Heredity of Gout and Hyperuricemia BARUCH S. BLUMBERG, M.D., PH.D. University of Pennsylcania School of Medicine Investigations of the genetics of gout and hyperuricemia have usually started on the assumption that the disease and trait are inherited and that the purpose of the investigation is to determine the mode of inheritance. This undoubtedly arises from the fact that the heritability of gout is one of the most ancient notions of medicine. Galen stated that gout was due to an “hereditary trait” as well as “debauchery and intemperance” and his opinions have prevailed through the centuries. This assumption apparently had arisen because the disease often occurs in families. The clustering of a disease in families does not of itself mean that the disease is inherited, although it is consistent with this explanation. For example, many infectious diseases with little or no inherited component have a striking family clustering. Other environmental causes of a disease can also lead to a familial cluster. If gout and hyperuricemia were due to a dietary factor, it would be expected that families would have more similar dietary habits than randomly selected individuals and therefore a clustering of the disease and trait in families could occur. To reiterate, although familial clustering of a disease is consistent with a genetic etiology it does not exclude the possibly of nongenetic causes for the disease, or the possibility of both genetic and nongenetic factors causing the disease. The collection of information on the familial distribution of a disease permits statistical analysis of the data to determine if the trait segregates in a predictable manner consistent with Mendelian inheritance. There have been surprisingly few studies in which this technic has been applied to large numbers of families ascertained in an epidemiologically sound manner. The ideal way to study the inheritance of gout and hyperuricemia is to select a defined population and determine the individuals in the population with the disease or trait (the latter would require the sampling of blood from the population). If the whole population is too large to study with the facilities available, an appropriate random sample of the population could be tested. The individuals with the trait or disease are designated propositii. An equal or larger number of controls, who do not differ from those with the disease except for the lack of the disease or trait, are selected, and the families of both of the groups studied for the presence of the disease, trait, and other factors. The frequency of the disease and trait in the families of the affected and control group could then be compared to determine if there were in fact a higher frequency in the patient group. To my knowledge, extensive studies of this kind have not been published. Such studies, I believe, are in progress and perhaps those completed by Drs. O’Brien and Burch on North American Indians will include this information. Most genetic studies thus far reported fall short of this ideal, and have given inconsistent results. The investigations of Stecker and his colleagues1 are sub627 ARTHRITIS AND RHEUMATISM,VOL. 8, NO. 5-pART 1 (OCTOBER), 1965 628 GOUT AND PURINE METABOLISM ject to a variety of criticisms, primarily in their failure to use satisfactory methods of ascertainment. Even allowing for the methods which were used they found it necessary to invoke the explanation of diminished penetrance to arrive at a genetic explanation; that is, the expected results were not exactly consistent with Mendelian inheritance and the discrepancy was explained by stating that in a percentage of the cases the genes were present but were not expressed and therefore could not be detected. It should be clear that the results are also compatible with a nongenetic explanation. The data of Smyth et aL2 were consistent with the explanation that hyperuricemia was inherited as a simple autosomal trait. However, hyperuricemia was generally not evident before the age of sixteen and females were less likely to develop hyperuricemia than males. These studies were based on families of only 19 individuals with a total of 87 relatives, a relatively small number on which to establish firmly the simple genetic hypothesis for the entire disease entity. (There has been a follow-up on these original investigations which has been reported in a recent abstract3 and is to be published in full shortly.) Furthermore, in these studies, it was not possible to show that in all cases the heterozygote was an asymptomatic individual with hyperuricemia. Hauge and Harvald; using a sib-pair method of analysis did not support the simple autosomal dominant mode of inheritance. Their results, in fact, suggested that the inheritance was polygenic; that is, it was due to a series of genes whose effect in total added up to the observed levels of uric acid measured in the subjects. These differences in results have never, to my knowledge, been resolved. It is rather surprising that more substantial family studies have not been undertaken but the rather discouraging comments of Hauge and Harvald may in part explain this. They state, “Continued studies on the serum urate levels in families with hyperuricemia can hardly contribute more towards elucidating the problem of the heredity of this condition. More would probably be gained by studying the individual steps of uric acid synthesis as well as its excretion in the individuals of affected families.” That is, they wisely suggested that investigative energy should rather be devoted to finding out more about the metabolic steps that are involved in the development of hyperuricemia before extensive genetic studies of the sort which they did are undertaken. This statement emphasizes one of the inherent problems in the study of the genetics of gout and hyperuricemia. In a genetic analysis, it is assumed that the trait under study is homogenous; that is, gout or hyperuricemia in one individual or family is the same as gout or hyperuricemia in another individual or family. This type of analysis obviously will fall down if one is dealing with a group of diseases or traits, each of which may have a different etiology. For example, let us say that there are three causes of hyperuricemia: 1) excess production; 2 ) decreased urinary excretion; and 3 ) decreased destruction. If hyperuricemia 1) is due to a simple inherited Mendelian trait, hyperuricemia 2 ) to some nongenetic trait, and hyperuricemia 3 ) to a simple recessive trait; and if all of the traits were lumped together, the genetic analysis and results of the study would be invalid. What has been done is that several different HEREDITY OF GOUT AND HYPERURICEMIA 629 diseases have been studied under the same name. It is now clear that there are several different causes of hyperuricemia. Inhomogeneity of the material studied is a basic problem in all of the genetic studies reported. In addition to these complications, individuals may, usually for environmental reasons, be hyperuricemic at one point in time and have normal uric acid levels at another. Some of the factors that lead to these elevations are known but others are not. In population studies in which simples of blood are taken at only one or perhaps two points in time, these inconsistent levels could lead to error. Dr. Seegmiller and his colleagues, in a recent article,5 have emphasized this point. They said, “It is difficult to conceive of a single inherited biochemical defect that is expressed in some persons by an excessive production of uric acid and in others by diminished renal handling of uric acid. We are therefore left with the hypothesis that there are a number of genetic factors causing hyperuricemia in gout.” There may be, in addition to these, nongenetic factors that further complicate the picture. It is possible and perhaps probable that a single gene controlling a single biochemical defect might be segregating in one family or in a closely related kindred or in fact in a particular population, and SeegmilleF has cited a possible example of this. This may explain the rather impressive pedigrees that have occasionally been published ( see, for example, Emmerson7). It should be clear, however, that one has to be very careful in the interpretation of such single pedigree data. Gout and hyperuricemia are relatively common, particularly in some populations, and by chance may occur in more than one individual in the same family. If, for example, in a population hyperuricemia occurs in two of one hundred people, the chance that it will occur in two people in the same family would be four in ten thousand. When a single pedigree is reported, what is in effect done is that the majority of the families are discarded and the single family in which concurrence of hyperuricemia occurs is noted and reported. Hence the finding of one or two families with a high frequency of a trait which is relatively common in a population does not in itself prove inheritance or even familial clustering. It is imperative, as pointed out earlier, that appropriate comparisons with control groups be made. To my knowledge, there have been no extensive studies of hyperuricemia and gout in twins. This is surprising because such studies could contribute to our understanding of heritability and, furthermore, might give some intimation as to the different varieties of the disease and trait. For example, if known identical twins were ascertained in which one member of the twin pair had gout or hyperuricemia and the other did not, they would have one form of the trait or disease which was probably not inherited. That is, one could identify the noninherited kinds of gout or hyperuricemia by the study of nonconcordant identical twins. Identification of the inherited forms of hyperuricemia or gout using this technic would be difficult but, generally speaking, one could start with the assumption that the kind of gout that is more common in concordant identical twins than in discordant identical twins or nonidentical twins is more likely to be inherited. If such an extensive twin study were undertaken, the same material could be used for dealing with hypemricemia as a quantitative trait and determining what proportion of the uric acid levels is due to inherit- 630 GOUT AND PURINE METABOLISM SERUM URIC ACID Ilb FILIPINO MALES LOWER COMPONENT MEAN 5.802 S E 0.15 SD 0.89tSE 0.11 PROPORTION 0.782 SE 0.08 UPPER COMPONENT MEAN 8 0 7 t S E O 2 8 SD 0.89tSE 0.1I 3 4 5 6 7 9 10 mgj100ml. Fig. 5.-Frequency distribution of serum uric acid in Filipino males, with fitted composite curve dissected into two normal components. (From Decker, Lane, and Reynolds.10) ance. This again presumes homogeneity of the disease which, as we have seen, is probably an incorrect assumption to make, except in particular families or populations. During the past few years some interesting studies on the distribution of gout and hyperuricemia in several Pacific populations have been reported. Since Dr. Decker and others who have done this work are here today, I shan’t go into detail in this paper but will discuss some possible genetic implications of the findings. In brief, Dr. Decker found that gouty arthritis is more common in Filipino adult males admitted to a large city hospital than in a comparable non-Filipino population.x It was also found that hyperuricemia is very common in Filipinos living in North The distribution curve of the uric acid levels in this Filipino population did not fit a normal distribution curve but was skewed slightly to the right (Fig. 5). Using the method of moments and maximum likelihood methods, two normal distribution curves were constructed which suggested that the trait was dimorphic; that is, that there were two kinds of distribution in populations, one high and one low. The suggestion was made that this trait was, in fact, inherited and that it represented a genetic polymorphic trait. There is an interesting analogy to this analysis and it provides some guide as to how one might proceed with a further study of the problem. Beta-amino-isobutyric acid (BAIB) is an end product of pyrimidine metabolism. Some individuals excrete very large amounts of this material and others excrete smaller quantities. Figure 6 shows the distribution of BAIB excretion in two populations, U. S. Whites and Marshall Islanders. There is a striking 631 HEREDITY OF GOUT AND HYPERURICEMIA 35 - 30 - 25 - n 20 -* - z W - h c c 0, 2 4l AMtRlCAN WHITES V 3 15 z ti 10 - 5 - 0- I I I *5 I I 7 2.0 2.5 MARSHALL ISLANDERS (Rongelap) N = 188 1.0 1.5 2.0 2.5 3.0 LOG (mgm. ~-mino-isobu:yric ocid/mgm. creatinine) Fig. 6.-Frequency distributions of urinary excretion-rates of beta-amino-isobutyric acid of Marshall Islanders (Rongelap) and U S A . Whites. N = number of subjects. (From Blumberg and Gartler.14) difference between these two populations in respect to the levels of BAIB excretion. This is analogous to the findings in the US. Filipino and the nonFilipino population that Dr. Decker reported. I n Figure 7 the results of a study of Eskimos living in the village of Wainright, Alaska, are shown. I n this curve there is a distinct bimodality and it is Gour 632 AND PURINE METABOLISM ESKIMOS 1.5 2.0 2.5 Log (mgm. ,B-amino-isobutyricacid/mgm. creatinhe) 1.0 Fig. T.-Frequency distributions of urinary excretion-rates of beta-amino-isobutyric acid of Alaskan Eskimos. (From Allison, Blumberg, and Gartler.15) possible to separate the population into one group with high excretion and a second with low excretion. A great deal of the work on the metabolism and genetics of BAIB has been done by Dr. Stanley Gartler of the University of Washington in Seattle.ll In an appropriate population he was able to test a genetic hypothesis by selecting families in which the parents were of different or the same excretor phenotype. These, and studies by others, indicate that although the environment may affect BAIB excretion, most of the variation between individuals is under genetic control. Low excretors are either homozygous or heterozygous for a dominant gene and high excretors are homozygous for the recessive allele. Similar studies on the gene which determines the rate of metabolism of the anti-tuberculosis drug, isoniazid, have also been made.12 A study of the population distribution permitted the distinction of two types of individuals, those who rapidly metabolize isoniazid and those who metabolize it at a slower rate. By appropriate family studies it was shown that the rate of metabolism of isoniazid is an inherited autosomal trait. A similar method of analysis could be used to test a genetic hypothesis for the hyperuricemia trait in an appropriate population. It may be interesting to conjecture on possible mechanisms of genetic influences in gout and hyperuricemia. Studies in New Zealand have shown a high frequency of gout and hyperuricemia in the contemporary Maori.13 On the basis of historical evidence it was suggested that gout was not common in pre-European contact Maoris, but that it has increased in frequency with Westernization of the country. Decker has noted that there have been no reports of high uric acids levels or high frequency of gout in Filipinos living in the Philippines. This again suggests that there may be some change in the environment and/or cultural habits of these populations which leads to an increase in the uric acid levels. This does not necessarily mean that there is no genetic control of hyperuri- 633 HEREDITY O F GOUT AND HYl33RURICEMIA cemia. It is quite possible for some individuals to have an inherited trait which predisposes to high uric acid levels but which is not manifest unless certain environmental conditions prevail. An analogous example of this is the well known glucose-6-phosphate dehydrogenase deficiency trait. The gene determining G6PD deficiency is fairly common in some populations. Individuals with this trait appear normal under ordinary circumstances and the gene presumably is present in the population because of some selective advantage for it, either currently present or present in the past. However, when these individuals are exposed to an exogenous hazard such as fava bean or antimalarial drugs, they may develop a severe hemolytic anemia. Only when they are exposed to a particular foreign exogenous agent, in this case foodstuffs or drugs, does the gene represent a hazard to its bearer. It is possible that a similar gene exists for uric acid levels. Under ordinary circumstances, it would not result in hyperuricemia and gout. However, if individuals with the gene are exposed to an at present unknown agent the biochemical defect could become detectable. If the observations on the Filipinos and Maoris are correct, one could conjecture that this postulated exogenous agent was not present in the original ecological situation, but when the individuals moved to a new environment or when a new environmental situation developed in their original home, the gene became a hazard. ACKNOWLEDGMENTS Figure 5 is republished with the permission of Dr. John L. Decker and the Journal of Arthritis G Rheumatism. Figures 6 and 7 are republished with the permission of Nature, London. REFERENCES 7. Emmerson, B. T.: Heredity in Primary 1. Stecker, R. hl., Hersh, A. H., and Solomon, W. hl.: The Heredity of Gout and its Relationship to Familial Hypernricemia. Ann. Int. Med. 31:595, 1959. 2. Smyth, C. J., Cotterman, C. W., and Freyberg, R. H., Jr.: The Genetics of Gout and Hyperuricemia-an Analysis of 19 Families. J. Clin. Incest. 27:749, 1948. 3. Rakic, M. T., Davidson, F. H., Valkenburg, H. A., Engels, J., Neel, J. V., and Duff, I. F.: The Natural History of- Serum Uric Acid Levels in the Relatives of Patients Suffering from Gout. Arth. G Rheumat. 5:315, 1962. 4. Hauge, M., and Harvald, B.: Heredity in Gout and Hyperuricemia. Acta hied. Scand. 152:247, 1955. 5. Seegmiller, J. E.: Gout. New England J. of LMed. 268:772, A p d 4, 1963. 6. -: Discussion of Clinical Staff Conference: Biochemical Abnormalities in Hereditary Diseases. Ann. Int. Med. 57:472, 1962. Gout. Australian Annals of Medicine, 9:168, 1960. 8. Decker, J. L., and Lane, J. J., Jr.: Gouty Arthritis in Filipinos. New England J. of hied. 261:805, October 15, 1959. 9. -: Gout and Hyperuricemia in a Pacific Population Group. Eugenics Quarterly 9:54-58, 1962. 10. -, Lane, J. J., Jr., and Reynolds, W. E.: Hyperuricemia in a Filipino Population. Arth. G Rheumat. 5:144, 1962. 11. Gartler, S. M.: Comments on the Population Dynamics of Beta-Aminoisobutyric Aciduria. Proc. of the Conf. on Genetic Polyinorphisms a d Geographic Variations in Disease. New York, Grune & Stratton, 1962, p. 192. 12. Price Evans, D. A., Storey, P. B., and McKusick, V. A.: Further Observations on the Determinaton of the Isoniazid Inactivator Phenotype. Bull. Johns Hopkins Hospital 108: 634 60, 1961. 13. Lennane, G. A. Q., Rose, B. S., and Isdale, I. C.: Gout in the Maori. Ann. Rheum. Dis. 19:120, 1960. 14. Blumberg, B. S., and Gartler, S. M.: High Prevalcnce of High-Level BetaAmino-Isobutyric Acid Excretors in GOUT AND PURINE METABOLISM Micronesians. Nature 184:1990, December 26, 1959. 15. Urinary Excretion of Beta-Amino-Isobutyric Acid in Eskimo and Indian Populations of Alaska. Nature 183: 118, January 10, 1959. Discussion DR. BARTELS: It seems to me that geneticists are attempting to disprove something that most clinicians have accepted as fact. When questioned carefully regarding familial tendencies, a group of patients with gout personally seen since 1960 indicated that 21 per cent of them had a direct family history of typical gout. If we add to this the number of patients with a history of kidney stone (which in the case of uric acid stone may be due to hyperuricemia), the figure will be higher than 21 per cent. Again, if we add to this figure family members whom we occasionally see who have elevated uric acid levels but without gout or kidney stones, the figure reaches a highly significant level. In another group of 501 patients with definite gout seen prior to 1960, 119 (24 per cent) reported a history of gout in the immediate family. Of these two groups, 21 per cent and 24 per cent had a family history of gout. These figures compare with the incidence of diabetes in families. The Joslin group, in studying their diabetics, observed a 25 per cent family tendency. I have felt that our statistics, like those of many other students of gout, would suggest that gout does indeed run in families, and I am surprised at attempts to disprove this. DR. MIKKELSEN: At Michigan, several studies have been conducted which have some bearing on this problem. The first is the study of 19 gouty families begun some three decades ago by Smyth, Cotterman and Freybergl and continued in 1961-62 when Rakic and others2 reinvestigated the members of these families. The second is the Tecumseh Community Health conducted by the Department of Epidemiology of the School of Public Health, The University of Michigan, and in which members of our Unit have participated. In considering the genetic implications of these studies, we have been very fortunate to have the interest and participation of Dr. James V. Neel, Chairman of our Department of Human Genetics. Most of the remarks which I would like to make stem from discussions with him and from a paper which he has in press.5 I would say first of all that there is close general agreement between Dr. Blumberg’s comments and the thinking of Dr. Neel and others in our Unit. The family study indicates several general principles. First, although it has serious defects as a genetic marker, hyperuricemia does seem to persist as a fixed trait in those family members who manifest it. Secondly, it would appear that hyperuricemia does subject these subjects to an increased risk of developing clinical gout. 635 HEREDITY OF GOUT AND HYPERURICEMIA DISTRIBUTION OF SERUM URIC ACID VALUES Tecumseh, 1959-1960 X W v)221 D h w .---a Females ( 3,O I I ) c-.Males (2,983) !k Ya ’ 8 v) 4- a I- e LL 0 z W 0 [L W a 0.5-0.9 1.5-1.9 2.5-2.9 3.5-3.9 4.5-4.9 5.5-5.9 6.5-6.9 SERUM URIC ACID IN 1/2mgms O/O 7.5-79 8.5-89 95-99 CLASSES Fig. 8. Following the completion of Dr. Rakic’s study, Dr. Neel, recognizing the many difficulties that Dr. Blumberg has noted, utilized these data in an exploratory way to consider further the role of familial factors. He investigated the degree of relationship of the subject to the propositus in relation to the serum uric acid level. It was his thesis that if familial or genetic factors were involved the mean serum uric acid level should decrease as the degree of relationship of the subjects to the propositi becomes more remote. Generally, this trend was apparent in these data, although the number of subjects in certain of the age and sex groups is rather small. Examination of the distribution of serum uric acid values for the larger group of relatives examined in 1961-1962 failed to show unequivocal evidence of bimodality although the possibility of two overlapping normal curves could not be completely dismissed, especially in the females. Dr. Neel regarded these findings as consistent with those of other investigators that familial factors were of importance, but cautioned, as Dr. Rlumberg has, that the demonstration that familial factors are operative is not necessarily proof that they are genetic in nature. The second investigation which I would like to describe briefly to you is the population study in Tecumseh, Michigan. Evaluation of hyperuricemia and clinical gout was included in the design of this total community study because we also recognized the bias in the way in which gouty families are selected, and that this may lead to erroneous conclusions. Some 9,000 individuals were examined in this study and serum uric acid determinations were performed by the enzymatic spectrophotometric method6 on 6,000 individuals of both sexes and age four years and over. In Figure 8 are shown the distribution curves for serum uric values for male and female subjects. Although there is an 636 GOUT AND PURINE METABOLISM appreciable difference in the two curves, as would be expected, it was considered that both showed evidence of skewing toward the higher values but that neither showed any apparent evidence of bimodality. This was also true of the distribution curves for males and females in the age groups 4 through 19, 20 through 49, and 50 years and over. These findings are regarded as consistent with the hypothesis that serum uric acid concentration is distributed as a continuous variable, although the possibility of two overlapping distribution curves cannot be rigorously excluded. This is preliminary information. We have not yet had an opportunity to extend these observations to gouty kindreds in comparison with matched control kindreds, as suggested by Dr. Blumberg, but such studies are contemplated for the future. REFERENCES Study in a Total Community. Uniu. 1. Smyth, C. J., Cotterman, C. W., and Freyberg, R. H.: The Genetics ot Mich. Med. Bull. 26:307, 1960. Gout and Hyperuricemia-An Anal- 5 . Ned, J. V., Rakic, hl. T., Davidson, R. ysis of Nineteen Families. J. Clin. Inu. T., Valkenburg, H. A., and Mikkelsen, 27:749, 1948. W. M.: Studies on Hyperuricemia. 11. 2. Rakic, M. T., Valkenburg, H. A., DavidA Reconsideration of the Distribution son, R. T., Engels, J. P., Mikkelsen, of Serum Uric Acid Values in the W. M., Ned, J. V., and Duff, I. F.: of Smyth, Cotterman and Families Observations on the Natural History Freyberg. Am. J. Human Genetics, in of Hyperuricemia and Gout, I. An press. Eighteen Year Follow-up of Nineteen 6. Liddle, L., Seegmiller, J. E., and LasGouty Families. Am. J. M e d . 37~862, ter, L.: The Enzymatic Spectrophoto1964. metric Method for Determination of 3. Franciy, T., Jr.: Aspects of the Tecumseh Study. Pub. Health Rep. 76~963,1961. Uric Acid. J. Lab. and Clin. Med. 54: 4. Epstein, F. H.: An Epidemiological 903, 1959. DR. DECKER: Our studies mentioned by Dr. Blumberg dealt with a special population-the Filipinos. I should say immediately that the study had nothing to do with gout. It was a survey of serum uric acid levels in a randomly selected population, limited to males. You will recall the original findings that Dr. Blumberg alluded to, showing hyperuricemia in Filipinos as compared to a group of white male pris0ners.l It is quite clear that the diet and the activity of the prisoners were different from those of the active Filipino workers. Accordingly, we considered environmental factors to represent a possible explanation. With this in mind, we have just completed but have not yet fully analyzed another study, in which we sought a population where a number of different races were on the same diet for an extended period of time, and their environment was similar.2 This kind of a population is a bit unusual and not easy to come by. The best one that we could locate was in the Hawaii State Hospital, a mental institution at Kaneohe, Oahu. Here we studied some 365 males, recording a good deal of information about what drugs they were taking, blood pressures, weight, height, and so on, and obtained serum samples on which uric acid analysis have been performed. This population is very interestingly 637 HEREDITY OF GOUT AND HYPERURICEMIA broken down into Chinese, Japanese, Filipino, Polynesian and Caucasians, as well as others. In this study, the same degree of hyperuricemia was found in Filipinos as had been previously reported. There is no question about it; they are strikingly more hyperuricemic than the population as a whole or any other subsample of the population. REFERENCES 1. Deckcr, J. L., Lane, J. J., Jr., and Reynolds, W. E.: Hyperuricemia in a Filipino Population. Arth. G Rheumat. 5:144, 1962. 2. Decker, J. L., Caner, J. E. Z., and Hel- sey, L. A.: Racial Variations in Serum Uric Acid in a Controlled Environment. (Abstract). Arth. G R h e u ~ t . 7:730, 1964. DR. RAKIC:Since it has been reported that distribution curves for serum uric acid levels may vary in different populations, and that this variation may be due either to environmental factors or to the action of multiple genes, our study was undertaken in Surinam (formerly Dutch Guiana), where a unique situation exists in that several different racial groups live under similar environmental factors. Surinam is situated on the northeastern coast of South America, near the equator. The population consists of one-third Creoles, onethird Hindustani, one-fifth Indonesians, one-tenth Bush Negroes; the remaining population consists of American Indians, Europeans, Chinese, Jews and others. Our survey was conducted on three racial groups and Dutch soldiers: of 332 specimens obtained (all from adults), 76 were from Djukas (Bush Negroes), 101 from Hindustani (47 males and 54 females), 123 from Indonesians (68 male and 55 females), and 32 from Dutch soldiers. Djukas (the Bush Negroes ) were African slaves imported as plantation labor. After the abolition of slavery, most of the freed slaves refused to work on the plantations and went into the interior to resume living as in Africa. This resulted in the immigration of other racial groups to work on the plantations. The Hindustani represent all people who emigrate from India, whatever their religion. Indonesians were brought to Surinam between 1891 and 1939 to work on the plantations as laborers and some of them became small independent farmers. Serum uric acid levels were determined by the enzymatic spectrophotometric method. The method was compared with one used in the Rackham Arthritis Research Unit, University of Michigan, and similar values were found. Therefore, a comparison could be made with a Caucasian population from North America studied at Tecumseh, Michigan. All individuals in the study were in good health, free of gout, and had not currently taken any medication. The distribution of mean serum uric acid levels in males and females according to age is presented in Figures 9 and 10. There is no significant difference in serum uric acid levels in the various racial groups. Figure 11 shows the distribution of serum uric acid levels for the males and females of all the races studied in Surinam. The distribution has the same general pattern as the Tecumseh population. The average serum uric acid was 5.7 638 GOUT AND PURINE METABOLIShl REGRESSION OF SERUM URIC ACID M E A N S ON AGE MAL ZS H"DUS T A N ' - 47 INDONESIANS-6 8 SUA 6 TECUMSEH- 874 4 1D J U K A S - 7 6 I I I I I I I REGRESSION OF SERUM U R I C ACID f €MAL € . 1 M E A N S ON AGE HINDUSTANI SUA - 54 INDONESIANS- 55 6 h .t TJ', rECUMSEH - 899 3 'I I u 0 10 I I I I I d 20 30 40 IN YEARS 50 60 70 AGE Fig. 10. mg. per cent for males and 4.9 mg. per cent for females. When the distribution of hyperuricemia among the various populations was compared at the arbitrary cut off point of 6 or 7 mg. per cent for males and 5 or 6 mg. per cent for females, no significant differences between the different racial groups were observed. Of the total group of 191 males, 38.7 per cent had a serum uric acid of 6.0 mg. per cent or higher, and 6.8 per cent of the 191 males had a value of 7.0 mg. per cent or higher. Of the total group of 109 females, 42.2 per cent had a 639 HEREDITY OF GOUT AND HYPERURICEMIA D~STRIBUTION OF SERUM URIC ACID LEVELS TEGUMSEH ..... . all ages 2 4 6 8 1 0 2 MALES 4 6 8 1 0 FEMALES SURINA M.?...ages I8 and over I 2 * H,nd"stoni, I09 I r 4 6 8 1 0 2 4 6 8 SERUM URIC ACID LEVELS in m g m / I O O m l 1 0 Indoneslon, DJukOS Fig. 11. serum uric acid value of 5.0 mg. per cent or more; 20.2 per cent of the 109 females had a value of 6.0 mg. per cent or higher. These values are surprisingly high and raise the question whether our upper normal limits for serum uric acid by the enzymatic spectrophotometric method (6.0 mg. per cent for males and 5.0 mg. per cent for females) are too low. DR. SEEGMILLER:The mean value for over 800 male army inductees who were studied by Dr. Stetten was 5 mg. per cent, and two standard deviations would glace the upper range of normal at 6.8 mg. per cent. This raises a question which I would be interested in hearing our population geneticists discuss. How do they decide whether or not 6.0 mg. per cent is a legitimate figure for dividing their populations into the hyperuricemic or the nonhyperuricemic? I 640 GOUT AND PURINE METABOLISM think that there are evidences in the literature that investigators may go astray by failing to take into account the upper range of normal for the particular method that is being used. One reason why the enzymatic method gives somewhat higher values than many of the colorimetric methods lies in the fact that it requires no protein precipitation, with its attendant coprecipitation and loss of uric acid. An added value of the enzymatic method is its specificity. None of the substances that can give rise to chromogens in the colorimetric methods are measured in the enzymatic determination. I would like to ask Dr. Decker whether or not there is any evidence of a uniform cause of hyperuricemia in the Filipino population. DR. DECKFR: There was no evidence of intermarriage in Filipinos. The majority of the ones that we studied had their origin in the Philippine Islands, coming from a normal environment, It is a large population and there is no reason to believe that they are all related. The studies that were reported today employed the enzymatic method for determination of uric acid. And SO far a5 our own definition of hyperuricemia is concerned-we assiduously avoid the word and speak only of population curves and means. It is very dangerous and, I think, altogether misleading to pick a value and talk about those above and those below. DR. BLUMBERG: I’m pleased to hear that the conclusions Dr. Nee1 arrived at were similar to our own. I think many of the questions revolve around the interpretation of a simple normal distribution curve versus two normal distribution curves. The kind of curve that John Decker found in his population raised this issue. A single skewed curve could be divided into two overlapping normal distribution curves. The extreme of this situation would exist when the two normal distribution curves were completely separate. Then there is no difficulty in distinguishing them. Now, when one talks about hyperuricemia and normal uricemia, one is assuming in effect that there is a dimorphic curve and that at some point a semiarbitrary decision can be made as to what constitutes a high level and what constitutes a low level. One then uses those designations to call an individual the phenotype hyperuricemia or the phenotype normal uricemia. However, this kind of dimorphic curve has never been definitely shown, with the possible exception of the one that Dr. Decker found. But it is quite possible that it exists nevertheless, and it would not be an obvious distinction in a population where hyperuricemia was quite rare. Let us say that there was a dimorphic curve. If hyperuricemia was quite rare, it would be very di5cult to see it in a study of that population. In studies in which small numbers are used-such as those from Surinam-the dimorphic distribution of uric acid levels, if it exists, would be difficult to see. DR. WYNGAARDEN: There was a suggestion of dimorphism in the original Stecker paper and also in the Cobb studies in Pittsburgh, I believe, DR. BLUMBERG: Yes, and also in Dr. Smyths paper. There was an interesting observation in Dr. Decker’s paper. He noted that the serum uric acid levels were elevated in Filipinos who were taking drugs but they were not elevated in non-Filipino controls who were receiving HEREDITY OF GOUT AND HYPERURICEMIA 641 the same drugs. Although it is a rather thin thread, this observation is conceivably related to the conjectural hypothesis I put forward. Perhaps Filipinos develop hyperuricemia when exposed to drugs because of some inherited metabolic trait which is not manifest until some exogenous material such as a drug is ingested. This is analogous to G6PD deficiency and antimalarial drugs. I think the question of what constitutes hyperuricemia is DR. MIKKELSEN: an important one, but extremely difficult to answer. In considering serum uric acid levels we are dealing, I think, with a measurement that appears to be continuously distributed. It seems very unlikely that we can find any arbitrary cutting-off point that will sharply define two groups. In this respect we are at a considerably disadvantage in comparison to the situation in many of the other genetic disorders. Data from the Tecumseh Study indicate mean serum uric acid values and standard deviations that are not greatly different from those mentioned by Dr. Seegmiller. The mean value for 6,000 determinations in subjects of both sexes age 4 years and over was 4.5 mg. per 100 ml. with a standard deviation of 1.3. Male subjects had a mean of 4.9 k 1.4 mg. per 100 ml. and females a mean of 4.2 t 1.2 mg. per 100 ml. One approach would be to attempt to define normal boundaries on a statistical basis, utilizing such data as these. One could base such a definition on the mean value plus and minus two standard deviations or on the percentile distribution of values. To the present we have adhered to the philosophy previously mentioned by Dr. Decker and avoided classification of individuals as “normouricemic” or “hyperuricemic,” especially on the basis of a single casual determination. In considering normal values there are significant age as well as sex effects (Fig. 12). As indicated, at age 4 the values are significantly lower than in later years and there is no sex difference. Values rise, as others have reported, during childhood and puberty, and the magnitude of this rise is greater in males than in females. Throughout the middle years of life the mean values are approximately 1.0 mg. per 1010 ml. higher in males. Females show a secondary gradual rise that appears to bracket the years associated with the menopause. Thus, in considering a definition of hyperuricemia, one needs to consider age as well as sex. Dr. H. J. Dodge, who has worked closely with us in this study, has attempted to take these variables into consideration by calculating a serum uric acid “score,” which attempts to adjust the individual value on the basis of these age and sex differences. Our present approach is to look for correlations of serum uric acid values with other variables and within families on the basis of both the observed value and the calculated “score.” DR. HALL:The distribution curves for uric acid in the Framingham Heart Study population, both for the population as a whole and for each age group, are very similar to the ones already shown. The Framingham Heart Study is a long-term study and four serum uric acid determinations have been performed. The means and standard deviations are about the same as have been reported. A 14 year follow-up of the Framingham Heart Study population, consisting of seven examinations at two-year intervals, has enabled us to see what has happened to the population in terms of gouty arthritis. In de- * GOUT AND PURINE METABOLISM 642 SEX-AGE SPECIFIC MEAN SERUM URIC ACID Tecumseh, 1959-1960 s.Ol a9 In 5.0- - - 5E 4.0- z P - 0- --. Males (2987) Females ( 30 13) 1 0 ’ 1 I 4 l 10-14 l I 20-24 I I 30-34 I I 40-44 l l 50-54 I I 60-64 I I 70-74 I i 80+ AGE GROUP IN YEARS Fig. 12. fining the limits of normal values obtained in a test such as the uric acid determination one may use statistical methods and select as abnormal, for example, all values differing from the mean by more than two standard deviations. It is desirable for our purposes, however, to define the limits in terms of values beyond which there is an association with disease. We were interested in defining, if possible, the normal values as opposed to the abnormal values in terms of association with gouty arthritis (Table 2). The Framingham Heart Study population was classified on the basis of the highest serum uric acid value obtained at any time in a series of four determinations two years apart. It is apparent that there is an increase in the cumulative prevalence of gouty arthritis as the serum uric acid values go up. Note, however, that 19 (29 per cent) of the 65 men who have developed gouty arthritis have not had a value above 6.9 mg. per cent. A much smaller per cent (8 per cent) have not had a value above 5.9 mg. per cent. When we made this analysis on the basis of a single examination or on the basis of the mean uric acid value for each subject, many more of the gouty subjects had values below 6 mg. per cent. For example, on the basis of a single examination 70 per cent of the gouty subjects had values below 7 mg. per cent and 26 per cent had values below 6 mg. per cent. It is obvious that the determination of the level above which the uric acid value might be considered abnormal depends, in part, on how many examinations have been carried out, how old the population is, and whether one con- 643 HEREDITY OF GOUT AND HYPERURICEMIA Table 2 PREVALENCE O F GOUTY ARTHRITLS BY HIGHEST URIC ACID VALUE AT ANYTIME FRAMINGHAM HEART STUDY Less than 6 mq.55 Men 5 (1383) (0.3%) Women 2 (2660) (0.08%) 6 - 6.9 mq.% 14 (790) (1.6%) 5 (151) (3.3%) 7 - 7.9 mg.55 27 (162) 4 (23) (16.7%) (17.4%) 8 mq. W or more 19 (50) (36%) 0 (5) TOTAL Men 65 (2283) (2.8%) Women 11 (2844) (0.4%) siders a single determination the highest value obtained or a mean value for each subject. We prefer, therefore, to say that there is no definite upper limit of normal, but that the risk of developing gouty arthritis increases as the serum uric acid goes up, whether the values considered are random values, maximal values or mean values. We prefer to talk about what happens to people in a certain serum uric acid range, just as Dr. 14ikkelsen has said. DH. HUGHSMYTHE:We also have carried out population studies involving two groups, university students and healthy males of the age of 45 to 60, and have shown no essential difference in means or in distribution in those two groups. But also we’ve used hospital populations-people who do not have gout, who are not taking drugs associated with changes in serum uric acid, and do not have renal failure-and in these the means were not different, but the distribution was quite different. There was quite a skew into the higher uric acid values. In the younger populations or the healthy populations, we would have about five per cent of the values falling above 7 mg. per cent, but in this other nongouty population there would be 15 per cent of the values falling above 7.0 mg. per cent. With the lack of a normal distribution, it becomes impossible to describe these findings in terms of a normal and two standard deviations. Some other statistical treatment is needed if you’re going to compare this with other groups. This is important also from a diagnostic point of view because we’re often asked to discuss the diagnostic significance of a high serum uric acid value in a hospital population. DR. WYNGAARDEN: Dr. Hall, in your population studies have you studied the subjects with the higher urate values in terms of renal function or of drugs they might have been taking? DR. HALL:The drugs that were being taken at the time of each examination are well documented. We can be reasonably certain that drugs did not ma- 644 GOUT AND PURINE METABOLISM terially affect the serum uric acid values we have described. Most of the values were determined before the advent of probenecid and all before the advent of the thiazides. Renal function, for all the reasons that were already mentioned, we really can’t say much about. With very few exceptions there does not seem to be any difference between the gouty subjects or the subjects with high uric acid values and the population as a whole in terms of renal function. On the other hand, the only way we have of evaluating this is in terms of repeated urinalyses and in terms of what had happened to the subjects over the years of follow-up. We think that impaired renal function has not been very important in our data, partly because only five of the 240 subjects with uric acid values of 7 mg. per cent had any clinical evidence of renal disease on admission to the study (only two of whom appeared to have their hyperuricemia possibly on the basis of renal disease) and partly because the subjects with renal disease did not show a progressive rise in uric acid values. DK. WYNGAARDEN: One point which might be restressed here is that Burch, OBrien and Bunim found that many nongenetic factors were at play in the hyperuricemia of the populations they studied. The serum urate values appeared to correlate with weight, and even more specifically with body surface area. Have you taken these factors into account in your studies? DR. HALL:Yes. For each subject in the Framingham Heart Study population an index called “The Framingham Relative Weight” was determined on admission. Observed Weight FRW = _ _ Median Weight for Height and Sex - x 100 The people with hyperuricemia and gouty arthritis tended to have a little higher index than the population at risk. The difference is not very great, but is statistically significant. The gouty patients were not heavier than the whole group of subjects with serum uric acid values of 7.0 mg. per cent or over, and it is difficult to know how strong the relationship is in terms of other risks associated with gouty arthritis or hyperuricemia. One observation (Fig. 13) which is a little difficult to interpret may be related to Dr. Blumberg’s remarks about environmental factors. We were surprised to find that such a high percentage of people who have high serum uric acid values have not yet developed overt gout. If the development of gouty arthritis is dependent only on a high serum uric acid and time, we would expect to see more and more cases develop every year until most of the subjects at risk have had an attack. The mean age of the Framingham Heart Study population on admission to the study was 44 years. As you can see, during the years between 1947 and 1957, when the mean population age was 42-52, there was a fairly high incidence of new cases of gouty arthritis. The incidence has now dropped off so that whereas we were getting nine or ten new cases every two years, we are now getting only one or two. Whether this means that there are two populationsone of which is hyperuricemic on the basis of the normal distribution curve, and another which is hyperuricemic on the basis of a genetic defect-I don’t know. It is also possible that there was an environmental factor in the com- HEREDITY OF GOUT AND HYPERURICEMIA 645 CUMULATIVE INCIDENC€ OF GOUTY ARTHRITIS IN RELATION TO MEAN AGE OF POPULATION HEN 30-59 AT ENTRY: FRAMINCHAM HEARTSTUDY 70 60 Ln a w 50 4 40 a 30 20 10 ~~ ~ Zb283OJo343636 4 0 4 2 1 4 4 6 4 5052545bJO MEAN AGE OF POPULATION Fig. 13. munity which was present between 1947 and 1957 but which is absent now. In any event, the incidence of new attacks is dropping off, even though only about one third of those with serum uric acid levels of 8.0 mg. per cent or more have had an attack. DR. BLUMBERG:One possible explanation of these observations is, I think, an extremely interesting one. That is, that nearly everyone who can get gout has already succumbed to the disease. There are a limited number of people who can get gout, given the environmental exposure, and the Framingham population is now reaching the upper limits of this number. DR.HALL:That would have to be on the basis of two kinds of people though, wouldn’t it? DR.BLUMBERG: Those who can get gout and those who can’t. DR. HALL:Yes. And determining the serum uric acid level is a reasonable way of finding people who are going to get gouty arthritis. On the other hand, that method may include a lot of people who aren’t particularly at risk. Another aspect of this question is the fact that there are some differences between those who get gouty arthritis and those who merely have high serum uric acid levels, in terms of the incidence of coronary heart disease. That fact may constitute an additional argument for the existence of two populations. 646 GOUT AND PURINE METABOLISM In any event, we have no ready explanation for the fact that there are some people on the high end of the serum uric acid distribution curve who are not developing gouty arthritis, and that there has bcen a striking drop in the incidence of new cases even in terms of the per cent of those left at risk (hyperuricemics who have never had arthritis ) who are developing new attacks. DR. DECKER: What is the total number of people involved? DIL HALL:There were 2,283 men between the ages of 30 and 59 at entry. DR. BLUMBERC: Another possible explanation is that there are two kinds of hyperuricemia. DR. HALL:That’s right. One kind reprcsented by those who are predisFosed to gouty arthritis and another represented by those who have no such predisposition. DR. RLUMBERG: By continuing your follow-up you’ll know the ones that don’t predispose to gout. DR. HALL:Yes, in retrospect. But it brings up the question of how often gout is secondary. We know that under particular stress, such as fasting, an operation, or chlorthiazide administration, some people develop attacks of gouty arthritis. Does this happen only to hyperuricemic subjects in the group who we have suggested are at particular risk? Are the other hyperuricemic subjects not liable to develop arthritis under the same stressful conditions? If such is the case, we should not expect the subjects who are presumably hyperuricemic on the basis of uremia to develop arthritis, and yet they do. On the other hmd, we have a certain number of subjects who were later admitted to hospitals and called secondary hyperuricemics who were definitely hyperuricemic years earlier. Of the 240 subjects with serum uric acid levels of 7.0 mg. per cent or more, in none can we be sure the hyperuricemia is secondary to another disease. Un. C. SMYTH:There is another factor I think we had better consider. Everybody is talking about gout as though gout were one thing. I think we should ask in these studies whether or not these patients had hereditary gout, whether any of them had so-called “pseudo gout,” and how many of them were secondary to some hematopoietic illness. Gout, I used to think, was a relatively pure culture, but I’m sure the more we learn, the more it will turn out not to be. DR. HALL: We had one subject with polycythemia after splenectomy for congenital hemolytic anemia, but he entered the study after these events and we have no serum uric acid values prior to the events. We do not have in the Framingharn Heart Study reliable ways of detecting positive family histories for gouty arthritis. Histories obtained about relatives from the subjects in the study are pretty unreIiabIe and we haven’t a good way of examining the relatives, many of whom are scattered all over the world. We may get some information in the future because the Framingham Heart Study is going to start bringing in the children of the study population. We don’t think we have any cases of gouty arthritis which were drug-induced. On the first serum uric acid examination, for example, only two of the subjects who subsequently developed attacks of gouty arthritis were taking anything at all in the way of drugs and both were taking salicylates. One was taking uricosuric doses of sodium salicyl- HEREDITY OF GOUT AND HYPERURICEMIA 647 ate and had a high serum uric acid anyway. The other was taking only a few tablets a month. The other examinations were similar as far as drugs were concerned. DR. SEECMILLER: From the standpoint of the practicing clinician, it would be helpful if one could define empirically a concentration of serum urate which would be most helpful in separating gouty from nongouty individuals. Such a value might be somewhat different from the upper range of “normal” conventionally defined as two standard deviations above the mean. D R . HALL: Well, of 790 men whose highest serum uric acid was between 6.0 and 6.9 mg. per cent only 1.8 per cent have had an attack of gouty arthritis so far. Of 162 men whove highest level fell between 7.0 and 7.9 mg. per cent, 16.7 per cent have developed gouty arthritis. Of the 50 men with uric acids over 8.0 mg. per cent, 36 per cent have developed gouty arthritis. Now part of the problem involved in answering your question is related to what you wish to establish as the “risk in which you are interested. If you are interested in a one out of three chance of developing gouty arthritis, you pick 8.0 mg. per cent. If one out of ten is your risk level, then the serum uric acid is going to be lower. You have to decide what value in the continuum is the most useful one for your purposes. Here again these values represent the highest of our determinations. DR.DECKER: I would simply suggest that gout is a clinical diagnosis-gouty arthritis-and it is not any too well correlated with the serum uric acid level. It is clear that you have a curve of people who get gout, which has to be superimposed on a curve of population serum uric acid levels. There is going to be a big overlap, but apparently they are not exclusive to each other. D R . HALL:When you talk to people who are interested in the epidemiology of coronary disease, they say they wish they had a method of detecting subjects with a one out of three chance of developing coronary disease. The correlation between uric acid levels and gouty arthritis is not a bad correlation by any means. It is just not as good as clinicians in general have hoped and assumed. DR. MCCARTY:I just looked at it from the opposite point of view. Persons who are hyperuricemic with arthritis-approximately two-thirds of that group are gouty. The remaining third has diagnoses other than gout but is hyperuricemic, and we’ve gone on the supposition that the diagnosis is a clinical one -yes, but certainly substantiated by pathologic examination of joint fluid.