点击显示 收起
1 From the Osteoporosis Research Center, Creighton University, Omaha.
2 Supported by Creighton University Intramural Research funds. 3 Address reprint requests to RP Heaney, Creighton University, 601 North 30th Street, Suite 4841, Omaha, NE 68131. E-mail: rheaney{at}creighton.edu.
ABSTRACT
Background: Calcium-fortified soy milk is growing in popularity, particularly among vegetarians, but the bioavailability of its calcium was not previously known. Additionally, the validity of isotopic labeling methods for fortified liquid products had not been established.
Objectives: The objectives of this study were to compare the bioavailability of the calcium in fortified soy milk with that of calcium in cow milk and to evaluate the method of labeling soy milk for bioavailability testing.
Design: A within-subject comparison of extrinsically labeled cow milk with intrinsically and extrinsically labeled soy milks was undertaken in 16 healthy men. In all tests, 300-mg Ca loads were given as a part of a light breakfast after an overnight fast. The milks were physically partitioned into liquid and solid phases to enable evaluation of tracer distribution.
Results: Calcium from intrinsically labeled soy milk was absorbed at only 75% the efficiency of calcium from cow milk. Extrinsic labeling of soy milk did not produce uniform tracer distribution throughout the liquid and solid phases and resulted in a 50% overestimate of true absorbability.
Conclusion: Calcium-fortified soy milk does not constitute a calcium source comparable to cow milk, and extrinsic labeling of such calcium particulate suspensions does not produce the uniform tracer distribution needed for bioavailability testing. Hence, intrinsic labeling of the fortificant is required for such liquid suspensions.
Key Words: Calcium • cow milk • soy milk • absorption • bioavailability • isotopic labeling • extrinsically labeled milk • intrinsically labeled milk • healthy men
INTRODUCTION
There is a growing awareness of the importance of maintaining a high calcium intake throughout life, not just for bone health (1) but also for the health of other body systems (2–5). Although dairy products account for 70% of the calcium sources of the US diet, total calcium intake remains inadequate. This realization has led to the voluntary calcium fortification of an expanding number of foods. Particularly pressing is the need to ensure adequate calcium intakes for vegans and for those with milk aversions because the calcium intakes of these groups tend to be lower than those of the general population. For some of these individuals, soy milk (technically, "soy beverage") has been gaining popularity.
Native soy milk contains only 10 mg Ca/serving. Hence, to provide needed calcium, food processors offer a variety of calcium-fortified soy milks with calcium contents ranging from 80 to 500 mg/serving. Generally, the calcium source is tricalcium phosphate. Because the bioavailability of this salt has not been well characterized, particularly in a soy milk matrix, and because of the importance of ensuring an adequate calcium intake in all individuals irrespective of dietary preferences, we embarked on a study of the bioavailability of calcium-fortified soy milk, using isotopic calcium as a tracer. In the process we uncovered findings important to tracer labeling of such extrinsically enriched calcium sources.
SUBJECTS AND METHODS
Subjects
The subjects were 16 healthy men aged 22–51 y who were taking no regular medications known to affect gastrointestinal function. Each subject gave written, informed consent. The project was approved by the human subjects committee of Creighton University's Institutional Review Board. Subjects were assigned to receive either extrinsically labeled 2%-fat cow milk or extrinsically labeled, calcium-fortified, 2%-fat soy milk (West Soy; Westbrae Natural Foods, Carson, CA) in a random sequence; 11 of the 16 subjects returned for a third test of soy milk containing an intrinsically labeled fortificant. All tests were completed between April 15 and June 30, 1999. Each test was performed as part of a breakfast meal after an overnight fast. The test meal consisted of 3 pieces of low-calcium, Italian-style white bread toasted with butter, a cup of coffee (with artificial sweetener if desired by the subjects), and a quantity of the test calcium source sufficient to provide a calcium load of 300 mg. The calcium source was ingested in its entirety over a 1–2 min period after half the toast was eaten. The milk containers were rinsed 2–3 times with deionized water and the rinsings were consumed. The other half of the buttered toast was then consumed, which served to sweep the mouth and esophagus of residual tracer and to carry it into the stomach.
Source labeling
Cow milk was labeled by adding a microgram-level quantity of high-specific-activity 45CaCl2 (Amersham, Arlington Heights, IL) to each weighed serving of milk, producing a tracer concentration of 185 kBq (5 µCi)/dose; the mixture was gently agitated and the container was then capped and stored at 4°C for 14 h before dosing. Soy milk was labeled extrinsically (precisely as for cow milk) and intrinsically [by incorporating the tracer into the tricalcium phosphate (see below), adding the fortificant to individual weighed servings of unfortified soy milk, and blending by using a tissuemizer (model SDT1810; Tekmar Co, Cincinnati) at high speed for 3 min]. The resulting product was capped and stored at 4°C until dosing.
Preparation of labeled tricalcium phosphate
Labeled tricalcium phosphate (TCP) was synthesized by suspending calcium hydroxide in deionized water, adding a microgram-level quantity of 45CaCl2, and precipitating TCP by slow, drop-wise addition of phosphoric acid, with constant stirring, keeping the pH always >7.0. The precipitate was washed with deionized water and dried at 90°C overnight. The dry powder was ground in a mortar until it would pass through a 200-mesh screen (75 µm). The resulting product was analyzed and found to contain 37.3% calcium by weight (theoretical, 33–39%; typical, 37.0%).
Physical fractionation of sources
All 3 products were centrifuged at 8740 x g for 20 min at room temperature, producing 3 zones in the centrifuge tube: an upper zone consisting of fat, a much larger intermediate fluid zone comprising the bulk of the volume of both sources, and the sediment or pellet (which was substantially larger for the soy milk than for the cow milk). The fat was aspirated and discarded and the other 2 zones were separated and analyzed for 40Ca and 45Ca. Separation was not absolutely quantitative because some of the fluid zone was aspirated and lost with the fat and, to avoid contaminating the fluid phase, some of the fluid was left on the pellet.
Chemical analysis
40Ca was analyzed by atomic absorption spectrophotometry (model AA100; Perkin-Elmer, Norwalk, CT) and 45Ca was analyzed by liquid scintillation counting (model 1900TR; Packard Instrument Corporation, Meriden, CT). The milks and the milk fractions were first ashed at 600°C and the ash was taken up in hydrochloric acid before analysis. Serum samples were analyzed directly without further processing.
Calculation and statistical analysis
Absorption was evaluated by an established method (6, 7) on the basis of the concentration of 45Ca in serum exactly 5 h after oral dosing. The method used has been calibrated for women and the values given here were adjusted upward by 15% to provide for the greater proportion of body water in men. Data were summarized by simple descriptive statistics; within-subject differences between sources were evaluated by the paired t test.
RESULTS
Mean (±SEM) fractional absorption was 0.306 ± 0.015 for the calcium in cow milk, 0.358 ± 0.0167 for the calcium in extrinsically labeled soy milk, and 0.237 ± 0.0153 for the calcium in intrinsically labeled soy milk. By pairwise comparisons, each source showed absorbability significantly different from the others (P < 0.01).
The probable basis for the different values for soy milk under the 2 labeling conditions is shown in Table 1. The sediment in the extrinsically labeled soy milk, which contained 90% of the total calcium of the source, was not well labeled. Approximately three-quarters of the tracer remained in the supernate, which contained <10% of the total calcium of the source.
View this table:
TABLE 1.. Partition of calcium isotopes in labeled cow and soy milks
The specific radioactivities of the 2 phases for all 3 milks are shown in Figure 1, which illustrates how different was the tracer equilibration with the major calcium phases of the 3 preparations. Note that the specific activity of the supernate from the extrinsically labeled soy milk was >25 times higher than that of the sediment calcium. The intrinsically labeled soy milk showed the converse behavior. Nearly 95% of the total calcium and 98% of the tracer was in the pellet. Accordingly, the specific radioactivity of the fluid phase was only about one-third that of the pellet. By contrast, the tracer in the cow milk was proportionately distributed between the calcium of supernate and sediment and both had very nearly the same specific radioactivity. Moreover, 84% of the calcium in the cow milk remained in suspension in the liquid phase.
FIGURE 1. . Specific radioactivity values (units of radioactivity/g Ca) for the fluid and pellet phases of the 2 labeled soy milks and for cow milk, showing very similar values for the cow milk phases but quite different values for each phase with the TCP-fortified soy milk.
DISCUSSION
Nickel et al (8) showed previously that extrinsic and intrinsic labeling of cow milk yielded the same results for calcium absorption. This indicates that the physical and chemical state of the calcium in cow milk are such that a tracer introduced extrinsically exchanges with virtually all of the calcium moieties in the milk, irrespective of their physical state and chemical ligands. Although centrifugation accomplishes only a crude, physical separation, the data in Table 1 are consistent with the absorption results of Nickel et al inasmuch as the tracer in our experiment was proportionately partitioned between the separable liquid and sedimentary phases of the cow milk.
It was also shown previously that extrinsic labeling is effective for certain calcium sources, such as those in solution [eg, calcium citrate malate in orange juice (9)]. Weaver et al (10) also showed that extrinsic labeling is valid for the calcium in wheat products made from a flour dough. On the other hand, extrinsic labeling does not produce uniform tracer distribution in food sources such as calcium-rich, leafy, green vegetables (11). It had been hoped that extrinsic labeling would be effective for an aqueous suspension such as the calcium-fortified soy milk we tested. As the data in Table 1 show, tracer added to the liquid phase of soy milk did not equilibrate with most of the fortificant calcium over several hours of refrigerated storage. This was the case whether the tracer was introduced directly into the liquid phase or via the TCP fortificant. Hence, blood tracer concentrations from an extrinsically labeled source cannot validly be used to quantify total calcium absorption. This nonrepresentativeness is seen clearly in the calculated absorption results. Had we relied on the extrinsically labeled product, we would erroneously have concluded that the absorbability of the TCP-fortified product was 50% higher than it actually was and, moreover, that the product was absorbed significantly better than was the calcium of cow milk. However, the opposite was the case for the intrinsically labeled product: the soy milk we fortified with TCP had only about three-quarters of the absorbability of milk calcium.
It can be argued that the same caution should be applied to conclusions drawn from the intrinsically labeled product, ie, that the absorbability value relates only to the moiety labeled. That is technically correct. However, in this case, the labeled component accounted for 95% of the total calcium of the product. Moreover, the relatively small mass of calcium exchanged between the solid and liquid phases produced a proportionately greater enrichment (ie, labeling) of the liquid phase calcium than could have been produced in the much larger pellet by the same mass of calcium moving from the labeled fluid to the unlabeled pellet. Hence the absorbability of the intrinsically labeled component is effectively the same as that of the composite product.
The very limited exchange between the liquid-phase and the solid-phase calcium in this system should not have come as a surprise. Although the TCP particle size was <75 µm, it is not known how fully crystalline the particles were. If even moderately well crystallized, TCP particles of that size would still have been perhaps 1000–10000 unit cells thick, and exchange with liquid-phase calcium would necessarily have been confined to superficial positions in the TCP crystal.
The lower absorbability of intrinsically labeled, calcium-fortified soy milk could mean that TCP is itself inherently less absorbable than are other calcium salts or it could reflect the presence in the soy milk of antiabsorbers, or both. We are inclined to favor some interference from the soy vehicle because, in unpublished experiments in our laboratory, we evaluated an identical labeled TCP preparation in a candy syrup matrix—also using a crossover, within-subject design—and found that this preparation had an absorbability of 90% that of cow milk calcium. Additionally, we previously tested an earlier batch of labeled TCP, added as a fortificant to yogurt, and found that absorbability was not significantly different from that of calcium from cow milk.
It is hazardous to generalize from these observations to all fortified soy milks because different producers may use different salts or fortificants may have different particle sizes. However, for all of the 5 brands of soy milk we were able to find in the local market area, the labels indicated that TCP was the fortificant, and there are only a few major suppliers of TCP operating in the United States. Moreover, TCP is the fortificant regularly used for soy milk in various infant formulas. Schanler et al (12) and Schanler and Abrams (13) found lower fractional absorption of calcium from such formulas in very-low-birth-weight infants. Our findings are thus consistent with theirs and may be representative of most of the fortified soy milks on the market today.
In conclusion, our findings show that calcium fortification of soy milk, at least by some producers, does not result in a calcium source comparable to cow milk in terms of either physical properties or absorbability. However, it must also be noted that absorption equivalent to that of cow milk can readily be achieved by fortifying soy milk to a higher concentration than the nominal 300 mg/serving. With use of the load-absorbability relation we defined previously (14), it can be calculated that a fortification of 500 mg/serving would result in the same mass of calcium absorbed from a serving of soy milk fortified with TCP as would have been absorbed from a serving of cow milk containing 300 mg Ca. Our findings also show that extrinsic labeling, even of liquid calcium sources, cannot be depended on to produce a uniformly labeled product suitable for bioavailability testing, and that explicit testing of extrinsically labeled against intrinsically labeled sources is necessary before extrinsic labeling can validly be used.
REFERENCES