Gout is an inflammatory arthritis caused by the crystallization of uric acid within the joints.(1) Historically, gout was predominantly observed among the wealthy upper class who were able to afford the comforts—and excesses—of life.(2) As this lifestyle has become widely prevalent in modern Western societies, gout’s changing epidemiology, together with the obesity epidemic, closely correlated with the evolving lifestyle trends of a Western diet and sedentary behaviour.(2) Thus, once considered a “disease of kings,” gout has become a “disease of commoners” and is now the most common form of inflammatory arthritis worldwide with the prevalence of gout rising in many countries.(3) For example, coinciding with the introduction of high-fructose corn syrup in the 1970s as well as increased consumption of sugar-sweetened beverages in the United States, both the incidence and prevalence of gout have more than doubled(4, 5), paralleling the rising obesity epidemic; hyperuricemia as well as gout prevalence have continued to increase into the late 2000s.(6) Similarly, gout was previously rare in rural African communities where traditional agricultural diets were consumed, but the prevalence in these regions is also now increasing, most notably in urban communities.(2, 7) Correspondingly, increased serum urate (SU; the causal precursor of gout) has been reported among Japanese immigrants who moved to the United States, which again coincides with the increased availability of a diet rich in meats and saturated fat; meanwhile, an elevated SU was not observed among those who continued to live in Japan.(2, 8) Finally, contemporary epidemiology data from Canada, Europe, New Zealand, and China all suggest gout incidence and prevalence are increasing,(9-12) also correlating with their obesity trends.(13)
These findings are at odds with Major et al.’s conclusion that the proportion of population SU variance explained by dietary components is markedly smaller than that explained by genetic factors.(14, 15) How can dietary changes over time (together with a Western lifestyle) cause a disease epidemic, yet also paradoxically such risk factors appear to be much less important as reflected by the population variance, even though they remain important in terms of causation? A key answer to this paradox stems from the fallacy of employing the “proportion of population variance explained” as a measure of effect.(16) The proportion of population variance explained is not a valid effect measure and tells us nothing about which risk factor is more important, as a series of recent methodology papers and textbooks have cautioned.(16-19) In a simple example of a study population where the vast majority of individuals are smokers (e.g., prevalence of 90%), the proportion of population variation in lung cancer explained by smoking is very low, even approaching zero (due to low exposure distribution variability, with 90% smoking); however, the population attributable fraction expectedly remains high (~90%).(17) Similar to this paradoxical result, the variability of dietary exposures in the United States (the most extreme “Western lifestyle” country) is small, with most people having overwhelmingly unhealthy and gout-prone diets, leading to a very small proportion of the population variance being explained, as seen in the study by Major et al.(14) As Pearce noted, “sometimes the most important causes of disease are invisible because they are everywhere.”(17) This would differ from the hypothetical variance explained among the aforementioned pre-Westernization communities like those in rural Africa. This paradox does not apply to regression coefficients (for continuous outcomes) or relative risks (for dichotomous outcomes), which would be valid effect measures for causal inference in both populations (i.e., the expected change in SU level or the relative risk of gout). Thus, the “proportion of variation explained” is not a valid or generalizable measure of effect or relative importance. Furthermore, it is overly simplistic to partition the population variation into the percentage explained by genetics (nature) versus environment (nurture) because almost all diseases involve a combination of genetic and environmental factors.(17) Simply put, nearly every case of every disease has some environmental and some genetic causes whose contributions sum to more than 100%; it is therefore inappropriate to allocate a portion of causation to either genes or environment when both act together.(19)
Beyond these key conceptual flaws, there are additional methodologic issues in Major et al.’s study that would have further contributed to the low proportion of population variance explained by dietary factors. First, despite dietary factors (together with physical activity) playing a critical role in energy balance, weight gain, and subsequent gout risk, the study did not consider the proportion of SU variability explained by obesity, thus substantially underestimating the net impact of dietary factors. Furthermore, compared with the high accuracy of genetic data, measurement error when assessing dietary exposures is substantially higher, which would bias associations toward the null. Here, in pooling the dietary intake data collected from each of the five cohorts, the authors aggregated the dietary data (despite differences between the questionnaires administered), and discarded some incompatible exposure data, which would have worsened existing measurement error. Finally, comparing the heritability derived from the entire set of GWAS genes (e.g., >300,000 genetic variants) with the proportion of SU variance explained by only 63 food items further added to the unfair comparison.
Given these critical issues, what are we left with in terms of gout causation and prevention? Despite measurement error, the reported effect estimates (as opposed to population variance estimates) associated with dietary factors should be valid in terms of the direction and ordinal ranking of the SU effect estimates. To that end, Major et al. confirmed many previously identified dietary factors and further suggested new food items that require replication. This provides valuable information for potential intervention studies, as has been seen in type 2 diabetes and cardiovascular disease (comorbidities of hyperuricemia and gout), where healthy diets have been shown to mitigate the effects of genetic susceptibility to these conditions. For now, the totality of evidence continues to suggest that genes are not our destiny, and modifiable lifestyle factors have an important place in the primary prevention of hyperuricemia and gout, and that the conclusion of this paper regarding the relative importance of genes versus diet is invalid.
Sharan K. Rai
Doctoral Student in Nutrition
Harvard T.H. Chan School of Public Health
Yuqing Zhang
Professor of Medicine in Residence
Harvard Medical School and Massachusetts General Hospital
Frank B. Hu
Professor of Nutrition and Epidemiology
Harvard T.H. Chan School of Public Health
Neil Pearce
Professor of Epidemiology and Biostatistics
London School of Hygiene and Tropical Medicine
Hyon K. Choi (Corresponding author)
Professor of Medicine
Harvard Medical School and Massachusetts General Hospital
References:
1. Choi HK, Mount DB, Reginato AM. Pathogenesis of gout. Annals of Internal Medicine. 2005;143(7):499-516.
2. Johnson RJ, Rideout BA. Uric acid and diet--insights into the epidemic of cardiovascular disease. N Engl J Med. 2004;350(11):1071-3.
3. Kuo CF, Grainge MJ, Zhang W, Doherty M. Global epidemiology of gout: prevalence, incidence and risk factors. Nat Rev Rheumatol. 2015;11(11):649-62.
4. Lawrence RC, Felson DT, Helmick CG, Arnold LM, Choi H, Deyo RA, et al. Estimates of the prevalence of arthritis and other rheumatic conditions in the United States. Part II. Arthritis Rheum. 2008;58(1):26-35.
5. Arromdee E, Michet CJ, Crowson CS, O'Fallon WM, Gabriel SE. Epidemiology of gout: is the incidence rising? J Rheumatol. 2002;29(11):2403-6.
6. Zhu Y, Pandya BJ, Choi HK. Prevalence of gout and hyperuricemia in the US general population: the National Health and Nutrition Examination Survey 2007-2008. Arthritis Rheum. 2011;63(10):3136-41.
7. Cassim B, Mody GM, Deenadayalu VK, Hammond MG. Gout in black South Africans: a clinical and genetic study. Ann Rheum Dis. 1994;53(11):759-62.
8. Kagan A, Harris BR, Winkelstein W, Jr., Johnson KG, Kato H, Syme SL, et al. Epidemiologic studies of coronary heart disease and stroke in Japanese men living in Japan, Hawaii and California: Demographic, physical, dietary and biochemical characteristics. Journal of Chronic Diseases. 1974;27(7):345-64.
9. Kuo CF, Grainge MJ, Mallen C, Zhang W, Doherty M. Rising burden of gout in the UK but continuing suboptimal management: a nationwide population study. Ann Rheum Dis. 2015;74(4):661-7.
10. Rai SK, Avina-Zubieta JA, McCormick N, De Vera MA, Shojania K, Sayre EC, et al. The rising prevalence and incidence of gout in British Columbia, Canada: Population-based trends from 2000 to 2012. Semin Arthritis Rheum. 2017;46(4):451-6.
11. Miao Z, Li C, Chen Y, Zhao S, Wang Y, Wang Z, et al. Dietary and lifestyle changes associated with high prevalence of hyperuricemia and gout in the Shandong coastal cities of Eastern China. J Rheumatol. 2008;35(9):1859-64.
12. Klemp P, Stansfield S, Castle B, Robertson M. Gout is on the increase in New Zealand. Annals of the Rheumatic Diseases. 1997;56(1):22-6.
13. NCD Risk Factor Collaboration. Worldwide trends in body-mass index, underweight, overweight, and obesity from 1975 to 2016: a pooled analysis of 2416 population-based measurement studies in 128.9 million children, adolescents, and adults. Lancet. 2017;390(10113):2627-42.
14. Major TJ, Topless RK, Dalbeth N, Merriman TR. Evaluation of the diet wide contribution to serum urate levels: meta-analysis of population based cohorts. BMJ. 2018;363.
15. Watson L, Roddy E. The role of diet in serum urate concentration. Bmj. 2018;363:k4140.
16. Greenland S, Schlesselman JJ, Criqui MH. The fallacy of employing standardized regression coefficients and correlations as measures of effect. Am J Epidemiol. 1986;123(2):203-8.
17. Pearce N. Epidemiology in a changing world: variation, causation and ubiquitous risk factors. Int J Epidemiol. 2011;40(2):503-12.
18. Rose G. Sick individuals and sick populations. Int J Epidemiol. 2001;30(3):427-32; discussion 33-4.
19. Rothman KJ, Greenland S, Lash TL (eds). Modern Epidemiology. 3rd edn. Philadelphia: Lippincott Williams & Wilkins, 2008.
Competing interests:
No competing interests
25 October 2018
Hyon K Choi
Professor of Medicine
Sharan K. Rai, Yuqing Zhang, Frank B. Hu, Neil Pearce, Hyon K. Choi
Harvard Medical School and Massachusetts General Hospital
Rapid Response:
The Paradox of Ubiquitous Risk Factors for Gout
Gout is an inflammatory arthritis caused by the crystallization of uric acid within the joints.(1) Historically, gout was predominantly observed among the wealthy upper class who were able to afford the comforts—and excesses—of life.(2) As this lifestyle has become widely prevalent in modern Western societies, gout’s changing epidemiology, together with the obesity epidemic, closely correlated with the evolving lifestyle trends of a Western diet and sedentary behaviour.(2) Thus, once considered a “disease of kings,” gout has become a “disease of commoners” and is now the most common form of inflammatory arthritis worldwide with the prevalence of gout rising in many countries.(3) For example, coinciding with the introduction of high-fructose corn syrup in the 1970s as well as increased consumption of sugar-sweetened beverages in the United States, both the incidence and prevalence of gout have more than doubled(4, 5), paralleling the rising obesity epidemic; hyperuricemia as well as gout prevalence have continued to increase into the late 2000s.(6) Similarly, gout was previously rare in rural African communities where traditional agricultural diets were consumed, but the prevalence in these regions is also now increasing, most notably in urban communities.(2, 7) Correspondingly, increased serum urate (SU; the causal precursor of gout) has been reported among Japanese immigrants who moved to the United States, which again coincides with the increased availability of a diet rich in meats and saturated fat; meanwhile, an elevated SU was not observed among those who continued to live in Japan.(2, 8) Finally, contemporary epidemiology data from Canada, Europe, New Zealand, and China all suggest gout incidence and prevalence are increasing,(9-12) also correlating with their obesity trends.(13)
These findings are at odds with Major et al.’s conclusion that the proportion of population SU variance explained by dietary components is markedly smaller than that explained by genetic factors.(14, 15) How can dietary changes over time (together with a Western lifestyle) cause a disease epidemic, yet also paradoxically such risk factors appear to be much less important as reflected by the population variance, even though they remain important in terms of causation? A key answer to this paradox stems from the fallacy of employing the “proportion of population variance explained” as a measure of effect.(16) The proportion of population variance explained is not a valid effect measure and tells us nothing about which risk factor is more important, as a series of recent methodology papers and textbooks have cautioned.(16-19) In a simple example of a study population where the vast majority of individuals are smokers (e.g., prevalence of 90%), the proportion of population variation in lung cancer explained by smoking is very low, even approaching zero (due to low exposure distribution variability, with 90% smoking); however, the population attributable fraction expectedly remains high (~90%).(17) Similar to this paradoxical result, the variability of dietary exposures in the United States (the most extreme “Western lifestyle” country) is small, with most people having overwhelmingly unhealthy and gout-prone diets, leading to a very small proportion of the population variance being explained, as seen in the study by Major et al.(14) As Pearce noted, “sometimes the most important causes of disease are invisible because they are everywhere.”(17) This would differ from the hypothetical variance explained among the aforementioned pre-Westernization communities like those in rural Africa. This paradox does not apply to regression coefficients (for continuous outcomes) or relative risks (for dichotomous outcomes), which would be valid effect measures for causal inference in both populations (i.e., the expected change in SU level or the relative risk of gout). Thus, the “proportion of variation explained” is not a valid or generalizable measure of effect or relative importance. Furthermore, it is overly simplistic to partition the population variation into the percentage explained by genetics (nature) versus environment (nurture) because almost all diseases involve a combination of genetic and environmental factors.(17) Simply put, nearly every case of every disease has some environmental and some genetic causes whose contributions sum to more than 100%; it is therefore inappropriate to allocate a portion of causation to either genes or environment when both act together.(19)
Beyond these key conceptual flaws, there are additional methodologic issues in Major et al.’s study that would have further contributed to the low proportion of population variance explained by dietary factors. First, despite dietary factors (together with physical activity) playing a critical role in energy balance, weight gain, and subsequent gout risk, the study did not consider the proportion of SU variability explained by obesity, thus substantially underestimating the net impact of dietary factors. Furthermore, compared with the high accuracy of genetic data, measurement error when assessing dietary exposures is substantially higher, which would bias associations toward the null. Here, in pooling the dietary intake data collected from each of the five cohorts, the authors aggregated the dietary data (despite differences between the questionnaires administered), and discarded some incompatible exposure data, which would have worsened existing measurement error. Finally, comparing the heritability derived from the entire set of GWAS genes (e.g., >300,000 genetic variants) with the proportion of SU variance explained by only 63 food items further added to the unfair comparison.
Given these critical issues, what are we left with in terms of gout causation and prevention? Despite measurement error, the reported effect estimates (as opposed to population variance estimates) associated with dietary factors should be valid in terms of the direction and ordinal ranking of the SU effect estimates. To that end, Major et al. confirmed many previously identified dietary factors and further suggested new food items that require replication. This provides valuable information for potential intervention studies, as has been seen in type 2 diabetes and cardiovascular disease (comorbidities of hyperuricemia and gout), where healthy diets have been shown to mitigate the effects of genetic susceptibility to these conditions. For now, the totality of evidence continues to suggest that genes are not our destiny, and modifiable lifestyle factors have an important place in the primary prevention of hyperuricemia and gout, and that the conclusion of this paper regarding the relative importance of genes versus diet is invalid.
Sharan K. Rai
Doctoral Student in Nutrition
Harvard T.H. Chan School of Public Health
Yuqing Zhang
Professor of Medicine in Residence
Harvard Medical School and Massachusetts General Hospital
Frank B. Hu
Professor of Nutrition and Epidemiology
Harvard T.H. Chan School of Public Health
Neil Pearce
Professor of Epidemiology and Biostatistics
London School of Hygiene and Tropical Medicine
Hyon K. Choi (Corresponding author)
Professor of Medicine
Harvard Medical School and Massachusetts General Hospital
References:
1. Choi HK, Mount DB, Reginato AM. Pathogenesis of gout. Annals of Internal Medicine. 2005;143(7):499-516.
2. Johnson RJ, Rideout BA. Uric acid and diet--insights into the epidemic of cardiovascular disease. N Engl J Med. 2004;350(11):1071-3.
3. Kuo CF, Grainge MJ, Zhang W, Doherty M. Global epidemiology of gout: prevalence, incidence and risk factors. Nat Rev Rheumatol. 2015;11(11):649-62.
4. Lawrence RC, Felson DT, Helmick CG, Arnold LM, Choi H, Deyo RA, et al. Estimates of the prevalence of arthritis and other rheumatic conditions in the United States. Part II. Arthritis Rheum. 2008;58(1):26-35.
5. Arromdee E, Michet CJ, Crowson CS, O'Fallon WM, Gabriel SE. Epidemiology of gout: is the incidence rising? J Rheumatol. 2002;29(11):2403-6.
6. Zhu Y, Pandya BJ, Choi HK. Prevalence of gout and hyperuricemia in the US general population: the National Health and Nutrition Examination Survey 2007-2008. Arthritis Rheum. 2011;63(10):3136-41.
7. Cassim B, Mody GM, Deenadayalu VK, Hammond MG. Gout in black South Africans: a clinical and genetic study. Ann Rheum Dis. 1994;53(11):759-62.
8. Kagan A, Harris BR, Winkelstein W, Jr., Johnson KG, Kato H, Syme SL, et al. Epidemiologic studies of coronary heart disease and stroke in Japanese men living in Japan, Hawaii and California: Demographic, physical, dietary and biochemical characteristics. Journal of Chronic Diseases. 1974;27(7):345-64.
9. Kuo CF, Grainge MJ, Mallen C, Zhang W, Doherty M. Rising burden of gout in the UK but continuing suboptimal management: a nationwide population study. Ann Rheum Dis. 2015;74(4):661-7.
10. Rai SK, Avina-Zubieta JA, McCormick N, De Vera MA, Shojania K, Sayre EC, et al. The rising prevalence and incidence of gout in British Columbia, Canada: Population-based trends from 2000 to 2012. Semin Arthritis Rheum. 2017;46(4):451-6.
11. Miao Z, Li C, Chen Y, Zhao S, Wang Y, Wang Z, et al. Dietary and lifestyle changes associated with high prevalence of hyperuricemia and gout in the Shandong coastal cities of Eastern China. J Rheumatol. 2008;35(9):1859-64.
12. Klemp P, Stansfield S, Castle B, Robertson M. Gout is on the increase in New Zealand. Annals of the Rheumatic Diseases. 1997;56(1):22-6.
13. NCD Risk Factor Collaboration. Worldwide trends in body-mass index, underweight, overweight, and obesity from 1975 to 2016: a pooled analysis of 2416 population-based measurement studies in 128.9 million children, adolescents, and adults. Lancet. 2017;390(10113):2627-42.
14. Major TJ, Topless RK, Dalbeth N, Merriman TR. Evaluation of the diet wide contribution to serum urate levels: meta-analysis of population based cohorts. BMJ. 2018;363.
15. Watson L, Roddy E. The role of diet in serum urate concentration. Bmj. 2018;363:k4140.
16. Greenland S, Schlesselman JJ, Criqui MH. The fallacy of employing standardized regression coefficients and correlations as measures of effect. Am J Epidemiol. 1986;123(2):203-8.
17. Pearce N. Epidemiology in a changing world: variation, causation and ubiquitous risk factors. Int J Epidemiol. 2011;40(2):503-12.
18. Rose G. Sick individuals and sick populations. Int J Epidemiol. 2001;30(3):427-32; discussion 33-4.
19. Rothman KJ, Greenland S, Lash TL (eds). Modern Epidemiology. 3rd edn. Philadelphia: Lippincott Williams & Wilkins, 2008.
Competing interests: No competing interests