Background While several studies have reported a link between the presence of gout and adverse cardiovascular (CV) events in the general population, none has addressed the question of whether the mortality risk of patients with gout is influenced by disease severity.
Methods We applied survival analysis methodology to prospectively collected data on clinical and radiographic measures of disease severity and mortality in a specialty clinic based cohort of 706 patients with gout (1992–2008). Standardised mortality ratios (SMR) were calculated to assess the magnitude of excess mortality among patients with gout compared with the underlying general population.
Results Mean follow-up was 47 months. Tophaceous deposition was present in 30.5% of patients; >4 joints were involved in 34.6% of cases. Mean annual flare rate was 3.4. Arterial hypertension (41.2%), hyperlipidaemia (42.2%), diabetes mellitus (20.1%), renal function impairment (26.6%) and a previous CV event (25.3%) were recorded. 64 (9.1%) patients died, death being attributed to vascular causes in 38 (59%) patients. SMR for gout patients was 2.37 (95% CI 1.82 to 3.03), 1.57 (1.18 to 2.05) and 4.50 (2.06 to 8.54) overall, and in men and women, respectively. The presence of tophi and the highest baseline serum urate (SU) levels were independently associated with a higher risk of mortality, in addition to age, loop diuretic use and a history of a previous vascular event. In the multivariable survival regression models, with time varying covariates, the presence of tophi remained a significant mortality risk after adjustment for baseline SU levels (1.98; 1.24 to 3.20).
Conclusions High baseline SU level and the presence of subcutaneous tophi were both associated with an increased risk of mortality in patients with gout, in most cases attributed to a CV cause. This suggests a plausible pathophysiological link between greater total body urate load and CV disease.
- Cardiovascular Disease
- Outcomes research
Statistics from Altmetric.com
Numerous large studies have documented that having a diagnosis of gout was associated with a higher risk of incident cardiovascular (CV) outcomes, such as myocardial infarction, systolic dysfunction, heart failure and death.1–5 Also, the presence of gout is an adverse prognostic indicator for those who have recently suffered a coronary event.6 Hyperuricaemia and gout are independent risk factors for coronary artery calcification, a epidemiological surrogate measure of coronary atherosclerosis.7 ,8 While impressive as a whole, there are several limitations with the individual studies, the most important of which is the lack of information on important disease characteristics of gout.
The pathophysiological pathways that link gout with CV disease are thought to be based on the presence of chronic inflammation9 during the intercritical periods or on acute inflammation related to the frequency of severe flares.10 However, hyperuricaemia may also cause CV events through induction of endothelial dysfunction and the oxidative stress caused by xanthine oxidase activation.11 If the observed link between gout and CV outcomes is causal, one could expect to document that the frequency of adverse CV outcomes is directly correlated with worsening gout severity, as indicated by the presence of tophi (elevated total body urate load) and uncontrolled hyperuricaemia. The objective of the present study was to test this hypothesis.
Patients and methods
Study settings and patient cohort
The patient cohort comprised all patients with diagnosis of gout by a gout specialist, evaluated at a specialty gout clinic located in a university hospital. This hospital serves patients referred from an industrial and suburban area from the Basque Country with half a million inhabitants, and is also a third stage referral hospital for 2 million other citizens. The patient cohort represents a population with severe gout.12
Inclusion and exclusion criteria
Patients with at least one follow-up visit between 1 January 1992 and 31 December 2008 were included in the present analysis. Patients in whom gout was associated with a haematological malignancy or other malignant disease not confirmed to be in remission were excluded.
Gout was diagnosed by one physician specialising in this condition (FP-R). Wherever feasible, monosodium crystals were demonstrated in synovial fluid samples or material aspirated from suspected tophi. When samples for crystal observation could not be obtained (previous long term treatment or patient unwillingness for joint aspiration), gout was defined as recurrent acute episodes of joint inflammation associated with sustained (>1 year) hyperuricaemia (>0.42 mmol/l) and remission of flares during follow-up after proper control of serum urate (SU) levels.
This study was performed following approval and supervision of the Clinical Investigation Ethics Committee of the Hospital Universitario Cruces. All participants provided signed informed consent.
All patients were evaluated at baseline, followed-up and treated by the same physician (FP-R). Data collected during the first visit to the clinic (baseline visit) included age, gender, time from onset of gout symptoms, body mass index, details of previous treatment history with urate lowering drugs (ULDs), number of joints affected, presence of subcutaneous tophi, radiographic evidence of chronic gout (bone erosions or arthropathy in plain radiographs), number of gout flares in the year preceding evaluation, previous diagnosis of CV disease (including coronary heart disease, heart failure, stroke, peripheral vascular disease or current treatment with acenocumarol as anticoagulant), diuretic use, alcohol intake, diabetes mellitus (fasting glucose level >7.8 mmol/l and/or taking insulin and oral antidiabetic drugs), hypertension (defined as a systolic pressure >140 mm Hg or diastolic >90 mm Hg or on antihypertensive treatment) and hyperlipidaemia (defined as total cholesterol >240 mg/dl or triglycerides >150 mg/dl or on hypolipidaemic drugs). Renal function was assessed at baseline by clearance of creatinine and clearance of uric acid (using 24 h urinary collections), and spot urine sediment examination. Clinically relevant renal impairment was defined as creatinine clearance <60 ml/min/1.73 m2. SU levels were monitored during follow-up, and average SU was used as a variable. All of these variables were included in the analysis.
Medications for gout were also recorded. Usually, all patients were treated in a similar manner, with prophylaxis with low dose colchicine depending on renal function (1 mg orally, each day, for patients with creatinine clearance >60 ml/min, 0.5 mg/day for patients with creatinine clearance 30–60 ml/min and 0.25 mg/day for patients with creatinine clearance 10–29 ml/min) prescribed for 6–12 months, or prednisone (2.5 mg orally, twice daily) if colchicine was not tolerated or contraindicated (as for patients with creatinine clearance <10 ml/min). ULDs were prescribed to patients with SU levels >0.36 mmol/l and to those who agreed to start treatment. ULDs were prescribed and doses adjusted targeting SU levels of <0.36 mmol/l during follow-up. The highest doses of ULDs ever used during follow-up were also updated during follow-up visits.
Follow-up and outcomes
Follow-up visits were established every 3–12 months depending on the patient's clinical status, treatment and comorbidities. Patients who did not attend a follow-up visit were contacted by telephone to ascertain the reason for not attending. Observations of patients who refused further follow-up at the hospital or who were discharged to primary care or to other physicians were censored at the last visit. For those who died, cause of death was ascertained from the patients’ medical records, their families or local registries if needed. Cause of death was then categorised into cancer, vascular (including stroke, heart failure and myocardial infarction), infections and other causes. Sudden death was not assumed, although plausible, to be a CV cause. The cut-off day for the observation period was 31 December 2008. Follow-up spanned from the first visit to the clinic until censor day, date of death or end of follow-up, whichever occurred first.
Overall and cause specific standardised mortality ratios (SMR) were calculated using the Basque Country population mortality data for the year 2005, available from the Spanish Institute of Statistics.
Kaplan–Meier estimates of survival and log rank test were used to initially identify those variables associated with mortality. Variables found to have a possible statistical association with mortality in the bivariate Cox regression analyses (p<0.20) were selected for a multivariate Cox proportional hazard regression analysis using a manual backward stepwise model, so that the variable showing the highest non-significant p value at every step was withdrawn from the multivariate model, until all remaining variables showed statistical significance (p<0.05). HRs and 95% CIs from bivariate and multivariate models are presented. The robustness of the models was checked with partial residual plots to support the assumptions of proportionality of hazards. For numerical variables, the proportional risk hypothesis was tested categorising variables into quartiles. Statistical analyses were run using the statistical package SPSS V.18.
The sample consisted of 706 patients with at least one follow-up visit—662 men (93.8%) and 44 women. Missing values were few (<8% at most). A diagnosis of gout was confirmed by urate crystal observation in 550 patients (79.9%). At the first visit, 436 patients (61.8%) had never received ULDs, 358 patients (50.8%) had 2–4 joints involved and 244 patients (34.6%) had more than four joints involved. The presence of at least a subcutaneous tophus was observed in 215 patients (30.5%), A diagnosis of any CV event was present in 178 (25.3%), arterial hypertension in 291 (41.3%), hyperlipidaemia in 298 (42.5%), diabetes mellitus in 142 (20.2%), loop diuretic use in 83 (11.8%) and heavy alcohol intake in 239 (34.0%). Renal function impairment was observed in 188 (26.6%) patients. Description of the baseline characteristics of the cohort are shown in table 1.
Over the follow-up period, lifestyle changes alone were sufficient to achieve target SU levels in 28 patients (3.9%). ULDs were prescribed in 612 patients (86.7%), of whom 392 (64%) were prescribed allopurinol (doses ranging from 50 to 600 mg/day), 203 (33%) were prescribed benzbromarone (doses ranging from 50 to 200 mg/day) and 17 (3%) were prescribed a combination of different ULDs. The rest refused pharmacotherapy. An average target SU of 0.36 mmol/l was achieved in 505/628 (80.3%) patients during follow-up.
Mean follow-up duration was 47±46 months (median 40; range 1–204 months). During this time, 64 (9.1%) patients died, of whom 55/662 (8.3%) were men and 9/44 (20.5%) were women. The expected number of deaths in this period, given the age and sex distribution, was 27 (SMR 2.37 (95% CI 1.82 to 3.03)). Of the 64 deaths, 38 (69.8%) were attributed to a CV cause (SMR 3.88; 95% CI 2.70 to 5.40). In men (55 cases vs 35 expected), the SMR was 1.57 (95% CI 1.18 to 2.05) and in women (nine cases vs two expected), the SMR was higher (4.50 (95% CI 2.06 to 8.54)). Causes of death are shown in table 2.
These variables, selected using the log rank test (table 3), underwent bivariate and multivariate model Cox regression analysis: age, gender, body mass index, time from onset, number of flares, presence of subcutaneous tophi, polyarticular distribution, gout related arthropathy in radiographs, high ethanol intake, hypertension, diabetes, hyperlipidaemia, renal function impairment, loop diuretic use and previous CV disease. Only four of these variables were found to be independently associated with mortality: age, SU level at baseline, loop diuretic use, presence of tophus and previous CV event (table 4). There was no statistically significant difference in the HR for the two lower quartiles of SU level distribution, but the HR remained statistically significant for the two upper quartiles. HRs for patients showing SU in the third quartile (0.52–0.55 mmol/l) and in the fourth quartile (0.55–0.91 mmol/l) at baseline were 2.44 (95% CI limits 1.14 to 5.23) and 2.93 (95% CI limits 1.41 to 6.08), respectively.
Hyperuricaemia is being considered as an independent, although modest, risk factor for CV events in the context of gout when the diagnosis of gout is included in the variables studied.13 ,14 This consideration remains, however, controversial, as the SU cut-off associated with CV events has yet to be defined, in both men and women.15 Two meta-analyses have shown that hyperuricaemia may be a marginal factor for ischaemic heart disease and mortality.15 ,16 Some authors suggest that the risk conferred by elevated SU varies depending on the overall risk profile of the patient population.17–19 Very high SU levels were reported in association with atherogenesis, independently of hypertension, but the contribution of SU levels in patients with hypertension was also considered to be modest.20 In contrast, hyperuricaemia is a well established risk factor for gout,21 although no study in the meta-analysis15 ,16 included the variable ‘gout’, the most frequent cause of arthritis in the adult population and closely related to persistent hyperuricaemia.21
Compelling evidence supports the fact that hyperuricaemia of gout is undertreated. Less than one in four patients attending general practice keep SU levels on target during follow-up22 and the maintenance rate on urate lowering treatment is low.23 The natural evolution of gout in this context of no treatment is towards the development of radiographic tophaceous gout in 40% of patients at 5 years of follow-up, and up to 75% after 15 years from the onset of gout.24
One could easily follow our hypothesis based on previous findings: gout is a common disease, frequently undertreated and related to CV events, whether directly or through hyperuricaemia. We then went further: maybe it is the inflammation of chronic gout, and not solely hyperuricaemia, that is associated with an increased risk of CV mortality.
Gout is the paradigm of acute inflammation; however, sustained inflammation is an insufficiently highlighted aspect of gout. An enhanced Doppler signal is present in asymptomatic joints of patients with gout, thus supporting the evidence of hidden or non-clinically apparent inflammation.25 Additionally, higher counts of white cells in synovial fluid samples from joints of patients with untreated gout and clinically inactive disease were observed compared with those of patients on proper control of hyperuricaemia.26
Given all of the above, if gout is, by itself, a CV risk factor, would its severity be correlated with poorer CV outcomes? In our study, we confirmed excess mortality in patients with gout compared with the general population, especially in women, probably due to the higher frequency of CV risk factors in patients with gout compared with the general population.27 Gout in women is well known to be associated with comorbidity, mainly hypertension, heart failure and renal insufficiency, but also with tophaceous gout.28 As expected, we found that gout induced by comorbidities, such as previous CV disease and the use of loop diuretics (more commonly associated with chronic heart failure and renal insufficiency), were associated with an increased risk of mortality.
Interestingly, baseline SU level, but not mean SU during follow-up, was associated with increased mortality, but restricted to the two higher quartiles. A relationship between SU levels and the urate pool has been reported.29 This may be in agreement with a study in which gout was associated with the risk of acute myocardial infarction, among gout patients with and without hyperuricaemia.1 Most of our patients achieved optimal SU levels during therapy, and unfortunately there was no long term follow-up of patients not accepting ULDs, so the effect of treating hyperuricaemia of gout could not be compared with not treating it. But this was not an a priori analysis.
The present study seems to confirm the hypothesis of a higher CV risk associated with the severity (presence of subcutaneous tophi) of gout, showing similar figures to those of observational studies in rheumatoid arthritis.30 Despite the fact that number of flares and number of ever inflamed joints were associated with mortality in the bivariate analysis, only the presence of tophi remained associated with an increased risk of mortality in the multivariate analysis. Although the definition of severe gout is not well established, the presence of subcutaneous tophi and a high rate of flares are both considered absolute indications for long term urate lowering treatment by evidence based expert recommendations.31 Subcutaneous tophi are associated with the presence of articular tophi32 and also with radiographic findings.33 Nevertheless, simple radiography has been shown to be less sensitive than ultrasound in detecting tophaceous erosions and tophaceous deposits in the joints.34 Finally, tophaceous gout was associated with Q wave myocardial infarction only in a subgroup of young patients with gout,35 although in that study mortality was not evaluated.
This study has two strengths. First, patients were prospectively followed-up by the same physician for the entire observational period, thus making the evaluations homogenous throughout the study. Second, the severity of gout was evaluated at baseline in all patients, data which are unavailable in all databases that have currently provided data on mortality in gout.
Some limitations were also inherent in this study. We were not able to observe any effect of ULDs on mortality, as most patients were treated, and also SU while on therapy did not seem to have an impact on mortality in hyperuricaemic patients in a recent study.19 Also, the study was not designed or powered to test such an effect. Other variables that may also be associated with increased mortality, such as smoking or previous treatment of gout flares with non-steroidal anti-inflammatory drugs, were not included in the protocol database, and it should be taken into consideration that patients with the most severe gout would be expected to use these drugs more frequently, as shown by audits on treatment.36 Some could argue that there was selection bias as we used a hospital based population of gout patients. Although such selection exists, it favoured our hypothesis testing, as patients had more severe disease than in a general practice setting, and it allowed us to include nearly one-third of severe patients and patients with a high prevalence of previous CV events and morbid conditions known to be CV risk factors in the analysis. Indeed, despite the high impact on mortality in our patients derived from the high prevalence of CV comorbidities and shown by the high SMR compared with our general population, the presence of tophaceous deposition (as a variable for gout severity) still remained statistically significant as an independent risk factor for mortality in our cohort.
In summary, we have observed an excess of mortality in gout, and that CV risk factors play an important role, but also that the severity of gout—presence of subcutaneous tophi—and previous exposure to high SU levels may play a role in mortality. This suggests a plausible pathophysiological link between greater total body urate load and CV disease. Therefore, avoidance of delayed treatment of gout to avoid the development of tophaceous gout should be considered.
Handling editor Tore K Kvien
Contributors FP-R and AMH-B: conception and design. FP-R and LM-I: analysis of the data. FP-R, LC, AMH-B and EK: drafting the article. All authors: interpretation of the data, revising the manuscript critically for important intellectual content and final approval of the version to be published.
Funding This work was supported by the Asociación de Reumatólogos del Hospital de Cruces, grant No 2008/03.
Competing interests FP-R: grants from Ministerio de Sanidad, Gobierno de España and Asociación de Reumatólogos del Hospital de Cruces; advisory board for Menarini; consultancies for Arde, Menarini, Metabolex, Novartis and Pfizer; speakers bureau for Ardea, Menarini and Novartis; educational presentations for Savient. AMH-B: grant from Asociación de Reumatólogos del Hospital de Cruces. LC: speaker for Menarini. EK: consultancies for Takeda, Metabolex and URL Pharma; grants from Takeda.
Ethics approval The study was performed following approval and supervision from the Clinical Investigation Ethics Committee of the Hospital Universitario Cruces.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.