Abstract
Background:
Increased high sensitivity C-reactive protein (hsCRP), a marker of inflammation, is associated with incident cardiovascular events. We aim to determine if baseline or trajectory of hsCRP levels over time predict incident HF hospitalization.
Methods:
Jackson Heart Study (JHS) participants’ (n=3920 Black adults) hsCRP levels were measured over 3 visits (from 2000 to 2013). We assessed the association of hsCRP at baseline (visit-1) with incident HF hospitalization using Cox proportional hazard models. Furthermore, we assessed the association of trajectory of hsCRP over repeated measurements (visits 1–3) with incident HF using joint models. Hazard ratios (HR) are reflective of an increase in hsCRP by 1 standard deviation on a Log2 scale. We also assessed the association of change in hsCRP between visit-1 and visit-3 with Cox-proportional hazards models by grouping patients by low (<2 mg/L) and high (≥2 mg/L) hsCRP levels. The 4 groups were low-to-low (referent), low-to-high, high-to-low, and high-to-high.
Results:
Mean baseline age of participants was 54±13 years and 63.8% were women. Over a median follow-up of 12 years, 308 (7.9%) participants were hospitalized with incident HF. Baseline hsCRP was not associated with incident HF (aHR 1.08; 95%CI: 0.96 to 1.22). However, increasing hsCRP levels over repeated measures were associated with a higher risk of incident HF overall (aHR 1.22; 95%CI: 1.03 to 1.44) and HF with preserved ejection fraction (HFpEF, aHR 1.30; 95%CI: 1.02 to 1.65) but not HF with reduced ejection fraction (p>0.05). Furthermore, change in hsCRP from low-to-high and high-to-low levels were associated with incident HF (p<0.05).
Conclusion:
While baseline hsCRP was not associated with incident HF, an increasing trajectory of hsCRP over time was associated with increased risk for incident HF (particularly HFpEF). Temporal change in hsCRP may be an important marker of risk for incident HFpEF in Black adults.
Keywords: Heart Failure, Inflammation, CRP, Biomarker, Incident, Trajectory
Introduction
Systemic inflammation is an intermediary step in the pathogenesis of heart failure (HF). C-reactive protein (CRP) is an inflammatory marker and acute phase protein1 and is a strong predictor of cardiovascular (CV) events and death from CV diseases (CVD).2 The predictive value of high sensitivity CRP (hsCRP) is recognized by the American Heart Association (AHA) in CV risk stratification.3
Several studies demonstrate an association of CRP levels with incident HF and its utility in HF risk assessment.4–8 However, others demonstrate no association of CRP with HF.9–11 Most of these studies have only evaluated CRP at a single time point and changes in CRP (or trajectories) have not been adequately assessed. Additionally, few studies have assessed the association of CRP with subtypes of HF, namely HF with reduced ejection fraction (HFrEF) and HF with preserved ejection fraction (HFpEF).8 Furthermore, Black adults remain an under-studied population, although they have a higher incidence of HF and greater risk of HF hospitalization compared with White adults.12
Inflammation is a variable state of the body and is subject to variations in blood pressure, alcohol consumption, or lipid levels among many other factors.3 Increasing trajectory of CRP, is associated with metabolic disease and hypertension.13 Thus, repeated measures of hsCRP may be required to uncover associations with CVD. While mixed data on the association of CRP and HF exists, the association of trajectory of CRP and HF is an understudied area. Thus, the aim of our study was to assess the association of trajectory of hsCRP with incident HF (including HFrEF and HFpEF) in a large cohort of Black adults.
Methods
Study population
The Jackson Heart Study (JHS) is a large, prospective community-based observational study designed to assess the risks and associations of CVD in Black Americans. The JHS recruited 5306 participants residing in the Jackson, Mississippi tri-county area (Hinds, Rankin, and Madison counties). Participants were evaluated at baseline (visit 1: 2000 to 2004) and 2 follow-up visits (visit 2: 2005 to 2008; visit 3: 2009 to 2013).14 Further details of the JHS study design, recruitment, and data collection have been previously described.15,16 The data, study materials, and analytic methods are available to other researchers for purposes of reproducing the results or replicating the procedure by following the JHS publications procedures and data use agreements.17
The study was approved by the Institutional Review Boards of Jackson State University, Tougaloo College, and the University of Mississippi Medical Center. All participants provided written informed consent. Furthermore, this manuscript has been reviewed by the JHS publications and presentations committee and has been approved for publication.
Inclusion/Exclusion criteria
Among 5306 participants at visit 1, we excluded 1386 participants with missing parameters (hsCRP measurements and covariates of interest), baseline HF, and severe valvular disease. Participants with severe valvular disease were excluded as a competing risk for HF. After exclusions, 3920 participants were available for the main analysis (Figure S1). For our secondary analysis of change in hsCRP measurements between visit 1 and visit 3, 935 participants were excluded who had missing hsCRP at visit 3 or a HF event before visit 3. This resulted in 2985 participants (Figure S1). Comparison of baseline characteristics of included and excluded participants is shown in Table S1.
Echocardiography
Echocardiograms were conducted in the JHS clinic facilities during the baseline clinical examination using standardized methods as previously described.18 Echocardiographic variables assessed as outcomes in our study included LV mass (LVM), LV ejection fraction (LVEF), posterior wall thickness (PWT), interventricular septal thickness (IVST), left atrial diameter (LAD), systolic/diastolic LV internal dimensions (LVID), LVM index (LVMI), LVM indexed to height, and relative wall thickness (RWT). LVM calculation was based on the standard formula: LVM (g)=0.8×1.04 [(diastolic LVID + interventricular septal wall thickness + PWT)3 − (diastolic LVID)3] + 0.6. LVEF was derived semi-quantitatively by the primary cardiologist using a modified Quinones technique and visual assessment of LV apex. LVMI was calculated by LVM indexed to body surface area (BSA).
hsCRP measurements
hsCRP was measured by the immunoturbidimetric CRP-Latex assay (Kamiya Biomedical Company, Seattle, WA) using a Hitachi 911 analyzer (Roche Diagnostics, Indianapolis, IN) as previously described.19 Measurement was done in duplicates, and any duplicates that were not within a 3 assay SD from one another were rerun. The interassay coefficient of variation on control samples was 4.5% and 4.4%. Approximately 6% of the samples were measured as masked replicates to assess repeatability of measurements. The reliability coefficient for masked quality control replicates was 0.95 for the hsCRP assay. hsCRP data is presented in mg/L.
Covariates
In-depth home interviews, clinic visits, clinical examinations, and laboratory tests were conducted on all JHS participants at baseline visit using standardized measures and protocols as previously described.20 Demographic characteristics including age, sex, coronary heart disease (CHD) history, smoking status, physical activity level, alcohol use, insurance status, income, and educational level were self-reported.15 Body mass index (BMI) was calculated as weight (kg)/height (m)2. Estimated glomerular filtration rate (eGFR) was calculated using the CKD-Epi equation.
Heart failure outcomes
The primary outcome was time to incident HF hospitalization. In the JHS cohort, HF hospitalization surveillance began January 1, 2005. Potential HF hospitalizations were identified and adjudicated as previously described.21 In brief, hospitalization data were obtained from the hospital discharge index from all catchment area hospitals and annual follow-up information. The self-reported data from annual follow-up were confirmed with the hospital discharge index data. The primary diagnoses based on International Classification of Diseases codes were reviewed by trained medical personnel and adjudicated by trained adjudicators based on signs and symptoms, clinical documentation, laboratory tests, chest x-ray films, and other imaging modalities. 22 Overall HF events included all incident HF hospitalizations. HF events were further classified as HFrEF (left ventricular ejection fraction [LVEF] <50%) or HFpEF (LVEF >50%) according to available cardiac imaging data before and at the time of the initial hospitalization for HF.23
Statistical analysis
Continuous variables were summarized as means, geometric means, standard deviations, medians, and interquartile ranges. Categorical variables were summarized as frequencies and percentages. We compared the baseline characteristics amongst the groups using chi-square, t-test, ANOVA, or Kruskal-Wallis tests, as appropriate based on the underlying distribution.
As hsCRP values were non-normally distributed (right skewed), hsCRP values were log-transformed for analyses.3 The association of log-transformed hsCRP was then analyzed with echocardiographic parameters via multivariable regression models. These models adjusted for Model 1= Age + sex; Model 2= Model 1 + systolic blood pressure (SBP), antihypertensive medication, BMI, diabetes status, eGFR, smoking history, CHD history, cholesterol to high density lipoprotein cholesterol (HDL-C) ratio, and statin use.
For our longitudinal analysis we utilized joint modelling to assess the association between variability of hsCRP through repeated measurements at JHS visits 1 through 3. A joint model combines a linear mixed-effects model for serial measurements to incorporate estimated hsCRP (at time-t) with a Cox proportional hazards model to estimate risk of incident HF hospitalization (time-t). A joint model can account for trajectory of change (via incorporation of random slope and random intercept) in hsCRP as opposed to isolated measurements. For the joint model analysis with repeated hsCRP measurements, we performed univariate analyses as an unadjusted model and then adjusted models 1 and 2 as described above. Model 3 additionally incorporated the random slope to account for trajectory of change. The results are presented as adjusted hazard ratios (HR) per 1 standard deviation (SD) increase of the hsCRP level (on a log2 scale) at any point in time with 95% confidence intervals.24
For our secondary analysis we dichotomized hsCRP into high (≥2 mg/L) and low (<2 mg/L) levels similar to the Canakinumab Antiinflammatory Thrombosis Outcome Study (CANTOS) trial25 to assess if a high level at baseline is associated with incident HF, with low hsCRP levels as the referent. Similarly, we also assessed the association of a change in high or low levels of hsCRP between visit 1 and visit 3. Hence, these participants were stratified into 4 groups of change, i.e. low-to-low (referent), low-to-high, high-to-low, and high-to-high hsCRP levels. These groups of change in hsCRP were used to analyze association of trajectory of change in hsCRP with incident HF hospitalization using a Kaplan Meier survival plot and Cox proportional hazards models. These models were adjusted for models 1 and 2 as described for the primary analysis.
We also conducted a sensitivity analysis with exclusion of participants with prevalent CHD as well as incident CHD (during study period) as participants with CHD may have an increased risk of HF at baseline and a bi-directional association with hsCRP. We included another sensitivity analysis for social determinants of health. including Model 3 + insurance, education, and income variables.
Joint modelling analysis and survival analysis was performed with R statistical software (R Core Team, 2017). While remaining analyses were performed using SAS version 9.4 (SAS Institute, Cary, North Carolina). A 2-sided p-value of <0.05 was considered statistically significant.
Results
Patient demographics
Overall, 3920 participants were included in our analyses, 308 (7.9%) of these participants developed HF over a median 12.0 years at an incidence rate of 7.5 HF events per 1000 person-years. At baseline, the mean age of the overall population was 54±13 years, mean BMI of 31.6±7.1 kg/m2, 63.8% were women, 53.7% had hypertension, 18.0% had diabetes, 4.5% had prevalent CHD, had a median hsCRP of 2.6 [1.1 to 5.6] mg/L, geometric mean hsCRP of 3.5 (95%CI 3.3 to 3.7) mg/L, and 12.5% were taking statins. At baseline, the participants who developed HF over the study were older, had higher BMI, SBP, HbA1c, hsCRP levels, and a greater proportion of patients were taking statins compared with participants who did not develop HF. Participants who developed HF also had a greater prevalence of hypertension, as well as a greater than two-fold prevalence of diabetes and CHD compared with participants who did not develop HF (Table 1).
Table 1.
Characteristics of Jackson Heart Study participants with or without heart failure at baseline
| All participants | Heart failure status | p-value | ||
|---|---|---|---|---|
| No heart failure | Heart failure | |||
| Participants, n (%) | 3920 | 3612 (92.1) | 308 (7.9) | |
| Mean age, years | 54±13 | 54±12 | 64±11 | <0.001 |
| Women, n (%) | 2502 (63.8) | 2302 (63.7) | 200 (64.9) | 0.67 |
| Hypertension status, n (%) | 2104 (53.7) | 1862 (51.6) | 242 (78.6) | <0.001 |
| Diabetes status, n (%) | 706 (18.0) | 580 (16.1) | 126 (40.9) | <0.001 |
| Median hsCRP, mg/L | 2.59 [1.06 – 5.59] | 2.56 [1.04 – 5.51] | 3.15 [1.37 – 6.23] | 0.08 |
| Geometric mean hsCRP (95% CI), mg/L | 3.53 (3.33 to 3.73) | 3.44 (3.24 to 3.65) | 4.70 (3.88 to 5.69) | 0.004 |
| Mean systolic blood pressure, mmHg | 127±17 | 126±16 | 135±19 | <0.001 |
| Mean diastolic blood pressure, mmHg | 75.9±8.5 | 76.0±8.4 | 74.7±9.3 | 0.017 |
| Mean HbA1c level, % | 5.9±1.2 | 5.8±1.1 | 6.5±1.8 | <0.001 |
| Mean eGFR, mL/min | 96±20 | 97±20 | 84±26 | <0.001 |
| Mean BMI, kg/m 2 | 31.6±7.1 | 31.6±7.1 | 32.6±7.5 | 0.019 |
| Cholesterol to HDL-C ratio | 4.1±1.4 | 4.1±1.3 | 4.2±1.6 | 0.28 |
| Antihypertensive medication use, n (%) | 1919 (49.0) | 1691 (46.8) | 228 (74.0) | <0.001 |
| Antidiabetic medication use, n (%) | 447 (11.4) | 350 (9.7) | 97 (31.5) | <0.001 |
| Statin medication use, n (%) | 489 (12.5) | 413 (11.4) | 76 (24.7) | <0.001 |
| Physical activity classification, n (%) | <0.001 | |||
| Poor Health | 1850 (47.2) | 1662 (46.0) | 188 (61.0) | |
| Intermediate Health | 1280 (32.7) | 1196 (33.1) | 84 (27.3) | |
| Ideal Health | 790 (20.2) | 754 (20.9) | 36 (11.7) | |
| Current smoker, n (%) | 470 (12.0) | 424 (11.7) | 46 (14.9) | 0.10 |
| Prevalent CHD, n (%) | 177 (4.5) | 144 (4.0) | 33 (10.7) | <0.001 |
BMI= body mass index; CHD= coronary heart disease; CI= confidence interval; eGFR= estimated glomerular filtration rate; HbA1c= glycated hemoglobin A1c; HDL-C= high density lipoprotein cholesterol; hsCRP= high sensitivity C-reactive protein
Amongst 2985 participants included in the change in hsCRP analysis by high (≥2 mg/L) and low (<2 mg/L) levels from visit 1 to visit 3, participants within the high-to-high group had the highest proportion of women, highest BMI, highest hsCRP, and the least proportion of statin use amongst all the groups. Participants within the high-to-low group had the highest age, the highest prevalence of hypertension, diabetes, and the highest HbA1c levels amongst all the groups (Table 2).
Table 2.
Baseline characteristics of Jackson Heart Study participants stratified by change in hsCRP by high (≥ 2 mg/L) and low (<2 mg/L) levels between visit 1 and visit 3
| Change in hsCRP level between visit 1 and visit 3 | p-value | ||||
|---|---|---|---|---|---|
| Low-to-low | Low-to-high | High-to-low | High-to-high | ||
| Participants, n (%) | 885 (29.6) | 394 (13.2) | 280 (9.4) | 1426 (47.8) | |
| Mean age, y | 54±12 | 51±13 | 57±12 | 53±12 | <0.001 |
| Women, n (%) | 413 (46.7) | 213 (54.1) | 184 (65.7) | 1078 (75.6) | <0.001 |
| Hypertension status, n (%) | 396 (44.7) | 176 (44.7) | 167 (59.6) | 782 (54.8) | <0.001 |
| Diabetes status, n (%) | 114 (12.9) | 42 (10.7) | 63 (22.5) | 247 (17.3) | <0.001 |
| Visit 1 median hsCRP, mg/L | 0.69 [0.39 – 1.18] | 1.24 [0.85 – 1.61] | 3.47 [2.54 – 5.02] | 5.47 [3.52 – 9.07] | <0.001 |
| Visit 1 geometric mean hsCRP (95% CI), mg/L | 0.53 (0.49 to 0.56) | 1.01 (0.89 to 1.15) | 7.32 (6.59 to 8.13) | 13.81 (13.07 to 14.59) | <0.001 |
| Visit 3 median hsCRP, mg/L | 0.84 [0.53 – 1.28] | 3.39 [2.44 – 5.12] | 1.22 [0.89 – 1.61] | 6.06 [3.74 – 11.44] | <0.001 |
| Visit 3 geometric mean hsCRP (95% CI), mg/L | 0.73 (0.69 to 0.77) | 7.00 (6.45 to 7.59) | 1.18 (1.09 to 1.28) | 15.95 (15.07 to 16.88) | <0.001 |
| Mean systolic blood pressure, mmHg | 125±15 | 125±16 | 127±16 | 127±16 | 0.022 |
| Mean diastolic blood pressure, mmHg | 76.3±8.4 | 76.7±8.9 | 75.1±8.5 | 75.9±8.3 | 0.07 |
| Mean HbA1c level, % | 5.6±0.9 | 5.7±1.0 | 6.0±1.3 | 5.9±1.1 | <0.001 |
| Mean eGFR, mL/min | 96±18 | 98±20 | 93±21 | 98±19 | <0.001 |
| Mean BMI, kg/m 2 | 28.3±5.0 | 30.1±5.7 | 31.4±6.2 | 34.3±7.6 | <0.001 |
| Cholesterol to HDL-C ratio | 4.1±1.3 | 4.1±1.2 | 4.2±1.5 | 4.1±1.3 | |
| Antihypertensive medication use, n (%) | 359 (40.6) | 161 (40.9) | 159 (56.8) | 727 (51.0) | <0.001 |
| Antidiabetic medication use, n (%) | 71 (8.0) | 29 (7.4) | 36 (12.9) | 145 (10.2) | 0.032 |
| Statin medication use, n (%) | 123 (13.9) | 45 (11.4) | 35 (12.5) | 140 (9.8) | 0.026 |
| Physical activity classification, n (%) | <0.001 | ||||
| Poor Health | 364 (41.1) | 167 (42.4) | 137 (48.9) | 704 (49.4) | |
| Intermediate Health | 281 (31.8) | 147 (37.3) | 86 (30.7) | 470 (33.0) | |
| Ideal Health | 240 (27.1) | 80 (20.3) | 57 (20.4) | 252 (17.7) | |
| Current smoker, n (%) | 82 (9.3) | 44 (11.2) | 27 (9.6) | 178 (12.5) | 0.09 |
| Prevalent CHD, n (%) | 38 (4.3) | 22 (5.6) | 14 (5.0) | 47 (3.3) | 0.15 |
BMI= body mass index; CHD= coronary heart disease; CI= confidence interval; eGFR= estimated glomerular filtration rate; HDL-C= high density lipoprotein cholesterol; HbA1c= glycated hemoglobin A1c; hsCRP= high sensitivity C-reactive protein
hsCRP distribution
Participants who developed HF had greater geometric mean hsCRP levels at all visits compared with participants who did not develop HF during the study. The participants who developed HF during the study had a greater rising trend in geometric mean hsCRP from visit 1 to visit 2 (4.7 to 6.6 mg/L) compared with participants who did not develop HF (3.4 to 4.3 mg/L). However, from visit 2 to visit 3, geometric mean hsCRP levels in participants who later developed HF had a decreasing trend (6.6 to 5.6 mg/L), while participants who did not develop HF continued to have an increasing trend (4.3 to 4.5 mg/L; Figure 1).
Figure 1.
hsCRP levels of the participants at A) visit number B) years from baseline.
Change in hsCRP at high and low levels between visit 1 and visit 3 was divided into 4 groups. Participants within the high-to-high group had the highest geometric mean hsCRP at visit 1 and visit 3 with a level of 13.8 and 16.0 mg/L, respectively. Participants in the high-to-low group started at a geometric mean hsCRP level of 7.3 mg/L at visit 1 and decreased to a level of 1.2 mg/L at visit 3. Whereas, participants in the low-to-high group started at a geometric mean hsCRP level of 1.0 mg/L at visit 1 and increased to a level of 7.0 mg/L at visit 3 (Table 2).
Association of hsCRP with echocardiographic parameters
In unadjusted models, we observed that higher hsCRP levels were associated with increased LVM, LAD, LVEF, and IVST as well as decreased LVMI. These associations remained significant in our minimally adjusted model (age and sex) for LVM, LVMI, LAD, and IVST. However, these associations were attenuated to a non-significant difference for all echocardiographic parameters in our fully adjusted models (Table 3).
Table 3.
Association of hsCRP with markers of cardiac structure and function
| Echocardiographic parameter | β coefficient (95% confidence interval)* | |||||
|---|---|---|---|---|---|---|
| Unadjusted | p-value | Model 1 | p-value | Model 2 | p-value | |
| Left ventricular mass (LVM), g | 0.06 (0.03 to 0.10) | <0.001 | 0.11 (0.08 to 0.15) | <0.001 | −0.02 (−0.05 to 0.02) | 0.31 |
| Left ventricular mass index (LVMI) | −0.18 (−0.23 to −0.14) | <0.001 | −0.31 (−0.35 to −0.27) | <0.001 | 0.03 (−0.03 to 0.09) | 0.28 |
| Diastolic left ventricular diameter (LVID), mm | −0.19 (−0.48 to 0.09) | 0.19 | −0.04 (−0.31 to 0.23) | 0.76 | −0.22 (−0.48 to 0.04) | 0.09 |
| Systolic left ventricular diameter (LVIS), mm | −0.02 (−0.14 to 0.10) | 0.78 | 0.08 (−0.04 to 0.19) | 0.19 | 0.07 (−0.04 to 0.18) | 0.20 |
| Posterior wall thickness (PWT), mm | 1.07 (−0.33 to 2.48) | 0.14 | 1.19 (−0.14 to 2.52) | 0.08 | 0.57 (−0.71 to 1.84) | 0.38 |
| Left atrial diameter (LAD), mm | 0.10 (0.02 to 0.19) | 0.020 | 0.09 (0.01 to 0.17) | 0.03 | 0.01 (−0.07 to 0.09) | 0.85 |
| Left ventricular ejection fraction (LVEF), % | 0.09 (0.03 to 0.14) | 0.002 | 0.01 (−0.04 to 0.07) | 0.60 | 0.01 (−0.04 to 0.06) | 0.76 |
| Relative wall thickness (RWT) | −18.83 (−50.84 to 13.19) | 0.25 | −1.60 (−4.63 to 1.44) | 0.30 | −14.15 (−43.07 to 14.78) | 0.34 |
| Interventricular septal wall thickness (IVST), mm | 0.63 (0.12 to 1.13) | 0.015 | 0.86 (0.38 to 1.34) | <0.001 | 0.20 (−0.26 to 0.67) | 0.39 |
The β coefficients in the table have been scaled per 10 units increase
Model 1= Age + sex
Model 2= Model 1 + systolic blood pressure, antihypertensive medication, BMI (body mass index), diabetes status, eGFR (estimated GFR), smoking history, CHD (coronary heart disease) history, cholesterol-HDL-C ratio, and statin medication
Association of hsCRP levels with incident heart failure
Over a median follow-up of 12.0 years, 308 (7.9%) participants were hospitalized for incident HF. Of which, 131 had HFrEF and 139 had HFpEF.
A single baseline hsCRP measurement was not associated with incident HF (overall, HFrEF, or HFpEF) in fully adjusted models. However, a single baseline measurement of hsCRP was associated with incident HF in unadjusted and minimally adjusted models (Table 4). This attenuation was largely driven by BMI, SBP, eGFR, diabetes, smoking status, CHD, and antihypertensive medications.
Table 4.
Risk of incident heart failure with increasing levels of hsCRP on a continuous scale
| Baseline hsCRP | Repeated measures of hsCRP | |||
|---|---|---|---|---|
| Hazard ratio (95% confidence interval) | p-value | Hazard ratio (95% confidence interval) | p-value | |
| Overall HF | ||||
| Unadjusted | 1.18 (1.06 to 1.32) | 0.003 | 1.24 (1.07 to 1.44) | 0.004 |
| Model 1 | 1.21 (1.07 to 1.36) | 0.002 | 1.4 (1.20 to 1.64) | <0.001 |
| Model 2 | 1.08 (0.96 to 1.22) | 0.21 | 1.23 (1.03 to 1.45) | 0.020 |
| Model 3 | − | − | 1.22 (1.03 to 1.44) | 0.018 |
| HFrEF | ||||
| Unadjusted | 1.04 (0.88 to 1.22) | 0.65 | 1.05 (0.84 to 1.32) | 0.66 |
| Model 1 | 1.10 (0.92 to 1.32) | 0.29 | 1.22 (0.96 to 1.56) | 0.10 |
| Model 2 | 1.06 (0.87 to 1.28) | 0.58 | 1.18 (0.9 to 1.54) | 0.24 |
| Model 3 | − | − | 1.18 (0.91 to 1.53) | 0.21 |
| HFpEF | ||||
| Unadjusted | 1.38 (1.17 to 1.62) | <0.001 | 1.45 (1.16 to 1.81) | 0.001 |
| Model 1 | 1.38 (1.16 to 1.65) | <0.001 | 1.60 (1.26 to 2.03) | <0.001 |
| Model 2 | 1.16 (0.97 to 1.4) | 0.10 | 1.33 (1.03 to 1.71) | 0.028 |
| Model 3 | − | − | 1.30 (1.02 to 1.65) | 0.036 |
Model 1= Age + sex
Model 2= Model 1 + systolic blood pressure, antihypertensive medication, BMI, diabetes status, eGFR, smoking history, CHD history, cholesterol-HDL-C ratio, and statin medication
Model 3= Model 2 + random slope (to account for trajectory of change)
The above values are presented as adjusted hazard ratios (HR) per 1 standard deviation (SD) increase of the hsCRP level (on a log2 scale)
HF = Heart failure; HFpEF = Heart failure preserved ejection fraction; HFrEF = Heart failure reduced ejection fraction; hsCRP = high sensitivity C-reactive protein
Repeated measures of hsCRP (visits 1 through 3), incorporating the trajectory of hsCRP were associated with an increased risk of incident HF overall (HR 1.22; 95%CI: 1.03 to 1.44) and HFpEF (HR 1.30; 95%CI: 1.02 to 1.65). However, there was no significant association between repeated hsCRP levels and incident HFrEF in any model (Table 4).
With the incorporation of random slope in model 3, the Akaike and Bayesian information criterion of all joint models decreased, confidence intervals narrowed, and the likelihood ratio test demonstrated a significant difference in the repeated measures of hsCRP with and without slope. These findings suggest a better fit for joint models with incorporation of slope.
Association of a baseline hsCRP ≥2 mg/L with incident HF
A high hsCRP level (≥2 mg/L) was not associated with incident HF in fully adjusted models. However, our unadjusted model had a lower 95% CI of 0.99 and our minimally adjusted model had a significant association with incident HF (HR 1.28; 95%CI: 1.01 to 1.62). Similarly, we observed no significant association of high hsCRP levels with incident HFrEF or HFpEF in fully adjusted models (Table 5).
Table 5.
Risk of incident HF with a high hsCRP (≥2 mg/L) at baseline visit
| Hazard ratio (95% confidence interval) | p-value | |
|---|---|---|
| Overall HF | ||
| Unadjusted | 1.24 (0.99 to 1.56) | 0.07 |
| Model 1 | 1.28 (1.01 to 1.62) | 0.038 |
| Model 2 | 1.08 (0.84 to 1.38) | 0.55 |
| HFrEF | ||
| Unadjusted | 1.01 (0.71 to 1.42) | 0.98 |
| Model 1 | 1.14 (0.80 to 1.62) | 0.48 |
| Model 2 | 1.07 (0.74 to 1.55) | 0.72 |
| HFpEF | ||
| Unadjusted | 1.62 (1.13 to 2.32) | 0.008 |
| Model 1 | 1.59 (1.10 to 2.29) | 0.013 |
| Model 2 | 1.22 (0.84 to 1.79) | 0.30 |
Model 1= Age + sex
Model 2= Model 1 + systolic blood pressure, antihypertensive medication, BMI, diabetes status, eGFR, smoking history, CHD history, cholesterol-HDL-C ratio, and statin medication
The above values are presented as adjusted hazard ratios (HR) per 1 standard deviation (SD) increase of the hsCRP level (on a log2 scale)
HF = Heart failure; HFpEF = Heart failure preserved ejection fraction; HFrEF = Heart failure reduced ejection fraction; hsCRP = high sensitivity C-reactive protein
Association of change in hsCRP (high and low) and incident HF
Participants amongst the low-to-high group had the highest risk for incident HF (HR 2.12; 95%CI: 1.25 to 3.61) and participants in the high-to-low group (HR 1.78; 95%CI: 1.02 to 3.10) also had an increased risk for incident HF in adjusted models compared with participants in the low-to-low group. Participants in the high-to-high group had no significant association of risk for incident HF in our fully adjusted models but had an increased risk of incident HF in the minimally adjusted model (Table 6 and Figure 2). Furthermore, participants in the low-to-high group were associated with an increased risk of HFrEF (HR 2.34; 1.12 to 4.92), while participants in the high-to-low group were associated with an increased risk of HFpEF (HR 2.44; 1.04 to 5.7) compared with participants in the low-to-low group (Table 5).
Table 6.
Risk of incident HF by change in hsCRP at high (≥2 mg/L) and low levels (<2 mg/L) from visit 1 to visit 3
| Low-to-low | Low-to-high | High-to-low | High-to-high | ||||
|---|---|---|---|---|---|---|---|
| Hazard ratio (95% confidence interval) | p-value | Hazard ratio (95% confidence interval) | p-value | Hazard ratio (95% confidence interval) | p-value | ||
| Overall HF | |||||||
| Unadjusted | Ref. | 1.80 (1.07 to 3.04) | 0.028 | 2.17 (1.25 to 3.77) | 0.006 | 1.33 (0.87 to 2.02) | 0.19 |
| Model 1 | Ref. | 2.21 (1.31 to 3.74) | 0.003 | 1.91 (1.10 to 3.32) | 0.022 | 1.60 (1.04 to 2.45) | 0.034 |
| Model 2 | Ref. | 2.12 (1.25 to 3.61) | 0.005 | 1.78 (1.02 to 3.10) | 0.044 | 1.41 (0.89 to 2.21) | 0.14 |
| HFrEF | |||||||
| Unadjusted | Ref. | 1.86 (0.90 to 3.88) | 0.10 | 1.86 (0.82 to 4.20) | 0.14 | 0.86 (0.45 to 1.63) | 0.64 |
| Model 1 | Ref. | 2.30 (1.10 to 4.79) | 0.026 | 1.80 (0.79 to 4.09) | 0.16 | 1.18 (0.61 to 2.29) | 0.62 |
| Model 2 | Ref. | 2.34 (1.12 to 4.92) | 0.025 | 1.84 (0.8 to 4.22) | 0.15 | 1.24 (0.62 to 2.47) | 0.55 |
| HFpEF | |||||||
| Unadjusted | Ref. | 1.68 (0.68 to 4.18) | 0.26 | 3.32 (1.44 to 7.65) | 0.005 | 1.98 (1.01 to 3.91) | 0.048 |
| Model 1 | Ref. | 2.06 (0.83 to 5.12) | 0.12 | 2.84 (1.23 to 6.59) | 0.015 | 2.29 (1.15 to 4.58) | 0.019 |
| Model 2 | Ref. | 1.96 (0.78 to 4.90) | 0.15 | 2.44 (1.04 to 5.7) | 0.040 | 1.78 (0.87 to 3.67) | 0.12 |
Model 1= Age + sex
Model 2= Model 1 + systolic blood pressure, antihypertensive medication, BMI, diabetes status, eGFR, smoking history, CHD history, cholesterol-HDL-C ratio, and statin medication
The above values are presented as adjusted hazard ratios (HR) per 1 standard deviation (SD) increase of the hsCRP level (on a log2 scale)
HF = Heart failure; HFpEF = Heart failure preserved ejection fraction; HFrEF = Heart failure reduced ejection fraction; hsCRP = high sensitivity C-reactive protein
Figure 2.
Kaplan Meier survival plots demonstrating association of participants changing in high (≥2 mg/L) and low levels (<2 mg/L) of hsCRP with incident HF hospitalization. HF = heart failure; hsCRP= high sensitivity C-reactive protein.
Sensitivity analysis
In our first sensitivity analysis we excluded participants with prevalent or incident CHD and found repeated measures of hsCRP were still associated with an increased risk (HR 1.26; 95%CI: 1.05 to 1.51) of incident HF.
In our second sensitivity analysis we adjusted for social determinants of health. To include these variables, as many were missing, our analytic sample size was decreased to 3320 participants who encountered 259 HF events leading to a sample size reduction of 600 participants and 49 HF events. After exclusion of these participants, we observed all repeated measures models (unadjusted, model 1, and model 2) had a decrease in HR. We observed an association with increased risk of incident HF in the minimally adjusted model (HR 1.33; 95%CI: 1.12 to 1.58). However, with inclusion of insurance, education level, and income variables we found that repeated measures of hsCRP were no longer associated with incident HF (HR 1.13; 95%CI: 0.94 to 1.36).
Discussion
In this study, we demonstrate the association of repeated hsCRP measurements with incident HF hospitalization in a large community-based cohort of Black adults. We demonstrate that single measurements of baseline hsCRP are not associated with incident HF in Black adults. However, repeated measurements of hsCRP can predict the risk of incident HF. This association is primarily evident for incident HFpEF. Furthermore, an increasing trend in hsCRP (while starting at low levels) is a risk profile associated with the highest risk of incident HF.
The Atherosclerosis Risk in Communities (ARIC) study,8 Malmö Diet and Cancer study,4 Cardiovascular Health Study (CHS),5 and other cohorts7,10 have demonstrated an association of single CRP and hsCRP levels at baseline with incident HF. Our study, as well as the Strong Heart Study (SHS),9 Health ABC Study,10 and Prevention of Renal and Vascular End-stage Disease (PREVEND)11 have demonstrated no association of single CRP and hsCRP levels at baseline with incident HF. One possible explanation for variation in findings may be secondary to a predominate Black population. Black adults have a disproportionately high risk of incident HF and HF hospitalization but are under-represented in population-based study cohorts.12 While the proportion of Black adults in the ARIC study and multicohort studies ranged between 12–31%, these are still far lower than our study and the Health ABC Study which had similar findings.10 The Health ABC Study also assessed change in CRP at 1 year follow-up but did not find an association with incident HF. In comparison, our data utilized hsCRP values up to 13 years apart and a joint modelling analysis which incorporates the slope of change in hsCRP to uncover the role of variation of hsCRP and incident HF.
Racial-ethnic differences in patterns of inflammation and markers such as hsCRP may have also played a role explaining the variation in our findings. Black adults have higher hsCRP levels and interleukin-6 levels compared with White adults.26, 27 This difference in levels of inflammation may partly be explained by genetic variation, the TREM locus is involved in the regulation of proinflammatory cytokines. African ancestry in Black adults is associated TREM locus alleles which lead to higher levels of CRP.28
Another possible explanation is the high proportion of obesity in our population with an average BMI of 31.6 kg/m2. As demonstrated in the PREVEND study, BMI has a significant association with CRP and inflammation.11 We observed significant attenuation in HR and β-coefficients with adjustment for BMI, and as result minimally adjusted models for the single hsCRP analysis and hsCRP ≥2 mg/L analysis demonstrated a significant association, but no longer demonstrated any significant association after full model adjustment (includes BMI). Our findings are supported by the SHS which also had a majority of patients with obesity.9 To adjust for patient-level baseline hsCRP levels, our results suggest that hsCRP trends have greater utility for HF risk stratification in this high-risk group while single baseline measurements may not.
The relationship between HF and inflammation is complex and bi-directional. Inflammation causes HF and is a consequence of hemodynamic stress from HF. There are several proposed mechanisms by which inflammation may contribute to the pathogenesis of HF through proinflammatory cytokines (interleukin-1, interleukin-6, and tumor necrosis factor-alpha), activation of the innate and humoral immune responses, as well as pro-inflammatory monocyte release from the spleen and adipose tissue.29 In turn, endothelial inflammation leads to release of reactive oxygen species which results in reduced soluble guanylate cyclase activity to cause cardiomyocyte stiffness and hypertrophy. Inflammatory cytokine release also leads to cell hypoxemia and neuroendocrine activation which lead to myocardial apoptosis and reduced ventricular wall compliance.30 Altogether, these changes lead to LV diastolic dysfunction and HF.
Participants with low levels of hsCRP initially, but later high levels had a greater than 2-fold increased risk for incident HF. Similarly, participants in the low-to-high group also had an increased risk of incident HF. However, participants in the high-to-high group had a higher risk of incident HF which did not achieve statistical significance. The participants in the high-to-high group had the highest BMI. This indicates that their stable high levels of hsCRP may be driven largely by their obesity. The findings of our change analysis demonstrated participants of the low-to-high group had an increased risk of HFrEF while those in the high-to-low group had an increased risk of HFpEF. The HFrEF and HFpEF analysis in these two groups must be interpreted with caution due to lower events given 4 group subdivision (by inflammation) and further subdivision (by HF type) as observed by the wide confidence intervals.
Our findings suggest that people whose true baseline hsCRP is a low level (<2 mg/L) but experience variation from this level (to high levels) are representative of inflammatory damage and risk of incident HF. Increasing trends (low-to-high) may be reflective of worsening stages of chronic diseases associated with inflammation (obesity, diabetes, hypertension, and chronic kidney disease) which impair skeletal muscle oxygen extraction, worsen anemia, and promote sodium retention leading to dyspnea which symptomatically trigger an incident HF event.31 Subclinical ischemia from microvascular dysfunction and subclinical CHD may also lead to inflammation. Whereas decreasing trends (high-to-low) may indicate inflammatory recovery after an inflammatory insult where a person’s baseline lies in a low inflammatory state but retains some damage from the insult. This group may also reflect variation in lifestyle, which has a major role in regulation of inflammation with increased physical activity and healthy eating habits demonstrating anti-inflammatory effects leading to lower levels of hsCRP.32,33 These data highlight the importance of personalized inflammatory biomarker trends in the prediction of HF.
With inclusion of social determinants of health, our sensitivity analysis also demonstrated an increased risk of incident HF, however this was not statistically significant. Data on social determinants of health was not available for a large number of participants- 15.3% participants and 15.9% HF events of our originally included population did not have data available. After excluding these participants, we observed a decrease in the observed HR of all models. The loss of these participants and events likely led to loss of statistical significance.
CRP levels have also been associated with higher LVMI and RWT.34,35 While our study did not demonstrate any significant association with increased LVMI or RWT, we observed large negative shifts in β-coefficients for LV mass and IVST with adjustment for BMI and our full adjusted model. This is likely secondary to the high levels of obesity in our population.29
While few analyses on CRP and HF subtypes exist, prior studies demonstrate associations of CRP with HFrEF but not HFpEF or no association with either.7,8 Contrary to these findings, we found that hsCRP is consistently associated with HFpEF when hsCRP trends are assessed as observed in our joint model and change analysis.
hsCRP is variable and is subject to environmental factors.3 Through our joint modelling analysis, we leveraged the benefit of using multiple values to account and subsequently adjust for variability. Our model allows us to predict hsCRP values prior to an incident HF event and assess its association with incident HF events. Novel findings from our study include a 22% higher risk for incident HF and a 30% higher risk of incident HFpEF for each SD increase in hsCRP on a log2 scale. We also demonstrated that participants who have low baseline hsCRP levels but experience an increase in hsCRP to high levels over time, have a more than 2-fold risk of incident HF. These findings highlight the highly variable nature of hsCRP, the importance of personalized inflammatory biomarker baselines, and the value of repeated measures on follow-up visits to risk stratify patients who may be at increasing risk for HF.
Our study is subject to several limitations. First, JHS is a cohort of only Black adults residing within the Jackson, Mississippi metropolitan area; thus, the findings may not be generalizable to other racial/ethnic groups or geographic regions. Secondly, ascertainment and adjudication of incident HF events started in 2005, i.e. 1–5 years after visit 1 hsCRP measurements (2000–2004) and continued through 2016. Therefore, time differences of performed hsCRP measurements may limit the causal inference of the effect of hsCRP on HF hospitalization.
Conclusion
In this large cohort of Black adults, repeated measures of hsCRP and hsCRP trajectories were associated with an increased risk of incident HF and were more predictive than a single hsCRP measurement. This association is particularly evident for HFpEF. Furthermore, People who start at a low hsCRP level and experience increases to high levels (≥2 mg/L) have the highest risk for incident HF. Our findings highlight that multiple measurements of hsCRP obtained over follow-up clinic visits may help better risk stratify patients in need of close follow-up and medical optimization to prevent the development of HF.
Supplementary Material
Figure S1. Exclusion criteria and number of participants. HF= heart failure, hsCRP = high sensitivity C-reactive protein.
Table S1. Baseline characteristics of Jackson Heart Study participants included or excluded in the study.
Clinical Perspectives.
What is New?
Inflammation is a variable state. There have been mixed findings regarding the association of a single high sensitivity c-reactive protein (hsCRP) measurement and risk of incident heart failure (HF).
This study demonstrates that multiple measurements of hsCRP over time while accounting for trajectory of change are more predictive for incident HF than an isolated single measurement of hsCRP in Black adults and that this association is primarily evident for HFpEF.
What are the clinical implications?
Patients at risk of HF should have hsCRP measurements assessed over time in the outpatient setting to risk stratify them for their risk of incident HF.
A single hsCRP ≥2 mg/L is not predictive on its own for a Black adult’s risk of incident HF, but Black adults who start at levels <2 mg/L and experience increases in hsCRP to ≥2 mg/L carry a greater than 2-fold higher risk of incident HF.
Future studies of anti-inflammatory therapies are needed to determine their impact in primary prevention of HF.
Sources of funding
The Jackson Heart Study is supported and conducted in collaboration with Jackson State University (HHSN268201800013I), Tougaloo College (HHSN268201800014I), the Mississippi State Department of Health (HHSN268201800015I) and the University of Mississippi Medical Center (HHSN268201800010I, HHSN268201800011I and HHSN268201800012I) contracts from the National Heart, Lung, and Blood Institute and the National Institute for Minority Health and Health Disparities. Dr. Michael E. Hall is funded by support from the National Institute of General Medical Sciences 5U54GM115428. Dr. Shah was supported by NIH/NHLBI grants R01HL135008, R01HL143224, R01HL150342, R01HL148218 and K24HL152008.
The authors would like to thank the participants of the Jackson Heart Study for their time and participation.
Abbreviations
- AHA
American Heart Association
- BMI
Body mass index
- CI
Confidence interval
- CHD
Coronary heart disease
- CRP
C-reactive protein
- CV
Cardiovascular
- CVD
Cardiovascular disease
- eGFR
Estimated glomerular filtration rate
- hsCRP
High sensitivity C-reactive protein
- HF
Heart failure
- HFpEF
Heart failure with preserved ejection fraction
- HFrEF
Heart failure with reduced ejection fraction
- HR
Hazard ratio
- LAD
Left atrial diameter
- LV
Left ventricle
- LVID
Left ventricular internal diameter
- LVM
Left ventricular mass
- LVEF
Left ventricular ejection fraction
- PWT
Posterior wall thickness
- RWT
Relative wall thickness
- SBP
Systolic blood pressure
Footnotes
Disclosures
Dr. Robert J. Mentz received research support and honoraria from Abbott, American Regent, Amgen, AstraZeneca, Bayer, Boehringer Ingelheim/Eli Lilly, Boston Scientific, Cytokinetics, Fast BioMedical, Gilead, Innolife, Medtronic, Merck, Novartis, Relypsa, Respicardia, Roche, Sanofi, Vifor, Windtree Therapeutics, and Zoll. Dr. Shah reports research support not related to this study from Novartis and Philips Ultrasound, and consulting fees from Philips Ultrasound. Dr. Virani receives research support from the Department of Veterans Affairs, National Institutes of Health, and Tahir and Jooma Family. Dr. Virani has also received honorarium from the American College of Cardiology in his role as the Associate Editor for Innovations, acc.org. Dr Butler reported personal fees from Abbott, Adrenomed, Amgen, Array, AstraZeneca, Bayer, Boehringer Ingelheim, Bristol Myers Squibb, CVRx, G3 Pharmaceutical, Impulse Dynamics, Innolife, Janssen, LivaNova, Luitpold, Medtronic, Merck, Novartis, Novo Nordisk, Relypsa, Roche, V-Wave Limited, and Vifor outside the submitted work. All other authors have no relevant disclosures.
Disclaimer
The views expressed in this manuscript are those of the authors and do not necessarily represent the views of the National Heart, Lung, and Blood Institute; the National Institutes of Health; the U.S. Department of Health and Human Services; or the Department of Veterans Affairs.
References
- 1.Sproston NR, Ashworth JJ. Role of C-Reactive Protein at Sites of Inflammation and Infection. Front Immunol. 2018;9:754. doi: 10.3389/fimmu.2018.00754 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Ridker PM, Rifai N, Rose L, Buring JE, Cook NR. Comparison of C-reactive protein and low-density lipoprotein cholesterol levels in the prediction of first cardiovascular events. N Engl J Med. 2002;347:1557–1565. doi: 10.1056/NEJMoa021993 [DOI] [PubMed] [Google Scholar]
- 3.Pearson TA, Mensah GA, Alexander RW, Anderson JL, Cannon RO 3rd, Criqui M, Fadl YY, Fortmann SP, Hong Y, Myers GL, et al. Markers of inflammation and cardiovascular disease: application to clinical and public health practice: A statement for healthcare professionals from the Centers for Disease Control and Prevention and the American Heart Association. Circulation. 2003;107:499–511. doi: 10.1161/01.cir.0000052939.59093.45 [DOI] [PubMed] [Google Scholar]
- 4.Smith JG, Newton-Cheh C, Almgren P, Struck J, Morgenthaler NG, Bergmann A, Platonov PG, Hedblad B, Engstrom G, Wang TJ, et al. Assessment of conventional cardiovascular risk factors and multiple biomarkers for the prediction of incident heart failure and atrial fibrillation. J Am Coll Cardiol. 2010;56:1712–1719. doi: 10.1016/j.jacc.2010.05.049 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Suzuki T, Katz R, Jenny NS, Zakai NA, LeWinter MM, Barzilay JI, Cushman M. Metabolic syndrome, inflammation, and incident heart failure in the elderly: the cardiovascular health study. Circ Heart Fail. 2008;1:242–248. doi: 10.1161/CIRCHEARTFAILURE.108.785485 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Maio R, Perticone M, Suraci E, Sciacqua A, Sesti G, Perticone F. Endothelial dysfunction and C-reactive protein predict the incidence of heart failure in hypertensive patients. ESC Heart Fail. 2021;8:399–407. doi: 10.1002/ehf2.13088 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.de Boer RA, Nayor M, deFilippi CR, Enserro D, Bhambhani V, Kizer JR, Blaha MJ, Brouwers FP, Cushman M, Lima JAC, et al. Association of Cardiovascular Biomarkers With Incident Heart Failure With Preserved and Reduced Ejection Fraction. JAMA Cardiol. 2018;3:215–224. doi: 10.1001/jamacardio.2017.4987 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Cohen AJ, Teramoto K, Claggett B, Buckley L Jr., Solomon S, Ballantyne C, Selvin E, Shah AM. Mid- to Late-Life Inflammation and Risk of Cardiac Dysfunction, HFpEF and HFrEF in Late Life. J Card Fail. 2021;27:1382–1392. doi: 10.1016/j.cardfail.2021.07.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Barac A, Wang H, Shara NM, de Simone G, Carter EA, Umans JG, Best LG, Yeh J, Dixon DB, Devereux RB, et al. Markers of inflammation, metabolic risk factors, and incident heart failure in American Indians: the Strong Heart Study. J Clin Hypertens (Greenwich). 2012;14:13–19. doi: 10.1111/j.1751-7176.2011.00560.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kalogeropoulos A, Georgiopoulou V, Psaty BM, Rodondi N, Smith AL, Harrison DG, Liu Y, Hoffmann U, Bauer DC, Newman AB, et al. Inflammatory markers and incident heart failure risk in older adults: the Health ABC (Health, Aging, and Body Composition) study. J Am Coll Cardiol. 2010;55:2129–2137. doi: 10.1016/j.jacc.2009.12.045 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Suthahar N, Meems LMG, Groothof D, Bakker SJL, Gansevoort RT, van Veldhuisen DJ, de Boer RA. Relationship between body mass index, cardiovascular biomarkers and incident heart failure. Eur J Heart Fail. 2021;23:396–402. doi: 10.1002/ejhf.2102 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Ziaeian B, Kominski GF, Ong MK, Mays VM, Brook RH, Fonarow GC. National Differences in Trends for Heart Failure Hospitalizations by Sex and Race/Ethnicity. Circ Cardiovasc Qual Outcomes. 2017;10. doi: 10.1161/CIRCOUTCOMES.116.003552 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Metti AL, Yaffe K, Boudreau RM, Simonsick EM, Carnahan RM, Satterfield S, Harris TB, Ayonayon HN, Rosano C, Cauley JA, Health ABCS. Trajectories of inflammatory markers and cognitive decline over 10 years. Neurobiol Aging. 2014;35:2785–2790. doi: 10.1016/j.neurobiolaging.2014.05.030 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kamimura D, Cain LR, Mentz RJ, White WB, Blaha MJ, DeFilippis AP, Fox ER, Rodriguez CJ, Keith RJ, Benjamin EJ, et al. Cigarette Smoking and Incident Heart Failure: Insights From the Jackson Heart Study. Circulation. 2018;137:2572–2582. doi: 10.1161/CIRCULATIONAHA.117.031912 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Taylor HA Jr., Wilson JG, Jones DW, Sarpong DF, Srinivasan A, Garrison RJ, Nelson C, Wyatt SB. Toward resolution of cardiovascular health disparities in African Americans: design and methods of the Jackson Heart Study. Ethn Dis. 2005;15:S6–4–17. [PubMed] [Google Scholar]
- 16.Payne TJ, Wyatt SB, Mosley TH, Dubbert PM, Guiterrez-Mohammed ML, Calvin RL, Taylor HA Jr., Williams DR. Sociocultural methods in the Jackson Heart Study: conceptual and descriptive overview. Ethn Dis. 2005;15:S6–38–48. [PubMed] [Google Scholar]
- 17.JHS data access. Jackson Heart Study. https://www.jacksonheartstudy.org/Research/Study-Data/Data-Access. Accessed 1/21/2020.
- 18.Carpenter MA, Crow R, Steffes M, Rock W, Heilbraun J, Evans G, Skelton T, Jensen R, Sarpong D. Laboratory, reading center, and coordinating center data management methods in the Jackson Heart Study. Am J Med Sci. 2004;328:131–144. doi: 10.1097/00000441-200409000-00001 [DOI] [PubMed] [Google Scholar]
- 19.Fox ER, Benjamin EJ, Sarpong DF, Rotimi CN, Wilson JG, Steffes MW, Chen G, Adeyemo A, Taylor JK, Samdarshi TE, et al. Epidemiology, heritability, and genetic linkage of C-reactive protein in African Americans (from the Jackson Heart Study). Am J Cardiol. 2008;102:835–841. doi: 10.1016/j.amjcard.2008.05.049 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Fuqua SR, Wyatt SB, Andrew ME, Sarpong DF, Henderson FR, Cunningham MF, Taylor HA Jr. Recruiting African-American research participation in the Jackson Heart Study: methods, response rates, and sample description. Ethn Dis. 2005;15:S6–18–29. [PubMed] [Google Scholar]
- 21.Mentz RJ, Greiner MA, DeVore AD, Dunlay SM, Choudhary G, Ahmad T, Khazanie P, Randolph TC, Griswold ME, Eapen ZJ, et al. Ventricular conduction and long-term heart failure outcomes and mortality in African Americans: insights from the Jackson Heart Study. Circ Heart Fail. 2015;8:243–251. doi: 10.1161/CIRCHEARTFAILURE.114.001729 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Keku E, Rosamond W, Taylor HA Jr., Garrison R, Wyatt SB, Richard M, Jenkins B, Reeves L, Sarpong D. Cardiovascular disease event classification in the Jackson Heart Study: methods and procedures. Ethn Dis. 2005;15:S6–62–70. [PubMed] [Google Scholar]
- 23.Patel KV, Simek S, Ayers C, Neeland IJ, deFilippi C, Seliger SL, Lonergan K, Minniefield N, Mentz RJ, Correa A, et al. Physical Activity, Subclinical Myocardial Injury, and Risk of Heart Failure Subtypes in Black Adults. JACC Heart Fail. 2021;9:484–493. doi: 10.1016/j.jchf.2021.04.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Baggen VJM, Baart SJ, van den Bosch AE, Eindhoven JA, Witsenburg M, Cuypers J, Roos-Hesselink JW, Boersma E. Prognostic Value of Serial N-Terminal Pro-B-Type Natriuretic Peptide Measurements in Adults With Congenital Heart Disease. J Am Heart Assoc. 2018;7. doi: 10.1161/JAHA.117.008349 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Ridker PM, Everett BM, Thuren T, MacFadyen JG, Chang WH, Ballantyne C, Fonseca F, Nicolau J, Koenig W, Anker SD, et al. Antiinflammatory Therapy with Canakinumab for Atherosclerotic Disease. New England Journal of Medicine. 2017;377:1119–1131. doi: 10.1056/NEJMoa1707914 [DOI] [PubMed] [Google Scholar]
- 26.Amit K, Darren KM, Sabina AM, Harold GS, Sandeep RD, Wanpen V, Frank HW, Scott MG, James A. Race and Gender Differences in C-Reactive Protein Levels. Journal of the American College of Cardiology. 2005;46:464–469. doi: 10.1016/j.jacc.2005.04.051 [DOI] [PubMed] [Google Scholar]
- 27.Paalani M, Lee JW, Haddad E, Tonstad S. Determinants of inflammatory markers in a bi-ethnic population. Ethn Dis. 2011;21:142–149. [PMC free article] [PubMed] [Google Scholar]
- 28.Reiner AP, Beleza S, Franceschini N, Auer PL, Robinson JG, Kooperberg C, Peters U, Tang H. Genome-wide association and population genetic analysis of C-reactive protein in African American and Hispanic American women. American journal of human genetics. 2012;91:502–512. doi: 10.1016/j.ajhg.2012.07.023 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Sean PM, Rahul K, Cian PM, James LJ. Inflammation in Heart Failure: JACC State-of-the-Art Review. Journal of the American College of Cardiology. 2020;75:1324–1340. doi: 10.1016/j.jacc.2020.01.014 [DOI] [PubMed] [Google Scholar]
- 30.Mocan M, Mocan Hognogi LD, Anton FP, Chiorescu RM, Goidescu CM, Stoia MA, Farcas AD. Biomarkers of Inflammation in Left Ventricular Diastolic Dysfunction. Disease markers. 2019;2019:7583690. doi: 10.1155/2019/7583690 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Murphy SP, Kakkar R, McCarthy CP, Januzzi JL Jr. Inflammation in Heart Failure: JACC State-of-the-Art Review. J Am Coll Cardiol. 2020;75:1324–1340. doi: 10.1016/j.jacc.2020.01.014 [DOI] [PubMed] [Google Scholar]
- 32.Gleeson M, Bishop NC, Stensel DJ, Lindley MR, Mastana SS, Nimmo MA. The anti-inflammatory effects of exercise: mechanisms and implications for the prevention and treatment of disease. Nature Reviews Immunology. 2011;11:607–615. doi: 10.1038/nri3041 [DOI] [PubMed] [Google Scholar]
- 33.Graff E, Vedantam S, Parianos M, Khakoo N, Beiling M, Pearlman M. Dietary Intake and Systemic Inflammation: Can We Use Food as Medicine? Current Nutrition Reports. 2023;12:247–254. doi: 10.1007/s13668-023-00458-z [DOI] [PubMed] [Google Scholar]
- 34.Iwashima Y, Horio T, Kamide K, Rakugi H, Ogihara T, Kawano Y. C-reactive protein, left ventricular mass index, and risk of cardiovascular disease in essential hypertension. Hypertens Res. 2007;30:1177–1185. doi: 10.1291/hypres.30.1177 [DOI] [PubMed] [Google Scholar]
- 35.Tsioufis C, Stougiannos P, Kakkavas A, Toutouza M, Mariolis A, Vlasseros I, Stefanadis C, Kallikazaros I. Relation of left ventricular concentric remodeling to levels of C-reactive protein and serum amyloid A in patients with essential hypertension. Am J Cardiol. 2005;96:252–256. doi: 10.1016/j.amjcard.2005.03.054 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1. Exclusion criteria and number of participants. HF= heart failure, hsCRP = high sensitivity C-reactive protein.
Table S1. Baseline characteristics of Jackson Heart Study participants included or excluded in the study.