Variation in Serum Urate Levels in the Absence of Gout and Urate Lowering Therapy: An Analysis Using Data from a Randomized Controlled Trial


 Background: Previous studies have noted significant variation in serum urate (sUA) levels, and it is unknown how this influences the accuracy of hyperuricemia classification based on single data points. Despite this known variability, hyperuricemic patients are often used as a control group in gout studies. Our objective was to determine the accuracy of hyperuricemia classifications based on single data points versus multiple data points given the degree of variability observed with serial measurements of sUA.Methods: Data was analyzed from a cross-over clinical trial of urate-lowering therapy in young adults without a gout diagnosis. In the control phase, sUA levels used for this analysis were collected at 2-4 week intervals. Mean coefficient of variation for sUA was determined, as were rates of conversion between normouricemia (sUA ≤6.8 mg/dL) and hyperuricemia (sUA >6.8 mg/dL). Results: Mean study participant (n = 85) age was 27.8 ± 7.0 years, with 39% female participants and 41% African-American participants. Mean sUA coefficient of variation was 8.5% ± 4.9% (1% to 23%). There was no significant difference in variation between men and women, or between participants initially normouricemic and those who were initially hyperuricemic.Among those initially normouricemic (n=72), 15% converted to hyperuricemia during at least one subsequent measurement. The subgroup with initial sUA <6.0 (n=54) was much less likely to have future values in the range of hyperuricemia compared to the group with screening sUA values between 6.0-6.8 (n=18) (7% vs 39%, p = 0.0037).Of the participants initially hyperuricemic (n=13), 46% were later normouricemic during at least one measurement.Conclusion: Single sUA measurements were unreliable in hyperuricemia classification due to spontaneous variation. Those with an sUA of <6.0 mg/dL were less likely to demonstrate future hyperuricemic measurements and this could be considered a safer threshold to rule out intermittent hyperuricemia based on a single measurement point.Trial registration: Data from parent study ClinicalTrials.gov Identifier: NCT02038179

was no signi cant difference in variation between men and women, or between participants initially normouricemic and those who were initially hyperuricemic.
Among those initially normouricemic (n=72), 15% converted to hyperuricemia during at least one subsequent measurement. The subgroup with initial sUA <6.0 (n=54) was much less likely to have future values in the range of hyperuricemia compared to the group with screening sUA values between 6.0-6.8 (n=18) (7% vs 39%, p = 0.0037).
Of the participants initially hyperuricemic (n=13), 46% were later normouricemic during at least one measurement.
Conclusion: Single sUA measurements were unreliable in hyperuricemia classi cation due to spontaneous variation. Those with an sUA of <6.0 mg/dL were less likely to demonstrate future hyperuricemic measurements and this could be considered a safer threshold to rule out intermittent hyperuricemia based on a single measurement point.
Trial registration: Data from parent study ClinicalTrials.gov Identi er: NCT02038179 Background In addition to its well-described role in the pathogenesis of gout, hyperuricemia (de ned as a serum urate [sUA] > 6.8 mg/dL) is associated with vascular, cardiac and renal disease [1][2][3][4][5][6][7][8] . This hyperuricemic threshold, or similar ones based on sUA distributions in populations of men and women, are commonly used in epidemiological and clinical research 9,10 .
However, the reliability of a single sUA measurement in de ning hyperuricemia is unclear. Previous studies have found signi cant variability in the measurement of serum urate over time, without known causes for the uctuation such as variations in diet, alcohol intake, body weight, time of day, and hydration status, although hourly 11 and seasonal variation in serum urate levels has also been suggested 12-14 . Many epidemiological studies which have explored the question of whether sUA levels are associated with cardiovascular outcomes have used a single-point threshold (such as 6.8 mg/dL or 7.0 mg/dL) to classify individuals as hyperuricemic 9,10 . Even when hyperuricemia is not de ned, sUA levels are often strati ed based on a single measurement 15,16 .
The large number of factors affecting SUA likely account for their propensity to uctuate. Because of this variability, more research is needed as to the value of a single SUA level in con rming or discarding hyperuricemia. Our study results will have implications for epidemiological and clinical studies in hyperuricemia and gout, as explained above hyperuricemia de nitions usually rely on a single measurement.

Methods
Data from the "Serum Urate Reduction to Prevent Hypertension (SURPHER)" study were used. The study protocol has been described in detail 17 . Brie y, SURPHER is a single center, cross-over clinical trial conducted in the city of Birmingham, Alabama (USA) that tested the hypothesis that serum urate reduction with allopurinol could be bene cial in elevated blood pressure in young adults. It was approved by the Institutional Review Board (IRB) at the University of Alabama at Birmingham (UAB -IRB approval number 130408004). Oversight was provided by a Data Safety Monitoring Board. The trial was conducted in accordance with the provisions of the Declaration of Helsinki and the International Conference on Harmonization Good Clinical Practice Guidelines and registered in clinicaltrials.gov (NCT02038179). Written informed consent was obtained from all participants. Key enrollment criteria for the SURPHER trial were 1) The mean of two clinic measurements with systolic blood pressure (SBP) ≥ 120 and <160 mm Hg or diastolic blood pressure (DBP) ≥ 80 and <100 mm Hg; 2) a serum urate ≥5.0 mg/dL for men or ≥4.0mg/dL for women; and 3) age between 18 and 40 years old. Major exclusion criteria included 1) current pharmacological treatment for hypertension (calcium channel blockers were only allowed); 2) prior diagnosis of gout or past use of ULT for gout; 3) >2 alcoholic drinks per day; 4) glomerular ltration rate (GFR)<60 mL/min/1.73 m 2 . Full description of enrollment criteria and study procedures have been published previously 17 .
Participants were initially selected via a telephone screen, followed by a study visit in which a screening sUA was rst measured at an appointment in either morning or afternoon. After a 2 to 4 week run-in period during which participants received a daily dose of placebo to assess adherence, a sUA level was measured a second time prior to any intervention.
Study participants were subsequently randomized to receive oral allopurinol 300mg daily or placebo for three weeks, and a sUA was measured a third time. After a 2 to 4 week washout period, sUA was re-measured prior to crossing over to the other study arm for another 3 weeks. A sUA was then measured a fth and nal time at completion of the study.
The 4-week washout interval allowed for full dissipation of the effects of allopurinol, which is converted to oxypurinol and has a half-life of 18-30 hours 2 . This dissipation was anticipated to take place over no longer than 2 weeks after the last dose of allopurinol. To con rm this in our study, carry-over effects were examined using a two sample t-test.
Samples for sUA level measurements were processed using two Beckman Coulter instruments. Precision statistics performed on these instruments in 2014 demonstrated an observed coe cient of variation in serum urate of 0.3-1.2%.
Except for the initial screening visit, which occurred throughout the day, serum urates were planned to be drawn during appointments between the hours of 0700 and 1100. Participants were asked to fast during the day prior to the screening visit. The samples were collected at a consistent time within the menstrual cycle of premenopausal females enrolled in the study.
For this secondary analysis, sUA levels collected at screening, pre-intervention, pre-placebo, and postplacebo were considered. Levels collected following intervention were not considered. Only the data from the participants that had measurements of all four sUA levels occurring without intervention was used for our analysis. Mean coe cient of variation for sUA was determined and compared across groups (sUA levels at screening, gender) using t-test statistics to determine differences.
The rates of conversion from normouricemia (sUA ≤6.8 mg/dL) to hyperuricemia (sUA >6.8 mg/dL), and from hyperuricemia to normouricemia at were calculated. The rates of conversion to hyperuricemia were then compared across subgroups de ned by the sUA level at initial screening, as well as between subgroups with only one initial normouricemic reading (and a second check which was hyperuricemic) versus two initial normouricemic readings. Comparisons between count data in groups were performed using Fischer's Exact tests, due to small cell size.

Results
Of the 99 SURPHER study participants enrolled, 85 completed all four sUA measurements that occurred in the absence of allopurinol (during the placebo period). The 14 participants who did not complete all four checks were excluded from further analysis. Time of day data was available for 76% of participants, and showed that 84% of measurements outside of the screening sUA occurred during the targeted time frame of 0700 to 1100. Mean + standard deviation study participant (n=85) age was 27.8 ± 7.0 years and mean body mass index was 31.1 ± 7.9. 39% of participants were women. 41% of participants were African-American. (table 1). The mean coe cient of variation in sUA for all qualifying participants was 8.5% ± 4.9% (1% to 23%).
To determine whether a second normouricemic sUA measurement decreases the rate of subsequent conversion to hyperuricemia, a group with one initial normouricemic sUA (and a second check which was hyperuricemic) was compared to a group with two initial normouricemic sUAs. There was a nonsigni cant trend between the group with only one normouricemic check (n=6) vs. two normouricemic checks (n=66) in conversion to hyperuricemia during the 3 rd to 4 th sUA checks, with the latter more likely to remain normouricemic (14% vs. 50%, p=0.0541) ( gure 3).

Discussion
In this study we demonstrated that single-point sUA measurements are unreliable in hyperuricemia classi cation, since about 15% of participants classi ed as normouricemic were found to be hyperuricemic on subsequent checkups and a large nearly half of hyperuricemic participants were normouricemic on subsequent checkups. Our ndings could in uence the way studies using serum urate as an enrollment criteria or outcome are conducted or planned. Our study differed from previous examinations in its calculation of a threshold for screening sUA level more helpful in ruling out future hyperuricemia. We used 6.8 mg/dL as our threshold for hyperuricemia, as this is related to the physiologic saturation at which urate begins to precipitate.
We found a sUA threshold < 6.0 mg/dL, reliable in ruling out subsequent hyperuricemia, as the conversion was just over 7%. Our somewhat small number of subject did not allow us to calculate a comparable threshold value above which persistent hyperuricemia appears to be more likely. A second sUA measurement did not add reliability in excluding future hyperuricemia, as there was no signi cant difference in the conversion rates between those with one initial value below threshold and those with two values below threshold.
There has been a paucity of published data on the rate of conversion from normouricemia to hyperuricemia and from hyperuricemia to normouricemia after an initial laboratory check in the absence of intervention.
A 2004 study examined the serum urate levels and 24-h urinary uric acid levels monthly for a 12-month period in 12 healthy men on self-selected diets, without medications known to effect urate levels, as well as abstinence from alcohol 7 days prior to measurement of levels. Seven of the twelve subjects (58.3%) had transient hyperuricemia at some point during the study period 18 . A similarly designed study found that 10/12 subjects experienced transient hyperuricemia at some point during the course of one year 14 .
The Framingham heart study concluded that a higher percentage of the male population had hyperuricemia when considering four biennial determinations rather than a single determination 6 .
The mean coe cient of variation for sUA was also examined in these two similar studies 14,18 . Our mean coe cient of 8.5% was comparable to the ndings of the 2004 study (9%, with a range of 5-12%) 18 , and is less than that found in the second study mentioned (17.5%, with a range of 15-22%) 14 . In addition to variation over months, it has also been demonstrated that uric acid levels uctuate over the course of a single day. Serum urate levels were signi cantly higher when measured in the morning than when measured in the afternoon, with a decrease of up to 30% seen in the subgroup of diabetic patients 11 .
The cause of the variation in sUA found by our study and previous studies is likely multifactorial. Multiple elements are known to affect sUA levels, including diet and medications. Intake of alcohol, purine-rich foods (including seafood and organ meats such as liver), xylitol and fructose have all been associated with increased serum urate. Medications that increase urate include low-dose aspirin, pyrazinamide, cytotoxic chemotherapy, diuretics (particularly loop and thiazide diuretics), immunosuppressants such as cyclosporine and tacrolimus, nicotinic acid, testosterone, levodopa and theophylline 19,20 . Lead exposure also increases urate and can cause saturnine gout 21 . Disease processes that increase serum urate include renal failure, polycythemia vera, chronic myeloid leukemia and other hematologic malignancies, genetic diseases such as Lesch Nyhan, and any malady that results in acidemia 11 . Alternately, medications that can decrease SUA levels include losartan, amlodipine, feno brate and high dose aspirin 20 .
Even though our study participants were part of a clinical trial and sUA measurements were collected in a relatively short time frame (12-14 weeks), the observed variability in sUA levels could possibly be a result of subtle factors such as time of day, diet, changes in weight, renal function and undisclosed use of medications. As more data becomes available regarding the degree to which these variables affect sUA, a more standardized protocol for checking sUA may become common practice, such as checking sUA early in the morning, prior to any meals.
This variability in sUA may also in uence treatment guidelines for gout in the future. Perhaps in part due to increased recognition of the imprecision of a single spot serum urate level, there is controversy regarding the choice between titration of urate lowering therapy to minimize acute intermittent symptoms and the treat-to-target approach with speci ed urate goals. However, because our study population excluded individuals with gout, our ability to extrapolate in this regard is limited.
Our study has a number of limitations. The initial screening sUA was not standardized with regards to time of day as the rest of the measurements were, and not all the subsequent measurements occurred during the targeted time frame. This discrepancy could account for some of the measured variability in sUA levels, as urate concentration is highest in the morning 11 . However, in current clinical practice it is not standard of care to draw sUA levels at a speci ed time of day, and sUA levels measured at different times are often compared to one another to assess response to treatment. In this sense, our analysis may more accurately represent variability in sUA often appreciated in clinical practice.
The young age of our participants could limit generalizability to older populations. We used the same de nition for hyperuricemia in both males and females, due to the physiologic precipitation of urate at this level; we also attempted to control for changes in urate due to hormonal uctuations by timing our sample collections at similar points in the menstrual cycles of our participants.
A total of 13 study participants presented with an initial sUA in the hyperuricemic range, which could make the conclusions obtained from that group imprecise. This small sub-population size could be the reason why a threshold sUA level above which conversion to hyperuricemia is less likely could not be proposed.
The presence of placebo and allopurinol dosing in between analyzed measurements could have affected our results, though we attempted to minimize these effects by con rming dissipation of the effects of allopurinol by examining carry-over effects with a two sample t-test.
There was also exclusion of 14 out of 99 individuals who did not complete all required sUA checks, which could affect our data if the reason for these omissions were associated with factors affecting serum urate uctuations.

Conclusion
Our data is consistent with previous studies which have found signi cant variation in sUA levels over time, even when controlling for commonly identi ed in uencers of sUA levels. This nding diminishes somewhat the value of a single sUA check, and even a second measurement was not found to add signi cant reliability in excluding future hyperuricemia in those who are initially normouricemic. Future studies could examine the reliability of sUA checks in those with gout, as well as those on ULT, to determine whether a treat-to-target approach for gout are suppression may be subject to these same uctuations in sUA levels.