Abstract
Objectives
Serum uric acid has been suggested to be associated with cancer risk. We aimed to study the association between serum uric acid and cancer incidence in a large Swedish cohort.
Results
A positive association was found between uric acid levels and overall cancer risk, and results were similar with adjustment for glucose, triglycerides and BMI. Hazard ratio (HR) for overall cancer for the 4th quartile of uric acid compared to the 1st was 1.08 (95% CI: 1.05–1.11) in men and 1.12 (1.09 – 1.16) in women. Site-specific analysis showed a positive association between uric acid and risk of colorectal, hepatobiliary, kidney, non-melanoma skin, and other cancers in men and of head and neck and other cancers in women. An inverse association was observed for pulmonary and central nervous system (CNS) cancers in men and breast, lymphatic and haematological, and CNS malignancies in women.
Materials and Methods
We included 493,281 persons aged 20 years and older who had a measurement of serum uric acid and were cancer-free at baseline in the AMORIS study. Multivariable Cox proportional hazards regression was used to investigate sex-specific quartiles of serum uric acid in relation to cancer risk in men and women. Analysis was further adjusted for serum glucose, triglycerides and, where available, BMI. Site-specific analysis was performed for major cancers.
Conclusions
Altered uric acid levels were associated with risk of overall and some specific cancers, further indicating the potential role of uric acid metabolism in carcinogenesis.
Keywords: uric acid, cancer, prospective study
INTRODUCTION
Uric acid is the product of the metabolism of dietary or endogenous adenine-based and guanine-based purines, and is excreted by the kidney and gut [1, 2]. Uric acid has paradoxically been found to have the characteristic of being an antioxidant in the extracellular environment, whilst having prooxidative effects in the intracellular environment [3, 4]. As an antioxidant, uric acid acts as a scavenger of oxygen radicals, and thus may serve to reduce carcinogenic reactive oxygen species (ROS) [2, 5]. ROS are carcinogenic as they increase the mutation rate in cells, and therefore increase their oncogenic potential [5, 6]. As a prooxidant, uric acid contributes to tumourigenesis by entering normal cells and promoting tumour cell proliferation, migration, and survival, mediated by ROS and inflammatory stress [7]. Recent epidemiological studies looking at the association of serum uric acid and cancer incidence have found conflicting results. A cohort study of 354,110 participants in Taiwan supported the hypothesis that uric acid was cancer-protective as it suggested that low serum uric acid levels were associated with a higher risk of all cancer mortality relative to high serum uric acid levels [8]. Conversely, a cohort study of 83,683 participants in Austria found that high levels of serum uric acid were linked to a higher risk of all cancer mortality [9]. Furthermore, a cohort study of 8,274 participants in China found that incidence of cancer increased with serum uric acid levels, but only in females with diabetes [10]. A recent systematic review and meta-analysis of 17 studies on the link between serum uric acid and cancer incidence and mortality was inconclusive [2].
A link between uric acid and cancer that involves metabolic syndrome (MetS) has also been speculated. MetS is characterised by insulin resistance, hypertension, abnormal lipids, and chronic inflammation [11]. MetS is associated with higher frequencies of elevated serum uric acid levels (hyperuricemia), and, likewise, hyperuricemia is associated with subsequent MetS [12, 13]. Furthermore, studies using genetic uric acid scores as instrumental variables suggested uric acid has a bystander role in MetS [14]. Despite this strong relationship, the effect of MetS on the association between uric acid and cancer has not yet been explored.
High consumption of animal protein and seafood have been associated with higher prevalence of hyperuricemia [15]. Such consumption habits may reflect an unhealthy diet and lifestyle [16], suggesting serum uric acid is an inverse marker of health. Several studies have shown that high serum uric acid is an independent risk factor for hypertension [17], diabetes [18], CVD, and mortality [19].
We studied the association between serum uric acid levels and the incidence of overall and site-specific cancer in the Swedish Apolipoprotein-related MOrtality RISk study (AMORIS) including 493,281 adults, who were followed-up for up to 25 years. Additionally, we took into account serum glucose and triglycerides levels to account for the potential confounding by MetS on the association between circulating uric acid and cancer risk.
RESULTS
A total of 72,349 persons developed cancer during follow up (mean: 19.47 years). More cancers were observed in men (56.96%). The mean age of participants at baseline was 45 years. The majority of the study population (90.7%) was gainfully employed (Table 1). Overall, participants with cancer diagnosis during follow-up had higher baseline uric acid levels compared to the group without cancer.
Table 1. Descriptive characteristics of study population by cancer status.
| No Cancer N=420,932 | Cancer N=72,349 | |
|---|---|---|
| Age (years) | ||
| Mean (SD) | 43.22 (13.75) | 52.56 (11.51) |
| Gender – N (%) | ||
| Male | 223592 (53.12) | 41207 (56.96) |
| Female | 197340 (46.88) | 31142 (43.04) |
| Education Level – N (%) | ||
| Low | 102391 (25.43) | 22088 (32.02) |
| Medium | 182446 (45.31) | 28896 (41.90) |
| High | 117833 (29.26) | 17988 (26.08) |
| Socioeconomic Status – N (%) | ||
| White collar | 194672 (46.25) | 37281 (51.53) |
| Blue collar | 185694 (44.11) | 29787 (41.17) |
| Not gainfully employed/missing | 40566 (9.64) | 5281 (7.30) |
| Serum Uric Acid* – N (%) | ||
| Quartile 1 | 105080 (24.96) | 15770 (21.80) |
| Quartile 2 | 107181 (25.46) | 17114 (23.65) |
| Quartile 3 | 105913 (25.16) | 18528 (25.61) |
| Quartile 4 | 102758 (24.41) | 20937 (28.94) |
| Follow up time (years) | ||
| Mean (SD) | 20.51 (5.35) | 13.43 (6.68) |
| CCI category – N (%) | ||
| 0 | 399736 (94.96) | 67444 (93.22) |
| 1 | 15218 (3.62) | 3597 (4.97) |
| 2 | 3666 (0.87) | 795 (1.10) |
| 3+ | 2312 (0.55) | 513 (0.71) |
* Serum uric acid quartiles (μmol/L): Quartile 1 (<281), quartile 2 (281-319), quartile 3 (319-362), quartile 4 (>362) in men, quartile 1 (<207), quartile 2 (207-240), quartile 3 (240-279), quartile 4 (>279) in women. CCI = Charlson comorbidity. index.
Higher cancer risk was observed with increasing levels of serum uric acid (Table 2), with a HR of 1.09 (95%CI: 1.06 – 1.13) for each log unit increase of serum uric acid in the multivariable model. Results were similar with the crude levels of uric acid (data not shown). Further adjustment for serum glucose and TG showed similar results (e.g. HR for overall cancer for each log unit increase of serum uric acid was 1.07 (95%CI: 1.03 – 1.11)). Analysis in the subgroup with information of BMI showed similar albeit weaker results owing to smaller sample size, for instance, HR for every log unit increase of serum uric acid was 1.11 (95%CI: 1.00 – 1.24)) whereas the trend for the quartiles of serum uric acid was weaker (Ptrend = 0.07). No marked difference was seen with adjustment for BMI. A sensitivity analysis excluding participants with follow up less than 2 years was consistent with these results (Supplementary Table 2, Supplementary Information). A positive association was also found between uric acid levels and risk of death (Supplementary Table 3, Supplementary Information). When analysis was stratified by gender, similar associations were observed in men (e.g. HR for log unit increase of serum uric acid: 1.18 (95%CI: 1.12 – 1.24)) and women (e.g. HR for log unit increase of serum uric acid: 1.23 (95%CI: 1.17 – 1.29)), although we found a statistically significant interaction between serum uric acid level and gender (Table 3).
Table 2. Hazard ratios (HR) and 95% Confidence Intervals (95%CI) for the risk of cancer by log and sex-specific quartiles of serum uric acid.
| HR (95%CI) | ||||
|---|---|---|---|---|
| Model 1 | Model 2* | Model 3† | Model 4§ | |
| N (Cancer/Total Participants) | 72,349/493,281 | 72,349/493,281 | 72,349/493,281 | 9,447/66,931 |
| Log serum uric acid | 1.94 (1.88 – 2.00) | 1.09 (1.06 – 1.13) | 1.07 (1.03 – 1.11) | 1.11 (1.00 – 1.24) |
| Serum uric acid | ||||
| Quartile 1 | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) |
| Quartile 2 | 1.05 (1.03 – 1.07) | 1.00 (0.98 – 1.03) | 1.00 (0.98 – 1.03) | 0.98 (0.92 – 1.03) |
| Quartile 3 | 1.16 (1.14 – 1.19) | 1.01 (0.99-1.03) | 1.01 (0.99 – 1.03) | 1.01 (0.96 – 1.08) |
| Quartile 4 | 1.46 (1.43 – 1.49) | 1.05 (1.03 – 1.07) | 1.04 (1.02 – 1.06) | 1.05 (0.99 – 1.11) |
| Ptrend | <0.0001 | <0.0001 | 0.0002 | 0.07 |
*Adjusted for age, gender, education level, SES and CCI category.
†Adjusted for age, gender, education level, SES, CCI category, serum glucose and triglycerides.
§Adjusted for age, gender, education level, SES, CCI category and BMI in the subgroup with BMI.
Table 3. Gender stratified hazard ratios (HR) and 95% Confidence Intervals (95%CI) for the risk of cancer by log and sex-specific quartiles of serum uric acid.
| No. of cancer/Total participants | Men HR (95%CI) | No. of cancer/Total participants | Women HR (95%CI) | |
|---|---|---|---|---|
| Total group | 41207/264799 | 31142/228482 | ||
| Log serum uric acid | 1.18 (1.12 – 1.24) | 1.23 (1.17-1.29) | ||
| Serum uric acid | ||||
| Quartile 1 | 9471/64570 | 1.00 (Ref) | 6299/56280 | 1.00 (Ref) |
| Quartile 2 | 9945/67089 | 1.00 (0.97 – 1.03) | 7169/57206 | 1.03 (1.00 – 1.07) |
| Quartile 3 | 10596/66812 | 1.05 (1.02 – 1.07) | 7932/57629 | 1.03 (0.99 – 1.06) |
| Quartile 4 | 11195/66328 | 1.08 (1.05 – 1.11) | 9742/57367 | 1.12 (1.09 – 1.16) |
| Ptrend | <0.0001 | <0.0001 | ||
| Pinteraction | <0.0001 |
All models were adjusted for age, education level, SES and CCI category.
Tables 4 and 5 show the association between serum uric acid levels and different site-specific cancers in men and women, respectively. In men, statistically significant positive associations were found for colorectal, hepatobillary, kidney, and non-melanoma skin cancers, and statistically significant inverse associations were found for pulmonary cancers. In women, statistically significant positive associations were found for head and neck cancers, and statistically significant inverse associations were found for breast, and lymphatic and haematological cancers in women. The hazard ratios (HRs) for positive associations were the largest for hepatobillary cancers in men (e.g. HR for log unit increase of serum uric acid: 2.03 (95%CI: 1.38 – 2.99)) (Table 4) and for head and neck cancers in women (e.g. HR for log unit increase of serum uric acid: 1.90 (95%CI: 1.24 – 2.92)) (Table 5). The inverse HRs were the largest for CNS cancers in both men and women (e.g. HR for log unit increase of serum uric acid: 0.60 (95%CI: 0.43 – 0.83) and 0.67 (95%CI: 0.50 – 0.89), respectively).
Table 4. Hazard ratios (HR) and 95% Confidence Intervals (95%CI) for the risk of different types of cancer by log and categories of serum uric acid in men.
| Site-specific cancer | Log Serum Uric Acid | Quartile 1 | Quartile 2 | Quartile 3 | Quartile 4 | Ptrend |
|---|---|---|---|---|---|---|
| Prostate | ||||||
| Events – n | 3673 | 3860 | 4058 | 3988 | ||
| HR (95%CI) | 0.92 (0.85 – 1.00) | 1.00 (Ref) | 0.97 (0.93 – 1.02) | 1.00 (0.96 – 1.05) | 1.02 (0.97 – 1.06) | 0.08 |
| Pulmonary | ||||||
| Events – n | 745 | 714 | 746 | 756 | ||
| HR (95%CI) | 0.79 (0.66 – 0.95) | 1.00 (Ref) | 0.89 (0.80 – 0.99) | 0.91 (0.82 – 1.00) | 0.93 (0.84 – 1.03) | 0.03 |
| Colorectal | ||||||
| Events – n | 897 | 1012 | 1162 | 1269 | ||
| HR (95%CI) | 1.55 (1.33 – 1.81) | 1.00 (Ref) | 1.05 (0.96 – 1.15) | 1.17 (1.07 – 1.28) | 1.31 (1.20 – 1.42) | <0.0001 |
| Gastroesophageal | ||||||
| Events – n | 290 | 319 | 336 | 341 | ||
| HR (95%CI) | 1.04 (0.79 – 1.38) | 1.00 (Ref) | 1.02 (0.87 – 1.20) | 1.05 (0.90 – 1.23) | 1.08 (0.92 – 1.27) | 0.83 |
| Hepatobillary | ||||||
| Events – n | 143 | 142 | 159 | 241 | ||
| HR (95%CI) | 2.03 (1.38 – 2.99) | 1.00 (Ref) | 0.92 (0.73 – 1.16) | 1.00 (0.80 – 1.26) | 1.56 (1.27 – 1.92) | 0.0001 |
| Pancreas | ||||||
| Events – n | 188 | 219 | 210 | 258 | ||
| HR (95%CI) | 1.38 (0.98 – 1.94) | 1.00 (Ref) | 1.08 (0.89 – 1.32) | 1.01 (0.83 – 1.23) | 1.26 (1.05 – 1.52) | 0.19 |
| Kidney | ||||||
| Events – n | 220 | 247 | 276 | 303 | ||
| HR (95%CI) | 1.52 (1.11 – 2.08) | 1.00 (Ref) | 1.05 (0.87 – 1.25) | 1.13 (0.95 – 1.35) | 1.27 (1.06 – 1.51) | 0.02 |
| Bladder | ||||||
| Events – n | 569 | 593 | 647 | 718 | ||
| HR (95%CI) | 1.30 (1.06 – 1.59) | 1.00 (Ref) | 0.97 (0.86 – 1.08) | 1.03 (0.92 – 1.15) | 1.16 (1.04 – 1.30) | 0.10 |
| Head and neck | ||||||
| Events – n | 223 | 200 | 214 | 280 | ||
| HR (95%CI) | 1.48 (1.06 – 2.08) | 1.00 (Ref) | 0.83 (0.69 – 1.01) | 0.87 (0.72 – 1.05) | 1.16 (0.97 – 1.38) | 0.08 |
| Melanoma | ||||||
| Events – n | 428 | 479 | 505 | 503 | ||
| HR (95%CI) | 1.17 (0.93 – 1.48) | 1.00 (Ref) | 1.04 (0.91 – 1.18) | 1.07 (0.94 – 1.21) | 1.09 (0.95 – 1.23) | 0.19 |
| Non-melanoma Skin | ||||||
| Events – n | 416 | 476 | 521 | 630 | ||
| HR (95%CI) | 1.59 (1.27 – 1.99) | 1.00 (Ref) | 1.06 (0.93 – 1.21) | 1.14 (1.00 – 1.29) | 1.42 (1.25 – 1.60) | 0.0007 |
| Central nervous system | ||||||
| Events – n | 260 | 242 | 257 | 213 | ||
| HR (95%CI) | 0.60 (0.43 – 0.83) | 1.00 (Ref) | 0.87 (0.73 – 1.03) | 0.89 (0.75 – 1.06) | 0.74 (0.62 – 0.89) | 0.005 |
| Lymphatic and Haematological | ||||||
| Events – n | 658 | 709 | 737 | 782 | ||
| HR (95%CI) | 1.03 (0.86 – 1.25) | 1.00 (Ref) | 1.00 (0.90 – 1.11) | 1.01 (0.91 – 1.12) | 1.09 (0.99 – 1.21) | 0.34 |
| Other Cancer | ||||||
| Events – n | 761 | 733 | 768 | 913 | ||
| HR (95%CI) | 1.23 (1.03 – 1.47) | 1.00 (Ref) | 0.92 (0.83 – 1.02) | 0.99 (0.90 – 1.10) | 1.28 (1.16 – 1.41) | 0.009 |
All models were adjusted for age, education level, SES and CCI category.
Table 5. Hazard ratios (HR) and 95% Confidence Intervals (95%CI) for the risk of different types of cancer by log and categories of serum uric acid in women.
| Site-specific cancer | Log Serum Uric Acid | Quartile 1 | Quartile 2 | Quartile 3 | Quartile 4 | Ptrend |
|---|---|---|---|---|---|---|
| Breast | ||||||
| Events – n | 2295 | 2570 | 2747 | 2782 | ||
| HR (95%CI) | 0.94 (0.86 – 1.02) | 1.00 (Ref) | 1.00 (0.94 – 1.05) | 0.99 (0.94 – 1.05) | 0.93 (0.88 – 0.98) | 0.03 |
| Pulmonary | ||||||
| Events – n | 444 | 483 | 519 | 698 | ||
| HR (95%CI) | 0.90 (0.74 – 1.08) | 1.00 (Ref) | 0.97 (0.85 - 1.10) | 0.97 (0.85 – 1.10) | 1.22 (1.08 – 1.37) | 0.35 |
| Colorectal | ||||||
| Events – n | 555 | 725 | 784 | 1077 | ||
| HR (95%CI) | 0.92 (0.79 – 1.08) | 1.00 (Ref) | 1.16 (1.04 – 1.30) | 1.17 (1.05 – 1.31) | 1.50 (1.35 – 1.66) | 0.16 |
| Gastroesophageal | ||||||
| Events – n | 91 | 122 | 138 | 215 | ||
| HR (95%CI) | 1.15 (0.80 – 1.65) | 1.00 (Ref) | 1.19 (0.91 – 1.56) | 1.25 (0.96 – 1.63) | 1.79 (1.40 – 2.29) | 0.40 |
| Hepatobillary | ||||||
| Events – n | 82 | 112 | 118 | 209 | ||
| HR (95%CI) | 1.55 (1.06 – 2.27) | 1.00 (Ref) | 1.23 (0.92 – 1.63) | 1.20 (0.91 – 1.60) | 1.96 (1.52 – 2.53) | 0.17 |
| Pancreas | ||||||
| Events – n | 119 | 136 | 171 | 283 | ||
| HR (95%CI) | 1.51 (1.09 – 2.10) | 1.00 (Ref) | 1.02 (0.79 – 1.30) | 1.19 (0.94 – 1.50) | 1.82 (1.47 – 2.26) | 0.08 |
| Kidney | ||||||
| Events – n | 88 | 104 | 121 | 193 | ||
| HR (95%CI) | 1.83 (1.24 – 2.71) | 1.00 (Ref) | 1.05 (0.79 – 1.40) | 1.14 (0.87 – 1.50) | 1.68 (1.31 – 2.16) | 0.08 |
| Bladder | ||||||
| Events – n | 118 | 149 | 176 | 265 | ||
| HR (95%CI) | 1.09 (0.79 – 1.52) | 1.00 (Ref) | 1.13 (0.88 – 1.43) | 1.24 (0.98 – 1.57) | 1.74 (1.40 – 2.16) | 0.42 |
| Gynaecological | ||||||
| Events – n | 833 | 890 | 1016 | 1337 | ||
| HR (95%CI) | 1.22 (1.06 – 1.40) | 1.00 (Ref) | 0.95 (0.86 – 1.04) | 1.01 (0.92 – 1.11) | 1.23 (1.12 – 1.34) | 0.12 |
| Head and neck | ||||||
| Events – n | 75 | 86 | 105 | 160 | ||
| HR (95%CI) | 1.90 (1.24 – 2.92) | 1.00 (Ref) | 1.02 (0.75 – 1.39) | 1.16 (0.87 – 1.57) | 1.65 (1.25 – 2.17) | 0.004 |
| Melanoma | ||||||
| Events – n | 314 | 320 | 371 | 337 | ||
| HR (95%CI) | 0.90 (0.71 – 1.15) | 1.00 (Ref) | 0.91 (0.78 – 1.06) | 0.98 (0.85 – 1.14) | 0.83 (0.71 – 0.96) | 0.63 |
| Non-melanoma Skin | ||||||
| Events – n | 216 | 281 | 324 | 456 | ||
| HR (95%CI) | 1.04 (0.81 – 1.33) | 1.00 (Ref) | 1.16 (0.97 – 1.39) | 1.25 (1.05 – 1.49) | 1.67 (1.42 – 1.96) | 0.20 |
| Central nervous system | ||||||
| Events – n | 228 | 238 | 224 | 213 | ||
| HR (95%CI) | 0.67 (0.50 – 0.89) | 1.00 (Ref) | 0.93 (0.77 – 1.11) | 0.81 (0.67 – 0.97) | 0.71 (0.59 – 0.85) | 0.003 |
| Lymphatic and Haematological | ||||||
| Events – n | 404 | 414 | 445 | 595 | ||
| HR (95%CI) | 0.80 (0.65 – 0.98) | 1.00 (Ref) | 0.91 (0.79 – 1.05) | 0.91 (0.80 – 1.05) | 1.13 (1.00 – 1.28) | 0.01 |
| Other Cancer | ||||||
| Events – n | 438 | 539 | 673 | 922 | ||
| HR (95%CI) | 1.78 (1.50 – 2.11) | 1.00 (Ref) | 1.20 (1.06 – 1.37) | 1.52 (1.35 – 1.72) | 2.32 (2.07 – 2.59) | <0.0001 |
All models were adjusted for age, education level, SES and CCI category.
After Bonferroni correction for multiple testing, we found that the associations in colorectal, hepatobillary, and non-melanoma skin cancers remained significant in men, and associations in CNS and other cancers remained significant in women.
DISCUSSION
We observed that elevated levels of serum uric acid were associated with an increased cancer incidence in comparison to normal values. Adjustment for glucose and TG did not alter the results, and trends were similar when adjusted for BMI. Cancer site specific and gender stratified analysis showed that positive associations between serum uric acid levels and risk of colorectal, hepatobillary, kidney, and non-melanoma skin cancers in men. For women this was observed for head and neck cancers.
Uric acid has been found to act as both an antioxidant and a prooxidant, both of which may affect carcinogenesis. Uric acid behaves as an antioxidant as it scavenges free radical and chelating transitional metal ions by preventing peroxynitrite-induced protein nitrosylation, lipid and protein peroxidation and inactivating tetrahydrobiopterin [20]. Experiments in which uric acid was administered to healthy volunteers found it to decrease ROS production [21]. Conversely, cells exposed to uric acid generate oxidative stress [22, 23]. Production of ROS is associated with local inflammation, impaired nitric oxide (NO) generation, activation of the renin-angiotensin-aldosterone system, insulin resistance, and fat accumulation [20], and promotes tumour cell proliferation, migration, and survival [7]. Tumourigenesis may occur because ROS increase cellular mutation rate, giving rise to oncogenic potential [5, 6].
We observed that elevated levels of serum uric acid were associated with increased incidence at some cancer sites and decreased incidence at other cancer sites, in comparison to normal values, supporting a paradoxical role for uric acid. Cancer site specific and gender stratified analysis showed positive associations between serum uric acid level and risk of colorectal, hepatobillary, kidney, and non-melanoma skin cancers in men, and head and neck cancers in women. Inverse associations between serum uric acid level and risk of pulmonary cancers in men, and breast, and lymphatic and haematological cancers in women were also shown.
In the current study, the positive associations between serum uric acid levels and cancer incidence support the role of uric acid as a prooxidant for the relevant cancer sites. Hyperuricemia is, therefore, a potential risk factor for colorectal, hepatobillary, kidney, non-melanoma skin, and other cancers in men, and for head and neck and other cancers in women. Conversely, hyperuricemia may be protective for pulmonary and CNS cancers in men, and for breast, lymphatic and haematological, and CNS cancers in women. The differing directions of association may imply variation in cellular responses to oxidative stress, or in the extent of involvement of oxidative stress and antioxidants in specific cancer development. Variation in cellular responses has previously been observed in other cellular stressors, for example, heterogeneous responses to anoxia in neurones [24].
Our results show some consistency with other similar studies. A meta-analysis of five prospective cohort studies [25–28] found that hyperuricemia was associated with an increased incidence of overall cancer (relative risk (RR): 1.03 (95%CI: 1.01 – 1.05, P=0.007)) [2]. Stratification by gender in the meta-analysis showed that hyperuricemia was significantly associated to risk of cancer in men, but not in women (RR: 1.05 (95%CI: 1.02 – 1.08) and 1.01 (95%CI: 0.98 – 1.04), for men and women, respectively), however only one study specifically studied women. Furthermore, the meta-analysis suggested substantial heterogeneity between the studies (I2: 44.7 for cancer risk in men and women, 53.8 for cancer risk in men) [2]. Stratification by cancer site found a significant positive association between hyperuricemia and lymphatic and haematological cancers only [2]. This is inconsistent with our finding that hyperuricemia may be protective for lymphatic and haematological cancers in women, possibly due to differences in sample populations and sample sizes. The heterogeneity within our results, and between the previously described studies, may also be caused by confounders, in addition to plausible biological processes.
Serum uric acid concentration reflects the balance between uric acid synthesis and excretion. Increased production of serum uric acid may follow consumption of diets high in purines, acute alcohol consumption, chronic fructose consumption, and severe exercise [20]. Impaired function of the kidney, which primarily excretes uric acid, may also cause hyperuricemia [29]. Lifestyle factors and other chronic diseases potentially have an effect on the association between serum uric acid and cancer, for example, MetS is associated with the aforementioned dietary factors that cause increased production of serum uric acid [11, 20]. Reverse causation between uric acid and MetS has also been indicated given the fact that uric acid causes mitochondrial oxidative stress, stimulating accumulation of fat that is independent of excessive caloric intake [11, 30]. As MetS has been suggested as a risk factor for some cancers [26, 31], this complex association implies that serum uric acid levels may also be a result or a proxy for underlying MetS and/or lifestyle risk factors for cancer. Nevertheless, we obtained similar results when we adjusted our analysis for MetS components, serum glucose and triglycerides.
To our knowledge, this is the largest study investigating serum uric acid and cancer risk in both men and women. We were able to account for MetS components by adjustments for serum glucose and lipids. The current study was strengthened by the large number of individuals with prospective measurements of serum uric acid available in the AMORIS database, all measured at one and the same laboratory. Complete follow up information, detailed information on cancer diagnosis, time of death and emigration was available for each participant through the use of national registers. Exposure and outcome information was obtained independently and assessed in an accurate manner [32]. The AMORIS population was selected following analysis of blood samples from health check-ups in non-hospitalised persons and is representative of the general working population of the greater Stockholm area in terms of SES and ethnicity. Over the course of the study, all-cause mortality was about 14% lower in the AMORIS participants than in the general population of the greater Stockholm area, accounting for age, gender and calendar year [32]. The internal validity of our study is unaffected by the selection of a healthy cohort. There were no repeated measurements for serum uric acid available and no information on tumour severity. We did not have data on other lifestyle factors such as alcohol consumption, smoking status and diet for all study participants. Nevertheless, adjustment for CCI was considered as a proxy for other lifestyle-related disorders and did not alter the results. However, residual confounding may still have occurred which may have resulted in underestimation of the associations observed. Genetic information was unavailable for our study population, and it would be interesting to expand future analysis by combining genetic markers for serum uric acid. Finally, we observed higher risk of mortality with higher uric acid levels. Although we have adjusted our analysis for other co-morbidities, lower risks of some cancers with increasing serum uric acid need to be interpreted carefully due to the remaining possibility of competing risk, which is beyond the scope of this study. Future studies needs to address the above limitations to clarify any causal link between circulating uric acid and specific cancers.
In summary, increased levels of serum uric acid level were associated with risk of overall cancer in this large prospective cohort. The different directions of observed associations may corroborate the paradoxical role of uric acid as both an antioxidant and a prooxidant. Nevertheless, further investigations are needed to assess biological mechanisms that may clarify the link between uric acid metabolism and cancer.
MATERIALS AND METHODS
Study population and data collection
Blood samples of men and women from the greater Stockholm area were collected by the Central Automation Laboratory (CALAB) between 1985 and 1996. No other clinical data were included in the CALAB database for all participants [33]. The sample population was representative of the general working population of Stockholm. All participants were either healthy individuals undergoing clinical laboratory testing as part of a general health check-up or outpatients referred for laboratory testing. None of the participants were inpatients at the time their blood samples were taken. The linkage of Swedish national registries, using the Swedish 10-digit personal identity number, to the CALAB database formed the AMORIS study and has been described in detail elsewhere [33–39]. Linkage to registries such as the National Cancer Register and the Hospital Discharge Register provided information on socio-economic status (SES), education level, vital status, cancer diagnosis and emigration. After a recent update, the AMORIS study now includes follow up information until 31st December 2011 for 812,073 individuals [40].
We included 493,281 men and women aged 20 and older with baseline measurements of uric acid without any previous diagnosis of cancer. We excluded those with follow up less than 3 months to reduce reverse causation (i.e. serum uric acid levels can be affected by an undiagnosed cancer). An additional sensitivity analysis, in which the first 2 years of follow up were excluded, was also carried out to address the issue of reverse causation (n = 485,820). Follow up time was defined as the time from baseline measurement until date of cancer diagnosis, death from any cause, emigration or end of study (31st December 2011), whichever occurred first. This study complied with the Declaration of Helsinki and was approved by the Ethics Review Board of the Karolinska Institute.
Definition of outcomes
The outcomes of this study were overall incident cancer and site-specific cancers. Diagnosis of cancer was obtained from the Swedish National Cancer Register. International Classification of Diseases, 7th revision (ICD-7) codes were used to classify major cancer sites (Supplementary Table 1, Supplementary Information).
Assessment of exposure and covariates
Uric acid was measured by enzymatic uricase method. Coefficients of variation for uric acid determinations were <2.8% at 164 μmol/L (2.76 mg/dL), 2.3% at 470 μmol/L (7.90 mg/dL) and 1.8% at 624 μmol/L (10.49 mg/dL). Due to gender difference in serum uric acid levels [2, 20, 41, 42], we used sex-specific quartiles to classify levels of uric acid into four quartiles. Uric acid levels in men were grouped into quartile 1 (<281 μmol/L), quartile 2 (281-319 μmol/L), quartile 3 (319-362 μmol/L) and quartile 4 (≥362 μmol/L). Uric acid levels in women were grouped into quartile 1 (<207 μmol/L), quartile 2 (207-240 μmol/L), quartile 3 (240-279 μmol/L), and quartile 4 (≥279 μmol/L). We additionally assess uric acid as a continuous variable by converting it logarithmically to ensure the normal distribution of the variable. Glucose was measured enzymatically with a glucoseoxidase/peroxidase method and Triglycerides (TG) were measured enzymatically with a glycerol-phosphate-oxidase after hydrolysis with lipoprotein lipase. Glucose (mmol/L) and triglycerides (TG; mmol/L) were assayed in 489,613 and 493,281 persons, respectively. Weight and height measurements were assessed in 66,931 participants and body mass index (BMI) was calculated. All methods were fully automated with automatic calibration and performed at one accredited laboratory [35].
SES (white collar, blue collar, unemployed or unknown) was based on the national censuses [43]. Education level was categorised into low (less than high school), medium (high school equivalent), and high (higher education). Charlson comorbidity index (CCI) was calculated to account for other diseases given the association between serum uric acid and metabolic diseases. CCI was based on information from the National Patient Register and consisted of 17 groups of diseases with a specific weight assigned to each disease category, which were summed to obtain an overall score between 0 and 3+ corresponding to comorbidity level [43].
Statistical analysis
Cox proportional hazards regression was used to estimate hazard ratios (HRs) and their 95% confidence intervals (95% CI) of overall cancer incidence by sex-specific quartiles of uric acid as well as the log transformation of uric acid. The assumption of proportionality of hazard was satisfied upon plotting levels of uric acid on Kaplan-Meier curves. In the multivariate model, we adjusted our analysis for age, gender, education level, SES, and CCI category. To assess whether this association was robust against adverse metabolic states, we performed further analyses adjusting for metabolic markers, i.e. serum glucose and triglycerides, and also adjusting for BMI as a continuous variable in the subgroup with information on BMI available.
We further stratified our analyses by gender to assess the incidence of overall cancer and each major cancer site and assessed the remaining cancers as one group (other cancers). We performed a Bonferroni correction, by dividing the significance threshold, 0.05, by the number of cancer sites assessed, fourteen in men and fifteen in women, to account for multiple testing in our cancer site-specific analyses. We also performed a test for interaction by including the product of uric acid levels and gender in the models. To observe association between serum uric acid and mortality, a similar model assessing the association between uric acid levels and risk of overall death was carried out. All analyses were conducted with Statistical Analysis Systems (SAS) release 9.4 (SAS Institute, Cary, NC).
SUPPLEMENTARY MATERIALS FIGURES AND TABLES
Footnotes
Authors’ contributions
AY, MVH, HG and WW conceptualised and designed the study. HM, HG, ML, NH, GR and IJ provided materials used in the study. AY and WW performed the statistical analysis with critical feedback from other authors. AY wrote the first draft of the manuscript. All authors reviewed, edited and finalised the manuscript.
CONFLICTS OF INTEREST
Niklas Hammar is employed by AstraZeneca. However, the views expressed in this study are his own and not those of AstraZeneca’s.
FUNDING
The research was funded/supported by the National Institute for Health Research (NIHR) Biomedical Research Centre based at Guy's and St Thomas’ NHS Foundation Trust and King's College London. Open access for this article was funded by King's College London.
REFERENCES
- 1.Becker BF. Towards the physiological function of uric acid. Free radical biology & medicine. 1993;14:615–631. doi: 10.1016/0891-5849(93)90143-i. [DOI] [PubMed] [Google Scholar]
- 2.Yan S, Zhang P, Xu W, Liu Y, Wang B, Jiang T, Hua C, Wang X, Xu D, Sun B. Serum Uric Acid Increases Risk of Cancer Incidence and Mortality: A Systematic Review and Meta-Analysis. Mediators of inflammation. 2015;2015:764250. doi: 10.1155/2015/764250. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Lanaspa MA, Sanchez-Lozada LG, Choi YJ, Cicerchi C, Kanbay M, Roncal-Jimenez CA, Ishimoto T, Li N, Marek G, Duranay M, Schreiner G, Rodriguez-Iturbe B, Nakagawa T, et al. Uric acid induces hepatic steatosis by generation of mitochondrial oxidative stress: potential role in fructose-dependent and -independent fatty liver. The Journal of biological chemistry. 2012;287:40732–40744. doi: 10.1074/jbc.M112.399899. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Sautin YY, Nakagawa T, Zharikov S, Johnson RJ. Adverse effects of the classic antioxidant uric acid in adipocytes: NADPH oxidase-mediated oxidative/nitrosative stress. American journal of physiology. Cell physiology. 2007;293:C584–96. doi: 10.1152/ajpcell.00600.2006. [DOI] [PubMed] [Google Scholar]
- 5.Ames BN, Cathcart R, Schwiers E, Hochstein P. Uric acid provides an antioxidant defense in humans against oxidant- and radical-caused aging and cancer: a hypothesis. Proceedings of the National Academy of Sciences of the United States of America. 1981;78:6858–6862. doi: 10.1073/pnas.78.11.6858. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Fabbrini E, Serafini M, Colic Baric I, Hazen SL, Klein S. Effect of plasma uric acid on antioxidant capacity, oxidative stress, and insulin sensitivity in obese subjects. Diabetes. 2014;63:976–981. doi: 10.2337/db13-1396. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Fini MA, Elias A, Johnson RJ, Wright RM. Contribution of uric acid to cancer risk, recurrence, and mortality. Clin Transl Med. 2012;1:16. doi: 10.1186/2001-1326-1-16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Kuo CF, See LC, Yu KH, Chou IJ, Chiou MJ, Luo SF. Significance of serum uric acid levels on the risk of all-cause and cardiovascular mortality. Rheumatology (Oxford, England) 2013;52:127–134. doi: 10.1093/rheumatology/kes223. [DOI] [PubMed] [Google Scholar]
- 9.Strasak AM, Rapp K, Hilbe W, Oberaigner W, Ruttmann E, Concin H, Diem G, Pfeiffer KP, Ulmer H. VHM&PP Study Group. Serum uric acid and risk of cancer mortality in a large prospective male cohort. Cancer causes & control. 2007;18:1021–1029. doi: 10.1007/s10552-007-9043-3. [DOI] [PubMed] [Google Scholar]
- 10.Deng Z, Gu Y, Hou X, Zhang L, Bao Y, Hu C, Jia W. Association between uric acid, cancer incidence and mortality in patients with type 2 diabetes: Shanghai diabetes registry study. Diabetes/metabolism research and reviews. 2016;32:325–332. doi: 10.1002/dmrr.2724. [DOI] [PubMed] [Google Scholar]
- 11.Johnson RJ, Nakagawa T, Sanchez-Lozada LG, Shafiu M, Sundaram S, Le M, Ishimoto T, Sautin YY, Lanaspa MA. Sugar, uric acid, and the etiology of diabetes and obesity. Diabetes. 2013;62:3307–3315. doi: 10.2337/db12-1814. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Choi HK, Ford ES. Prevalence of the metabolic syndrome in individuals with hyperuricemia. The American Journal of Medicine. 2007;120:442–447. doi: 10.1016/j.amjmed.2006.06.040. [DOI] [PubMed] [Google Scholar]
- 13.Facchini F, Chen YD, Hollenbeck CB, Reaven GM. Relationship between resistance to insulin-mediated glucose uptake, urinary uric acid clearance, and plasma uric acid concentration. Jama. 1991;266:3008–3011. [PubMed] [Google Scholar]
- 14.Sluijs I, Holmes MV, van der Schouw YT, Beulens JW, Asselbergs FW, Huerta JM, Palmer TM, Arriola L, Balkau B, Barricarte A, Boeing H, Clavel-Chapelon F, Fagherazzi G, et al. A Mendelian Randomization Study of Circulating Uric Acid and Type 2 Diabetes. Diabetes. 2015;64:3028–3036. doi: 10.2337/db14-0742. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Villegas R, Xiang YB, Elasy T, Xu WH, Cai H, Cai Q, Linton MR, Fazio S, Zheng W, Shu XO. Purine-rich foods, protein intake, and the prevalence of hyperuricemia: The Shanghai Men's Health Study. Nutrition, Metabolism, and Cardiovascular Diseases. 2012;22:409–416. doi: 10.1016/j.numecd.2010.07.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.American Heart Association Nutrition Committee. Lichtenstein AH, Appel LJ, Brands M, Carnethon M, Daniels S, Franch HA, Franklin B, Kris-Etherton P, Harris WS, Howard B, Karanja N, Lefevre M, et al. Diet and lifestyle recommendations revision 2006: a scientific statement from the American Heart Association Nutrition Committee. Circulation. 2006;114:82–96. doi: 10.1161/CIRCULATIONAHA.106.176158. [DOI] [PubMed] [Google Scholar]
- 17.Johnson RJ, Kang DH, Feig D, Kivlighn S, Kanellis J, Watanabe S, Tuttle KR, Rodriguez-Iturbe B, Herrera-Acosta J, Mazzali M. Is there a pathogenetic role for uric acid in hypertension and cardiovascular and renal disease? Hypertension. 2003;41:1183–90. doi: 10.1161/01.HYP.0000069700.62727.C5. [DOI] [PubMed] [Google Scholar]
- 18.Nakanishi N, Okamoto M, Yoshida H, Matsuo Y, Suzuki K, Tatara K. Serum uric acid and risk for development of hypertension and impaired fasting glucose or Type II diabetes in Japanese male office workers. European journal of epidemiology. 2003;18:523–530. doi: 10.1023/a:1024600905574. [DOI] [PubMed] [Google Scholar]
- 19.Niskanen LK, Laaksonen DE, Nyyssonen K, Alfthan G, Lakka HM, Lakka TA, Salonen JT. Uric acid level as a risk factor for cardiovascular and all-cause mortality in middle-aged men: a prospective cohort study. Archives of Internal Medicine. 2004;164:1546–1551. doi: 10.1001/archinte.164.14.1546. [DOI] [PubMed] [Google Scholar]
- 20.Kang DH, Ha SK. Uric Acid Puzzle: Dual Role as Anti-oxidantand Pro-oxidant. Electrolytes & Blood Pressure. 2014;12:1–6. doi: 10.5049/EBP.2014.12.1.1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Waring WS, Webb DJ, Maxwell SR. Systemic uric acid administration increases serum antioxidant capacity in healthy volunteers. Journal of cardiovascular pharmacology. 2001;38:365–371. doi: 10.1097/00005344-200109000-00005. [DOI] [PubMed] [Google Scholar]
- 22.Kang DH, Park SK, Lee IK, Johnson RJ. Uric acid-induced C-reactive protein expression: implication on cell proliferation and nitric oxide production of human vascular cells. Journal of the American Society of Nephrology. 2005;16:3553–3562. doi: 10.1681/ASN.2005050572. [DOI] [PubMed] [Google Scholar]
- 23.Yu MA, Sanchez-Lozada LG, Johnson RJ, Kang DH. Oxidative stress with an activation of the renin-angiotensin system in human vascular endothelial cells as a novel mechanism of uric acid-induced endothelial dysfunction. Journal of hypertension. 2010;28:1234–1242. [PubMed] [Google Scholar]
- 24.Nathan PW, Sears TA. The susceptibility of nerve fibres to analgesics. Anaesthesia. 1963;18:467–476. doi: 10.1111/j.1365-2044.1963.tb13570.x. [DOI] [PubMed] [Google Scholar]
- 25.Horsfall LJ, Nazareth I, Petersen I. Serum uric acid and the risk of respiratory disease: a population-based cohort study. Thorax. 2014;69:1021–1026. doi: 10.1136/thoraxjnl-2014-205271. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Strasak AM, Lang S, Kneib T, Brant LJ, Klenk J, Hilbe W, Oberaigner W, Ruttmann E, Kaltenbach L, Concin H, Diem G, Pfeiffer KP, Ulmer H, et al. Use of Penalized Splines in Extended Cox-Type Additive Hazard Regression to Flexibly Estimate the Effect of Time-varying Serum Uric Acid on Risk of Cancer Incidence: A Prospective, Population-Based Study in 78,850 Men. Annals of Epidemiology. 2009;19:15–24. doi: 10.1016/j.annepidem.2008.08.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Kolonel LN, Yoshizawa C, Nomura AM, Stemmermann GN. Relationship of serum uric acid to cancer occurrence in a prospective male cohort. Cancer epidemiology, biomarkers & prevention. 1994;3:225–228. [PubMed] [Google Scholar]
- 28.Hiatt RA, Fireman BH. Serum uric acid unrelated to cancer incidence in humans. Cancer research. 1988;48:2916–2918. [PubMed] [Google Scholar]
- 29.Kim IY, Lee DW, Lee SB, Kwak IS. The role of uric acid in kidney fibrosis: experimental evidences for the causal relationship. BioMed research international. 2014;2014:638732. doi: 10.1155/2014/638732. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Wang DD, Sievenpiper JL, de Souza RJ, Chiavaroli L, Ha V, Cozma AI, Mirrahimi A, Yu ME, Carleton AJ, Di Buono M, Jenkins AL, Leiter LA, Wolever TM, et al. The effects of fructose intake on serum uric acid vary among controlled dietary trials. The Journal of nutrition. 2012;142:916–923. doi: 10.3945/jn.111.151951. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.You J, Liu WY, Zhu GQ, Wang OC, Ma RM, Huang GQ, Shi KQ, Guo GL, Braddock M, Zheng MH. Metabolic syndrome contributes to an increased recurrence risk of non-metastatic colorectal cancer. Oncotarget. 2015;6:19880–19890. doi: 10.18632/oncotarget.4166. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Van Hemelrijck M, Jassem W, Walldius G, Fentiman IS, Hammar N, Lambe M, Garmo H, Jungner I, Holmberg L. Gamma-glutamyltransferase and risk of cancer in a cohort of 545,460 persons - the Swedish AMORIS study. European journal of cancer (Oxford, England) 1990;2011;47:2033–2041. doi: 10.1016/j.ejca.2011.03.010. [DOI] [PubMed] [Google Scholar]
- 33.Walldius G, Jungner I, Holme I, Aastveit AH, Kolar W, Steiner E. High apolipoprotein B, low apolipoprotein A-I, and improvement in the prediction of fatal myocardial infarction (AMORIS study): a prospective study. Lancet (London, England) 2001;358:2026–2033. doi: 10.1016/S0140-6736(01)07098-2. [DOI] [PubMed] [Google Scholar]
- 34.Walldius G, Jungner I, Kolar W, Holme I, Steiner E. High cholesterol and triglyceride values in Swedish males and females: increased risk of fatal myocardial infarction. First report from the AMORIS (Apolipoprotein related MOrtality RISk) study. Blood Press Suppl. 1992;4:35–42. [PubMed] [Google Scholar]
- 35.Jungner I, Marcovina SM, Walldius G, Holme I, Kolar W, Steiner E. Apolipoprotein B and A-I values in 147576 Swedish males and females, standardized according to the World Health Organization-International Federation of Clinical Chemistry First International Reference Materials. Clinical chemistry. 1998;44:1641–1649. [PubMed] [Google Scholar]
- 36.Holme I, Aastveit AH, Jungner I, Walldius G. Relationships between lipoprotein components and risk of myocardial infarction: age, gender and short versus longer follow-up periods in the Apolipoprotein MOrtality RISk study (AMORIS) Journal of internal medicine. 2008;264:30–38. doi: 10.1111/j.1365-2796.2008.01925.x. [DOI] [PubMed] [Google Scholar]
- 37.Holme I, Aastveit AH, Hammar N, Jungner I, Walldius G. Relationships between lipoprotein components and risk of ischaemic and haemorrhagic stroke in the Apolipoprotein MOrtality RISk study (AMORIS) Journal of internal medicine. 2009;265:275–287. doi: 10.1111/j.1365-2796.2008.02016.x. [DOI] [PubMed] [Google Scholar]
- 38.Van Hemelrijck M, Garmo H, Binda E, Hayday A, Karagiannis SN, Hammar N, Walldius G, Lambe M, Jungner I, Holmberg L. Immunoglobulin E and cancer: a meta-analysis and a large Swedish cohort study. Cancer causes & control. 2010;21:1657–1667. doi: 10.1007/s10552-010-9594-6. [DOI] [PubMed] [Google Scholar]
- 39.Van Hemelrijck M, Garmo H, Holmberg L, Walldius G, Jungner I, Hammar N, Lambe M. Prostate cancer risk in the Swedish AMORIS study: the interplay among triglycerides, total cholesterol, and glucose. Cancer. 2011;117:2086–2095. doi: 10.1002/cncr.25758. [DOI] [PubMed] [Google Scholar]
- 40.Wulaningsih W, Holmberg L, Garmo H, Malmstrom H, Lambe M, Hammar N, Walldius G, Jungner I, Ng T, Van Hemelrijck M. Serum lactate dehydrogenase and survival following cancer diagnosis. British journal of cancer. 2015;113:1389–1396. doi: 10.1038/bjc.2015.361. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Holme I, Aastveit AH, Hammar N, Jungner I, Walldius G. Uric acid and risk of myocardial infarction, stroke and congestive heart failure in 417,734 men and women in the Apolipoprotein MOrtality RISk study (AMORIS) Journal of internal medicine. 2009;266:558–570. doi: 10.1111/j.1365-2796.2009.02133.x. [DOI] [PubMed] [Google Scholar]
- 42.Wu CY, Hu HY, Chou YJ, Huang N, Chou YC, Lee MS, Li CP. High Serum Uric Acid Levels Are Associated with All-Cause and Cardiovascular, but Not Cancer, Mortality in Elderly Adults. Journal of the American Geriatrics Society. 2015;63:1829–1836. doi: 10.1111/jgs.13607. [DOI] [PubMed] [Google Scholar]
- 43.Wulaningsih W, Michaelsson K, Garmo H, Hammar N, Jungner I, Walldius G, Holmberg L, Van Hemelrijck M. Inorganic phosphate and the risk of cancer in the Swedish AMORIS study. BMC cancer. 2013;13:257. doi: 10.1186/1471-2407-13-257. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.