Abstract
Background
Angiopoietin-like 3/4/8 (ANGPTL3/4/8) proteins play critical roles in modulating lipid metabolism through calorically sensitive and tissue-specific regulation of lipoprotein lipase (LPL) activity via formation of ANGPTL3/8 and ANGPTL4/8 complexes. ANGPTL3/4/8 proteins are also associated with cardiovascular risk. Circulating levels of ANGPTL3, ANGPTL4/8, and C-terminal domain-containing ANGPTL4 (CD-ANGPTL4) have been associated with overall cardiovascular mortality. In this study, we investigated the associations of ANGPTL3/4/8 proteins and complexes with heart failure (HF) and HF death.
Methods
We studied 2,394 participants of the LUdwigshafen RIsk and Cardiovascular Health (LURIC) study, a cohort of patients referred for coronary angiography. HF with reduced ejection fraction (HFrEF) and HF with preserved ejection fraction (HFpEF) were diagnosed at baseline. There was a follow-up period with a median (interquartile range) duration of 9.80 (8.75–10.40) years. ANGPTL3/4/8 proteins and complexes were measured at baseline using dedicated immunoassays, and their serum concentrations were compared to HF data.
Results
There was an inverse association of ANGPTL3/8 with the N-terminal prohormone of B-type natriuretic peptide (NT-proBNP). ANGPTL3 was positively associated with NT-proBNP, HFpEF, and the New York Heart Association (NYHA) functional class. ANGPTL4/8 was positively associated with HFrEF, HFpEF, and NT-proBNP and was inversely associated with the left ventricular ejection fraction (LVEF). CD-ANGPTL4 was positively associated with the NYHA functional class, HFrEF, HFpEF, and NT-proBNP and was inversely associated with LVEF. A total of 101 participants died from HF. ANGPTL3/8 and ANGPTL4/8 were not associated with HF mortality. In contrast, ANGPTL3 and especially CD-ANGPTL4 were positively associated with HF death.
Conclusions
We found positive associations of ANGPTL3 and CD-ANGPTL4 with HF and HF mortality. Of particular interest, serum levels of CD-ANGPTL4 demonstrated positive associations with HFrEF, HFpEF, NT-proBNP, and NYHA functional class and an inverse association with LVEF. CD-ANGPTL4 also demonstrated the strongest positive association with HF mortality. The explanation for these associations is not currently apparent. Further investigation will be required to understand more fully the possible biochemical mechanisms that may be responsible for these associations.
Keywords: ANGPTL3, ANGPTL4/8, CD-ANGPTL4, Heart failure, Mortality
Introduction
Mendelian randomization studies have shown that lipid traits are causally related to heart failure (HF) [1]. For LDL-cholesterol (LDL-C), this association is mediated by its role as a risk factor for atherosclerosis leading to ischemic cardiomyopathy [1]. For triglycerides (TG), the mechanism is less clear and may involve cardiac energy metabolism required for normal heart function [2]. Decreased lipoprotein lipase (LPL) activity has also been implicated in HF [3], and the angiopoietin-like 3, 4, and 8 proteins (ANGPTL3/4/8), which regulate LPL activity, have been associated with circulating TG levels and cardiovascular risk [4–7].
In the postprandial state, ANGPTL3 and ANGPTL8 form an ANGPTL3/8 complex that inhibits LPL in oxidative tissues, so TG are routed for storage in adipose tissue, and the TG-lowering apolipoprotein A5 (APOA5) suppresses ANGPTL3/8-mediated LPL inhibition [8–20]. In contrast, ANGPTL4 inhibits LPL in the fat during fasting so TG can be available as energy substrates for oxidative tissues [21–26]. After feeding, ANGPTL8 forms a complex with ANGPTL4 (ANGPTL4/8) that decreases the LPL-inhibitory effect of ANGPTL4 so TG can be hydrolyzed and stored in adipose tissue [27, 28]. ANGPTL4 overexpression in the heart has been associated with inhibition of lipoprotein utilization and left-ventricular dysfunction [29].
We have developed dedicated immunoassays to measure ANGPTL3/8, ANGPTL3, ANGPTL4/8, and C-terminal domain-containing ANGPTL4 (CD-ANGPTL4), the major circulating isoform of ANGPTL4 [30, 31]. We have previously used these assays to measure these ANGPTL proteins and complexes in the LUdwigshafen RIsk and Cardiovascular Health (LURIC) study of patients referred for coronary angiography. ANGPTL3/8 was positively associated with TG and LDL-C but not cardiovascular mortality [30]. In contrast, ANGPTL4/8 and especially CD-ANGPTL4 were positively associated with diabetes, inflammation, and cardiovascular mortality [30]. The aim of the present study was to evaluate the associations of each of these ANGPTL3/4/8 proteins and complexes with HF and HF mortality in LURIC study participants during a median (inter-quartile range) follow-up of 9.8 (8.7–10.4) years.
Methods
Availability of data and materials
All data supporting the conclusions of this article are contained within the article and will be made available to other researchers upon reasonable request to the corresponding author.
LURIC study subjects
The LURIC study included 3,316 patients referred for coronary angiography to the Ludwigshafen Heart Center in South-West Germany [30]. The LURIC study complied with the Declaration of Helsinki and was approved by the Ethics Committee of the ‘Landesärztekammer Rheinland-Pfalz’ (Mainz, Germany). Participants were recruited between July 1997 and January 2000, and written informed consent was obtained from all participants, Inclusion criteria were as follows: German ancestry, clinical stability except for acute coronary syndromes, and the availability of a coronary angiogram. Indications for angiography in individuals with clinically stable disease were chest pain and/or non-invasive test results suggestive of myocardial ischemia.
Individuals suffering from any acute illness other than acute coronary syndromes, chronic non-cardiac diseases, or malignancy within the past five years, and those unable to understand the purpose of the study were excluded. Information on the vital status was obtained from local population registries. Cardiovascular mortality was defined as death due to fatal myocardial infarction (FMI), sudden cardiac death, death after cardiovascular intervention, stroke, or other causes of death caused by cardiovascular diseases.
Heart failure data and laboratory analyses
Heart failure with reduced ejection fraction (HFrEF) and HF with preserved ejection fraction (HFpEF) were categorized as previously described [32]. Information on the vital status was obtained from local population registries, and causes of death were categorized based on death certificates. The severity of HF was assessed using the New York Heart Association (NYHA) classification and evaluation of left ventricular ejection fraction (LVEF). The median (inter-quartile range) duration of follow-up was 9.8 (8.7–10.4) years. The N-terminal prohormone of B-type natriuretic peptide (NT-proBNP) was measured using an electro-chemiluminescent assay (Roche Diagnostics, Mannheim, Germany).
Patients with the following characteristics were evaluated for the current study: symptoms and signs of HF, LVEF less than 45% (echocardiographic or invasive), and the presence of diastolic HF according to the definition previously published by Paulus et al. [33]. Diastolic dysfunction was diagnosed in 388 patients based on hemodynamic criteria (mean pulmonary capillary wedge pressure greater than 12 mmHg or left ventricular end-diastolic pressure greater than 16 mmHg). In an additional 71 patients, diastolic dysfunction was identified by an elevated NT-proBNP concentration (greater than 220 pg/mL) and electrocardiographic evidence of atrial fibrillation. Echocardiography criteria for HFpEF and HFrEF have changed significantly since the LURIC population was enrolled more than 25 years ago. While the 45% LVEF threshold for HFrEF was consistent with a previous LURIC study that examined the association of C-reactive protein (CRP) with HF [32], the latest criteria for HFrEF now include LVEF < 40% [34]. Likewise, the definition of HFpEF by Paulus and colleagues [33] that was used when assessing LURIC study population is not completely consistent with the most modern guidelines [34].
Measurements of ANGPTL3/4/8 proteins and complexes
Baseline ANGPTL3/4/8 protein levels were measured in 2394 participants (with adequate serum remaining) using immunoassays as previously described [30], with minor modifications. Assays were performed using streptavidin-coated MesoScale Discovery (MSD) plates on which biotinylated capture antibodies were allowed to bind for 1 h. After washing, recombinant standard curves and serum samples diluted 1:10 in assay diluent were added to the plates for a 1-hour incubation. Following washing, ruthenium-labelled detection antibodies were added for a 1-hour incubation, after which plates were washed again, and MSD read buffer was added to enable electrochemiluminescent (ECL) signal detection on an MSD plate reader. For ANGPTL3/8, a monoclonal anti-ANGPTL3/8 antibody was used for capture, and a second monoclonal anti-ANGPTL3/8 antibody recognizing a separate ANGPTL3/8 epitope was used for detection. For ANGPTL3, a monoclonal anti-ANGPTL3 antibody was used for capture, and a polyclonal anti-ANGPTL3 antibody was used for detection. For ANGPTL4/8, a monoclonal anti-ANGPTL4 antibody was used for capture, and a monoclonal anti-ANGPTL8 antibody was used for detection. For CD-ANGPTL4, a polyclonal anti-CD-ANGPTL4 antibody was used for capture, and a polyclonal anti-CD-ANGPTL4 antibody was used for detection.
Statistical analysis
For all ANGPTL3/4/8 protein immunoassays, MSD software (4.0.13) was used for curve fitting (5-parameter fit with 1/y2 weighting). Correlations of ANGPTL proteins with each other and NT-proBNP were assessed using a two-tailed correlation t-test The χ²-test and analysis of variance were used to compare distributions of categorical parameters and NT-proBNP concentrations across tertiles of each of the ANGPTL3/4/8 proteins and complexes, and NT-proBNP levels were transformed logarithmically before performing the analysis. The Cox proportional hazards model was used to examine associations of each of the ANGPTL3/4/8 proteins and complexes with HF mortality. Two models of adjustment were used. Model 1 was derived from a Dagitty causal diagram that included adjustment for sex, age, body mass index (BMI), and diabetes, while model 2 included additional adjustment for hypertension, smoking, Friesinger scores, cerebrovascular disease, and peripheral arterial disease [30]. In addition, we also performed three additional analyses for CD-ANGPTL4. These included model 2 plus further adjustment for CRP, model 2 plus further adjustment for log-transformed NT-proBNP, and model 2 plus further adjustment for both CRP and log-transformed NT-proBNP.
Results
Baseline characteristics of LURIC subjects
Of 3,316 subjects enrolled in the LURIC study, 2,394 had serum samples available for ANGPTL3/4/8 protein analyses while 922 did not. To investigate the possibility of a sample selection bias, we compared the baseline characteristics of both groups. The two groups (2,394 subjects analysed versus 922 subjects not analysed) were similar with respect to percentage of male participants (68.3% versus 73.1%), age (62.8 ± 10.4 years versus 62.3 ± 11.2 years), BMI (27.6 ± 4.1 kg/m2 versus 27.3 ± 3.9 kg/m2), percentage of subjects with HFrEF (17.0% versus 18.7%), percentage of subjects with HFpEF (15.8% versus 13.9%), and percentage of HF deaths (4.2% versus 5.1%). These comparisons indicated that the baseline characteristics of the two groups were similar, thus suggesting that a sample selection bias was unlikely.
Correlations of ANGPTL3/4/8 proteins and complexes with each other and HF data
Table 1 shows the correlations of each of the ANGPTL3/4/8 proteins and complexes with each other and NT-proBNP. The strongest correlation amongst the ANGPTL proteins themselves was that of CD-ANGPTL4 with ANGPTL4/8 (r = 0.449, p < 0.001), followed by those of CD-ANGPTL4 with ANGPTL3 (r = 0.276, p < 0.001) and ANGPTL3 with ANGPTL4/8 (r = 0.274, p < 0.001). Interestingly, the strongest overall correlation observed was that of CD-ANGPTL4 with NT-proBNP (r = 0.542, p < 0.001).
Table 1.
Correlations of ANGPTL3/8, ANGPTL3, ANGPTL4/8, and CD-ANGPTL4 with each other and NT-proBNP
| LURIC study | ANGPTL3/8 (ng/mL) |
ANGPTL3 (ng/mL) |
ANGPTL4/8 (ng/mL) |
CD-ANGPTL4 (ng/mL) |
NT-proBNP (pg/mL) |
|
|---|---|---|---|---|---|---|
|
ANGPTL3/8 (ng/mL) |
r-value | 1 | 0.18 | 0.199 | −0.035 | −0.104 |
| p -value | – | < 0.001 | < 0.001 | 0.087 | < 0.001 | |
| N | 2,394 | 2,394 | 2,394 | 2,394 | 2,367 | |
|
ANGPTL3 (ng/mL) |
r-value | 0.18 | 1 | 0.274 | 0.276 | 0.208 |
| p -value | < 0.001 | – | < 0.001 | < 0.001 | < 0.001 | |
| N | 2,394 | 2,394 | 2,394 | 2,394 | 2,367 | |
|
ANGPTL4/8 (ng/mL) |
r-value | 0.199 | 0.274 | 1 | 0.449 | 0.109 |
| p -value | < 0.001 | < 0.001 | – | < 0.001 | < 0.001 | |
| N | 2,394 | 2,394 | 2,394 | 2,394 | 2,367 | |
|
CD-ANGPTL4 (ng/mL) |
r-value | −0.035 | 0.276 | 0.449 | 1 | 0.542 |
| p -value | 0.087 | < 0.001 | < 0.001 | – | < 0.001 | |
| N | 2,394 | 2,394 | 2,394 | 2,394 | 2,367 | |
|
NT-proBNP (pg/mL) |
r-value | −0.104 | 0.208 | 0.109 | 0.542 | 1 |
| p -value | < 0.001 | < 0.001 | < 0.001 | < 0.001 | – | |
| N | 2,367 | 2,367 | 2,367 | 2,367 | 2,367 | |
Legend: Correlations of ANGPTL3/4/8 proteins and complexes with each other and NT-proBNP were assessed using a two-tailed correlation t-test. NT-proBNP, The r-value, p-value, and N for each correlation are indicated in bold. N-terminal prohormone of B-type natriuretic peptide
Table 2 shows the associations of ANGPTL3/4/8 protein tertiles with HF data. There was an inverse association of the ANGPTL3/8 complex with NT-proBNP and a trend toward an inverse association with HFrEF. ANGPTL3 was positively associated with HFpEF, the NYHA functional class, and NT-proBNP. Modest, positive associations were observed for ANGPTL4/8 with HFrEF, HFpEF, and NT-proBNP, while there was an inverse association observed with LVEF. CD-ANGPTL4 was positively associated with HFrEF, HFpEF, the NYHA functional class, and NT-proBNP, while being inversely associated with LVEF.
Table 2.
Associations of ANGPTL3/4/8 proteins and complexes with HF parameters
| ANGPTL3/8 tertile | 1 | 2 | 3 | p-value |
|---|---|---|---|---|
| Number | 798 | 798 | 798 | |
| NYHA | 0.062 | |||
| I | 445 (55.8) | 407 (51.0) | 401 (50.3) | |
| II | 228 (28.6) | 233 (29.2) | 240 (30.1) | |
| III | 107 (13.4) | 131 (16.4) | 142 (17.8) | |
| IV | 18 (2.3) | 27 (3.4) | 15 (1.9) | |
| LV-Echo | 0.001 | |||
| Normal | 445 (63.5) | 464 (66.0) | 538 (73.5) | |
| Modestly impaired | 120 (17.1) | 124 (17.6) | 91 (12.4) | |
| Moderately impaired | 104 (14.8) | 77 (11.0) | 78 (10.6) | |
| Severely impaired | 32 (4.6) | 38 (5.4) | 28 (3.8) | |
| HFrEF | 148 (18.5) | 143 (17.9) | 116 (14.5) | 0.072 |
| HFpEF | 105 (13.2) | 118 (14.8) | 155 (19.4) | 0.002 |
| NT-proBNP, pg/mL | 423 (128–1252) | 302 (108–900) | 230 (85–595) | < 0.001 |
| ANGPTL3 tertile | 1 | 2 | 3 | p-value |
|---|---|---|---|---|
| Number | 798 | 798 | 798 | |
| NYHA | 0.003 | |||
| I | 419 (52.5) | 452 (56.6) | 382 (47.9) | |
| II | 232 (29.1) | 217 (27.2) | 252 (31.6) | |
| III | 133 (16.7) | 114 (14.3) | 133 (16.7) | |
| IV | 14 (1.8) | 15 (1.9) | 31 (3.9) | |
| LV-Echo | 0.281 | |||
| Normal | 488 (68.1) | 487 (70.2) | 472 (64.8) | |
| Modestly impaired | 116 (16.2) | 93 (13.4) | 126 (17.3) | |
| Moderately impaired | 83 (11.6) | 86 (12.4) | 90 (12.4) | |
| Severely impaired | 30 (4.2) | 28 (4.0) | 40 (5.5) | |
| HFrEF | 128 (16.0) | 133 (16.7) | 146 (18.3) | 0.465 |
| HFpEF | 114 (14.3) | 115 (14.4) | 149 (18.7) | 0.024 |
| NT-proBNP, pg/mL | 235(98–691) | 296 (102–737) | 390 (128–1175) | < 0.001 |
| ANGPTL4/8 tertile | 1 | 2 | 3 | p-value |
|---|---|---|---|---|
| Number | 798 | 798 | 798 | |
| NYHA | 0.231 | |||
| I | 435 (54.5) | 420 (52.6) | 398 (49.9) | |
| II | 233 (29.2) | 231 (28.9) | 237 (29.7) | |
| III | 118 (14.8) | 125 (15.6) | 137 (17.2) | |
| IV | 12 (1.5) | 23 (2.9) | 25 (3.1) | |
| LV-Echo | < 0.001 | |||
| Normal | 510 (71.4) | 475 (67.1) | 462 (64.4) | |
| Modestly impaired | 115 (16.1) | 119 (16.8) | 101 (14.1) | |
| Moderately impaired | 62 (8.7) | 87 (12.3) | 110 (15.3) | |
| Severely impaired | 27 (3.8) | 27 (3.8) | 44 (6.1) | |
| HFrEF | 103 (12.9) | 126 (15.8) | 178 (22.3) | < 0.001 |
| HFpEF | 112 (14.0) | 110 (13.5) | 156 (19.6) | 0.002 |
| NT-proBNP, pg/mL | 233(95–607) | 301 (108–883) | 370 (130–1174) | < 0.001 |
| CD-ANGPTL4 tertile | 1 | 2 | 3 | p-value |
|---|---|---|---|---|
| Number | 798 | 799 | 797 | |
| NYHA | < 0.001 | |||
| I | 474 (59.4) | 444 (55.6) | 335 (42.0) | |
| II | 221 (27.7) | 235 (29.4) | 245 (30.7) | |
| III | 93 (11.7) | 102 (12.8) | 185 (23.2) | |
| IV | 10 (1.3) | 17 (2.1) | 33 (4.1) | |
| LV-Echo | < 0.001 | |||
| Normal | 517 (73.4) | 506 (70.6) | 424 (59.1) | |
| Modestly impaired | 107 (15.2) | 117 (16.3) | 111 (15.5) | |
| Moderately impaired | 59 (8.4) | 74 (10.3) | 126 (17.5) | |
| Severely impaired | 21 (3.0) | 20 (2.8) | 57 (7.9) | |
| HFrEF | 91 (11.4) | 112 (14.0) | 204 (25.6) | < 0.001 |
| HFpEF | 84 (10.5) | 137 (17.2) | 157 (19.7) | < 0.001 |
| NT-proBNP, pg/mL | 157 (72–433) | 270 (103–709) | 634 (212–1716) | < 0.001 |
Legend: The table shows the correlations of each of the ANGPTL3/4/8 proteins and complexes with the HF parameters that were measured. HFpEF HF with preserved ejection fraction, HFrEF HF with reduced ejection fraction, LV left ventricular, NT-proBNP N-terminal prohormone of B-type natriuretic peptide, NYHA New York Heart Association
Associations of ANGPTL3/4/8 proteins and complexes with HF mortality
A total of 101 participants in the LURIC study died from HF. Table 3 shows the HF mortality according to the tertiles of baseline circulating concentrations of ANGPTL3/8, ANGPTL3, ANGPTL4/8, and CD-ANGPTL4. Of all ANGPTL3/4/8 proteins measured, CD-ANGPTL4 demonstrated the highest third-to-first tertile hazard ratio (HR) for HF mortality. Because we observed a strong correlation between CD-ANGPTL4 and NT-proBNP, which is considered the gold-standard biomarker for HF severity and a powerful predictor of HF mortality, we considered the possibility that CD-ANGPTL4 might merely act as a surrogate marker by correlating with NT-proBNP.
Table 3.
Heart failure death (HFD) according to tertiles of ANGPTL3/8, ANGPTL3, ANGPTL4/8, and CD-ANGPTL4
| LURIC study | Model 1 | Model 2 | ||||
|---|---|---|---|---|---|---|
| ANGPTL3/8 | N | HFD (N/%) | HR (95% CI) | p | HR (95% CI) | p |
| 1st tertile | 798 | 35 (4.4) | 1.00 reference | – | 1.00 reference | – |
| 2nd tertile | 798 | 34 (4.3) | 0.88 (0.54–1.42) | 0.587 | 0.94 (0.58–1.52) | 0.785 |
| 3rd tertile | 798 | 32 (4.0) | 0.84 (0.51–1.40) | 0.511 | 0.85 (0.51–1.41) | 0.533 |
| ANGPTL3 | N | HFD (N/%) | HR (95% CI) | p | HR (95% CI) | p |
|---|---|---|---|---|---|---|
| 1st tertile | 798 | 18 (2.3) | 1.00 reference | – | 1.00 reference | - |
| 2nd tertile | 798 | 41 (5.1) | 2.18 (1.25–3.81) | 0.006 | 2.11 (1.21–3.69) | 0.008 |
| 3rd tertile | 798 | 42 (5.3) | 1.96 (1.12–3.45) | 0.019 | 2.04 (1.15–3.59) | 0.014 |
| ANGPTL4/8 | N | HFD (N/%) | HR (95% CI) | p | HR (95% CI) | p |
|---|---|---|---|---|---|---|
| 1st tertile | 798 | 28 (3.5) | 1.00 reference | – | 1.00 reference | – |
| 2nd tertile | 799 | 36 (4.5) | 1.21 (0.73–1.99) | 0.457 | 1.29 (0.79–2.12) | 0.312 |
| 3rd tertile | 797 | 37 (4.6) | 1.07 (0.64–1.78) | 0.792 | 1.09 (0.66–1.82) | 0.731 |
| CD-ANGPTL4 | N | HFD (N/%) | HR (95% CI) | p | HR (95% CI) | p |
|---|---|---|---|---|---|---|
| 1st tertile | 798 | 12 (1.5) | 1.00 reference | – | 1.00 reference | – |
| 2nd tertile | 798 | 23 (2.9) | 1.44 (0.71–2.91) | 0.311 | 1.53 (0.75–3.10) | 0.242 |
| 3rd tertile | 798 | 66 (8.3) | 3.81 (2.01–7.20) | < 0.001 | 3.91 (2.06–7.41) | < 0.001 |
| Additional adjustments | Model 2 + CRP | Model 2 + NT-proBNP |
Model 2 + CRP + NT-proBNP |
|||
|---|---|---|---|---|---|---|
| CD-ANGPTL4 | HR (95% CI) | p | HR (95% CI) | p | HR (95% CI) | p |
| 1st tertile | 1.00 reference | – | 1.00 reference | – | 1.00 reference | – |
| 2nd tertile | 1.52 (0.75–3.08) | 0.250 | 1.43 (0,69-2.95) | 0.341 | 1.42 (0.69–2.96) | 0.343 |
| 3rd tertile | 3.80 (1.99–7.25) | < 0.001 | 2.96 (1.50–5.80) | 0.002 | 2.96 (1.51–5.83) | 0.002 |
Legend: Hazard ratios (HR) were calculated using Cox regression. Model 1 included adjustment for sex, age, BMI, and diabetes. Model 2 included model 1 plus additional adjustments for Friesinger scores, smoking, arterial hypertension, cerebrovascular disease, and peripheral arterial disease. Further adjustments were performed for CD-ANGPTL4 for model 2 to also include CRP, NT-proBNP, and CRP plus NT-proBNP and are shown at the bottom of the table. BMI body mass index, CI confidence interval, CRP C-reactive protein, HFD heart failure death, HR hazard ratio, N number, NT-proBNP N-terminal prohormone of B-type natriuretic peptide
We therefore performed an additional analysis for CD-ANGPTL4 and HF death using model 2 plus further adjustment for log-transformed NT-proBNP. Additional adjustment of model 2 for log-transformed NT-proBNP reduced the third-to-first tertile HR for CD-ANGPTL4 from 3.91 (p < 0.001) to 2.96 (p = 0.002), however, the association remained significant. We also performed similar analyses for model 2 plus CRP. Adding CRP to model 2 had a minimal effect, with the third-to-first tertile HR for CD-ANGPTL4 decreasing slightly from 3.91 (p < 0.001) for model 2 to 3.80 (p < 0.001) for model 2 plus CRP. Adding both CRP and log-transformed NT-proBNP to model 2 had an almost identical effect to that observed when only log-transformed NT-proBNP was added. These data, which are shown at the bottom of Table 3, suggested that CD-ANGPTL4 may provide additional prognostic information beyond already established clinical biomarkers.
Kaplan-Meier plots for each of the ANGPTL3/4/8 proteins and complexes are shown in Fig. 1. No association with HF mortality was observed for the ANGPTL3/8 or ANGPTL4/8 complexes. In contrast, ANGPTL3 and CD-ANGPTL4 were associated with an increased risk of death from HF. Of all the ANGPTL3/4/8 proteins and complexes examined, CD-ANGPTL4 demonstrated the strongest association with HF mortality.
Fig. 1.
Kaplan-Meier plots for survival free from heart failure death (HFD) according to tertiles of ANGPTL3/8 (A), ANGPTL3 (B), ANGPTL4/8 (C), and CD-ANGPTL4 (D). The y-axis represents survival free from HFD, and the x-axis represents follow-up in years. Circulating CD-ANGPTL4 concentrations demonstrated the strongest association with HFD
Discussion
The association observed for CD-ANGPTL4 with HF and HF death is consistent with our previous observation for the association of CD-ANGPTL4 with overall cardiovascular mortality and fatal myocardial infarction [30]. Interestingly, the positive association of CD-ANGPTL4 with HF mortality remained significant even after further adjustment of model 2 for CRP and NT-proBNP, suggesting that CD-ANGPTL4 may provide additional prognostic information beyond that available from already established clinical biomarkers.
The reason for the positive association of CD-ANGPTL4 with HF, which also aligns with previous animal model studies [29], remains incompletely understood. It may have to do with the regulation of LPL activity, but this seems unlikely since CD-ANGPTL4 lacks the LPL-inhibitory specific epitope 1 (SE1) domain that is present in full-length ANGPTL4 and circulating CD-ANGPTL4 concentrations show little correlation with serum lipids [30, 31].
Alternative mechanisms, perhaps related to diabetes and inflammation might be possible considering the association of CD-ANGPTL4 with diabetes mellitus and C-reactive protein (CRP) levels and progression of coronary artery calcification [30, 31, 35]. However, the positive association of CD-ANGPTL4 with HF mortality included adjustment for diabetes and decreased only slightly after further adjustment for CRP. It also remains unclear whether the mechanisms accounting for the association of CD-ANGPTL4 with HFREF, HFPEF, and HF mortality are the same as those related to overall cardiovascular mortality or represent different mechanisms. Finally, the tissues responsible for CD-ANGPTL4 generation are also uncertain. CD-ANGPTL4 could arise from plasmin cleavage of ANGPTL4/8 in adipose tissue or furin cleavage of ANGPTL4 in the liver, intestine, kidney, or other tissues [27, 28, 36].
It may appear counterintuitive that circulating concentrations of the ANGPTL3/8 complex, the most potent circulating inhibitor of LPL enzymatic activity and the inhibition of which has been shown to reduce serum TG and LDL-C levels [37, 38], were not significantly associated with death from HF. This might be due to confounding factors, as low LDL-C and TG levels may on one hand predict lower cardiovascular risk but on the other hand could reflect the frailty that often accompanies late-stage HF. In this respect, previous studies have indicated that CRP was an independent predictor of mortality in HFpEF and have suggested that inflammation (as assessed by CRP measurements) may in fact be a better predictor of cardiovascular mortality than hyperlipidemia [39, 40]. Considering the positive associations of ANGPTL3 and especially of CD-ANGPTL4 with HF and HF mortality, it will be important to understand the biochemical mechanisms accounting for these observations.
Our study has several limitations. It was observational in nature and cannot prove causality. There were a relatively small number of HF deaths (n = 101) which limits the statistical power of multivariable adjustments. The single-center nature of the LURIC study that enrolled more than 25 years ago limits its generalizability to other patient populations, and thus a cohort selection bias cannot be ruled out. In addition, echocardiography criteria for HFpEF and HFrEF have changed since the LURIC study began. The LURIC study definitions of the LVEF threshold used for HFrEF and the criteria for HFpEF do not precisely align with the most current clinical guidelines [34]; this could affect interpretation of data regarding correlations of the ANGPTL3/4/8 protein tertiles with HFrEF and HFpEF. Also, ANGPTL3/4/8 proteins and complexes were measured only at baseline. Therefore, our study does not account for changes that might have occurred during the follow-up period. Finally, although our Kaplan-Meier plots were consistent with the adjusted Cox models, they represent unadjusted descriptions of the HF mortality data.
Conclusion
Of all the ANGPTL3/4/8 proteins and complexes measured, CD-ANGPTL4 showed the strongest association with HF and HF mortality. The exact tissue source of CD-ANGPTL4 and reasons for its strong positive association with HF and HF death are unclear. Further research will be needed to elucidate the biochemical mechanisms responsible for these observations.
Acknowledgements
The authors thank the LURIC subjects for participation in this study.
Authors’ contributions
G.S. and R.J.K. conceived and designed the study with contributions from A-C.M., M.E.K., G.D., A.P.M., W.M. and H.S., and A-C.M., H.L., D.L., Y.Q.C., E.Z., and Y.W. were responsible for data acquisition. G.S. analysed and interpreted the data, and G.S. and R.J.K. wrote the manuscript. Funding was obtained by G.S., W.M., and R.J.K. All authors read, edited, and approved the final version of the manuscript.
Funding
This study was supported by the Austrian FWF (KLI 1117 Programm Klinische Forschung [KLIF]). LURIC received support by the 7th Framework Program of the European Union (integrated projects AtheroRemo, grant number 201668 and RiskyCAD, project number 305739), by the German Federal Ministry of Education and Research (project e: AtheroSysMed [Systems medicine of coronary heart disease and stroke], grant number 01ZX1313A-K) and as part of the Competence Cluster of Nutrition and Cardiovascular Health (nutriCARD), funded by the German Federal Ministry of Education and Research (grant number 01EA1411A).
Data availability
All data supporting the conclusions of this article are contained within the article and will be made available to other researchers upon reasonable request to the corresponding author.
Declarations
Ethics approval and consent to participate
The LURIC study complied with the Declaration of Helsinki and was approved by the Ethics Committee of the ‘Landesärztekammer Rheinland-Pfalz’ (Mainz, Germany). Written informed consent was obtained from all participants.
Competing interests
H.L., D.L., Y.Q.C., E.Z., Y.W., and R.J.K. are Eli Lilly and Company employees and own Lilly stock. G.S. has received research funding from Numares AG, Sanofi, Bayer, and Eli Lilly, consulting fees from Sanofi, honoraria from Sanofi and Daiichi-Sankyo, meeting travel expenses from Bayer, Amgen, Daiichi-Sankyo, and research materials from Eli Lilly. H.S. has received research grants and support from Abbott Diagnostics and Amgen and honoraria from Amgen and Sanofi. M.E.K. has received employment from SYNLAB Holding, Deutschland. W.M. has received grants from Siemens, AstraZeneca, Bayer Vital, Bestbion Dx, Boehringer Ingelheim, Immundiagnostik, Merck Chemicals, and Olink, grants and personal fees from Aegerion, Amgen, Sanofi, Amryt Pharma, BASF, Abbott Diagnostics, Numares AG, Berlin-Chemie, Akzea Therapeutics, and Novartis, personal fees from Vifor Pharma, and employment from SYNLAB Holding Deutschland. The other authors declare no disclosures.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Günther Silbernagel and Ann-Cathrin Maas contributed equally to this work.
References
- 1.Xiao J, Ji J, Zhang N, et al. Association of genetically predicted lipid traits and lipid-modifying targets with heart failure. Eur J Prev Cardiol. 2023;30:358–66. [DOI] [PubMed] [Google Scholar]
- 2.Lopaschuk GD, Karwi QG, Tian R, et al. Cardiac energy metabolism in heart failure. Circ Res. 2021;128:1487–513. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Khan RS, Lin Y, Hu Y, et al. Rescue of heart lipoprotein lipase-knockout mice confirms a role for triglyceride in optimal heart metabolism and function. Am J Physiol Endocrinol Metab. 2013;305:E1339–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Romeo S, Pennacchio LA, Fu Y, et al. Population-based resequencing of ANGPTL4 uncovers variations that reduce triglycerides and increase HDL. Nat Genet. 2007;39:513–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Romeo S, Yin W, Kozlitina J, et al. Rare loss-of-function mutations in ANGPTL family members contribute to plasma triglyceride levels in humans. J Clin Invest. 2009;119:70–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Dewey FE, Gusarova V, O’Dushlaine C, et al. Inactivating Variants in ANGPTL4 and Risk of Coronary Artery Disease. N Engl J Med. 2016;374:1123–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Dewey FE, Gusarova V, Dunbar RL, et al. Genetic and Pharmacologic Inactivation of ANGPTL3 and Cardiovascular Disease. N Engl J Med. 2017;377:211–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Quagliarini F, Wang Y, Kozlitina J, et al. Atypical angiopoietin-like protein that regulates ANGPTL3. Proc Natl Acad Sci U S A. 2012;109:19751–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Wang Y, Quagliarini F, Gusarova V, et al. Mice lacking ANGPTL8 (Betatrophin) manifest disrupted triglyceride metabolism without impaired glucose homeostasis. Proc Natl Acad Sci U S A. 2013;110:16109–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Gusarova V, Banfi S, Alexa-Braun CA et al. (2017) ANGPTL8 blockade with a monoclonal antibody promotes triglyceride clearance, energy expenditure, and weight loss in mice. Endocrinology. 2017;158:1252–1259. [DOI] [PMC free article] [PubMed]
- 11.Banfi S, Gusarova V, Gromada J, et al. Increased thermogenesis by a noncanonical pathway in ANGPTL3/8-deficient mice. Proc Natl Acad Sci U S A. 2018;115:E1249–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Chen YQ, Pottanat TG, Siegel RW, et al. Angiopoietin-like protein 8 differentially regulates ANGPTL3 and ANGPTL4 during postprandial partitioning of fatty acids. J Lipid Res. 2020;61:1203–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Oldoni F, Cheng H, Banfi S, et al. ANGPTL8 has both endocrine and autocrine effects on substrate utilization. JCI Insight. 2020;5:e138777. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Zhang R, Zhang K. An updated ANGPTL3-4-8 model as a mechanism of triglyceride partitioning between fat and oxidative tissues. Prog Lipid Res. 2022;85:101140. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wen Y, Chen YQ, Konrad RJ. The regulation of triacylglycerol metabolism and lipoprotein lipase activity. Adv Biol. 2022;6:e2200093. [DOI] [PubMed] [Google Scholar]
- 16.Zhang R, Zhang K. A unified model for regulating lipoprotein lipase activity. Trends Endocrinol Metab. 2024;35:490–504. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Chen YQ, Pottanat TG, Zhen EY, et al. ApoA5 lowers triglyceride levels via suppression of ANGPTL3/8-mediated LPL inhibition. J Lipid Res. 2021;62:100068. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Yang Y, Beigneux AP, Song W, et al. Hypertriglyceridemia in Apoa5-/- mice results from reduced amounts of lipoprotein lipase in the capillary lumen. J Clin Invest. 2023;133:e172600. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Chen YQ, Yang Y, Zhen EY, et al. Carboxyl-terminal sequences in APOA5 are important for suppressing ANGPTL3/8 activity. Proc Natl Acad Sci U S A. 2024;121:e2322332121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Yang Y, Beigneux AP, Lowe TL, et al. A neutralizing APOA5 monoclonal antibody reduces amounts of lipoprotein lipase in capillaries and triggers hypertriglyceridemia. Proc Natl Acad Sci U S A. 2026;123:e2528664123. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Kersten S, Mandard S, Tan NS, et al. Characterization of the fasting-induced adipose factor FIAF, a novel peroxisome proliferator-activated receptor target gene. J Biol Chem. 2000;275:28488–93. [DOI] [PubMed] [Google Scholar]
- 22.Dijk W, Kersten S. Regulation of lipoprotein lipase by Angptl4. Trends Endocrinol Metab. 2014;25:146–55. [DOI] [PubMed] [Google Scholar]
- 23.Dijk W, Beigneux AP, Larsson M, et al. Angiopoietin-like 4 promotes intracellular degradation of lipoprotein lipase in adipocytes. J Lipid Res. 2016;57:1670–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Dijk W, Schutte S, Aarts EO, et al. Regulation of angiopoietin-like 4 and lipoprotein lipase in human adipose tissue. J Clin Lipidol. 2018;12:773–83. [DOI] [PubMed] [Google Scholar]
- 25.Ruppert PMM, Michielsen CCJR, Hazebroek EJ, et al. Fasting induces ANGPTL4 and reduces LPL activity in human adipose tissue. Mol Metab. 2020;40:101033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Kersten S. Role and mechanism of the action of angiopoietin-like protein ANGPTL4 in plasma lipid metabolism. J Lipid Res. 2021;62:100150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Zhen EY, Chen YQ, Russell AM, et al. Angiopoietin-like protein 4/8 complex-mediated plasmin generation leads to cleavage of the complex and restoration of LPL activity. Proc Natl Acad Sci U S A. 2023;120:e2214081120. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Chen YQ, Zhen EY, Russell AM, et al. Decoding the role of angiopoietin-like protein 4/8 complex-mediated plasmin generation in the regulation of LPL activity. J Lipid Res. 2023;64:100441. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Yu X, Burgess SC, Ge H, et al. Inhibition of cardiac lipoprotein utilization by transgenic overexpression of Angptl4 in the heart. Proc Natl Acad Sci U S A. 2005;102:1767–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Silbernagel G, Chen YQ, Li H, et al. Associations of circulating ANGPTL3, C-terminal domain-containing ANGPTL4, and ANGPTL3/8 and ANGPTL4/8 complexes with LPL activity, diabetes, inflammation, and cardiovascular mortality. Circulation. 2024;151:218–34. [DOI] [PubMed] [Google Scholar]
- 31.Silbernagel G, Higbie H, Meininger T, et al. Associations of ANGPTL proteins and complexes with progression of coronary artery calcification and coronary events. Atherosclerosis. 2025;409:120485. [DOI] [PubMed] [Google Scholar]
- 32.Koller L, Kleber M, Goliasch G, et al. C-reactive protein predicts mortality in patients referred for coronary angiography and symptoms of heart failure with preserved ejection fraction. Eur J Heart Fail. 2014;16:758–66. [DOI] [PubMed] [Google Scholar]
- 33.Paulus WJ, Tschöpe C, Sanderson JE, et al. How to diagnose diastolic heart failure: a consensus statement on the diagnosis of heart failure with normal left ventricular ejection fraction by the Heart Failure and Echocardiography Associations of the European Society of Cardiology. Eur Heart J. 2007;28:2539–50. [DOI] [PubMed] [Google Scholar]
- 34.Ostrominski JW, DeFilippis EM, Bansal K, et al. Contemporary American and European Guidelines for Heart Failure Management: JACC: Heart Failure Guideline Comparison. JACC Heart Fail. 2024;12:810–25. [DOI] [PubMed] [Google Scholar]
- 35.Tjeerdema N, Georgiadi A, Jonker JT, et al. Inflammation increases plasma angiopoietin-like protein 4 in patients with the metabolic syndrome and type 2 diabetes. BMJ Open Diabetes Res Care. 2014;2:e000034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Yin W, Romeo S, Chang S, et al. Genetic variation in ANGPTL4 provides insights into protein processing and function. J Biol Chem. 2009;284:13213–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Balasubramaniam D, Schroeder O, Russell AM, et al. An anti-ANGPTL3/8 antibody decreases circulating triglycerides by binding to a LPL-inhibitory leucine zipper-like motif. J Lipid Res. 2022;63:100198. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Gaudet D, Gonciarz M, Shen X, et al. Targeting the angiopoietin-like protein 3/8 complex with a monoclonal antibody in patients with mixed hyperlipidemia: a phase 1 trial. Nat Med. 2025;31:2632–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Ridker PM, Bhatt DL, Pradhan AD, et al. PROMINENT, REDUCE-IT, and STRENGTH Investigators. Inflammation and cholesterol as predictors of cardiovascular events among patients receiving statin therapy: a collaborative analysis of three randomized trials. Lancet. 2023;401:1293–301. [DOI] [PubMed] [Google Scholar]
- 40.Ridker PM, Lei L, Louie MJ, et al. Inflammation and cholesterol as predictors of cardiovascular events among 13970 contemporary high-risk patients with statin intolerance. Circulation. 2024;149:28–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
All data supporting the conclusions of this article are contained within the article and will be made available to other researchers upon reasonable request to the corresponding author.
All data supporting the conclusions of this article are contained within the article and will be made available to other researchers upon reasonable request to the corresponding author.