The Korean Acute Heart Failure (KorAHF) registry was a prospective multicenter cohort study that consecutively enrolled 5,625 patients hospitalized for acute HF syndrome in 10 tertiary university hospitals throughout the country between March 2011 and December 2014. Detailed information on the study design and results has been reported previously (ClinicalTrials.gov NCT01389843) (27, 28). Briefly, patients who had signs or symptoms of HF and met one of the following criteria were eligible for this study: (1) lung congestion or (2) objective left ventricular systolic dysfunction or structural heart disease findings. For assessment of structural heart disease chamber sizes and wall thickness, valve anatomy and functions, presence of cardiomyopathies, and for assessment of diastolic dysfunction left atrial volume index, e′ velocity, and E/e′, among others, were evaluated during the echocardiography by the attending physicians. In the present study, we only included patients whose acute decompensation was not triggered by infection and those with availability of data regarding CRP levels or LVEF. Regarding the aggravating factor for acute decompensation, the responsible physician was asked to choose one of the following factors as the most-likely trigger for acute decompensation, which included acute coronary syndrome, severe hypertension, atrial or ventricular tachyarrhythmia, bradycardia, infection, pulmonary emboli, renal failure, anemia/bleeding, medication (e.g., NSAIDs), non-compliance, endocrinal abnormality, and recent addition of negative inotropic agents. The information on the aggravating factor was prospectively collected and was adjudicated before discharge by the investigators (28).

The study protocol was approved by the Ethics Committee and institutional review board (IRB) of each hospital (Seoul National University Bundang Hospital, IRB No. B-1104-125-014). The study was conducted in accordance with the Declaration of Helsinki and all patients provided written informed consent upon enrollment.

All echocardiographic studies were performed by cardiologists who were certified by Korean Society of Echocardiography, using a standard ultrasound machine with a 2.5-MHz probe. Standard techniques were adopted to obtain M-mode, 2-dimensional, and Doppler measurements in accordance with the American Society of Echocardiography’s guidelines (29). LVEF was measured using Simpson’s biplane method, unless Simpson’s method was not possible. Most patients underwent echocardiography on the day of the admission (the median time interval between admission and echocardiographic exam was 1 day with an interquartile range of 0–2 days). Based on echocardiography findings, HFrEF, HF with mildly reduced EF (HFmrEF) and HFpEF were defined as LVEF ≤ 40%, 41–49%, and ≥50%, respectively (12). Routine blood sampling and tests were conducted by laboratories at each center that were certified by the Korean Association of Quality Assurance for Clinical Laboratory. CRP levels were measured at the index admission. CRP and high-sensitivity CRP levels were measured using a high-sensitivity immunoturbidimetric method. Patients were classified according to the type of HF and CRP tertiles: first tertile [T1]: CRP < 0.30 mg/dL, T2: CRP level 0.30–1.14 mg/dL, T3: CRP > 1.14 mg/dL. The use of statins was evaluated at hospital discharge. The primary endpoints were all-cause, in-hospital, and post-discharge mortality. The mortality data of patients who were lost to follow-up were collected from the Korean Statistical Information Service and Microdata Integrated Service managed by Statistics Korea, a government agency.

Data are presented as numbers and frequencies for categorical variables and as means ± standard deviation for continuous variables. For comparison among groups, the χ² test (or Fisher’s exact test when any expected count was <5 for a 2 × 2 table) was applied for categorical variables and the unpaired Student’s t-test or one-way analysis of variance was applied for continuous variables.

A receiver operating characteristic (ROC) curve was obtained to compare the prognostic performance of CRP in both the types of HF. The Get With The Guidelines-Heart Failure (GWTG-HF) score was calculated for each patient (30) to estimate the risk. The relationship between CRP and other variable was evaluated using Pearson correlation coefficient. Kaplan–Meier curves were plotted and compared using the log-rank test for the evaluation of post-discharge outcomes. Multivariable logistic regression and Cox proportional hazards regression models were used to determine the independent effects of CRP levels on in-hospital and post-discharge outcomes, respectively. We adjusted for variables associated with mortality except variables with >10% missing values or variables that were closely related to other clinical variables. Thus, we adjusted for age, sex, new-onset HF, diabetes, ischemic heart disease, chronic obstructive pulmonary disease, cerebrovascular disease, New York Heart Association class, systolic blood pressure, heart rate, white blood cell (WBC) count, hemoglobin, blood urea nitrogen, E/e′, LVEF, CRP tertiles, and the use of RAS inhibitors, beta-blockers, and MRAs. Since GWTG-HF score stratify the risk of patients, we developed models with adjustment for the GWTG-HF score as a sensitivity analysis.

Statistical significance was set at a two-sided P-value < 0.05. Statistical analyses were performed using SPSS version 25.0 (IBM Corp., Armonk, NY, USA) and R programming version 4.0.2 (The R Foundation for Statistical Computing, Vienna, Austria).

Altogether, 139 (3.6%) patients died during the index admission. Mortality was higher among patients with HFrEF than among those with HFmrEF or HFpEF (4.4% vs. 2.7% vs. 2.5%, P = 0.013). Among patients with HFrEF, in-hospital mortality increased gradually from T1 to T3 (T1: 2.2, T2: 3.6, and T3: 7.4%; χ² test for linear-by-linear association, P < 0.001). Similar findings were observed for HFmrEF and HFpEF or whole population (Figure 1).

In the multivariable analysis, an increase in the CRP tertiles was independently associated with in-hospital outcomes in the whole population [odds ratio (OR) 1.78, 95% CI 1.33–2.39] as well as in the subgroups of HFrEF (OR 1.58, 95% CI 1.09–2.30), HFmrEF (OR 1.51, 95% CI, 0.72–3.52), HFpEF (OR 2.98, 95% CI, 1.46–6.33) (Table 2).

Since the WBC count increases during inflammation, we also investigated the prognostic value of WBC count. There was a weak, but significant correlation between WBC count and CRP (r = 0.27, P < 0.001). WBC count was not associated with increased in-hospital mortality (P = 0.118). However, it was associated with post-discharge mortality (P < 0.001). There was a weak correlation between CRP and BNP (r = 0.11, P < 0.001) and between CRP and N-terminal-pro-BNP (r = 0.18, P < 0.001), suggesting that CRP levels are less dependent on the natriuretic peptide levels.

Among 3,692 patients who discharged alive, 1,269 (34.4%) patients died during a median follow-up of 995 days (interquartile range: 365–1,386 days). In contrast to the in-hospital mortality, the post-discharge mortality did not differ between the HFrEF, HFmrEF, and HFpEF groups (34.4% vs. 33.1% vs. 35.1%, P = 0.730). After stratification according to the CRP tertiles, there was a gradual increase in mortality with an increase in the CRP tertiles in HFrEF (T1: 27.3, T2: 32.6, T3: 43.9%; P < 0.001), HFmrEF (T1: 24.5, T2: 29.4, and T3: 46.8%; P < 0.001), and HFpEF (T1: 27.8, T2: 39.0, and T3: 40.6; P = 0.001) (Figure 2).

In the Cox proportional hazards regression analysis after adjustment for significant covariates, an increase in the CRP tertiles was independently associated with post-discharge mortality in the whole population [hazard ratio (HR) 1.22, 95% CI 1.13–1.32] as well as in the subgroups of HFrEF (HR 1.20, 95% CI 1.08–1.33), HFmrEF (HR 1.38, 95% CI, 1.12–1.70), and HFpEF (HR 1.19, 95% CI, 1.02–1.38) (Table 2). The results of the sensitivity analysis performed with adjustment for the GWTG-HF score were consistent with those of the main analyses. We also performed sensitivity analysis by classifying HF as HFrEF (LVEF ≤ 40%) and HFpEF (LVEF > 40%) because when KorAHF registry was designed in 2010, the LVEF cutoff for HFpEF was 40% which was also recommended by 2005 ESC-HF guideline. These results by two categories were similar as primary results (Supplementary Figures 1, 2).

Among total 3,692 patients who discharged alive, 2,096 (56.8%) patients received statin. Baseline characteristics according to statin use are shown in Supplementary Table 2. Overall, patients with statin-use were more likely to be old and male, and had more co-morbidities than those without statin-use. By multivariable Cox proportional hazard regression models, statin use was not associated with all-cause mortality in the whole population or in patients with HFrEF (Table 3). However, in patients with LVEF > 40% (HFmrEF plus HFpEF), statin users showed better survival trend than those without statin use with a marginal statistical significance. This trend only observed in patients with elevated CRP levels (CRP tertile 1 and 2: HR 0.87, 95% CI 0.69–1.10, P = 0.254; tertile 3: HR 0.76, 95% CI 0.56–1.02, P = 0.065). Among these patients with LVEF > 40% plus CRP tertile 3, statin use was associated with 46% reduced risk for mortality in those with ischemic heart disease (HR 0.54, 95% CI 0.31–0.97).

Currently, HF is classified according to EF (31), mainly because patients with similar EF show similar responses to pharmacologic and non-pharmacologic treatments (12). The difference in the cardiac anatomy contributes to the differences in hemodynamic and neurohumoral effects (12). Patients having HFrEF with enlarged LV diameter exhibit higher BNP levels than patients with HFpEF, implying a higher degree of neurohumoral activation (11). In this study the CRP levels were similar between HFrEF, HFmrEF, and HFpEF, although HFmrEF and HFpEF had numerically higher CRP level than HFrEF patients. Previous studies investigating the differential effects of inflammatory markers according to HF phenotypes showed inconsistent results (14–16, 23). For example, Tromp and colleagues showed 96 patients with chronic HFpEF, defined as LVEF ≥ 45%, had higher CRP level than 364 patients with HFrEF, defined as LVEF < 45%. The difference in the study population may have led to different results. However, our study has a large sample size, and the patients had been carefully selected so those with infection as trigger for the acute decompensation were excluded. In this study, CRP seems to be independent of LVEF, since CRP level and the magnitude of its prognostic impact did not differ between HFrEF and HFpEF. This key finding suggests that the neurohumoral and inflammatory pathways are two distinct pathways. The third pathway is the cardiometabolic pathway, which also seems to be independent of LVEF. While drugs that modulated neuro-humoral activation improved outcomes in HFrEF who have higher BNP levels than HFpEF (12), empagliflozin, an SGLT2-inhibitor which mainly acts on cardiometaboilc pathway, improved the outcomes in both HFrEF and HFpEF with similar impact size (18, 19).

We observed a gradual increase in both in-hospital and post-discharge mortality with an increase in the CRP levels, suggesting a dose-dependent relationship. However, the exact mechanisms are still not well defined. It is also unclear whether CRP is a marker or mediator of HF progression. CRP is an acute-phase protein that mediates and perpetuates inflammation and may accelerate the cardiac remodeling process. Interleukin-6 (IL-6) induces CRP production in the liver (32, 33) and IL-1β is a central inflammatory cytokine that drives the IL-6 signaling pathway. Canakinumab is a fully human monoclonal antibody that inhibits IL-1β and has been shown to reduce CRP levels and the composite of death, myocardial infarction, or stroke (34).

Another important finding in the present study is that statins seem to be beneficial in patients with elevated CRP levels. In our analysis patients with LVEF > 40% and CRP tertile 3 seem to benefit from statins. Among these patients, statin use was associated with 44% reduced risk for mortality in those with ischemic heart disease. Although statins have robustly improved outcomes in high-risk patients (35), their effect seems to be neutral in patients with HF. In the Controlled Rosuvastatin Multinational Trial in Heart Failure (25) and the GISSI-HF study (26), statins did not improve the outcomes in patients with HF. Nevertheless, in the Justification for the Use of Statins in Primary Prevention study (24), the use of 20 mg of rosuvastatin in patients with CRP ≥ 2.0 mg/dL reduced the CRP levels by 37% and the composite of myocardial infarction, stroke, arterial revascularization, hospitalization for unstable angina, or death from cardiovascular causes by 46%. It is well known that not all patients benefit equally from a specific therapy. For example, the effect of beta-blockers seems to be greater in patients having HFrEF with a higher heart rate (36). Similarly, statins that modulate inflammation may be more effective in patients having HF with elevated CRP levels, as shown in the present study. This hypothesis should be evaluated in clinical trials.

The present study has several limitations. Although we adjusted for significant clinical factors and the GWTG-HF score, there might be residual confounding factors that could have influenced the relationship between CRP levels and clinical outcomes due to the observational nature of the study. In patients with stable HF, the CRP level remains relatively stable (24). Therefore, we excluded patients whose acute decompensation was triggered by infection. Although data regarding the triggers in each patient were prospectively collected, confirmed, and adjudicated by the investigators, their interpretation remains subjective and was based on the judgment of the investigators, which might limit the reproducibility of the study results. Since we enrolled only the patients hospitalized for acute HF in East Asia, the generalizability of our results to other racial groups and to patients with chronic stable HF might be limited. Finally, the number of patients with HFmrEF was relatively small (n = 559), thus predisposes to type II error, i.e., false negative results.