Association of inflammatory markers with incident heart failure or cancer in the HUNT3 and Health ABC population studies

Abstract

Aims

To investigate the relationship between chronic low-grade inflammation, as measured by high-sensitivity C-reactive protein (hsCRP) levels, and incident heart failure (HF) or cancer.

Methods and results

We assessed the relationship between baseline hsCRP concentrations and subsequent HF or cancer in two community-based cohorts, the Trøndelag Health Study (HUNT3) and the Health, Aging and Body Composition (ABC) study. In the latter, the analysis was replicated with interleukin (IL)-1, IL-6, or tumour necrosis factor (TNF)-α instead of hsCRP. In HUNT3, hsCRP was measured in 47 163 subjects (mean age 52.3 ± 15.8 years). During a median follow-up of 12.1 years, 2034 (4.3%) individuals developed HF and 5024 (10.7%) cancer, with 442 (0.9%) being diagnosed with both. After adjusting for age, male sex, diabetes, obesity, previous or current smoking, and comorbidities, elevated baseline hsCRP was associated with a higher risk of HF or cancer [hazard ratio (HR) 1.09; 95% confidence interval (CI), 1.07–1.10]. In the Health ABC study, hsCRP levels were assessed in 2803 participants, who had a mean age of 72.6 ± 2.9 years and a higher burden of comorbidities than in HUNT3. During a median follow-up of 8.2 years, HF and cancer were diagnosed in 346 (12.3%) and 776 (27.7%) subjects, respectively, with 77 (2.7%) having both conditions. After adjusting for the same variables used for the HUNT3 cohort, hsCRP remained significantly associated with incident HF or cancer (HR 1.11; 95% CI, 1.05–1.18), as were IL-1 (HR 1.15; 1.07–1.24), IL-6 (HR 1.09; 1.02–1.17), and TNF-α (HR 1.15; 1.07–1.24).

Conclusion

A state of chronic, low-grade inflammation captured by an increase in hsCRP levels is associated with an increased risk of developing HF or cancer, with potential implications for clinical trials with anti-inflammatory therapies.

See the editorial comment for this article ‘Inflammation at the crossroad between cancer and heart failure’, by A. Cuomo et al., https://doi.org/10.1093/eurjpc/zwae166.

Introduction

Several cardiovascular (CV) risk factors, such as obesity and diabetes, portend an increased risk of both heart failure (HF) and cancer.¹ This shared risk profile suggests that common underlying mechanisms may contribute to the development of both HF and malignancy.² Most CV risk factors are either pro-inflammatory stimuli, such as smoking, or correspond to a state of chronic, low-grade inflammation, as seen in obese individuals,³ and inflammation is causally implicated in atherosclerosis,⁴ maladaptive cardiac remodelling,⁵ and oncogenesis.⁶ Ageing, the single most potent predictor of CV risk, is associated with a state of low-grade inflammation characterized by increased levels of high-sensitivity C-reactive protein (hsCRP) and interleukin (IL)-6.⁷ Recent work has uncovered clonal haematopoiesis of indeterminate potential (CHIP) as a driver of inflammation during ageing,⁷ and has linked CHIP with an increased risk of atherosclerotic CV disease,⁸ incident HF,⁹ and cancer.¹⁰ Further evidence connecting inflammation to HF and cancer comes from the Canakinumab Anti-inflammatory Thrombosis Outcome Study (CANTOS) trial, in which blockade of IL-1β reduced the rates of HF hospitalization and HF-related death as well as the incidence of lung cancer in patients with prior myocardial infarction (MI) and hsCRP ≥ 2 mg/L.^11,12

On these grounds, HF and cancer can be provocatively viewed as distinct manifestations of a common disease spectrum driven by chronic inflammation.¹³ Following this reasoning, we determined the relationship between hsCRP concentrations at baseline and the occurrence of HF or cancer in two large, prospective cohorts representative of the middle-aged and elderly general population of industrialized countries.

Methods

Study population

For this analysis, we used data from the 3rd wave of the Trøndelag Health Study (HUNT3) and from the Health, Aging and Body Composition (Health ABC) study.^14–16

The 3rd wave of HUNT3 enrolled 50 803 ≥ 20-year-old individuals residing in the Nord-Trøndelag County (Norway) between October 2006 and June 2008. HUNT3 comprises data from questionnaires, interviews, clinical measurements, and biological samples (blood and urine), which we linked with two nationwide and mandatory registries, the Cancer Registry of Norway and the regional hospital electronic medical records registry, to obtain information on incident cases of cancer and HF, respectively.

The Health ABC study included 3075 individuals, aged 70–79 years at inception, identified from a random sample of residents in designated zip code areas surrounding Pittsburgh, PA and Memphis, TN (USA). To be eligible for recruitment, participants had to be free of functional limitations, defined as self-reported any difficulty in walking 0.25 mile or climbing 10 stairs without resting. Subjects were also excluded if they described difficulties in their activities of daily living, had cognitive impairment, or were unable to communicate with the interviewer. Baseline data were collected from 1997 to 1998. Participants were followed up by yearly clinic examinations for 6 years and received phone calls every 6 months to provide updates on their functional and health statuses.¹⁶

Serum biomarkers measurements

In both HUNT3 and the Health ABC study, hsCRP levels were assessed at the time of cohort entry. In the Health ABC study, concentrations of IL-1, IL-6, and tumour necrosis factor (TNF)-α were also determined at baseline.

In HUNT3, non-fasting serum samples were collected from all participants at the HUNT3 health examination station and transported by cold chain courier to the Central Laboratory, Levanger Hospital, where hsCRP levels were assessed using latex immunoassay methodology reagent kit 6K26–41 CRP Vario (Abbot, Clinical Chemistry, USA). The detection limit was 0.03 mg/L, and samples without detectable CRP were assigned this value.¹⁷ In the Health ABC study, cytokines were measured using an ELISA kit from R&D Systems (Minneapolis, MN). The detectable limit was 0.10 pg/mL for IL-6 (by HS600 Quantikine kit) and 0.18 pg/mL for TNF-α (by HSTA50 kit). Serum levels of hsCRP were also assessed by ELISA based on purified protein and polyclonal anti-CRP antibodies (Calbiochem, San Diego, CA). The lower limit of detection of this assay was 0.007 mg/L.¹⁸

Study outcomes

In the HUNT3 cohort, incident HF was defined as the first HF diagnosis (ICD10 codes: I11.0, I13.0, I13.2, I50) in structured electronic medical records regarding all inpatient and outpatient admissions/contact to hospitals in the study area, and incident cancer was defined as the first diagnosis of cancer as reported in the Cancer Registry of Norway.

In the Health ABC cohort, incident HF was defined as a first admission with an overnight stay. Both incident HF and cancer diagnoses were adjudicated by a physician and confirmed by medical record documentations and, in the case of cancer, pathology reports.¹⁶

Statistical analysis

After excluding subjects with prevalent HF or cancer as reported in records collected at enrolment, the incidence of HF or cancer (whichever came first) was compared between patients with baseline hsCRP levels ≥ 2 mg/L and <2 mg/L, a cut-off based on the inclusion criteria of the CANTOS trial and of the ongoing ZEUS and HERMES trials (NCT05021835 and NCT05636176, respectively),^11,12 and across tertiles of baseline hsCRP by the Kaplan–Meier method. Moreover, the association between base 2 log-transformed hsCRP (owing to non-normal distribution) or other baseline characteristics and the composite outcome of HF or cancer was evaluated by means of Cox proportional hazards regression models, using time from enrolment as the time scale. A sensitivity analysis excluding patients with hsCRP ≥ 50 mg/L was performed to exclude patients with a possible acute phase response at the time of entry in the HUNT3 study. In the analysis with hsCRP, IL-1, IL-6, or TNF-α in the Health ABC study, the hazard ratios are expressed for 1 SD increase in each biomarker to allow indirect comparisons between them.

Statistical analysis was performed using R software, version 4.0.2.

Results

Characteristics of the HUNT3 population stratified by high-sensitivity C-reactive protein levels

In HUNT3, hsCRP was measured in 50 049 individuals at enrolment. Of these, we excluded 2478 (5.0%) with cancer and 498 (1.0%) with HF at baseline, leaving a sample of 47 163 subjects. Overall, the mean age was 52.3 ± 15.8 years and 45% of subjects were males. The median hsCRP value was 1.2 mg/L [interquartile range (IQR) 0.6–2.7 mg/L]. Table 1 shows the characteristics of the sample stratified according to hsCRP levels at study entry. The frequency of CV risk factors including diabetes, obesity, and previous or current smoking habit was higher in subjects with hsCRP ≥ 2 mg/L. Similarly, the prevalence of previous MI and stroke, as well as of non-CV comorbidities including chronic kidney disease (CKD) and chronic obstructive pulmonary disease (COPD), was higher in the high hsCRP group. The same trends were observed when we excluded 637 subjects with hsCRP levels ≥ 50 mg/L (see Supplementary material online, Table S1), which might reflect an acute and transient inflammatory state rather than chronic low-grade inflammation. Similarly, when we stratified the study population based on hsCRP tertiles, the prevalence of CV risk factors, CV disease, CKD, and COPD was higher with increasing levels of hsCRP (see Supplementary material online, Table S2), even after excluding subjects with hsCRP levels ≥ 50 mg/L (see Supplementary material online, Table S3).

Risk of incident heart failure or cancer in HUNT3

During a median follow-up of 12.1 (IQR 11.6–12.6) years, there were 2034 (4.3%) cases of incident HF and 5024 (10.7%) cancer diagnoses, with 442 (0.9%) individuals developing both HF and cancer. The risk of developing HF or cancer was higher in patients with hsCRP ≥ 2 mg/L (Figure 1A) and progressively increased with increasing levels of hsCRP when the study population was stratified based on hsCRP tertiles (see Supplementary material online, Figure S1A). In multivariable analysis, age, male sex, diabetes, obesity, previous or current smoking, a history of MI, angina or stroke, and non-CV comorbidities were all associated with an increased risk of incident HF or cancer (Table 2). The increased risk of incident HF or cancer associated with diabetes, previous MI or stroke, COPD, and CKD was driven by an increased risk of HF, but not cancer, whereas age, male sex, obesity, and previous or current smoking portended an increased risk of both HF and cancer. Even after adjustment for all these factors, hsCRP remained significantly associated with either incident HF or cancer and the composite outcome of HF or cancer. Each fold-increase of base 2 log-transformed hsCRP was associated with a 9% higher risk of HF or cancer (Table 2). In a sensitivity analysis separating the combined outcome of HF or solid cancer and that of HF or haematologic cancer, hsCRP remained significantly associated only with the former (see Supplementary material online, Table S4).

Characteristics of the Health ABC population stratified by high-sensitivity C-reactive protein levels

Compared with the HUNT3 cohort, subjects from the Health ABC study sample were on average 20 years older (72.6 vs. 52.3 years) and had a markedly higher prevalence of diabetes (14.6% vs. 4.1%), previous MI (10.6% vs. 2.8%) or angina (10.7% vs. 3.4%), and CKD (26.6% vs. 3.2%) (Table 3). In this cohort, hsCRP concentrations were assessed in 3037 participants, 237 of whom were excluded because of HF (94, 3.1%) or cancer (143, 4.7%) at baseline. The median hsCRP value was 1.66 mg/L (IQR 0.99–3.07 mg/L). Furthermore, 2365 subjects had IL-1, IL-6, and TNF-α measured. As shown in Table 3, diabetes, obesity, CKD, and COPD were more common among subjects with hsCRP ≥ 2 mg/L at study entry. Similarly, the prevalence of diabetes, obesity, CKD, and COPD increased across tertiles of hsCRP (see Supplementary material online, Table S5). In contrast, the prevalence of a history of MI, angina, or stroke did not change with hsCRP concentrations.

Risk of incident heart failure or cancer in Health ABC

During a median follow-up of 8.2 (IQR 4.2–12.2) years, there were 346 (12.4%) diagnoses of HF and 776 (27.7%) cases of incident cancer, with 77 (2.8%) participants having both HF and cancer. Akin to the HUNT3 cohort, the risk of developing HF or cancer was higher in patients with hsCRP levels ≥ 2 mg/L (Figure 1B) and increased across tertiles of hsCRP levels (see Supplementary material online, Figure S1B). Applying the same multivariable model as in Table 2, age, male sex, diabetes, current or former smoking, a history of MI, and COPD were associated with an increased risk of incident HF or cancer (Table 4). When considering cancer and HF as independent outcomes, only male sex was associated with an increased risk of both conditions; diabetes and MI portended an increased risk of incident HF, whereas smoking and COPD were associated with an increased risk of cancer (Table 4). After adjustment for these factors, we confirmed the association of hsCRP with the risk of HF or cancer (Table 4). Furthermore, no significant heterogeneity in the association of inflammation with HF or cancer was observed when IL-1, IL-6, or TNF-α were used instead of hsCRP [Cochran’s Q(df = 3) = 1.68, P = 0.64; Figure 2]. Remarkably, we did not find an interaction between hsCRP and sex on the risk of developing HF or cancer (HUNT3: P = 0.55, Health ABC: P = 0.15).

Discussion

In the present study, we show that hsCRP concentrations are associated with an increased risk of incident HF or cancer, independently of age, CV risk factors, pre-existing CV disease, and key comorbidities, in two large and well-defined populations. Associations of hsCRP with incident HF and cancer were of similar size, and were consistent between HUNT3 and the Health ABC study despite the different ages and demographic characteristics of the two cohorts.

These results add to epidemiological evidence that an inflammatory milieu, reflected by increased levels of circulating inflammatory markers, portends a high risk of HF¹⁶ and of different types of malignancies.^19,20 The concept that systemic inflammation is a shared pathogenic pathway for HF and cancer implies that anti-inflammatory treatments may be an effective approach to prevent both conditions. In fact, besides its potential cancer-preventive activity revealed by the CANTOS trial, IL-1 blockade might represent an effective treatment for several types of cancers, and other cytokine-targeting agents are being tested in clinical trials for solid and haematologic malignancies.²¹ However, the benefit–harm trade-off of large-scale treatments targeting inflammation remains to be ascertained.²²

The role of pro-inflammatory cytokines in CV disease emerged from large epidemiological studies that identified IL-6 and hsCRP as potent predictors of CV events.^23,24 The mechanisms underlying this association were dissected in preclinical studies that implicated pro-inflammatory mediators in atherogenesis and plaque rupture and, more recently, in maladaptive ventricular remodelling following MI and in the pathogenesis of diastolic dysfunction.^25,26 The CANTOS trial demonstrated that IL-1β inhibition with canakinumab reduces the risk of CV events, including HF-related hospitalization and mortality. Notably, these effects were independent of lipid and blood pressure lowering, and their magnitude correlated with the decrease in hsCRP levels.^12,27 Furthermore, the anti-inflammatory agent colchicine reduced CV events in patients with either a recent MI or stable coronary artery disease in the Colchicine Cardiovascular Outcomes Trial (COLCOT) and Low-Dose Colchicine 2 (LoDoCo2) trials, respectively.^28,29 In addition to hindering neutrophil migration by interfering with tubulin polymerization, colchicine exerts its anti-inflammatory activity by inhibiting the nod-like receptor pyrin domain containing protein-3 (NLRP3) inflammasome, which catalyses the proteolytic cleavage and maturation of pro-inflammatory cytokines, including IL-1β. It remains unclear whether colchicine decreased hsCRP levels in the two trials, since this endpoint was assessed only in small subgroups of patients.^29,30 In contrast, in the Cardiovascular Inflammation Reduction Trial (CIRT), low-dose methotrexate did not reduce CV events nor hsCRP levels in patients with previous MI or multivessel coronary disease and metabolic syndrome or diabetes mellitus.³¹

Altogether, these studies buttress the concept that targeted anti-cytokine therapies, rather than broad-spectrum anti-inflammatory agents, might abate CV risk related to low-grade, chronic inflammation that is not affected by controlling traditional risk factors. The magnitude of this residual inflammatory risk is captured by hsCRP. This inflammatory biomarker does not represent a treatment target per se, but reflects the activity of the IL-1 and IL-6 signalling cascades, which are instead mechanistically involved in the pathophysiology of both CV disease and cancer.³² On these grounds, the CV effects of IL-6 inhibition with ziltivekimab are being tested in patients with CV disease and CKD (NCT05021835) or HF (NCT05636176) and hsCRP ≥ 2 mg/L.

The results of our study emphasize the importance of hsCRP as a downstream clinical biomarker that captures a state of chronic, low-grade inflammation, which is associated with an increased risk of both HF and cancer. In line with this concept, in the Health ABC study sample, three inflammatory markers other than hsCRP, namely IL-1, IL-6, and TNF-α, performed similarly to hsCRP in predicting the risk of incident HF or cancer (Figure 2). Of note, it was recently shown that the predictive value of hsCRP for both HF and cancer can be strengthened by monitoring its longitudinal changes over time, rather than using a single baseline measurement.³³

This analysis has shortcomings. Only a limited number of covariates, common to HUNT3 and the Health ABC study, were taken into account, thereby leaving the possibility of residual confounders underlying the association between inflammation and incident HF or cancer. Second, HF was not characterized, e.g. by left ventricular ejection fraction value, and cancer was only further defined as solid or haematologic in HUNT3. Nonetheless, the results presented here are biologically plausible and were consistent in two cohorts with remarkable differences regarding median age, frequency of CV and non-CV disorders, and rate of outcomes, suggesting their generalizability.

In conclusion, our results suggest that a state of chronic, low-grade inflammation, captured by an increase in hsCRP levels, is associated with an increased risk of developing HF or cancer. Despite not being a therapeutic target, hsCRP is a reliable biomarker to identify those individuals at higher risk of HF or cancer, who are more likely to benefit from anti-inflammatory treatments such as IL-6 inhibition in ongoing clinical trials.

Supplementary material

Supplementary material is available at European Journal of Preventive Cardiology.

Acknowledgements

The Trøndelag Health Study (HUNT) is a collaboration between HUNT Research Centre (Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology NTNU), Trøndelag County Council, Central Norway Regional Health Authority, and the Norwegian Institute of Public Health.

Author contribution

All authors made substantial contributions to the conception and design of the work. E.B. and L.C., and P.A. contributed to the acquisition, analysis, and interpretation of data. E.B. and P.A. drafted the manuscript. All authors have approved the submitted version and have agreed both to be personally accountable for the author’s own contributions and to ensure that questions related to the accuracy or integrity of any part of the work.

Funding

C.M. was supported by the German Research Foundation (DFG; Ma 2528/7–1 and 8–1;SFB 894, TRR 219, SFB 1525, project #453989101) and Deutsches Zentrum für Herz-Kreislaufforschung (DZHK; EX22). P.A. was supported by the Italian Ministry of Health (GR-2018-12365661).

Italian Ministry of Health (GR-2018-12365661); German Research Foundation (DFG; Ma 2528/7-1 and 8-1;SFB 894, TRR 219, SFB 1525, project #453989101) and Deutsches Zentrum für Herz-Kreislaufforschung (DZHK; EX22).

Data availability

The data underlying this article were provided by the Norwegian University for Science and Technology (for the HUNT3 study) and the National Institute on Ageing (for the Health ABC cohort) under licence. The HUNT3 data can be accessed by submitting an application to the HUNT Data Access Committee. Health ABC data can be accessed by submitting a research proposal to the National Institute on Aging (https://www.nia.nih.gov/healthabc-study).

References

Author notes