Abstract
BACKGROUND:
Inflammation and protein-energy malnutrition are associated with heart failure (HF) mortality. The Metabolic Vulnerability Index (MVX) is derived from markers of inflammation and malnutrition and measured by nuclear magnetic resonance (NMR) spectroscopy. MVX has not been examined in HF.
OBJECTIVES:
To examine the prognostic value of MVX in patients with HF.
METHODS:
We prospectively assembled a population-based cohort of patients with HF from 2003 to 2012 and measured MVX scores with an NMR scan from plasma collected at enrollment. Patients were divided into four MVX score groups and followed until March 31, 2021.
RESULTS:
We studied 1,382 patients (median age: 78, 48% women). The median MVX score was 64.6. Patients with higher MVX were older, more likely to be male, have atrial fibrillation, have higher NYHA class, and HF duration >18 months. Higher MVX was associated with mortality independent of MAGGIC score, ejection fraction, and other prognostic biomarkers. Compared to those with lowest MVX, the hazard ratios for MVX groups 2, 3, and 4 were 1.2 (95% CI, 0.9–1.4), 1.6 (95% CI, 1.3–2.0), and 1.8 (95% CI, 1.4–2.2), respectively (Ptrend < 0.001). Measures of model improvement document the added value of MVX in HF for classifying the risk of death beyond the MAGGIC score and other biomarkers.
CONCLUSION:
In this HF community cohort, MVX was strongly associated with mortality independently of established clinical factors and improved mortality risk classification beyond a clinically validated markers. These data underscore the potential of MVX to stratify risk in HF.
Keywords: Heart failure, metabolomics, NMR, inflammation, malnutrition, mortality
Introduction
The “cytokine hypothesis”, formulated almost two decades ago, proposes that HF progresses as a result of the overexpression of inflammatory molecules, e.g., cytokines, which reflect systemic inflammation and are often associated with protein-energy malnutrition [1]. Consistent with this hypothesis, several published studies of individual markers of inflammation have reported prognostic associations in HF [2–4]. Further, studies have suggested an association between inflammation and wasting syndromes related to malnutrition, such as cachexia and sarcopenia, and HF prognosis [5, 6]. However, these studies mostly investigated one single marker at a time and their clinical utility in routine practice has not been fully delineated [3, 4]. It stands to reason that precision phenotyping could improve our understanding of the prognostic role of inflammation and malnutrition in HF, thereby augmenting the information provided by clinical risk scores, which most often focus on short-term mortality and/or only consider clinical characteristics [7–9]. Within this context, we hypothesized a multimarker, reflecting inflammation and wasting syndromes associated with malnutrition (which we will refer to as “metabolic malnutrition”) would improve mortality risk stratification in HF [10].
Nuclear magnetic resonance (NMR) spectroscopy can generate targeted high-throughput metabolomics data suitable for epidemiological research. The Metabolic Vulnerability Index (MVX), a novel NMR multimarker developed for mortality risk stratification, is comprised of biomarkers of systemic inflammation and metabolic malnutrition [10]. To date, the prognostic value of MVX has not been evaluated in patients with HF. Therefore, we aim to report the distribution of MVX scores in a HF community cohort as well as the association of MVX with clinical characteristics, and with death (Central Illustration). We further examined the incremental clinical value of MVX beyond an established mortality risk score and other biomarkers of risk.
Central Illustration: Study design to determine the prognostic value of the Metabolic Vulnerability Index (MVX) in HF.
MVX group 1: ≤50 (N = 171); MVX group 2: (50–60] (N = 339); MVX group 3: (60–70] (N = 445); MVX group 4: >70 (N = 427). Abbreviation: MAGGIC= Meta-analysis global group in chronic heart failure. Image created with BioRender.com
Methods
Patient Population
Our HF community cohort is derived from the record linkage system of the Rochester Epidemiology Project, an optimal setting to conduct population research as it captures nearly all clinical diagnoses, procedures, results, and outcomes in its catchment area [11, 12]. Our approach to identify cases, assemble the cohort, and collect data was previously published [13, 14]. In brief, potential patients with HF were identified with natural language processing of electronic medical record text [15]. We identified patients who were ≥20 years old and resided in Olmsted, Dodge, and Fillmore Counties in Minnesota. This approach yielded 100% sensitivity compared with billing data, a reference method for case finding [15]. Research nurses reviewed and validated HF diagnosis with Framingham criteria [16]. Patients were approached in the hospital or after an outpatient encounter to provide written consent to participate in the study, including a blood draw, between September 2, 2003, and June 16, 2012. The Mayo Clinic and Olmsted Medical Center Institutional Review Boards approved of this study.
Data collection
Clinical information from inpatient and outpatient records from all providers in the Rochester Epidemiology Project [17] were collected by nurse abstractors. Clinical information included cardiovascular risk factors (e.g., smoking status, hypertension, hyperlipidemia, and diabetes), comorbid conditions included in the Charlson comorbidity index, and laboratory values [17]. N-terminal pro-brain natriuretic peptide (NT-proBNP) levels were measured using a multiplex immunoassay (Meso Scale Diagnostics). Left ventricular ejection fraction was obtained from the closest available echocardiogram value within six months prior to or two months following date of enrollment. Body mass index (BMI) was calculated using weight (kilograms) from the last outpatient prior to enrollment divided by their earliest recorded adult height (meters) squared. Electronic retrieval of international classification of disease codes from inpatient and outpatient encounters were used to ascertain chronic conditions identified as a public health priority by the U.S. Department of Health and Human Services [18, 19]. The Meta-Analysis Global Group in Chronic HF (MAGGIC) score was calculated using sex, age, ejection fraction, systolic blood pressure, BMI, creatinine, New York Heart Association (NYHA) class, smoking status, diabetes, chronic obstructive pulmonary disease, HF diagnosis >18 months ago, and use of beta blocker, angiotensin converting enzyme inhibitors, an/or angiotensin-receptor blockers [8].
Ascertainment of Death
Patients were followed through March 31, 2021 for death. The date of death was determined from obituary notices, county death certificates, and electronic files of death certificates obtained from the State of Minnesota Department of Vital and Health Statistics. We considered all-cause death and cardiovascular death. The latter was defined by using the underlying cause of death classified by ICD-10 codes 100–199. Patients who were alive at the last follow-up were censored at their date of last medical contact.
MVX Measurement
NMR LipoProfile analyses of frozen EDTA plasma collected from community patients with HF at the time of enrollment were performed on the high-throughput 400 MHz Vantera® clinical analyzer platform at the NHLBI Lipoprotein Metabolism Laboratory (Bethesda, MD) using the LP4 algorithm (Labcorp, Morrisville, NC) and sex-specific MVX scores were calculated using the MVX software algorithm [20]. Development of the MVX algorithm and the association of MVX scores with mortality in subjects at high risk of cardiovascular disease has been previously reported [10]. A brief description of the analytes that comprise the MVX scores are as follows. GlycA and small high-density lipoprotein particles (S-HDLP), measured by the NMR LipoProfile® scan, were associated with an increased risk of mortality in the CATHGEN (CATHeterization GENetics) cohort [21, 22]. GlycA arises from the glycan residues of several acute-phase glycoproteins and reflects systemic inflammation [23]. S-HDLP mediates protective functions of anti-inflammatory and immune response proteins [24, 25]. GlycA and S-HDLP were combined into an Inflammation Vulnerability Index (IVX) [10]. Further analysis in the CATHGEN and Intermountain Heart studies found four malnutrition metabolites that are associated with mortality, including citrate and the branched-chain amino acids: valine, leucine, and isoleucine, which were further combined into a score termed the Metabolic Malnutrition Index (MMX) [10]. MMX and IVX were combined as a composite score called MVX (Metabolic Vulnerability Index). MVX, IVX and MMX scores, as well as the analytes that are used to generate the scores, are stable for up to 12 years in EDTA plasma samples when frozen at <−70°C for up to 12 years. MVX scores are dimensionless ranging from 1 to 100 with a higher score indicating greater metabolic vulnerability.
Statistical Analysis
Baseline characteristics and individual MVX components were reported as frequencies (percent) for categorical variables and continuous variables were reported as median (interquartile range [IQR]). NT-proBNP values were log2 transformed for analyses. Continuous variables and categorical variables were compared across MVX groups using Kruskal-Wallis and Chi-squared tests, respectively.
Median follow-up time was calculated using the reverse Kaplan-Meier method [26]. Survival by MVX group was estimated by the Kaplan-Meier method and compared across groups by the log-rank test. Multiple imputations by chain equations was performed to account for missing clinical data used to calculate MAGGIC scores, including BMI (2.8%), NYHA class (0.4%), HF duration (0.1%), and ejection fraction (1.8%) [27].
Multivariable Cox proportional hazard regression was used to examine the association between MVX group and mortality adjusting for age and sex, MAGGIC score, NT-proBNP (logtransformed) and hemoglobin. Analysis for cardiovascular death was conducted using Fine-Gray competing risk models. Sensitivity analysis was conducted using complete cases only. Additional stratified analyses were performed by ejection fraction group and MAGGIC score subgroups based on published cut-points [8]; the two highest groups were combined, and the two lowest groups were combined due to low sample sizes. Wald tests for trend were performed by assigning midpoints of the 4 MVX groups (1 to 4) to assess the linearly increasing trend of the HR across MVX groups. To assess the linear association between MVX and mortality, we evaluated the p-value for non-linearity based on the likelihood ratio test between a model with and without restricted cubic splines [28, 29]. The number of knots was determined based on the Akaike’s information criteria.
Several measures of model improvement, including Uno C-statistic [30], Net Reclassification Improvement (NRI), Integrated Discrimination Improvement (IDI) and their corresponding 95% CIs were calculated to estimate the incremental prognostic value of MVX group beyond MAGGIC score and other biomarkers of risk in HF for mortality risk prediction at three years as the MAGGIC score is designed to estimate mortality at three years.
Analyses were performed using RStudio version 1.3.1093 with a two-sided p-value < .05 considered statistically significant.
Results
Clinical characteristics and MVX
The cohort included 1,389 patients, seven of whom did not have sufficient plasma volume, leaving 1,382 patients for analysis. Median (IQR) age was 78 (68, 84) years and 51.7% were male (Table 1). Key cardiometabolic risk factors were highly prevalent, including hypertension, hyperlipidemia, and diabetes. The median (IQR) MAGGIC score was 26 (22, 30), most patients were in NYHA class III or IV, 72% of patients were recruited in-hospital and patients presenting with preserved ejection fraction (≥ 50%) comprised 55.7% of the cohort.
Table 1. Baseline Characteristics by Metabolic Vulnerability Index (MVX) Group.
Values are N (%) or median (IQR)
| Total (N=1,382) | MVX Group 1 N = 171 | MVX Group 2 N = 339 | MVX Group 3 N = 445 | MVX Group 4 N = 427 | p-value | |
|---|---|---|---|---|---|---|
| Age (years) | 78 (68, 84) | 72 (61, 82) | 77 (67, 84) | 79 (69, 85) | 79 (70, 85) | <0.001 |
| Men | 715 (52) | 74 (43) | 162 (48) | 226 (51) | 253 (59) | <0.001 |
| Cardiovascular Risk Factors | ||||||
| Hypertension | 1261 (91) | 154 (90) | 316 (93) | 414 (93) | 377 (88) | 0.038 |
| Current smoker | 144 (10) | 21 (12) | 34 (10) | 46 (10) | 43 (10) | 0.800 |
| Diabetes mellitus | 493 (36) | 51 (30) | 132 (39) | 165 (37) | 145 (34) | 0.200 |
| Hyperlipidemia | 1171 (85) | 149 (87) | 306 (90) | 373 (84) | 343 (80) | 0.001 |
| Body mass index (kg/m 2 ) | 28 (25, 34) | 30 (27, 34) | 29 (25, 34) | 28 (24, 33) | 28 (24, 32) | <0.001 |
| Medical History | ||||||
| Myocardial Infarction | 391 (28) | 34 (20) | 96 (28) | 133 (30) | 128 (30) | 0.068 |
| Chronic obstructive pulmonary disease | 395 (29) | 41 (24) | 94 (28) | 130 (29) | 130 (30) | 0.400 |
| Atrial fibrillation | 493 (36) | 34 (20) | 120 (35) | 162 (36) | 177 (41) | <0.001 |
| HF characteristics | ||||||
| HF duration > 18 months | 495 (36) | 51 (30) | 108 (32) | 148 (33) | 188 (44) | <0.001 |
| Ejection fraction, % | 54 (35, 63) | 55 (31, 65) | 55 (36, 65) | 50 (35, 62) | 54 (35, 62) | 0.110 |
| NYHA Class | <0.001 | |||||
| I or II | 426 (31) | 66 (39) | 133 (39) | 124 (28) | 103 (24) | |
| III or IV | 950 (69) | 103 (61) | 205 (61) | 320 (72) | 322 (76) | |
| MAGGIC Score | 26 (22, 30) | 23 (18, 27) | 25 (22, 29) | 26 (22, 30) | 27 (23, 30) | <0.001 |
| Charlson Comorbidity Index | 7 (5, 9) | 5 (4, 7) | 6 (5, 8) | 7 (5, 9) | 7 (5, 9) | <0.001 |
| Laboratory values | ||||||
| eGFR (mL/min) | 53 (40, 68) | 59 (50, 70) | 57 (44, 70) | 53 (41, 68) | 48 (33, 60) | <0.001 |
| NT-proBNP (pg/mL) | 8,896 (4,205, 16,301) | 2,891 (812, 6,758) | 6,208 (3,355, 11,202) | 10,094 (5,464, 16,996) | 13,889 (7,768, 21,824) | <0.001 |
| Hemoglobin (g/dl) | 12 (11, 14) | 13 (12, 15) | 13 (12, 14) | 12 (11, 14) | 12 (10, 13) | <0.001 |
| MVX Components | ||||||
| GlycA (μmol/l) | 453 (391, 537) | 389 (349, 427) | 419 (384, 472) | 475 (419, 559) | 505 (423, 606) | <0.001 |
| S-HDLP (μmol/l) | 8.8 (5.3, 11.9) | 13.7 (11.5,15.7) | 11.5 (9.5,13.5) | 8.4 (6.1, 10.4) | 4.6 (2.4, 7.0) | <0.001 |
| Leucine (μmol/l) | 141 (115, 170) | 161 (142, 186) | 147 (128, 174) | 145 (122, 174) | 116 (95, 144) | <0.001 |
| Isoleucine (μmol/l) | 61 (48, 74) | 65 (55, 79) | 65 (53, 77) | 61 (50, 74) | 55 (42, 68) | <0.001 |
| Valine (μmol/l) | 208 (173, 246) | 244 (218, 271) | 221 (191, 257) | 212 (181, 245) | 172 (146, 206) | <0.001 |
| Citrate (μmol/l) | 121 (97, 149) | 111 (96, 129) | 121 (100, 144) | 123 (97, 151) | 127 (97, 167) | <0.001 |
MVX group 1: ≤50; MVX group 2: (50–60]; MVX group 3: (60–70]; MVX group 4: >70. Abbreviations: MVX= Metabolic Vulnerability Index; HF= heart failure; HFpEF= Heart failure with preserved ejection fraction; HFrEF= Heart failure with reduced ejection fraction; NYHA= New York Heart Association; MAGGIC= Meta-analysis global group in chronic heart failure; eGFR: Estimated glomerular filtration rate; NT-proBNP: N-terminal pro-brain natriuretic peptide; S-HDLP: Small high-density lipoprotein particles
MVX scores were normally distributed in the entire cohort with a median (IQR) of 64.6 (55.9, 71.7) (Figure 1). The relationship between MVX and mortality was linear (p-value of nonlinearity = 0.60) (Figure 2), and we divided the cohort into groups using MVX increments of 10 (group 1: ≤50; group 2: >50 and ≤60; group 3: >60 and ≤70; group 4: >70) for ease of clinical interpretation.
Figure 1.
Distribution of Metabolic Vulnerability Index (MVX) Scores among 1,382 community-dwelling persons with HF.
Figure 2. Association between MVX and mortality using a cubic spline (left panel) or study defined cut-points (right panel).
Hazard ratios (HR) and 95% confidence intervals are shown on both plots.
In univariable analyses, compared to the lowest MVX group, higher MVX was associated with older age, male sex, higher NYHA class, higher MAGGIC score, higher Charlson comorbidity index, NT-proBNP, a higher prevalence of atrial fibrillation, HF duration > 18 months, lower BMI, lower hemoglobin and eGFR, and a lower prevalence of hypertension and hyperlipidemia. In multivariable analyses, age, sex, NYHA class, atrial fibrillation, and HF duration remained independently associated with higher MVX (p < 0.05). Notably, we did not detect an association between MVX and ejection fraction modeled continuously or categorically.
MVX and Mortality
Over a median (IQR) follow-up of 13.9 (11.5, 15.4) years, 1,158 patients died equating to a 5-year all-cause mortality rate of 51.8% (95% CI: 49.1–54.4%). This corresponds to 14.5 (95% CI: 13.6–15.3) deaths per 100 patient-years. Mortality also varied by MVX group with a graded positive association between MVX group and mortality (Ptrend < 0.001). The 5-year mortality rate in MVX group 1 was 23.5% (95% CI: 16.8–29.6%), compared to 69.0% (95% CI: 64.3–73.1%) in MVX group 4 (Figure 3). After adjustment for age and sex, patients in MVX group 4 had a nearly threefold increase in the risk of death compared to group 1 (HR 2.8; 95% CI: 2.3–3.4) (Table 2). Adjustment for the MAGGIC score only minimally attenuated this association (HR 2.5; 95% CI: 2.0–3.1). After sequential adjustment for NT-proBNP and hemoglobin, MVX group 4 remained associated with a large increase in the risk of death (HR 1.8; 95% CI: 1.40–2.2). Results were similar when the follow-up was restricted to three or five years and when a complete case analysis was carried out. Of the individual MVX components, the inflammation vulnerability index had the highest HR (1.4; 95% CI: 1.3–1.5) per one standard deviation (Figure 4).
Figure 3. Kaplan-Meier Survival Curves by Metabolic Vulnerability Index (MVX) Group.
MVX group 1: ≤50 (N = 171); MVX group 2: (50–60] (N = 339); MVX group 3: (60–70] (N = 445); MVX group 4: >70 (N = 427).
Table 2. Association between Metabolic Vulnerability Index (MVX) and mortality.
Hazard ratios (HR) and 95% confidence intervals; MVX group 1 is the reference group.
| Model | MVX Group 1 (N = 171) | MVX Group 2 (N=339) | MVX Group 3 (N=445) | MVX Group 4 (N=427) | P trend |
|---|---|---|---|---|---|
| Deaths per 100 patient-years | 7.3 (6.0–8.7) | 10.8 (9.5–12.1) | 17.2 (15.5–18.8) | 22.6 (20.3–24.8) | N/A |
| Univariate HR | 1.00 (reference) | 1.47 (1.18–1.84) | 2.33 (1.89–2.87) | 3.03 (2.45–3.74) | <0.001 |
| HR adjusted for age and sex | 1.00 (reference) | 1.37 (1.10–1.71) | 2.19 (1.78–2.71) | 2.78 (2.25–3.44) | <0.001 |
| HR adjusted for MAGGIC score | 1.00 (reference) | 1.31 (1.05–1.63) | 2.01 (1.63–2.48) | 2.52 (2.03–3.11) | <0.001 |
| HR adjusted for MAGGIC score + NT-proBNP | 1.00 (reference) | 1.17 (0.93–1.47) | 1.64 (1.32–2.05) | 1.94 (1.55–2.45) | <0.001 |
| HR adjusted for MAGGIC score + NT-proBNP + hemoglobin | 1.00 (reference) | 1.15 (0.92–1.44) | 1.57 (1.26–1.96) | 1.80 (1.39–2.21) | <0.001 |
MVX group 1: ≤50 (N = 171); MVX group 2: (50–60]; MVX group 3: (60–70]; MVX group 4: >70. Abbreviations: MAGGIC= Meta-analysis global group in chronic heart failure
Figure 4. Association of Metabolic Vulnerability Index (MVX) individual components with mortality adjusted for MAGGIC score.
Hazard ratios and 95% confidence intervals are shown per one standard deviation. All biomarker units are μmol/l. Abbreviations: MAGGIC: Meta-analysis group in chronic heart failure; IVX: Inflammation vulnerability index; MMX: Metabolic malnutrition index; S-HDLP: Small high-density lipoprotein particles
Notably, we observed no significant interaction between ejection fraction and MVX group in survival analyses stratified by ejection fraction group (reduced <50% and preserved ≥50%). The HR for the association between cardiovascular death (45% of all deaths) and MVX group 4 was 1.6 (95% CI: 1.2–2.1) compared to MVX group 1, similar to the HR of 1.8 (95% CI: 1.4–2.2) for all-cause mortality. In survival analyses stratified by MAGGIC score subgroups (Table 3), we observed a positive association between MVX group and increased mortality across all groups. Notably, MVX group 4 HR was highest among patients in the lowest-risk MAGGIC score subgroup (HR 2.8; 95% CI: 1.5–5.2).
Table 3. Metabolic Vulnerability Index (MVX) group association with mortality across MAGGIC score subgroups.
Hazard ratios (HR) and 95% confidence intervals; MVX Group 1 is the reference. Cox models were adjusted for NT-proBNP and hemoglobin.
| MAGGIC Score Subgroup | N Deaths / N | Deaths per 100 patient-years | MVX Group 2 (N=339) | MVX Group 3 (N=445) | MVX Group 4 (N=427) | P trend |
|---|---|---|---|---|---|---|
| < 21 | 145 / 267 | 6.1 (5.1–7.1) | 1.30 (0.74–2.31) | 2.22 (1.29–3.83) | 2.83 (1.54–5.21) | <0.001 |
| [21–25) | 243 / 293 | 12.7 (11.1–14.3) | 1.00 (0.64–1.55) | 1.33 (0.86–2.06) | 1.33 (1.85–2.09) | 0.60 |
| [25–29) | 340 / 378 | 17.9 (16.0–19.8) | 1.28 (0.82–2.00) | 1.48 (0.94–2.35) | 1.82 (1.13–2.91) | <0.001 |
| ≥ 29 | 430 / 444 | 24.2 (21.9–26.5) | 1.12 (0.73–1.71) | 1.55 (1.04–2.32) | 1.78 (1.18–2.70) | <0.001 |
MVX group 1: ≤50; N = 171. MVX group 2: (50–60]; MVX group 3: (60–70]; MVX group 4: >70. Abbreviation: MAGGIC= Meta-analysis global group in chronic heart failure, NT-proBNP: N-terminal pro-brain natriuretic peptide
Finally, we evaluated the incremental value of adding MVX group to reference models including MAGGIC score, NT-proBNP and hemoglobin at three years. The Uno C-statistic, net reclassification improvement, and integrated discrimination improvement indicate that MVX group improves model performance beyond these clinical variables (Table 4).
Table 4.
Measures of Prognostic Model Improvement.
| MAGGIC score | MAGGIC score + NT-proBNP | MAGGIC score + NT-proBNP + hemoglobin | MAGGIC score + NT-proBNP + hemoglobin + MVX group | |
|---|---|---|---|---|
| Uno’s C-statistic (95% CI) | 0.61 (0.59–0.64) | 0.66 (0.64–0.69) | 0.67 (0.65–0.70) | 0.69 (0.67–0.71) |
| * p-value | N/A | < 0.001 | 0.035 | < 0.001 |
| IDI, % (95% CI) | N/A | 3.6 (2.3–4.8) | 1.2 (0.5–2.1) | 1.6 (0.7–2.7) |
| NRI, % (95% CI) | N/A | 20 (16–25) | 14 (10–20) | 18 (11–24) |
p-value indicates whether c-statistic is significantly increased by adding a new variable to the previous model
Abbreviations: MAGGIC= Meta-analysis global group in chronic heart failure, NT-proBNP: N-terminal pro-brain natriuretic peptide, IDI: integrated discrimination improvement, NRI: net reclassification improvement
Discussion
We report that MVX score is associated with a large increase in the risk of death in a community cohort representing the entire spectrum of the HF syndrome. Patients in the highest MVX group were nearly three times more likely to die compared to those in the lowest MVX group showing a graded positive association between MVX and mortality. The strong association persisted after adjustment for the MAGGIC score and other biomarkers of risk, and in stratified analyses. MVX added substantial information on the risk of death in all categories relying on predetermined values of the MAGGIC score. Further, we found evidence that MVX improved the classification of the risk of death at three years over the MAGGIC score and validated prognostic biomarkers using three distinct measurements of model performance. Our results are consistent with those of the CATHGEN observational cohort; among 1,556 patients with HF in CATHGEN, there was a strong positive association between MVX and mortality, in a fully adjusted model for several clinical factors, the HR for MVX was 1.95 (95% CI: 1.7–2.2) per one SD [10]. Collectively, our findings suggest that MVX can provide substantial clinical benefit for mortality risk stratification across the entire spectrum of the HF syndrome as all associations were independent of ejection fraction.
Determinants of MVX in HF
Patients had a median MVX score of 64.6 and a mean score of 63, notably higher than the mean of 50 reported in the CATHGEN cohort, reflecting differences in study populations [10]. Indeed, patients in CATHGEN were younger, more likely to be men, had a lower prevalence of hypertension and diabetes and only a minority of patients had HF. In the present cohort, higher MVX was associated with older age, male sex, greater HF duration, higher NYHA class, and atrial fibrillation, all clinical indicators of more advanced HF [31–33]. The distribution of MVX did not differ by ejection fraction.
Inflammation and Malnutrition in Patients with HF
Prior studies mainly focused on the association of single markers of inflammation, such as C-reactive protein (CRP), interleukin-6, tumor necrosis factor-α, and Galectin-3, with HF severity and prognosis [2–4], with few studies of S-HDLP and GlycA in HF [34–37]. Reports of an inverse association between S-HDLP and mortality in HF reflected heterogenous designs, varying population size, endpoint definition, and follow-up duration [34–36]. Data on GlycA in HF are scarce, with a positive association between elevated GlycA and a composite endpoint of hospital readmission and mortality in nonischemic patients only noted in a small, convenience sample of ambulatory patients with chronic HF [37].
Studies of malnutrition markers in HF are equally scarce, with limited data on citrate and the branched-chain amino acids (leucine, isoleucine, and valine). In a referral population of 130 acute HF patients, citrate was positively associated with 3-month mortality (odds ratio: 11.74; 95% CI: 1.44–113.20) [38]. A study of 41 chronic HF patients found an inverse association between the branched-chain amino acids and NYHA class, suggesting an association with worse prognosis [39]. However, the patient selection, small sample size, and wide confidence intervals compromise inference and validity.
Composite biomarker indices provide more comprehensive mechanistic “coverage” than an individual biomarker. The Glasgow Prognostic Score [40] is a categorical scoring system based on CRP and albumin initially proposed to assess inflammation and malnutrition in cancers. It was recently evaluated in two studies of HF. The first study, of 443 patients presenting with chronic stable HF with reduced ejection fraction at a tertiary care center, found increased Glasgow Prognostic Score to predict mortality at three years independently of age and NT-proBNP [41]. Likewise, in a multicenter sample of 870 patients hospitalized with acute decompensated HF, those with the highest Glasgow Prognostic Score had a nearly three-fold increased risk of short term (18 months) death compared to patients with the lowest scores independently of clinical risk factors [42].
The Glasgow Prognostic Score relies on a point system categorically integrating two variables each with two levels. Conversely, MVX captures six biomarkers as continuous variables, which conceptually provide more comprehensive metabolic information while allowing for greater range of values [28, 43, 44]. Specifically for inflammation, there is evidence that GlycA and high-sensitivity CRP have distinct inflammation-related metabolic effects [45], and several epidemiological studies reported associations between branched-chain amino acids and cardiovascular risk [46, 47].
Therefore, our findings provide novel evidence that a composite biomarker index that encompasses comprehensive measures of inflammation and metabolic malnutrition provides important prognostic information in a large community cohort of optimal clinical generalizability.
A key challenge in biomarker research is to identify markers with predictive capabilities that are substantial enough to change clinical practice. We acknowledge the challenge in doing so given controversies surrounding the preferred approach to assess model performance [48]. These challenges notwithstanding, the substantial incremental value of MVX over the MAGGIC score, a class 2.a. recommendation in the 2022 HF Guidelines [49] is particularly notable as it is independent from ejection fraction. These data thus suggest that the MVX score, an NMR-based assessment of inflammation and metabolic malnutrition, may have a broad applicability to stratify risk across the entire spectrum of the HF syndrome.
Limitations and Strengths
Our cohort was predominantly of European ancestry, limiting the generalizability of our findings in other populations and warranting research in a more racially and ethnically diverse population. As in any observational study, we cannot rule out residual confounding. Additionally, more contemporary HF guideline-directed medical therapy were not assessed given the time of the study. Finally, these results require replication in a different cohort.
Our study has several important strengths. We examined the association of MVX and mortality in a population-based cohort that represents the community practice and has strong clinical relevance. Nearly all in-patient and outpatient encounters within the Rochester Epidemiology Project were captured, providing us a rich clinical data set which enabled comprehensive adjustments for known indicators of risk in HF.
Clinical Perspectives: Competency in Medical Knowledge
In this HF community cohort, MVX, a composite measure of inflammation and metabolic malnutrition, conferred strong incremental prognostic information over clinically validated biomarkers and the MAGGIC score, adding risk prediction information even among patients considered to be low risk. Thus, MVX may offer a feasible and scalable method to measure inflammation and metabolic malnutrition, which can improve risk stratification in HF.
Translational Outlook
Further studies are warranted to define the relationship between MVX and other clinical indicators of inflammation and malnutrition in HF (e.g., frailty, sarcopenia, cachexia).
Acknowledgements
We would like to extend our appreciation to Mary Walter, Yuhai Dai, the NIDDK Clinical Laboratory Core, and Rebecca Oyetoro of the NHLBI whose assistance was critical to measuring key laboratory variables.
Funding:
The investigators were supported by the Intramural Research program of the National Heart Lung and Blood Institute of the National Institutes of Health. This study also used in part the resources from the Rochester Epidemiology Project (REP) medical records-linkage system, which is supported by the National Institute on Aging (NIA; AG 058738), by the Mayo Clinic Research Committee, and by fees paid annually by REP users. The funding institution did not play a role design, conduct, analysis, or reporting nor in the decision to submit this manuscript for publication. The study was approved by the Mayo Clinic and Olmsted Medical Center Institutional Review Boards.
ABBREVIATIONS AND ACRONYMS
- HF
heart failure
- MVX
Metabolic Vulnerability Index
- NMR
nuclear magnetic resonance
- MAGGIC
Meta-Analysis Global Group in Chronic HF
- NYHA
New York Heart Association
- S-HDLP
small high-density lipoprotein particles
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Disclosures: MAC is an employee of and holds stock in LabCorp. JDO is a consultant, stockholder, and former employee of LabCorp.
References
- 1.Seta Y, et al. , Basic mechanisms in heart failure: the cytokine hypothesis. J Card Fail, 1996. 2(3): p. 243–9. [DOI] [PubMed] [Google Scholar]
- 2.Murphy SP, et al. , Inflammation in Heart Failure: JACC State-of-the-Art Review. J Am Coll Cardiol, 2020. 75(11): p. 1324–1340. [DOI] [PubMed] [Google Scholar]
- 3.Schumacher SM and Naga Prasad SV, Tumor Necrosis Factor-alpha in Heart Failure: an Updated Review. Curr Cardiol Rep, 2018. 20(11): p. 117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Shah RV, et al. , Galectin-3, cardiac structure and function, and long-term mortality in patients with acutely decompensated heart failure. Eur J Heart Fail, 2010. 12(8): p. 826–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Conraads VM, Bosmans JM, and Vrints CJ, Chronic heart failure: an example of a systemic chronic inflammatory disease resulting in cachexia. Int J Cardiol, 2002. 85(1): p. 33–49. [DOI] [PubMed] [Google Scholar]
- 6.Yin J, et al. , New insights into the pathogenesis and treatment of sarcopenia in chronic heart failure. Theranostics, 2019. 9(14): p. 4019–4029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Levy WC, et al. , The Seattle Heart Failure Model: prediction of survival in heart failure. Circulation, 2006. 113(11): p. 1424–33. [DOI] [PubMed] [Google Scholar]
- 8.Pocock SJ, et al. , Predicting survival in heart failure: a risk score based on 39 372 patients from 30 studies. Eur Heart J, 2013. 34(19): p. 1404–13. [DOI] [PubMed] [Google Scholar]
- 9.Peterson PN, et al. , A validated risk score for in-hospital mortality in patients with heart failure from the American Heart Association get with the guidelines program. Circ Cardiovasc Qual Outcomes, 2010. 3(1): p. 25–32. [DOI] [PubMed] [Google Scholar]
- 10.Otvos JD, Shalaurova I, May HT, et al. Multimarkers of metabolic malnutrition and inflammation and their association with mortality risk in cardiac catheterisation patients: a prospective, longitudinal, observational, cohort study. Lancet Healthy Longev. 2023;4(2):e72–e82. doi: 10.1016/S2666-7568(23)00001-6 [DOI] [PubMed] [Google Scholar]
- 11.Rocca WA, et al. , History of the Rochester Epidemiology Project: half a century of medical records linkage in a US population. Mayo Clin Proc, 2012. 87(12): p. 1202–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Rocca WA, et al. , Data Resource Profile: Expansion of the Rochester Epidemiology Project medical records-linkage system (E-REP). Int J Epidemiol, 2018. 47(2): p. 368–368j. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Dunlay SM, et al. , Activities of daily living and outcomes in heart failure. Circ Heart Fail, 2015. 8(2): p. 261–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Dunlay SM, et al. , Prognostic value of biomarkers in heart failure: application of novel methods in the community. Circ Heart Fail, 2009. 2(5): p. 393–400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Pakhomov SV, Buntrock J, and Chute CG, Prospective recruitment of patients with congestive heart failure using an ad-hoc binary classifier. J Biomed Inform, 2005. 38(2): p. 145–53. [DOI] [PubMed] [Google Scholar]
- 16.McKee PA, et al. , The natural history of congestive heart failure: the Framingham study. N Engl J Med, 1971. 285(26): p. 1441–6. [DOI] [PubMed] [Google Scholar]
- 17.Charlson ME, et al. , A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis, 1987. 40(5): p. 373–83. [DOI] [PubMed] [Google Scholar]
- 18.Goodman RA, et al. , Defining and measuring chronic conditions: imperatives for research, policy, program, and practice. Prev Chronic Dis, 2013. 10: p. E66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.US Department of Health and Human Services., Multiple chronic conditions - a strategic framework: optimum health and quality of life for individuals with multiple chronic conditions. December, 2010: Washington, DC. [Google Scholar]
- 20.Garcia E, et al. , The extended lipid panel assay: a clinically-deployed high-throughput nuclear magnetic resonance method for the simultaneous measurement of lipids and Apolipoprotein B. Lipids Health Dis, 2020. 19(1): p. 247. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.McGarrah RW, et al. , A Novel Protein Glycan-Derived Inflammation Biomarker Independently Predicts Cardiovascular Disease and Modifies the Association of HDL Subclasses with Mortality. Clin Chem, 2017. 63(1): p. 288–296. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.McGarrah RW, et al. , High-density lipoprotein subclass measurements improve mortality risk prediction, discrimination and reclassification in a cardiac catheterization cohort. Atherosclerosis, 2016. 246: p. 229–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Otvos JD, et al. , GlycA: A Composite Nuclear Magnetic Resonance Biomarker of Systemic Inflammation. Clin Chem, 2015. 61(5): p. 714–23. [DOI] [PubMed] [Google Scholar]
- 24.Otvos JD, et al. , Relations of GlycA and lipoprotein particle subspecies with cardiovascular events and mortality: A post hoc analysis of the AIM-HIGH trial. J Clin Lipidol, 2018. 12(2): p. 348–355 e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Rohatgi A, et al. , HDL in the 21st Century: A Multifunctional Roadmap for Future HDL Research. Circulation, 2021. 143(23): p. 2293–2309. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Altman DG, et al. , Review of survival analyses published in cancer journals. Br J Cancer, 1995. 72(2): p. 511–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.van Buuren S, & Groothuis-Oudshoorn K mice: Multivariate Imputation by Chained Equations in R. Journal of Statistical Software, 2011. 45(3), 1–67. [Google Scholar]
- 28.Bennette C and Vickers A, Against quantiles: categorization of continuous variables in epidemiologic research, and its discontents. BMC Med Res Methodol, 2012. 12: p. 21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Woodward M, Tunstall-Pedoe H, and Peters SA, Graphics and statistics for cardiology: clinical prediction rules. Heart, 2017. 103(7): p. 538–545. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Uno H, Cai T, Pencina MJ, D’Agostino RB, & Wei LJ (2011). On the C‐statistics for evaluating overall adequacy of risk prediction procedures with censored survival data. Statistics in Medicine, 30(10), 1105–1117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Lam CSP, et al. , Sex differences in heart failure. Eur Heart J, 2019. 40(47): p. 3859–3868c. [DOI] [PubMed] [Google Scholar]
- 32.Kalantar-Zadeh K, et al. , Reverse epidemiology of conventional cardiovascular risk factors in patients with chronic heart failure. J Am Coll Cardiol, 2004. 43(8): p. 1439–44. [DOI] [PubMed] [Google Scholar]
- 33.Horwich TB, Fonarow GC, and Clark AL, Obesity and the Obesity Paradox in Heart Failure. Prog Cardiovasc Dis, 2018. 61(2): p. 151–156. [DOI] [PubMed] [Google Scholar]
- 34.Potocnjak I, et al. , Serum Concentration of HDL Particles Predicts Mortality in Acute Heart Failure Patients. Sci Rep, 2017. 7: p. 46642. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Hunter WG, et al. , High-Density Lipoprotein Particle Subfractions in Heart Failure With Preserved or Reduced Ejection Fraction. J Am Coll Cardiol, 2019. 73(2): p. 177–186. [DOI] [PubMed] [Google Scholar]
- 36.Teis A, et al. , Particle size and cholesterol content of circulating HDL correlate with cardiovascular death in chronic heart failure. Sci Rep, 2021. 11(1): p. 3141. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Cediel G, et al. , GlycA and GlycB as Inflammatory Markers in Chronic Heart Failure. Am J Cardiol, 2022. 181: p. 79–86. [DOI] [PubMed] [Google Scholar]
- 38.Stryeck S, et al. , Serum Concentrations of Citrate, Tyrosine, 2- and 3- Hydroxybutyrate are Associated with Increased 3-Month Mortality in Acute Heart Failure Patients. Sci Rep, 2019. 9(1): p. 6743. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Aquilani R, et al. , Plasma Amino Acid Abnormalities in Chronic Heart Failure. Mechanisms, Potential Risks and Targets in Human Myocardium Metabolism. Nutrients, 2017. 9(11). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Forrest LM, et al. , Evaluation of cumulative prognostic scores based on the systemic inflammatory response in patients with inoperable non-small-cell lung cancer. Br J Cancer, 2003. 89(6): p. 1028–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Cho A, et al. , The inflammation-based modified Glasgow prognostic score is associated with survival in stable heart failure patients. ESC Heart Fail, 2020. 7(2): p. 654–662. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Itagaki T, et al. , Inflammation-based assessment for the risk stratification of mortality in patients with heart failure. Sci Rep, 2021. 11(1): p. 14989. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Royston P, Altman DG, and Sauerbrei W, Dichotomizing continuous predictors in multiple regression: a bad idea. Stat Med, 2006. 25(1): p. 127–41. [DOI] [PubMed] [Google Scholar]
- 44.Altman DG and Royston P, The cost of dichotomising continuous variables. BMJ, 2006. 332(7549): p. 1080. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Levine JA, et al. , Associations of GlycA and high-sensitivity C-reactive protein with measures of lipolysis in adults with obesity. J Clin Lipidol, 2020. 14(5): p. 667–674. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Ruiz-Canela M, et al. , Plasma Branched-Chain Amino Acids and Incident Cardiovascular Disease in the PREDIMED Trial. Clin Chem, 2016. 62(4): p. 582–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Tobias DK, et al. , Circulating Branched-Chain Amino Acids and Incident Cardiovascular Disease in a Prospective Cohort of US Women. Circ Genom Precis Med, 2018. 11(4): p. e002157. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Cook NR, Quantifying the added value of new biomarkers: how and how not. Diagn Progn Res, 2018. 2: p. 14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Heidenreich PA, et al. , 2022 AHA/ACC/HFSA Guideline for the Management of Heart Failure: Executive Summary: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. J Am Coll Cardiol, 2022. 79(17): p. 1757–1780. [DOI] [PubMed] [Google Scholar]