Tobacco Documents Online

The Saccharin case Prof. Cliford 1. Chappel

BLADDER CANCER IN RATS FED SODIUM SACCHARIN - MECHANISTIC DATA AND THEIR APPLICATION IN RISK ANALYSIS Clifford I. Chappel Oakville, Ontario, Canada and Samuel M. Cohen Department of Pathology and Microbiology and Eppley Institute University of Nebraska Medical Center Omaha, NE, USA Saccharin, first synthesized in 1879 by Remsen and Fahlberg, has been in use as a non-nutritive sweetener since 1890. Although the use of saccharin in foods has at~ times been controversial (Arnold, 1983), the first indications of a potential safety problem associated with this additive were reported inthe early 1970's. Two-generation lifetime feeding, studies conducted at that time showed that, in the rat, high dietary levels of sodium saccharin (NaS) were associated with, an increased incidence of bladder tumors (Taylor et al., 1980; Tisdel et al., 1974,; Arnold et al:, 1980; Schoenig et al., 1985). As a result of the sustained interest in saccharin by scientists, regulators, the food and pharmaceutical industries, politicians, press and the public, it has been one of the most intensively studied of all compounds of toxicological interest. The extensive data base on saccharin covers biochemistry, physiology, pharmacokinetics, metabolism and toxicology in both animals and humans andlprovides a sound scientific basis upon which to judge the safety of saccharin for human consumption. During the past 20 years the results of numerous safety studies of saccharin have become available, including a two-generation study in the rat which provided previously unavailable information regarding the dose-response relationship between saccharin intake and bladder tumor development (Schoenig et al., 1985). These and other studies that have elucidated the mechanism of action of saccharin and provided insight into the relevance of the observed tumors in the rat to humans have been the subject of several recentreviews (Weisburger, 1988; Ellwein and Cohen, 1990; Chappel, 1992). The purpose of this paper is to provide an overview of the safety data for saccharin. Particular emphasis is given to studies that throw light on the relevance of the findings in animals for humans and those that are critical to a risk analysis. I

Studies of Carcinogenicity in Laboratory Animals Saccharin or its sodium or calciuml salts have been the subject of numerous carcinogenicity bioassays in animals: These include 12 one-generation studies in the rat, five one-generatiom studies in the mouse and one-generation studies in the hamster and monkey. With one exception, one-generation studies of NaS in the rat did not cause significant increases in tumors or other toxicological effects in males or females. The results of one-generation studies in the mouse, hamster and non-human primates were also negative. In contrast to the results of the one-generation studies in the rat, a consistent finding of four two-generation studies was a significant increase in the incidence of bladder tumors in male rats fed 3% NaS or more (Table 1). The most recent of the two-generationi bioassays of saccharin defined~ the dose-response relationship for second-generation male rats using an, unbalanced study, design with very large numbers of animals at the lowest dose levels (Schoenig et al., 1985). This study, which was one of:the largest two-generation bioassays ever undertaken, showed a steep dose-response curve with an incidence of bladder tumors similar to controls at a calculated level of 2.7% in the diet (Carlborg, 1985). The no observed' effect level (NOEL) in the study was 1% NaS in the diet, while at the 3% dietary level the effect of NaS on bladder tumor incidence was considered to be marginal (Schoenig et al., 1985) or not statistically significant (Squire, 1i985). This study demonstrated that in utero exposure of rats to NaS was not necessary for bladder tumor development. Male rats fed 5% NaS from birth had an incidence of bladder tumors identical to that of rats fed the same dietary level prior to conception, in utero, and throughout life. A comparison of the results of the one-generation~ and two-generation bioassays in the rat makes it clear that~ exposure in the neonatal period is an essential factor in: the development of bladder tumors. Tumor development in the bladder in rats fed NaS occurred primarily in the male. Small numbers of malignancies were present in females in some of the studies and none i6 controls, but the incidence was not statistically significant in any of the studies. The results of the various bioassays support the following conclusions: Lifetime ingestion of high doses of NaS beginning in utero or at birth leads to the development of bladder tumors in the rat, predominantly in males. A strong organ specificity is present - tumors have been observed only in the bladder. Tumors were observed only at dietary concentrations equal to or greater than 3%, equivalent to a lifetime exposure of 1.75 g of saccharin/kg bw/day. The dose response curve for NaS is linear and very steep, suggesting an overloading of physiological systems. 2

The tumor response to NaS is relatively weak. NaS is 10,000 times less potent than~ acetyl aminofluorene. NaS does not cause bladder tumors ia the mouse, 'hamster or monkey. The NOEL for NaS in the male rat was 1% and the calculated no effect level was 2.7% (Carlborg, 1985). ~ Acid saccharin did not cause bladder tumors in the rat or the mouse. Metabolism Saccharin is a strong acid (pKa about 2) that is sparingly soluble in water and which forms highly water-soluble sodium, potassium or calcium salts. Pharmacokinetic studies have shown that it is incompletely absorbed from the gut and rapidly excreted unchanged in the urine in humans or rats. Saccharin does not accumulate in body tissues and studies have shown that it is not concentrated in the bladder compared to other tissues (Sweatmani and Renwick, 1982). There is no evidence of excessive accumulation of saccharin in the bladder in utero or during lactation that would explain the sex specificity of the tumorigenic response or the importance of neonatal exposure. The metabolism of saccharin is not'modified by P450 enzyme inducers (Byard and Goldberg, 1973); Sweatman and Renwick, 1979) and feeding saccharin itself does not~ affect liver P450 enzyme activity (Hasegawa et al., 1984; Ball et al:, 1977). Mutagenicity, Genotoxicity, etc. Saccharin does not bind to bladder DNA in vivo (Lutz and'~ Schlatter, 1977). It is a nucleophilic molecule that is excreted unchanged. Metabolism would be required to transform it to an electrophilic species that could act as an alkylating, genotoxic chemical that would react with DNA. Many mutagenicity and genotoxicity studies of saccharin have been conducted in order to clarify the underlying carcinogenic mechanism. These have been the subject of a number of reviews (JECFA, 1984; IARC, 1980, 1982; Arnold et al., 1983; Ashby, 1985; Arnold and Boyes, 1989). The extensive mutagenicity and genotoxicity tests conducted since the early 1970's have yielded generally negative, but occasionally conflicting results. Saccharin appears to be non- mutagenic in bacterial systems and to be clastogenic or genotoxic only at very high doses or concentrations in short-term in vitro or in vivo test systems. Ashby and Ishidate (1986) found that clastogenicity was not specifically associated with NaS since isotonic solutions of other potassium, calcium and magnesium salts had similar potency. All of the positive responses with NaS have beeni of an exceptionally weak nature, requiring dose levels in the g/kg range, which were up to 1000 fold' higher than the dose levels of known genotoxic carcinogens required too elicit a similar response. 3

Although mutagenic impurities have been detected in commercially produced saccharin, they were present at extremely low levels (IARC, 1980; Herbold, 1981; Riggin et al., 1983). 0- toluenesulfonamide is a notable impurity in saccharini synthesized by the Remsen-Fahlberg process, however, carcinogenicity bioassays of this substance have been negative (Arnold et al., 1980). It is now generally accepted that impurities do not provide an explanation for the bladder tumors in male rats fed high dietary levels of NaS. It is known that protooncogenes may contribute to the development of malignancy when their structure or expression is altered. Ras gene activation has been demonstrated in human urinary tract tumors (Fujita et al., 1984, 1985), and similar activation of the ras gene has been demonstrated in animals treated with the bladder carcinogens, n-butyl-n-(4- hydroxybutyl)nitrosamine (BBN) and N-[4-(5-nitro-2-furyl)-2-thiazolyl]formamide (FANFT). Studies of the expression of the ras gene product, p21, in bladder tumors in rats from long-term studies using immunohistochemical techniques have shown that although immunoreactivity was present in 20-50% or more of the lesions in~ rats fed FANFT or FANFT plus NaS, it was not present in rats fed NaS alone (Masui et al., 1990, 1991). It has been demonstrated that plasma membrane receptors are present on the urothelial surface for phorbol ester tumor promoters (Verma et al., 1985) and epidermal-growth factor (Neal et al., 1985; Momose et al:, 1991)L The possibility that saccharin might bind to urothelial cells has also been investigated using AY-27 transitional cell carcinoma cells (Williamson and Cohen,, 1989, unpublished). No evidence for a: specific celli receptor for saccharin was obtained. This is not surprising given the high concentrations which are required~ for the effects on the urothelium. Based on the results of the animali studies, efforts have been made to determine whether artificial sweetener or saccharin use has been associated with human cancer. The extensive epidemiological literature that has evolved (approximately 25 studies) has been reviewed by Morgan and Wong (1985), IARC (1980, 1982) and more recently by Elcock and Morgan (1993). Of all the studies conducted to date, only one case control study (Howe et al., 1977) reported an increased relative risk for bladder cancer among,consumers of products containing saccharin. A more recent and larger case control study by the same authors (Risch et al., 1988) failed to confirm the increased risk of bladder cancer for consumers of artificial sweeteners that had been reported earlier. Reviews of non-nutritive sweeteners by IARC (1980, 1982) concluded that "the epidemiology data provide no clear evidence that saccharin alone, or in combination with cyclamates, causes urinary bladder cancer" and "there is no consistent evidence that the risk of cancer is increased among users of saccharin". More recent analyses of the epidemiological findings (Armstrong, 1985) and Elcock and Morgan (1993) reflect similar conclusions: 4

Mechanistic Studies It is clear that long-term studies of NaS in the rat provide evidence of a potential carcinogenic hazard. Observations made during the course of several of these studies have also provided clues as to the mechanism involved. Arnold et al. (1980) reported that the urine of male, but not female, rats fed NaS in their two-generation lifetime feeding study contained a white flocculent precipitate that dissolved in acetic acid. Chowaniec and Hicks (1979) observed "sandy mineralization" adherent to each of the bladder tumors that occurred in rats fed 4% NaS. Further evidence of changes in the urine of rats fed NaS in long-term studies was reported by Schoenig et al. (1985). They observed that in the group fed 7.5 % NaS, the animals that went on to develop bladder tumors excreted a significantly higher volume of urine of lower osmolality compared to animals which did not develop tumors. These findings served to focus attention on changes in the urine and bladder as etiological factors in~ the development of bladder tumors. Further evidence of the potential importance of urinary physiological changes was obtained in a study of different salt forms of saccharin (Hasegawa and Cohen; 1986). Feeding the equivalent of 5% of the sodium, potassium, calcium and acid forms of saccharin to the rat for 10 weeks produced very different effects on the urine and bladder epithelium. The sodium and potassium salts were associated with higher urine pR values, increased urine volume, urothelial hyperplasia and am increase in the thymidine labelling index. The calcium salt had a questionable and non-significant effect on the bladder, while the acid form was without effect. These differences were not correlated with exposure of the urothelium to the saccharin anion as determined by urine concentration. Anderson et al. (1988) found similar results with saccharin salt forms in a subsequent study. The critical change that leads to the development of bladder tumors in rats fed NaS appears to be urothelial hyperplasia. Proliferative changes in the bladder have been observed at the same dose levels associated with tumor development in the two-generation feeding studies reported by Taylor et al. (1980), Arnold et al. (1980) and Squire (1985). Fukushima and'! Cohen (1980) studied the sequence of changes that lead to hyperplasia in the bladder of male rats fed 5% NaS and found that hyperplastic foci' were present in all rats fed this level of saccharin by nine weeks. Garland et al. (1989) found a no effect level for hyperplasia and increased thymidine labelling index in rats fed NaS at the 3% dietary level. This corresponds to the bladder tumor no effect level of 3% observed in the two-generation IRDC study of NaS by Squire (1985). As indicated above, hyperplastic changes in the bladder in rats fed saccharin are dependent on the salt form which has been fed (Hasegawa and Cohen, 1986). Other studies have also shown that the hyperplastic response of the bladder to saccharin is also subject to strain, species and dietary influences. Garland et al. (1989) found that the bladder hyperplastic response to NaS was more marked in Fischer 344 rats than in Sprague Dawley rats and for both strains the response was greater when NaS was fed in the Prolab®' than in the Purina* diet. Hyperplasia: was not seen in the bladders of mice, hamsters or guinea pigs fed~ 5%a NaS. 5

Conflicting evidence that may reflect strain and/or dietary effects was also reported' by Nakanishi et al.. These authors found no evidence of hyperplasia in Fischer 344 rats fed NaS at levels up to 5% in a stock diet (Nakanishi et al., 1980) or in F344 rats fed' 5% NaS in the Oriental M diet (Nakanishi et al., 1982). On the other hand, these investigators reported tiyperplasia affecting a high percentage of Wistar rats fed NaS in a commercial diet (Oriental Yeast Co.). Garland et al. (1989) found that the bladder proliferative response that is present in rats fed 5 or 7.5°16 NaS in the Prolab®' diet did not occur when the same levels of NaS are fed in the AIN- 76A diet. The latter is a semipurified diet which is associated with low urinary pH values (5.5 - 6.5) when fed to rats. The effects of diet and'saccharin salt form on the bladder response to NaS have been confirmed in a tumor promotion study reported by Cohenl et al: (1991a). Male weanling rats were fed FANFT in the diet for six weeks followed by various treatments for an additional 72 weeks. The bladder tumor incidence was increased by 3 or 5% NaS; a slight increase was noted with calcium saccharin while no increase occurred in rats fed acid saccharin. The bladder tumor promoting effect of 5% NaS was abolished when~ it was fed in a diet containing ammonium chloride that lowered urine pH. A number of experiments have been conducted which point to changes in the urine in rats fed NaS as the critical mediators of the proliferative and hyperplastic changes that occur in the bladder. As indicated above, Arnold et al: (1980a) reported that male, but not female, rats fed 5% NaS had a white flocculent' precipitate in the urine. This finding has recently been confirmed by Cohen et al. (1991b) who found silicon-containing crystals and precipitate in the urine of rats fed 7.5 % NaS. Garland (to be published) used light transmittance at 650 nm, and visual inspection to measure the flocculi in the urine of rats fed NaS. They observed precipitate and decreased transmittance in the urine of male rats fed 5 or 7.5 9b NaS but not in the urine of rats fed 1 or 3% NaS. Precipitate formation and decreased light transmittance did not occur in rats fed acid saccharin or in rats with acid urine as a result of feeding the AIN-76A diet or a diet to which ammonium chloride had been added. It is important to note (Figure 1) that significantly increased incidences of bladder tumors, bladder hyperplasia, bladder labelling index and decreased urine light transmittance as a result of precipitate formation are all present at dietary levels of NaS above 3%. These changes do not occur when NaS is fed at lower dose levels. The concurrence of these changes over the same dose range is strong evidence of an etiologic relationship. More than a decade ago when precipitate was first observed in the urine of rats fed NaS, Arnold et al. (1980) demonstrated that it contained protein and saccharin. More recently, although the ~ presence of saccharin and proteinl in the precipitate has been confirmed, the major components Q have been shown to be calcium, phosphate and magnesium and occasionally silicate (Cohen, ~ ~ 6 ~ ~ ~ W

unpublished). lt is suggested that the precipitates cause necrosis of urothelial cells and that the subsequent regenerative hyperplasia is the basis for tumor development. Recent, studies also implicate the male rat-specific protein, a2µ-globulin, in the formation of urinary crystals and precipitates in rats fed NaS. Saccharin binds to rat urinary proteins, predominantly low molecular weight proteins like aZ,-globulin (Eklund et al., 1992). Studies in the NBR male rat which does not synthesize au-globulin and in castrated male F344 rats that have decreased urinary a2M globulin levels have shown that in these animals NaS has less effect on the bladders (Garland et al., 1993). Protein binding of saccharin in male rats may provide the nucleus for precpitate and crystal formation and thus explain the gender and species specificity of the saccharin tumorigenicity. As indicated above, Arnold et al.(1980),, and more recently, Cohen et al. (unpublished) have shown that precipitates and crystalluria occur in the urine of male, but not female, rats fed 5% NaS. The difference between the responses of the sexes in this respect may be critical in terms of the difference between: the sexes in tumor development. Female rats fed 7.5 % NaS have urinary saccharin concentrations at least equivalent~ to those of males; similarly, there are no marked differences between the sexes in urinary pH, urinary sodium levels or osmolality (Schoenig and Anderson, 1985; Cohen, unpublished). Female rats fed 5% NaS also have less urothelial proliferation, as measured by the labelling index, than do male animals (Cohen, 1985)) and~ females fed 5% NaS have fewer and less severe microscopic changes, as determined by SEM (Cohen, unpublished). Since precipitates and crystalluria do not occur, or occur to a lesser extent, in urine of female rats fed NaS, an additional factor(s) appears to be necessary. The high urinary protein level in male rats compared to females, and perhaps the aU-globulin in the males, may explain the high levels of precipitate and crystal formation in males compared to females. Studies in laboratories in Japan an6 Holland, as well as Cohen's laboratory, have shown that the effects of high doses of NaS on the urine and bladder of the male rats also occur when high doses of other sodium salts of organic acids are fed. Sodium ascorbate, erythorbate, aspartate, citrate, glutamate and bicarbonate, as well as sodium~ chloride and dibasic sodium phosphate, have similar proliferative and tumorigenic effects on the male rat bladder (Fukushima et al., 1986a, 1986b; Lina and Woutersen, 1989;, Cohen, et al., 1991a; Shibata et al., 1991). These changes do not occur with the parent acids: Furthermore, the urinary precipitates that occur in rats fed NaS also occur in rats fed sodium ascorbate and they are prevented when sodium ascorbate is given to rats fed the AIN-76A diet or a diet containing ammonium chloride (Garland et al., personal communication). In summary, the findings to date indicate that bladder tumors in~ the male rat fed NaS are dependent on a combination of factors in the urine that cause precipitates, urothelial cytotoxicity, regenerative hyperplasia and bladder tumors. Development of bladder tumors in male rats fed NaS occurs in the presence of a pH level above 6.5, increased urine volume and~ high urinary sodium or potassium values. Recent studies also suggest that high urinary protein levels and 7

particularly the male rat urinary protein, cr2,,-globulin, may play a role in the development of urinary precipitates and crystals (Figure 2). Bladder changes are not directly related to urinary saccharin levels since they do not occur to a significant degree in female rats, and not at all in rats fed acid saccharin or in rats fed diets that result in acidic urine even though, under these conditions, urinary saccharin levels are equal to, or higher than that in male rats fed equivalent levels of NaS in control diets. The urinary and bladder changes that occur when high dietary levels of NaS are fed are also specific for the rat. Feeding comparable dose levels of NaS to the male or female mouse does not cause urinary precipitates or urothelial hyperplasia and NaS is not a bladder carcinogen or bladder tumor promoter in mice. Risk Analysis for Humans As summarized above, there is a large body of evidence that indicates that the bladder tumors that occur in the male rat fed high dietary levels of NaS occur as the result of the formation of precipitates and crystals in the urine, urothelial cytotoxicity and hyperplasia. The bladder tumors in the rat~ fed high dietary levels of NaS appear to be specific for this species since they do not occur in the mouse, hamster or monkey. Furthermore, the mechanism shown to be involved in the development of the tumors in the male rat provides an explanation for the decreased urothelial effect in female rats and the lack of effect in mice. The mechanistic studies also suggest that the sequence of changes underlying the tumor development is unlikely to be operative on humans. The long-term bioassays of NaS in the male rat have demonstrated that bladder tumors occur only at dietary concentrations of 3% and higher. A 1% dietary concentration of NaS in the rat is a no effect level for tumors. It is also clear that the 3% dietary level is a threshold dose for the urinary changes, precipitate formation, urothelial cytotoxicity and hyperplasia: which appear to be prerequisites for tumor development and the risk of tumors occurring at lower dietary levels is zero. The 1% dietry no effect level of NaS in the rat is equivalent to a lifetime consumption of approximately 500 mg/kg bw/day. High levels of human consumption have been estimated at 2 mg/kg bw/day. This is equivalent to three diet sodas per day or two sodas and 12 sweetener tablets. High human consumption levels are 250 fold less than the no effect level for the rat. There are thus.two considerations in determining the safety of NaS for humans: first, that the mechanism responsible for development of tumors in male rats is unlikely to be relevant for humans; and second, there is a: clear threshold for the tumorigenic and related effects of NaS in the rat and consumption of a: large amount' of NaS by humans on a daily basis would be necessary if this threshold were to be exceeded. 8

The experimental data support the application of the conventional 100 fold~ safety factor to arrive at an acceptable daily intake (ADI) for NaS. Renwick (1993), in a recent analysis of the safety data for NaS has proposed that a 50 fold safety factor would be appropriate. At the 41st meeting of the Joint FAO/WHO Expert Committee on Food Additives (JECFA) in February 1993, saccharin safety data were reviewed and an ADI of 0-5 mg/kg bwLday was established for saccharin. A statistical analysis of the quantitative tumor data from tlte 1RDC study (Schoenig, 1985) was reported by Carlborg (1985). It was concluded that the observed steep response curve was not consistent with a one-hit or a linear dose-response model. The data were adequately described by a threshold model that produced 2.7% as a best estimate of a no effect threshold. The TRDC data were also adequately described by the non-threshold Weibull and~ multistage models. A risk assessment based on the Weibull model predicts that a 0.01 % level of saccharin in the diet of the male rat would lead to an~ increased risk of bladder tumors of 2.5 X 1G10 (upper confidence limit 1.2 x 10'a. Converting this level directly to man, approximately, 2.3 cans of soda per day which is the 90th percentile for consumption in the USA, yields a lifetime risk of 2.5 x 10' for bladder tumors. Extrapolation by a multistage model with a forced linear term resulted in a lifetime'risk of 5.0 x 10-s (upper confidence limit 9.1 x 10-5. A scientific review panel convened to analyze the data on saccharin in 1983 concluded that "given the steepness of the dose-response curve, forced linearity seems to be an excessively cautious means of extrapolation". The expert panel concluded that "the present level of exposure of humans to saccharin through its use as a food additve is unlikely to present a cancer risk". Research on saccharin conducted since 1983 strongly reinforces that view. 9