Acute human studies describe variable postprandial glycemic and/or insulinemic responses to RS ingestion. In general, it is accepted that RS consumption lowers postprandial glucose concentrations marginally and postprandial insulin concentrations markedly. Many groups report a decrease in postprandial glycemic or insulinemic responses to RS ingestion relative to digestible starch (DS) consumption [2-7], whereas some report no change [8-11]. It is important to note that the fat content of the diet has a significant impact on the glycemic response to a meal and some meal tests contained no fat or the fat content of the meal varied among the different RS diets making results from these studies difficult to interpret [2-4]. Also, there are many sources of RS, such as beans, high amylose corn starch, and potatoes, which possess different physicochemical properties. So, the source of RS can influence the glycemic/insulinemic response to RS ingestion.

Many studies have examined the relationship between RS ingestion and postprandial metabolite and hormone concentrations. Fewer studies have documented the effect of RS on lipid metabolism. In humans, five weeks of RS feeding lowered fasting cholesterol and triglyceride concentrations and postprandial plasma insulin concentrations relative to digestible starch (DS) feeding [12,13]. It has also been reported that chronic RS feeding in rats causes a decrease in adipocyte cell size relative to DS feeding [14,15]. In addition, expression of fatty acid synthase was lower in rats fed a RS-based diet than in those fed a DS-based diet [16]. Taken together, these studies provide evidence that RS intake has an effect upon the activity of key lipogenic enzymes and adipocyte morphology. Thus, it seems that the effects of this carbohydrate subtype on lipid metabolism should be carefully examined in human studies.

Overall, the dose of RS in the meal had a significant influence on ΔRQ (respiratory quotient) values (F-test, 0.04; Figure ​Figure2).2). This overall effect was due to a significantly lower ΔRQ at the 5.4% RS dose than the 0% (p = 0.02) or 10.7% (p = 0.009) RS doses, indicating an increase in fat oxidation in response to the 5.4% RS meal relative to the 0% and 10.7% RS doses (Figure ​(Figure2).2). ΔRQ was significantly lower for the 5.4% RS meal than 0% RS meal at 120, 240, 300 and 360 minutes (p = 0.05, 0.03, 0.02 and 0.04, respectively) whereas significant differences occurred at 120, 180, 240, 300 and 360 minutes (p = 0.01, 0.01, 0.005, 0.02, and 0.03, respectively) for the 5.4% RS versus 10.7% RS meals. These data are reflected in total macronutrient oxidation rates (Figure ​(Figure3),3), which show a significant increase in the amount of fat oxidized at the 5.4% RS dose relative to the 0% RS meal, with a concomitant decrease in total carbohydrate oxidation.

Similarly, the oxidation of [¹⁴C]-triolein to ¹⁴CO₂ was different between RS doses (F-test, 0.0005). Meal fat oxidation at the 5.4% RS dose was significantly higher than both the 0% (p = 0.0062) and 10.7% doses (p < 0.0001). Separate tests at 6 h or 24 h following the test meal gave comparable results (Figure ​(Figure4a).4a). Taken together, these independent measurements of fat oxidation (indirect calorimetry, oxidation of [¹⁴C]-triolein to ¹⁴CO₂) suggest that the inclusion of 5.4% RS in the meal elevated postprandial fat oxidation. Unexpectedly, this effect was lost if the dose was increased to 10.7% RS.

There was no difference in postprandial glucose (Figure ​(Figure1a),1a), FFA (Figure ​(Figure1e),1e), triglyceride (Figure ​(Figure1f),1f), or insulin (Figure ​(Figure1c)1c) concentrations at any RS dose examined. This concurs with data from other acute human studies using complete, mixed meals which showed no difference in postprandial glycemia/insulinemia in response to RS content of the diet [8-11]. Although this seems contrary to the general perception that RS ingestion reduces postprandial insulinemia and glycemia, many of the studies indicating this did not balance test diets for total fat and/or fiber content [17]. However, in the current study all diets were carefully matched for total fat and fiber content. This an important distinction between this and other studies as fiber has extensively been shown to reduce postprandial glycemia/insulinemia and increasing the RS content of the diet intrinsically increases the total fiber content. Also, dietary fat can have potent effects on the accessibility of dietary carbohydrate to digestive enzymes and on the rate of gastric emptying/gut motility. Thus, the glucose- and insulin-lowering effects of RS that have been observed in other studies may be due to changes in fiber and/or fat between test meals which have been extensively shown to lower postprandial glycemic and insulinemic responses. So, the balanced conditions used in the meal tests for the study described herein, which included baked products and processed foods as part of a complete, mixed meal, balanced for total fat and fiber content, could account for the lack of difference in insulinemia and glycemia in response to increased RS content in the diet.

Both indirect calorimetry and ¹⁴C-tracer data indicate that there was an increase in fat oxidation between the 0% and 5.4% RS doses (Figures ​(Figures2,2, ​,3,3, and ​and4a).4a). This increase in total and meal fat oxidation in response to the 5.4% RS meal is not driven by disparate responses amongst subjects as 11 of the 12 subjects studied showed the greatest fat oxidation in response to the 5.4% RS meal, relative to the 0% and 10.7% RS meals (see Figure S1, Additional File 1, for individual responses). Tracer data showed that the addition of 5.4% of RS to a meal increased meal fat oxidation by more than 20% over the 6 h and 24 h post-meal ingestion period (Figure ​(Figure4a).4a). The increase in fat oxidation at 6 h accounted for approximately one-half of the total increase over 24 h, indicating that the increase in meal fat oxidation in response to a single meal containing 5.4% RS is a prolonged, sustained effect. In addition, comparison of total and meal fat oxidation (Figures ​(Figures3a3a and ​and4a)4a) indicates that endogenous fat stores were the predominant source of fat utilized for energy, contributing approximately 80% of the total fat oxidized, with a much lower contribution from ingested meal fat. Figure ​Figure33 shows that this increase in fat oxidation at the 5.4% RS dose is accompanied by a relative reduction in carbohydrate oxidation (does not reach statistical significance).

The increase in fat oxidation at the 5.4% RS dose relative to the 0% dose was not driven by any disparity in circulating glucose, insulin or FFA concentration (Figure ​(Figure1;1; see Figures S2, S3, S4, Additional Files 2, 3, 4, respectively, for individual subject responses) nor by a difference in available carbohydrate between the 0% and 5.4% RS meals. If decreased carbohydrate availability was responsible for the observed increase in fat oxidation, the 10.7% RS meal, which has the least available carbohydrate, would show the greatest increase in fat oxidation. However, there was no difference in fat oxidation between the 0% and 10.7% RS meals. Thus, carbohydrate availability cannot be a contributing factor to the increase in fat oxidation observed at the 5.4% dose of RS. It is possible that this increase may be due to an increase in circulating SCFAs from the fermentation of RS reaching the large bowel. The observed increase in fat oxidation is not due to oxidation of these SCFAs per se as it was measured directly from conversion of ¹⁴C-labeled meal fat to ¹⁴CO₂ (Figure ​(Figure3a).3a). Such a measurement would not detect any increase in SCFA oxidation. Rather, it may be that the metabolic effects of increased SCFA production cause an increase in fat oxidation.

RS consumption has been shown to alter the acetate:butyrate:propionate ratio compared to fermentation of non-starch polysaccharides [29]. In particular, the amount of butyrate is substantially elevated in response to RS fermentation [30,31]. In humans fed a low or high RS diet for three days, the concentration of excreted SCFA rose from 20 mmol/d to 33 mmol/d, respectively [19]. This increase in total SCFA concentration was caused by a doubling of the acetate and butyrate content changing the acetate:butyrate:propionate ratio from 12:3:3 to 21:6:4 in response to the low and high RS diets, respectively.

In vitro data from isolated animal tissues provide convincing evidence for the role of SCFAs in carbohydrate and lipid metabolism [26,32-34]. Acetate and/or butyrate have been shown to decrease glycogenolysis and glycolysis in isolated rat and sheep hepatocytes [35-37]. So, it is plausible that the fermentation of RS from the 5.4% RS diet increases the net production of SCFAs which inhibit glycolysis in the liver. In this scenario, the liver, deprived of carbohydrate-derived acetyl CoA would be more reliant on fat-derived acetyl CoA as a fuel source, thereby contributing to an overall increase in fat oxidation [17]. This possibility needs to be investigated in future studies.

No difference in fat oxidation was evident between the maximal 10.7% dose of RS and the 0% dose. This is an unexpected result that is difficult to explain. The loss of any effect on fat oxidation when the RS dose in the meal was increased to 10.7% may occur because this dose is at the threshold of the starch's properties as RS. That is, at the 10.7% dose of RS, the starch may not be completely fermented in the large bowel thereby causing a loss of energy from the diet via the feces. If this is the case, the strong physical association between RS and dietary lipid may cause excretion of lipid and therefore, less dietary fat to be available for oxidation at the 10.7% dose. Indeed, it has previously been shown that intake of high-amylose maize starch, such as that used in this study, caused an increased number of bowel actions per day [18]. RS has also been shown to decrease colonic transit time and, as more RS enters the large bowel, more starch is also excreted [19,20]. This indicates that, at higher levels of RS consumption, only a portion of the RS can be fermented and the remainder passes through the colon as an insoluble fiber. Furthermore, if indeed RS at the 10.7% dose is being excreted as insoluble fiber, less fermentation and SCFA production would be occurring. As SCFA are hypothesized to be the cause of the observed increase in fat oxidation in response to the 5.4% RS meal, this would have a large impact on the fat oxidation potential of the 10.7% RS diet.

The hypothesis that RS is acting like dietary fiber and being excreted can be tested by measuring the amount of fat excreted in the feces. As this outcome was not predicted, fecal samples were not collected from subjects during this study. It is important to consider that it is difficult to add 10.7% RS to a standard diet without the use of specially designed foods and/or without significantly increasing caloric intake. Therefore, this level would be difficult to attain in a free-living situation and the lower doses used in this study are more reflective of predicted levels if normal, starchy foods in the diet were to be replaced with commercially available RS products.

In addition, not all biological processes display linear dose-response curves. Dose-response curves can vary from sigmoidal to 'U'-shaped curves for processes as diverse as drug absorption/clearance [21], low dose radiation effects on cells [22], DNA repair following double-strand breaks [23], and metabolic parameters. Metabolic processes that are non-linear functions include the level of illuminance and plasma melatonin levels [24], caffeine intake versus plasma caffeine metabolite concentrations [25], allergen exposure (concentration) and histamine response [26], zinc-stimulated histamine release from mast cells [27], and fructose-1,6-diphosphate metabolism in cardiomyocytes [28]. Thus, it is possible that the lack of any effect on fat oxidation at the 10.7% RS dose may indicate that the relationship between RS intake and fat oxidation is indeed a 'U'-shaped curve. However, more RS doses between 5.4% and 12% must be tested to accurately define the shape of this dose response curve.

It must be noted that the calculation of oxidation of [¹⁴C]-triolein via measurement of ¹⁴CO₂ did not take into account the dilution of tracer in vivo due to the incorporation of labeled carbons into intermediates of the TCA cycle and endogenous bicarbonate pools. Generally, an acetate correction factor is used to account for this effect. In this study, subjects consumed all four test meals under the same conditions and it was assumed that there was no difference in tracer recovery between tests. Also, these TCA intermediate and bicarbonate pools were not pre-labeled prior to the ingestion of the label in the meal which would cause a total underestimation of total fat oxidation. Therefore, the rate of fat oxidation calculated from ¹⁴CO₂ recovery in the breath was probably underestimated in all subjects but remains valid to compare differences between test meals.