Article Text

Download PDFPDF

Extended report
Impact of diabetes against the future risk of developing gout
  1. García Rodríguez1,
  2. Lucia Cea Soriano1,
  3. Hyon K Choi2
  1. 1Spanish Centre for Pharmacoepidemiological Research (CEIFE), Madrid, Spain
  2. 2Section of Rheumatology and the Clinical Epidemiology Unit, Boston University School of Medicine, Boston, Massachusetts, USA
  1. Correspondence to Dr Hyon K Choi, Section of Rheumatology and the Clinical Epidemiology Unit, Boston University School of Medicine, 650 Albany Street, Suite 200, Boston, Massachusetts 02118, USA; hchoius{at}bu.edu

Abstract

Objective Although hyperuricaemia and gout are associated with an increased future risk of diabetes, diabetes may reduce the future risk of gout through the uricosuric effect of glycosuria or the impaired inflammatory response. The present work evaluates the impact of diabetes on the future risk of developing gout.

Methods A case-control study nested in a UK general practice database (the health improvement network) was conducted by identifying all incident cases of gout (N=24 768) and randomly sampling 50 000 controls who were 20–89 years between 2000 and 2007. The independent effect of type 1 and type 2 diabetes on the development of incident gout was examined.

Results After adjusting for age, sex, body mass index, general practitioner visits, smoking, alcohol intake, ischaemic heart disease and presence of cardiovascular risk factors, the RR for incident gout among patients with diabetes, as compared with individuals with no diabetes was 0.67 (95% CI 0.63 to 0.71). The multivariate RRs with the duration of diabetes of 0–3, 4–9 and ≥10 years were 0.81 (95% CI 0.74 to 0.90), 0.67 (95% CI 0.61 to 0.73) and 0.52 (95% CI 0.46 to 0.58), respectively. The inverse association was stronger with type 1 diabetes than with type 2 diabetes (multivariate RR, 0.33 vs 0.69) and stronger among men than women (p value for interaction <0.001).

Conclusions Individuals with diabetes are at a lower future risk of gout independent of other risk factors. These data provide support for a substantial role of the pathophysiology associated with diabetes against the risk of developing gout.

Statistics from Altmetric.com

Request Permissions

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.

Introduction

Although diabetes is considered a comorbid condition of gout, a common and excruciatingly painful inflammatory arthritis,1,,3 pathophysiological mechanisms associated with diabetes suggest that having diabetes may actually reduce the future risk of gout. For example, several previous studies have shown that individuals with type 2 diabetes (or blood glucose >10 mmol/litre (180 mg/dl)) have substantially lower serum uric acid levels than normal individuals, likely due to the uricosuric effect of glycosuria.4,,8 Furthermore, the impaired inflammatory response observed in diabetes may play a protective role against the risk of gout, which is the direct result of an intense inflammatory response to urate crystals.9 While these data suggest that diabetes could lead to a lower future risk of gout, no study has investigated this potential influence of diabetes on the future risk of developing gout.5 To examine this link, we analysed a cohort of 24 768 patients newly diagnosed as having gout and 50 000 controls from the health improvement network (THIN) database.

Methods

Study population

The THIN database contains computerised medical records entered by general practitioners (GPs) in the UK.10 Data on approximately 4 million patients are systematically recorded and sent anonymously to THIN. THIN collects and organises this information in order to be used for research projects. The computerised information includes demographics, details from GP visits, diagnoses from specialist referrals and hospital admissions, results of laboratory tests and a free text section (information available on request from the authors). Prescriptions issued by the GP are directly generated from the computer. An additional requirement for participating practices is recording of the indication for new courses of therapy. The Read classification is used to code specific diagnoses and a drug dictionary based on data from the Multilex classification is used to code drugs (National Health Service (NHS) terminology service; http://www.connectingforhealth.nhs.uk/terminology/readcodes) (Multilex drug data file; http://www.firstdatabank.co.uk/products/multilex/). The current study was approved by the NHS South-East Multi-Centre Research Ethics Committee.

We conducted a nested case-control study using data from the THIN database between January 2000 and December 2007. The cohort included all individuals aged 20–89 years with a registration status of permanent or died in the last THIN update. Study cohort members were required to have ≥2 years of enrolment with the GP, at least one GP visit and one prescription in the 2 years before entering the study period. Start date corresponded to the date when all these eligibility criteria were met. Individuals had to be free of gout and cancer before entering the study. Members of this cohort were followed until the date of one of the following end points: gout detection, 90th birthday, death, or end of study period, whichever came first. The final cohort consisted of 1 775 505 persons followed for an average of 5.2 years.

Ascertainment of gout

We identified all individuals with a first ever diagnosis of gout recorded by a GP with an automatic search using Read codes. The date of gout onset (index date) was defined as the earliest of the date of first diagnosis of gout or first antigout treatment (allopurinol, colchicine and uricosuric drugs) among individuals with a diagnosis of gout. We considered incident cases to be all patients with gout with an index date (gout onset) occurring after the cohort entry start date (N=24 768).

To evaluate the robustness of gout ascertainment, we performed a sensitivity analysis restricting gout cases to those receiving antigout treatment. To this end, we used the following operational definition: we identified within 90 days after first ever diagnosis of gout (index date) any antigout treatment (colchicine, probenecid and uricosuric drugs) and/or a prescription of non-steroidal anti-inflammatory drugs on the same index date. A similar case definition of gout has been shown to have a validity of 90% in the General Practice Research Database (GPRD).11 12

Control sampling

We selected 50 000 controls that were randomly sampled from the same cohort where gout cases were ascertained. A date encompassed within the study period was generated at random for each of the members of the cohort, if the random date for a study member was included in his/her eligible person-time (follow-up period), we marked that person-day as an eligible control. The same exclusion criteria were applied to controls as to cases. Controls were frequency matched to cases by age within 1 year, sex and calendar year (year of first ever gout diagnosis). The random date for controls was used as index date in the case-control analysis.

Exposure assessment (type 1 and type 2 diabetes)

We identified all individuals with diabetes mellitus in the computerised medical records occurring before the index date, as described elsewhere in detail.13 The date of diabetes onset was defined as the minimum of the date of first ever recorded diagnosis of diabetes or first antidiabetic prescription. The types of diabetes have been described elsewhere in detail.13 Briefly, if diabetes type was recorded and the GP had recorded only one specific diabetes type, it was assigned to that type of diabetes. When no diabetes type was recorded or both types were recorded in one patient, we used age at diagnosis to determine the type. For example, individuals diagnosed as having both types, younger than 30 years and treated with insulin were assigned to type 1, and individuals older than 50 years treated primarily with oral antidiabetics were assigned to type 2. For the rest of the patients, we manually reviewed each patient's electronic medical records to assign a diabetes type. There were no patients with type 1 diabetes without at least one insulin prescription. For type 2 diabetes we subdivided patients according to treatment into the following groups: no pharmacological treatment prior to index date, treated only with oral antidiabetics and treated with insulin. We calculated duration of diabetes as the time interval between the date of first diagnosis of diabetes and the index date.

Covariate assessment

Demographics and traditional life style factors such as alcohol use, smoking and body mass index (BMI), as well as comorbidities such as ischaemic heart disease (IHD), hypertension, hyperlipidaemia and renal failure, were collected from the database before the index date. The numbers of GP visits, referrals and hospitalisations in the year prior to the index date were also ascertained.

Statistical analysis

We estimated the OR and 95% CI for gout associated with diabetes by means of unconditional logistic regression, adjusted for the frequency matched variables, GP visits and the following potential risk factors: BMI (five categories), alcohol use (six categories), smoking (non-smoker, current smoker, past smoker and unknown), IHD, hypertension, hyperlipidaemia and renal failure. Under our nested case control study design, using incidence density sampling, the OR is an unbiased estimator of the incidence rate ratio of the underlying cohort. Thus, we have reported OR estimates as RRs throughout this paper.

We performed stratified analyses by gender because previous reports suggested the inverse associations between diabetes and serum uric acid levels were significantly stronger among men than women.5 14 We tested the significance of the interaction with a likelihood ratio test by comparing a model with the main effects of the two variables (diabetes and sex) and the interaction term with a reduced model with only the main effects. Similarly, we examined the potential interaction between diabetes and diuretic use, as the latter is a strong risk factor of hyperuricaemia and gout.

Results

The study included 24 768 newly diagnosed cases of gout and 50 000 controls, and their characteristics are shown in table 1. As anticipated, gout was associated with increased GP visits, alcohol use, increased adiposity, IHD, hypertension, hyperlipidaemia and renal failure.

Table 1

Characteristics of cases of gout and frequency matched controls

The proportion with a history of type 1 diabetes among gout cases was lower than that among controls (0.22% vs 0.4%). The proportion with type 2 diabetes among gout cases was numerically higher than that among controls (9.48% vs 7.37%), with a notable presence of confounders among individuals with type 2 diabetes (eg, high BMI, renal insufficiency, hypertension and GP visits). After adjusting for age, gender, calendar year and GP visits, the RR for gout among individuals with overall diabetes was 0.85 (95% CI 0.80 to 0.90). After adjusting for other covariates, the RR became more protective (0.67; 95% CI 0.63 to 0.71). The inverse association was stronger with type 1 diabetes than with type 2 diabetes (multivariate RR, 0.33 vs 0.69; p value <0.001) (table 2). Furthermore, the inverse association was stronger with treated type 2 diabetes than with untreated type 2 diabetes (multivariate RR, 0.61 vs 0.91; p value <0.001). The magnitude of association with treated type 2 diabetes did not materially differ between cases treated with insulin and with oral antidiabetic agents (eg, multivariate RR among men, 0.52 vs 0.55). In our sensitivity analysis, restricting gout cases to those with antigout treatment (N=19 749), the multivariate RR was 0.69 (95% CI 0.64 to 0.73) with overall diabetes, 0.37 (95% CI 0.26 to 0.52) with type 1 diabetes and 0.70 (95% CI 0.66 to 0.75) with type 2 diabetes.

Table 2

Relative risk of incident gout associated with diabetes

There was a graded, inverse association between the duration of diabetes and the future risk of gout (p for trend <0.001). The multivariate RRs with the duration of all diabetes of 0–3, 4–9 and ≥10 years were 0.81 (95% CI 0.74 to 0.90), 0.67 (95% CI 0.61 to 0.73) and 0.52 (95% CI 0.46 to 0.58), respectively. This graded association remained evident for type 2 diabetes (p for trend <0.001), whereas the strong inverse association with type 1 diabetes was evident from the first year of the diagnosis without a substantial change with a longer duration of diabetes (p for trend=0.84) (table 3).

Table 3

Relative risk of gout by duration of diabetes

In our stratified analyses, the inverse association was significantly stronger among men than women (multivariate RR, 0.59 vs 0.90; p value for interaction <0.001) (table 2). Similarly, the association was stronger among individuals without diuretic use than among diuretic users (multivariate RR, 0.58; 95% CI 0.53 to 0.64 vs multivariate RR, 0.72; 95% CI 0.66 to 0.79; p value for interaction, 0.004). In all subgroups we evaluated, the inverse association with type 1 diabetes remained stronger than with type 2 diabetes.

Discussion

In this large general practice cohort, we found that men with diabetes had a 41% lower future risk of gout. This inverse association was independent of age, BMI, smoking, alcohol intake, presence of IHD, hyperlipidaemia, hypertension and renal failure. Furthermore, the inverse association was evident with type 1 and type 2 diabetes, although the risk reduction associated with type 1 diabetes was substantially larger (73% vs 39% reduction). Similarly, treated type 2 diabetes, a marker for severe diabetes or high levels of glucose, showed a lower risk of gout than untreated diabetes. Type 2 diabetes showed a clear dose response relationship with the duration of diabetes, whereas type 1 diabetes revealed a reduced risk of gout from the first year of the diagnosis. Finally, the inverse association was weaker among women, as suggested by previous findings based on serum uric acid measurements.5 14 The current study provides the first large-scale evidence base for the independent protective effect of diabetes against the future risk for gout.

Individuals with diabetes are at a lower future risk of gout likely through the uricosuric effect of glycosuria, which occurs when the blood glucose level is greater than approximately 10 mmol/litre (180 mg/dl).5 6 For example, a nationally representative study showed that a haemoglobin A1c (HbA1c) level ≥9% was associated with a lower serum uric acid level than a HbA1c level <5% by 1.1 mg/dl among men and by 0.4 mg/dl among women (p value for interaction <0.0001). Corresponding ORs for hyperuricaemia as an outcome showed 0.24 (95% CI 0.09 to 0.67) among men and 0.51 (95% CI 0.25 to 1.03) among women.5 Similarly, studies have shown that a history of diabetes was significantly associated with lower levels of serum uric acid4,,8 and the inverse association was significantly stronger among men than women.5 These previous data based on serum uric acid, the causal intermediate of gout, closely agree with the current study findings based on gout, the clinical outcome.

Our findings might appear counterintuitive given the strong cross-sectional associations between hyperuricaemia, gout and the metabolic syndrome,9,15,,20 which is considered a pre-diabetes status. However, it is important to note that this strong association exists during pre-diabetes status when increased serum urate levels are primarily influenced by increased insulin levels associated with insulin resistance, without the counteracting uricosuric effect of glycosuria, which becomes evident when type 2 diabetes develops. For example, a large prospective cohort study showed that, as compared with normal individuals, the prevalence of hyperuricaemia was 74% higher in men with pre-diabetes, but was 63% lower in men with type 2 diabetes.4 The corresponding serum uric acid level differences compared with normal individuals were 0.3 mg/dl and −0.77 mg/dl.4 This difference became increasingly negative with increasing duration of type 2 diabetes.4 Thus, these data suggested that individuals with pre-diabetes are at a higher risk of developing gout, but once they develop diabetes their risk drops to a lower level than that of non-diabetic individuals, as our data also indicate. The latter notion was further supported by the stronger protective association seen with type 1 diabetes and a longer duration of type 2 diabetes, likely due to a prolonged period of relative hypouricaemia in these cases. Also, it is conceivable that the lack of pre-diabetes status with increased insulin resistance in the pathogenesis of type 1 diabetes makes its inverse association with gout stronger and larger than with type 2 diabetes, as observed in our study.

Furthermore, the impaired inflammatory response observed in diabetes may play a protective role against the risk of gout, which is the direct result of an intense inflammatory response to urate crystals.9 Urate crystals are able to directly initiate, amplify and sustain an intense inflammatory attack because of their ability to stimulate the synthesis and release of humoral and cellular inflammatory mediators.9 Many of these inflammatory components have been found to be impaired in diabetes, including decreased microvascular responses to inflammatory mediators such as histamine and bradykinin,21 22 reduced protein leakage and oedema formation,23,,25 reduced mast cell degranulation,26 impairment of neutrophil adhesion to the endothelium and migration to the site of inflammation,27,,30 production of reactive oxygen species and reduced release of cytokines and prostaglandin by neutrophils,31,,33 increased leucocyte apoptosis and reduction in lymph node retention capacity.34 35 These findings further provide potential biological mechanisms that may underlie the inverse association observed in this study.

Our longitudinal findings add an interesting twist to the accumulating evidence for the independent impact of hyperuricaemia and gout on the risk of type 2 diabetes.36,,38 For example, in a lifestyle intervention study of high-risk middle-aged subjects with impaired glucose tolerance, baseline uric acid and its changes predicted a twofold increase in the likelihood of developing type 2 diabetes.36 Similarly, men with gout have been found to have a higher future risk of type 2 diabetes independent of other known risk factors.37 Taken together with our current findings, one could conclude that while gout increases the future risk of diabetes, diabetes decreases the future risk of gout. These opposing directions of effect explain well the cross-sectional associations between the two conditions, which were surprisingly modest despite their strong associations with the metabolic syndrome. For example, a cross-sectional analysis based on the UK GPRD found an 11% increased prevalence of type 2 diabetes among individuals with gout,3 and a study of male health professionals reported that the prevalence of diabetes was 5% among men with gout and 3% among men without gout.39 These modest, positive cross-sectional associations are likely to be the residual associations after cancelling out the opposing directions of risk posed by one condition on the future risk of the other.

Strengths and limitations of our study deserve comment. This study was performed using a large UK general practice; therefore, findings are likely to be applicable to the general population. Because the definition of gout was based on GP diagnoses, a certain level of misclassification of exposure is inevitable. A diagnosis of gout could have often been recorded based on the suggestive clinical presentation of gout without documentation of monosodium urate crystals. Any non-differential misclassification of these diagnoses would have biased the study results toward the null and would not explain the strong associations and dose-response relationships observed in this study. Furthermore, when we used GP diagnoses of gout combined with antigout medication use as our case definition, our results remained similar. Finally, a potential detection bias (ie, increased detection of gout due to clinical care associated with diabetes) would not explain the observed inverse association.

In conclusion, our findings suggest that individuals with diabetes are at a lower future risk of gout independent of other known risk factors and the protective effect was stronger among men with type 1 diabetes or with long standing type 2 diabetes. These data provide support for a substantial role of the pathophysiology associated with diabetes (eg, the uricosuric effect of glycosuria and the impaired inflammatory response) against the risk of developing gout.

References

View Abstract

Footnotes

  • Competing interests HKC has received research funding from TAP Pharmaceuticals for other research projects. In addition, HKC has received honoraria from, and serves on advisory boards to, Takeda Pharmaceuticals and Savient.

  • Funding This work was supported in part by grants from the National Institute of Health (AR47785). Spanish Centre for Pharmacoepidemiological Research has received an unrestricted research grant from Novartis to work on other projects related to gout. The funding sources had no role in the design, conduct, or reporting of the study or in the decision to submit the manuscript for publication.

  • Ethics approval This study was conducted with the approval of the NHS South-East Multi-Centre Research Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.