|Home | About | Journals | Submit | Contact Us | Français|
Hyperuricemia, a known correlate of oxidative stress, is a marker for adverse prognosis among individuals with heart failure. However, the relationship between hyperuricemia and the risk for incidence of heart failure in a community-based population has not been studied.
We prospectively analyzed the relationship between serum uric acid concentration at baseline and subsequent heart failure among the participants of the Framingham Offspring cohort (mean baseline age 36 years, women 52%). Using Cox regressions we calculated the risk of heart failure with increasing serum uric acid after adjusting for sex, age, smoking, body mass index, renal dysfunction, diuretics, systolic blood pressure, valvular heart disease, diabetes, alcohol, and use of anti-hypertensive medications. The incidence rates of heart failure was ~6 fold higher among those at the highest quartile of serum uric acid (>6.3 mg/dl) compared to those at the lowest quartile (<3.4mg/dl). The adjusted hazard ratio for the highest quartile of serum uric acid compared to the lowest was 2.1 (1.04–4.22). The relationship between hyperuricemia and heart failure was found in participants without metabolic syndrome and other subgroups as well.
Hyperuricemia is a novel, independent, risk factor for heart failure in a group of young general community dwellers. This has implications for development of preventive strategies for heart failure.
Nearly 5 million Americans currently suffer from heart failure and approximately 550,000 new cases of heart failure are now diagnosed each year1. Heart failure is associated with high risk of morbidity, mortality and hospital utilization in the United States2. The established risk factors for heart failure include male sex, hypertension, valvular heart disease, coronary artery disease, and obesity3. In spite of the progress made in its management, the mortality from heart failure remains high, underlining the need for identification of novel risk factors that may be amenable to intervention.
Earlier studies have shown that heart failure is often associated with hyperuricemia4, 5. Hyperuricemia is associated with worse hemodynamic measures such as increased left atrial pressure and decreased cardiac index among patients with primary pulmonary hypertension, cor pulmonale and dilated cardiomyopathy in a small case series6. Among those with established heart failure, hyperuricemia is a risk factor for adverse outcomes including mortality5, 7–15.
Serum uric acid may be useful for prognostication among those with preexisting heart failure5, 10–15. Hyperuricemia can predict heart failure among those with preexisting hypertension16. There have not been any studies that examined hyperuricemia as independent risk factors for heart failure risk among the general population. The single available study from Austria, did not account for confounders such as valvular heart disease and diuretics, and renal disease suggested that highest quantiles of serum uric acid was associated with elevated risk for death from heart failure17.
Hyperuricemia can be easily detected in routine medical care. If indeed presence of hyperuricemia provides additional information on future heart failure risk (over and above other risk factors), it has the potential to be a screening tool. Accordingly, we hypothesized that hyperuricemia is a risk factor for heart failure independent of other known risk factors.
We used data from the Framingham Offspring Study, a longitudinal observational study of children of the original Framingham Heart Study cohort and their spouses18. All participants of the Framingham Offspring Cohort that began in 1971 were eligible to be included in the present study. The data for our analyses were obtained from the National Heart Lung and Blood Institute (NHLBI) limited access dataset program. This analysis protocol was approved by the Stanford University institutional review board. We excluded all the subjects who did not have uric acid measurement.
Participants were under surveillance for cardiovascular events and were followed up approximately every four years by study visits that included medical review, physical examination and laboratory testing. In the present analyses we used data collected from the first through seventh study visit. The exact number of days from baseline to each study visit/outcome event was utilized in our analyses. The median follow up of this cohort was 29 years and the cumulative observation time was 135,991 person-years.
For analyses of the effects of serum uric acid, each observation started in the first (baseline) visit and ended at the day of outcome event, death or last contact with the study. In using such a definition, we acknowledge that the duration of hyperuricemia for each individual in the observation period is an underestimation of the true duration of hyperuricemia. In all analyses, the observation ended at the time of death, last contact or the outcome event.
Exercise, diet, drugs, and state of hydration, may result in transient fluctuations of uric acid levels and one measurement of uric acid may not be an accurate metric of the hyperuricemic `trait'. We examined this possibility in our data by calculating the probability of an individual participant changing the quartile of uric acid during the time interval between the first and second visit (i.e. transition probability) each of the uric aid stratum. Since this estimate was ~20%,, we deemed the variability to be too high and serum uric acid was measured at the first and the second visits and averaged to arrive at a mean value that replaced the single baseline measurement. Serum uric acid was assayed using the uricase method. Information on renal dysfunction, obesity measures, blood pressure, serum lipids, serum glucose, smoking, alcohol, aspirin, antihypertensive, and anti-diabetic medication use were available at all visits. Detailed information on individuals' diabetes/hypertesion medications such as name, dosage, duration of treatment were not available. For the purpose of this study, participants with a cardiac murmur at the time of the first study visit were assessed to have valvular heart disease, a risk factor for heart failure. Participants were evaluated for coronary artery disease at baseline and at subsequent visits by medical history, clinician assessment, and electrocardiogram.
The determination of renal dysfunction at baseline was made by the study physician. Serum creatinine or other laboratory measures of renal function was not available for this analysis. Gout was defined as a study physician diagnosis of definite gouty arthritis19.
Heart failure events (both hospitalized and non-hospitalized) were adjudicated by a study physician panel according to predetermined Framingham criteria shown in Table 1. 20, 21. Heart failure was considered to be present if two major or one major and two minor criteria were present in the absence of alternative explanation for the clinical picture (please see Table 1 for further details). There were no participants with heart failure at baseline.
Our primary analyses addressed the question-Does elevated serum uric acid independently predict the risk for incident heart failure? We used Cox proportional hazards regression model to study the relationship between baseline serum uric acid level, and heart failure. In these regressions, the time variable was defined as the period (number of days) from the baseline date to the date of incidence of heart failure or the date of last study visit. Observations of patients who did not die or develop heart failure were censored at the time of last observation. In the primary analyses the baseline values of the covariates were used to adjust for confounding. However, the relationship between hyperuricemia and other cardiovascular risk factors are complex (Figure 1) since hyperuricemia is a risk factor for kidney disease, hypertension, and atherosclerotic cardiovascular diseases22–27. Changes in health conditions over time such as increased blood pressure and worse renal function can potentially be a cause and a consequence of hyperuricemia. Thus using time varying measures of these covariates may be problematic. Therefore, in addition to Cox regressions with time-varying values of covariates, we preformed extensive stratified analyses such as for those who did not meet the ATP criteria for metabolic syndrome baseline28, and who died of any cause during the follow-up, and those who survived until the cut-off date for observation (visit 7).
Overall, of the 4989 participants in the Offspring study there were 4912 eligible participants with 196 incident cases of heart failure. Participants who developed heart failure were more likely to be older, male and with a worse traditional risk factor profile, have gout and currently used allopurinol, a uric acid reducing medication. These individuals had a greater prevalence of gout and higher serum uric acid concentration. Increasing serum concentrations of serum uric acid was associated with worse cardiovascular risk (Table 2).
Over the follow-up period, the cumulative incidence of gout were 12.6% (n=171) and 4.5% (n=192) among heart failure and no heart failure groups respectively (p<0.001). Overall, 155 participants with gout reported using allopurinol during the follow up. Only 2 participants without gout reported using allopurinol.
Figure 2 shows the heart failure-free survival curve. Those in the higher quartiles of serum uric acid had greater incidence of heart failure (Table 3). Proportional hazards assumptions were met in all the Cox regression models. In these models, increasing level of serum uric acid was associated with increased risk for heart failure, in unadjusted, and age-sex adjusted models (Table 3). In multivariable regressions, the increased risk relationship between uric acid level and heart failure was most evident in the highest quartile.
In this cohort of relatively young adults (median baseline age 36 years, inter-quartile range 28–44), the prevalence of documented coronary artery disease at baseline was infrequent (n=6) and exclusion of these individuals did not change our overall risk estimate. There were no participants with renal dysfunction at the baseline. Multivariable Cox regressions were performed for each of the following sub groups: participants who did not use diuretics, any blood pressure medications, non-diabetics, participants who did not develop renal dysfunction anytime during follow up, and those who did not develop the metabolic syndrome (Table 4). The link between hyperuricemia and heart failure was consistent across all these analyses.
When the multivariable analyses were repeated separately among the 892 participants who died during follow-up from any cause and those who survived until the 7th visit, each unit increase in serum uric acid increased the risk for incident heart failure for the deceased (n=125, hazard ratio 1.2(0.9–1.5)) and survivors (n=76, hazard ratio 1.1(0.9–1.5)) although neither reached statistical significance.
In multivariable regressions, the impact of such statistical interaction was tested for but was found to be statistically insignificant (p=0.21). Further, when data were analyzed for men and women separately, the risk estimates were greater than unity but not statistically significant in both groups owing to small number of events in each.
We report for the first time that hyperuricemia is a risk factor for heart failure in a large prospective study of a community dwelling population. Similar to the Vorarlberg study, this risk was most evident at serum uric acid levels greater than ~6 mg/dl- a cut-off point close to the solubility of urate in the normal human body17.
Our observation is not unexpected given the knowledge about the significance of hyperuricemia as a marker of abnormal oxidative metabolism29. Serum uric acid level is an index of oxidative stress in the human body 30. Serum uric acid is known to contribute to endothelial dysfunction by impairing nitric oxide production31. Serum uric acid has also been shown to be inversely correlated with the measures of functional capacity and maximal oxygen intake5. Among patients with chronic heart failure, serum uric acid concentrations are associated with greater activity of superoxide dismutase and endothelium dependent vasodliatation32.
Another potential pathophysiological link between hyperuricemia and heart failure might be through inflammation. Asymptomatic hyperuricemia is a pro-inflammatory state associated with higher levels of serum markers of inflammation, such as CRP, interleukin-6, and neutrophil count 31, 33 34. Among patients with heart failure, hyperuricemia is associated with higher levels of markers of endothelial activation such as the soluble intercellular adhesion molecule(ICAM)-1 and inflammatory markers such as interleukin-6, tumor necrosis factor- α and its receptors 12. Similar observations have been made in other population-based studies 35 and hospital-based studies 11, 12. The risk of heart failure was proportionate to the degree of elevation of serum uric acid among patients with gout 36. Locally, even when there is no active arthritis, the synovial fluid of patients with gout show low grade inflammatory activity37.
Elevated levels of serum uric acid among normal individuals predict hypertension38 38, 39, renal dysfunction 25 coronary artery disease22, and portends reduced life expectancy40. Lowering of serum uric acid with allopurinol can reduce blood pressure among hypertensives41, 42. This raises the possibility of the hyperuricemia-heart failure link being mediated by hypertension, a hypothesis that cannot be directly tested in observational studies such as ours. Nevertheless, other studies have shown that hyperuricemia is an independent risk factor for heart failure among those who already have hypertension 16. In our study, this link was consistently observed in a) time-varying Cox models where incident hypertension was adjusted for and b) in stratified analyses of participants who did not develop hypertension.
The significance of our observation lies in its use for developing a risk prediction rule for heart failure. While observations we have made raise the possibility of primary prevention of heart failure, the literature is conflicting on whether a reduction in serum uric acid will result in measurable clinical benefit among those with established heart failure43, 44. Some even argue that increased serum uric acid cause by diuretic use might have a beneficial role in itself 44. On the other hand, the uricosuric property of Losartan, an antihypertensive has been thought to have a beneficial effect among patients with hypertension and left ventricular hypertrophy in the LIFE study45. The putative mechanisms by which uric acid reduction treatments have shown benefit is also unclear. Specifically, it is unclear if the observed benefit from the use of Xanthine Oxidase inhibitors is mediated through reduction in serum uric acid levels or some other mechanism. Inhibition of Xanthine Oxidase enzyme by allopurinol has beneficial effects in terms of improved peripheral vasodilator capacity, systemic blood flow, and clinical outcomes 46 47. Randomized controlled studies have also been unclear about the putative benefit of allopurinol or its metabolite oxypurinol on established heart failure. While La Plata study showed improvement in left ventricular ejection fraction with use of allopurinol48, the OPT-CHF study did not show an overall benefit49. In our study, the majority of patients with gout were treated with allopurinol; the number of participants with gout but not on allopurinol was too few for meaningful comparison. If indeed allopurinol is protective from heart failure, the excess risk for serum uric acid we have found is likely to be an underestimate.
Limitations apply to our analysis. Our observational data on serum uric acid are essentially left-truncated. In other words, we know the severity of hyperuricemia but not the duration of hyperuricemia. Additionally, the long interval between follow-up visits (~4 years) may be too long to capture heart failure that results in death in shorter time. : The distribution of serum uric acid concentrations among men and women were different, the former having higher concentrations. Thus the lowest quartile of the pooled data was constituted mainly by women and the highest quartile by men. The gender-uric acid statistical interaction was insignificant, but limitations in statistical power precluded a more detailed analysis.
In summary, this large prospective study found that hyperuricemia is associated with greater incidence of heart failure. Future studies of various urate reduction strategies with adequate power to detect small improvement in clinical outcomes would be needed to determine whether, if at all, heart failure is preventable. Given the increasing prevalence and serious health impact of heart failure, even such small clinical benefit can translate into substantial public health benefit.
Heart failure is an incurable condition that is responsible for at least 287,000 deaths annually. Heart failure is the most common reason for hospitalization among people on Medicare and the number of hospitalizations have been increasing over time. We propose that hyperuricemia is a useful biomarker for estimating risk for heart failure and tested this hypothesis using the data from Framingham Offspring Study. We observed that the incidence of heart failure among those with serum uric acid concentrations >6.3 mg/dl was 6 fold higher than that among participants with serum uric acid <3.4 mg/dl. The adjusted risk for heart failure was double among those with hyperuricemia. Our findings have implications for early identification of those at risk for heart failure and put forward a new target for intervention.
Funding Sources: The Framingham Offspring Study (FOS) is conducted and supported by the National Heart Lung Blood Institute (NHLBI) in collaboration with the FOS Study Investigators. This manuscript was prepared using a limited access dataset Dr. Krishnan obtained from the NHLBI and does not necessarily reflect the opinions or views of the FOS or the NHLBI. Dr Krishnan conceived the manuscript idea, designed the analysis plan, performed statistical analysis, interpreted the results, drafted the manuscript and will serve as the guarantor.
This publication was made possible by in part Grant Number KL2 RR024154-01 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH), and NIH Roadmap for Medical Research. Its contents are solely the responsibility of the authors and do not necessarily represent the official view of NCRR, NHLBI or the NIH. Information on NCRR is available at http://www.ncrr.nih.gov/. Information on Re-engineering the Clinical Research Enterprise can be obtained from http://nihroadmap.nih.gov/clinicalresearch/overview-translational.asp. No commercial products are discussed in this manuscript.
Disclosures Dr. Krishnan has received grant support from Takeda Pharmaceuticals of North America Inc. Deerfield, IL. (formerly TAP Pharmaceutical Products, Inc) and had held stock in Savient Pharmaceuticals. He has served as an advisor/consultant for both these companies. Proprietary products manufactured by these companies are not named/discussed in this manuscript.