Article ID: CJ-15-0964
Background: Hyperuricemia is a prognostic factor in patients with chronic heart failure, but whether uric acid level can predict clinical outcome of acute heart failure (AHF) remains to be elucidated. We therefore investigated the association of uric acid with mortality in patients hospitalized for AHF.
Methods and Results: Data for patients hospitalized for AHF were drawn from an intramural registry. Biochemistry data, echocardiographic characteristics, and uric acid level were collected. National Death Registry was linked for the identification of mortality data. Among a total of 1,835 participants (age, 75±13 years, 68% men), 794 patients died during follow-up. Patients who died were older, had lower hemoglobin and estimated glomerular filtration rate, and higher pulmonary artery systolic pressure, NT-proBNP, and uric acid. Uric acid was a significant predictor of mortality on univariate analysis (HR per 1 SD, 1.18; 95% CI: 1.11–1.26) and in multivariate Cox models (HR, 1.15; 95% CI: 1.02–1.29). Survival analysis showed an increasing risk of death along the quartile distribution of uric acid level. Given renal function, cardiac performance, and kidney perfusion as major determinants of hyperuricemia, the prognostic impact of uric acid level was diminished as renal function deteriorated.
Conclusions: Uric acid level was an independent predictor of mortality in patients hospitalized for AHF, but the prognostic impact of hyperuricemia was attenuated by worsening renal function.
Hyperuricemia has been proposed as an independent prognostic factor in chronic heart failure (CHF) and atrial fibrillation,1–3 while hyperuricemia may be related to xanthine oxidase activation and the subsequent inflammation pathway.4 Hyperuricemia could be a consequence of overproduction of uric acid from purine metabolism or undersecretion of uric acid due to impaired renal function or the use of diuretics. Hyperuricema related to overproduction was suggested to be associated with clinical outcome,5 when uric acid level in patients with CHF and preserved renal function serves as a marker of increased xanthine oxidase activity. This scenario may have supported the therapeutic effect of xanthine oxidase inhibition in CHF and inspired further randomized controlled trials,6,7 but the majority of patients with CHF may either have impaired renal function or be treated with diuretics, and therefore uric acid level would be a reflection of renal function rather than of xanthine oxidase activity. It might conflict with the prognostic impact of hyperuricemia in the overall group of CHF patients with or without chronic kidney disease (CKD).
Acute heart failure (AHF) is considered as a separate entity, with epidemiology and pathophysiology distinct from CHF. Given the higher prevalence of cardiorenal syndrome and the use of diuretics in AHF patients, uric acid level was not related to clinical outcome of AHF, regarding mortality and re-hospitalization.8,9 Málek et al further noted an interaction between the prognostic value of hyperuricemia and the use of allopurinol in AHF,10 and complicated associations of uric acid level with comorbidity and outcome in AHF. Hence, in this study, we investigated the determinants of hyperuricemia and its prognostic impact in patients hospitalized for AHF.
The study population was drawn from an intramural administrative registry of Taipei Veterans General Hospital, conducted to record digitally the medical information of every patient since 2002. We set up the Heart Failure Registry of Taipei Veterans General Hospital (HARVEST registry) in 2009 to retrospectively analyze and prospectively enroll patients hospitalized for AHF, fulfilling the Framingham criteria and the guidelines.11,12 From October 2003 to December 2012, there were 6,650 hospitalizations primarily for AHF, among which, 3,261 hospitalizations of 2,663 patients who underwent echocardiography before discharge were eligible for this analysis. Among 2,663 eligible patients, 1,835 subjects who had fasting uric acid measurement constituted the subject group. Patients with acute coronary syndrome, severe hepatic disease, or severe infection were excluded. Data on medical history, biochemistry, echocardiographic characteristics, and medication were all obtained from a Web-based electronic medical recording system. The investigation conformed to the principles outlined in the Declaration of Helsinki, and it was approved by the institutional review board of Taipei Veterans General Hospital.
Left ventricular ejection fraction (LVEF) was measured using the 2-D M-mode modified Ellipsoid method,13 and E/e’ was calculated as the ratio of early ventricular filling flow velocity (E) to the septal mitral annulus tissue velocity (e’). Subjects with LVEF <50% were defined as having systolic heart failure. Hemogram, renal function, serum electrolytes, and N-terminal pro-brain natriuretic peptide (NT-proBNP) were measured immediately at presentation to hospital. Lipid profile and uric acid were checked at fasting the next morning. Estimated glomerular filtration rate (eGFR) was then calculated using the modified glomerular filtration rate estimating equation for Chinese patients.14 The ratio of blood urea nitrogen (BUN) to serum creatinine, BUN/Crea, was also obtained. Given that commercialized measurement of NT-proBNP (Roche Diagnostics, Basel, Switzerland) was available only after 2009, some subjects were missing NT-proBNP data in this analysis. Renin-angiotensin system (RAS) blockade was carried out with angiotensin-converting enzyme inhibitors or angiotensin II receptor antagonist. Anti-hyperuricemic agents were either allopurinol or benzbromarone.
Follow-upThe National Death Registry was linked to identify mortality in the study population with a follow-up duration of up to 4 years. The National Death Registry database registers valid information according to the International Classification of Disease, Ninth Revision (ICD-9). The ICD-9 codes for cardiovascular death were 390–459.15
Statistical AnalysisContinuous variables are expressed as mean±SD and comparisons between groups were conducted using Student’s t-test. Categorical data are given as absolute number and relative frequencies and were compared using chi-squared test. The correlates of uric acid level were examined on linear regression analysis, and the determinants of uric acid level were evaluated on multivariate linear regression analysis. Kaplan-Meier accumulated survival analysis was carried out using uric acid quartiles. Cox proportional hazards models were used to identify the predictors of all-cause mortality. Given that uric acid level differs with gender, we defined hyperuricemia according to the cut-offs of 7.2 mg/dl for men and 6.2 mg/dl for women in prognostic analysis. Based on KDOQI guidelines,16 we further conducted subgroup analysis to evaluate the interaction between renal function and uric acid. Statistical analysis was carried out using SPSS v.16.0 (SPSS, Chicago, IL, USA). All the tests performed were 2-sided, and P<0.05 was considered statistically significant.
Among 2,663 eligible subjects, 1.835 patients (mean age, 75±13 years; 68% men, 24.6% de novo HF) had fasting uric acid measurement on the next morning after admission, and they were enrolled in the present study. During the mean follow-up duration of 27.1±17.6 months, 794 patients died. The patients who died were older, had higher pulmonary artery systolic pressure (PASP), NT-proBNP and uric acid, and lower hemoglobin, eGFR and cholesterol (Table 1). Gender and underlying comorbidities, including hypertension, diabetes, coronary artery disease and atrial fibrillation were similarly distributed in both groups. During the index hospitalization, the use of i.v. nitroglycerin was more common, but the use of dobutamine was less common in the patients who survived. In addition, the patients who died had a higher prevalence of respiratory failure and ventilator support. Otherwise, the rate of treatment with vasopressors, intra-aortic balloon pump and hemodialysis was similar between the groups. Left ventricular performance indices, including LVEF, fractional shortening, and E/e’ also did not differ between the groups. Regarding medications, there were fewer prescriptions of RAS blockade, β-blockers, spironolactone, dihydropyridine calcium channel blockers (CCB), statin, thiazide, and loop diuretics in the patients who died compared with the patients who survived.
| Survival (n=1,041) | Death (n=794) | P-value | |
|---|---|---|---|
| Age (years) | 72.9±14.5 | 77.7±10.9 | <0.001 |
| Male gender | 697 (67.0) | 551 (69.3) | 0.31 |
| De novo HF | 273 (26.2) | 183 (23.0) | 0.12 |
| Comorbidity | |||
| Hypertension | 629 (60.5) | 463 (58.2) | 0.33 |
| Diabetes mellitus | 376 (36.2) | 313 (39.4) | 0.15 |
| Coronary artery disease | 322 (31.0) | 246 (30.9) | 1.00 |
| Atrial fibrillation | 298 (28.7) | 228 (28.7) | 1.00 |
| In-hospital management | |||
| I.v. nitroglycerin | 361 (34.6) | 160 (20.1) | <0.001 |
| Dobutamine | 157 (15.1) | 151 (19.0) | 0.02 |
| Vasopressors | 233 (22.4) | 203 (25.5) | 0.12 |
| Ventilatory support | 16 (1.5) | 26 (3.3) | 0.01 |
| IABP | 9 (0.9) | 8 (1.0) | 0.80 |
| Hemodialysis | 38 (3.6) | 35 (4.4) | 0.47 |
| Echocardiography | |||
| Systolic heart failure | 342 (37.8) | 381 (41.3) | 0.12 |
| LVEF (%) | 54.47±20.9 | 53.0±21.0 | 0.14 |
| Fractional shortening (%) | 30.1±13.3 | 29.1±13.4 | 0.13 |
| E/A ratio | 1.19±0.79 | 1.13±0.82 | 0.48 |
| Septal E/e’ | 17.9±8.2 | 18.2±8.3 | 0.66 |
| PASP (mmHg) | 42.7±15.9 | 46.4±17.0 | <0.001 |
| Biochemistry | |||
| Hemoglobin (g/dl) | 12.1±2.3 | 11.5±2.1 | <0.001 |
| eGFR (ml/min/1.73 m2) | 56.5±31.2 | 45.9±26.5 | <0.001 |
| BUN/Crea | 19.7±7.5 | 20.8±7.9 | 0.003 |
| Sodium (mEq/L) | 139.2±4.3 | 138.8±5.1 | 0.07 |
| Potassium (mEq/L) | 4.0±0.6 | 4.1±0.6 | 0.83 |
| Total cholesterol (mg/dl) | 159.1±43.1 | 153.1±43.0 | 0.01 |
| LDL-C (mg/dl) | 97.2±33.6 | 91.8±34.7 | 0.004 |
| HDL-C (mg/dl) | 43.5±14.5 | 42.6±14.7 | 0.24 |
| Log NT-proBNP (pg/ml), n=501† | 4,838.3±3.6 | 7,956.0±3.5 | <0.001 |
| Uric acid (mg/dl) | 8.3±2.8 | 9.0±3.1 | <0.001 |
| Post-discharge medications | |||
| β-blocker | 722 (69.4) | 452 (56.9) | <0.001 |
| RAS blockade | 902 (86.7) | 636 (80.0) | <0.001 |
| Spironolactone | 619 (59.5) | 430 (54.1) | 0.02 |
| Dihydropyridine CCB | 562 (53.9) | 367 (46.1) | 0.001 |
| Antiplatelet | 733 (70.5) | 557 (70.1) | 0.87 |
| Warfarin | 272 (26.2) | 183 (23.0) | 0.12 |
| Statin | 461 (44.3) | 257 (32.3) | <0.001 |
| Digoxin | 356 (34.2) | 299 (37.6) | 0.14 |
| Thiazide | 485 (46.5) | 315 (39.6) | 0.003 |
| Loop diuretics | 869 (83.4) | 620 (77.9) | 0.003 |
| Anti-hyperuricemic agents | 432 (41.5) | 305 (38.3) | 0.09 |
Data given as mean±SD, †geometric mean±SD or n (%). BUN/Crea, blood urea nitrogen/serum creatinine; CCB, calcium channel blocker; E/A ratio, early/late ventricular filling velocities; E/e’, ratio of early ventricular filling velocity to early diastolic tissue velocity mitral annulus; eGFR, estimated glomerular filtration rate; HDL-C, high-density lipoprotein cholesterol; IABP, intra-aortic balloon pump; LDL-C, low-density lipoprotein cholesterol; LVEF, left ventricular ejection fraction; log NT-proBNP, log-transformed N-terminal prohormone brain natriuretic peptide; PASP, pulmonary artery systolic pressure; RAS, renin-angiotensin system; vasopressor, norepinephrine and dopamine.
The risk factors related to all-cause mortality are listed in Table 2. Age, PASP, hemoglobin, eGFR, sodium, uric acid and NT-proBNP were all significant predictors of all-cause death. In addition, prescription of RAS blockades, β-blockers, spironolactone, and statin significantly reduced mortality. Kaplan-Meier analysis indicated an increasing risk of death with quartile distribution of uric acid level, with cumulated survival rates of 62.2%, 61.0%, 55.1% and 48.2%, respectively, in a 4-year follow-up duration (Figure 1). In addition, the associated risks of hyperuricemia were also statistically significant in patients with reduced and preserved LVEF (Figure 2). On multiple Cox proportional hazard modeling, uric acid level was an independent predictor of mortality (HR per 1 SD, 1.146; 95% CI: 1.046–1.256) after accounting for age, gender, PASP, hemoglobin, eGFR, sodium, and prescribed medications of β-blocker, RAS blockade and spironolactone (Table 2, model 2). With further adjustments for NT-proBNP, uric acid level remained a significant prognostic factor of death (HR, 1.321; 95% CI: 1.053–1.659; Table 2, model 3) When the use of statin, dihydropyridine CCB, diuretics, and anti-hyperuricemia agents were considered, uric acid level was still predictive of mortality (HR, 1.318; 95% CI: 1.040–1.671).
| Model 1 | Model 2 | Model 3 | |
|---|---|---|---|
| HR (95% CI) | HR (95% CI) | HR (95% CI) | |
| Uric acid (1 SD=3 mg/dl) | 1.18 (1.10–1.26) | 1.15 (1.02–1.29) | 1.32 (1.05–1.65) |
| Age (1 SD=13.3 years) | 1.38 (1.27–1.51) | 1.30 (1.14–1.48) | 1.71 (1.28–2.29) |
| Sex (M vs. F) | 1.08 (0.93–1.26) | 1.11 (0.89–1.39) | 0.87 (0.56–1.35) |
| Hemoglobin (1 SD=2.26 g/dl) | 0.79 (0.73–0.85) | 0.84 (0.75–0.95) | 0.96 (0.76–1.21) |
| eGFR (1 SD=29.7 ml/min/1.73 m2) | 0.73 (0.67–0.79) | 0.97 (0.94–0.99) | 1.00 (0.95–1.05) |
| Sodium (1 SD=4.7 mEq/L) | 0.90 (0.84–0.96) | 0.92 (0.83–1.01) | 0.75 (0.60–0.93) |
| Log NT-proBNP (1 SD=0.56 pg/ml) | 1.44 (1.21–1.71) | – | 1.40 (1.12–1.74) |
| PASP (1 SD=16.59 mmHg) | 1.23 (1.14–1.31) | 1.16 (1.05–1.29) | 0.99 (0.80–1.23) |
| β-blocker | 0.64 (0.56–0.74) | 0.77 (0.61–0.96) | 0.84 (0.52–1.34) |
| RAS blockade | 0.62 (0.52–0.74) | 0.81 (0.61–1.06) | 0.74 (0.43–1.29) |
| Spironolactone | 0.80 (0.70–0.92) | 1.05 (0.84–1.30) | 1.09 (0.71–1.67) |
Model 1, univariate Cox proportional regression analysis; model 2, multivariate Cox proportional regression analysis; model 3, model 2 with further adjustment for NT-proBNP. Abbreviations as in Table 1.
Kaplan-Meier survival curve analysis according to uric acid quartile (quartile 1, <6.5 mg/dl; quartile 2, 6.5–8.3 mg/dl; quartile 3, 8.3–10.4 mg/dl; and quartile 4, ≥10.4 mg/dl).
Kaplan-Meier survival curve analysis according to uric acid quartile, in patients with reduced left ventricular ejection fraction (LVEF; quartile 1, <7.0 mg/dl; quartile 2, 7.0–9.0 mg/dl; quartile 3, 9.0–11.0 mg/dl; and quartile 4, ≥11.0 mg/dl); and patients with preserved LVEF (quartile 1, <6.4 mg/dl; quartile 2, 6.4–8.0 mg/dl; quartile 3, 8.0–9.9 mg/dl; and quartile 4, ≥9.9 mg/dl).
With regard to 362 patients who died of cardiovascular causes in this study, uric acid level was related to cardiovascular death (HR, 1.223; 95% CI: 1.120–1.334). After accounting for age, gender, PASP, hemoglobin, eGFR, sodium, and prescribed medications for HF, uric acid level remained an independent predictor of cardiovascular death (HR, 1.146; 95% CI: 1.002–1.311).
In addition, gender-defined hyperuricemia was also associated with both all-cause and cardiovascular mortality (HR, 1.233; 95% CI: 1.049–1.449; and HR, 1.531; 95% CI: 1.211–1.936, respectively).
Determinants of Uric AcidTo investigate the correlates of uric acid level, we performed a stepwise multiple linear regression analysis with age, hemoglobin, sodium, eGFR, BUN/Crea, high-density lipoprotein cholesterol, LVEF, PASP, E/A ratio, and NT-proBNP. Only eGFR, BUN/Crea and LVEF were identified as predictors of uric acid level, and the attributable proportions were 73.4%, 17.3%, and 9.3%, respectively (Figure 3).
Determinants of uric acid level and their proportional contributions, calculated using multiple linear regression stepwise analysis. BUN/Crea, blood urea nitrogen/creatinine; eGFR, estimated glomerular filtration rate; LVEF, left ventricular ejection fraction.
When the subjects were categorized according to KDOQI guidelines,16 the prognostic correlation of uric acid was abolished with progressively worsening renal function (Figure 4). When adjustment was made for age, sex, hemoglobin, eGFR, PASP, and sodium, uric acid level was related to mortality only in patients with CKD stage I/II but not stage III, IV or V.
Hazard ratio and 95% CI per 1 SD increase of uric acid level for all-cause mortality according to chronic kidney disease (CKD) stage, accounting for age, sex, hemoglobin, estimated glomerular filtration rate, pulmonary artery systolic pressure, and sodium.
This study clearly demonstrated the independent prognostic value of uric acid in patients hospitalized for AHF with either reduced or preserved LVEF. We further showed that renal function, LV performance and kidney perfusion were major determinants of uric acid level. Given the high correlation of uric acid level with renal function, a significant interaction was seen between renal function and uric acid in predicting mortality of AHF. Uric acid level would especially be an indicator of clinical outcome in AHF subjects with preserved renal function.
Uric Acid as a Clinical Risk Factor in AHFWhile hyperuricemia has been shown to be an independent risk for unfavorable outcome in patients with CHF,17,18 pathophysiologic mechanisms beyond hemodynamic impairment, involving imbalance in neurohormone, immune and metabolic functions, have been suggested.2 Pharmacological intervention for hyperuricemia with allopurinol would improve the prognosis of CHF, and it even showed a dose-response relationship.19 A further experimental study also supported the therapeutic advantage of xanthine oxidase inhibition, which might attenuate LV remodeling processes and dysfunction, independent of uric acid reduction.20
Data regarding the prognostic value of uric acid in patients hospitalized for AHF, however, are contradictory.8–10,21,22 Pascual-Figal et al showed that uric acid was a prognostic marker in AHF patients with reduced LVEF during a mean follow-up of 2 years.22 It seemed that there was a threshold effect rather than a dosing effect of uric acid level in predicting mortality and/or recurrent HF. Given the relatively small sample size, it is possible that over-adjustment may have occurred in the multivariate Cox proportional hazard models.22 In another study of 1,255 patients in the Czech AHF registry, Málek et al noted an interaction between uric acid level and the use of allopurinol in predicting clinical outcome,10 but a detrimental rather than beneficial effect of using allopurinol for hyperuricemia was observed. A recent randomized double-blind placebo-controlled study also failed to demonstrate the benefit of allopurinol in hyperuricemic HF patients with reduced ejection fraction.7 Although hyperuricemia was related to short-term and long-term mortality, the independence of the prognostic value of uric acid levels was not evaluated on multivariate modeling in the Málek et al study.10
In contrast, in the present study of 1,835 patients with a mean follow-up >2 years, we confirmed that admission uric acid level was an independent predictor of long-term all-cause and cardiovascular mortality in AHF. In addition, the associations between uric acid level and clinical outcome remained statistically significant in patients with either reduced or preserved LVEF.
Hyperuricemia and CKDFilippatos et al analyzed 2,645 chronic systolic HF patients, who participated in the Beta-Blocker Evaluation of Survival Trial; they found that hyperuricemia was associated with all-cause mortality or re-hospitalization for HF only in patients with preserved renal function rather than in those with CKD.5 While kidney is the major organ to excrete uric acid, there is, theoretically, an association of uric acid level with renal function. Filippatos et al further proposed that only hyperuricemia in subjects with preserved renal function might represent hyperactive xanthine oxidase, and therefore it was related to clinical outcome.5 The “under-secreted” hyperuricemia in patients with CKD or treated with diuretics might carry little information related to outcome. Málek et al, however, noted the conflicting result that hyperuricemia was related to long-term mortality only in subjects with eGFR=30–60 ml/min/1.73 m2 but not in those with eGFR <30 or >60 ml/min/1.73 m2.10 In the present study, we demonstrated an obvious attenuation of prediction of long-term mortality for uric acid level with deteriorating renal function in patients with AHF, which was consistent with the majority of published data in patients with CHF. The results suggest that hyperuricemia in patients with CKD has a more complex relationship. Further studies are needed to clarify the role of uric acid as an indicator for risk classification in AHF with or without CKD.
Determinants of HyperuricemiaHyperuricemia is a complex presentation in patients with CHF, involving not only renal function but also age, gender, functional class, metabolism, use of diuretics, and cardiovascular hemodynamics.22,23 It was further suggested that ischemic kidney damage might be the dominant factor for hyperuricemia in AHF.24 In the present study, the major determinants of uric acid level were eGFR, BUN/Crea, and LVEF on multiple linear regression stepwise analysis. In addition to renal function, we also showed that ventricular performance independently correlated with uric acid level, consistent with the previous findings in either CHF or AFH.10,23 While BUN/Crea was suggested as an indicator of renal perfusion,25–27 acute ischemic kidney challenge in AHF was also identified as a determinant of uric acid level.24 Therefore, we may conclude that hyperuricemia is a complicated phenomenon involving cardiac performance, kidney function, and cardiorenal interaction.
Study LimitationsStudy limitations were as follows. First, given the nature of an observational study, selection bias would have been present due to unobserved variables, but we adjusted for all available confounders to confirm the independent prognostic value of hyperuricemia in AHF. Second, NT-proBNP data were available only for 27% of the subjects. In this subset of 501 patients with available NT-proBNP data, there were 139 mortalities. We therefore still had sufficient power to demonstrate that hyperuricemia was related to outcome, independent of NT-proBNP. Third, uric acid was measured routinely at fasting in hospital.28 Although uric acid level would have been affected by the use of loop diuretics at the very first presentation of AHF, such information bias of non-differential measurement errors toward the null may strengthen the study results.29 Last, the study lacked incidental morbidities, such as re-hospitalization for HF. Further work is needed to address the risks of hyperuricemia for mortality and morbidity.
In patients hospitalized for AHF with either reduced or preserved LVEF, hyperuricemia correlated with increased mortality, independent of traditional risk factors, as well as of NT-proBNP. This may be because hyperuricemia presentation involves renal function, cardiac performance, and kidney perfusion in AHF. Worsening renal function would hence attenuate the prognostic value of uric acid level.
The study was supported by Taipei Veterans General Hospital (V103B-017), and Ministry of Health and Welfare, Taiwan with grant (MOHW-104-TDU-B-211-113003) and the death registry.
None.
