Elevated Levels of Interleukin 6 and C-Reactive Protein Associated With Cognitive Impairment in Heart Failure


Ponrathi Athilingam, PhD, College of Nursing, University of South Florida, 12901 Bruce B. Downs Boulevard, Tampa, FL
E-mail: pathilin@health.usf.edu


©2012 Wiley Periodicals, Inc.

There is abundant evidence on inflammatory mechanisms in heart failure (HF) that are used for prognostication of the disease; however, data are lacking regarding the association between elevated cytokines, C-reactive protein (CRP), and cognition in HF. A cross-sectional pilot study of 38 patients with HF, aged 62 years (standard deviation± 9 years), predominantly men (68%) and Caucasian (79%) were screened for cognitive function using the Montreal Cognitive Assessment (MoCA). The study aimed to examine cognitive scores on MoCA with cytokines, interleukin 6 [IL-6] and tumor necrosis factor α [TNF-α], and CRP as indicators of early cognitive changes in HF. The result showed no direct correlation between cardiac variables and the MoCA score. The MoCA score, however, was inversely associated with IL-6 (r=−0.53, P=.001) and CRP (r=−0.34, P=.04), with no association to TNF-α. Regression analysis on the MoCA score and log-transformed IL-6 accounted for an additional 11% variation and remained statistically significant (P=.008) after controlling for covariates of education, living arrangements, and loneliness. The large effect size (R2=0.87) found in this pilot study provides rationale for a larger exploratory study to examine associations between cognitive function, cytokines, and CRP levels and help design future intervention studies.

Heart failure (HF) is a progressive disorder and is often the terminal stage of cardiac disease.1 It is estimated that nearly 5.8 million Americans have HF, and the prevalence approximately doubles with each decade of life.2 Importantly, persons with HF have a 4-fold increased risk of cognitive impairment compared with the general population,3 with the prevalence of cognitive impairment ranging from 25% to 70%.4 The odds of dementia in persons with HF were significantly higher in a recent study, even after adjustment for age, education level, net worth, and prior stroke (odds ratio, 1.52; 95% confidence interval [CI], 1.14–2.02).5 Management of HF requires active participation of patients in their self-care, including use of sound decision making.5 Unfortunately, cognitively impaired persons with HF are 30% more likely to have inadequate self-care confidence and self-care ability to recognize and seek prompt medical attention for early HF symptoms, which predisposes to frequent readmissions.7 Cytokines and C-reactive protein (CRP) play a central role in neuroimmmunoendocrine processes. It has been hypothesized that these molecules influence cognition via diverse cytokine-mediated interactions between neurons and glial cells in animal studies.8,9 Inflammatory cytokines, including interleukin (IL) 1, IL-6, and tumor necrosis factor α (TNF-α) and CRP have been implicated in the progression of HF and exert direct toxic effects on the heart that may exacerbate hemodynamic abnormalities.10 Because of multiple organ involvement in HF and altered tissue perfusion, inflammatory biomarkers important in the pathophysiology of HF may be produced both by the myocardium and extramyocardial sources.11,12 Mechanical overload and myocardial shear stress may also induce cytokine expression.13 In addition, cardiac ischemia may potentiate the inflammatory processes.14

Although there is a large body of evidence about cytokines and CRP as prognostic markers in HF,15,16 sparse evidence is available that supports any association between inflammatory biomarkers (cytokines and CRP) and cognitive impairment in HF. Therefore, the purpose of this pilot correlational study was to examine associations between cognitive screening scores of the Montreal Cognitive Assessment (MoCA) and IL-1, IL-6, TNF-α, and CRP and to calculate an effect size for a larger exploratory study. Although several cytokines and biomarkers may be associated with cognitive function, IL-1, IL-6, TNF-α, and CRP were selected as variables in this study on the basis of their significant association to cognitive function in comorbid conditions similar to HF.17

We tested the following hypotheses:

Hypothesis 1: Elevated levels of inflammatory cytokines (IL-1, IL-6, and TNF-α) and CRP will be associated with severity of HF (measured by noninvasive cardiac output, cerebral perfusion pressure, New York Heart Association [NYHA] class, and ejection fraction) and cognitive function (measured by the MoCA).

Hypothesis 2: The MoCA score will be negatively associated with IL-1, IL-6, TNF-α, and CRP in persons with HF after controlling for covariates.


Study Design, Sample, and Inclusion and Exclusion Criteria

A cross-sectional correlational design was used for this pilot study. Forty adults with a clinical diagnosis of HF (as defined by the International Statistical Classification of Diseases and Related Health Problems–Ninth Revision [ICD-9] codes) who were cared for by cardiologists affiliated with a tertiary hospital were enrolled from 90 patients who participated concurrently in a primary cross-sectional study during a period of 4 months.18 Each cytokine kit allowed 40 tests; hence, only 40 participants who consented to provide blood samples for cytokine and CRP were included in this pilot study. Patients were enrolled if they were 50 years or older and NYHA classification II or III. The age inclusion criteria was chosen because the incidence of HF among the US population increases with age, affecting about 6% of people in their 60s and 10% of people in their 80s.19 Participants were excluded if they were: (1) diagnosed with an acute infection; (2) enrolled in a palliative or hospice care program; (3) diagnosed with Alzheimer’s disease; or (4) had a clinical history of stroke (due to higher incidence of multi-infarct dementia).

Variables/Instruments and Data Collection

Cognitive Impairment.  Participants were screened for cognitive impairment (the primary outcome of interest) using the MoCA.20 Although neuropsychological batteries are the gold standard for cognitive testing, such measures are often not feasible in a busy outpatient or clinic setting. The MoCA is a simple cognitive screening tool that includes 8 cognitive domains (visuo-spatial/executive function, naming pictures, memory, attention, abstraction, language, delayed recall, and orientation) with 13 item tests that are most commonly impaired in patients with HF. The MoCA has been validated as a screening tool for mild cognitive impairment, with sensitivity ranging from 90% to 96% and specificity from 87% with 95%.20 The MoCA has been identified as a sensitive tool to screen persons with HF18,21,22 and high-risk populations with cardiovascular diseases.23 The MoCA has a scoring range of 0 to 30. A MoCA score greater than 26 is considered normal, a score between 22 and 26 indicates mild cognitive impairment, between 17 and 21 indicates moderate cognitive impairment, and a score below 17 indicates dementia.20 The MoCA is a 1-page paper-and-pencil test administered in question and answer format by an interviewer. To minimize the effects of education, 1 point is added to the score as a correction factor for patients with less than high school education.20

Severity of HF.  Severity of HF was determined by a noninvasive measurement of cardiac output and cerebral perfusion pressure (CPP), an assessment of the NYHA functional classification and ejection fraction (EF) that are associated with cognitive impairment in HF.24–27 EF was obtained from medical record, and NYHA functional class was assessed on the day of data collection by an experienced cardiology nurse practitioner. CPP was measured using transcranial Doppler (ST3 Transcranial Doppler [TCD] 2006 with DICOM software, Spencer Technologies, Seattle, WA). Jesus and colleagues reported a significant correlation between right middle cerebral artery (MCA) mean flow velocity measured by TCD and Mini-Mental Status Examination score (r=0.23, P=.039) in 83 cardiomyopathy patients.28 Ten measures of bilateral MCA blood flow velocity were recorded using the TCD in semi-Fowler’s position with the head of the bed at a 35° angle. Every patient’s TCD values were confirmed and validated by a neurologist. The CPP was then estimated using the formula derived by Edouard and colleagues using the mean and diastolic MCA velocity (V) and mean arterial blood pressure and diastolic blood pressure (AP) {CPPe=(Vmean/{Vmean−Vdiastolic})  × (AP mean−AP diastolic)}.29

Cardiac output and cardiac index were obtained noninvasively assessed by impedance cardiography (ICG) technique (BioZ Dx Diagnostic system 2006; Cardiodynamic, Bothell, WA). ICG measures were confirmed and validated by a cardiologist. Our primary correlational study of 90 HF patients compared cardiac output and cardiac index measured by thermodilution during cardiac catheterization and ICG in a subgroup of participants (n=12) with a strong correlation among both the measures (r=0.830, P=.011).18 Additionally, both of these measures were validated in prior studies.30,31

Inflammatory Biomarkers.  The inflammatory cytokines (IL-1, IL-6, TNF-α, and CRP) were treated as independent variables to ascertain inflammatory process. In addition, IL-1, IL-6, TNF-α, and CRP were explored for their relationships with cardiac functions and MoCA scores. About 6 mL of blood was obtained from the antecubital vein of patients in 2 red-topped vacutainer tubes. Serum was stored at −80°C until assayed by standard enzyme-linked immunoassay (ELISA) kits and protocols (R&D Systems Inc, Minneapolis, MN). Interassay and intra-assay variability for the assays was <10%. Each of these assays were run at the Rochester Center for Mind-Body Research Laboratory, who funded this project.


Demographic data (ie, age, sex, education, race, living arrangements, and socioeconomic status) and medication list were collected for inclusion as covariates because of their significant association with cognitive impairment, cytokines, and CRP.32–34 Coexisting depression was measured by the Geriatric Depression Scale-15 (GDS-15) using a cutoff score of 6 or above to indicate depression.35 Noncardiac comorbidities were measured using a 14-item modified version of the Cumulative Illness Rating Scale (CIRS) that includes 14 organ systems considering the hematopoietic system separately from the vascular system with an intra-class correlation of 0.78 to assess comorbidity.36 Objective functional capacity and physical activity was measured using the 6-Minute Walk Test (6MWT; intra-class correlation coefficient of 0.90).37 Loneliness was measured using the UCLA Loneliness Scale Version 3 with scores ≤10 to 40 indicating lonesomeness.38

Study Procedures

The study was approved by the University of Rochester’s human subject review board. Participants were recruited from the cardiology practices affiliated with the university. This study used the same protocol as the primary correlation study that compared MoCA and Mini-Mental Status Examination as a cognitive screening tool in HF.18 Forty patients who provided additional consent to have 6 mL of blood drawn in 2 tubes for cytokines (IL-1, IL-6, and TNF-α) and biomarker (CRP) were included in this pilot study and were identified by a code number.

Data Collection

Data were collected by the principal investigator at the University’s Clinical Research Center. The principal investigator received training and certification to use BioZ Dx and TCD. A demographic and clinical data questionnaire was administered first, followed by the MoCA and other questionnaires. Cerebral perfusion pressure, cardiac output/index, and blood samples for cytokines, and CRP were collected and the 6MWT was conducted last to minimize fatigue while obtaining data on cognitive measures and other questionnaires.

Data Analyses

Means and standard deviations were calculated for continuous variables. Frequencies and percentages were calculated for categorical variables. Based on the relative size of standard deviations, levels of IL-6, TNF-α, and CRP were highly variable and thus the data were skewed. The data were log 10-transformed for analysis. An outlier with a CRP score of 239.52 mg/L was excluded from the final regression analysis. Correlations were calculated among cytokines (IL-6, TNF-α), CRP, MoCA score, and continuous measures of cardiac function (cardiac index, cardiac output, and EF). The Kruskal-Wallis test was used to examine the relationship between categorical measures of cardiac function (NYHA class) and IL-6, TNF-α, CRP, and MoCA scores. Multiple linear regression analysis was used to examine the independent relationships between cytokines, CRP, and MoCA score. In an initial model, forward stepwise regression was used to identify and select covariates associated with MoCA score (P<.10) to minimize type I error. After this initial model was fit, IL-6, TNF-α, and CRP were added in a separate model to examine their independent contribution to variation in MoCA scores. The regression effect size was calculated using Cohen’s f2 from multiple regression.39 The Statistical Package for the Social Sciences (version 19; SPSS, Inc, Chicago, IL) was used for analyses.


Of the 40 participants enrolled in the pilot study, blood samples could not be collected for 2 participants due to difficulty during venipuncture. Thus, the total analytical sample included 38 participants. Participants had a mean age of 62 years; 68% were men and 70% were Caucasian. With regard to clinical data, 55% had nonischemic HF and 79% had an EF ≤40. The mean MoCA score was 24.9 (standard deviation±3.2) and scores ranged from 17 to 30. A total of 55.3% (n=21) of the patients scored <26 on the MoCA (a screening value suggestive of mild cognitive impairment). Only 1 patient (2.6%) reported being lonely, with a score above 10 on the loneliness scale, and 3 patients (8%) scored above 6 on the GDS-15, indicating clinical depression. Table I provides details on demographic and clinical data. IL-1α or IL-1β were not quantifiable in this sample of community-dwelling stable HF participants and hence could not be included for further analyses.

Table I. Demographic and Clinical Characteristics of the Sample (N=38)
  1. Abbreviations: ACC/AHA, American College of Cardiology/American Heart Association; ACE inhibitor, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; CPP, cerebral perfusion pressure; CRP, C-reactive protein; EF, ejection fraction; GDS-15, Geriatric Depression Scale-15; IL, interleukin; NYHA class, New York Heart Association classification; 6MWT, 6-minute walk test; SD, standard deviation;TNF-α, tumor necrosis factor α.

Age, y>651231.661.7±8.850–82
Education, yHigh school or less2257.913.4±2.510–22
Marital statusMarried3078.9  
Living arrangementAlone615.8  
Spouse/significant other3284.2  
Etiology of HFIschemic1744.7  
EF, %EF <40%3078.927.7±13.710–55
NYHA classClass I410.5  
Class II2052.6  
Class III1436.8  
ACC/AHA HF stageStage B615.8  
Stage C3078.9  
Stage D25.3  
Cardiac resynchronizationYes924  
ACE inhibitorYes3181.621.7±17.865–80
Aldosterone antagonistYes1745.631.0±12.7612.5–50
CPP, mm/HgHypoperfusion CPP <501642.1  
Normal perfusion CPP ≥502257.953.8±15.228–85
Cardiac indexLow ≤2.41128.9  
L/min/m2Normal ≥2.52771.12.65±0.391.9–3.8
6MWT, mDistance walked below 75th percentile513.2376.8±87.1180–540
UCLA loneliness scaleLoneliness score <1412.614.2±4.010–25
GDS-15GDS score <637.92.95±2.73–11
IL-6, pg/mL   11.41±12.00.93–38.9
TNF-α, pg/mL   0.73±1.260.0–5.42
CRP, mg/L   17.5±15.22.5–66.4

Association Between Severity of HF, MoCA Score, Cytokines, and CRP

Severity of HF measured by cardiac output, CPP, NYHA class, EF, and cognitive function as measured by the MoCA were examined first by performing bivariate correlational analyses and the Kruskal-Wallis test. The MoCA score was not associated with continuous or categorical measures of cardiac function indicative of HF severity (Table II and Table III). Thus, this lack of association indicated no direct correlation between severity of HF and MoCA scores in this study. Although EF was not associated with MoCA score in this sample, EF had an inverse association with IL-6 (r=−0.45, P≤.01) having no association with CRP and TNF-α (P>.05).

Table II. Bivariate Pearson Correlations Between Continuous Measures of Cardiac Function, Cytokines, and Montreal Cognitive Assessment Score
Measure of Cardiac FunctionCytokine/BiomarkerCognitive Measure
Interleukin 6Tumor Necrosis Factor αC-Reactive ProteinMontreal Cognitive Assessment Score
  1. aSignificance with P<.05.

Cardiac index−0.12 (0.48)0.03 (0.87)−0.35 (0.03)a0.05 (0.77)
Left ventricular ejection fraction−0.18 (0.27)−0.11 (0.52)−0.30 (0.07)0.21 (0.21)
Table III. Spearman Correlations Between Categorical Measures of Cardiac Function, Cytokines, and Montreal Cognitive Assessment Score
Measure of Cardiac FunctionCytokine MarkerCognitive Measure
Interleukin 6Tumor Necrosis Factor αC-Reactive ProteinMontreal Cognitive Assessment Score
Median10, 90% P ValueMedian10, 90% P ValueMedian10, 90% P Value R P Value
New York Heart Association Class
1 (n=4)4.71.3, 8.9.470.80.6,, 17.4.390.02.09
2 (n=20)3.01.5,, 1.411.93.3, 38.8
3 (n=14)11.01.5,, 3.317.54.2, 39.0

Association Between MoCA Score, Cytokines, and CRP

The second hypothesis stating that MoCA scores would be negatively associated with IL-6, TNF-α, and CRP after controlling for covariates was tested using a regression model. The result indicated a strong inverse association with MoCA score, IL-6 (r=−0.53, P≤. 001), and CRP (r=−0.34, P<.05). There was no correlation between MoCA score and TNF-α (r=−0.30, P>.05). The results indicated that higher levels of IL-6 and CRP are associated with worse MoCA scores. Bivariate correlations and the Kruskal-Wallis test indicated that education (r=0.59, P<.001), living arrangement (RS=0.36, P<.05), and UCLA loneliness score (r=0.36, P<.05) were associated with MoCA score and were included as covariates in the multiple regression models. Age, sex, total GDS-15 depression score, and distance walked by 6MWT were not associated with MoCA score.

In the multiple linear regressions, the covariate set consisting of education, living arrangements, and UCLA loneliness score explained 46% of the total variation in MoCA scores (Table IV). When IL-6 was added to the model, an additional 11% of the variation in MoCA score was explained and IL-6 was independently associated with MoCA score (P<.001). Parameter estimates of the covariates and IL-6 (log-transformed) are provided (Table V). IL-6 was strongly but inversely associated with MoCA score after controlling for covariates education, living arrangements, and UCLA loneliness score. When IL-6, TNF-α, and CRP were entered step-wise, the model was significant (P<.001) and remained significant for IL-6 (P<.05) but was not significant for TNF-α and CRP (data not shown).

Table IV. Summary of the Multiple Regression Model for the Montreal Cognitive Assessment and Interleukin 6
Model R R2 Adjusted R2Standard Error R2 Change F ChangeP Valuea
  1. aStatistical significance P<.05.

Step 1: Covariates0.680.460.402.440.469.51.001
Step 2: Log-interleukin 60.750.560.512.
Table V. Coefficients Statistics for IL-6 on Multiple Regression Model for MoCA Score
 Unstandardized CoefficientStandardized CoefficientSignificance95% CI
βStandard Errorβ t LowerUpper
  1. Abbreviations: CI, confidence interval; IL, interleukin; MoCA, Montreal Cognitive Assessment. aStatistical significance P<.05.

Living arrangement1.560.620.292.53.016a0.312.85
UCLA Loneliness Scale0.−0.150.29
Log IL-6−0.880.31−0.34−2.81.008a−1.51−0.24

Effect Size

The regression effect size was calculated using Cohen’s f2 from multiple regressions. Cohen’s f2 effect size indicates a large effect size (R2=0.87).

Associations Between MoCA Score, Cytokines, CRP, and Cardiac Medications

Bivariate correlational analyses revealed a significant inverse association between prescription for angiotensin-converting enzyme inhibitor and IL-6 (rS=−0.21, P<.05) and CRP (rS=−0.39, P<.01), with no dose association and no associations with TNF-α (P>.05). Prescription for angiotensin-converting enzyme inhibitors had no association with MoCA scores. Other medications including β-blockers, angiotensin receptor blockers, and statins had no associations with MoCA score, cytokines, or CRP.


The hypothesis that inflammatory processes (as evidenced by elevated levels of cytokines and CRP) may be associated with severity of HF and cognitive function (MoCA score) was not supported in this study. Cardiac variables that indicated severity of HF (including cardiac output, cardiac index, NYHA class, and EF) were not associated with MoCA scores. The lack of a direct correlation between cardiac variables and MoCA score could be due in part to the relatively small sample of stable community-dwelling adults. The chronicity of HF among these stable participants may have dampened the levels of cytokines and CRP.

In addition, our data had no participants in NYHA class IV, the most severe pathological state of HF. This limited our ability to examine cytokine levels during an acute stress response and determine the potential association to cognitive scores. Suzuki and colleagues reported an elevation of IL-6 in acute decompensated NYHA class IV patients with HF compared with controls in NYHA class I and II HF (P<.001), and IL-6 levels peaked within 12 hours after HF admission (r=0.33, P<.001).40 Earlier studies that examined cytokines as prognostic markers in HF involved only the hospitalized sickest patients to follow the progression of the disease.15 Also, serum levels of TNF-α, IL-6, and IL-10 were reported to be significantly higher in NYHA class IV patients compared with those with mild HF (P<.001).41 IL-1β or IL-1β were not quantifiable in this sample of community-dwelling stable HF participants, indicating no acute insult since IL-1 is rapidly induced within minutes of acute injuries or insults.42 However, these cytokines and biomarkers and the inflammatory processes have not been studied for their influence on cognitive function in HF. To our knowledge, there are no studies that have explored cytokines and CRP and/or inflammatory process as potential mediators of cognitive function in HF. The lack of statistical significance in this study may be due to type II error.

The second hypothesis of this pilot study was to examine cognitive scores on MoCA with cytokines (IL-6 and TNF-α) and CRP as indicators of early cognitive changes among stable HF patients living in the community. The result from this pilot study provided evidence that cognitive impairment measured by the MoCA was strongly associated with IL-6 on bivariate analysis and was independently associated in multiple linear regression analyses. This inverse association between IL-6 and cognitive impairment supports findings of other studies of individuals who are at high risk for HF, including the Framingham study43 and the Northern Manhattan study (which focused on a multiethnic population-based cohort of patients older than 39).44 In addition, a recent study of aging rodents (mice) implicated elevated IL-6 in age-related neuron loss and reduced cognitive performance.45

Although cognitive changes are multifactorial in etiology,15,19,20 this pilot study focused only on inflammatory cytokines and CRP for their association with cognitive impairment in HF and did not include other factors that may influence cognition, such as sleep apnea46 and plasma concentration of B-type natriuretic peptide.47 Angiotensin-converting enzyme inhibitor was the only medication that was associated with IL-6 and CRP, although there was no dose association. This supported an earlier study that angiotensin-converting enzyme inhibitors modify the level of cytokines IL-6 (P<.05) in HF patients compared with healthy controls48 and influence levels of CRP.34

The results support the currently available evidence indicating a high prevalence of cognitive impairment among HF patients.3,4 Although literature suggests a high prevalence of depression among persons with HF,49 only 8% of the study participants had a depression score (GDS-15) higher than 6, and depression total score was not associated with MoCA score, cytokines, or CRP in this study, contrary to available evidence.50 This may reflect a restriction in the range of depression scores and antidepression medications that may have influenced this relationship, but they were not available to include in the analyses.

Currently available evidence from neuro-biological research shows that the cytokines IL-1β, IL-6, TNF-α, and CRP play a significant role in complex cognitive processes at the molecular level, such as synaptic plasticity, neurogenesis, and neuromodulation under physiological conditions.14–16 However, there are scant data available in HF literature to support any association between cognitive function and cytokine and CRP levels in HF, an area that needs further exploration. The large effect size (R2=0.87) found in this pilot study provides rationale for a larger exploratory study to examine associations between cognitive function using neurocognitive batteries, cytokines, and CRP levels and may potentially help design intervention studies to reduce cytokine and CRP levels.


This pilot study of cross-sectional data was not designed to compare our patients with controls but instead to calculate an effect size for a future study and to examine association cognitive function and elevated cytokines and CRP in community-dwelling, chronic, stable HF patients. This study used the MoCA as a screening tool for cognitive impairment, and thereby lacks data from a gold standard neuropsychological test battery to more thoroughly evaluate cognitive function. This study also did not collect data on HF duration, recent hospitalizations for HF, or other cardiac or noncardiac events to examine inflammatory marker levels following an acute stressful cardiac event or other pathological conditions.

Conclusions and Practical Implication

These inflammatory/cytokine-mediated cognitive processes may have implications in the long-term development and pathogenesis of mild cognitive impairment related to chronic physiological conditions such as HF and cardiovascular disease.10,11 Appropriately designed longitudinal studies may have the potential to demonstrate early elevation in cytokines and CRP levels as potential markers of impending mild cognitive impairment in HF. Understanding the existence and potential mediating relationships of these cytokines and CRP on cognitive impairment in HF may provide valuable insight into future preventive interventions designed to reduce stressors that release cytokines and CRP.