First draft prepared by

Dr D. Benford
Food Standards Agency, London, England

Dr J.B. Grieg
Food Standards Agency, London, England

Dr J. R. Lupton
Faculty of Nutrition
Texas A&M University, College Station, Texas, USA

and Dr C. Leclercq
National Institute for Food and Nutrition Research, Rome, Italy

Salatrim is the acronym for short- and long-chain acyl triglyceride molecules. Salatrim preparations are tailored mixtures of triglycerides containing at least one long-chain fatty acid moiety (principally stearic acid) and one or two short-chain fatty acid moieties (acetic, propionic and/or butyric acid). Salatrim is synthesized by non-enzymatic inter-esterification of triacetin, tripropionin, tributyrin or their mixtures with hydrogenated canola, soya bean, cottonseed or sunflower oils. The physical properties of salatrim may be fitted to its food applications by selecting specific short-chain fatty acids and hydrogenated oils for synthesis. Salatrim is intended for use as a low-calorie fat in soft sweets, coatings (e.g., wafers and confectionery), dairy products (including spreads), margarines and bakery products.

Salatrim of various compositions was evaluated by the Committee at its forty-ninth meeting (Annex 1, reference 131), when it concluded that the available studies did not provide an adequate basis for evaluating the safety and the nutritional effects of salatrim. At that meeting, the Committee recommended that additional, appropriately designed studies be performed to assess fully both the toxicological and nutritional consequences of ingestion of salatrim.

The Committee noted at its forty-ninth meeting that the specifications for salatrim permit compositions that include a triglyceride mixture with up to 0.87 g of stearate per gram of fat, but that the available biological data did not provide information on materials of such a composition. Since that time, the sponsor has requested that the specifications for salatrim be limited to preparations with < 70% by weight of total saturated long-chain fatty acids. Studies were available on the biological effects in humans of salatrim containing up to 0.61 g of stearate per gram of fat.

The Committee previously evaluated studies on the caloric value of salatrim, taking into account the fact that short-chain fatty acids supply fewer kilocalories per gram than long-chain fatty acids. Moreover, it concluded that the claim of reduced absorption of stearic acid has not been proven for humans. In addition, as there is no specific composition for salatrim, a single caloric value could not be assigned.

In evaluating the safety of salatrim at its forty-ninth meeting, the Committee considered studies in which salatrim of various compositions was administered to rats or minipigs at concentrations of up to 10% in the diet. These studies showed no toxicologically significant effects, but the study in minipigs was considered to have been of insufficient duration. The Committee also discussed one home-based trial and four clinic-based studies with volunteers who ingested up to 60 g salatrim per person per day. The studies indicated that the consumption of diets containing salatrim, particularly at 60 g per day, was associated with an increased incidence of mild gastrointestinal symptoms and significantly elevated serum enzyme activities.

Thus, the Committee concluded at its forty-ninth meeting that the available studies in experimental animals and humans did not provide an adequate basis for a toxicological evaluation. In addition, as the studies were not optimized to detect potential nutritional effects, the Committee concluded that they did not provide an adequate basis for a nutritional evaluation.

In response to the Committee’s earlier request, four additional home-based trials in humans and one study of nutrition in rats were submitted for consideration at the present meeting.

Table 1 shows the nomenclature of salatrim products. The products that have been used in safety evaluations are listed in Table 2. The results of all these studies were published in The Journal of Agricultural and Food Chemistry, Volume 42 (2) in 1994. The papers contained numerous typographical and transcription errors, and three sets of errata were subsequently published in The Journal of Agricultural and Food Chemistry, Volume 42 (4, 9 and 12) in 1994, and the Committee was informed of additional corrections. Two of the newly submitted studies have been published (Nestel et al., 1998; Sanders et al., 2001), whereas the remaining newly submitted studies in humans were unpublished. The summaries given below contain information from the published accounts and also from the full, unpublished study reports.

Table 1. Typical molar ratios of short- and long-chain acid sources used to prepare the salatrim family of edible oils

The salatrim family name defines the sources of the short-chain and long-chain fatty acids, the numerals representing the carbon lengths of the short-chain acids in decreasing proportion in the mix, and the letters defining the oil that provides the source of the long-chain fatty acids; e.g., in salatrim 43SO, tributyrin and tripropionin are the short-chain fatty acids and hydrogenated soya bean oil is the source of long-chain fatty acids. The molar ratio of the mix that is used to prepare salatrim is 11 parts tributyrin:1 part tripropionin:1 part hydrogenated soya bean oil.

Table 2. Materials used in studies of the metabolism and toxicity of salatrim

^a Plus supplementary 17-day test of effects on transaminases

In an initial review of the metabolism of fats and long- and short-chain fatty acids, it was proposed that short-chain fatty acids in salatrim would be released after hydrolysis of the triglyceride in the stomach. A proportion of those released would be absorbed in the stomach and used as an energy source, while the remainder would be taken up by the liver. Hydrolysis of long-chain fatty acids from salatrim fats would occur predominantly in the small intestine. The absorption of stearate would be limited. A proportion of absorbed stearic acid would be converted to oleic acid. Any short-chain fatty acids released in the small intestine would enter the hepatic portal vein (Hayes et al., 1994a).

Experiments in which salatrim 23CA was administered to rats were designed to test this hypothesis. Salatrim 23CA was chosen because it contains triacyl-glycerides unique to the salatrim family, whereas triglycerides containing butyrate are commonly consumed as part of the diet.

2.1.1 Biotransformation

Rats

Salatrim products are not absorbed intact, as they are hydrolysed and metabolized in an identical manner to triacylglycerides present in the diet. Monoacylglycerides containing stearate derived from the hydrolysis of salatrim products can be absorbed from the small intestine. After administration of a single oral dose of [¹⁴C-stearate]salatrim (1.4 g/kg bw) to male Sprague-Dawley rats, 0.4% of the radiolabel was present in fat 72 h later, of which half was in oleate (Musick & Peterson, 1993; Hayes et al., 1994b).

A single oral dose of 1.4 g/kg bw of salatrim 23CA labelled with ¹⁴C in the acetate, propionate, stearate or glycerol moiety or 1.4 g/kg bw of triolein labelled with ¹⁴C in the oleate or glycerol moiety was administered to groups of five male Sprague-Dawley rats by gavage. Radiolabel elimination was followed for 72 h. Acetate and propionate from salatrim were exhaled as CO₂ (82% and 89% of the dose, respectively). Stearate was released from salatrim 23CA, and oleate was released from radiolabelled triolein, 22% of the labelled stearate and 44% of the labelled oleate being exhaled as CO₂. In experiments with salatrim or triolein labelled in the glycerol moiety, 75% of the dose was exhaled as CO₂. Faecal excretion of radiolabel after administration of salatrim labelled in the acetate, propionate or glycerol moiety and glycerol-labelled triolein approximated to 4–5% of the dose. More faecal excretion of radiolabel was seen when [¹⁴C-stearate]salatrim (55%) and [¹⁴C-oleate]triolein (38%) were administered to rats. Less than 4% was excreted in the urine, and this consisted of radiolabel that had presumably been incorporated into intermediary metabolites. Urinary excretion of radiolabel was greater in animals treated with salatrim radiolabelled at the short-chain fatty acid moiety than in those given salatrim or triolein radiolabelled in the long-chain fatty acid moiety. Approximately 10% of the radiolabel in experiments with stearate- or glycerol-labelled salatrim and oleate- or glycerol-labelled triolein was recovered from the carcass. Very small amounts of radiolabel were found in the liver, blood and fat in these studies (total, < 1.8%). Feeding a diet containing 10% salatrim 23CA for 2 weeks before dosing with radiolabelled salatrim or triolein did not affect their metabolism. Slightly greater faecal excretion of radiolabel was reported in all the investigations with animals fed salatrim before testing, and the authors suggested that competition for metabolism by dietary salatrim and radiolabelled salatrim administered by gavage was responsible.

The authors concluded that the absorption, distribution and elimination of salatrim are identical to those of other triglycerides found in the diet and that the data support the observation that stearate is less well absorbed than oleate. No conclusions could be derived about the absorption of salatrim 23CA and its component fatty acids, since absorption was not measured directly. The value of this study with respect to the assessment of absorption of salatrim is limited, because the test solutions were prepared by mixing radiolabelled triolein (control fat) into the test fat (salatrim) matrix, and the absorption of the control fat may therefore have been affected. The radiolabelled triolein should have been incorporated into non-radiolabelled triolein. The data do, however, support the view that the metabolism of salatrim 23CA is similar to that of triolein once it has been absorbed (Musick & Peterson, 1993; Hayes et al., 1994b).

In vitro

The hydrolysis of a number of salatrim products by porcine pancreatic lipase was studied over 30 min. Chloroform solutions (100 mg/ml) of salatrim 4CA, 23CA, 32CA and 234CA were incubated at 37 °C for 2, 5, 10 or 30 min, and the predominant triacylglycerides, diacylglycerides, monoacylglycerides and long- and short-chain fatty acids were measured by gas chromatography with mass spectrometry. Each of the salatrim products tested showed a consistent pattern of hydrolysis, consisting mainly of a peak in diacyl-glyceride formation after about 2 min and a concurrent rapid rise in free stearate over 5 min. The rate of stearic acid formation slowed during the remainder of the 30-min period. In experiments with salatrim 23CA and 32CA, hydrolysis of the triglyceride containing two short-chain fatty acids (i.e. di-short triglyceride) was faster and more complete than that of the corresponding di-long triglyceride, which contained two stearate esterifications. The hydrolysis of triacylglycerides containing butyrate was faster than that of those containing acetate. The authors concluded that salatrim molecules undergo lipolysis in a predictable manner.

The authors speculated that the higher rate of release of short-chain as compared with long-chain fatty acids was due to the greater hydrophilicity of short-chain fatty acid-rich triacylglycerides within fat droplets and the faster diffusion of the released short-chain fatty acids from the active site of the enzyme into the aquatic phase surrounding fat droplets. The authors concluded that rapid release of short-chain fatty acids would occur in the stomach and upper intestine (Phillips, 1992; Sequeria & Gordon, 1993)

Caloric availability in rats and humans

Studies on this topic are summarized in Appendix 1 to this monograph.

2.1.2 Effects on enzymes and other biochemical parameters

Twenty-four Crl:CD BR VAF strain rats of each sex were fed a diet containing 10% salatrim 23SO for 17 days. An additional 24 rats of each sex were fed a diet containing 10% corn oil, and 12 rats of each sex received only the basal diet throughout the study. The rats were observed twice daily, and body weights were recorded weekly. Blood was collected, and serum concentrations of aspartate aminotransferase (AST) and alanine aminotransferase (ALT) and gamma-glutamyltranspeptidase (GGT) were determined 12 and 4 days before and on days 3, 6, 9, 13 and 17 after initiation of the study. Neither salatrim nor corn oil had any effect on enzyme activities at any interval (Kiorpes, 1993a; Hayes et al., 1994c).

2.2.1 Short-term studies of toxicity

Rats

(i) Salatrim 4CA

Diets containing 0, 2, 5 or 10% salatrim 4CA or 10% corn oil were fed to groups of 20 Crl:CD BR VAF rats of each sex for 13 weeks. The salatrim-containing diets were supplemented with varying amounts of vitamins A, E, D and K according to the level of salatrim 4CA incorporation. A vitamin control group was also included, the level of supplementation being equivalent to that used in the 10% salatrim diet. A further group of 10 rats of each sex were fed salatrim-containing diets for 5 weeks and were used for interim histopathology and measurement of calcium, copper, iron, magnesium, phosphorus, sodium, strontium and zinc in defatted femur (10% salatrim or corn oil only). Blood and urine samples were collected at 4 and 13 weeks for clinical chemistry, urinary analysis and haematology. Bone minerals were also measured at 13 weeks. The animals were observed for signs of toxicity daily, and body-weight gain and food intake were measured weekly. At autopsy, the adrenals, brain, liver, kidneys and testes were weighed. The 10% dose represented the highest concentration that the authors considered would not cause excessive dilution of micronutrients.

No treatment-related deaths occurred, and no effects on body weight were observed in salatrim-fed animals, although increased weight gain was noted in most weeks of the study for animals fed corn oil. Decreased food consumption was observed for males fed 10% salatrim 4CA and males and females fed corn oil. There were no effects on haematological, clinical chemical or urinary end-points. The serum levels and urinary clearance of minerals were unaffected by treatment with salatrim or corn oil. No treatment-related changes in the levels of fat-soluble vitamins in serum or liver were documented in animals given 10% salatrim 4CA when compared with vitamin-supplemented controls that received the same amounts of vitamins in the diet (the occasional differences seen in single sexes at week 5 were not confirmed at termination). No definitive conclusions can be drawn regarding any potential effect on fat-soluble vitamin absorption, since an appropriate unsupplemented salatrim group was not included in the study. The concentrations of strontium and zinc in bone were higher in animals of each sex at 10% salatrim than in either control group, while the sodium concentration was higher only in females in comparison with the unsupplemented control group. Animals of each sex given 10% corn oil had higher concentrations of strontium in bone than either control group, while the concentration of zinc in bone was lower in males than in unsupplemented controls. No treatment-related effects on organ weights or histopathological changes were documented at the interim or final necropsy. A number of animals fed 10% corn oil showed hepatocellular vacuolation. The authors concluded that the changes in the concentrations of minerals in bone were directly related to the quantity of unsaturated fatty acids in the diet fed to animals. The NOEL was 10% in the diet, equal to 7.3 g/kg bw per day (Williams, 1992a; Hayes et al., 1994d).

(ii) Salatrim 23CA and 32CA

Diets containing 0, 2, 5 or 10% salatrim 23CA or 32CA or 10% corn oil were fed to groups of 20 Crl:CD BR VAF rats of each sex for 13 weeks, when blood and urine samples were collected from a subgroup of 10 rats of each sex per group for haematology, serum and urine chemistry and urinary analysis. Blood was obtained from the remaining 10 rats of each sex per group for measurement of fat-soluble vitamins in serum. Animals were observed for signs of toxicity daily, and body-weight gain and food intake were measured weekly. The 10% dose represented the highest concentration that was considered would not cause excessive dilution of micronutrients. Adrenals, brain, kidneys, liver and testes were weighed at autopsy. The caecum of each rat was ligated at the distal ileum, at the proximal colon and at approximately the distal one-third of the blind end. The distal portion of the caecum was collected for histological examination, and the remaining ligated portion of the caecum from each rat was used for a special study of effects on gut microflora (see section 2.2.5). Minerals were measured (as for salatrim 4CA) in samples of bone (defatted femur) from 10 animals per group at necropsy.

No treatment-related deaths occurred. The mean body-weight gain and feed consumption were similar to those of control animals. No toxicologically significant effects on haematology or clinical chemistry were reported. A dose-related trend to slightly increased urinary phosphorus clearance was noted in rats fed salatrim fats. A statistically significant increase in urinary phosphorus clearance was seen in males fed 10% salatrim 23CA and in males and females fed 10% salatrim 32CA. No changes in the levels of fat-soluble vitamins in serum or liver were documented, except for a reduced level of vitamin A in animals fed corn oil. The mean strontium concentration in bone was significantly higher in females fed 10% salatrim 23CA than in controls. A slight, statistically nonsignificant increase in the mean concentration of strontium in bone was also found in males fed 10% salatrim 23CA. The mean zinc concentration in bone was significantly higher in females fed 10% salatrim 23CA or 10% salatrim 32CA than in control females. The mean zinc concentration in bone was significantly lower in males given the 10% corn oil diet than in controls. No treatment-related effects on organ weights were noted. An increased incidence of renal mineralization was noted microscopically in females of groups receiving corn oil and 5% and 10% salatrim 23CA and salatrim 32CA when compared with controls. The incidence and severity of this lesion was similar in each of these groups of triacylglycerol-treated females. No treatment-related renal mineralization was noted in any group of corn oil-treated or salatrim-treated males. The authors concluded that the changes in levels of minerals in bone and renal mineralization were directly related to the quantity of unsaturated fatty acids in the salatrim diet fed to animals. A number of animals fed 10% corn oil had hepatocellular vacuolation. The Committee concluded that these two salatrim products did not induce any toxicologically significant effects. The concentration of 10% salatrim 23CA or 32CA was equal to 7.5 g/kg bw per day (Williams, 1992b; Hayes et al., 1994c).

(iii) Salatrim 234CS and 234CA

Diets containing 0 (control), 2, 5 or 10% salatrim 234CS or 234CA or 10% corn oil were fed to groups of 20 Crl:CD BR VAF plus rats of each sex for 13 weeks. At that time, blood and urine samples were collected from a subgroup of 10 rats of each sex per group for haematology, serum and urine chemistry and urinary analysis. Blood was obtained from the remaining 10 rats of each sex per group for measurement of fat-soluble vitamins in serum. Animals were observed for signs of toxicity daily, and body-weight gain and food intake were measured weekly. The 10% dose represented the highest concentration that it was considered would not cause excessive dilution of micronutrients. Adrenals, brain, kidneys, liver and testes were weighed at autopsy. The levels of minerals were measured (as for salatrim 4CA) in samples of defatted bone (femur) from 10 animals per group at autopsy.

No treatment-related deaths occurred. The mean body-weight gain and feed consumption of salatrim-treated groups were similar to those of untreated control animals, although controls given corn oil and females fed 10% salatrim 234CS gained slightly more weight. A significant decrease in food consumption was noted in animals of each sex fed 10% corn oil. No toxicologically significant effects on haematological or clinical chemical parameters were reported. Slight, statistically nonsignificant increases in urinary phosphorus clearance were seen in all groups given 10% salatrim. The mean serum vitamin A level was significantly higher in male rats fed 2% salatrim 234CA or 10% corn oil than in controls. The mean vitamin A concentration in liver was significantly lower in males fed 10% salatrim 234CA and males and females fed 10% salatrim 234CS or 10% corn oil than in controls. The mean serum 25-hydroxy vitamin D concentrations were significantly lower than those of controls in females fed 2 or 10% salatrim 234CA, 2 or 10% salatrim 234CS or 10% corn oil. The authors noted inconsistencies in the results for males and females with regard to the effects of the salatrim products on fat-soluble vitamin levels. They considered that corn oil induced a similar reduction in serum 25-hydroxy vitamin D levels and a larger increase in serum vitamin A levels in comparison with salatrim and concluded that the salatrim products tested did not substantially alter fat-soluble vitamin absorption. The mean concentration of sodium in bone was significantly lower in females fed 2% salatrim 234CA, and the mean concentration of zinc in bone was significantly higher in males fed 2% salatrim 234CA and in females fed 10% of either salatrim fat when compared with controls.

Reduced ratios of liver- and brain-to-body weight in females fed 10% salatrim 234CS were considered by the authors to be related to increased terminal body weights. Macroscopically, no treatment-related effects were observed in salatrim-treated rats, but an increased incidence of renal mineralization was seen in females fed 5% or 10% salatrim 234CA and a slightly higher incidence of renal mineralization in females fed 5 or 10% salatrim 234CS compared with controls. Except in the groups treated with 10% salatrim 234CA, the renal mineralization was similar in appearance in all groups. In females fed 10% salatrim 234CA, the severity of renal mineralization was slightly greater than in other groups. The authors concluded that the changes in the levels of minerals in bone and renal mineralization were directly related to the quantity of unsaturated fatty acids in the diet fed to treated animals. A number of animals fed 10% corn oil had hepatocellular vacuolation. The Committee concluded that these two salatrim products did not induce any toxicologically significant effects. The concentration of 10% salatrim 234CS or 234CA was equal to 7 g/kg bw per day (Williams, 1992c; Hayes et al., 1994g).

Minipigs

(i) Salatrim 23SO

Diets containing 0, 3, 6 or 10% salatrim 23SO or 10% corn oil were fed for 28 days to groups of four Hanford minipigs, 3.5–7 months old and weighing 17–30 kg at initiation of treatment. A control group was fed the basal diet. The group fed 10% corn oil served as a reference for those fed the high fat content salatrim diets. All diets (including the diet used for the untreated control group) were supplemented with 2% (w/w) corn oil, considered necessary by the authors to avoid possible induction of essential fatty acid deficiency due to dilution with the test fat. Each pig was given 500 g of the appropriate diet twice a day. The test diets were prepared every 2 weeks and stored frozen (–20 ± 10 °C) until use. After removal from the freezer, the diets were maintained at room temperature for 1–6 days (average, 3.3 days) before being fed to the pigs. Evidence of significant degradation of the test diet was reported in one stability trial in which samples of the 3% and 10% test diets were stored frozen for 63 days. The authors considered that the storage conditions used in the study would have resulted in minimal degradation (approximately 7%) of the test diets. The results of additional stability trials reported in the original unpublished account of the study support the view that limited degradation occurred during 2 weeks of frozen storage. However, the data suggest that degradation may have occurred during storage at room temperature and that the stability may have differed significantly between batches of diet. Thus, it is difficult from the available information to estimate the precise doses given to the minipigs.

Blood was collected from the vena cava of each pig 2 weeks and 3 days before initiation of salatrim feeding and on days 3, 7, 14, 21 and 29 after initiation of feeding. The pigs were fasted overnight before blood collection. Haematological and clinical chemical parameters were determined in these samples. After 28 days of treatment, all pigs were subjected to gross necropsy, and adrenals, brain, kidneys, liver, ovaries, spleen, testes, thymus and thyroid were weighed. The entire femur was removed and stored frozen at –20 ± 10 °C, and dry weight and percentage ash were determined. Each femur was assayed for calcium, phosphorus, strontium and zinc concentrations by inductively coupled plasma spectrometry.

No treatment-related effects were seen during daily physical examinations, and all pigs survived to the scheduled terminal sacrifice. The mean body weights, body-weight gains and feed consumption of pigs receiving salatrim 23SO and corn oil were comparable to those of untreated controls. Haematological and clinical chemical evaluation revealed no treatment-related effects. Fairly wide variations in ALT and AST activity in serum were seen in individual animals: 2 weeks before initiation of treatment, the activity of AST was significantly higher in males in the group destined to be fed 10% salatrim, and the activity of ALT was lower in males both before and 7 days after feeding 6% salatrim. This variation was unexplained but could have been random. The mean serum cholesterol concentration at day 3 in females given 6% salatrim 23SO was lower than that of untreated controls, and the serum concentrations of low-density lipoprotein at day 3 in females given 6 or 10% salatrim 23SO or 10% corn oil were lower than those of untreated controls. At day 29, the mean serum cholesterol and high-density lipoprotein cholesterol concentrations were higher in female pigs fed 10% corn oil than in untreated controls. No biologically significant findings were reported with respect to serum and liver vitamin A and E concentrations. No differences in percentage ash or bone concentrations of calcium, phosphorus or strontium were found between treated and untreated control pigs of either sex.

No differences in organ weights were found between pigs fed salatrim 23SO and untreated controls. Macroscopically, no treatment-related effects were observed in any animals treated with either salatrim or corn oil. Microscopically, a slight increase in the severity of focal vacuolation in hepatocytes was found in one male fed 10% corn oil and one male fed 10% salatrim. The authors considered this to be either a spurious finding, since it occurred in only two males, or a non-specific effect of fat, as it occurred only in pigs given 10% fat diets (corn oil or salatrim). The Committee concluded that 10% salatrim 23SO produced neither toxicologically nor nutritionally significant effects. The concentration of 10% salatrim 23SO was equal to 3.3 g/kg bw per day (Kiorpes, 1993b; Hayes et al., 1994e).

2.2.2 Genotoxicity

The results of tests on the genotoxicity of salatrim fats are given in Table 3.

Table 3. Results of tests for genotoxicity with salatrim fats

^a Maximum dose restricted by precipitation at 1000 µg/ml (or 1000 µg/plate).

^b No evidence of cytotoxicity

2.2.3 Special studies: Effects on gut microflora

Diets containing 10% salatrim 23CA or 32CA or 10% corn oil were fed to groups of 20 Crl:CD BR VAF rats of each sex for 13 weeks. At necropsy after a 24-h fast, the caecum was ligated at the distal tip, proximal to the ileocecal junction and distal to the exit into the colon. The caecal tip was removed for histological examination, and the remainder of the caecum was removed and frozen at –20 °C until use. After thawing, the contents of each caecum were thoroughly mixed by kneading within the caecum and then removed, and a portion of the caecal contents from five male rats in each dietary group were examined by scanning electron microscopy for changes in the dominant bacterial morphotypes. The caecal contents from all animals were analysed for pH, bile acids, neutral sterols (cholesterol and its secondary metabolite coprostanol) and phytosterols, and measurements were made of the primary bile acids cholic and alpha- and beta-muricholic acids; the secondary bile acids deoxycholic, lithocholic, hyodeoxycholic, omega-muricholic and unsaturated omega-muricholic acids; the primary phytosterols 24beta-ethylcholesterol and 24beta -methylcholestero; and the secondary phytosterol metabolites 24alpha-methylcoprostanol (campestanol), 24beta-methylcoprostanol, 24alpha-ethylcoprostanol (stigmastanol) and 24beta-ethylcoprostanol (sitostanol).

The authors noted wide inter-animal variation in the concentrations of bile acids. No significant differences in caecal pH or in the level of secondary bile acids as a percentage of total bile acids were reported. Increased coprostanol concentrations were found in male rats fed salatrim 23CA, 32CA and corn oil, but with no effect on the ratio of coprostanol to cholesterol. Increased concentrations of all four secondary phytosterols were found in rats fed corn oil, whereas the level of only one (24alpha-methylcoprostanol) was increased in rats fed salatrim 32CA. In general, salatrim-fed rats of each sex produced slightly less of the three remaining secondary phytosterols than chow-fed rats, while rats fed corn oil produced more. No evidence of any alteration in the population of bacterial morphotypes was reported, although the authors considered that the inter-animal variation limited the sensitivity of scanning electron micrography to the detection of major changes (Scheinbach et al., 1994).

2.3.1 Clinic-based studies

Four clinical studies of controlled diets that were reviewed by the Committee at its forty-ninth meeting were:

• an acute test for tolerance with a double-blind cross-over design and doses of 45 g and 60 g salatrim 23CA;
• a 7-day test with a double-blind protocol and doses of 45 g and 60 g salatrim 23CA;
• a four-way triple cross-over test with 30 g or 60 g of salatrim 23SO, hydrogenated soya bean oil as the control and hydrogenated coconut oil as the wash-out vehicle, with a Latin-square treatment sequence; and
• an acute bolus dose of salatrim 23SO to measure the effects on serum ketones.

(i) Acute tolerance

A randomized, double-blind, cross-over design was used, in which volunteers (six men and four women) aged 18–65 (mean, 38 years) received salatrim 23CA or coconut oil for 1 day, at either 60 g/day (for eight individuals consuming a 2500-kcal diet) or 45 g/day (for two, both female, consuming a 1800-kcal diet). The materials were introduced into the diet in the form of vanilla sandwich biscuits and chocolate bonbons (or bars), each containing 5 g of either salatrim or control fat. On day 4, five volunteers received the test material, while five received the control material. The substitution of test material was reversed on day 8. Thus, there were two treatment groups in this study, one on day 4 (group 1) and one on day 8 (group 2), for whom changes in clinical parameters on days 5 and 9, respectively, would suggest a treatment-related effect. A maintenance diet (either 1800 or 2500 kcal/day) including the control material was administered on all other study days. A standardized 4-day meal plan was repeated for three cycles.

The results from the full, unpublished report of this study are summarized here. Salatrim 23CA increased the mean activities of serum lactate dehydrogenase and gamma-GGT in both treated groups and increased the mean activities of serum ALT and AST and alkaline phosphatase in group 2. The changes were modest, and the group means did not exceed the reference range. A slight increase in mean serum cholesterol concentration was reported at the end of the study in both treated groups. The authors of the published report considered that the small size of the group did not permit a conclusion about the palatability of the salatrim foods; however, the unpublished report stated that the volunteers rated the salatrim-containing foods lower than identical control food carriers. Miild adverse gastrointestinal symptoms such as flatulence, nausea and diarrhoea were reported by a number of volunteers. In the published report, these symptoms were not related to consumption of salatrim 23CA when the data were analysed by time of onset (Besselaar Clinical Research Unit, 1993a,b; Finley et al., 1994a).

(ii) 7-day test

In a randomized, double-blind design, 36 volunteers (19 men and 17 women) aged 18–65 (mean, 33 years) received either salatrim 23CA or coconut oil for 7 days, at 60 g/day (for those on a 2500-kcal diet) or 45 g/day (for those on a 1800-kcal diet). The materials were introduced into the diet in the form of biscuits, bonbons (or bars) and chocolate ice cream. All volunteers received a maintenance diet (1800 or 2500 kcal/day) containing the control material on days 1–7. On days 8–14, 12 men received the test material at 60 g/day and six women received it at 45 g/day; nine men and nine women continued on the maintenance diet with food carriers containing control fat. One woman on the test material withdrew for reasons unrelated to the test on day 10. On days 15–24, all the volunteers were returned to the maintenance diet. A standardized 7-day meal plan was followed for three cycles throughout the study.

Except where stated, the results summarized below were taken from the published report. Increases in mean serum ALT (19%) and AST (< 41%) activity over that seen before intake of salatrim 23CA were recorded during treatment: three persons had ALT values above the normal maximum of 35 mU/ml, and one had a value for AST above the normal maximum of 50 mU/ml. Lactate dehydrogenase activity was also increased during intake of salatrim 23CA,although the values remained within the normal range throughout the test. The authors reported that the serum AST and lactate dehydrogenase activities declined steadily to control levels; the unpublished report states that all parameters declined to near baseline levels after withdrawal of salatrim 23CA. Mean corpuscular volume, number of monocytes, serum calcium and carbonate concentrations and GGT activity were all significantly altered in the volunteers receiving salatrim, but all values remained within the normal range, and none of the changes was considered by the authors to be clinically relevant.

A significant increase in total cholesterol was found before the test, which was associated with ingestion of hydrogenated coconut oil. On days 8–14, a significant decrease in total and low-density lipoprotein cholesterol was found in the group receiving salatrim 23CA, whereas the values for the control group remained elevated. No significant changes in urinary parameters were reported. Large increases in the faecal excretion of fats and stearic acid were seen when salatrim 23CA was added to the diet.

The group fed salatrim 23CA reported more headaches and gastrointestinal symptoms during the test period, nausea, abdominal pain and headaches being the most frequent symptoms described. The authors considered that these effects were mild, none leading to refusal to continue the study or requiring clinical intervention. The unpublished report showed that 14/17 volunteers (eight men and six women) given salatrim 23CA (78%; 56% of controls) reported one or more adverse effect, regardless of study material. Flatulence and nausea were reported by 61% (22% of controls) and 67% (17% of controls) of volunteers given salatrim 23CA, respectively, and headache by 56% (11% of controls). The adverse gastrointestinal tract symptoms were moderate in 8/17 individuals given salatrim 23CA and mild in the remainder. None of the reported symptoms was considered to be severe. The unpublished report also suggested that the salatrim carrier foods were considered by the subjects to be of poorer palatability than the control carrier foods, and they may have been able to distinguish between the food products used in the trial. Hence, the study may have been biased (Besselaar Clinical Research Unit, 1993c,d; Finley et al., 1994b).

(iii) 4-day triple cross-over test

Volunteers received chocolate raisin crisp bars and hot chocolate drink prepared with salatrim 23SO at 30 g/day (plus 30 g of control fat) or 60 g/day or 60 g/day of hydrogenated soya bean material (control) for 4 days. Before and after receiving the test or control vehicle, the volunteers were given 60 g of hydrogenated coconut oil for 4 days, which was prepared from the same coconut oil used in the two studies summarized above and served as a wash-out medium between the salatrim and control treatments. A standardized 8-day meal plan was repeated three times throughout the study.

No clinically significant differences were reported between volunteers receiving salatrim 23SO at 30 g/day and controls; however, those given salatrim at 60 g/day had statistically significant increases in mean serum ALT, AST and lactate dehydrogenase activities and decreased mean serum cholesterol concentration. These changes were well within the normal ranges and were considered by the authors to be clinically unimportant. The values returned to pre-test levels when the volunteers were transferred to the coconut oil wash-out diets. The total stool weight was significantly higher in all volunteers given salatrim at 60 g/day than in those given salatrim at 30 g/day or soya bean oil at 60 g/day. The values for stool water, total fat and stearic acid were significantly higher in women and men given salatrim at 60 g/day than in those given 30 g/day or soya bean oil at 60 g/day. More frequent stool softening and abnormal stools were reported by women given the higher concentration of salatrim as compared with the lower concentration and by all volunteers given the higher concentration of salatrim as compared with either the lower concentration or the soya bean oil. A small but statistically significant increase in mean serum beta-hydroxybutyrate concentration was reported in volunteers given 60 g of salatrim per day (0.2 ± 0.10 mmol/l, compared with 0.1 ± 0.06 mmol/l in controls). Adverse gastrointestinal effects (abdominal pain, diarrhoea and nausea) were reported by 10 women and 5 men given salatrim at 60 g/day. The authors considered that the lower body weights of the women might have exlained their more frequent complaints. These results support the conclusion that salatrim at < 30 g/day does not cause significant gastrointestinal symptoms. The Committee noted the limited duration of salatrim intake in this study.

The volunteers considered that the hot chocolate drink containing salatrim was acceptable but disliked the other salatrim food carriers. No evidence of carry-over in subjective assessments of food carriers was reported. The taste of the salatrim-containing food carriers may have biased the study (Besselaar Clinical Research Unit, 1993a; Finley et al., 1994b).

(iv) Effects on ketones

A randomized study was conducted with seven groups of six volunteers to determine the effect of a single dose of salatrim 23SO, hydrogenated soya bean oil or a medium-chain triglyceride on the serum concentrations of acetate, acetoacetate and hydroxybutyrate (monitored for up to 4 h). All fat samples were delivered in one cup of chocolate-flavoured beverage in the morning after a 10-h fast. The volunteers were assigned randomly (with equal distribution by age and sex) to salatrim 23SO at 7.5, 10, 12.5 or 15 g; control hydrogenated soya bean oil at 7.5 or 15 g; or medium-chain triglyceride at 15 g. A slight increase in serum acetate concentration was seen in volunteers receiving 15 g of salatrim, but no increase in serum ketone concentration was seen at any concentration of salatrim. As expected, slight increases in acetoacetate and beta-hydroxybutyrate concentration were observed in volunteers receiving medium-chain triglyceride. Adverse gastrointestinal effects were reported by five individuals: one each given 10 g, 12.5 g and 15 g salatrim and one each given medium-chain triglyceride and soya bean oil. The authors considered that salatrim was not ketogenic at the concentrations used (Besselaar Clinical Research Unit, 1993b; Finley et al., 1994b).

2.3.2 Home-based studies

The Committee at its forty-ninth meeting evaluated a randomized, double-blind, multiple-dose, parallel comparison of fat replacement by salatrim 23SO, 4SO or 43SO with a control soya bean oil. Four additional clinical studies were conducted subsequently, comprising:

• a study in hypercholesterolaemic individuals given diets containing margarines enriched with either salatrim or palm oil, including an initial 2-week period on a low-fat diet and 5 weeks on each high-fat diet with a cross-over design;
• a randomized cross-over trial to compare the effect of a single experimental meal containing 30 g of salatrim 23SO, cocoa butter or high-oleic sunflower oil;
• a study of five men given 30 g of salatrim 4SO from whom blood samples were taken every 20 min for 6 h for measurement of plasma cholesterol, triglycerides, clotting parameters and butyric acid; and
• a study in which eight normolipidaemic volunteers were fed a single meal supplemented with either 30 g of salatrim 23SO or cocoa butter in a cross-over design and five men were given a meal containing salatrim 4SO.

(i) Gastrointestinal effects and liver enzymes

Fat replacement by salatrim 23SO, 4SO or 43SO oils was compared in a randomized, double-blind, multiple-dose study, with that by soya oil. Groups of at least 12 female and 12 male volunteers aged 19–63 (mean, 35 years) were given diets containing 30, 45 or 60 g of 23SO oil, 60 g of 4SO oil or 60 g 43SO oil for 6 weeks; two control groups were included to account for the anticipated diversity in clinical values for a typical population. The total fat intake of volunteers in all groups was 60 g/day. All persons received the control fat (soya oil) during weeks 1 and 6, and either control or test fat as assigned per group in weeks 2–5. The food products were changed weekly on a 2-week cycle to ensure variety. The volunteers were given food products for consumption each week, five of which were to be integrated into the daily diet: four contained 15 g of control or test oil, and one product (crackers or cornflakes) did not deliver test or control oil. The volunteers were free to consume a normal diet in addition to the food provided, the only restriction being that the amount of alcohol consumed be limited to no more than two glasses of wine or two servings of beer per day.

After screening, subject selection and initial check-up, which included investigations of drug use and pregnancy (day 0), the volunteers received products and diaries for reporting food consumption and health over the next 7 days. Body weight was also recorded, and blood was drawn for analysis. The volunteers returned after a 10-h fast every 7 days thereafter (days 8, 15, 22, 28 and 36) to receive food products for the next week, to return the diaries, to be weighed and to have blood drawn. On the final day of the study (day 43), the volunteers returned to the clinic to hand in the diaries, be weighed and have final samples of blood drawn. Daily health was assessed for a number of general categories.

The daily diaries were used to record the type and amount and the time of day on which all foods and beverages were consumed, to rate the palatability of the foods provided and to record side-effects and the quality of daily life. The food carriers used were ice cream, chocolate milk, pudding and yoghurt produced by conventional processes. Cinnamon raisin muffins, chocolate cake, lemon cake and waffles were prepared before the study and held frozen until they were dispensed to the volunteers. Chocolate milk was prepared in multiple batches every 2 weeks as needed. All other products were produced in a single lot before initiation of the study. The formulations of the products were adjusted so that each individually packaged serving delivered 15 g of control oil or salatrim in each unit. The food carriers were rotated on a 2-week basis:

• weeks 1, 3, 5: muffins, chocolate cake, ice-cream, yoghurt;
• weeks 2, 4, 6: waffles, lemon cake, chocolate milk/pudding.

Control diets were administered in weeks 1 and 6.

Of the 183 volunteers who started, 149 completed the study. The results of clinical assessments were reported only for those who completed the study. The Committee noted differences between the published and unpublished reports in the numbers of individuals in each group who dropped out. The data given here are derived from the unpublished clinical report prepared by the study authors. A total of 34 individuals dropped out of the study: 1/26 (control 1), 3/27 (control 2), 4/27 (30 g 23SO), 5/27 (45 g 23SO), 2/25 (60 g 23SO), 7/26 (60 g 4SO) and 12/25 (60 g 43SO). After exclusion of the controls who dropped out of the study, 20 of the 31 volunteers who dropped out and who had received salatrim in their diet eported that adverse effects due to the test material had been the reason for leaving the study. The authors reported that the salatrim food products were well tolerated.

Transient increases in mean serum AST and ALT activity over time were observed in controls and all treated groups; however, the magnitude of the increases in ALT and AST activity over that at day 8 was greater in the groups given 60 g of salatrim than in the control groups. The authors reported that the ALT and AST activities in all groups by the end of the 4-week treatment approached values equivalent to those recorded on day 8. There was no clear evidence of a reversal in enzyme activity, and appropriate statistical tests would be required to evaluate the data further. None of the group means exceeded the normal clinical limit for AST or ALT. A small, transient reduction in mean serum cholesterol concentration was recorded in the unpublished report among volunteers given salatrim 23SO or 43SO at 60 g/day.

A substantial number of volunteers who consumed 60 g/day reported stomach cramps and nausea. The authors calculated the percentage of days during treatment on which adverse effects were reported [number of individuals with adverse effect × number of days with effect / 28 × total number of individuals]. Among those at 60 g of salatrim per day, stomach cramps and nausea were reported during approximately 17–21% and 19–25%, respectively, of the 28-day treatment period, depending on which salatrim product was being evaluated. Among volunteers at 45 g/day and 30 g/day, stomach cramps and nausea were reported during approximately 8–9% and 10–11%, respectively, of the period. A similar analysis of other symptoms was not presented in the published report. The authors considered that subjects given 30 g/day salatrim 23SO did not report nausea that impaired daily function. Three volunteers out of 23 who completed the study and who were given salatrim 23SO at 30 g/day experienced stomach cramps or nausea for at least 10 days of the trial compared with one volunteer out of 49 who completed the study in the combined control groups (Harris Laboratories, 1993; Sourby, 1993; Finley et al., 1994c).

(ii) Effects of stearic acid-rich and palmitic acid-rich triacylglycerols on plasma lipid concentrations

In a study submitted for evaluation by the Committee at its present meeting, 20 middle-aged volunteers (12 men and 8 women) who had had a high plasma cholesterol concentration during the preceding year were invited to participate in a study to compare the effects on plasma lipids of diets containing margarines enriched with either salatrim or palm oil (rich in palmitic acid). The study included an initial 2-week period on a low-fat diet and 5 weeks on each high-fat diet, with a cross-over design. The average target intake of lipid was 30 g/day, but the actual intake was not specified. Weighed food records were maintained for 3 days in each period. Five volunteers failed to keep adequate records and were therefore excluded from the final analysis. The subjects’ plasma total, high-density and low-density lipopotein cholesterol and triacylglycerol concentrations did not differ significantly with the salatrim and palm oil diets. None of the volunteers reported adverse symptoms such as bowel irritation, although procedures for assessing such symptoms were not described in the paper (Nestel et al., 1998).

Although the above study was designed to provide information on blood lipid concentrations, additional analyses of serum enzymes were performed and reported to the Committee on Toxicity of Chemicals in Food, Consumer Products and the Environment (1999) in the United Kingdom. The data were confidential and have not been published. It was concluded that consumption of high doses of salatrim had resulted in slight increases in AST and ALT activities in serum but that the increases were within the normal reference range.

(iii) Effect of a single intake of 23SO oil, cocoa butter or high-oleic sunflower oil on serum triacylglycerols

Thirty-five healthy adults (17 men and 18 women) participated in a randomized cross-over trial to compare the effects of a single intake of experimental meals containing 30 g of 23SO salatrim, cocoa butter or high-oleic sunflower oil. The volunteers were allocated randomly to six treatment sequences, the three test meals being consumed 1 week apart over 3 weeks, and they were asked to avoid foods with a high fat content the day before and to fast overnight before each test meal. Blood samples were taken immediately before and 3 and 6 h after the test meal. A smaller post-prandial increase in serum triacylglycerol was found after the salatrim meal than after intake of oleate or cocoa butter meals. Possible side-effects were not investigated (Sanders et al., 2001).

(iv) Effect of 30 g of 4SO oil on plasma cholesterol, triglycerides, clotting parameters and butyric acid

In an unpublished study, five men consumed 30 g of 4SO salatrim (equal to 9 g of butyric acid) in cupcakes, with a fat-filled milkshake to balance the fatty acid intake. Blood samples were taken every 20 min for 6 h for measurement of plasma cholesterol, triglycerides, clotting parameters and butyric acid. The authors reported that the post-prandial lipaemia was approximately 50% lower than expected from normal fat under these conditions, but there were no concurrent controls. The volunteers reported no gastrointestinal complaints or marked feelings of satiation.

The Advisory Committee on Novel Foods and Processes in the United Kingdom expressed concern that ingestion of butyrate-containing salatrim might increase the butyrate content of the blood, as certain short-chain fatty acids have teratogenic potential in vitro (Coakley et al., 1986; Auerbach & Frier, 1999). In a study conducted in response to that concern, free butyrate was not detected in the plasma of these men (detection limit, 1 µmol/l) at any time. It was concluded that ingestion of salatrim containing butyrate would have no effect on serum butyrate concentrations. Although the Committee would have preferred that a positive control had been included in which butyrate was detected in plasma, it agreed that butyrate from salatrim is highly unlikely to have teratogenic effects in humans with normal consumption (Pronczuk et al., 2001).

(v) Response to a single breakfast meal

Another unpublished paper reported two studies involving four male and four female volunteers with normolipidaemia, who were fed a single breakfast meal supplemented with either 30 g of 23SO salatrim or cocoa butter in a cross-over design. The second study addressed the response of five men given a 4SO salatrim meal alone. In the first study, the women complained of feeling ‘bloated, cramped or even nauseated’ 2–4 h after the meal, with recovery by 7 h. One also complained of general gastric discomfort after the cocoa butter. The men did not experience discomfort but noted that they felt satiated and averse to eating until the early evening after the salatrim breakfast. The post-prandial lipaemia was again about 50% lower after the salatrim meal. Less satiation was felt in the second study. There was no concurrent control group, but the post-prandial lipaemia was reported to be lower than that in the study with salatrim 23SO (Hayes et al., 2001).

The data on projected intake of salatrim submitted to the Food and Drug Administration of the USA for affirmation of ‘generally recognized as safe’ status were made available to the Committee at its forty-ninth meeting: consumption at the 90th percentile for all ages and for 3–5-year-olds was reported to be 37 and 26 g/day, respectively.

The data submitted to the Food and Drug Administration were resubmitted with small revisions and complemented with projections of intake based on consumption data for the United Kingdom. In the USA, salatrim is used as a fat replacer on a weight-equivalent basis within good manufacturing practice in confectionery, fine bakery wares, dairy products, salty snacks, chips, nuts, margarine and fat-based spreads (Food and Drug Administration, 1994). The level of use ranges from 1–2 g/100 g in some dairy products and crackers to > 80 g/100 g in margarine. Use of salatrim to replace current fats would not introduce new fatty acids but would result in a different profile of fatty acids, with more saturated acids (mainly stearic and butyric). Data on the estimated intake of salatrim, stearic acid and total saturated fatty acids are thus presented (Table 4).

Table 4. Estimate of intake of salatrim and of fatty acids on basis of individual surveys

Substitution of all added fats with salatrim (worst case) would thus lead to a twofold increase in saturated fats and a threefold increase in fats with 18 carbons. In the USA, the highest intake of salatrim would be that of adolescents aged 13–17 (30 g/day at the 90th percentile) and that of children aged < 2 years on a grams per kilogram body weight basis (1.5 g/kg bw at the 90th percentile) (Finley, 1993).

The main contributors to the mean estimated intake of salatrim according to the simulations for the United Kingdom (Tennants & Douglass, 1999) and the USA (Finley, 1992) would be table margarine, followed by buns and pastries. In the USA, chocolate and confections were the individual food categories with the greatest single intake at the 90th percentile, on the basis of both grams per day and grams per kilogram body weight. Substitution of fats with salatrim in this food category would lead to intake of 37 g/day of salatrim by adolescent boys and 1.1 g/kg bw by children aged 1–6 years at the 90th percentile (Finley, 1993).

Toxicological data

The Committee noted that the newly submitted study of nutrition in rats did not resolve the issue of the extent to which salatrim might reduce energy intake since, as stated previously, the rat is not a good model for humans with respect to stearic acid absorption.

The aim of the home-based trials was to investigate plasma lipid profiles in small groups of adults ingesting salatrim at 30 g/day for up to 5 weeks. Diets containing salatrim did not affect the blood concentration of total high-density or low-density lipoprotein cholesterol. The post-prandial increase in serum triglyceride concentration was lower after ingestion of a meal containing 30 g of salatrim than after ingestion of a meal containing 30 g of oleate or cocoa butter. In one study, it was reported that some of the volunteers complained of gastric discomfort, nausea or feeling ‘bloated’ or ‘satiated and averse to eating until the early evening’ after a breakfast containing 30 g of salatrim. However, adequate controls were not included in this study, and it was not clear if these subjective symptoms were related to ingestion of salatrim. The Committee was informed that, in one of these trials, serum liver enzyme activities were found to be within the normal range; however, these results were not included in the published manuscript.

Overall, the Committee concluded that the new studies did not respond to the questions raised at its forty-ninth meeting and the available information was inadequate for toxicological evaluation. The Committee noted that, in general, toxicological studies in experimental animals should be conducted at doses that allow an adequate margin of safety in relation to anticipated human intake. Such safety margins cannot feasibly be attained with substances, such as salatrim, that are intended for use as major components of food. Additionally the rat is not a good model for humans with respect to stearic acid absorption, and further studies in experimental animals would not contribute to the safety evaluation of salatrim. The Committee therefore paid particular attention to the results of the studies in humans. These studies indicated that adverse effects (gastrointestinal and hepatic effects) might be experienced by adults consuming diets containing salatrim at 60 g/day (equivalent to 1 g/kg bw per day). The available information was not adequate to evaluate whether effects might occur at intakes of less than 60 g/day, nor if the effects differ with the composition of salatrim.

Intake

Data on the projected intake of salatrim were assessed by the Committee at its forty-ninth meeting. These data, with small revisions, and with projections of intake based on data on consumption in the United Kingdom, were submitted to the present meeting. Manufacturers’ use levels for salatrim vary from 1–2 g/100 g in some dairy products and crackers to more than 80 g/100 g in margarine.

The mean intake of salatrim after hypothetical substitution of added fats by this substance in all food categories would be 45 g/day, but this value is unrealistic. The mean intake of salatrim based on hypothetical substitution of fats in specified food categories would vary from 11 to 13 g/day. Intake by adolescent consumers at the 90th percentile would reach 35 g/day. On a gram per kilogram body-weight basis, the highest intake would be that of children aged less than 2 years, as those at the 90th percentile of consumption would have an intake of 1.5 g/kg bw per day. Children aged 1–6 years who are consumers of the category ‘chocolates and confectionery’ at the 90th percentile would have an intake of salatrim from this food category alone of 1.1 g/kg bw per day, which is higher than the intake reported to have effects in adults in clinical studies.

The Committee concluded that the available data did not provide an adequate basis for evaluating the safety of salatrim and recommended that additional studies be performed in humans to assess fully the safety of salatrim covering the range of compositions intended for use. The studies should be designed appropriately for safety evaluation and should include investigation of the gastrointestinal and hepatic effects observed in previous studies.

Auerbach, M.H. & Frier, H.I. (1999) Analysis of the teratogenic potential of propionate and butyrate vs likely levels of short chain organic acid levels in sytemic circulation arising from consumption of salatrim. Manuscript provided by Danisco Cultor USA, Inc., Ardsley, New York, USA.

Besselaar Clinical Research Unit (1993a) Randomised, 3-way crossover, double blind tolerance study of fat replacement compound TAG A9300 versus soybean oil administered to non-sedentary subjects by substituting 30 g/day or 60 g/day at 1800 or 2500 kcal/day diets. Unpublished report No. 8024 from Besselaar Clinical Research Unit. Submitted to WHO by Cultor Food Science, Ardsley, New York, USA).

Besselaar Clinical Research Unit (1993b) Randomized double blind study to determine the effects of a single administration of fat replacement salatrim 23SO at four different levels compared to three levels of control fat compounds on serum acetotacetate, acetate, beta-hydroxybutyrate and glucose. Unpublished report No. 8165 from Besselaar Clinical Research Unit. Submitted to WHO by Cultor Food Science, Ardsley, New York, USA.

Besselaar Clinical Research Unit (1993c) Statistical report. Randomized-double blind safety and tolerance study of fat replacement compound TAG 8200 versus coconut oil administered at 60 g (2500 kcal) and 45 g (1800 kcal) utilising an acute (1-day) crossover phase and a chronic (7-day) non-cross-over phase. Protocol GHBA-105, 28 June 1993. Submitted to WHO by Cultor Food Science, Ardsley, New York, USA.

Besselaar Clinical Research Unit (1993d) Clinical report. Randomized-double blind safety and tolerance study of fat replacement compound TAG 8200 versus coconut oil administered at 60 g (2500 kcal) and 45 g (1800 kcal) utilising an acute (1-day) crossover phase and a chronic (7-day) non-cross-over phase. Unpublished report No. 6270-164 from GHBA/Hazelton, dated 27 October 1993. Submitted to WHO by Cultor Food Science, Ardsley, New York, USA.

Coakley, M.E., Rawlings, S.J., & Brown, N.A. (1986) Short-chain carboxylic acids, a new class of teratogens: Studies of potential biochemical mechanisms. Environ. Health Perspectives, 70, 105–111.

Committee on Toxicity of Chemicals in Food, Consumer Products and the Environment (1999) Statement for the Advisory Committee on Novel Foods and Processes on toxicological aspects of a submission on short and long chain triacylglycerol molecules (salatrims)—A family of low calorie fats submitted for approval under the EC Novel Food Regulation. Department of Health, London, England.

Danisco Cultor USA, Inc. (2001a) Salatrim, 29 November 2001. Submitted to WHO by Danisco Cultor USA, Inc., Ardsley, New York, USA.

Danisco Cultor USA, Inc. (2001a) Salatrim, attachment B, 30 November 2001. Submitted to WHO by Danisco Cultor USA, Inc., Ardsley, New York, USA..

Finley, J.W. (1992) Estimation of dietary exposure (of) stearic acid from salatrim. Unpublished report from Nabisco Foods Group, East Hanover, New Jersey, USA. Submitted to WHO by Danisco Cultor USA, Ardsley, New York, USA.

Finley, J.W. (1993) The potential exposure of salatrim in the diet. Unpublished report from Nabisco Foods Group, East Hanover, New Jersey, USA. Submitted to WHO by Danisco Cultor USA, Ardsley, New York, USA.

Finley, J.W. & Leveille, G.A. (1993) Exposure assessment of various foods and food categories based on the 1987–88 USDA NFCS database. Unpublished report from Nabisco Foods Group, East Hanover, New Jersey, USA. Submitted to WHO by Danisco Cultor USA, Ardsley, New York, USA.

Finley, J.W., Leveille, G.A., Dixon, R.M., Walchak, C.G., Sourby, J.C., Smith, R.E., Francis, K.D. & Otterburn, M.S. (1994a) Clinical assessment of salatrim, a reduced-calory triacylglycerol. J. Agric. Food Chem., 42, 581–596.

Finley, J.W., Klemann, L.P., Leveille, G.A., Otterburn, M.S. & Walchak, C.G. (1994b) Caloric availability of salatrim in rats and humans. J. Agric. Food Chem., 42, 495–499.

Finley, J.W., Walchak, C.G., Sourby, J.C., & Leveille, G.A. (1994c) Clinical study of the effects of exposure of various salatrim preparations to subjects in a free-living environment. J. Agric. Food Chem., 42, 597–604.

Finley, J.W., Leveille, G.A., Klemann, L.P., Sourby, J.C., Ayres, P.H. & Appleton, S. (1994d) Growth method for estimating the caloric availability of fats and oils. J. Agric. Food Chem., 42, 489–494.

Food and Drug Administration (1994) GRAS Affirmation Petition GRP 4G0404, Washington DC, USA, Government Printing Office.

Harris Laboratories (1993) Clinical report. Randomized, crossover, double blind tolerance study of salatrim fat replacements. A 7 day pre-exposure to control soybean oil followed by a 28 day exposure to salatrim oils or control soybean oil followed by a 7 day exposure to control soybean oil (Project No. 15472 7). Unpublished report from Harris Clinical Laboratories, Lincoln, Nebraska, USA. Submitted to WHO by Cultor Food Science, Ardsley, New York, USA.

Hayes, J.R. & Riccio, E.S. (1994) Genetic toxicology studies of salatrim structured triacylglycerols. 1. Lack of mutagenicity in Salmonella/microsome reverse mutation assay. J. Agric. Food Chem., 42, 515–520.

Hayes, J.R., Pence, D.H., Scheinbach, S., D’Amelia, R.P., Klemann, L.P., Wilson, N.H. & Finley, J.W. (1994a) Review of triacylglycerol digestion, absorption, and metabolism with respect to salatrim triacylglycerols. J. Agric. Food Chem., 42, 474–483.

Hayes, J.R., Finley, J.W. & Leveille, G.A. (1994b) In vivo metabolism of salatrim fats in the rat. J. Agric. Food Chem., 42, 500–514.

Hayes, J.R., Wilson, N.H., Pence, D.H. & Williams, K.D. (1994c) Subchronic toxicity studies of salatrim structured triacylglycerols in rats: 2. Triacylglycerols composed of stearate, acetate, and propionate. J. Agric. Food Chem., 42, 539–551.

Hayes, J.R., Wilson, N.H., Pence, D.H. & Williams, K.D. (1994d) Subchronic toxicity studies of salatrim structured triacylglycerols in rats: 1. Triacylglycerols composed of stearate and butyrate. J. Agric. Food Chem., 42, 528–538.

Hayes, J.R., Wilson, N.H., Roblin, M.C., Mann, P.C. & Kiorpes, A.L. (1994e) 28-day continuous dosing study in minipigs with a salatrim structured triacylglycerol composed of stearate, acetate, and propionate. J. Agric. Food Chem., 42, 563–571.

Hayes, J.R., Rudd, C.J., Mirsalis, J.C., Bakke, J.P., Winegar, R.A. & Murli, H. (1994f) Genetic toxicology studies of salatrim structured triacylglycerols. 2. Lack of genetic damage in in vitro mammalian cell assays and the in vivo micronucleus assay. J. Agric. Food Chem., 42, 521–527.

Hayes, J.R., Wilson, N.H., Pence, D.H. & Williams, K.D. (1994g) Subchronic toxicity studies of salatrim structured triacylglycerols in rats: 3. Triacylglycerols composed of stearate, acetate, propionate, and butyrate. J. Agric. Food Chem., 42, 552–562.

Hayes, K.C., Lipinski, B., Pronczuk, A. & Nestel, P. (2001) Structured trigyceride substitute for cocoa butter (salatrim) reduces postprandial lipemia and improves the acute, but not chronic, hemostatic profile in humans. Manuscript provided by Danisco Cultor Inc., Ardsley, New York, USA.

Kiorpes, A.L. (1993a) Final report: Exploratory study with A9300 in rats. Unpublished report No. HWI 6270-172 from Hazelton, Wisconsin, USA. Submitted to WHO by Cultor Food Science, Ardsley, New York, USA.

Kiorpes, A.L. (1993b) Final report: 28 day dietary toxicity study with A9300 in Hanford minipigs. Unpublished report No. HWI 6270-174 from Hazelton, Wisconsin, USA. Submitted to WHO by Cultor Food Science, Ardsley, New York, USA.

Musick, T. & Peterson, M. (1993) Tissue distribution and excretion of radiolabel following a single oral dose of ¹⁴C-triacylglycerol in rats. Unpublished report No. HW 6270-166 from Hazelton, Wisconsin, USA. Submitted to WHO by Cultor Food Science, Ardsley, New York, USA.

Nestel, P.J., Pomeroy, S., Kay, S., Sasahara, T. & Yamashita, T. (1998) Effect of a stearic acid-rich, structured triacylglycerol on plasma lipid concentrations. Am. J. Clin. Nutr., 68, 1196–1201.

Phillips, K.D. (1992) In vitro hydrolysis of emulsified TAG A8200 by porcine pancreatic lipase. Unpublished report No. CR001-KMP-003: Final report from Tech Lab VPI Research Park, Blacksburg, Virginia, USA. Submitted to WHO by Cultor Food Science, Ardsley, New York, USA.

Pronczuk, A., Lipinski, B. & Hayes, K.C. (2001) Butyrate-rich salatrim induces minimal lipemia without altering fibrinogen or fibrinolysis in men. Submitted to WHO by Cultor Food Science, Ardsley, New York, USA.

Sanders, T.A.B., Oakley, F.R., Cooper, J.A. & Miller, G.J. (2001) Influence of a stearic acid-rich structured triacylglycerol on postprandial lipemia, factor VII concentrations, and fibrinolytic activity in healthy subjects. Am. J. Clin. Nutr., 73, 715–721.

Scheinbach, S., Hayes, J.R., Carman, R.J., Zhou, D., Van Tassell, R.L. & Wilkins, T.D. (1994) Effects of structured triacylglycerols containing stearic, acetic, and propionic acids on the intestinal microflora of rats. J. Agric. Food Chem., 42, 572–580.

Sequeria, A. & Gordon, B.M. (1993) In vitro pancreatic lipase hydrolysis of salatrim; cocoa butter, and triolein: Quantification by HTGC/F1D. Unpublished report from Experimental Pathology Laboratory, Inc., dated 17 March 1993. Submitted to WHO by Cultor Food Science, Ardsley, New York, USA.

Sourby, J.C. (1993) Statistical report. Randomised, crossover, double blind tolerance study of salatrim fat replacements. A 7 day pre-exposure to control soybean oil followed by a 28 day exposure to salatrim oils or control soybean oil followed by a 7 day exposure to control soybean oil. Salatrims under study are 23SO at exposure levels of 30 g, 45 g, and 60 g, salatrim 4SO at 60 g, and salatrim 43SO at 60 g compared to a 60 g exposure level of control soybean oil. Unpublished report from Nabisco Foods, East Hanover, New Jersey, USA .Submitted to WHO by Cultor Food Science, Ardsley, New York, USA.

Tennants, D.R. & Douglass J.S. (1999) Estimation of intakes of fatty acids from the UK diet. Unpublished report from Cultor Food Science. Submitted to WHO by Danisco Cultor USA, Ardsley, New York, USA.

Williams, K.D. (1992a) Final report: 13 week dietary toxicity study with A7200 in rats. Unpublished report No. HW 6270-157 from Hazelton, Wisconsin, USA. Submitted to WHO by Cultor Food Science, Ardsley, New York, USA.

Williams, K.D. (1992b) Final report: 13 week dietary toxicity study with A8200 and A8200 in rats. Unpublished report No. HWI 6270-162 from Hazelton, Wisconsin, USA. Submitted to WHO by Cultor Food Science, Ardsley, New York, USA.

Williams, K.D. (1992c) Final report: 13 week dietary toxicity study with A9100 and A7800 in rats. Unpublished report No. HW 6270-168 from Hazelton, Wisconsin, USA. Submitted to WHO by Cultor Food Science, Ardsley, New York, USA.

Salatrim is a family of structured triglycerides prepared with combinations of short- and long-chain fatty acids, which can serve as a total replacement for conventional fats and oils in many foods. It has been claimed that it provides about half the energy content of the fats and oils it replaces. An evaluation was provided of the caloric availability of salatrim triglycerides (Howlett, 1997), and the supporting materials were reviewed by the Committee at its forty-ninth meeting (Annex 1, reference 131). At that time, the Committee concluded that the available studies did not provide an adequate basis for evaluating the caloric availability of salatrim, since a major justification of the purportedly lowered caloric availability of salatrim was reduced absorption of stearic acid, and the rat is not a good model for stearic acid absorption. In addition, it considered that it was not possible to assign a single caloric value to salatrim as there are many different compositions. The Committee recommended that appropriate studies be conducted in humans. In response, one additional study of nutrition in rats was performed and is summarized below; furthermore, corrections of typographical and transcription errors in the reports and publications provided for the discussion at the forty-ninth meeting were provided to the Committee at its present meeting (Danisco Cultor USA, Inc., 2001a).

According to the specifications, there is no specific requirement for the amounts of short-chain fatty acids and stearic acid in the final product, although a request has been made (Auerbach, 2000) to limit salatrim compositions to those with  70% total saturated long-chain fatty acids. Unless the composition is specified, it is not possible to assign a single caloric value to salatrim.

The lower energy content of salatrim triglycerides, in comparison with conventional fats and oils, is purportedly due to the lower caloric value of short-chain fatty acids and the reduced absorption of stearic acid. The following discussion evaluates the evidence for these two purported effects.

A3.1 Caloric value of short-chain fatty acids

It is claimed that acetic acid, propionic acid and butyric acid contribute fewer kilocalories per gram than longer-chain fatty acids. On the basis of their heats of combustion (Weast, 1992–93), these fatty acids contribute 3.5 kcal/g (acetate), 4.9 kcal/g (propionate) and 5.8 kcal/g (butyrate), while the standard physiological fuel value of lipids is 9 kcal/g. The Committee agreed with that statement, but noted that the fatty acid composition of salatrim has not been designated and therefore no specific caloric value can be attributed to it. Also, since short-chain fatty acids contribute less energy per gram than long-chain fatty acids, it can be predicted that the energy content of a salatrim product would decrease as the short-chain fatty acid:long-chain fatty acid in a salatrim product increased. In fact, in the studies that related the ratio of short- to long-chain fatty acids in salatrim to weight gain in rats, the opposite was seen. Thus, as the amount of short-chain fatty acid in salatrim increased, the energy contribution of that salatrim product increased. A ratio of 0.51 for short-chain:long-chain fatty acid was considered to provide 2.6 kcal/g, whereas a ratio of 1.99 short-chain:long-chain fatty acid was considered to provide 6.4 kcal/g (Klemann et al., 1994). In general, short-chain fatty acids provide fewer kilocalories per gram than long-chain fatty acids; however, the ratio of short-chain: long-chain fatty acids in a triglyceride may also influence its caloric availability, and, unless a salatrim product is specified, an overall energy content cannot be assigned.

A3.2 Caloric value of stearic acid

On the basis of its heat of combustion (Weast, 1992–93), stearic acid should provide 9.5 kcal/g. It has been contended that stearic acid in salatrim contributes much less energy because stearic acid is poorly absorbed (Klemann et al., 1994). Studies on the absorption of stearic acid are described below.

Rats

Reduced stearate absorption has been reported in studies of the absorption: excretion balance in rodents and humans.

Four groups of 10 rats were given a NIH-07 diet supplemented with 10% corn oil (controls) or with 5, 10 or 15% salatrim 23SO. The rats were acclimated to the diets for 5 days, and then faeces were collected for 5 days. Stearic acid was determined in the faeces by method Ce 1-62 of the American Oil Chemists’ Society (1990), but the method for determining intake of stearic acid was not described. The rats given the 5% salatrim diet consumed 0.56 g of stearic acid per day and excreted 0.40 g in the faeces, representing absorption of 29%. The rats given the 10% salatrim diet consumed 1.0 g of stearic acid per day and excreted 0.67 g in the faeces, representing absorption of 35%. The rats given the 15% salatrim diet consumed 1.5 g of stearic acid and excreted 1.0 g, for a total absorption of 32% (Finley et al., 1994a). Stearic acid appeared to be poorly absorbed from salatrim 23SO by rats in this assay, but the actual absorption is unknown as the method for determining stearate intake was not described and no account was taken of the microbial contribution to stearate formation from other 18-carbon fatty acids. Furthermore, this study of stearic acid excretion was performed with only one salatrim formulation. Data for various salatrim products strongly suggest that stearic acid absorption depends on the ratio of short-chain:long-chain fatty acids in the product (Klemann et al., 1994).

Radiolabelled salatrim fats that mimic salatrim 23CA were synthesized and purified and designated salatrim APS. [¹⁴C]Triolein was used as the reference fat. Unlabelled salatrim 23CA was used to dilute the radiolabelled salatrim and triolein to the appropriate specific activity. Groups of five male Sprague-Dawley were given radiolabelled salatrim APS or triolein at a single dose of 1.4 g/kg bw by gavage. After dosing, they were housed individually in glass metabolism cages designed for collection of expired carbon dioxide, urine and faeces. One group of rats received a diet containing 10% salatrim 23CA by weight for 2 weeks before administration of the radiolabelled fats to see if pre-feeding the fat influenced its disposition. After a 72-h sample collection period, the rats were killed, and radiolabel was determined in duplicate samples of all tissues, urine, faeces and carbon dioxide. Faecal samples from the rats fed salatrim APS with radiolabelled stearate were analysed for recovery of stearate, and the amounts of radiolabel from triolein (control fat) and from the stearate carbonyl in salatrim in faecal material were plotted over time.

Although the amount of label from stearate was numerically higher than that from oleate in faeces at 12 h, the difference was not statistically significant. As the percentage of the radiolabelled dose recovered from oleate was 38% and that from stearate was 55%, oleate in triolein appeared to have been 62% absorbed, whereas stearate in salatrim 23CA was 45% absorbed. In other comparisons of the calculated caloric value of various salatrim products and other fats, however, an absorption coefficient of 99% was used for oleic acid (Klemann et al., 1994). The advantage of the study with radiolabelled stearate over that of stearate balance without radiolabel, described above and by Finley et al. (1994a), was that excretion of stearate was not overestimated in the first study because microbially derived stearate was not measured. The disadvantage of the study with radiolabel was that it involved only five rats per group instead of 10. However, when an additional five rats per group, differing from the others only in that they were pre-fed salatrim 23CA for 14 days before dosing, were also tested for excretion of labelled stearate, the values were similar to those for animals fed the basal diet or salatrim 23CA, which justifies pooling of the data on excretion. These results suggest that the stearic acid in salatrim 23CA is approximately 45% absorbed in male rats (Hayes et al., 1994).

Humans

A study was conducted without a cross-over design, in which two groups of 18 volunteers consumed products containing hydrogenated coconut oil for 7 days and then received salatrim 23CA at either 45 or 60 g/day, depending on their total caloric need (1800 kcal/day or 2500 kcal/day), or products made with hydrogenated coconut oil. During a final 10-day period, all volunteers received products made with hydrogenated coconut oil. Faecal samples were collected from each person during the last 3 days of each 7-day test period and pooled. Persons consuming 1800 kcal/day ingested 27 g/day of stearic acid (method for arriving at this value not specified) and excreted 7.6 g/day, for a net absorption of 72%. Those consuming 2500 kcal/day ingested 34 g/day of stearic acid and excreted 12 g/day, for a net absorption of 64%. The authors concluded that 64–72% of the stearic acid in salatrim was absorbed, resulting in an apparent caloric availability of 4.7–5.1 kcal/g (Finley et al., 1994a). A recent report (Danisco Cultor USA, Inc., 2001b) indicated slightly different values, of 65–74%, for the absorption of stearic acid from salatrim in the same study. The way in which the amount of stearic acid consumed in the diets was determined was not described. Without an accurate determination of the amount consumed, it is difficult to assess stearic acid absorption. Also, no data on the weight gain of the volunteers were presented. Thus, no conclusions can be drawn about the caloric value of salatrim from this study.

Olubajo et al. (1986) reported that the absorption of stearic acid in a study of 30 men aged 34–61 years was 82–88%. Jones et al. (1985) measured the absorption of ¹³C-labelled stearic, oleic and linoleic acids in six healthy men and reported absorption coefficients of 78% for stearate, 97% for oleate and 100% for linoleic acid. In an investigation among in-patients in a metabolic ward, Denke & Grundy (1991) fed four different fats (butter, beef tallow, cocoa butter and olive oil) as part of liquid diets. Each diet was fed for 3 weeks. The dietary intakes and faecal excretion rates of three major fatty acids (palmitic, stearic and oleic) were determined and used to estimate their absorption. The highest absorption rate was that of oleic acid (approximately 99%); palmitic acid was absorbed at a rate of 95–97%, whereas the absorption of stearic acid was 90–94%. These authors made an important observation about the method for measuring stearic acid absorption. Their initial analyses were based on a method in which the fatty acid content of a sample was determined by extracting total lipids from a faecal sample and then determining the proportion of the lipid sample that was found to be stearic acid by gas chromatography. They stated, however, that faecal fatty acids could be overestimated by 100% when calculated as the percentage of extractable lipid weight. They then measured the actual mass of stearic acid by adding a known amount of 17:0 to the sample.

Bonanome & Grundy (1989) evaluated the absorption of stearic acid, relative to other fatty acids, in a group of 10 volunteers fed a meal with a high or a relatively low stearic acid content. Plasma chylomicrons were isolated 2,4, 6 and 8 h after ingestion of the meals. The fatty acid patterns of chylomicron lipids were determined and compared with those of the ingested lipids. The percentages of palmitic acid (16:0) and stearic acid, relative to other fatty acids, were only slightly lower in the lipids from chylomicrons than in those in the meal. The authors concluded that the absorption of stearic acid is similar to that of palmitic acid and that both are absorbed almost as well as oleic acid.

Stearic acid absorption in humans may thus be as low as 64–72%, as reported by Finley et al. (1994a), or, more realistically, as high as 90–94% (Bonanome & Grundy, 1989; Denke & Grundy, 1991). The rat is apparently not a good model for humans with respect to stearic acid absorption, since most studies on rats showed lower coefficients of absorption of stearic acid. Danisco Cultor USA, Inc. (2001b) have stated that stearic acid in salatrim is less well absorbed than stearic acid in conventional fats and oils, the proposed mechanism being the rapid hydrolysis of short-chain fatty acid from the triglyceride, leaving a higher melting-point lipid which is less likely to form micelles and more likely to form indigestible calcium or magnesium soaps. However, there are no experimental data to support this hypothesis.

A3.3 Caloric value of salatrim products in an assay for rat growth

The main method by which the caloric value of various salatrim products was determined was a rodent growth assay, the basic protocol of which was described by Finley et al. (1994b). A later modification to this protocol is described below. In the first set of experiments, a basal NIH-07 diet was given to groups of 10 rats, with daily feed consumption restricted to 50% of that of rats fed ad libitum. All rats consumed the same amount of basal diet containing different amounts of corn oil or test fat. A regression curve was calculated for the weight gain of the rats fed corn oil (considered to supply 9 kcal/g), and the caloric value of salatrim was calculated from a comparison of the weight gain of the animals fed salatrim and those fed corn oil.

The rationale behind use of this assay was that rats are an appropriate model for humans and that balance studies were not conducted because they are ‘cumbersome and do not lend themselves to evaluating large numbers of materials’; studies with radiolabelled materials were not conducted because they are ‘expensive and time-consuming and are not practical as a routine screening tool’ (Finley et al., 1994b). The rodent growth assay has, however, a number of limitations for determining the caloric content of salatrim products for humans. First and most important, no evidence was provided that humans receiving equivalent amounts of salatrim substituted for traditional fats had decreased weight gain. As shown above, stearic acid appears to be less well absorbed in rats than in humans, and thus studies in rats are not appropriate for determining the caloric content for humans of triglycerides containing stearic acid. In addition, restriction of the food intake of the rats to 50% of that of rats fed ad libitum may not result in normal growth. Finally, the test fats were added to the basal diet, whereas it would have been more suitable to adjust all diets to maintain similar ratios therefore not the only variable in growth.

In summary, it must be questioned whether the growth assay in rats provides meaningful data for determining the caloric contribution of salatrim in human diets.

In a further unpublished study, the diets were less restrictive and allowed 60–80% normal weight gain as compared with 50% in the study of Finley et al. (1994b), and the test fats were blended with a defined fat in a 3:1 ratio, obviating any concern about essential fatty acid deficiency. Further, the ratios of protein plus carbohydrate:fat, fat:vitamin and fat:mineral were kept constant. The average caloric value of the four salatrims tested was 5.2 kcal/g during the 4-week feeding period. This study confirmed the finding of Finley et al. (1994b) that the higher the ratio of short-chain:long-chain fatty acids the higher the caloric value. As described above, this is counterintuitive, as short-chain fatty acids are known to have a lower energy content per gram than long-chain fatty acids. Also, in this study, the amount of stearic acid excreted in the faeces was not negatively correlated with the energy per gram of test fat, as would be expected. In fact, the diet with 82% dilong-18:0 yielded a low caloric value but was associated with low fecal 18:0 and normal total fat excretion. The results of this study confirm that the energy contribution of a particular salatrim fat cannot be predicted by assigning an absorption coefficient for stearic acid and a standard low value for short-chain fatty acids (Treadwell et al., 2001).

A3.4 Caloric value of salatrim products on the basis of stearic acid absorption coefficient

The 14-day rat growth assay was used to determine the absorption coefficient for stearic acid. Eleven compositions of salatrim were generated by inter-esterification of various starting molar ratios of tributyrin and hydrogenated canola oil. The molar short-chain:long-chain ratios of the compositions varied between 0.51 and 1.99. The caloric content of these 11 salatrim 4CA samples was determined in rats fed 50% of the basal diet of the control rats plus 21% salatrim by weight of the diet. Weight gain was calculated over 14 days, and the weight gain of rats on each of the 11 salatrim diets was compared with that of rats on 21% corn oil, considered to supply 9 kcal/g. A previously developed regression equation (Finley et al., 1994c) was used to estimate the number of kilocalories per gram for each salatrim product by comparing it to a standard curve generated with various amounts of corn oil added to the basal diet.

Stearic acid absorption was estimated from the energy (in kcal/g) of the various salatrim products in the rodent growth assay and the ratio of short-chain:long-chain fatty acids in the test salatrim products. A table was constructed of the assayed composition of a salatrim product with a ratio of 0.51:1 for short-chain:long-chain fatty acids. For each portion of the triglyceride, the component of the triglyceride, the mass fraction of that component, the gross energy from the study in rats and the absorption coefficient of that component were provided. All the absorption coefficients were derived from the literature, and the only ‘unknown’ was considered to be the absorption of stearic acid. An absorption coefficient for stearic acid was derived from this matrix, the 11 ratios of short-chain:long-chain fatty acid and the estimated caloric values for the various products. The absorption coefficient ranged from 0.15 with the lowest ratio to 0.70 with the highest ratio (Klemann et al., 1994).

These calculations of the absorption of stearic acid reinforced the Committee’s view that a general caloric value cannot be applied to salatrim in the absence of knowledge of the specific fatty acid composition of the product.

A4. Other nutritional considerations: Potential cholesterol-raising effect of stearic acid

Although not considered by the Committee at its forty-ninth meeting, a review by the Committee on Toxicity of Chemicals in Food, Consumer Products and the Environment (1999) of the European Commission raised the issue that the stearic acid in salatrim could potentially increase blood cholesterol concentrations. This potential effect was addressed in four clinical trials, reviewed in section 2.3 above. Nestel et al. (1998) compared the effect of a stearic acid-rich diet (23SO) with that of a diet rich in palmitic acid in 15 volunteers in a cross-over study, each dietary period being 5 weeks. The outcome measures were total, low-density lipoprotein and high-density lipoprotein cholesterol and plasma triacylglycerols concentrations. The dietary intervention followed a 2-week period in which the volunteers were instructed to eat low-fat diets. Although the values for total cholesterol in the usual diets were higher than those in the low-fat diet, they did not differ significantly in the palmitic acid- and stearic acid-rich diets as compared with the low-fat diet. Furthermore, the values for triglycerides in the palmitic acid- and strearic acid-rich diets were similar. The Committee found these surprising, as it has been widely reported that cholesterol concentrations are increased by high-fat as compared with low-fat diets. Also, although some controversy exists about whether dietary stearic acid is hypercholesterolaemic, the cholesterol-raising effects of palmitic acid are well documented. The fact that neither high-fat diet raised cholesterol concentrations over those with low-fat diets, and in particular the finding that palmitic acid fed at high levels resulted in cholesterol values similar to those with a low-fat diet, suggests a lack of power to detect such differences in a sample of 15 volunteers. The report that stearic acid from salatrim does not raise plasma cholesterol levels is therefore considered not to be definitive.

Sanders et al. (2001) compared the same salatrim product (23SO) with meals enriched with cocoa butter or high-oleate sunflower oil in 35 volunteers in a randomized cross-over design. Each person received three experimental meals 1 week apart over 3 weeks. The increase in plasma triacylglycerol and in factor VII coagulant activity was less marked after consumption of a meal containing 23SO than after a meal enriched with cocoa butter or oleate. The Committee observed that the stearic acid-rich diet and the other two diets contained 77% long-chain fatty acids. As it is the long-chain fatty acids that are incorporated into chylomicrons and would thus be picked up as triglycerides in blood, it is not surprising that the diet with less long-chain fatty acids resulted in 79% less trigylcerides at 3 h than with the other two diets.

American Oil Chemists’ Society (1990) Fatty acid comparison by gas chromatography: Method Ce 1-62. In: Official Methods of the American Oil Chemists’ Society, Champaign, Illinois, USA.

Auerbach, M.H. (2000) Letter submitted to Ms M. de L. Costarrica, Senior Officer, FAO Food and Nutrition Division, requesting a change to the specifications for salatrim. Submitted by Danisco Cultor USA, Inc., Ardsley, New York, USA.

Bonanome, A. & Grundy, S.M. (1989) Intestinal absorption of stearic acid after consumption of high fat meals in humans. J. Nutr., 119, 1556–1560.

Committee on Toxicity of Chemicals in Food, Consumer Products and the Environment (1999) Statement for the Advisory Committee on Novel Foods and Processes on toxicological aspects of a submission on short and long chain triacylglycerol molecules (salatrims)—A family of low calorie fats submitted for approval under the EC Novel Food Regulation. Department of Health, London, United Kingdom.

Danisco Cultor USA, Inc. (2001a) Salatrim, 29 November 2001. Submitted to WHO by Danisco Cultor USA, Inc., Ardsley, New York, USA.

Danisco Cultor USA, Inc. (2001a) Salatrim, Attachment B, 30 November 2001. Submitted to WHO by Danisco Cultor USA, Inc., Ardsley, New York, USA.

Denke, M.A. & Grundy, S.M. (1991) Effects of fats high in stearic acid on lipid and lipoprotein concentrations in men. Am. J. Clin. Nutr., 54, 1036–1040.

Finley, J.W., Klemann, L.P., Leveille, G.A., Otterburn, M.S. & Walchak, C.G. (1994a) Caloric availability of salatrim in rats and humans. J. Agric. Food Chem., 42, 495–499.

Finley, J.W., Leveille, G.A., Klemann, L.P., Sourby, J.C., Ayres, P.H. & Appleton, S. (1994b) Growth method for estimating the caloric availability of fats and oils. J. Agric. Food Chem., 42, 489–494.

Finley, J.W., Leveille, G.A., Dixon, R.M., Walchak, C.G., Sourby, J.C., Smith, R.E., Francis, K.D. & Otterburn, M.S. (1994c) Clinical assessment of salatrim, a reduced-calory triacylglycerol. J. Agric. Food Chem., 42, 581–596.

Hayes, J.R., Pence, D.H., Scheinbach, S., D’Amelia, R.P., Klemann, L.P., Wilson, N.H. & Finley, J.W. (1994) Review of triacylglycerol digestion, absorption, and metabolism with respect to SALATRIM triacylglycerols. J. Agric. Food Chem., 42, 474–483.

Howlett, J. F. (1997) Memorandum to Dr J.L. Herrman, 27 March 1997.

Jones, P.J.H., Pencharz, P.B. & Clandinin, M.T. (1985) Absorption of ¹³C-labeled stearic, oleic, and linoleic acids in humans: Application to breath tests. J. Lab. Clin. Med., 105, 647–652.

Klemann, L.P., Finley, J.W. & Leveille, G.A. (1994) Estimation of the absorption coefficient of stearic acid in salatrim fats. J. Agric. Food Chem., 42, 484–488.

Nestel, P.J., Pomeroy, S., Kay, S., Sasahara, T. & Yamashita, T. (1998) Effect of a stearic acid-rich, structured triacylglycerol on plasma lipid concentrations. Am. J. Clin. Nutr., 68, 1196–1201.

Olubajo, O., Marshall, M.W., Judd, J.T. & Adkins, J.T. (1986) Effects of high and low fat diets on the bioavailability of selected fatty acids, including linoleic acid, in adult men. Nutr. Res., 6, 931–955.

Sanders, T.A.B., Oakley, F.R., Cooper, J.A. & Miller, G.J. (2001) Influence of a stearic acid-rich structured triacylglycerol on postprandial lipemia, factor VII concentrations, and fibrinolytic activity in healthy subjects. Am. J. Clin. Nutr., 73, 715–721.

Treadwell, R.M., Pronczuk, A. & Hayes, K.C. (2001) Dietary stearic acid content and glyceride structure affect caloric value of glycerides in growing rats. Unpublished report. Submitted to WHO by Danisco Cultor USA, Inc., Ardsley, New York, USA.

Weast, R.C., ed. (1992–93) CRC Handbook of Chemistry and Physics, 73rd Ed., Boca Raton, Florida, USA: CRC Press.

    See Also:
       Toxicological Abbreviations
       SALATRIM (JECFA Evaluation)