- Top of page
- Study Limitations
Few studies have explored the clinical potentials of combined Cystatin C (Cys) and uric acid (UA) in heart failure (HF). The authors evaluated Cys and UA as predictors of clinical outcomes compared with conventional renal biomarkers. This prospective cohort study included 587 HF patients presenting with dyspnea. At admission, Cys, UA, and other renal measures including serum urea nitrogen (BUN), creatinine (Cr), and glomerular filtration rate (GFR) were obtained. The primary endpoint was the composite of cardiac death and rehospitalization for worsening HF. During a 25-month median follow-up period, 68 patients experienced clinical outcomes: 9 cardiac deaths and 59 HFs. They showed higher BUN and Cr values and lower GFR. Within these parameters, Cys and UA had the most favorable area under the curves, and patients with Cys ≥0.8 mg/L and UA ≥6.6 mg/dL showed more frequent events. The net reclassification improvement analysis showed the combination of Cys and UA had a greater incremental effect for cardiac prognosis. On multivariate Cox hazard analysis, Cys and UA were independent predictive markers for clinical outcomes. In HF patients presenting with dyspnea, Cys and UA appear to be more useful predictors of clinical events than other renal measures.
Patients with acute heart failure (HF) usually have concomitant diseases, such as hypertension, diabetes, or peripheral vascular disorders. Among these diseases, renal dysfunction has played an important role in predicting future adverse cardiac events. Importantly, “cardiorenal syndrome,” which is the interaction between the heart and renal function, has clarified the risk-stratification and prognosis of cardiac patients.1,2 Indeed, the presence of even subtle or mild renal dysfunction is usually associated with a higher risk for adverse cardiac outcomes.3,4
Serum urea nitrogen (BUN), creatinine (Cr), and glomerular filtration rate (GFR) are classic biomarkers that have been widely used to evaluate renal function. Similarly, uric acid (UA) and cystatin C (Cys) have been studied for use in evaluating renal dysfunction. UA has been considered a metabolic marker, while Cys has been considered an early marker for acute kidney injury.5–8 Based on these characteristics, UA and Cys may be used for the prognosis of HF or ischemic heart disease (IHD).9–12
The potential value that Cys and UA may offer for prognosis has been unclear; however, it is unknown how they could improve the evaluation of cardiac events in acute HF patients with mild to moderate renal dysfunction. With regard to the interaction between the heart and renal function, we hypothesized that admission Cys and UA would be related to long-term prognosis. Therefore, the aim of this study was to assess the clinical impact of using UA and Cys to predict adverse clinical outcomes and compare their performance to conventional renal biomarkers.
- Top of page
- Study Limitations
A total of 587 patients were enrolled in this study, and during a median follow-up period of 25 months, 68 patients (11.5%) experienced the clinical outcomes of 9 cardiac deaths and 59 HFs. The baseline clinical and laboratory characteristics of the study population are presented in Table I. Patients with cardiac events were older and more often had experienced prior HF and IHD compared with patients without events. They had lower systolic blood pressure, lower EF, and worse New York Heart Association (NYHA) functional class. The prevalence of diabetes, hypertension, atrial fibrillation, and types of prescribed medication taken were similar between the two groups. Based on conventional renal measures, including BUN, Cr, and GFR, renal function was more deteriorated in the cardiac events group. In addition, that group showed higher levels of Cys, UA, and NT-ProBNP.
Table I. Baseline Characteristics of Study Population According to Clinical Outcomes
|Variables||Events (−) (n=519)||Events (+) (n=68)|| P Value|
|Systolic BP, mm Hg||125±18||120±20||.046|
|Diastolic BP, mm Hg||75±11||73±15||NS|
|Heart rate, beats per min||74±12||78±16||.022|
|Ejection fraction, %||58.6±12.3||48.5±16.4||<.001|
|NYHA class III/IV, %||48.7||83.8||<.001|
| Admission for HF, %||15.7||46.7||<.001|
| IHD, %||40.5||56.0||.014|
| Diabetes, %||29.6||33.2||NS|
| Hypertension, %||42.4||39.1||NS|
| Atrial fibrillation, %||28.9||32.1||NS|
| BUN, mg/dL||17.4±6.2||20.4±8.2||.005|
| Creatinine, mg/dL||1.1±0.3||1.2±0.4||.001|
| GFR, mL/min/1.73 m2||71.5±22.9||63.4±28.0||.008|
| Cystatin C, mg/L||0.78±0.25||1.12±0.56||<.001|
| Uric acid, mg/dL||5.6±1.8||7.0±2.3||<.001|
| NT-ProBNP, pg/mL||1.029±3.739||4.998±8.101||<.001|
| ACE inhibitor or ARB, %||58.0||47.1||NS|
| BB, %||59.9||55.9||NS|
| CCB, %||23.3||25.0||NS|
| Diuretics, %||30.2||35.3||NS|
| Aldosterone antagonist, %||5.2||2.9||NS|
| Amiodarone, %||3.7||5.9||NS|
| Aspirin, %||73.6||76.5||NS|
| Coumadin, %||28.9||25.0||NS|
By ROC curve analyses, BUN ≥18 mg/dL, Cr ≥1.1 mg/dL, GFR ≤57.8 mL/min/1.73 m2, UA ≥6.6 mg/dL, and Cys ≥0.8 mg/L were the best cutoff values for predicting clinical outcomes (Figure 1, left). Although the differences of areas under the ROC of each UA and Cys were not significant, both AUCs were significantly greater than that of BUN, Cr, and GFR. With regard to the validity of modeling, C indices between each model were not significantly different, but the Cys- or UA-based modeling in both NRI and IDI could provide the addictive prediction probability more over 10% than those of conventional renal measures (Table II, Figure 2). Clinical outcomes were compared by Kaplan-Meier analysis according to the cutoff values of UA and Cys: UA <6.6 mg/dL and Cys <0.8 mg/L, UA ≥6.6 mg/dL or Cys ≥0.8 mg/L, and UA ≥6.6 mg/dL and Cys ≥0.8 mg/L (Figure 1, right). Serum UA and Cys levels independently and significantly predicted unfavorable prognosis; thus, combining the best UA and Cys cutoff values could improve the prediction of adverse clinical outcomes.
Figure 1. Receiver-operating characteristic curve analysis of renal measures to identify clinical adverse outcomes (left). Kaplan-Meier survival curves presenting both uric acid and cystatin C level (right).
Download figure to PowerPoint
Table II. Comparison of Cox Regression Models in Prediction of the Occurrence of Cardiovascular Adverse Events
|Prognostic Modela With Risk Factors at Presentation||Harrell’s C Index||Net Reclassification Improvement|| P Value||Integrated Discrimination Improvement|| P Value|
Figure 2. Net reclassification improvement of patients with or without events. In the event group, 8 (21%) patients were correctly reclassified as at lower risk for events (green) and 4 (10%) patients incorrectly moved downward (pink). In the event-free group, 53 (10%) patients were reclassified in a desirable direction (green) and 25 (5%) patients incorrectly moved upward with a higher risk (pink). The Established Prognostic Model includes a previous history of heart failure and ischemic heart disease, New York Heart Association functional class III or IV, and N-terminal Pro–B-type natriuretic peptide level.
Download figure to PowerPoint
Univariate and multivariate analyses were performed to identify independent predictors of each measure (Table III). All 5 measures for renal function were significantly associated with increased clinical outcomes in the univariate analysis (data not shown). In the multivariate analysis, however, UA and Cys remained the significant prognostic factors, after adjusting for NYHA class and previous history of HF and IHD. Furthermore, we used UA and Cys to evaluate the impact that both parameters had on clinical events. Pooled estimates of adjusted hazard ratios for clinical outcomes (using categories for both UA and Cys) are shown in Figure 3. Throughout the full range of hazard ratios, UA and Cys were associated strongly and positively with clinical events and appeared to follow a log-linear model. The risk of clinical events was substantially higher for patients with a UA level of ≥7.0 mg/dL than for patients with UA concentrations of 3.0 to 6.5 mg/dL. We noted similar findings for Cys. Events were strongly clustered in Cys levels >0.8 mg/L, increasing more than 3-fold over the full range.
Table III. Cox’s Proportional Hazards Analysis for Prediction of the Occurrence of Cardiovascular Adverse Events
|HR||95% CI|| P Value||HR||95% CI|| P Value|
Figure 3. Hazard ratio of clinical outcomes by category of uric acid (upper) and cystatin C (lower) level. Data are adjusted for New York Heart Association (NYHA) class and previous history of heart failure and ischemic heart disease. Error bars represent 95% confidence intervals.
Download figure to PowerPoint
- Top of page
- Study Limitations
The important finding of the current study is that in acute HF, Cys and UA could provide the incremental prognostic benefits over the conventional classic renal measures: BUN, Cr, and GFR. Despite the small but significantly different impacts of BUN and Cr, the renal dysfunction could signal worsening HF. Together with direct hemodynamic impact, the decreased renal function might accelerate the progression of atherosclerosis or IHD. Cys has been regarded as a useful marker for renal dysfuction and could also be a valuable tool for predicting cariovascular clinical outcomes.7,12–15 Apart from renal dysfunction, high Cys level may increase the risk for cardiac events in patients with peripheral arterial disesase or IHD.16,17 On the contrary, Cys levels may be increased by arterisoclerosis lesion or aging, although these factors have not been conclusively proven.18 In addition, Cys seems to reflect hemodynamic changes or myocardial damage.11,19 In the present study, we found a clear correlation between Cys level and cardiac events in patients with a history of HF admission.
The mechanism that allows UA and Cys to predict future cardiovascular events has been recently evaluated. Although the exact mechanism that causes Cys to be linked with the cardiovascular system was not suggested, cardiorenal hemodynamic interaction is primarily responsible for the relationship between Cys and the adverse outcomes.1,2,8,20 The level of Cys is elevated in proportion to the level of renal dysfunction. Cys levels also increase during cardiac injury, such as ischemic conditioning.11,21 Like Cys, UA is also associated with cardiovascular risk factors.5,9 This is because the xanthine oxidase system is stimulated by cardiac disorders. It remains unknown, however, whether UA and Cys at admission might be the active markers reflecting acute HF or only the result of the renal dysfunction. UA has been associated with inflammatory marker or oxidative products. Similarly, Cys level could show a significant correlation of the inflammatory markers suggesting that Cys generation might be accelerated by inflammatory cytokines.22,23 Indeed, from these findings, UA and Cys may provide the comprehensive information across a broad spectrum that could be observed with conventional renal measures.
With regard to renal measures, we demonstrated that only UA and Cys showed a strong prediction for outcomes. In fact, renal estimation by BUN, Cr, or GFR is usually reliable on stable conditions, but it is less useful when sudden renal function changes develop. However, UA and Cys could provide a more stable evaluation of renal function in patients with acute HF.10,24–26 Cys is a marker of renal function evaluation, but it has been a more sensitive tool for reflecting renal dysfunction, particularly even at the atmosphere of normal or mild elevation of Cr.27 Likewise, the role of UA as a predictior of outcome has been examined in previous many studies.5,9,28 In the current study, we included UA and Cys and compared them with conventional renal measures; their ability to determine adverse outcomes may vary slightly. Therefore, for a full comparision of their usefulness, further studies are needed.
While past studies have shown that the highest quartile of Cys was associated with clinical outcome more than the lowest quartile, the best cutoff point is still under debate.12,14,29 The current study showed the cutoff level of 0.8 mg/L of Cys, which is lower compared with levels determined by previous studies.12,25,30 One possible explanation is that the baseline characteristics of our study population showed the relatively normal or mildly decreased mean Cr and GFR levels, respectively. However, these findings may imply that cardiac events could increase steadily from low Cys and UA levels. Despite the cutoff value of 6.7 mg/dL of UA, the hazard ratio increased in proportion to the UA level. When it comes to exploration of the individual potential effect, we could not find individual exact values indicating the steep increasing hazard ratio, possibly due to the fact that clinical outcomes seem to occur even within normal ranges or at the low levels of UA and Cys. Thus, it may be desirable that we should consider both parameters for early identification of a subgroup of patients at risk for outcomes.
Regarding predicting prognosis, we found that combining UA and Cys would provide more useful information than each one alone. This is partially explained by the good correlation between two parameters. Cys showed the favorable correlation with UA (r=0.457, P<.001), which is as significant as those of GFR or Cr. Thus, the role of Cys could be comparable to those of UA. Particularly, when we graphically explored the impact of Cys and UA for the whole patients on Kaplan-Meier analysis, the clinical outcomes occurred more significantly than when using each parameter, even after more than 1.5 years of follow-up. Therefore, we found that the combination of both UA and Cys may allow long-term prognostic information, while each parameter could provide relatively short-term prognosis <1 year.