Atherosclerosis 297 (2020) 40–46

Contents lists available at ScienceDirect

Atherosclerosis
journal homepage: www.elsevier.com/locate/atherosclerosis

Association of variability in uric acid and future clinical outcomes of patient T

with coronary artery disease undergoing percutaneous coronary
intervention
Su Shen Lima,e, Ya-Ling Yanga,d,e, Su-Chan Chena,d,e, Cheng-Hsueh Wua,b,e,
Shao-Sung Huanga,b,d,e, Wan Leong Chana,c,d, Shing-Jong Lina,b,d,e, Jaw-Wen Chena,b,c,d,e,
Chia-Yu Choua, Ju-Pin Pana,d,e, Min-Ji Charnga,d,e, Ying-Hwa Chena,d,e, Tao-Cheng Wua,d,e,
Tse-Min Lua,d,e, Pai-Feng Hsua,d,e, Po-Hsun Huanga,d,e, Hao-Min Chenga,c,d,e,
Chin-Chou Huanga,c,d,e, Shih-Hsien Sunga,d,e, Yenn-Jiang Lina,d,e, Hsin-Bang Leua,b,d,e,∗
a
Division of Cardiology, Department of Medicine, Taiwan
b
Healthcare and Management Center, Taiwan
c
Department of Medical Education, Taipei Veterans General Hospital, Taipei, Taiwan
d
Faculty of Medicine, Taiwan
e
Institute of Clinical Medicine and Cardiovascular Research Center, National Yang-Ming University, Taipei, Taiwan

HIGHLIGHTS

• High uric acid (UA) variability is associated with increased risk of developing future adverse outcome in coronary artery disease (CAD) patients after percu-
taneous coronary intervention (PCI).
• Both higher UA variability and average UA value were significantly associated with a higher CV event occurrence.
• High UA variability had greater impact than high average UA value in patients with CAD.

ARTICLE INFO ABSTRACT

Keywords: Background and aims: Hyperuricemia is independently associated with cardiovascular disease (CVD) and is
Inter-visit uric acid variability considered to be one of the major risk factors for CVD. However, the impact of inter-visit uric acid (UA)
Coronary artery disease variability on cardiovascular risk remains undetermined.
Major adverse cardiovascular events Methods: We enrolled 3202 patients with coronary artery disease (CAD), who received successful coronary in-
Hyperuricemia
tervention, in a cohort from Taipei Veterans General Hospital from 2006 to 2015. All post-baseline visits UA
Mortality
measurements using standard deviation (SD) were analyzed to correlate with long-term outcome. The primary
outcome was the composite of cardiac death, nonfatal MI, nonfatal stroke (MACE). The secondary event was
MACE and hospitalization for heart failure.
Results: During an average 65.06 ± 32.1-month follow-up, there were 66 cardiovascular deaths, 175 nonfatal
myocardial infarctions, 64 nonfatal strokes, 287 hospitalizations for heart failure, and 683 revascularization
procedures. There was a linear association between high UA SD and future adverse events. Compared to the
lowest quartile SD, subjects in the highest quartile SD had a higher risk of MACE (HR: 2.53, 95% CI: 1.78–3.59),
myocardial infarction (HR: 2.43, 95% CI: 1.53–3.86), cardiovascular death (HR: 6.45, 95% CI: 2.52–16.55),
heart failure-related hospitalization (HR: 3.43, 95% CI: 2.32–5.05), and total major CV events (HR: 2.72, 95% CI:
2.09–3.56). Furthermore, compared to the average achieved on-treatment UA value, increasing UA SD had a
stronger association of higher risk of developing MACE (HR: 1.51, 95% CI: 1.36–1.68), myocardial infarction
(HR: 1.37, 95% CI: 1.38–1.68), ischemic stroke (HR: 1.43, 95% CI: 1.13–1.82), CV death (HR: 1.77, 95% CI:
1.50–2.11), HF (HR: 1.43, 95% CI: 1.29–1.58), and total major CV events (HR: 1.46, 95% CI: 1.34–1.58).
Conclusions: High UA variability is associated with a higher risk of developing future cardiovascular events,

∗
Corresponding author. Healthcare and Management Center, Division of Cardiology, Taipei Veterans General Hospital, No. 201, Section 2, Shih-Pai Road, Taipei,
Taiwan.
E-mail address: [email protected] (H.-B. Leu).

https://doi.org/10.1016/j.atherosclerosis.2020.01.025
Received 23 August 2019; Received in revised form 9 January 2020; Accepted 29 January 2020
Available online 30 January 2020
0021-9150/ © 2020 Elsevier B.V. All rights reserved.
S.S. Lim, et al. Atherosclerosis 297 (2020) 40–46

suggesting the importance of maintaining stable serum UA levels and avoiding large fluctuations in CAD patients
after percutaneous coronary intervention (PCI).

1. Introduction To evaluate the inter-visit variability of these markers, they were de-
termined at baseline and upon every outpatient visit during the follow-
Cardiovascular disease (CVD) is one of the leading causes of mor- up period.
tality worldwide, accounting for approximately 31% of all deaths [1].
Atherosclerosis is no longer seen as cholesterol simply accumulating 2.3. UA variability measurement
within blood vessels but as a sustained, dynamic, and inflammatory
process affecting the vasculature [2,3]. Therefore, the identification of Every post-baseline visit SUA measurement was included for ana-
patients at high risk for developing future complications is important lysis. Variability was assessed every three months at the outpatient
toward ensuring optimal patient outcome, especially after coronary clinic. Inter-visit variability was defined as the standard deviation (SD)
intervention. of the UA values between visits. For the primary analysis, we used the
Uric acid (UA) is the final breakdown product of purine metabolism. intra-individual SD of uric acid across visits as our variability of uric
The degradation of purine nucleotides is regulated by the xanthine- acid. For patients with missing UA values, the available values from
oxidoreductase enzyme, which converts hypoxanthine to xanthine and other time points were used to calculate variability. The number of
xanthine to UA [4]. Accumulation of UA is well known to cause measurements per participant ranged as follows: 3 measurements
monosodium urate (MSU) crystal deposition in the joints and kidneys, (n = 1378 or 43%), 4 measurement (n = 346 or 11%) and more than 5
leading to the development of gout and kidney stones [5]. Moreover, measurements (n = 1478 or 46%). Similar variability measurements
experimental and epidemiological studies suggest that elevated levels of for other biochemical markers using SD as the primary means of re-
UA are associated with developing hypertension, obesity, insulin re- presenting variability (e.g., lipoprotein variability and blood pressure
sistance, dyslipidemia, diabetes mellitus, kidney disease, cardiovascular variability measurements) have also been used in other studies [11].
and cerebrovascular events [6]. These findings are indicative of a po-
tentially critical connection between UA and CVD.
Recently, clinical observations have demonstrated that increased 2.4. Clinical follow-up and endpoints
variability in cardiovascular parameters including blood pressure, body
mass index (BMI), glucose and cholesterol levels, are associated with an All participants were followed-up every three months after enroll-
increased risk of future cardiovascular events [7–9]. While fluctuations ment. The primary outcome was MACE, a composite of cardiovascular
in serum UA (SUA) are known to prolong the inflammatory process of a death, nonfatal myocardial infarction, or nonfatal stroke. The key sec-
gout attack [10], the impact of UA variability on cardiovascular risk ondary outcome was a composite of the primary outcome plus hospi-
remains undermined. Furthermore, whether SUA variability or overall talization for congestive heart failure (total major CV events).
level has more prognostic relevance in terms of CVD remains unclear. Myocardial infarction was confirmed in patients presenting with is-
Therefore, the present study aimed to explore the association between chemic symptoms with elevated serum cardiac enzyme levels and/or
SUA variability and nonfatal myocardial infarction, ischemic stroke, characteristic ECG changes. Ischemic stroke was confirmed as an ob-
cardiovascular death, and hospitalization due to congestive heart struction within a blood vessel supplying blood to the brain with ima-
failure in patients with CAD after PCI. ging evidence by either MRI or CT scan and new neurological deficit
lasting for at least 24 h. The protocol for CV event follow-up was per-
formed as previously described [12,13].
2. Materials and methods
2.5. Statistical analyses
2.1. Study population
The baseline characteristics of patients were compared according to
This retrospective, single-center observation study included parti-
quartiles of baseline UA variability (SD). Quantitative variables were
cipants with symptomatic CAD who received coronary intervention
expressed as a mean and standard deviation in the presence of normal
between July 2006 and December 2015 at the Taipei Veteran General
or median distribution and interquartile range in the presence of
Hospital, Taiwan. Diagnosis for CAD included positive results to stress
asymmetric distribution. Qualitative variables were presented in both
test, history of angina with ischemic change on ECG recording, myo-
absolute frequencies (number of patients) and relative frequencies
cardial infarction attack, or angina symptom with significant stenosis
(percentage). Comparisons of continuous variables between groups
lesion in coronary computed tomography angiography (CCTA).
were achieved using an ANOVA test, while subgroup comparisons of
Symptomatic patients who received percutaneous coronary interven-
categorical variables were assessed by χ2 or Fisher's exact test. The
tion (PCI) either with coronary stenting or balloon angioplasty were
primary and secondary outcomes were described by an overall per-
eligible for enrollment.
centage and expressed by means of proportions with a 95% confidence
All procedures were conducted in accordance with the Declaration
interval (CI). The event-free survival rate was calculated using the
of Helsinki and were approved by the Ethics Committee and
Kaplan–Meier (KM) method, with significance evaluation using log-
Independent Review Board of the Taipei Veterans General Hospital.
rank tests. For the primary analysis, an unstratified log-rank test was
used to compare overall survival between the quartiles of UA varia-
2.2. Baseline characteristics and biochemical data bility. Multiple regression analysis was carried out using a Cox pro-
portional hazard regression analysis adjusted for age, sex, BMI, history
After enrollment, coronary intervention procedures and treatment of hypertension, diabetes, medications including statins, and serum
strategies were recorded. Baseline characteristics were collected from levels of LDL and HDL to evaluate whether UA variability was an in-
every patient including sex, age, history of hypertension, diabetes, dependent factor in determining the occurrence of primary and sec-
hyperlipidemia, smoking, family history of premature CAD, and cere- ondary endpoints including cardiac death, nonfatal MI, nonfatal stroke,
bral vascular disease. Furthermore, biochemical data describing renal hospitalization for heart failure and revascularization procedures.
function, lipid profiles, UA levels, and medications were also collected. Subsequent subgroup analysis was performed to investigate the effects

41
S.S. Lim, et al. Atherosclerosis 297 (2020) 40–46

of UA variability on other risk factors for cardiovascular events, such as 25.39 kg/m2, and mean UA value was 6.52 ± 1.89 mg/dl. The SD of
age, sex, history of diabetes mellitus, hypertension, and baseline levels quartiles of UA were 0.31 ± 0.14 (Q1), 0.67 ± 0.10 (Q2),
of LDL and HDL. Statistical significance was inferred at a two-sided p- 1.12 ± 0.18 (Q3), and 2.35 ± 0.83 (Q4). Subjects in the higher
value < 0.05. To assess the predictive accuracy of the UA SD for car- quartiles of UA variability tended to be older, have higher systolic blood
diovascular outcome, we used a receiver operating characteristic (ROC) pressure (SBP), higher baseline UA, worse renal function, unfavorable
curve for censored data and the area under the ROC curve (AUC) as the lipid profiles, higher smoking rates, more underlying comorbidities
criterion. An AUC of 0.5 indicates no predictive ability, whereas a value including diabetes mellitus, chronic kidney disease, and peripheral ar-
of 1 represents perfect predictive ability. The statistical significance of terial occlusion disease (p for trend < 0.05).
the difference between 2 ROC curves was evaluated with the method of
DeLong et al. [14] Data were analyzed using SPSS software (version 20, 3.2. Major cardiovascular events during follow-up
SPSS, Chicago, IL). A p value of < 0.05 was considered to indicate
statistical significance. Clinical follow-up was carried out with all patients for a mean
period of 65.6 ± 32.1 months. During this time, there was 66 cardi-
3. Results ovascular deaths, 175 cardiac nonfatal myocardial infarctions, 64
nonfatal strokes, 287 hospitalizations for heart failure, and 683 re-
3.1. Baseline characteristics vascularization procedures. Supplementary Table 1 shows the baseline
and follow-up values for UA and lipid profiles. There were significant
A total of 3202 patients with CAD who underwent successful cor- improvements in the lipid profiles (p < 0.001) in the study population
onary intervention were enrolled in this study. The baseline char- but the UA values remained the same during the follow-up period.
acteristics of the participants according to quartiles of UA variability Supplementary Fig. 1 shows the incidence of an observed event ac-
are presented in Table 1. Mean patients’ age was 68.8 ± 12.5 years, of cording to UA variability and illustrates the increased incidence of
which 83.1% were male, 39.2% had diabetes mellitus, mean BMI was adverse events with high variability in UA level (p for trend < 0.05).

Table 1
Baseline characteristics of subjects at baseline and according to the uric acid variability.
All subjects (n = 3202) Q1 (n = 800) Q2 (n = 796) Q3 (n = 803) Q4 (n = 803) p-value

Uric acid variability, mg/dL 1.12 ± 0.89 0.31 ± 0.14 0.67 ± 0.10 1.12 ± 0.18 2.35 ± 0.83 < 0.001
Age, years 68.76 ± 12.49 66.95 ± 12.41 67.03 ± 12.89 69.54 ± 12.48 71.48 ± 11.61 < 0.001
Gender, male, n (%) 2660 (83.1) 663 (82.9) 687 (86.3) 648 (80.7) 662 (82.4) 0.025
Body mass index, kg/m2 25.39 ± 3.66 25.28 ± 3.42 25.33 ± 3.51 25.41 ± 3.84 25.53 ± 3.89 0.731
Systolic blood pressure, mmHg 128.03 ± 18.17 127.04 ± 17.47 125.84 ± 16.45 129.13 ± 19.41 130.15 ± 18.94 < 0.001
Diastolic blood pressure, mmHg 71.75 ± 11.00 72.47 ± 10.18 71.68 ± 10.59 71.81 ± 11.50 71.04 ± 11.65 0.090
Uric acid, mg/dL 6.52 ± 1.89 6.09 ± 1.50 6.30 ± 1.53 6.49 ± 1.81 7.20 ± 2.39 < 0.001
Total cholesterol, mg/dL 172.76 ± 40.67 175.49 ± 38.25 174.37 ± 42.47 170.40 ± 39.35 170.80 ± 42.29 0.026
Triglyceride, mg/dL 139.34 ± 87.87 131.39 ± 79.17 134.86 ± 82.42 144.93 ± 92.60 146.16 ± 95.45 0.001
HDL-C, mg/dL 42.90 ± 11.89 43.69 ± 11.54 43.16 ± 11.15 43.24 ± 12.72 41.54 ± 11.98 0.002
LDL-C, mg/dL 106.42 ± 34.41 108.66 ± 33.19 109.09 ± 36.17 103.63 ± 33.19 104.35 ± 34.71 0.001
Glucose, mg/dL 121.60 ± 41.54 119.16 ± 38.40 120.87 ± 42.42 122.78 ± 41.52 123.42 ± 43.49 0.265
eGFR, mL/min/1.73m2 63.36 ± 26.03 70.96 ± 23.33 70.32 ± 25.66 59.80 ± 25.56 52.45 ± 24.85 < 0.001
Chronic disease
Stroke, n (%) 182 (5.7) 42 (5.3) 39 (4.9) 53 (6.6) 48 (6) 0.461
Hypertension, n (%) 2849 (89) 710 (88.8) 701 (88.1) 729 (90.8) 709 (88.3) 0.287
Diabetes mellitus, n (%) 1254 (39.2) 268 (33.5) 269 (33.8) 346 (43.1) 371 (46.2) < .001
PAOD, n (%) 147 (4.6) 26 (3.3) 25 (3.1) 36 (4.5) 60 (7.5) < 0.001
Hyperlipidemia, n (%) 1498 (46.8) 386 (48.3) 404 (50.8) 372 (46.3) 336 (41.8) 0.003
ACS, n (%) 1111 (34.7) 228 (28.5) 270 (33.9) 270 (33.6) 343 (42.7) < 0.001
Medications
ACEI, n (%) 665 (20.8) 179 (22.4) 173 (21.7) 158 (19.7) 155 (19.3) 0.343
ARB, n (%) 1257 (39.3) 291 (36.4) 307 (38.6) 329 (41) 330 (41.1) 0.166
Beta blocker, n (%) 1528 (47.7) 388 (48.5) 397 (49.9) 380 (47.3) 363 (45.2) 0.288
CCB, n (%) 1178 (36.8) 275 (34.4) 268 (33.7) 330 (41.1) 305 (38) 0.007
Statins, n (%) 1890 (59) 508 (63.5) 514 (64.6) 454 (56.5) 414 (51.6) < 0.001
Diuretics, n (%) 363 (11.3) 73 (9.1) 66 (8.3) 102 (12.7) 122 (15.2) < 0.001
Urate lowering agent, n (%) 300 (9.4) 36 (4.5) 56 (7) 73 (9.1) 135 (16.8) < 0.001
CAD
SVD, n (%) 934 (29.2) 256 (32) 266 (33.4) 229 (28.5) 183 (22.8) < 0.001
DVD, n (%) 1035 (32.3) 260 (32.5) 246 (30.9) 270 (33.6) 259 (32.3)
TVD, n (%) 1232 (38.5) 283 (35.4) 284 (35.7) 304 (37.9) 361 (45)
Stents
DES stents, n 1.88 ± 1.10 1.85 ± 1.07 1.88 ± 1.14 1.83 ± 1.01 1.94 ± 1.19 0.559
DES stents size, mm 3.01 ± 0.36 3.04 ± 0.37 3.00 ± 0.34 3.01 ± 0.35 3.00 ± 0.36 0.323
DES stents length, mm 24.15 ± 5.89 23.91 ± 5.78 24.16 ± 6.17 24.36 ± 5.79 24.23 ± 5.83 0.697
BMS stents, n 1.48 ± 0.74 1.42 ± 0.71 1.50 ± 0.71 1.52 ± 0.80 1.49 ± 0.72 0.447
BMS stents size, mm 3.17 ± 0.54 3.14 ± 0.53 3.20 ± 0.55 3.20 ± 0.56 3.16 ± 0.51 0.461
BMS stents length, mm 21.64 ± 6.60 20.96 ± 6.35 21.58 ± 6.96 21.87 ± 6.28 22.03 ± 6.77 0.283

Data are expressed as mean ± SD, or n (%).

HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; eGFR, estimated glomerular filtration rate; PAOD, peripheral artery oc-
clusive disease; ACS, acute coronary syndrome; ACEI, angiotensin-converting-enzyme inhibitor; ARB, angiotensin II receptor blocker; CCB, calcium channel blocker;
SD, standard deviation, SVD, single vessel disease; DVD, double vessel disease; TVD, triple vessel disease; DES, drug-eluting stent; BMS, bare metal stent.

42
S.S. Lim, et al. Atherosclerosis 297 (2020) 40–46

Fig. 1. Kaplan–Meier estimates of cumulative survival by quartiles of UASD in future risk of (A) AMI, (B) MACE, (C) CV death, (D) CHF, (E) ischemic stroke and (F)
total CV event.
Adjusted for age, gender, hypertension, diabetes mellitus and HMG-CoA reductase inhibitors.

The incidence of future events including cardiovascular death, nonfatal 3.4. Comparing the strength of association of inter-visit UA variability to
MI, nonfatal stroke, major adverse cardiovascular event (MACE, com- overall average UA value
bined of CV death, MI and ischemic stroke), CHF and total events in-
creased in subjects with increasing UA variability. Fig. 1 shows the The results of this study indicate that higher UA SD is associated
Kaplan–Meier curves assessing the occurrence of future adverse events with an increased risk of adverse future events. Next, we wanted to
according to quartiles of UA SD. There were significant stepwise re- know whether UA SD or UA average on treatment contributes more to
lationships between increasing quartiles of UA SD and cumulative future CV risk. Fig. 2 depicts the relationship between CV risk and UA
events including myocardial infarctions, cardiovascular deaths, heart SD or UA average. UA SD significantly associated with an increased risk
failure hospitalizations, MACEs, ischemic stroke and total major CV of experiencing MACE (HR: 1.51, 95% CI: 1.36–1.68), myocardial in-
events (log-rank p < 0.01). farction (HR: 1.37, 95% CI: 1.38–1.59), ischemic stroke (HR: 1.43, 95%
CI: 1.13–1.82), CV death (HR: 1.77, 95% CI: 1.50–2.11), HF (HR: 1.43,
3.3. Risk of major cardiovascular events according to quartile of UA 95% CI: 1.29–1.58) and total major CV events (HR: 1.46, 95% CI:
variability 1.34–1.58). However, average UA value was not associated with is-
chemic stroke and CV death.
Table 2 shows the risk of future events including nonfatal MI,
nonfatal stroke, cardiovascular death, heart failure, and total major CV 3.5. Improvement in predictive performance
event according to the quartiles of UA variability after adjustment for
age, sex, hypertension, diabetes, BMI, renal function, statin and diure- To evaluate improvement in predictive performance of considering
tics use. Compared to the lower variability (reference group), patients UA average and UA SD, the AUCs of UA and UA SD were assessed by
with high variability (Q4) had a significantly higher risk of MACE (HR: ROC curve over the follow-up duration. Fig. 3 showed the ROC curves
2.53, 95% CI: 1.78–3.59), myocardial infarction (HR: 2.43, 95% CI: of UA SD and UA average related to CV outcomes. The AUC by adding
1.53–3.86), cardiovascular death (HR: 6.45, 95% CI: 2.52–16.55), heart UA SD to UA average was significantly increased for predicting MACE
failure-related hospitalization (HR: 3.43, 95% CI: 2.32–5.05), and total (0.616 vs. 0.535, p = 0.001), AMI (0.579 vs. 0.550, p = 0.049), is-
major CV events (HR: 2.72, 95% CI: 2.09–3.56) (Table 2 and chemic stroke (0.594 vs. 0.502, p = 0.037), CV death (0.707 vs. 0.539,
Supplementary Fig. 2). Supplementary Fig. 3 shows the subgroup p = 0.003) CHF (0.666 vs. 0.635, p < 0.001), and total CV events
analysis of the association of future risk of increasing UA SD quartile (0.636 vs. 0.583, p < 0.001).
among the different sexes, ages, BMIs, histories of DM, HTNs, lipid
profile, renal function, ejection function (EF) and whether enrollment 4. Discussion
from ACS. Higher UA variability (SD) was independently associated
with higher risk of developing adverse event, especially among these The present study is the first to demonstrate the association of inter-
with lower EF and ACS. visit UA variability and future cardiovascular outcome in patients with

43
S.S. Lim, et al. Atherosclerosis 297 (2020) 40–46

Table 2 baseline comorbidities and medication use. Furthermore, this study

Hazard ratios and 95% confidence intervals of MACE, myocardial infarction, showed that both higher UA SD and achieved average on-treatment UA
cardiovascular death, congestive heart failure, total cardiovascular events. value were significantly associated with a higher CV event occurrence.
Events (n) Follow-up Incidence rate (per Adjust HR UA SD had greater impact and improved predictive performance, sug-
duration (person- 1000 person years) (95% CI) gesting that maintaining low and consistent UA values in patients with
years) CAD associates with better clinical outcome.
High serum UA (SUA) levels are reported to strongly associate with
MACE
Q1 45 51,648 0.87 1 (ref) prevalent disorders including hypertension, diabetes, obesity, and renal
Q2 54 51,390 1.05 1.09 (0.73–1.62) disease [6,15–17]. Although the causal relationship between SUA va-
Q3 73 51,841 1.41 1.53 (1.05–2.22) lues and CV events has been long debated, population-based studies
Q4 114 51,841 2.20 2.53 (1.78–3.59)
have shown that hyperuricemia is an independent risk factor for CVD
p for trend < 0.001
Myocardial infarction
[18], a prospective cohort study of 90,393 subjects conducted in
Q1 26 52,048 0.50 1 (ref) Taiwan concluded high serum uric acid level is associated with all
Q2 41 51,788 0.79 1.45 (0.89–2.37) causes, total CVD, and ischemic stroke in the Taiwanese general po-
Q3 46 52,243 0.88 1.68 (1.04–2.73) pulation [19]. Recently, a Mendelian randomization (MR) study con-
Q4 62 52,243 1.19 2.43 (1.53–3.86)
ducted by Pan et al. suggested a possible causal relationship between
p for trend 0.001
Ischemic stroke UA and CVD in the Chinese population [20], emphasizing the im-
Q1 15 52,872 0.28 1 (ref) portance of avoiding high UA levels for risk reduction in patients with
Q2 8 52,608 0.15 0.45 (0.19–1.07) CAD. Although the biological mechanism is still not clear, investigation
Q3 17 53,070 0.32 1.06 (0.53–2.14)
studies have leaded to more understanding of UA metabolism and helps
Q4 24 53,070 0.45 1.61 (0.84–3.09)
p for trend 0.022
to elucidate the role of UA in the pathogenesis of CVD. UA can induce
Cardiovascular death oxidative stress by activating nicotinamide adenine dinucleotide
Q1 5 53,288 0.09 1 (ref) phosphate-oxidase (NADPH). Rats with hyperuricemia induced with an
Q2 9 53,022 0.17 1.71 (0.57–5.11) uricase inhibitor showed increased oxidative stress and developed hy-
Q3 16 53,488 0.30 2.90 (1.06–7.94)
pertension that could be blocked by antioxidant treatment [21,22]. In
Q4 36 53,488 0.67 6.45 (2.51–16.55)
p for trend < 0.001 addition, UA crystals can induce an inflammatory response by stimu-
Congestive heart failure lating the release of chemokines (e.g., monocyte chemoattractant pro-
Q1 33 51,056 0.65 1 (ref) tein-1, MCP-1), inflammatory markers (e.g., high-sensitivity C-reactive
Q2 44 50,801 0.87 1.23 (0.78–1.93) protein), and the activation of vasoconstrictive mediators (e.g.,
Q3 87 51,247 1.70 2.36 (1.58–3.53)
Q4 123 51,247 2.4 3.43 (2.32–5.05)
thromboxane, endothelin-1 and angiotensin II) [22,23]. Furthermore,
p for trend < 0.001 hyperuricemia is proven to cause endothelial dysfunction in humans by
Total cardiovascular event reducing NO bioavailability [24–26] and the use of UA-lowering agents
Q1 75 49,624 1.51 1 (ref) improves endothelial function [27,28]. Moreover, UA may also stimu-
Q2 92 49,376 1.86 1.13 (0.83–1.53)
late the renin-angiotensin system (RAS), thus further contributing to
Q3 141 49,810 2.83 1.76 (1.33–2.34)
Q4 205 49,810 4.11 2.72 (2.09–3.56) vascular smooth cell growth, arterial functional impairment, and stif-
p for trend < 0.001 fening [29,30].
Patients with higher UA variability had higher percentage users of
Adjusted for age, gender, hypertension, diabetes mellitus, renal function, diuretics, urate lowering agents, worse eGFR and more comorbidities,
diuretics and HMG-CoA reductase inhibitors. MACE, major adverse cardiovas- suggesting a group at risk. After considering comorbidities and medi-
cular events.
cation including chronic kidney disease, urate lowering agents and
diuretics use, higher UA variability was still independently associated
CAD after coronary intervention. High inter-visit UA variability, mea-
with future risk of developing myocardial infarction, ischemic stroke,
sured via higher SD of intra-individual SUA, was associated with a
cardiovascular mortality, HF hospitalization, and MACE. Furthermore,
graded increase in the risk of myocardial infarction, ischemic stroke,
this correlation is more prominent in patients with ACS and lower
cardiovascular mortality, HF hospitalization, and MACE independent of

Fig. 2. Hazard ratios and 95% confidence intervals of increasing UA SD and UA average on treatment to future risk of developing MACE, AMI, ischemic stroke, CV
death, CHF, and total CV event in CAD patients after PCI.

44
S.S. Lim, et al. Atherosclerosis 297 (2020) 40–46

Fig. 3. ROC curves of UA SD and UA average to CV outcomes. (A) UA SD and UA average to MACE. (B) UA SD and UA average to total AMI. (C) UA SD and UA
average to ischemic stroke. (D) UA SD and UA average to CV death. (E) UA SD and UA average to CHF. (F) UA SD and UA average to total CV event.

LVEF. In the current study, ROC curve analysis demonstrated sub- lifestyle medication [36]. In that study, lower and stable serum level of
stantial increase in the prognostic prediction value when adding UA uric acid was associated with lower risk of primary endpoints and it was
variability to serum UA values, indicates that UA variability seems to in accordance with our study finding of maintaining lower and con-
confer greater impact than overall average SUA value. Actually, the sistent serum level of uric acid is associated with lower future risk of
reason why greater UA variability results in higher risk remains un- cardiovascular events. However, a controversial study was reported
determined, but clinical observation showed the fluctuations in UA recently. The Cardiovascular Safety of Febuxostat and Allopurinol in
level may exacerbate the inflammatory process underlying gout [31], Patients with Gout and Cardiovascular Morbidities (CARES) trial, which
suggesting that fluctuations in UA level may prolong inflammation. included 6190 patients with gout and CVD, showed that patients who
Because SUA level has been related to oxidative stress, inflammation, received febuxostat had lower SUA levels but higher all-cause mortality
endothelial dysfunction, and the magnitude of activation of the RAS and cardiovascular mortality than those who received allopurinol [32].
system, UA fluctuation may antagonize these important issues and thus The exact mechanism remains unknown and study with placebo group
contribute to disease progression. A surge of UA in the blood is known is needed to elucidate whether the results of CARES study are due to
to increase the crystallization rate of urate, which stimulates an im- harmful effect of febuxostat or beneficial effects of allopurinol. Our
mune reaction and inflammatory response. This may indicate that a study demonstrated higher UA variability was associated with in-
large fluctuation in SUA level can be deleterious [32–34]. Additional creased higher future risk even considering medication and underlying
large-scale studies focusing on SUA level fluctuation are needed to comorbidities. However, our study is an observation study, not clinical
further elucidate these issues. A previous observational study focusing trial designed for investigating the causality. Further research is re-
on SUA and mortality in subjects without gout arthritis found a U- quired to uncover the causal mechanism underlying the association
shaped relationship between SUA and both all-cause and CV mortality, between UA variability and CVD.
indicating that individuals with SUA levels at either extreme are at In conclusion, in patients with CAD receiving PCI, greater inter-visit
higher risk for deleterious events [35]. Recently, the Febuxostat for variability in SUA level is significantly associated with future adverse
Cerebral and CaRdiorenovascular Events PrEvEntion StuDy (FREED) cardiovascular outcomes, including cardiovascular mortality, MACE,
compared febuxostat and conventional therapy with lifestyle mod- myocardial infarction, ischemic stroke, congestive heart failure, and
ification in subjects with hyperuricemia, showed febuxostat had lower total cardiovascular events. This observation suggests the UA variability
serum level of uric acid (4.50 ± 1.52 vs. 6.76 ± 1.45 mg/dL, p < is associated with increased risk and highlights the importance of
0.001) and lower risk of composite of cardiovascular events than achieving stable SUA levels and avoiding large fluctuations.

45
S.S. Lim, et al. Atherosclerosis 297 (2020) 40–46

Funding Information [10] T. Neogi, Clinical practice. Gout, N. Engl. J. Med. 364 (2011) 443–452.
[11] J. Whittle, A.I. Lynch, R.M. Tanner, et al., Visit-to-Visit variability of BP and CKD
outcomes: results from the ALLHAT, Clin. J. Am. Soc. Nephrol. : CJASN 11 (2016)
Taiwan Association for integration of cardiology and surgery, 471–480.
Grant/Award Number: TAICS-107-2-Y2. [12] H.-B. Leu, C.-P. Lin, W.-T. Lin, et al., Risk stratification and prognostic implication
of plasma biomarkers in nondiabetic patients with stable coronary artery disease:
the role of high-sensitivity C-reactive protein, Chest 126 (2004) 1032–1039.
CRediT authorship contribution statement [13] H.-B. Leu, C.-P. Lin, W.-T. Lin, et al., Circulating mononuclear superoxide produc-
tion and inflammatory markers for long-term prognosis in patients with cardiac
Su Shen Lim: Writing - original draft, Data curation, Formal ana- syndrome X, Free Radic. Biol. Med. 40 (2006) 983–991.
[14] E.R. DeLong, D.M. DeLong, D.L. Clarke-Pearson, Comparing the areas under two or
lysis. Ya-Ling Yang: Data curation, Formal analysis. Su-Chan Chen: more correlated receiver operating characteristic curves: a nonparametric ap-
Data curation, Formal analysis. Cheng-Hsueh Wu: Data curation, proach, Biometrics 44 (1988) 837–845.
Formal analysis. Shao-Sung Huang: Data curation, Formal analysis. [15] T.W. Yoo, K.C. Sung, H.S. Shin, et al., Relationship between serum uric acid con-
centration and insulin resistance and metabolic syndrome, Circ. J. : Off. J. Jpn. Circ.
Wan Leong Chan: Data curation, Formal analysis. Shing-Jong Lin:
Soc. 69 (2005) 928–933.
Data curation, Formal analysis. Jaw-Wen Chen: Data curation, Formal [16] R.J. Johnson, D.I. Feig, J. Herrera-Acosta, et al., Resurrection of uric acid as a causal
analysis. Chia-Yu Chou: Funding acquisition. Ju-Pin Pan: Funding risk factor in essential hypertension, Hypertension (Dallas, Tex, 45 (1979) 18–20
acquisition. Min-Ji Charng: Data curation, Formal analysis. Ying-Hwa 2005.
[17] J. Tuomilehto, P. Zimmet, E. Wolf, et al., Plasma uric acid level and its association
Chen: Data curation, Formal analysis. Tao-Cheng Wu: Data curation, with diabetes mellitus and some biologic parameters in a biracial population of Fiji,
Formal analysis. Tse-Min Lu: Data curation, Formal analysis. Pai-Feng Am. J. Epidemiol. 127 (1988) 321–336.
Hsu: Data curation, Formal analysis. Po-Hsun Huang: Data curation, [18] J. Fang, M.H. Alderman, Serum uric acid and cardiovascular MortalityThe NHANES
I epidemiologic follow-up study, J. Am. Med. Assoc. 283 (2000) 2404–2410 1971-
Formal analysis. Hao-Min Cheng: Data curation, Formal analysis. 1992.
Chin-Chou Huang: Methodology, Resources, Software. Shih-Hsien [19] J.H. Chen, S.Y. Chuang, H.J. Chen, et al., Serum uric acid level as an independent
Sung: Data curation, Formal analysis. Yenn-Jiang Lin: Supervision. risk factor for all-cause, cardiovascular, and ischemic stroke mortality: a Chinese
cohort study, Arthritis Rheum. 61 (2009) 225–232.
Hsin-Bang Leu: Conceptualization, Supervision, Writing - review & [20] K.M. Chiang, Y.C. Tsay, T.C. Vincent Ng, et al., Is hyperuricemia, an early-onset
editing, Data curation, Formal analysis. metabolic disorder, causally associated with cardiovascular disease events in han
Chinese? J. Clin. Med. 8 (2019).
[21] L.G. Sanchez-Lozada, V. Soto, E. Tapia, et al., Role of oxidative stress in the renal
Declaration of competing interest abnormalities induced by experimental hyperuricemia, Am. J. Physiol. Ren.
Physiol. 295 (2008) F1134–F1141.
The authors declared they do not have anything to disclose re- [22] M. Salvetti, A. Paini, C. Agabiti-Rosei, et al., Uric Acid and Cardiovascular Disease:
an Update vol. 11, European Cardiology Review, 2016.
garding conflict of interest with respect to this manuscript.
[23] M.J. Maclachlan, G.P. Rodnan, Effects of food, fast and alcohol on serum uric acid
and acute attacks of gout, Am. J. Med. 42 (1967) 38–57.
Appendix A. Supplementary data [24] S. Zharikov, K. Krotova, H. Hu, et al., Uric acid decreases NO production and in-
creases arginase activity in cultured pulmonary artery endothelial cells, Am. J.
Physiol. Cell Physiol. 295 (2008) C1183–C1190.
Supplementary data to this article can be found online at https:// [25] C. Gersch, S.P. Palii, K.M. Kim, et al., Inactivation of nitric oxide by uric acid,
doi.org/10.1016/j.atherosclerosis.2020.01.025. Nucleos Nucleot. Nucleic Acids 27 (2008) 967–978.
[26] M. Kanbay, M. Segal, B. Afsar, et al., The role of uric acid in the pathogenesis of
human cardiovascular disease, Heart 99 (2013) 759–766.
References [27] W. Doehner, N. Schoene, M. Rauchhaus, et al., Effects of xanthine oxidase inhibition
with allopurinol on endothelial function and peripheral blood flow in hyperur-
[1] World Health Organization, Cardiovascular Diseases, CVDs, 2017. icemic patients with chronic heart failure: results from 2 placebo-controlled studies,
[2] T. Willerson James, M. Ridker Paul, Inflammation as a cardiovascular risk factor, Circulation 105 (2002) 2619–2624.
Circulation 109 (2004) II-2-II-10. [28] C.A. Farquharson, R. Butler, A. Hill, et al., Allopurinol improves endothelial dys-
[3] N. Ruparelia, J.T. Chai, E.A. Fisher, et al., Inflammatory processes in cardiovascular function in chronic heart failure, Circulation 106 (2002) 221–226.
disease: a route to targeted therapies, Nat. Rev. Cardiol. 14 (2016) 133. [29] D.B. Corry, P. Eslami, K. Yamamoto, et al., Uric Acid Stimulates Vascular Smooth
[4] J. George, A.D. Struthers, Role of urate, xanthine oxidase and the effects of allo- Muscle Cell Proliferation and Oxidative Stress via the Vascular Renin–Angiotensin
purinol in vascular oxidative stress, Vasc. Health Risk Manag. 5 (2009) 265–272. System vol. 26, (2008), pp. 269–275.
[5] Y. Shi, A.D. Mucsi, G. Ng, Monosodium urate crystals in inflammation and im- [30] M.A. Yu, L.G. Sanchez-Lozada, R.J. Johnson, et al., Oxidative stress with an acti-
munity, Immunol. Rev. 233 (2010) 203–217. vation of the renin-angiotensin system in human vascular endothelial cells as a
[6] S.D. Nath, V.S. Voruganti, N.H. Arar, et al., Genome scan for determinants of serum novel mechanism of uric acid-induced endothelial dysfunction, J. Hypertens. 28
uric acid variability, J. Am. Soc. Nephrol. : JASN (J. Am. Soc. Nephrol.) 18 (2007) (2010) 1234–1242.
3156–3163. [31] A.T. Eggebeen, Gout: an update, Am. Fam. Physician 76 (2007) 801–808.
[7] M.K. Kim, K. Han, Y.-M. Park, et al., Associations of variability in blood pressure, [32] W.B. White, K.G. Saag, M.A. Becker, et al., Cardiovascular safety of febuxostat or
glucose and cholesterol concentrations, and body mass index with mortality and allopurinol in patients with gout, N. Engl. J. Med. 378 (2018) 1200–1210.
cardiovascular outcomes in the general population, Circulation 138 (2018) [33] B.T. Emmerson, The management of gout, N. Engl. J. Med. 334 (1996) 445–451.
2627–2637. [34] A.L. Weaver, M.A. Cheh, R.H. Kennison, How PCP education can impact gout
[8] M.K. Kim, K. Han, H.S. Kim, et al., Cholesterol variability and the risk of mortality, management: the gout essentials, J. Clin. Rheumatol. : Pract. Rep. Rheum.
myocardial infarction, and stroke: a nationwide population-based study, Eur. Heart Musculoskel. Dis. 14 (2008) S42–S46.
J. 38 (2017) 3560–3566. [35] C.F. Kuo, L.C. See, K.H. Yu, et al., Significance of serum uric acid levels on the risk
[9] D. Clark 3rd, S.J. Nicholls, J. St John, et al., Visit-to-visit cholesterol variability of all-cause and cardiovascular mortality, Rheumatology 52 (2013) 127–134.
correlates with coronary atheroma progression and clinical outcomes, Eur. Heart J. [36] S. Kojima, K. Matsui, S. Hiramitsu, et al., Febuxostat for cerebral and
39 (2018) 2551–2558. CaRdiorenovascular events PrEvEntion StuDy, Eur. Heart J. 40 (2019) 1778–1786.

46