = cardiac troponin T; cTnt = cardiac troponin T; CXCL9 = C-X-C motif chemokine ligand 9/monokine induced by gamma interferon; EF50 = left ventricular ejection fraction <50; eGFR = estimated glomerular filtration rate; HGF = hepatocyte growth factor; NT-proBNP = N-terminal pro-brain natriuretic peptide; ROC = receiver-operating characteristics curve; TNFRSF13B = tumor necrosis factor receptor superfamily member 13B.

Central illustration

High Levels of Hepatocyte Growth Factor Are Associated With Disease Severity and Survival in Cardiac Amyloidosis We used targeted proteomics to discover potential biomarkers in patients with cardiac amyloidosis. To validate the result, we evaluated the regulated proteins in a larger cohort, using immunoassays. Levels of hepatocyte growth factor were significantly higher in patients with cardiac amyloidosis than in patients with HF without amyloidosis and HCs. High levels of HGF were associated with echocardiographic and hemodynamic measures of poor cardiac function. High levels of HGF were independently associated with all-cause mortality. AL = amyloid light chain; HC = healthy control; HF = heart failure; other abbreviations as in Figure 1.