The epidemiology of favism

Bull. Org. mond. Santé 1973, 48, 1-13 Bull, Wid Hith Org. UAM Qs \Deras Y LA A. Vea “> The epidemiology of favism ! MARK A. BELSEY Favism is a potential obstacle to the use of the fava bean in the development of a locally produced, inexpensive weaning food for the Middle East and North Africa. The purposes of this study were to define the epidemiology offavism, to evaluate the advisability of using the fava bean in a weaning food, and to suggest ways of avoiding or.eliminating the toxic factor in the bean. Field observations, locally acquired data, and a literature review suggested that the use of the fava bean in a weaning food wouid be hazardous, but that the hazard might be overcome by using certain strains of the bean or, more particularly, by using old dried beans. The disease is usually directly related in time to the harvesting and availability of fresh beans, but it is also associated with fresh dried beans. On the basis of the age distribution of the disease, patterns of bean consumption, and local food taboos it appears that the toxic factor is concentrated in the skin of the bean, that it is heat-stable, that in dried beans it decreases with age, and that it crosses into the breast milk of lactating mothers. It also appears that disease expression may be a result of the interaction of several host factors, such as nutritional status and the consumption of other foods. These observations are consistent with the results of laboratory studies, which incriminate vicine, divicine, and DOP A4 in the etiology of favism. In order to provide an adequate source of protein for infants and children, a combination of a cereal intended to define the potential effect of favism on a nutrition programme based on the use of a weaning grain and broad beans (Vicia fava) has been sugfood containing V. fava. A second purpose is to gested as a weaning food for use in the Middle East describe and relate the epidemiology of favism to and North Africa, where both wheat and broad beans are grown extensively. Although the fava bean laboratory and clinical studies, thereby seeking clues to the approach to be taken in preventing the disease. is low in methionine content it is rich in lysine {Aykroyd & Doughty, 1964), and can thus compenDESCRIPTION OF THE DISEASE sate for the low level of lysine in wheat (FAO, 1955). However, there are potential risks in so using the fava bean. Some individuals who eat or are exposed Clinical description to it are reported to develop an acute haemolytic Favism is characterized most often by four signs anaemia, favism. The disease may be fatal, usually and symptoms: weakness or fatigue, pallor, jaundice, occurs in children (especially males), and is generand haemoglobinuria. The symptoms are distinct ally thought to be related to the genetic deficiency of erythrocyte glucose-6-phosphate dehydrogenase enough to be recognized as characterizing a specific (1.1.1.49) (G6PD). Favism therefore must be viewed disease. Five different forms of the disease, based on as a potential obstacle to this solution to the problem the degree of severity, have been described by Gasof malnutrition in the Middle Fast and North Africa. In some areas the disease itself is a major public barrini (1915). The interval between exposure to the fava bean health probiem (Donoso et al., 1969). and the onset of symptoms may be only a few hours The following review and field investigation of favism in the Middle East and North Africa is and is usually under 24 h in both clinical and disease entity among populations experiencing the experimental studies, but it may be as long as several days (Panizon ‘ Associate Professor of Epidemiology and Pediatrics, Tulane University, New Orleans, La., USA. Present address: Human Reproduction, World Health Organization, 1211 Geneva 27, Switzerland. 2971 _ & Vullo, 1961). In contrast, the interval between drug exposure and haemolysis in the G6PD-deficient individual is several days (Burka et al., 1966; Dern et al., 1954).

Bull. Org. mond. Santé Bull. Wid Hith Org. | 1973, 48, 1-13 The epidemiology of favism MARK A. BELSEY! Favism is a potential obstacle to the use of the fava bean in the development af a locally produced, inexpensive weaning food for the Middle East and Norih Africa. The purposes of this study were to define the epidemiology offavism, to evaluate the advisability of using the Java bean in a weaning food, and to suggest ways of avoiding or eliminating the toxic factor in the bean. Field observations, locally acquired data, and a literature review suggested that the use of the fava bean in a weaning food would be hazardous, but that the hazard might be overcome by using certain strains of the bean or, more particularly, by using old dried beans. The disease is usually directly related in time to the harvesting and availability offresh beans, but it ts also associated with fresh dried beans. On the basis of the age distribution of the disease, patterns of bean consumption, and local food taboos it appears that the toxic factor is concentrated in the skin of the bean, that it is heat-stable, that in dried beans it decreases with age, and that it crosses into the breast milk af Inctating mothers. It also appears that disease expression may be a result of the interaction of several host factors, such as nutritional status and the consumption of other foods. These observations are consistent with the results of laboratory studies, which incriminate vicine, divicine, and DOPA in the etiology of farisım. In order to provide an adequate source of protein for infants and children, a combination of a cereal grain and broad beans (Vicia fava) has been suggested as a weaning food for use in the Middle East and North Africa, where both wheat and broad beans are grown extensively. Although the fava bean is low in methionine content it is rich in lysine {Aykroyd & Doughty, 1964), and can thus compensate for the low fevel of lysine in wheat (FAO, 1955). However, there are potential risks in so using the fava bean. Some individuals who eat or are exposed to it are reported to develop an acute haemolytic anaemia, favism. The disease may be fatal, usually occurs in children (especially males}, and is generally thought to be related to the genetic deficiency of erythrocyte glucose-6-phosphate dehydrogenase (1.1.1.49) (G6PD). Favism therefore must be viewed as a potential obstacle to this solution to the problem of malnutrition in the Middle East and North Africa. In some areas the disease itself is a major public health problem (Donoso et al. 1969). The following review and field investigation of favism in the Middle East and North Africa is Associate Professor of Epidemiolosy and Pediatrics. Tulane University, New Orleans, La., USA. Present address. Human Reproduction, World Health Organization. Geneva 27, Switzerland, 2971 1211 intended to define the potential effect of favism on a nutrition programme based on the use of a weaning food containing F. fava. A second purpose is to describe and relate the epidemiology of favism 10 laboratory and clinical studies, thereby seeking clues to the approach to be taken in preventing the disease. DESCRIPTION OF THE DISEASE Clinical description Favism is characterized most often by four signs and symptoms: weakness or fatigue, pallor, jaundice, and haemoglobinuria. The symptoms are distinct enough to be recognized as characterizing a specific disease entity among populations experiencing the disease. Five different forms of the disease, based on the degree of severity. have been described by Gasbarrini (1915) The interval between exposure io the fava bean and the onser of symptoms may be only a few hours and is usually under 24 h in both cfinical and experimental studics, but if may be as long as several days (Panizon & Yullo, 1961), In contrast, the interval between drug exposure and haemolysts in the G6PD-deficient individual 15 several days (Burka et al, 1966; Dern et al, 1954).

Favism has been observed among individuals who had previously consumed fava beans without showing signs or symptoms of the disease; others have developed the disease on the first exposure (K.attamis et al., 1969; Angelov & Andrev, 1959). In Egypt most of the cases occur among infants on their first exposure to the bean, which is reflected in the high proportion of cases among children under 1 year old (Ghafsarpur, personal communications). These later reports do not support the hypothesis of Kantor & Arbesman (1959) that favism represents in part a hypersensitivity phenomenon among individuals who have previously ingested or been exposed to the bean. Repeated attacks of favism are not uncommon; second attacks of favism were noted in 10 of 120 patients studied by Kattamis et al. (1969). No note of repeated attacks has ever been made in the same year in any of the reported studies or personal communications, although a history of annual overt attacks or compatible symptoms was elicited from villagers and physicians in Egypt, Iran, and Tunisia. Family clustering of cases 1s noted, both among the sporadic cases (Vince-Ribarié, 1962; Gehrmann et al, 1963) and in larger series. Kattamis et al. 41969) noted the frequent occurrence of mild compensated symptoms in mothers at the time their male children developed favism. An acute haemolytic anaemia with symptoms identical to those in favism was ascribed by Lederer (1925) to an infectious process before the G6PD defect was known, Subsequent writers have speculated that many of the cases described as Lederer’s anaemia actually were cases of favism in which a history of F. fava ingestion was not specifically asked for, was unknown, or was denied despite actual ingestion {Gelin, 1952; Wharton & Duesselman, 1947), Other cases of Lederer’s anaemia may represent drug exposure or infection superimposed on G6PD deficiency (Burka et al., 1966). Laboratory findings Laboratory findings in faviem reflect the underlying G6PD defect, the haemolytic anaemia, and its consequences. They may also reflect the severity of the G6PD deficiency, the dose and form of fava bean exposure, and the acuteness and severity of the anaemia. Erythrocyte counts are markedly diminished among the hospitalized patients, most being between 1 and 2 million cells per mm? (Kattamis et al., 1969, Joannides, 1952; Messerschmitt et al., 1967). Since patients with symptoms severe enough to necessitate blood transfusion are hospitalized, the “ true ” distribution of either erythrocyte counts or haemoglobin levels among all patients with haemolysis following fava bean exposure is not known. Urinary findings are frequent in hospitalized patients. Haemoglobin is found in large amounts in the urine for 1-3 days; more prolonged haemoglobinuria is not usually observed. Small amounts of oxyhaemoglobin and methaemoglobin are noted in the urine, which appears dark brown, red, or even black. Oliguria and even anuria may occur in severe cases, with cancomitant azotaemia. Death may occur from renal failure. The G6PD level of patients with favism is generally low even during the acute episode (Kattamis et al., 1969; Larizza et al., 1958; Russo & Balsamo, 1962). The G6PD deficiency has also been noted in the granulocytes of patients with favism (Ideo et al., 1965), Nearly all studies on patients with favism indicate that there is a low level of reduced giutathione (GSH) and instability of GSH on in vitro incubation with l-acetyl-2-phenylhydrazine (Sansone & Segni, 1956). In the Mediterranean type of G6PD deficiency young erythrocytes appear to be almost as deficient of G6PD as old erythrocytes. During haemolytic crises among Caucasians with G6PD deficiency, the GSH instability of surviving erythrocytes is not decreased, nor is the GSH significantly increased (Zannos-Mariolea & Kattamis, 1961). This observation contrasts with that made among primaquinesensitive! G6PD-deficient Negroes, in whom the continued administration of primaquine results in a younger population of erythrocytes with a higher G6PD level (Kellermeyer et al., 1961). Other studies among patients with favism indicated a fall in GSH during haemolytic crises, but a later rise of GSH (Panizon & Pujatti, 1958). Genetics of G6PD deficiency The G6PD deficiency is clearly inherited as a sexlinked trait. This has been demonstrated in mammals other than man (Trujillo et al, 1965}. Marked variability occurs in the phenotypic expression of the female heterozygotes (Beutler, 1968). This variability has been explained by the observation that one X chromosome is euchromatic, whereas the other is heterochromatic and genetically inactive (Ohno et al., 1959), Since the pattern of X-linked inheritance tPrimaquine methoxyquinoline. 8-[(4-amino-1-methy]buty])amino]-6-

precludes the possibility that an inactive X chromosome was derived either from the paternal X or the maternal X exclusively, it would be necessary to assume the existence of mosaicism in which the paternal X was inactive in some cells and the maternal X in others (Beutler et al, 1962). In effect, the female heterozygote possesses two populations of erythrocytes. Techniques for detecting such heterogeneous populations of cells have been developed (Beutler et al., 1963) and heterozygous females have been shown to have a varying pattern of erythrocyte staining reflecting such heterogeneity; all the erythrocytes of homozygous subjects fail to stain (Caruso et al., 1967), The varying proportion of G6PD-deficient cells might then account for the very variable expression of disease among heterozygous females, Several forms of G6PD exist in human populations. The pattern of deficiency has been thought to correspond to the distribution of malaria caused by Plasmedium falciparum, Although this hypothesis is still in dispute (Kidson & Gorman, 1962) many studies support it (Choremis et al., (962; Siniscalco et al., 1961). In Sardinia, the variations in gene frequency for G6PD deficiency are directly related to the variations in the prevalence of malaria (Siniscalco et al., 1961), The high frequency (2155) of G6PD deficiency among Kuwaitis, in the absence of falciparum malaria, has been attributed to the high degree of intermingling in the Persian Gulf area of groups (1) glucose-6-phosphate (2) 6-phosphoglucomic acid G6PD TPN 3 from areas where falciparum malaria is highly endemic, e.g., Iraq and Iran (Shaker et al., 1966). G6PD and haemolysis The mechanism of haemolysis in individuals exposed to FV. fava or to certain drugs was elaborated by Sansone & Segni (1956), who showed that a lowered reduced glutathione (GSH) level was found in patients who had had favism or acute druginduced haemolytic anaemia. Carson et al, (1956) showed that the erythrocytes of individuals with a primaquine-sensitive haemolytic anaemia had low levels of G6PD, These two characteristics of the defictent cells are related to one another and to the further decrease of GSH after incubation with 1acetyl-2-phenylhydrazine (Zinham et al., 1958). The administration of primaquine or F. fava to individuals deficient in G6PD results in a fall in the GSH levels of their erythrocytes before haemolysis develops (Szeinberg et al, 1957), GSH makes up 95% of the reduced nonprotein mercapto compound in the erythrocyte and is thought to be important for the integrity of the cell (Fegler, 1952; Benesch & Benesch, 1954; Klebanoff, 1957). The role of the G6PD deficiency in the haemolytic disorder appears to be related to the important role of the enzymes in the pentose phosphate pathway, which is the only means by which glucose may be metabolized by mature erythrocytes. The following reactions generate reduced triphosphopyridine nucleotide (TPNH}: . 6-phosphogluconic acid -- TPNH phosphogluconate dehydrogenase (1.1.1.4) TPN Without G6PD, TPNH will not be formed. The TPNH ıs necessary with glutathione reductase (1,642) in the reduction of oxidized glutathione (GSSG): in G6PD-deficient cells the level of GSSG is increased (Srivastava & Beutler, 1968). Substances capable of oxidizing TPNH or GSH might then be expected to cause haemolysis. The defect in the erythrocytes per se may not be the only host-determinant of disease. Panizon & Yullo (1961) demonstrated that radioactive-chroribose-5-(dihydrogenphosphate) — TPNH favism after the ingestion of fava bean extract. Others have failed to demonstrate any haemolysis of such cells in normal recipients (Greenberg & Wong, 1961). Among the factors that may explain the erratic pattern af the disease and the difference between it and drug-induced haemolysis are (1) the toxic factor in the fava bean may have to be activated in the host or in the presence or absence of other dietary factors; (2) the intake of the toxic factor may vary although bean consumption may be constant; and (3) the mium-tagged G6PD-deficient cells transfused into a expression of disease may also be a function of normal individual lysed more slowly than the same cells transfused into a patient who hadrecovered from another enzyme system, which may be either genetically directed or nutritionally affected.

INCIDENCE, DISTRIBUTION, AND MORTALITY The reported incidence of favism in the Middle East varies considerably from one country to another and within each country, and depends on the distribution of the genetic defect, the presence of the fava bean in the local diet, and the availability and utilization of medical facilities. An incidence of 5 cases per 1 000 population has been estimated in Sardinia (Crosby, 1956). In 1965, the incidence in Guilan and Mazanderan, Iran, was 6.39 and 2.23 per 10000 population, respectively (Lapeysonnie & The annual variation of cases in Mazanderan and Guilan is illustrated in Fig, 1, The decline in 1964 appears to have been real, since the reporting system was in operation in 1963; so does the rise in 1965, The variation has in part been attributed to the availability of the local F. fara crop, which was poor in 1964 and 1966 in Guilan. In Khuzestan, although few cases of favism are reported, a decline from 20 cases in 1968 to 8 in 1969 paralleled the poor crop in 1969, Keyhan, 1966); in 1969 the incidence was 1.85 and 0.65, respectively (unpublished data). In the Rasht area of Guilan the 1965 incidence was 9,27 per 10 000. In view of the local awareness of the problem of favism, the availability of medical facilities, and the existence of a specific reporting system for the disease, the data from these Iranian estans (provinces) appear reliable and probably approximate to the true situations. Although other data on incidence are lacking, favism is known to be widespread in the Mediterranean and Middle East regions. The disease is frequently encountered in Algeria (Messerschmitt et al., 1967), Bulgaria (Angelov & Andrev, 1959), China (Chung, 1965), Cyprus (Joannides, 1952), Iran (Lapeysonie & Keyhan, 1965), and Lebanon (Shahid, 1960). Sporadic cases have been reported from the Federal Republic of Germany {Gehrmann et al, 1966), France (Auquier et al, 1968), Poland (Rokicka-Milewska et al, 1968), Romania (Schneer et al, 1966), Yugoslavia (VinceRibarié, 1962), and the USA (Wharton & Diesselmann, 1947; Burka et al., 1966). No large series of cases of favism have been reported from Egypt but the disease is common, accounting for approximately 1-29 of paediatric hospital admissions in Cairo and Alexandria (M. Gabr & M. Khalil, personal communication). On the basis of the number of hospital admissions of patients with favism to two of the paediatric services, the minimum incidence is 0.4 cases per 10 000 population per year. In Tunisia, in the past 8 years, fewer than one patient a year has been admitted to the children’s hospital with a diagnosis of favism (A. B. Hamza, personal communication). In the areas where the genetic defect is prevalent and the fava bean is a major food staple, only 10-20% of the population at risk ever appear to develop the disease. Kattamis et al. (1969) noted a history of favism in 10,3% of 223 G6PD-deficient males; Siniscalco et al. (1961) noted that 20% of 121 G6PD-deficient males had had episodes of favism. 7 sr i \ —— Gulan H — Mazanderan i BE Ss. \ | | 3 1 Zal Î = f - | sr i 8 a a \ x \\ 27 7 J i i =| v4 \ 12! 1963 = Î AI 4 u I 1964 | 1965 i 1966 Year | 1967 i 1968 1968 WHO 36090 Fig. 1. Incidence of favism in Guilan and Mazanderan, lran, 1963-69. Seasonal distribution Wherever favisin 1s endemic, the disease generally has a characteristic seasonal distribution corresponding to the harvesting of fresh fava beans (Crosby, 1956; Chung, 1965; Lapeyssonie & Keyhan, 1966; Kattamis et al., 1969), although most large series include cases during other seasons. The peak incidence of cases in Iran also occurs in the harvest season: March and April in the south-eastern area, May in Mazanderan, and 3 weeks later in Guilan. In Greece 42% of cases occur in May (Kattamis et al, 1969); in Bulgaria, 90% in June {Angelov & Andrev, 1959); in Cyprus, 57% in April and May (about 25% were seen in January to March, when small quantities of fresh beans are available) (Joannides, 1952); and in Kwangtung, China, 95%

between 11 March and 10 April, 50°% of them in the last 10 days of March (Chung, 1965). The direct relationship beween cases and the availability of beans is seen in the shift of the occurrence of cases by time of onset in Guilan in 1965, 1968, and 1969 (Fig. 2). These changes correspond to the shift in time of the fava bean harvest; in 1969, owing to floods and a long winter, the beans were planted late and ripened even later than usual (W. Kaiser, personal communication). 5 possible that the second seasonal increase in cases in November corresponds to the time of marketing of fava beans from Manuf that have been stored in a way peculiar to Egypt. In the areas where cases are related to the locally harvested crop, most occur within 3-4 weeks. The distribution of cases in time resembles that of an infecnous disease, 1.e,, the sudden introduction of an agent into a susceptible population leads to a rapid appearance of cases, followed by a less rapid decline, Age distribution BO | [=] I w Percentage of casas —— Guilan (total 338) — Mazanderan {total 182} | fi) { <1 I 1-4 | L 5-9 10-14 Age (years) I | 15-19 20+ AMO 100 Fig. 2. Occurrence of favism in Guilan (1965, 1968, and 1969). A unique seasonal pattern of disease occurs in Egypt. There are two seasonal peaks, and cases are not frequent in the spring until several weeks after the fresh beans are ripe (W. Gabr, personal communications). Analysis of 30 cases showed that 14 occurred from May to July and 10 from November to January. Although not documented, a similar pattern was noted in Alexandria {N. Hashem, personal communication}. The onset of cases in Mav and June corresponds to the time of harvesting the dry bean, most cases being related to the consumption of stewed dried fava beans (medamass). It is Favism is generally a paediatric disease; however, even within the paediatric age range there is a variation in the age distribution. which probably reflects the varying risk of exposure ta F. fava in different cultures and host factors that may enhance or inhibit expression of the disease. In Greece, 65% of the cases occurıed in the 25-year age group, 7.2% in the 10-15-year group, and 5,5% in infants (Kattamis et al, 1969). Similar patterns have been noted in Bulgaria (Angelov & Andrey, 1959), Lebanon (Shahid, 1960), Cyprus (Joannides, 1952) and parts of Iran (Lapeyssonie & Keyhan, 1966). There 15 a consistently larger percentage of cases in the 14-year age group in Guilan than in Mazandcran (Fig. 2), Several of the previously noted sources in Egypt report (unpublished data) a greater proportion of cases in children aged 0-1 and 1-4 years. In one hospital where 85 cases are seen each year, 50% of the patients were under 1 and 95% under 5 years of age. The occurrence of disease at these apes probably reflects the consumption pattern of V. fava in Egypt. Kattamis et al. (1969) reported one case in a 45day-old infant; Chung (1965) noted that the youngest of his 1464 patients was 6 months old, and Emanuel & Schoenfeld (1961) reported a case in a dmonth-old nursing infant. One of the oldesi patients who developed favism for the first time was a 67year-old woman (Faiola et al. 1969). Occurrence af cases by sex Favism is a disease to which male children are predisposed. the ratio of male to female cases ranged from 21:1 in Cyprus (Joannides, 1952) to 2,7: lin Mazanderan (Laperssonie& Keyhan, 1966) (Table 1). These facts are explicable on the hypothesis thar G6PD deficiency is a sex-linked trait with marked variability for phenotypic expression in the heterozygous female. The 1atio of male to female cases would then be expected io be slightly lower than the calculated ratio of males with G6PD defi-

© homozygote (1) 3 gane/ /? cases d cases(2) Ratio % Egypt (average) 334 14 24 Cyprus 32 21.3 1.5 Athens (average) 120 6.2 19.4 Fars 12.5 3.9 32 Persian Gulf 14.36 70 21 The consistent pattern of a greater than expected number of fernale cases, based on a ratio of male gene frequency to female homozygote deficiency {column 3 of Table 1) greater than 1.0, is not observed in the Caspian region of Iran, Either disease expression is more likely in males in this region or there is a markedly different exposure pattern by sex. The distribution of cases by age and sex suggests a changing pattern of either exposure or disease expression among females (Table 2). The ratio of male to female cases is lower in those aged under 10 years Guilan 239 3.38 71 than in those aged 10 years or over. The difference Mazanderan 2.39 270 .89 Table 1. Relationship between the ratio male gene: female homozygote and the ratio male cases : female cases in selected areas a Area . could reflect cultural patterns in respect of cooking, tasting, and other eating habits or biological changes affecting expression of the disease. The severity of disease, as judged by the severity of the anaemia, does not appear to be sex-dependent. ciency genes to female homozygotes for this gene because there would be some disease expression in female heterozygotes. For example, the calculated ratio! in Shiraz, Iran, would be 12.5: 1; the actual ratio of cases is approximately 4 : 1. These findings can be explained by any {or a combination) of the following hypotheses: (1) there are two female heterozygotes per female homozygote developing disease, (2) relative to female homozygotes only one third of all males with G6PD deficiency are at risk of developing disease, or (3) the risk of exposure to the bean is markedly less in the males than in the homozygous females. Most reports of large series of patients with favism. do not include a breakdown of the data according to the occurrence of cases in urban and rural areas. In Bulgaria cases are more frequent in rural areas (Angelov & Andrev, 1959), though specific rates are not reported; nor are they included in the extensive reports from China, Cyprus, Greece, Iran, or Italy. The cases reported from Egypt are nearly all from the urban areas of Cairo and Alexandria; no data exist on the rural distribution. " Calculated on the basis of 2g (1—g) + g° where g is The mortality from favism varies from area to area: in Sardinia it has been reported to be as high as the male gene frequency; g* represents the female homozygote. Urban-rural distribution Mortality Table 2. Distribution of cases of favism by sex and age in selected areas! Age 2:10 years Age <10 years Area Rasht Year No. of cases Rato & 8 No. of cases Ratio & é 3 1958-62 530 95 5.6 22 14 16. 1965 490 154 3.2 61 37 1.6 1968 243 68 36 19 7 27 1965 234 75 3.1 38 14 2.7 +968 106 40 2.6 22 15 15 È 8 (2 hospitals in Guilan) Guilan Mazanderan 1 Source: Lapeyssonie & Keyhan (1966), excopt for the 1968 data for Guilan and Mazanderan, which were provided by the Ministry of Health of Iran,

2 per 10 000 population (Crosby, 1956); in Tunisia there are few cases and practically no deaths (B. Hamza, personal communications), In Kwangtung the case mortality was 2,3% (Chung, 1965) and in Bulgaria it was 2,1% (Angelov & Andrey, 1959), Among Joannides’ (1952) patients, 3 of 67 died. In the Caspian region in 1965 the case mortality was 12% in Guilan and 1.8% in Mazanderan (Lapeyssonic & Keyhan, 1966}. In 1968 in Guilan only 1 death was recorded among 338 patients. Mortality is a function of age; in Guilan in 1965, the case mortality rate was 4.7 % for children less than 2 years old and 1.11% for those aged 2-4 years; among children aged 5-16 there were no deaths (Lapeyssome & Keyhan, 1966), The broad bean ts planted throughout the Middle East between October and December; only one crop a year is produced. The plant usuaily flowers 40-50 days after planting; few, if any, cases of favism are noted during the months of flowering. In a given region, fresh beans are harvested over a period of approximately 4-5 weeks, a period that corresponds to the usual interval of the epidemie curve of favism cases in each local area. The shifts in the epidemic curves of cases in Guilan correspond to the variations in the time of planting and harvesting (Fig. 3). 220 — 200 |- ' 180 7 Number of cases THE TAVA BEAN: PRODUCTION, CONSUMPTION, AND TOXIC FACTORS Production 240 140 - (estimated) © TT 1965 120 | — In Egypt approximately 5% of the dried bean crop is stored in such a way that the beans retain 10 all outward appearances the characteristics of fresh dried beans, from which they are indistinguishable in taste. It is possible that the introduction of these beans into the local markets in October and November accounts for the second peak in cases of favism in Cairo and Alexandria. Disease of F. fara, such as chocolate spot disease, caused by the fungus Botrytis fava, or rust, cannot be responsible for the production of the toxic factor in F. fava, since the years in which these diseases have been common have been those with a marked decrease in the number of cases of favism in both the Caspian and the Khuzestan regions of Iran (Kaiser et al., 1967, 1968). Consumption Form associated with acute haemolysis. The predominant type of exposure associated with favism 18 the ingestion of fresh beans at the time of harvest, WHO 30092 May June July Month and week Fig. 3 Age distribution of cases of favism in Guilan and Mazanderar, 1968 (data from the Department of Health, Mazanderan and Guilan). although dry beans may also be implicated (Gasbarrini, 1915: Kattamis et al. 1969; Chung, 1965). In addition, cases of favism have been well documented in suckling infants whose unaffected mothers had eaten fava beans (Chung, 1965; Angelov & Andrev, 1959; Kattamis et al., 1969; Emanuel & Schoenfeld, 1961}; such cases have been noted in the Caspian

Table 3. Form of V. fava contact associated with favism ! Area and year Breast fed Rite wh alin Eaten raw Ellen, Guilan, 1967 90 (14.5) 93 (15.0) 90 (145) 186 (288) 171 (27.6) Eaten, form Total 9 (1.5) 619 Mazanderan 1965 2 (08) 47 (18.1) 15 (8.8) 181 (699) 14 (54) 0 259 1968 1 (0,5) 37 (16.9) 6 (27) 154 (70,3) (9.6) o 219 1969 107 28 (20.1) 0 96 (891) 14 (101) 0 139 Fars, 1968-69 349 109 0 8 (13.1) 49 (802) 0 61 21 1 The table gives the number of cases and, in parentheses, the proportion of total number of cases associated with each form of contact. Source: Departments of Health Guilan, Mazanderan, and Fars {unpublished data). 2 The mothers had eaten fava beans. region of Iran (Table 3}. Luisada (1941) cites one case that was possibly associated with the ingestion of milk from a goat that was fed on fava beans. Thus, whatever the toxic factor, it is capable of crossing from the mother into breast milk. Although many authors indicate that cases are associated with inhalation of the pollen of the F. fava flower, most of them refer to the early papers of Gasbarrini (1915) (Luisada, 1941; Wharton & Duesselman, 1947). Recent reports on large series of patients fail to confirm the occurrence of pollenproduced cases of favism (Katiamis et al, 1969; Chung, 1965). In Guilan and Mazanderan associated cases have been claimed (Table 3), yet no reported cases have occurred in January or February, when the flowers are in bloom. The few cases reporied in Cyprus and in Shiraz, Iran, during December and January may have this cause, or may be cases of haemolytic anaemia related to drugs or infection. Fresh beans accounted for 68% of the cases studied by Kattamis (1969), but the report did not differentiate the forms in which the fresh beans were consumed. In the Caspian region fresh beans account for nearly all the cases; raw beans account for 26-70%, of them and cooked beans for 5-27% (Table 2). In Egypt nearly all cases are associated with dried stewed beans, almost the only form in which they are consumed. The form in which the bean is ingested does not affect the time interval between ingestion and the onset of symptoms (Kattamis et al, 1969), suggesting that, whatever the toxic factor, it is present in the fresh and in the dried beans in a similar form. No data exist relating the risk of favism to the form in which the bean ts consumed. Table 4 shows the attack rates of favism associated with the diflerent forms of exposure. Since the table is based on the total population and an assumed equal risk of Table 4. Attack rate of favism associated with different forms of contact (per 10 000 population) Area and year Guilan 1967 Breast-fed infants Passage Contact through with pollen 1} favatield or flower Eaten raw Eaten cacked Eaten, form : unknown Tota! 464 066 068 0.66 1.22 125 0.07 Mazanderan 1965 001 0.26 0.08 0.98 0.08 a 1.41 1969 0 095 020 0.03 064 011 Ô 0 99 1969 0.005 0.15 û 652

exposure for the different forms, it is a gross simplification. However, such a simplification 1s appropriate when the patterns of bean consumption and disease in Guilan and Mazanderan are compared (Tables 3 and 4). The higher attack rate in Guilan is due to exposure of breast-fed infants, partially to pollen (which is already questioned), and to the consumption of cooked beans, which on the basis of the seasonal pattern of disease and the consumption pattern would be fresh cooked beans (Table 4). The consumption of cooked fava beans is similar in Mazanderan and in Guilan, and the prevalence of the G6PD deficiency is the same, yet the disease is more frequent in Guilan. The explanation may lie in different folk attitudes and the manner in which the bean is prepared and consumed in the two areas, particularly since the attack rates based on raw bean ingestion are similar (assuming equal risk of exposure). Consumption patterns. The daily consumption of the fava bean in rural areas of Eevpt, Iran, and Tunisia is shown in Tables 5 and 6. Consumption is greatest in the urban areas of Egypt, where up to 46 g per day may be eaten by pregnant and lactating women {Abdou & Amer, 1965) and where there is little seasonal variation; the mean daily consumption is 28 g (Abdou, personal communication), The pattern of consumption in rural Egypt appears to resemble that in Iran (Iranian Statistical Center, 19657) and Tunisia (personal communications), In Iran the pattern varies markedly from one region to another and is in part dependent on the pattern of bean production. The data in Table 5 understate consumption in some areas since only Table 5. Seasonal consumption of fava and other beans (g per person per day) in Egypt and Iran Country and type of bean Winter Spring Sum. Autumn Egypt pulses 1 7 19 2 12 0.4 48 13 0 split dried beans 0.5 o a3 04 dried beans & chick peas 2 718 265 485 7.50 Iran (rural areas) fresh beans 9 Table 6. Consumption of fava beans (g per person per day) in Tunisia Area Fresh beans Dried beans 6 rural areas © north 1B central & south 10 average 14 large cities 167 7,4 villages 14.2 49 dispersed settlements average 45 131 3.6 147 5.4 Villages and dispersed settlements. about one-third of the villages sampled in the survey (Iranian Statistical Center, 19657} represented areas of heavy bean production and consumption: in many villages the bean was neither produced nor consumed, In Tunisia, where favism is rare, the per caput consumption of fava beans, particularly fresh beans, appears to exceed even that of Iran, where favism is common. The pattern varics in Tunisia; increasing consumption is associated with increasing urbanization and proximity to the growing areas. The implications are that in Tunisia (1) there is a lower frequency of G6PD deficiency (probable), (2) the bean is Jess toxic (possible), or (3) the preparation of the bean minimizes any toxic factor present. Different vaneties of fava bean are consumed in Egypt, Iran, and Tunisia, but in the medical literature the variety associated with cases of favism has never been specified. One problem in differentiating the species of F fava is that, among the pulses, F, fava is the only one that crass-pollenates between species (W. Kaiser, personal communication). Pure strains may be difficult to find in areas where more than one stratn is grown. In Egypt three different strains are consumed, and they are often eaten together in the dried form in the urban areas. In Iran there are three main strains; in addition, the Algerian strain has recently been introduced into several areas, particularly parts of Khuzestan, in order to increase the yield and attempt to decrease susceptibility to various F. fava diseases. 1 Data from 560 households (2800 persons). Whether the pulses were dried or fresh was not specified, 2 Separate data not available. In Tunisia there are two main strains; a third is produced but is fed only to animals. The form in which the bean is eaten varies

throughout the Middle East and North Africa. In the clinical descriptions and large series of cases, reference is made only to whether the beans associated with disease were raw or cooked and, if the latter, whether they were fresh or dried. The raw bean is consumed with or without the skin and rarely, if ever, with the pod. In most areas the skin of the raw fava bean is tough and inedible, yet in Guilan the raw bean is frequently eaten with the skin, particularly by younger children. Fresh beans, whether raw or cooked, are not considered an appropriate food for children in Egypt. In rural areas of Iran and Tunisia, raw beans are frequently consumed, particularly by children. The fresh fava bean is generally cooked in water, in or out of the pod; in the latter case, the skin may or may not be removed. Only in Guilan and occasionally in Mazanderan is the cooked bean eaten with the skin. In Tunisia, large fresh beans are always eaten without the skin, Fresh cooked beans are not commonly eaten in Egypt, but are widely eaten in Tran and Tunisia when in season. In Iran the fresh beans may be consumed without the skin though this is more common with the dried beans. The dried beans are eaten in many forms. Stewed dried beans form a main staple in the Egyptian diet and are often fed to infants; the skins are not removed when the beans are eaten. Most cases of favism in Egypt have been associated with this form skin; in other areas the eating of the raw bean alone is associated with the disease. A series of prohtbitions has evolved against the consumption of the fava bean by certain persons or under certain circumstances. In Tran there is an attitude that women should not eat the fava bean for at least 30-45 days after the delivery of a child. Many of the prohibitions cover other foods, but these patterns are not as constant as that relating to the fava bean. Villagers in most of the areas where the bean is eaten consider that it should not be given to young children and infants, especially when it is raw. In the Caspian region, and especially in areas of Guilan, this view is not particularly strong, In the Mazanderan area consumption of the skin, whether cooked or raw, is often thought to be dangerous for children; there is no such recognition of any inherent danger in the Rasht area of Guilan. Outside the Caspian region, the consumption of the raw fava bean is frowned upon and it is thought that only young children, particularly those who are helping in the fields, eat the raw bean. In Egypt there is a strong feeling against the consumption of the fresh fava bean by children. Folk remedies for the prevention or treatment of the disease have evolved in three widely separated areas, in Guilan and Fars in fran and in Tunisia. The consumption of large amounts of sweets in the form of grape juice, honey, dates, or sugar water is of bean, The dried beans may also be soaked in water advocated in these regions both to prevent and to overnight, the skins removed, and the beans incorporated into stew-like dishes, street vendors or in eating places in both Iran and cure the disease. The cure rate for this type of therapy has been reputed to be about 50% in both Guilan and Fars. This therapy is thought to be fairly effective in Tunisia, where few such patients ever Turusia. In Iran the dried beans are cooked in their come to medical attention. In Rasht, where all skins, but the latter are peeled off when the beans are eaten. Older children and adults are the principal consumers of these beans, the consumption of which is greatest during winter. In the Caspian region during winter, dried beans are soaked overnight; the skins are then removed classes consume the fava bean, the lower-class peasants look upon the disease as being imposed on them because they are too poor to afford the sweets and meats of the upper-class diet. Boiled dried beans are frequently purchased from and the beans cooked for a few hours, In Rasht and the surrounding areas the soaked dried beans may be eaten uncooked (without the skin). Tn much of Iran, wheat flour with added fava bean flour is used. Individual bakers, particularly in the Caspian region, may also use fava bean flour. Folk attitudes. In areas where the disease or its symptom complex is recognized by villagers the association with the fava bean has usually been made, In some areas villagers feel the disease is associated with eating the raw bean together with the Weaning practices and consumption by infants. In Egypt a large proportion of infants are exposed lo the fava bean before the age of 2 years (Abdou et al., 1965); in Iran and Tunisia only a small proportion of such infants are so exposed. Weaning usually takes place at about 6 months of age, although frequently, particularly in the rural areas, mothers will continue breast feeding at least through the first year and often into the second year of a child’s life, In urban areas of Egypt, 14% of the infants between 6 and 12 months old and 47% of those aged 12-14 months eat the fava bean (Abdou et al., 1965}. There are no precise data on the food-consumption patterns of

infants and children in Iran. In Tunisia, an infant the with is it unless beans fava get to ly unlike at family’s food; at 6-9 months of age only 11%, and food. family the eat infants the of 9-12 months 24%, The development of a weaning food containing Y. fava would appear to be a potential health hazard to infants with the G6PD deficiency. If it were to be 11 the incubation of G6PD-deficient erythrocytes with al pyrimidine aglycon of vicine and divicine is identic with cells such of tion incuba to that produced by the are found APH (Mager et al., 1965). The pyrimidines may arise as their glycosides in W. fava. The aglycons either in es, osidas B-glyc of from the hydrolytic action 1965). al, ct r (Mage tract ive digest the in or the bean of both The action of isourarmi in reducing the GSH potennormal and GéPD-deficient erythrocytes is ted inhibi is and ate ascorb or DOPA either by tiated er Kosow & er by glucose (Razin et al., 1968). Kosow with DOPA of tion incuba that ed (1967) demonstrat loss G6PD-deficient erythrocytes resulted in a sharp cells l norma in GSH of GSH and that oxidation of took place only in the absence of glucose. vivo, it If these in vitro models are relevant in ial potent of n natio becomes obvious that a combi sion, expres e diseas for ary necess be may toxic factors haemoj.e., both vicine and DOPA may produce the for ements requir le possib lysis of erythrocytes. Other lic metabo ated associ an be may sion such expres ycaemia, state, i.e, the presence or absence of hypogl as the such s, system e and the status of other enzym 2), (1.6.4. ase reduct hione elutat and es B-glycosidas as asand the presence of other substances, such preincorporated in such a food, the bean would dried of risk e relativ sumably have to be dried. The on {as compared to fresh) beans is not known, but In the basis of experience in Egypt the risk, is real. a have en childr young that most areas it is evident not is it but sion, expres e diseas of ility probab greater known whether this is determined biologically (e., stanwhether it is directly related to age) or circum , manner wn unkno an in ion ingest by tially (ie, that consumption of the skin, etc.). It is possible more exposure at an earlier age might result in even the age, r younge a at r greate is ity mortal Since cases. introduction of the fava bean into the infant’s diet would have a major effect on infant health. is The only way of diminishing the risk of favism then and bean to identify the toxic factors in the strain determine whether they can be removed by bean the which in way the ing modify by or on selecti corbic acid, in the diet. is prepared for use in à weaning food. of It has been demonstrated that the browning ed Toxic factors. The substances that have receiv is related to the fi-p-glucoside of DOPA, fava F. in coat, ihe greatest attention as the possible toxic factor which 1s located almost exclusively in the skin opytap-gluc -5-(fiamina [2,6-d t in presen and don favism are vicine cotyle the from absent being iamino (2,6-d ne divici and the dinol] of pyrimi tissue xy)-4green nosylo only small amounts in the inner 1961). al., 4,5-pyrimidinediol) (Razin et al., 1968; Lin & Ling, et sawa (Naga pod and the hilum ) skin 1962) and 3-(3,4-dihydroxyphenyDalanine (DOPA Broad beans when fresh have a green to buff Vi1967). er, Kosow & wer s (Koso darken ide slawly glycos seed its and and a buff to green interior. The ric reddark is year a cine, divieine, and isouramil (6-aminoisobarbitu after and ion because of oxidat have light) acid), another constituent of the fava bean, brown. Storage in the absence of oxygen (and in been shown to oxidize reduced glutathione (GSH) retards this effect. cells G6PD-deficient erythrocytes but not in normal 1968; DISCUSSION (Lin & Ling, 1962: Lin. 1963; Razin et al, ed Mager et al, 1965). Vicine has also been isolat a Throughout the Middle East favism constitutes but from beet juice and peas (Donoso et al., 1969), of the problem extent The m. proble health major ytic these have not been associated with acute hacmol ency depends on the prevalence of the G6PD defici anaemia. conis bean fava the whom by and how and on by On aging, vicine becomes brown, presumably has sumed. Only within the Caspian region of Iran 1965), al, et r (Mage y activit of loss a with ion, oxidat respon health public a as ed the problem been accept When incubated with normal erythrocytes in vitro, probThe iated. apprec tude magni full its and y sibilit & vicine inhibited 38% of the activity of GEPD (Lin resources lem could be aggravated, and many health Ling, 1962). of paement manag and ent treatm strained in the in The fava-bean pyrimidines can oxidize GSH the diet into uced introd were bean fava the af tients, henylpure solution, in contrast to L-acetyl-2-p of weaning infants without removal or neutralization e the hydrazine (APH) and primaquine, which requir of the toxic factor. patpresence of erythrocytes or haemoglobin. The The association of favism with fresh raw and the from ing result bance distur tern of metabo

cooked fava beans is well documented throu ghout the Middle East. The onset of cases invariably starts following initial exposure to the bean with the skin may not develop it when they subsequent ly consume with the harvesting of fresh beans and their appearthe bean without the skin. ance in the market. Individual case reports of favism Several factors in the status of the individual may after the ingestion of dry beans are found in Tran and affect disease expression. It is possible that the host's other countries; such an association is comm on in blood glucose level may affect the occur rence or Egypt. Since the consumption of dried beans is degree of haemolysis. The individual’s pre-existing constant in the four seasons in Egypt, the declin e of haemoglobin level may be a determinan t of the cases in the month of July and their disappearanc e in degree of disease expression. The presence of proAugust may be a result of a decrease of the toxic tein-calorie malnutrition might affect other enzyme factor in the bean. This appears to coincide with the systems and thus interfere with a child’ s reserve normal darkening of the bean, i.e, the breakdown of ability to reduce oxidized glutathione. DOPA or vicine or both. Although antibodies to F. Java have been demon The association of cases and of their frequency in strated clinically and experimentally (Kant or & areas where the bean is consumed with the skin, Hock, 1966), hypersensitivity to a factor in the fava whether raw or cooked, appears to indicate that at bean does not appear to be related to the developleast one toxic component of the bean is locate d in ment of the disease, since most of the cases in Egypt the skin. All the toxic components would have to be and a large proportion of those in the Caspi an region heat-stable, as disease is frequently associated with are associated with the first exposure of the infant to fresh or cooked dried beans. Furthermore, the toxic the bean, factors are capable of passing into breast milk. The risk of contracting favism decreases with age. The occurrence of favism may be erratic within a Individuals who have experienced an episode of population, a family, and even an individual. Indifavism will frequently abstain from further consumpviduals are frequently exposed to the fava bean many tion of the fava bean. Individuals who suffer mild times before they develop the disease, or they may attacks are known to have them repeatediy over develop it from initial but not from subseq uent Many years, indicating that desensitizat ion has not exposure. Part or much of this erratic pattern may be occurred. The form in which the bean is consumed is attributable to the form in which the bean is eaten likely to change with age. In most areas, even though and the season when it is consumed; in Egypt, for some children may consume raw or cooked fava example, disease is unlikely to develop as a result of beans with the skin on, it is the gener al practice for either initial or subsequent exposure during Augus t adults and older children to peel the skin off with and September. In Iran, persons who develop favism their teeth before eating the beans. ACKNOWLEDGEMENTS This study was supported by the UNIC EF/WHO/FAO Protein Advisory Group and was carried out in collaboration with WHO, The author thanks the many individuals in Egypt, Iran, and Tunisia who made the results of their observations available to him. RESUME L'ÉPIDÉMIOLOGIE DU FAVISME Le favisme représente un obstacle potentiel à l’incorporation des fèves (Vicia fava) dans un aliment de sevrage Moyen-Orient et en Afrique du Nord. Dans le présent susceptible d'être produit sur place et à bon compte au article, l’auteur décrit les aspects épidémiologiqu es de l'affection, évalue l'opportunité d'utiliser les fèves comme Composant d'un aliment de sevrage et suggère des moyens propres à neutraliser ou à éliminer les toxiques contenus dans ces légumineuses. facteurs Les observations faites sur le terrain et l'examen des publications consacrées au problème du favisme montrent que l’emploi des fèves dans un aliment de sevrage comporte des risques pour la santé, mais que le danger pourrait être fortement diminué par la sélect ion de certaines variétés de plantes et aussi, notamment, par l’utilisation de graines séchées depuis un temps suffisamment long. Les cas de favisme sont généraleme nt associés dans le temps a la récolte et à la mise sur Je marché des féves fraiches, mais ils surviennent aussi à la suite de la consommation de féves récemment séchées, De l'étud e de la

répartition de la maladie selon l'âge, des modalités de la consommation des féves et des tabous alimentaires locaux, il ressort que les facteurs toxiques, thermostables, sont concentrés dans la cuticule de la graine et qu'ils sont de moins en moins abondants avec le temps dans les fèves séchées. On les retrouve dans le lait des mères allaitantes. Le favisme peut sc manifester à la suite de l’inferaction 13 de différents facteurs individuels comme la carence en glucose-6-phosphate déshydrogénase {G6PD), l'état de uutrition et la consommation d'autres types d'aliments. Ces observations concordent avec les résultats des recherches de laboratoire qui mettent en lumière le rôle de la vicine, de la divicine et de la dihydroxyphénylalanine dans l’étiologie du favisme. REFERENCES Abdou, I. A. & Amer, A. K. (1965) Bell. Nutr. Inst. (Carro), 1, 21 Abdou, I. A. et al, (1965) Bull. Nutr. Inst. {Cairo}, 1. 39 Kantor, 5. Z. & Hoch, D, (1956) Int. Arch, Allergy, 29, 264 Katramis. C. A, et al. (1969) J. med. Genet, 6, 34 Ginek., No. 2, 7 Auquier, L. et al, (1968) Sem, Hôp. Paris, 44, 2037 Kidson. C. & Gorman, J. G. (1962) Nature 196, 49 Angelov, A. & Andrev, L (1959) Wop, Pediat., Akui. i Aykroyd, W.R. & Doughty, G. (1964) Legumes in human nutrition, Rome (FAO Nutritional Studies No. 19) Benesch, R. E. & Benesch, R. (1954) Arch, Biochem., 48, 38 Beutler, E. (1968) The genetics of glucase-6-phosphate dehydrogenase deficiency in hereditary disorders of erythrocyte nietabolism, New York, Grune & Stratton, p. 114 Beutler, E. et al. (1962) Proc, nat. Acad. Sct. (Wash. ), 48, 9 Beutler, E. et al. (1963) In: Proceedings af the Ninth Congress of ihe European Society for Haematology, Lisbou, Basle, Karger, p. 675 Burka, E. R. et al. (1966) Ann. intern. Med, 64, 817 Carson, P. E. ei al. (1956) Science, 124, 484 Caruso, P. el al. (1967) Haematologica, 82, 29 Choremis, C, et al. (1967) Lancer, 1, 17 Chung, 8. F. (1965) Paediat. indones., 5, Suppl, 880 Crosby, W. H. (1956) Blood. 11, 91 Dern, R. J. ct al. (1954) I, Lab. clin, Med., 44, 171 Donoso, G. et al. (1969) Bull, Wid Hlth Org., 40, 513 Emanuel, B. & Schoenfeld. A. (1961) J. Pediat., 58, 263 Faiola, G. et al. (1969) Policlinico. Sez. prat., 76, 65 Kellermeyer, R. W. et al. (1961) J. Lab. clin. Med, 58, 225 (Lond), Klebanoff, S, J, (1957) Biochem. J.. 65, 423 Kosower, N. 5. & Kosower, E, M. (1967) Narure (Lond.), 215, 285 Lapeyssonnie, L. & Keyhan. R. (1966) In: Proceedings of the First Seminar of Farism in Iran, Teheran, Food and Nutrition Institute of Iran, p. 36 Larizza, P. ct al. (1958) Haematologica, 43 205 Lederer, M. (1925) Amer, J. med. Sei, 70, 500 Lin, J. Y. (1963) J. Formosan med, Ass. 62, 777 Lin, JY. & Ling. K. H. (19623 J. Formosan med. Ass., 61, 579 Luisada, A. (1941) Medicine { Baltimore), 20, 229 Mager, J. et al. (1965) Biochem. biophvs. Res. Conmun, 20, 235 Messerschmutt, J. et al. (1967) Nour. Rev. frane, Hémat., 7, 827 Nagasawa, T. et al, (1961) Agric. Biol, Chem. 25, 441 Ohno, 5. et al. (1959) Exp. Cell Res, 18, 415 Panizon, F. & Pujaiti, G. (1958) Acta paediat, lat, 11, 71 Panizon, F, & Vullo, C. (1961) Acta haemar. (Basel), 26, 337 Razin, A. et al (1968) Israel J. med. Sel. 4, 852 Rokicka-Milewska, R. ot al. (1968) Pediat. pol., 43, 621 FAO Committee on Protein Requirements (1957) Report (FAO Nutrirfonal Studies No. 16) Russo, G, & Balsamo, V. (1962) Pediarria ‘Napali}, 70, 337 Fogler, Cr. (1952) Nature (Lond. ), 170, 624 Gasbarrini, A. (1915) Policlinico, Sez. prat. 22, 1505 Gehrnann, J. G. et al. (1963) Disch, med. schr, 88, Sansone, G. & Segni, G. (1956) Ball. Soc. Ital, Biol. sper, 1865 Gelin, P. (1952) Sang, 23, 622 Greenberg, M. 5. & Wong, H {1961) J. Lab. clin. Med, 57, 733 [deo, G. et al. (1965) Rass. med. sarda, 65, 327 iranian Statistical Center (19657) An analysis af results of househald budget survey, Issue No. 4, Teheran loannides, C. C. (1952) Cyprus med. J, 5, 793 Kaiser, W. J. et al. (1967) Plant Dis. Reptr 54, 595 Kaiser, W. J. et al. (1968) Plant Dis. Reper, 52, 68 Kantor, 8. Z. & Arbesman. €. E. (1959) J. Allerer, 30, 114 32, 456 Schneer, J. H. et al. (1966) Wed. interna fBue,), 18, 727 Shahid, M. (1960) Rer. med. Aloy. Or. 17 83 Shaker, Y. et al, (1966) diver. J. hum. Gener., 18, 609 Siniscalca, M, et al. (1961) Nare (Lond.}, 190, 1179 Srivastava, 5. K. & Beutler. E. (1968) Lancer, 2, 23 Szeinberg, A. et al. (1957) Bull. Res. Coun. Israel E, 6, 115 Trujillo, J. M, et al. (1965) Science, 148, 1603 Vince-Ribarié. V. (1962) Lieën. Veen, 84, 151 Wharton. H. J. & Duesselman, W. (1947) New Engl. J. Med., 236. 974 Zannos-Mariolea. L. & Kattamis. C, (1961) Blood, 18, 34 Zinham, W. H. et al. (1958) Bull, Johns Hopk. Hosp., 102,