Introduction

Target serum urate (sUA) levels for the treatment of hyperuricemia in gout have been set below 5–6 mg/dL [1–4]. In the European League Against Rheumatism (EULAR) recommendations, avoidance of sustained sUA levels below 3 mg/dL was recommended because of an association with mortality and neurodegenerative diseases in some epidemiological studies [1]. By contrast, no limitation to reaching very low sUA levels has been included in either the British Guidelines [2], the American College of Rheumatology (ACR) Guidelines [3], or the most recent British National Institute for Health and Care Excellence (NICE) Guideline [4].

This is a matter of controversy, as sUA in the upper and lower ranges have been associated with undesirable clinical outcomes [5], but only for hyperuricemia in longitudinal clinical studies [6] and not clearly in recent epidemiological studies [7].

While in the design of some clinical trials, low sUA levels < 3 mg/dL were a cause for interchanging therapy arms, as it happened in 14% of patients on febuxostat 120 mg/day [8], and even discouraged for long-term exposure [1], no clinical alert has been raised to date during long-term exposure to pegloticase, a medication that, when effective, induces extremely low sUA levels [9].

This study intends to examine whether exposure to low sUA levels, lower than 3 mg/dL as per the definition of hypouricemia, during urate-lowering therapy (ULT) in patients with gout is associated with increased risk of mortality.

Materials and methods

This study analyzes data from a bidirectional cohort of patients with gout. Patients were prospectively recruited and followed up from 1995 to 2025, at a university hospital integrated into the public healthcare system, with a reference population of approximately 380,000. This cohort was evaluated and approved by our institutional Clinical Ethics of Investigation Committee (CEIC). Written informed consent was obtained from patients whose data were included in the database.

Follow-up was accomplished at either a hospital gout-dedicated office or by retrieving data from the electronic files that include all active and previous treatments, dosing, and adherence as medication possession ratio, along with comorbid conditions.

Evaluation at baseline included retrospective data including general data, comorbid conditions, and variables specifically related to gout. Variables prospectively retrieved during follow-up included urate-lowering medications, medications for the prophylaxis of gout attacks, and sUA levels during treatment. Mortality was retrieved by automatic informatics communication from the national/country mortality database to the local healthcare system.

Comorbid conditions (hypertension, diabetes, hyperlipidemia, renal function impairment) were defined by a previous diagnosis by a practicing physician, prescription of specifically labelled medications for such comorbid conditions, or fulfilling international definitions at the moment of inclusion in the database; therefore, definitions varied from initiation of the cohort in the nineties and mainly during the first decade of the 21st century [10–15]. It is assured that comorbid conditions included were in accordance with international definitions at the time of entrance in the cohort and that recorded medications were correct with regard to prescriptions. Overall comorbidity is routinely evaluated electronically by our healthcare system using the Kaiser-Permanente Pyramid strata (including ongoing hypertension, diabetes, chronic obstructive lung disease, chronic heart failure, and chronic kidney disease [CKD]) and was available from 2010 onwards [16]. Alcohol intake was considered if daily intake was estimated to exceed 15 g/day. Clinically relevant CKD was defined as estimated glomerular filtration rate (eGFR) < 60 mL/min. Previous ULT was defined as a continuous prescription of any ULT lasting at least 6 months. Gout-related variables specifically included time from the onset of gout symptoms (patient-referred), number of joints ever involved by flares (monoarticular, oligoarticular, and polyarticular distribution), presence of subcutaneous tophi on physical examination, ULT previous to baseline visit and during follow-up, medications for the prophylaxis of gout flares at initiation of ULT, and sUA levels at baseline and during follow-up.

sUA level during ULT was selected when stable levels were reached during follow-up after up-titration of urate-lowering medications. Following a treat-to-target sUA approach, sUA < 3 mg/dL during follow-up was specifically pursued as the therapeutic target when patients showed clinical evidence of severe gout, defined as polyarticular joint involvement, presence of subcutaneous tophi, and radiographic erosions at baseline. ULT was maintained even if sUA < 3 mg/dL until remission of flares over 1 year and absence of subcutaneous tophi were achieved, not quite different from published preliminary gout remission criteria [17]. sUA levels were stratified according to the outcome to be studied (< 3 mg/dL), the therapeutic levels (3–5.9 mg/dL), then 6–8.9 mg/dL and ≥ 9 mg/dL to maintain the same 3 mg/dL range as in previous strata. Hospitalization periods, if any, were not considered for the sUA variable.

Statistical analysis was performed using an institutionally licensed V29.0 IBM SPSS package. Descriptive analysis is shown as mean, median, and interquartile range for continuous variables and as percentages for discrete variables. Mortality risk was evaluated initially using univariate Kaplan-Meier Log-Rank survival analysis (Mantel-Cox test) to select those variables that could be plausibly (p < 0.200) associated with mortality [18, 19]. Selected variables were then tested using multivariate Cox proportional hazards survival analysis in order to ascertain variables that were independently associated with increased risk of mortality. Patients with cancer not considered to be in remission or stable were excluded from analysis.

Results

In December 2025, the database included 1,816 patients; 78 were excluded from analysis due to cancer not considered to be in remission or stable at the moment of first visit. Data on sUA levels during therapy were available from 1,618/1,738 (93.1%) patients at the moment of statistical analysis. Over 99.5% of patients were Caucasian.

Death during a mean follow-up of 43.6 months (24, 12–60) was retrieved from 240/1,618 (14.8%) files. Main causes of death were cardiovascular (136/240, 56.7%), cancer (62/240, 25.8%), infections (29/240, 12.1%), and others (13/240, 5.4%).

Mean age at entrance in the cohort was 63.7 years (64, 54–74), and there were 164 (10.1%) women. The mean time from onset of gout was 6.5 years (4, 1–4); polyarticular involvement was present in 555/1,618 (34.3%), tophaceous disease in 511/1,618 (31.6%), and a previous prescription for ULT medication was retrieved in 725/1,618 (44.8%).

Comorbid conditions were frequent: hypertension (974/1,618, 60.2%), hyperlipidemia (917/1,618, 56.7%), diabetes (403/1,618, 24.9%), CKD (595/1,618, 36.8%), and previous vascular event (491/1,618, 30.3%). Significant alcohol intake was observed in 494/1,618 (30.5%) patients.

Univariate survival analysis showed that female gender, age, baseline sUA, sUA during follow-up, polyarticular joint involvement, hypertension, diabetes, previous vascular event, CKD, comorbidity, and tophaceous disease were associated with increased risk of mortality (Table 1). We found no association between ULT (allopurinol vs. febuxostat vs. benzbromarone) and mortality.

sUA levels < 3 mg/dL were more commonly observed in patients on febuxostat, benzbromarone, or combined therapy (Table 2). Mean duration of exposure to sUA < 3 mg/dL was 22 ± 12 months (range 6–60, interquartile range 12–24). In patients showing sUA < 3 mg/dL, the presence of tophi, polyarticular, and erosive disease was, as expected because this target was pursued based on severity, very high: 40/72 (55.6%), 35/72 (48.6%), and 41/72 (56.9%), respectively.

Multivariate survival analysis showed that age, gender, comorbidity, presence of subcutaneous tophi, sUA levels at baseline, and sUA strata were independently associated with increased risk of mortality. sUA levels in the therapeutic range were associated with a lower risk of mortality compared to sUA not in the therapeutic range.

In addition, whereas there was no difference between patients under or over 3 mg/dL but within the therapeutic range, there was a difference between patients out of the therapeutic range but over or under 9 mg/dL (Table 3). This is graphically shown in Figure 1.

Display full size

Figure 1. Survival curve by sUA strata. sUA: serum urate.

Discussion

Epidemiological studies showed an association of both high and low sUA levels with increased risk of mortality, but low sUA shows a similar association, defining a J-shaped mortality curve [20]. This association may not indicate a causality relationship, as hypouricemia is commonly associated to debilitating and potentially lethal diseases, as solid cancer, uncontrolled severe diabetes, liver failure, and severe neurologic diseases [21, 22]. Mortality associated to hypouricemia has been recently investigated, finding no association in women, and all-cause and diabetes-related mortality among hypouricemic diabetic men [7]. Also, no association between low sUA levels and cardiovascular mortality was retrieved from a systematic review and dose-response meta-analysis of cohort studies of over a million participants [23]. The variability of results in epidemiological studies may be explained by differences in population morbidities, but especially by the fact that the very low sUA level index may precede the finding of a yet undiagnosed severe comorbid condition associated by itself with increased risk of mortality, such as cancer, severe diabetes, liver failure, and severe neurologic disorders.

Genetically determined disorders, such as familial hypouricemia type 1 and type 2 due to knock-out of renal tubular transporters URAT1 and Glut 9, respectively [24], and xanthinuria [25], in which affected subjects show very low sUA levels from birth, are not associated with premature mortality or poor cardiovascular outcomes.

The controversy raises to whether this epidemiological association between very low sUA and mortality would suggest to avoid low sUA target during ULT [26] or not [27]. Very low sUA are associated with the fastest reduction of urate deposits, and reaching such levels would be useful to rapidly deplete urate deposition in patients with severe gout [28]. Contrary to what would be expected based on sUA levels, urate-lowering medications that induce low or very low sUA levels, such as febuxostat and pegloticase, showed the highest probability of reduction of major adverse cardiovascular events (MACEs) and all-cause mortality, respectively [29]. Also, there was no difference in the rate of cardiovascular events in patients responding to pegloticase, showing very low sUA levels, compared to non-responders [30].

We previously reported that failure to achieve therapeutic sUA levels was associated to increased mortality [18], but we did not explore very low sUA (< 3 mg/dL) or very high sUA (> 9 mg/dL) at that time. Exposure to sUA < 3 mg/dL was not associated with increased mortality in this bidirectional prospective follow-up cohort after adjustment for other variables. Very low (< 3 mg/dL) sUA levels were more frequently achieved with febuxostat [9], benzbromarone [31], and combined therapy with allopurinol plus benzbromarone [32] and febuxostat with benzbromarone [33].

Although we also reported the association between low sUA levels and mortality in a general population [34], it should not be assumed that low sUA is the cause, but a biomarker of higher risk of mortality due to coexisting comorbid conditions [35].

This is a very large, individual, prospectively followed-up cohort, thus reducing the risk of interpersonal bias. Compared to epidemiological studies, the cohort includes clinical variables regarding gout severity (number of flares, presence of tophi, distribution of joint involvement) and treatment (medications, doses) that are lacking in most databases used for epidemiological studies. Also, a naturalistic approach for ULT was intended, treating to target sUA and even achieving the lowest sUA levels for the most severely diseased patients in order to rapidly debulk urate deposits.

Evidently, there are some limitations. Exposure to sUA < 3 mg/dL (debulking period) was limited in time and not intended during the whole period of treatment. Although corrected for comorbidity, patients with the most severe or unstable diseases could be more prone to involuntary non-adherence to ULT (hospital inpatient periods, for example) that are not captured by the electronic adherence system, and that could be theoretically associated with high sUA levels and also with mortality. Also, definitions of comorbid conditions during a very long recruitment period may induce some noise in the variables.

In conclusion, exposure to very low sUA levels during ULT, at least limited to the debulking period, was not associated with increased mortality. By contrast, it was similar to that observed for the sUA therapeutic range, and lower than that observed in patients showing sUA levels out of the therapeutic range.