AJCN EB Program 2010
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Goda, T.
Right arrow Articles by Livesey, G.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Goda, T.
Right arrow Articles by Livesey, G.
Agricola
Right arrow Articles by Goda, T.
Right arrow Articles by Livesey, G.
American Journal of Clinical Nutrition, Vol. 83, No. 6, 1321-1330, June 2006
© 2006 American Society for Nutrition

ORIGINAL RESEARCH COMMUNICATION

Availability, fermentability, and energy value of resistant maltodextrin: modeling of short-term indirect calorimetric measurements in healthy adults 1,2,3

Toshinao Goda, Yuya Kajiya, Kazuhito Suruga, Hiroyuki Tagami and Geoffrey Livesey

1 From the Laboratory of Nutritional Physiology and the COE Program in the 21st Century, University of Shizuoka School of Food and Nutritional Sciences, Shizuoka, Japan (TG, YK and KS); Matsutani Chemical Industry Co, Ltd, Itami, Japan (HT); and Independent Nutrition Logic Ltd, Wymondham, United Kingdom (GL)

2 Supported by the Center of Excellence Program in the 21st Century, the Center of Excellence for Evolutionary Human Health Sciences, and the Ministry of Education, Culture, Sports, Science and Technology of Japan.

3 Address reprint requests and correspondence to T Goda, Laboratory of Nutritional Physiology, School of Food and Nutritional Sciences, The University of Shizuoka, 52-1 Yada Shizuoka-shi, Shizuoka 422–8526, Japan. E-mail: gouda{at}u-shizuoka-ken.ac.jp.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Background: Determination of the metabolizable (ME) and net metabolizable (NME) energy of total carbohydrate requires estimation of its available (AC) and fermentable (FC) carbohydrate content. Modeling of indirect calorimetric observations (respiratory gas exchange) and breath hydrogen would appear to make it possible to estimate noninvasively these nutritional quantities and the approximate time-course of availability.

Objective:We assessed the time-course of metabolism and energy availability from resistant maltodextrin (RMD) by modeling of respiratory gases after a single oral dose.

Design:Seventeen healthy adults (13 M, 4 F; aged 25–46 y) were randomly assigned to treatments (water, maltodextrin, or RMD) in a multiple-crossover, single-blinded trial with ≥7 d washout. We monitored 8-h nitrogen-corrected oxygen and carbon dioxide exchanges and breath hydrogen. All treatment groups took low-carbohydrate meals at 3 and 6 h.

Results:Indirect calorimetry alone provided only qualitative information about the nutritional values of carbohydrate. In contrast, modeling of gaseous exchanges along with the use of central assumptions showed that 17 ± 2% of RMD was AC and 40 ± 4% was FC. As compared with 17 kJ gross energy/g RMD, mean (±SE) energy values were 7.3 ± 0.6 kJ ME/g and 6.3 ± 0.5 kJ NME/g. The fiber fraction of RMD provided 5.2 ± 0.7 kJ ME/g and 4.1 ± 0.6 kJ NME/g.

Conclusions:Modeling with the use of this noninvasive and widely available respiratory gas–monitoring technique yields nutritional values for carbohydrate that are supported by enzymatic, microbial, and animal studies and human fecal collection studies. Improvement in this approach is likely and testable across laboratories.

Key Words: Indirect calorimetry • available carbohydrate • fermentable carbohydrate • energy value • modeling • humans


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Dietary fibers can contribute appreciably to the availability of energy from foods. This capability has nutritional importance and also receives interest from the food industry (1). Variation in the fermentability of different fiber sources has led to the recommendation of specific food energy conversion factors for added, novel, or functional fibers intended for specific dietary needs related to energy requirements, weight control, and regulation via food codes (2). Evaluation of energy availability from fiber in humans has resulted in the successful application of detailed indirect calorimetry (IDC) to quantify heat production (1, 3-6). This quantification shows a predictable rise in heat production that results from fermentation in humans, as occurs in animals. The rise is greater than that for available carbohydrate (AC), and, consequently, fiber has a net metabolizable energy (NME) value that is less than its metabolizable energy (ME) value (1, 2). Whether IDC can also be used to assess carbohydrate availability or fermentability (or both), from which energy values would be calculable by using established general energy conversion factors (2), is a question left open by studies conducted to date.

To gain insight into this possibility and to compare the human response in the short term to resistant maltodextrin (RMD) and maltodextrin (MD), we monitored carbohydrate oxidation by using the well-recognized noninvasive IDC technique (1,7). Stored glycogen is also calculable as AC intake in excess of oxidation. We described the principles of IDC elsewhere along with details of the corrections needed when different types of carbohydrate, fat, protein, or fiber are ingested (1, 7).

The complete combustion of RMD yields carbon dioxide in amounts equal to the amount of oxygen consumed [respiratory quotient (RQ) = 1], which is the same as IDC assumes for carbohydrate. However, the general assumptions do not hold during fermentation because combustion is incomplete with the excretion of the combustible end-products hydrogen, methane, and microbial biomass. In addition, there is unquantified temporal accumulation of short-chain organic acid intermediates in the colon (1). Here, for the first time, a post hoc account is given of carbohydrate oxidation measurements acquired by using IDC under these circumstances, with the use of breath-hydrogen measurement to monitor fermentation. Also for the first time, a separate estimation of carbohydrate utilization by absorption, fermentation, and likely excretion in feces, as distinct from the commonly reported oxidation, is given. Moreover, we make an evaluation of the assumptions used, the possible errors in calculated outcomes, and ways to minimize these errors in the future.

We used commercial RMD that has Generally Recognized as Safe status in the United States (21 CFR 184.1444, maltodextrin) and that was certified as approved for Foods for Specified Health Uses in Japan (8). The RMD is an aggregate of glucose polymers, averaging 12 degrees of polymerization (range: 2–62) and containing currently unspecified 1–2 and 1–3 glucosidic linkages (9). Animals digest and absorb {approx}10% of RMD in the small intestine (10), whereas {approx}40% of RMD appears in feces (11).


   SUBJECTS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Subjects
Seventeen adults (13 M, 4 F) with a mean (±SD) age of 33 ± 6 (range: 25–46) y, body weight of 61.7 (range: 46–74) kg, and body mass index (BMI; kg/m2) of 21.7 ± 2.0 (range: 17.3–24.8) participated in a principal 8-h study monitoring both carbohydrate oxidation and fermentation. Two preparatory studies were also undertaken, one focusing on carbohydrate oxidation in 5 (2 M, 3 F) subjects for 3 h, and another focusing on fermentation marked by breath hydrogen in 10 (8 M, 2 F) subjects for up to 24 h; each study was within the range of subject characteristics for the principal study. All subjects were from the Osaka district, habitually consumed a typical Japanese diet, and were healthy. None had taken medication likely to affect substrate metabolism or antibiotics that may affect fermentation for ≥1 mo.

All volunteers gave written informed consent. The study was approved by the Ethics Committee of the University of Shizuoka.

Study protocols
In the principal study, we investigated both carbohydrate oxidation (kJ/min) and breath hydrogen (ppm). The study was a multiple-crossover single-blinded design, and we familiarized the subjects with the procedures beforehand in a dummy run. Seventeen participants (13 M, 4 F) were randomly assigned by computer program, rested, and fasted overnight (11–12 h), and they took a single oral dose of water (3 mL/kg body wt) containing maltodextrin (0.3 g/kg) or RMD (0.6 g/kg) or water alone as control. Breath oxygen consumption and carbon dioxide production while supine were monitored hourly by IDC for 15-min periods over the next 8 h. During this time, we collected urine for nitrogen measurement, after storage at –20 °C. At 180 and 360 min after oral treatments, we provided low-carbohydrate meals, which were eaten completely. Both meals consisted of eggs, tuna fish canned in oil, and soybean curd; contained 22 g fat, 30 g protein, and 4.3 g AC; and provided 1437 kJ and ≤0.6 g fiber. For each subject, we made observations 9 times at intervals of ≥7 d. These observations comprised 3 repeated measurements (on different days) and 3 different treatments (control, MD, and RMD) for each of the 17 participants, for a total of 153 observations.

We undertook 2 preparatory studies whose general design was similar to that of the principal study. The first assessed only carbohydrate oxidation for 3 h but after different oral doses of maltodextrin providing 0, 0.15 or 0.3 or 0.6 g/kg body wt. Observations were made 8 times for each participant with intervals ≥ 7 d in a rotary multiple crossover design. The 8 observations comprised 2 repeat measurements (on different days) and the 4 doses, on each of the 5 participants, totaling 40 observations.

The second preparatory study assessed the breath-hydrogen response to RMD hourly except for a break from 15 to 21 h after the ingestion. The study had a sequential design in which 0.6 g RMD in 3 mL water/kg body wt was provided on the first occasion, and water alone was provided on the next. Thus, we made observations twice for each participant with an interval of ≥7 d. With 10 participants, this totaled 20 observations.

Materials
The RMD and maltodextrin preparations were from the Matsutani Chemical Industry Co (Itami, Hyogo, Japan). They were defined as Fibersol-2 and TK-16, respectively.

Indirect calorimetry
Respiratory exchange measurements were made hourly in the principal study by using a Vmax29 (SensorMedics, Yorba Linda, CA). After stabilization was confirmed ({approx}7 min), data were averaged for 8 min. In the preparatory study, minute-by-minute measurements were made. Calibration used standard gas of oxygen (26%) and carbon dioxide (4%) accurate to within 0.1%. We measured nitrogen in urine by using the Dumas (combustion) method in an NC-80 nitrogen analyzer (Sumitomo Chemical Industry Co, Tokyo, Japan). Carbohydrate oxidation was calculated by the method of Weir (12), according to the following equation. Adjustments for incomplete oxidation due to fermentation were made post hoc as described in Results.

Formula 1(1)
where C = carbohydrate oxidation, Formula 1CO2 = carbon dioxide uptake, and Formula 1O2 = oxygen uptake.

Breath hydrogen
To mark changes in the rate of fermentation, we measured breath hydrogen. End-expired air (200 mL) was sampled after calorimetric measurements, and the air sample was tested in a TGA-2000HC hydrogen-gas analyzer (Teramecs, Kyoto, Japan).

Statistical analysis
Data are given as means (±SE) or as means (95% CI). Linear (regression), nonlinear (exponential), and dynamic (delay) models were implemented with consideration for measurement error in the determinants (error-in-variables regression). All models used mean values for treatment groups with least-squares estimates weighted for SEs. In addition, estimation with error-in-variables regression used reliability estimates for the measured determinants (13). Carbohydrate oxidation was adjusted for covariation between initial rates at time zero (t0) and the rate at the ith time of measurements (ti). These adjustments were toward the initial rate averaged across all treatment groups. The principal study had a multiple crossover design with 9 periods, 6 sequences, and 3 entries. Each was without significant effect, so these factors were dropped from the analysis to maximize df; the same process was followed in the preliminary studies. Modeling, described as needed in Results, was explored initially in MS EXCEL with subsequent fitting and parameter estimation in STATA software (version SE9; Stata Corp, College Station, TX). Because the data on each occasion derive from a time series, we applied Durbin-Watson tests for positive and negative autocorrelations (14).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Carbohydrate oxidation after maltodextrin ingestion
Carbohydrate oxidation peaked at 60–80 min after maltodextrin (md) ingestion (Figure 1Go). The area under the curve above that for water (w) alone from t0 to the end of measurements (tmax) was summarized as the difference in incremental (i) area under the 2 carbohydrate (C) oxidation curves, denoted by the integral Formula 1. This outcome was related linearly to the amount of maltodextrin ingested (Figure 2Go A) in the range 0 to 0.6 g/kg body wt (Formula 1 range: 0–36 g maltodextrin in our participants). Similar results were obtained by integration over 1-min and 1-h intervals (Figure 2BGo).


Figure 1
View larger version (13K):
[in this window]
[in a new window]
  		FIGURE 1.. Oxidative utilization of maltodextrin in a human volunteer; minute-by-minute measurement was done with indirect calorimetry.

 

Figure 2
View larger version (11K):
[in this window]
[in a new window]
  		FIGURE 2.. Dose-dependent oxidative utilization of maltodextrin by using indirect calorimetry in 5 human subjects up to 3 h. (A) Areas after 1-min integrations, y = 0.26 x x (r2 = 1.000, P < 0.001). (B) Areas after 1-h integrations, y = 0.26 x x (r2 = 0.998, P < 0.001).

 
IDC monitors carbohydrate irretrievably used in oxidation (15). Carbohydrate used in de novo lipogenesis is portrayed in IDC as negative fat oxidation, which equals carbohydrate oxidation in IDC. Its use for glycogen storage or as replacement of endogenous glucose is not detected because no respiratory exchange of oxygen or carbon dioxide is involved. However, the linearity found (Figure 2Go) implies constancy in the ratio of monitored carbohydrate oxidation to unmonitored carbohydrate storage across the dose range examined.

The integrated appearance of oral carbohydrate in blood when assessed by using stable isotopes can follow a lag-rising exponential curve (16, 17) and thus approximate first-order kinetics. Such a curve was apparent for carbohydrate oxidation, assessed here also by using IDC (Figure 3Go), as in the following equation:

Formula 2(2)
where t is time elapsed since ingestion, tlag is a lag period between carbohydrate ingestion and breath carbon dioxide excretion, Cmax is a plateau marking the amount of ingested carbohydrate used in the oxidative process, and {lambda} is a fractional rate constant that is related to the process half-life, as in the following equation:

Formula 3(3)
Normalization of Formula 3by the amount of carbohydrate ingested gives the approach to Cmax as a fraction. No significant difference occurred between the trend for all doses and that for any one dose (Figure 3Go).


Figure 3
View larger version (12K):
[in this window]
[in a new window]
  		FIGURE 3.. Mean (±SE) estimated lag time for oxidative utilization of maltodextrin at 3 doses [0.15 ({diamond}), 0.30 ({triangleup}), and 0.60 ({square}) g/kg] fitted to a common lag-rising exponential curve. The estimated lag time was 0.64 ± 0.06 h; in this study, it reached the estimated plateau (Cmax) at 0.33 ± 0.03 g/g maltodextrin ingested.

 
Cmax occurred at 0.33 ± 0.03 g/g (or kJ/kJ). Thus, on average, approximately one-third of the maltodextrin was "seen" to be used by IDC during this small study. In the more definitive principal study, this value was 0.46 ± 0.05 g/g or {approx}45%; theremainder entered storage, presumably as glycogen or free glucose.

The breath-hydrogen response to RMD
The breath-hydrogen concentration is a well-accepted indicator of the presence of fermentable carbohydrate in the colon. Its appearance after RMD ingestion (0.6 g/kg, averaging 36 g) in the second of the preparatory studies was rapid, after an initial lag, and it persisted up to 24 h (Figure 4Go). Observations not made overnight (15–21 h) were interpolated by fitting a second-order polynomial to each subject's results to enable integration over 24 h. Integration gave the cumulative incremental area under the curve for RMD (rmd), above that for water, toward completion of RMD fermentation (ie, Formula 3). This quantity also fitted a lag-rising exponential curve (Figure 5Go), which was used to summarize the progress and endpoint (Hmax) of RMD fermentation, as shown in the following equation:

Formula 4(4)


Figure 4
View larger version (23K):
[in this window]
[in a new window]
  		FIGURE 4.. Mean (•), SE (—), and interpolated overnight ({circ}) values for the breath-hydrogen response to resistant maltodextrin (0.6 g/kg body wt).

 

Figure 5
View larger version (19K):
[in this window]
[in a new window]
  		FIGURE 5.. Observed means (•), interpolated overnight means ({circ}), and SE (—) for the commulative area under the curve for the breath-hydrogen response to resistant maltodextrin. Values are normalized by the endpoint maximum (Hmax). The central curves show the mean and SE extended toward the normalized mean and SE of Hmax.

 
Carbohydrate utilization after resistant maltodextrin ingestion
Rates of carbohydrate oxidation from RMD or maltodextrin in the principal study are compared in Figure 6Go. These observations followed an analysis of covariance to account for often-substantial correlations with initial rates (Table 1Go). Lower rates were obtained from RMD during the first 3 h, which reflects less AC from RMD than from maltodextrin. Later, oxidation was higher for RMD than formaltodextrin, which is attributable to the sum of oxidation from the anaerobic process in the large bowel and that from aerobic processes acting on absorbed products of carbohydrate fermentation.


Figure 6
View larger version (14K):
[in this window]
[in a new window]
  		FIGURE 6.. Mean (±SE) increments in oxidative utilization of maltodextrin (•, 0.3 g in 3 mL water/kg body wt) and resistant maltodextrin({circ}, 0.6 g in 3 mL water/kg body wt) above a water control (3 mL/kg body wt). Observations for maltodextrin are multiplied by 2 to avoid the illusion of a rate that is too low relative to that for resistant maltodextrin, which was fed at twice the dose.

 

View this table:
[in this window]
[in a new window]
  		TABLE 1. Correlation of the carbohydrate oxidation rate at hourly intervals with the initial rate1

 
Modeling of carbohydrate utilization
Modeling helps to identify various assumptions and limitations and provides additional, quantitative information about the routes of carbohydrate utilization and overall energy value in a form that is conventionally useful. Thus, a model was fitted by using weighted least-squares to assess the probable fractions of RMD used as AC (absorbed via the small intestine), as FC (absorbed as short-chain organic acids via the large intestine), and overall as energy. The last was derived in amounts that can either generate heat (ME) or can replace energy spent on work and maintenance (NME) and so indicates the value in maintenance of energy balance (2).

The simplest model for the present purpose, and one that describes the increase (i) in the rate of carbohydrate oxidation at any time point [iC (t)], is one that includes the sum of the increase from its component parts—available carbohydrate (ac) (iCac) and fermentable carbohydrate (fc) (iCfc)—according to the following equation:

Formula 5(5)
With respect to RMD, both iCac(t) and iCfc(t) are unknown (on the basis of the foregoing data), although they can be estimated when markers of their time-course and extent are available. The time-course of iCac(t) can be approximated initially by that of maltodextrin above that of water Formula 5. A possible delay in the digestion and oxidation of RMD compared with those of maltodextrin is considered below. The time-course of iCfc can be approximated by the breath-hydrogen concentration of RMD above that for water Formula 5; in this case, a possible delay is due to an accumulation of short-chain organic acids in the colon before their absorption and oxidation.

Assuming that there are no delays, then the adoption of the proportionality constants ß1 and ß2 allows rate estimates to be obtained by linear regression, as in the following equation, which assumes the determinants to be unbiased and reliable:

Formula 6(6)
The proportionality constant ß1 has meaning. Given linearity in carbohydrate oxidation via this route with dose (Figure 2Go), then, ß1 indicates the proportion of RMD that is available as carbohydrate, information that is used conventionally in nutrition. In contrast, ß2 simply describes the relation between the apparent rate of carbohydrate oxidation via fermentation and the appearance of hydrogen gas in breath. Equation 6 can be applied directly, although a more stable form for the purpose of regression is its integral, which replaces the rates with cumulative areas, as in the following equation:

Formula 7(7)
Equation 7 simply states that the incremental area under the curve for carbohydrate oxidation between t0 and tmax equals the sum of the separate incremental areas for AC and FC, which, again, assumes unbiased and reliable determinants.

Possible delays due to slow digestion of RMD or accumulation of short-chain organic acids in the colon were evaluated by using the amount entering the delay (Xin) and a fractional rate constant ({lambda}) for an amount leaving the delay (Xout). Thus, the amount leaving the delay was assumed to follow first-order kinetics and therefore to be a fraction of the sum of the amount entering the delay in a given period of time (ti to ti+1[r]) and the amount remaining in the delay (D) from the prior period ending at time ti (with D initially being zero), as shown in the following equation:

Formula 8(8)
This construct in practice was validated by a yield of Xout = Xin when {lambda} = 1 and of Xout = 0 when {lambda} = 0 and by an equality of Xin and the redifferentiated Xout for all other values of {lambda}. Replacement of Xin by iCwtmdgave a delay for AC oxidation, whereas replacement of Xin by iHwtrmdgave a delay for oxidation via fermentation. Values of {lambda} were used that minimized the weighted sum of squares in equation 7, when replacing inputs with delay outputs. This yielded models with 2, 3, or 4 parameters corresponding to those in equation 7 alone and to those in equation 7 with 1 or 2 delays, as shown in equation 8.

Carbohydrate availability from RMD
The estimates for ß1 and ß2, obtained by using the differential regression (equation 6), the integral regression (equation 7), and the latter with and without delays (equations 7 and 8), are shown in Table 2Go. These estimates immediately inform us that a significant fraction of RMD appeared to be available as carbohydrate; thus, ß1 was 0.16-0.17 g/g or kJ/kJ RMD—ie, {approx}17% of the RMD preparation. Implementation of the delays had no significant effect on this particular result. The models so far imply high reliability for the measured determinants. Application of reliability factors (ie, 1 minus the ratio of variance among subjects to total variance in the measurement) also had no effect because the values were close to unity as assumed in ordinary least-squares estimation. Because the data derive from a time series, the residuals were examined for autocorrelation by calculation of the Durbin-Watson statistic. This showed a critical value d = 2.66 > dU ({alpha}<0.01, k = 2)—where "dU" refers to the upper bounds, which indicated that significant positive autocorrelation was unlikely—and a critical value 4 – d = 1.33 > dU ({alpha} <0.01, k = 2)—which indicated that significant negative autocorrelation also was unlikely.


View this table:
[in this window]
[in a new window]
  		TABLE 2. Carbohydrate utilization from resistant maltodextrin (RMD)1

 
The half-life (see equation 3) for the process of maltodextrin oxidation was 0.9 h (Table 2Go). The possibility of oxidation of AC from RMD sooner rather than later than that from maltodextrin was apparent and was quantified as {lambda} >1, which corresponded to a smaller t1/2 for oxidation by 0.2 h. This parameter was neither significant nor practically different from zero and was therefore dropped to preserve df when other model parameters were fitted.

Time-course of carbohydrate availability from maltodextrin and RMD
Some idea of the time-course of carbohydrate availability and the potential errors involved can be obtained from the data. The fraction of maltodextrin used as AC can be assumed to be 1.0 (100% available) at tmax, but the time course is unknown. The fraction [here called phi, for fraction ({Phi})] made available from the maltodextrin (md) and used as carbohydrate (ac) from t0 to any one time (t) can be denoted by [{phi}ac (t)]md. This value is approximately equal to the ratio of the incremental area before time t to the incremental area under the curve before tmax, when all of the AC (ac) has been used from the digestive tract, as shown in the following equation:

Formula 9(9)
The time-course for maltodextrin given by equation 9 is not exact; whereas a constant proportion of dose is used in oxidation (Figure 2Go), uncertainty remains about constancy across time (although oxidation was similar with time across doses; see Figure 3Go). However, this unknown has little implication for the accuracy with which eventual carbohydrate availability and fermentability are quantified, provided the study lasts long enough to permit absorption of the AC to approach completion. By completion, 0.46 ± 0.05 g AC/g maltodextrin had been oxidized and was the denominator in equation 9.

On the basis of equation 9, an approximate time-course for carbohydrate availability is evident for maltodextrin, as shown in Figure 7Go. In this case, carbohydrate from RMD is less than fully available (Figure 7Go); this fraction, [{phi}ac (t)]rmd is given by the product of [{phi}ac (t)]md (from equation 9) and ß1 derived in equation 7, as shown in the following equation:

Formula 10(10)


Figure 7
View larger version (15K):
[in this window]
[in a new window]
  		FIGURE 7.. Mean (95% CI) estimates of the availability of carbohydrate from maltodextrin and resistant maltodextrin.

 
Time-course of carbohydrate fermentability from RMD
It helps to restate in full the expression for the time-course for carbohydrate availability from maltodextrin. Expanding [{phi}ac (t)]md in equation 10 by using the righthand side of equation 9 gives the following equation:

Formula 11(11)
The fraction of RMD used via fermentation (rather than that directly available) at any time, [{phi}fc (t)]rmd, can be derived in a similar way, which yields the following equation:

Formula 12(12)
As it stands, equation 12 is biased—it lacks important features, 2 of which tend to cancel one another and must be remedied simultaneously. When the errors do not cancel, they affect the value of ß2.[r] The first error results in ß2 being too large and arises because IDC fails to represent AC used in glycogen deposition or in the replacement of endogenous glucose. We can remedy this error simply by multiplying the numerator by the fraction of AC (ac) seen by the IDC, {gamma}1. The second error results in ß2 being too small, because IDC fails to represent the extent to which short-chain organic acids from carbohydrate fermentation contribute to glycogen deposition or endogenous glucose during the period of calorimetry. We can remedy this error by multiplying the denominator by a similar fraction, {gamma}2. Both of these remedies are illustrated in the following equation:

Formula 13(13)
These fractions ({gamma}1 and {gamma}2) are unknown at each time point. For simplicity, we can assume that both |{gamma}1| and |{gamma}2| are constants that vary with time between error bounds, ie, |{gamma}1| and |{gamma}2|. As constants, they can be outside the integrals, and thus the equation below replaces equation 13. A third factor, |{gamma}3|, also introduced in the equation below, corresponds to another bias—ie, the overestimation of FC oxidation by IDC because of the influence of accumulating combustible gases (including hydrogen) and biomass on the process RQ (1). The {gamma}2 factor is reasonably well established at 0.90 for the complete process of fermentation, but it may initially be 0.97 should a lag occur in time for the generation of biomass (1), as shown in the following equation:

Formula 14(14)
In the current study, {gamma}1 was derived as the ratio of maltodextrin apparently oxidized by tmax to the amount of maltodextrin ingested, and it was 0.46 ± 0.05 g/g (or kJ/kJ). At any time point, the true value of {gamma}1 would have been more or less than 0.46, and therefore we arbitrarily applied a range ± 20% of 0.46 (ie, 0.37–0.55). The lowest limit for {gamma}2 must be equal to {gamma}1 when short-chain organic acid mixtures are assumed to be as effective as exogenous glucose in displacing endogenous glucose in oxidation or storage. The highest possible value for {gamma}2 is at its limit, 1.0, when carbohydrate used via fermentation has no effect on endogenous glucose metabolism and when all short-chain organic acids are oxidized after absorption, so that the utilization of fermentable carbohydrate is then fully "seen" by IDC. A more representative intermediate value of 0.67 is suggested because it corresponded both to gluconeogenic propionate's contribution to 16% of the short-chain organic acids produced (18) and to the lack of effect of acetate (and presumably butyrate) on respiratory exchange from endogenous glucose (19). Short-chain organic acids potentially could be as effective as dietary fat in saving glucose from oxidation, suppressing oxidation by as much as 20% (20). Thus, the range of values used for |{gamma}2| was 0.37 to 1.0, and a suggested representative value was 0.67. The approximate rise in the amount of carbohydrate fermented with time and theerrors propagated through uncertainties in ß2, {gamma}1, {gamma}2, and {gamma}3 are shown in Figure 8Go.


Figure 8
View larger version (25K):
[in this window]
[in a new window]
  		FIGURE 8.. Mean (±SE; —) and 95% CI (bars) estimates of fermentable carbohydrate from resistant maltodextrin with {gamma}1, {gamma}2, and {gamma}3 at the upper and lower boundaries; SEs are given for assumptions of suggested values ({gamma}1 = 0.46, {gamma}2 = 0.67, and {gamma}3 = 0.9). The period of both indirect calorimetric and breath-hydrogen measurements, •; the period of breath-hydrogen measurement alone, {circ}.

 
In the principal study, observations on breath hydrogen were available up to 8 h only; beyond that time, the progress and endpoint of fermentation were forecast by using parameter estimates from equation 3 (justified in the longer preparatory study; see Figure 5Go). This process allowed the application of equation 14 to quantify carbohydrate fermentation in a useful form, as shown in Figure 8Go. When only the experimental error is considered (ie, given the central values for {gamma}1, {gamma}2, and {gamma}3), the likely mean (±SE) for RMD fermentation would be 40 ± 4 g/100 g RMD. However, the additional error propagated from the bounds for {gamma}1, {gamma}2, and {gamma}3 yields wide CIs (95% CI: 22, 77 g/100 g RMD), so that the true value falls within the vertical bars shown (Figure 8Go).

Fecal carbohydrate
An AC content of 17 g/100g RMD and a fermentable carbohydrate content of 40 g/100 g RMD imply that the amount destined to reach feces would be 43 g/100 g RMD (Table 2Go). The wide 95% CIs for this estimate (ie, 13, 53 g/100 g RMD) indicate that direct measurement of this quantity is warranted. Nevertheless, the model indicates that a considerable proportion of RMD is highly resistant even to the fermentation process, which is difficult to discern from estimates of carbohydrate oxidation alone, as shown in Figure 6Go.

Estimates of energy availability
With respect to the contribution of RMD to energy balance—ie, its contribution to the NME value (2)—it is clear that RMD carbohydrate entering feces contributes no energy. At the other extreme, all of the combustible energy (gross energy) in the AC is available energy at 17 kJ/g. It is now well established in conventions that only half of the combustible or gross energy in FC is useful energy (2), as shown in the following equation:

Formula 15(15)
and ME is obtained as shown in the following equation:

Formula 16(16)
An example is given in the following equation:

Formula 17(17)
Propagation of this estimate of 6.3 kJ NME/g with time is shown in Figure 9Go, along with the experimental SE alone (given unbiased central assumptions for {gamma}1, {gamma}2, and {gamma}3) and with the widest 95% CIs that arise when boundary values for {gamma}1, {gamma}2, and {gamma}3 are used. The corresponding ME value (see equation 16)—ie, the amount of heat that can be generated from RMD—is obtained by replacement of 0.5 in equation 15 with 0.65 in equation 16 (2). These results are summarized in Table 2Go. The ME estimate for the fiber fraction of RMD was 5.2 (2.9–10.7) kJ/g RMD, whereas the NME estimate was 4.1 (2.2–8.2) kJ/g.


Figure 9
View larger version (26K):
[in this window]
[in a new window]
  		FIGURE 9.. Mean (±SE; —) and 95% CI (bars) estimates of the availability of net metabolizable energy from resistant maltodextrin. {gamma}1, {gamma}2, and {gamma}3 are at the upper and lower boundaries, and SEs assume suggested values ({gamma}1 = 0.46, {gamma}2 = 0.67, and {gamma}3 = 0.9). The period of both indirect calorimetric and breath-hydrogen measurements, •; and the period of breath-hydrogen measurement alone, {circ}.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
We report on the metabolism of a carbohydrate that is mostly resistant to digestion and partially resistant to fermentation, as indicated by a short-term IDC technique and breath-hydrogen measurements. Modeling is used to estimate carbohydrate availability, fermentability, energy values, and the precision of the approach. A summary of our observations is given schematically in Figure 10Go.


Figure 10
View larger version (35K):
[in this window]
[in a new window]
  		FIGURE 10.. Summary of carbohydrate utilization from maltodextrin and resistant maltodextrin derived from the time course of gaseous exchanges. CHO, carbohydrate; Av, available; Unav, unavailable. Single quotes indicate values derived by calculation involving both parameter estimates and assumed values; double quotes indicate assumed values. Parameter estimates are from Table 2Go.

 
We confirm that the response of carbohydrate oxidation to AC intake is linear for the quantities consumed: 0–0.6 g/kg body wt. Hence, carbohydrate availability appears predictable from gaseous exchange measurements. Whereas hourly integration is sufficient for the present, we believe that more frequent sampling during the first 4 h would improve accuracy (Table 3Go), particularly for the parameters of time.


View this table:
[in this window]
[in a new window]
  		TABLE 3. Recommendations for standardization1

 
The linear dose response also implies a dose-independent distribution of AC between irretrievable oxidation (including, to a minor extent, de novo lipogenesis) and glycogen storage. Linearity also implies that the doses fall within the capacity of the homeostatic mechanisms for our participants. However, that may not be so at higher doses or in diabetic patients, and so we recommend standardization (Table 3Go), which excludes participants with abnormal carbohydrate metabolism and, possibly, aims for individuals of similar glucose tolerance.

Between studies (preparatory and principal) and in subjects (principal study), we found a trend for persons with a low fasting RQ, who tend to oxidize more lipid than carbohydrate, to oxidize less of the maltodextrin (ie, to store more maltodextrin as glycogen). The correlation of individual fasting RQs measured on different occasions is significant (P for trend = 0.013) although imperfect (r2 = 0.34) and thus is a source of experimental error. Such error may be reduced by consumption of a standard high-carbohydrate meal the evening before IDC, as is often practiced. Instead, the current study adjusts carbohydrate oxidation rates for covariance with the initial (fasting) rates. This is more representative of the free-living state, but penalties may remain for intergroup comparisons. In the future, both approaches may be used simultaneously to acquire greater control (Table 3Go).

We found that RMD consumption produced 2 peaks of carbohydrate oxidation, one within 3 h and another {approx}6–8 h after RMD ingestion. The latter is consistent with breath 14CO2 data from nondigestible but readily fermentable [14C]fructooligosaccharides (21), which were first detectable within 2 h and which reached a maximum at 6–7 h after ingestion. We consider it likely that our first peak with RMD is due in large part to the presence of AC, whereas the latter peak involves fermentation.

Breath-hydrogen appearance from RMD is similar in time-course to that of the fermentable tagatose (22). However, it occurs somewhat earlier than expected from the appearance in breath of 14CO2 after the ingestion of fructooligosaccharides (21). This earlier appearance likely is due to the so-called "bicarbonate delay" (23), in which the 14CO2 has to equilibrate with endogenous unlabeled bicarbonate before peaking in breath. Because of the temporal accumulation of short-chain organic acids in the colon, we expect some delay in the appearance of carbon dioxide in breath after the ingestion of FC. This accumulation likely explains the delay between breath-hydrogen appearance and the later peak of carbohydrate oxidation (t1/2: 1.3 h; Table 2Go).

Our model estimate for the AC content of RMD is {approx}17 g/100 g. This value accords with a glycemic (and insulinemic) response to RMD, which is {approx}10% of that for maltodextrin in humans (9). Furthermore, our observation of little difference in the time-course for oxidation of maltodextrin and AC from RMD (delay t1/2 does not differ significantly from zero) is consistent with little difference in the time-course for the glycemic (and isulinemic) response (9). Moreover, our in vivo data agree with in vitro digestion studies using a gastric juice, salivary {alpha}-amylase, pancreatic {alpha}-amylase, and intestinal mucosal enzymes, which indicate that 10.2 g/100 g RMD is available as carbohydrate (11). We cannot exclude method bias as a cause of moderately higher availabilities in vivo than in vitro. The value in vivo is sensitive to baseline information, and, thus, more time points at the start of IDC would be helpful (Table 3Go).

The model indicates that fermentable carbohydrate in RMD is {approx}40 g/100 g. This finding is supported by a recent study of similar design in rats, which gave an identical result (T Goda et al, manuscript in preparation). RMD not accounted for as AC or FC is 43 g/100 g, and we assume that this quantity enters the feces. It is interesting that Satouchi et al (24) reported that 100 g RMD consumed over 5 d causes fecal dry weight to increase by 43 g. This increase corresponds to 40 g more fecal carbohydrate and 7 g more biomass per 100 g RMD consumed [given 90% recovery of nondegradable matter in 5 d and appearance of 20 g biomass/100 g fermented carbohydrate ({approx} 30 kJ bioenergy/100 kJ RMD)]. Further support comes from animal studies in which 38 g/100 g RMD consumed is recovered in feces (11). The current model estimates are therefore consistent with prior studies in both humans and animals.

Our observations are also consistent with those of in vitro studies. With human fecal inocula, fermentability of RMD is 50–66 g/100 g that of 97.4% fermentable pectin (18, 25). This amount in vitro corresponds to the sum of AC and FC in RMD and is in good agreement with the current sum for AC (17 g/100 g) and FC (40 g/100 g), which is 57 g/100 g. Similarly, the sum of the nondigestible but fermentable and nonfermentable carbohydrate in RMD in the current in vivo study represents the total fiber content and is 83 g/100 g RMD. This value is comparable to, although somewhat less than the 95 g/100 g RMD assessed by the total dietary fiber method in collaborative studies of the Association of Official Analytical Chemists (10); it is more consistent with 89.8 g/100 g after the deduction of enzymatically digestible carbohydrate (11). Although such agreements support the knowledge gained from this human study, we expect to obtain stronger evidence with the use of the current method by following the general recommendations given in Table 3Go.

In conclusion, our observations support an earlier view (1) of the need for detail to obtain adequate interpretation of often-used IDC data when FC is present. Under such a circumstance, IDC together with breath hydrogen can give reasonably precise information about the AC content of the total carbohydrate consumed. Information on FC content, in contrast, is less precise, largely because of uncertainty in the {gamma} factors in equation 14. Here we set particularly wide error bounds for the effect of short-chain organic acids from the colon on glucose metabolism ({gamma}2)—from 0% to 100% as effective as glucose in modifying glycogen storage. Otherwise, errors are quite small when {gamma} factors that fit our current knowledge are used, and results are in good agreement with observations from a variety of in vitro, animal, and other human studies of RMD. For the present, therefore, such comparative support data remain helpful in the assessment of the nutritional value of carbohydrate, such data are essential when providing the totality of evidence, and the need still exists to assess resistance to overall digestion and fermentation directly by using fecal collection and analysis. However, fecal data alone would not provide information about the partition between AC and FC or about the time-course of in vivo availability.


   ACKNOWLEDGMENTS
 
We thank Norimasa Hosoya and Hidemasa Hidaka for valuable comments on an innovative procedure for estimating the availability of energy from carbohydrates.

TG was responsible for the experimental design, supervision of collections and analyses, data analysis, and writing of the manuscript. YK was responsible for specimen collection, preparation, and indirect calorimetry operation. KS was manager of laboratory operations and was responsible for specimen collection and preparation. HT supervised specimen collection and gas analysis and edited the manuscript. GL was responsible for modeling and for editing of the manuscript. None of the authors had any personal or financial conflict of interest.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 SUBJECTS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Livesey G. Thermogenesis associated with fermentable carbohydrate in humans, validity of idc, and implications of dietary thermogenesis for energy requirements, food energy and body weight. Int J Obes Relat Metab Disord 2002;26:1553–69.[Medline]
  2. Food and Agriculture Organization. Food energy—methods of analysis and conversion factors. FAO Food and Nutrition Paper no. 77. Rome, Italy: Food and Agriculture Organization, 2003.
  3. Castiglia-Delavaud C, Verdier E, Besle JM, et al. Net energy value of non-starch polysaccharide isolates (sugarbeet fiber and commercial inulin) and their impact on nutrient digestive utilization in healthy human subjects. Br J Nutr 1998;80:343–52.[Medline]
  4. Poppitt SD, Livesey G, Elia M. Energy expenditure and net substrate utilization in men ingesting usual and high amounts of nonstarch polysaccharide. Am J Clin Nutr 1998;68:820–6.[Abstract]
  5. Sinaud S, Montaurier C, Wils D, et al. Net energy value of two low-digestible carbohydrates, Lycasin HBC and hydrogenated polysaccharide fraction of Lycasin HBC in healthy human subjects and their impact on nutrient digestive utilization. Br J Nutr 2002;87:131–9.[Medline]
  6. Vermorel M, Coudray C, Wils D, et al. Energy value of a low-digestible carbohydrate, NUTRIOSE FB, and its impact on magnesium, calcium and zinc apparent absorption and retention in healthy young men. Eur J Nutr 2004;43:344–52.[Medline]
  7. Livesey G, Elia M. Estimation of energy expenditure, net carbohydrate utilization, and net fat oxidation and synthesis by idc: evaluation of errors with special reference to the detailed composition of fuels. Am J Clin Nutr 1988;47:608–28.[Abstract/Free Full Text]
  8. Shimizu T. Health claims and substantiation of functional foods—Japanese system aiming the global standard. Curr Topics Nutraceut Res 2003;1:213–24.
  9. Ohkuma K, Matsuda I, Katta Y, Hanno Y. Pyrolysis of starch and its digestibility by enzymes. Denpun Kagaku 1990;37:107–14.
  10. Gordon DT, Okuma K. Determination of total dietary fiber in selected foods containing resistant maltodextrin by enzymatic-gravimetric method and liquid chromatography: collaborative study. J AOAC Int 2002;85:435–44.[Medline]
  11. Tsuji K, Gordon DT. Energy value of a mixed glycosidic linked dextrin determined in rats. J Agric Food Chem 1998;46:2253–9.
  12. Weir JBV. New methods for calculating metabolic rate with special reference to protein metabolism. J Physiol 1949;109:1–9.[Free Full Text]
  13. Stata base reference manual. College Station, Texas: Stata Corp, 2005;1:321–6.
  14. Montgomery D, Peck EA, Vining GG. Introduction to linear regression analysis. New York: Wiley & Sons, Inc, 2001.
  15. Elia M, Livesey G. Theory and validity of idc during net lipid synthesis. Am J Clin Nutr 1988;47:591–607.[Abstract/Free Full Text]
  16. Livesey G, Wilson PDG, Roe MA, et al. Splanchnic retention of intraduodenal and intrajejunal glucose in healthy adults. Am J Physiol 1998;275(4 Pt 1):E709–16.
  17. Livesey G, Wilson PDG, Dainty JR, et al. Simultaneous time- varying systemic appearance of oral and hepatic glucose in adults monitored with stable isotopes. Am J Physiol 1998;275(4 Pt 1):E717–28.
  18. Flickinger EA, Wolf BW, Garieb KA, et al. Glucose-based oligosaccharides exhibit different in vitro fermentation patterns and affect in vivo apparent nutrient digestibility and microbial populations in dogs. J Nutr 2000;130:1267–73.[Abstract/Free Full Text]
  19. Scheppach W, Wiggins HS, Halliday D, et al. Effect of gut-derived acetate on glucose turnover in man. Clin Sci (Lond) 1988;75:363–70.[Medline]
  20. McDevitt RM, Poppitt SD, Murgatroyd PR, Prentice AM. Macronutrient disposal during controlled overfeeding with glucose, fructose, sucrose, or fat in lean and obese women. Am J Clin Nutr 2000;72:369–77.[Abstract/Free Full Text]
  21. Hosoya N, Dhorranintra B, Hidaka H. Utilization of [U-14C]fructooligosaccharides in man as energy resources. J Clin Biochem Nutr 1998;5:67–74.
  22. Buemann B, Toubro S, Astrup A. D-Tagatose, a stereoisomer of D-fructose, increases hydrogen production in humans without affecting 24-hour energy expenditure or respiratory exchange ratio. J Nutr 1988;128:1481–6.
  23. Elia M, Fuller NJ, Murgatroyd PR. Measurement of bicarbonate turnover in humans: applicability to estimation of energy expenditure. Am J Physiol 1992;263:E676–87.
  24. Satouchi M, Wakabayashi S, Ohkuma K, Fujiwara K, Matsuoka A. Effects of indigestible dextrin on bowel movements. Jpn J Nutr 1993;51:31–7.
  25. Barry JL, Hoebler C, Macfarlane GT, et al. Estimation of the fermentability of dietary fibre in vitro: a European interlaboratory study. Br J Nutr 1995;74:303–22.[Medline]
Received for publication December 19, 2005. Accepted for publication February 28, 2006.





This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Goda, T.
Right arrow Articles by Livesey, G.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Goda, T.
Right arrow Articles by Livesey, G.
Agricola
Right arrow Articles by Goda, T.
Right arrow Articles by Livesey, G.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS