Abstract
Objective Gout is reportedly associated with a higher incidence of cancer. However, patients with gout tend to have several cancer-related factors including obesity, smoking, and alcohol consumption; thus, the precise association between gout and cancer risk remains unclear. We aimed to investigate the risk of cancer in Korean patients with gout.
Methods Based on the Korea Health Insurance Service database, the subjects comprised patients aged 41–55 years with gout newly diagnosed between 2003 and 2007. We used a multivariable-adjusted Cox proportional hazards model in gout patients and a 1:2 ratio for the matched controls by age, sex, and index year.
Results We compared 4176 patients with gout with 8352 controls. The mean age and follow-up duration were 48.8 years and 10.1 years in both groups. Overall cancer risk was significantly different between gout patients and controls (HR 1.224, 95% CI 1.073–1.398). The all-cause mortality (HR 1.457, 95% CI 1.149–1.847) and cancer mortality (HR 1.470, 95% CI 1.020–2.136) were higher in patients with gout. In the subgroup analysis, the cancer risks of the stomach (HR 1.710, 95% CI 1.221–2.395), head and neck (HR 1.850, 95% CI 1.071–3.196), and hematologic or lymphoid organ (HR 2.849, 95% CI 1.035–7.844) were higher in patients with gout.
Conclusion Patients aged 41–55 years with gout have a higher risk of cancer and all-cause and cancer mortality compared with the general population. Therefore, special attention should be paid to higher cancer risk and mortality in these patients who are diagnosed in middle age.
Gout is a common inflammatory disease characterized by hyperuricemia and recurrent acute arthritis.1 Several studies have reported that hyperuricemia or gout is associated with a higher incidence of cancer.2,3,4,5 In addition, a metaanalysis and a recent prospective cohort study of crystal-proven gout patients showed that hyperuricemia and gout were associated with increased cancer mortality.6,7 However, major confounding factors such as obesity, hypertension, diabetes mellitus, and heavy alcohol consumption are common in patients with gout.8,9,10,11 Metabolic syndrome is a significant risk factor for gout, especially in middle-aged persons,12 and the incidence of cancer increases with age, with a rapid increase beginning in middle age.13 It is therefore worthwhile to evaluate the effect of gout on cancer incidence and mortality in middle-aged patients.
Generally, incidence of cancer types associated with smoking, obesity, and alcohol intake, including colon, nasopharyngeal, and pancreatic cancer, has been reported to be higher in patients with gout.3 Further, in a Taiwan population study, the risk of prostate cancer was reported to be higher in patients with gout than in the general population.4 In a study of a Danish registry, the incidence of hematologic malignancies including multiple myeloma and leukemia was reported to be higher in patients with gout.14 This means that the prevalent type of cancer in gout could be dependent on cultures and vary between countries. Here, we investigated the risk of cancer, cancer mortality, and site-specific cancer types in Korean patients with gout, especially in a middle-aged population.
METHODS
Data sources. The present study was conducted using patient records extracted from the National Health Insurance Service-Health Screening Cohort (NHIS-HEALS) between 2002 and 2015 in Korea.15 The NHIS-HEALS is a cohort of Korean adults who participated in general health screening programs conducted by the NHIS. All subscribers (including medical aid beneficiaries) can receive general health screening biennially or annually (for manual employees) and, in particular, Korean adults aged 40 years or older are eligible for biennial medical checkups.16 This sample cohort first selected qualified individuals between 2002–2003 who were between the ages of 40–79 years and received general medical screening. This cohort consisted of 514,866 individuals who comprised a 10% simple random sample of all health screening participants in that period. The data collected included information on demographics (such as sex, age, region of residence, type of subscription, socioeconomic status [income], disability, death, cause of death), diagnosis (using the International Statistical Classification of Disease and Related Health Problems, 10th revision [ICD-10]), medical procedures, prescriptions, and medical checkup data.
A profile study of the NHIS-HEALS found that the age-standardized prevalence of anemia, diabetes, hypertension, obesity, and hypercholesterolemia in NHIS-HEALS participants was similar to the results in those aged 40 or older who participated in the Korean National Health and Nutrition Examination Survey (KNHANES).15,17 The KNHANES was a cross-sectional survey that consisted of nationally representative samples of the Korean population and had information on health, nutrition, physical conditions, laboratory tests, and chronic diseases.18 This indicates that the NHIS-HEALS is representative of the sample cohort of the general Korean population.
Study design and study subjects. As shown in Figure 1, we compared gout patients with non-gout individuals as the control group. The study cohort included patients aged 41–55 years with gout that was newly diagnosed between 2003 and 2007. Patients who were diagnosed with gout (ICD-10, M10.xx) were treated with gout-related medications, including colchicine, glucocorticoids, nonsteroidal antiinflammatory drugs (NSAIDs), allopurinol, febuxostat, or benzbromarone at the date of diagnosis. Previous validation studies had reported that the positive predictive value of diagnostic codes for gout was 61–86%,19,20 and that of the diagnostic codes in addition to a prescription claim for gout-related medications including colchicine, NSAIDs, or oral glucocorticoids was up to 99%.21 The cohort entry date of patients with gout was the date of the first diagnosis of gout and prescription for gout-related medications. Patients who were diagnosed with gout in 2002 were excluded. Subjects who had no uric acid test information, no measurement data, and who had measurement outliers were excluded. Patients were also excluded if they were diagnosed with a malignancy before or within 1 year of the index date, or if their follow-up duration was < 1 year.
A control cohort included individuals aged 41–55 years who received health checkups between 2003 and 2007 with no diagnostic code for gout and no medical claim for urate-lowering medications. We excluded outlier cases, those diagnosed with cancer before or within 1 year of the index date, and those with a follow-up duration of < 1 year. The cohort entry date of the control group was the date the individual received general medical screening.
Cohort follow-up. In the group with gout, the date of the first gout diagnosis and prescription for gout-related medications was defined as the index date. In the control group, the date of the health checkup was defined as the index date.
If the outcome was a diagnosis of cancer, follow-up started the day after the index date and continued until the occurrence of cancer, death, or end of the study. In case of a mortality outcome, follow-up began from the index date and continued until death or the end of the study.
Study outcome. The primary outcome was newly diagnosed cancer between 2004 and 2015 that required cancer as the main diagnostic code. The secondary outcome was all-cause mortality, cancer mortality, and cancer by site-specific cancer type. In the case of cancer mortality, we worked out the cause of death through “death_code1” and “death_code2” that indicated why the subjects died.
Baseline covariates. All covariates were assessed during the baseline. The following covariates were collected from demographic information including age at diagnosis of gout (or date of medical checkup), sex, income, and region of residence; and comorbid medical conditions including hypertension, diabetes mellitus, kidney disease, chronic obstructive pulmonary disease, and Helicobacter pylori-related diseases, including peptic ulcer (only for stomach cancer). In addition, medical checkup data including smoking, alcohol intake, and BMI, and examination and laboratory data including systolic blood pressure, diastolic blood pressure, glucose, total cholesterol, and levels of hemoglobin, were collected.
Statistical analyses. We created a matched control group by conducting age, sex, and index year 1:2 matching to balance these factors. Then, multivariable-adjusted Cox proportional hazards models and subdistribution hazard models with the other covariates except age, sex, and index year were used to minimize potential confounding factors.
Categorical variables in baseline characteristics are presented as frequencies with percentages, and continuous variables are shown as mean with SD or median with IQR. Categorical variables were compared using the chi-square test, and continuous variables were compared using t tests.
The incidence of outcomes (including cancer, all-cause mortality, cancer mortality, and incident cancer by site-specific cancer type) was calculated as a rate per 1000 person-years (PY). A Kaplan–Meier plot was used to show the cumulative incidence of cancer diagnosis by the 2 groups and to compare differences between them, we conducted a log-rank test. A multivariable-adjusted Cox proportional hazards model with covariates was also used to assess the association between gout and cancer diagnosis and all-cause mortality (HR [95% CI]). When a proportional hazards assumption was violated, we conducted stratified Cox proportional hazards only for covariates to solve this problem. To estimate the crude incidence of cancer mortality and competing mortality, we used the cumulative incidence function (CIF).22 We also conducted a subdistribution hazards model to assess the effect of covariates on the CIF for cancer mortality and mortality from other causes.
All statistical analyses were performed using SAS Enterprise Guide software (version 7.13; SAS Institute) and R software version 3.6.2 (The R Foundation). P < 0.05 was considered statistically significant.
Ethical approval. This study fulfilled the ethical guidelines of the Declaration of Helsinki, and was approved by the Institutional Review Board of Asan Medical Center (number: 2018-0299). The requirement for informed consent was waived because an existing database was used.
RESULTS
Baseline characteristics. According to the medical checkup cohort of the Korean Health Insurance Service database, 4176 patients were newly diagnosed with gout. We included 8352 matched controls in a 1:2 ratio by age, sex, and index year, who presented between 2003 and 2007 (Figure 1). As shown in Table 1, patients with gout were more likely to be nonsmokers and had a greater alcohol intake and more comorbidities than controls. In addition, BMI and systolic and diastolic blood pressure were higher than in the control group. The mean age and follow-up duration for cancer were 48.8 years and 10.1 years, respectively.
Overall cancer risk in patients with gout. As shown in Table 2, the overall cancer occurrence rates in the gout and control groups were 9.241 and 7.218 per 1000 PY, respectively. A Cox proportional hazards model showed that the risk of cancer was significantly different between patients with gout and controls (HR 1.224, 95% CI 1.073–1.398). According to the Kaplan-Meier plots (Figure 2A), the proportion of patients who developed cancer was significantly higher in those with gout than in controls (log-rank test: P < 0.001).
All-cause mortality and cancer mortality. Table 2 shows the all-cause mortality and cancer mortality in patients with gout compared with controls. The overall all-cause mortality rates in the gout group and control group were 2.997 and 1.970 per 1000 PY, respectively. The risks of all-cause mortality (HR 1.457, 95% CI 1.149–1.847) and cancer mortality (HR 1.470, 95% CI 1.020–2.136) were significantly higher in patients with gout than in controls. Also, the plot of the CIF (Figure 2B) indicated that death from cancer was significantly higher in those with gout than in controls (Gray test: P = 0.002).
Individual cancer risk in patients with gout. Table 3 shows the individual cancer risk in patients with gout compared with controls. The risk for stomach cancer was higher in patients with gout than in controls (HR 1.710, 95% CI 1.221–2.395). In addition, the risks of head and neck cancer (HR 1.850, 95% CI 1.071–3.196) and hematologic or lymphoid cancers (HR 2.849, 95% CI 1.035–7.844) were higher than in controls. The risk of other site-specific cancers was not significantly different between patients with gout and controls.
DISCUSSION
Cancer is the second-leading cause of death in the United States23 and the foremost cause of death in South Korea24; the incidence seems likely to increase in the future.25 Gout is a chronic disease that is preventable and manageable; thus, understanding the relationship between gout and cancer risk is important for the reduction of cancer mortality. In this Korean population-based study, patients aged 41–55 years with gout had a higher risk of cancer compared with the non-gout population. Also, all-cause mortality and cancer mortality were higher in patients with gout than in controls. In terms of cancer type, patients with gout have a higher risk of stomach, head and neck, and hematologic or lymphoid cancers compared with the non-gout population.
Previous studies have reported that overall cancer risk is higher by 15–25% in patients with gout compared to those without gout.3,4 Most patients with gout have several risk factors for cancer, including heavy alcohol consumption, a smoking habit, obesity, and metabolic syndrome that are confounding factors for evaluating the causal relationship between gout and cancer risk.10 We used a multivariable-adjusted Cox proportional hazards model in the 41- to 55-year age group to evaluate the relationship between gout and cancer risk more clearly. Patients with gout showed different characteristics compared with controls; however, after adjusting for the risk factors, patients with gout aged 41–55 years showed a higher risk of cancer than controls. In addition, there is an association between gout and higher cancer risk in patients with gout aged 56–79 years (HR 1.126, 95% CI 1.025–1.237; Supplementary Table 1, available with the online version of this article). Thus, all patients with gout aged over 40 years have a higher risk of cancer than those without gout. The results of this study are in line with the aforementioned studies on gout and cancer risk. However, to our knowledge, this is the first evidence to demonstrate the effect of gout on incidental cancer, especially in the middle-aged population. According to our findings, cancer screening may be important in middle-aged patients with newly diagnosed gout. Several studies have reported that gout is associated with a higher risk of all-cause mortality, which is explained by the elevated risk of death from cardiovascular disease.26,27 In regard to cancer mortality, a previous study reported that patients with gout have increased cancer mortality, compared with the general population.6 Similar to the previous study, the all-cause and cancer mortalities were higher in the middle-aged gout patients than the control cohort in the present study. There are several studies reporting that hyperuricemia is associated with an increased risk of cancer mortality2,28; thus, the increased cancer mortality in gout patients may be associated with hyperuricemia. However, the precise relationship between serum uric acid and cancer mortality is still inconclusive, and pharmacological modulation of hyperuricemia may have a role in reducing cancer mortality, but this remains unknown.
There are diverse results in the increased risk of site-specific cancers in patients with gout or hyperuricemia.4,29 In the present study, a higher risk of stomach, head and neck, and hematologic or lymphoid cancers was observed in patients with gout compared with controls. These diverse results may be influenced by methodology, ethnicity, and environmental exposure. Korea has the highest incidence of stomach cancer worldwide,30 and the crude incidence rate of stomach cancer was 59.7 per 100,000 in 2016.24Thus, the increased probability of developing the prevalent type of cancer in gout could be considered. On the other hand, a previous study reported that patients with H. pylori infections showed a higher incidence of gout and hyperuricemia than those without infection.31 Indeed, the incidence of H. pylori-related diseases, including peptic ulcers, is higher in patients with gout than those without it in the present study. However, the risk of stomach cancer was still higher in patients with gout after adjusting for H. pylori–related diseases, including peptic ulcers. Thus, other factors, including hyperuricemia, gout-related medication, or environmental exposure, could contribute to the increased risk of stomach cancer in patients with gout. For hematologic or lymphoid cancers, a previous metaanalysis reported a significant association between hyperuricemia and the risk of lymphoid and hematopoietic system cancers, suggesting that serum uric acid itself has a role in carcinogenesis.28 Another study reported that gout medications are associated with an increased risk of leukemia and non-Hodgkin lymphoma.32 Taken together, the association between the increased risk of site-specific cancer and gout is still inconclusive, and further studies are required to confirm these results.
The present study has several strengths. First, the major confounding factors that contribute to cancer risk were adjusted by the multivariable-adjusted Cox proportional hazards model. Second, the follow-up duration was sufficient to detect the occurrence of cancer. The study also has several limitations. First, the sample cohort included only those aged over 40 years, and the population was relatively small; therefore, there is a limit to the generalization of the results. Second, the serum uric acid level could not be determined in the medical checkup cohort of the Korean Health Insurance Service, and thus the association between hyperuricemia and cancer risk could not be clearly delineated. Third, other potential confounding factors that affect cancer risk, such as medications and various comorbidities, might exist. Indeed, different risk factors for site-specific cancers exist and are difficult to fully evaluate in the present study. Thus, the observed increased risk of site-specific cancer in patients with gout should be interpreted carefully.
In conclusion, patients with gout have a higher risk of cancer, as well as both all-cause and cancer mortality, compared with the general population aged 41–55 years. Therefore, special attention should be paid to higher cancer risk and cancer mortality in patients with gout who are diagnosed in middle age.
ACKNOWLEDGMENT
The authors would like to thank Enago (www.enago.co.kr) for the English language review.
Footnotes
This work was supported by a grant from the Asan Institute for Life Sciences, Asan Medical Center, Korea [2020IP0025]. This research was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (grant number: HI19C0790).
JSL and JM contributed equally to this manuscript.
The authors declare no conflicts of interest.
- Accepted for publication November 4, 2020.
- © 2021 by the Journal of Rheumatology
REFERENCES
ONLINE SUPPLEMENT
Supplementary material accompanies the online version of this article.