The prognostic impact of the uric acid level in patients who require cardiovascular intensive care – is serum uric acid a surrogate biomarker for critical patients in the non-surgical intensive care unit?

Background:

The prognostic impact of hyperuricemia and the factors that induce hyperuricemia in cardiovascular intensive care patients remain unclear.

Methods and results:

A total of 3257 emergency department patients were screened, and data for 2435 patients who were admitted to an intensive care unit were analyzed. The serum uric acid level was measured within 15 min of admission. The patients were assigned to a low-uric acid group (uric acid ⩽7.0 mg/dl, n=1595) or a high-uric acid group (uric acid >7.0 mg/dl, n=840) according to their uric acid level on admission. Thereafter, the patients were divided into four groups according to the quartiles of their serum uric acid level (Q1, Q2, Q3 and Q4), and uric acid levels and Acute Physiology and Chronic Health Evaluation II (APACHE II) score. A Kaplan–Meier curve showed a significantly lower 365-day survival rate in a high-uric acid group than in a low-uric acid group, and in Q3 than in Q1 or Q2 and in Q4 than in the other groups. The multivariate logistic regression model for 30-day mortality identified Q4 (odds ratio: 1.856, 95% confidence interval (CI) 1.140–3.022; p=0.013) as an independent predictor of 30-day mortality. The area under the receiver-operating characteristic curve values of the serum uric acid level and APACHE II score for the prediction of 30-day mortality were 0.648 and 0.800, respectively. The category-free net reclassification improvement and integrated discrimination improvement showed that the calculated risk shifted to the correct direction by adding the serum uric acid level to the APACHE II score (0.204, 95% CI 0.065–0.344; p=0.004, and 0.015, 95% CI 0.005–0.025; p=0.004, respectively). The prognosis, including the 365-day mortality, among patients with a high uric acid level and a high APACHE II score was significantly poorer in comparison with other patients.

Conclusion:

The serum uric acid level, which might be elevated by the various critical stimuli on admission, was an independent predictor in patients who were emergently hospitalized in the intensive care unit. The serum uric acid level is therefore useful as a surrogate biomarker for critical patients in the intensive care unit.

Introduction

The serum uric acid (UA) level has been recognized to predict the future incidence of hypertension, diabetes mellitus, dyslipidemia and metabolic syndrome.^1,–3 Hyperuricemia has therefore been reported to be closely associated with the carotid intima-media thickness, coronary artery atherosclerosis (calcification) and arterial stiffness,^4,–6 leading to life-threatening diseases such as cardiovascular disease, stroke and chronic kidney disease (CKD) with hemodialysis. Thus, hyperuricemia sometimes leads to adverse outcomes. The mechanisms through which the serum UA levels are elevated in these disease conditions had been considered to be the excessive intake of high-purine foods or failed excretion by the kidney. In our previous report, although hyperuricemia itself was an independent predictor of the long-term prognosis in patients with acute heart failure (AHF),⁷ the prognostic value of hyperuricemia was not associated with the atherosclerotic risk factors of AHF patients.⁸ Thus, another mechanism might be involved in the UA elevation in AHF patients.

Xanthine oxidoreductase (XOR), which is the generic term for xanthine oxidase (XO) and xanthine dehydrogenase (XDH), was newly suggested to be involved in the mechanism underlying the elevation in the serum UA level.⁹ Following the establishment of a method for measuring XOR in the human body, several reports were published on the XOR levels in patients with chronic heart failure and cardiac disease as well as in normal volunteers.^9,–11 Some tissue stimuli were reported to directly induce the mobilization of XDH to the blood.¹² Recently, the serum UA level was reported to be associated with the prognosis of clinically ill patients.¹³ In that study, the elevation of the UA level due to clinical stimuli was reported to have a prognostic impact in patients with sepsis and acute respiratory distress syndrome. We therefore hypothesized that various critical stress factors induce the elevation of serum UA, thereby leading to adverse outcomes. In the present study, we investigated the clinical data of patients receiving emergency care to test this hypothesis.

Methods

Subjects

We screened 3257 patients who visited the emergency room at Nippon Medical School Chiba Hokusoh Hospital and who required a medical examination by a cardiologist belonging to the intensive care unit (ICU), or who were admitted to the non-surgical ICU from general wards between May 2011 and February 2017. One hundred and eight patients who were re-admitted to the non-surgical ICU during the same hospitalization, 43 patients who did not undergo UA measurement within 15 min after admission, two patients whose in-hospital mortality could not be evaluated because they had been transferred to other institutes, and 669 patients who were admitted to the general ward or who returned home were excluded from the present study. Ultimately, a total of 2435 patients were enrolled in the present study (Figure 1).

The enrolled patients were admitted to the non-surgical ICU and all patients were treated by a cardiologist. There are two ICU departments (surgical or non-surgical) in our institute; both of them are ‘closed ICUs’. All of the physicians in the non-surgical ‘closed ICU’ are cardiologists. Thus, patients who were admitted to the surgical ICU (i.e. trauma, burn, drowing and cerebrovascular disease) were basically excluded from the present study. Patients with the following diseases were admitted to the non-surgical ICU in the present study: acute coronary syndrome (ACS), AHF, arrhythmia, acute aortic dissection (AAD), infectious disease (severe sepsis, septic shock, infective endocarditis, pneumonia, pericarditis or myocarditis), pulmonary embolism, coronary spasm angina, Takotsubo cardiomyopathy, respiratory emergent disease, acute kidney injury, and patients who had severe symptoms that required a differential diagnosis.

Serum UA measurement

Serum samples were obtained from all 2435 patients on the day of admission. The blood samples were collected into tubes within 15 min of admission and were centrifuged within 30 min of admission. If it was difficult to examine the blood samples on the day of sampling, they were cooled to 2–10°C. The samples of patients who were admitted between 09:00 h and 17:00 h (day-time) were usually examined immediately, while those of patients who were admitted between 17:00 h and 09:00 h (night-time) were examined after being cooled to 2–10°C. The timing of the examination did not affect the data. An absorptiometry kit (Sekisui Medical Company, Tokyo, Japan) was used to measure the UA level based on the hydrogen peroxide produced in the chemical reaction that occurs when UA is combined with uricase. In the present study, a UA level of >7.0 mg/dl was defined as high, according to the Japanese guidelines.¹⁴

Procedure

The 2435 ICU patients were divided into: the low-UA group (UA ⩽7.0 mg/dl; n=1595) and the high-UA group (UA >7.0 mg/dl; n=840) according to their serum UA level on admission.

The age, gender, etiology, medical history (diabetes mellitus, hypertension, dyslipidemia and hyperuricemia), vital signs and status (systolic blood pressure (SBP), diastolic blood pressure, pulse, respiratory rate, body temperature, body mass index and left ventricular ejection fraction (LVEF) upon admission), arterial blood gas (pH, PCO₂, PO₂, HCO₃⁻, SaO₂ and lactate), laboratory data (white blood cell (WBC) count, hemoglobin, blood urea nitrogen (BUN), creatinine, sodium, potassium, blood glucose, C-reactive protein (CRP) and brain natriuretic peptide (BNP)) and mechanical support during the ICU stay (non-invasive positive pressure ventilation (NPPV), endotracheal intubation (ETI), intra-aortic balloon pumping, percutaneous cardiopulmonary support (PCPS) and continuous hemodiafiltration (CHDF)) were compared between the low- and high-UA groups. The acute Physiology and Chronic Health Evaluation (APACHE II) score¹⁵ was also compared between these two groups.

The patients were further assigned to four groups according to the quartile of the UA level (Q1 (UA ⩽5.0 mg/dl, n=624), Q2 (UA 5.1–6.2 mg/dl, n=611), Q3 (UA 6.3–7.6 mg/dl, n=598) and Q4 (UA ⩾7.7, n=602)). The same clinical findings and treatments were compared among the four quartile groups.

The prognosis and the serum UA level

The long-term prognosis, including all-cause death within one year (365-day mortality), was evaluated using Kaplan–Meier curves and Cox regression analysis. Thereafter, the short-term prognosis (30-day mortality) was assessed by a multivariate logistic regression model. The patients were routinely followed up at an outpatient clinic. Among the patients who were followed up at other institutes, the prognosis was determined by telephone contact. The prognostic value of the serum UA levels for predicting 365-day mortality was compared in the high-UA and low-UA group using Kaplan–Meier curves. The survival rates according to the serum UA quartiles were also analyzed using Kaplan–Meier curves. Based on the results, a multivariate logistic regression analysis was performed to obtain the odds ratios (ORs) for 30-day mortality. To evaluate the short-term prognostic impact of the serum UA level, the configuration factors of the APACHE II score system¹⁵ (per one-point increase) and treatment strategy (performed cardiopulmonary resuscitation (CPR), use of vasopressors (i.e. dopamine, dobutamine, noradrenaline, adrenaline and vasopressin) and use of antibiotics (yes or no)) were selected for inclusion in a multivariate logistic regression model.

The receiver-operating characteristic (ROC) curves for the serum UA level and APACHE II score were calculated to predict the optimal cut-off values for predicting 30-day mortality, and the sensitivity, specificity and area under the ROC curve (AUC) were determined to indicate the optimal values for predicting 30-day mortality. The value of the addition of the UA measurement to the APACHE II score for predicting 30-day mortality was also investigated by calculating the change in the AUC.

We finally focused on the association between the UA level and the APACHE II score. Another set of four groups was constructed based on the serum UA levels and APACHE II score. The cut-off value of the serum UA level was defined by the low- or high-UA groups,¹⁴ and that of the APACHE II score used the optimal cut-off values determined by the ROC curve. The survival rates of these four groups were analyzed using Kaplan–Meier curves.

Statistical analyses

All statistical analyses were performed using the SPSS (version 22.0 J, SPSS Japan Institute, Tokyo, Japan) and Stata version 14 (Stata Corp., College Station, TX, USA). All numerical data were expressed as the median (range or 25–75% interquartile range) depending on the variable distribution. The Mann–Whitney U test was used to compare two groups, and the Kruskal–Wallis test was used to compare four groups. Normality was assessed using the Shapiro–Wilk W test. Comparisons of proportions were performed using chi-squared test for two groups and Pearson’s bivariate test for four groups. The prognostic value (30-day mortality) of the serum UA level in the Q2, Q3 and Q4 groups in comparison with the Q1 as the referent was assessed using a multivariate logistic regression model. The multivariate logistic regression analysis was performed to obtain the ORs for 30-day mortality by backward stepwise selection. ROC curves were calculated to determine the cutoff values, and the sensitivity, specificity and AUC were determined. p values of < 0.05 were considered to indicate statistical significance. The category-free net reclassification improvement (cf-NRI) and the integrated discrimination improvement (IDI) were performed to evaluate the prognostic value of the addition of the UA measurement to the APACHE II score for predicting 30-day mortality. cf-NRI counts the direction of change in the calculated risk for each patient (i.e. either +1 or –1 is counted depending on whether the change is in the correct direction (higher for those with events, lower for those without events)). The IDI separately considers the actual change in the calculated risk for each patient with and without events. Kaplan–Meier survival curves were used to estimate the cumulative mortality rates in the two groups (low- or high-UA), the four quartiles (Q1, Q2, Q3 and Q4) and another set of four groups. The significance of differences was calculated using a log-rank test. p values of < 0.05 were considered to indicate statistical significance. The long-term prognostic value (365-day mortality) of the serum UA levels in the Q2, Q3 and Q4 groups compared with those in the Q1 as the referent was assessed using a Cox regression hazard model. A Cox regression analysis was performed to obtain the hazard ratio for 365-day mortality.

Ethical concerns

The research ethics committee at Nippon Medical School Chiba Hokusoh Hospital approved the study protocol.

Results

Serum UA levels in ICU patients

The distribution of the serum UA levels in the ICU patients is described in Figure 2. Of the 2435 patients admitted to receive cardiovascular intensive care, a majority (n=1658, 68.1%) had serum UA levels of 4–8 mg/dl. The serum UA levels were ⩽ 4 mg/dl in 262 patients (10.8%) and ⩾ 8.0 mg/dl in 515 patients (21.1%).

The ICU patient cohort consisted of 1765 (72.5%) male patients (median age: 70 years). A total of 1105 (45.4%) patients had ACS, 522 (21.4%) had AHF, 147 (6.0%) had AAD, 182 (7.5%) had arrhythmia, 160 (6.6%) had infectious diseases, 74 (3.0%) had pulmonary embolism and 245 (10.0%) had other diseases, including coronary spasms (n=39, 1.6%), Takotsubo cardiomyopathy (n=29, 1.2%) and other diseases requiring intensive care (n=177, 7.3%) (Table 1). The relationships between the patients’ characteristics and the serum UA levels (low or high UA) on admission are shown in Table 1. The rate of male subjects in the high-UA group was significantly higher than that in the low-UA group (p <0.001). Interestingly, in comparison with the low-UA group the incidence of ACS was significantly lower (p <0.001) and the incidence of and AHF was significantly higher (p <0.001) in the high-UA group. The SBP and pulse in the high-UA group were significantly (both, p <0.001) lower than those in the low-UA group, and the LVEF in the high-UA group was significantly (p<0.001) lower than that in the low-UA group. In comparison with the low-UA group, the pH levels were significantly lower (p <0.001) and the lactate levels were significantly higher (p<0.001) in the high-UA group. In the high-UA group, the serum creatinine, BUN, potassium, CRP and BNP levels were significantly higher than those in the low-UA group (all p <0.001).

Interestingly, the APACHE II scores of the high-UA group were significantly higher than those of the low-UA group (p<0.001); furthermore, the mechanical support (including NPPV, ETI, PCPS and CHDF) during the ICU stay was significantly more likely to be required in the high-UA group than in the low-UA group (p <0.001). Almost the same result was obtained in the analysis of the quartile groups (Table 2). These results suggest that the high-UA group included a greater proportion of critical patients than the low-UA cohort.

The prognostic value of the serum UA level

The Kaplan–Meier survival curves, including for all-cause death, of the serum UA groups are shown in Figure 3. The 365-day survival rates in the low-UA and high-UA groups were 87.2% and 75.1%, respectively; those in Q1, Q2, Q3 and Q4 were 88.0%, 89.4%, 82.3% and 72.3%, respectively. The survival rate of the high-UA group was significantly lower than that of the low-UA group (p <0.001) (Figure 3(a)). The survival rates of the Q1 and Q2 groups were similar, that of the Q3 group was significantly lower than those of the Q1 and Q2 groups (p=0.006 and p <0.001, respectively) and that of the Q4 group was significantly lower than those of the Q1, Q2 and Q3 groups (all p <0.001).

The factors predicting 30-day mortality were assessed by a multivariate logistic regression analysis (Table 3). The multivariate logistic analysis indicated that Q4 (ORs: 1.856, 95% CI 1.140 to 3.022; p=0.013), GCS (per 1-point increase, ORs: 1.052, 95% CI 1.006 to 1.101; p=0.025), pulse (ORs: 1.151, 95% CI 1.014 to 1.307; p=0.030), respiratory rate (ORs: 1.381, 95% CI 1.212 to 1.574; p<0.001), body temperature (per 1-point increase; ORs: 1.367, 95% CI 1.040–1.796; p=0.025), WBC count (per 1-point increase; ORs: 1.402, 95% CI 1.065–1.845; p=0.016), creatinine (per 1-point increase; ORs: 1.265, 95% CI 1.130–1.416; p<0.001), sodium (per 1-point increase; ORs: 1.491, 95% CI 1.109–2.004; p=0.008), age (per 1-point increase; ORs: 1.180, 95% CI 1.052–1.323; p=0.005), performed CPR (yes; ORs: 1.751, 95% CI 1.017–3.014; 0.043) and use of vasopressors (yes; ORs: 4.857, 95% CI 3.305–7.138; p<0.001) were independent predictors of 30-day mortality.

The serum UA level and the APACHE II score

The value of serum UA that produced the optimal balance between sensitivity and specificity for predicting 30-day mortality (71.0% and 51.0%; AUC=0.648, p <0.001) was 6.2 mg/dl (Figure 4); meanwhile, the APACHE II score that produced the optimal balance between sensitivity and specificity (77.5% and 68.6%; AUC=0.800, p <0.001) was 14.0 points (Figure 4). The AUC was not significantly increased by the addition of the serum UA level to the APACHE II (0.805, 95% CI 0.774–0.837, p=0.356 vs. APACHE II alone). However, the addition of the serum UA level to the APACHE II score resulted in a significant cf-NRI and IDI (0.204, 95% CI 0.065–0.344, p=0.004 and 0.015, 95% CI 0.005–0.025, p=0.004, respectively) (Figure 4).

Based on the results of the ROC curve analysis, we divided the patients into four groups according to their UA level and APACHE II score. The Kaplan–Meier survival curves for both a low (⩽14) and a high (⩾15) APACHE II score showed that the prognosis, including all-cause death, in the high-UA (⩾ 7.1 mg/dl) patients was significantly poorer than that in the low-UA patients (⩽7.0 mg/dl) (p <0.001 in the low APACHE II group, p=0.016 in the high APACHE II group) (Figure 5). Interestingly, even if the patients had a high UA level, the prognosis, including all-cause death, was significantly better in those with a low APACHE II score than in those with a high APACHE II score. Having both a high APACHE II score and a high UA level was the independent factor that had the greatest influence on poor 365-day mortality in patients requiring intensive care (hazard ratio: 6.690, 95% CI 5.250–8.852; p<0.001).

Discussion

An elevated UA level was a predictor of 30-day and 365-day mortality in patients who required cardiovascular intensive care. Elevated UA levels, which have the potential to lead to a poor prognosis, were found to be associated with critical situations in our patients. Patients with both elevated UA levels and a high APACHE II score showed the worst outcome of the four UA/APACHE II combination groups. Of note, the serum UA elevation and a low APACHE II score was not associated with an adverse outcome. Elevated UA levels caused by acute severe stress on admission might be associated with a poor prognosis in patients who require cardiovascular intensive care, while elevated UA levels in association with atherosclerosis might not be associated with a poor prognosis. These findings suggest that the serum UA levels are a useful surrogate biomarker for patients who require intensive care.

The mechanisms underlying UA elevation

As UA is the final product of purine metabolism, the serum level of UA can easily be increased by diets containing a large amount of purine. Elevated UA levels have therefore been suggested to be associated with hypertension, diabetes, dyslipidemia, obesity and metabolic syndrome.^16,17 Patients with these diseases also tend to be complicated with atherosclerosis-associated diseases. Thus, the serum UA level has long been indicated to be associated with atherosclerosis.

However, patients with elevated serum UA levels included patients with excessive UA production and those with decreased excretion. It was reported that hyperuricemia caused by excretory failure accounts for approximately 60% of all cases of hyperuricemia.¹⁸ CKD has therefore been suggested to be a risk factor for hyperuricemia.¹⁹ We also found that UA elevation, which is associated with poor long-term outcomes, was common among patients complicated with CKD or with severely decompensated AHF who had been treated with loop-diuretics before admission. Excretion might therefore be an important factor in the mechanism of UA elevation.⁷

Adverse outcomes and the mechanisms underlying UA elevation in ICU patients

The serum levels of UA have been suggested to be a surrogate marker leading to an adverse outcome in patients with critical cardiovascular disease, such as chronic or acute heart failure and ACS.^7,20,–22 In the present study, we also showed that elevated serum UA levels were associated with long-term morality in patients requiring intensive care. However, the mechanisms associated with this poor prognosis have been unclear.

As mentioned above, diet, kidney disease, and medication are factors that might be associated with UA elevation in ICU patients. However, other mechanisms might also be associated with a poor prognosis in these patients. We therefore decided to focus on the mechanisms underlying hyperuricemia that were associated with a poor prognosis in the ICU.

Enhanced XOR activity was recently suggested to be a key factor involved in the development of hyperuricemia in patients with cardiovascular disease.⁹ XO and XDH are the most important enzymes in this metabolic system.^23,24 Although XO and XDH are both enzymes that catalyze UA production, their electron acceptors are different. XO and XDH both exist in human blood with different electron acceptors (oxygen and NAD⁺, respectively).²³ XOR is the generic term for XO and XDH. Some acute stressors (e.g. the increased production of lactate, or tissue hypoxia) directly induce the mobilization of XDH to the blood. The exchange to XO then leads to the enhancement of XOR activity, which subsequently increases the serum UA level. Furthermore, reactive oxygen species (ROSs), such as hydrogen peroxide (H₂O₂) and superoxide anion (O₂⁻), are generated through the productive reaction of UA, which is catalyzed by XOR.²³ These by-products lead to cell damage. From this perspective, an excessive increase in the XOR activity would lead not only to serum UA elevation but also to increased oxidative stress. Thus, the activation of ROS via the activation of XOR might be a mechanism that leads to an adverse outcome in ICU patients.

Based on this theory, we hypothesized that the critical status of the patients induced the increase in the serum UA level, which subsequently led to an adverse outcome due to the production of ROS in emergency patients. However, our present findings suggest that an elevated UA level with a high APACHE II score on admission in the emergency setting leads to a particularly poor outcome. Further studies will be required to clarify why the UA levels are elevated in the emergency setting and why these elevated UA levels lead to adverse outcomes in patients who require cardiovascular intensive care.

Study limitations

The present study is associated with several limitations. First, this was a retrospective study performed at a single center. The decision as to whether to admit each patient to the ICU was made by the individual physician. Thus, the criteria for ICU admission may have varied, which might have affected the prognosis. Second, the APACHE II scores of the patients were relatively low, even when they were admitted to the non-surgical ICU. The reason for this issue was that patients with ACS (i.e. Killip class I) accounted for the majority of the cohort in the present study. Furthermore, some physicians decided to admit patients to the ICU to follow up critical symptoms. This might be one reason why the APACHE II scores were low in comparison with previous reports. This is one of the limitations of the present study. Third, a few data, including data on the body temperature of 37 patients and the arterial blood gas values of 70 patients, were missing from the medical records. We were therefore obliged to exclude these 107 patients from the calculations of the APACHE II score. As a consequence, only 2328 patients were analyzed. Moreover, the 30-day follow-up data of 22 patients were missing; thus, 2306 patients were evaluated in the ROC curve analysis for 30-day mortality. Fourth, the calculated risk for 30-day mortality shifted to the correct direction in 20.4% of the patients in the present study (cf-NRI) when the serum UA value was added to the APACHE II score, while the average calculated risk of 30-day mortality shifted in the correct direction (IDI) by 1.5%. The NRI and IDI data were significant. However, the AUC value was not significantly increased by adding UA to the APACHE II score because the AUC of the APACHE II score was adequately high (0.800). Further discussion might be required regarding this issue. Finally, as the present study was a retrospective study, the serum UA levels were only assayed at one time point for emergency patients. We were unable to perform a follow-up evaluation (e.g. at seven or 14 days later). It might be important to evaluate the levels after treatment in order to assess the full impact of this parameter. Further studies will be needed to evaluate the UA level at discharge from the ICU as well as from the hospital.

Conclusion

The serum UA level was an independent predictor of 365-day mortality in patients who were emergently hospitalized in the cardiovascular ICU. The prognosis, including 365-day mortality, was significantly poorer in patients who had a high UA level with a high APACHE II score than in other patients. The serum UA level is therefore suggested to be a surrogate biomarker for critical patients in the cardiovascular ICU. Their critical status might induce the elevation of the serum UA level, and an elevated UA level itself might lead to an adverse outcome in emergency patients.

We are grateful to the staff of the ICU and medical records office at Chiba Hokusoh Hospital, Nippon Medical School for collecting the medical data.

References