Postprandial hypotension (PPH), defined as a fall in systolic blood pressure (BP) ≥ 20 mmHg within 2 h of a meal(Reference Jansen and Lipsitz1), leads to syncope and falls, and is recognised as a frequent and clinically important problem, particularly in the elderly and patients with autonomic dysfunction, the latter often secondary to diabetes mellitus(Reference Jansen and Lipsitz1, Reference Mathias2). PPH is distinct from, and occurs more frequently than, orthostatic hypotension(Reference Jansen and Lipsitz1). The mechanisms responsible for PPH are poorly defined; however, several factors including meal composition, gastric distension, small-intestinal nutrient delivery, splanchnic blood flow and neural and hormonal mechanisms appear important(Reference Jansen and Lipsitz1, Reference Sidery, Cowley and MacDonald3–Reference O'Donovan, Feinle and Tonkin7). An understanding of these mechanisms is pivotal for the effective management of PPH, which is currently suboptimal.
The onset of the fall in BP is usually evident soon after a meal, with a maximum response at 30–60 min(Reference Jansen and Lipsitz1), suggesting a relationship to the delivery of nutrients to the small intestine, which has proven to be the case. When glucose is administered intraduodenally to healthy older subjects at rates of 4·2 kJ/min (1 kcal/min) or 12·6 kJ/min (3 kcal/min)(Reference O'Donovan, Feinle and Tonkin7, Reference Vanis, Gentilcore and Rayner8), i.e. within the normal physiological range of gastric emptying (GE)(Reference Brener, Hendrix and McHugh9), the fall in BP is much greater in response to 12·6 kJ/min (3 kcal/min) when compared with 4·2 kJ/min (1 kcal/min). In contrast, gastric distension, probably even at low volumes, attenuates the fall in BP(Reference Rossi, Andriesse and Oey5, Reference Shannon, Diedrich and Biaggioni6, Reference Gentilcore, Meyer and Rayner10). Ingestion of carbohydrate, particularly glucose, was believed to have the greatest suppressive effect on BP(Reference Jansen, Peeters and Van Lier11) when compared with fat and protein(Reference Jansen, Peeters and Van Lier11), but recent studies by our group have shown that oral(Reference Visvanathan, Horowitz and Chapman12) and intraduodenal(Reference Visvanathan, Horowitz and Chapman12, Reference Gentilcore, Hausken and Meyer13) infusion of fat, protein and glucose(Reference Gentilcore, Hausken and Meyer13) induces comparable falls in BP in healthy older subjects, although the hypotensive response to glucose occurs earlier than with fat or protein(Reference Visvanathan, Horowitz and Chapman12, Reference Gentilcore, Hausken and Meyer13). There is little information about the effect of different carbohydrates on postprandial BP, particularly those that are absorbed more slowly than glucose. Xylose is a poorly absorbed pentose, commonly found in plant cell walls, which is used as a food additive to produce a ‘savoury’ flavour(Reference Arnoldi, Corain and Scaglioni14). Information relating to the effects of xylose on BP is inconsistent. It has been reported that there is no fall in BP after oral xylose in amounts of 42(Reference Robinson, Stowell and Purdie15) and 0·83 g/kg body weight(Reference Robinson and Potter16) in healthy older subjects who exhibited a fall in BP following oral glucose, whereas Mathias et al. (Reference Mathias, da Costa and McIntosh17, Reference Mathias18) suggested that there is a small fall in BP following oral xylose. A limitation of these studies(Reference Robinson, Stowell and Purdie15–Reference Mathias18) was that GE of glucose and xylose was not measured, and differences in the rate of carbohydrate delivery into the small intestine may have, accordingly, influenced the observations. In monkeys, the GE of xylose apparently occurs in a similar fashion to that of glucose; i.e. in an overall linear pattern and more slowly with increasing concentration, presumably as a result of inhibitory feedback arising from the small intestine(Reference Moran and McHugh19). In contrast, in humans, xylose (25 g) has been reported to markedly prolong GE when compared with the same amount of glucose(Reference Shafer, Levine and Marlette20).
The ‘incretin hormones’, glucose-dependent insulinotropic polypeptide (GIP) and glucagon-like peptide-1 (GLP-1), are responsible for the substantially greater insulin response to oral glucose compared with isoglycaemic intravenous glucose loads(Reference Holst and Gromada21). GLP-1 is secreted by L-cells located predominately in the distal small intestine and colon, and suppresses glucagon secretion, as well as stimulating glucose-dependent insulin secretion, while GIP is released from the K-cells, which are located predominantly in the proximal small intestine(Reference Holst and Gromada21, Reference Schirra, Nicolaus and Roggel22). Recent observations have suggested that GLP-1 may have a protective role in PPH. In humans(Reference Edwards, Todd and Ghatei23) and animals(Reference Barragan, Rodriguez and Eng24), exogenous administration of GLP-1 may increase BP. We have reported that the α-glucosidase inhibitor, acarbose, which is used frequently in the management of type 2 diabetes, attenuates the fall in BP induced by oral sucrose in healthy older subjects, slows GE and markedly stimulates the secretion of GLP-1(Reference Gentilcore, Bryant and Wishart25). The latter effect presumably reflects the presence of carbohydrate in the small intestine. In dogs, there was no increase in the release of GLP-1 following an infusion of xylose into an ileal loop(Reference Shima, Suda and Nishimoto26). The effects of carbohydrate on GLP-1 secretion may, however, be species-dependent(Reference Baggio and Drucker27), and there is no information about the effects of xylose on GIP and GLP-1 in humans.
The aims of the present study were to determine the effects of oral xylose on BP, GE and incretin hormone secretion, when compared with oral glucose and water, in healthy older subjects.
Materials and methods
Subjects
A total of eight healthy older subjects (six males and two females), with a median age of 70·5 (range 65–75) years and BMI of 23·5 (range 20·4–27·1) kg/m2, were recruited by advertisement. All were non-smokers, and none had a history of gastrointestinal disease or surgery, diabetes, significant respiratory, renal, hepatic or cardiac disease, intake of >20 g alcohol/d or was taking medication known to influence BP or gastrointestinal function.
Protocol
The protocol was approved by the Human Research Ethics Committee of the Royal Adelaide Hospital, and each subject provided written informed consent. All experiments were carried out in accordance with the Declaration of Helsinki.
Each subject was studied on three occasions in a randomised, double-blind order; each study day was separated by a minimum of 3 d. On each day, the subject attended the laboratory at 08.00 hours following an overnight fast (10 h for solids and 8 h for liquids). An intravenous cannula was placed in a left antecubital vein for blood sampling, and an automated BP cuff positioned around the right arm for the measurement of BP and heart rate. Each subject was then allowed to rest, seated in a chair, for about 30 min. At t = − 2 min, the subject consumed a 300 ml drink comprising either (1) water (50 ml low-energy lemon cordial (Bickford's, Adelaide, SA, Australia)+250 ml water) – ‘W’, (2) 50 g glucose monohydrate (dissolved in 50 ml low-energy lemon cordial+155 ml water+80 ml hypertonic saline (3 %)) – ‘G’ or (3) 50 g d-xylose (dissolved in 50 ml low-energy lemon cordial+235 ml water) – ‘X’, within 2 min. Both carbohydrate drinks were isoenergetic (approximately 782·4 kJ (187 kcal)) and iso-osmolar (approximately 1350 mOsmol). GE, BP (systolic and diastolic) and heart rate were then measured for 120 min. On one day, cardiovascular autonomic nerve function was evaluated immediately after the completion of the study(Reference Ewing and Clarke28, Reference Piha29).
Measurements
Blood pressure and heart rate
BP (systolic and diastolic) and heart rate were measured using an automated oscillometric BP monitor (DINAMAP ProCare 100; GE Medical Systems, Milwaukee, WI, USA) before the consumption of the drink and then every 3 min between t = 0 and 120 min(Reference O'Donovan, Feinle and Tonkin7). ‘Baseline’ BP and heart rate, i.e. ‘t = 0 min’, were calculated as the mean of measurements taken at t = − 9, − 6 and − 3 min. PPH was defined as a fall in systolic BP of ≥ 20 mmHg that was sustained for at least 30 min.(Reference Jansen and Lipsitz1)
Gastric emptying
GE was assessed using three-dimensional ultrasonography, using a Logiq™ 9 ultrasonography system (GE Healthcare Technologies, Sydney, Australia) with TruScan Architecture (i.e. built-in magnetically sensored three-dimensional)(Reference Gentilcore, Hausken and Horowitz30). For three-dimensional positioning and orientation measurement (POM), a transmitter was placed close to the subject, and a snap-on sensor attached to a 3.5C broad-spectrum 2·5–4 MHz convex transducer(Reference Gentilcore, Hausken and Horowitz30, Reference Tefera, Gilja and Olafsdottir31). As the transmitter produces a spatially varying magnetic field, and ferrous and conductive metals distort the magnetic field, all metal objects were removed from the subject and from the area directly between the POM transmitter and sensor(Reference Liao, Gregersen and Hausken32). The POM transmitter was placed behind (approximately 10 cm) the subject(Reference Gilja, Detmer and Jong33), at the level of the stomach, so that the subject was positioned between the ultrasound scanner and the transmitter. For three-dimensional data acquisition, the subject was scanned at t = − 2 and 0 min (i.e. immediately following drink consumption) and then at 15 min intervals between t = 0 and 120 min. A region of interest was drawn around the total stomach, and the volume of the drink in the total stomach was derived and expressed as a percentage of the original volume at t = 0 min (i.e. 100 %)(Reference Gentilcore, Hausken and Horowitz30). GE curves (expressed as % retention over time) were derived for the total stomach at t = 0, 15, 30, 45, 60, 75, 90, 105 and 120 min. The 50 % GE time was also determined.
Blood glucose, serum insulin, glucagon-like peptide-1 and glucose-dependent insulinotropic polypeptide concentrations
Venous blood samples were obtained before consumption of the drink (i.e. t = − 2 min) and at 15 min intervals between t = 0 and 120 min. Blood glucose concentrations (mmol/l) were determined immediately using a portable blood glucose meter (Medisense Precision Q·I·D™ System; Abbott Laboratories, Medisense Products, Inc., Bedford, MA, USA).
Serum was separated by centrifugation at 3200 rpm for 15 min at 4°C within 30 min of collection and stored at − 70°C until analysed. Serum insulin (mU/l) was measured by ELISA immunoassay (Diagnostics Systems Laboratories, Inc., Webster, TX, USA). Sensitivity of the assay was 0·26 mU/l, and CV was 2·6 % within assays and 6·2 % between assays(Reference O'Donovan, Doran and Feinle-Bisset34).
Serum GLP-1 (pmol/l) was measured by RIA (GLPIT-36HK; Linco Research, St Charles, MO, USA). Minimum detection limit was 3 pmol/l, intra-assay CV was 6·7 % and inter-assay CV was 7·8 %.
Serum GIP (pmol/l) was measured by RIA with some modifications to the original method(Reference Wishart, Morris and Horowitz35). The standard curve was prepared in buffer rather than in extracted charcoal-stripped serum, and the radio-iodinated label was supplied by Perkin Elmer (Boston, MA, USA). Minimum detection limit of the assay was 2 pmol/l, and both intra- and inter-assay CV were 11·2 and 11·6 %, respectively.
Autonomic function
Autonomic nerve function was assessed using standardised cardiovascular reflex tests(Reference Ewing and Clarke28, Reference Piha29). In brief, parasympathetic function was evaluated by the variation (R–R interval) of the heart rate during deep breathing and the response to standing (‘30:15’ ratio). Sympathetic function was assessed by the fall in systolic BP in response to standing. Each of the test results was scored according to age-adjusted predefined criteria as 0 = normal, 1 = borderline and 2 = abnormal for a total maximum score of 6. A score >3 was considered to indicate autonomic dysfunction(Reference Ewing and Clarke28, Reference Piha29).
Statistical analysis
Systolic and diastolic BP and heart rate were expressed as changes from baseline. GE, blood glucose, serum insulin, and GLP-1 and GIP concentrations were analysed as absolute values. One-way ANOVA was used to analyse the effects of ‘time’ on GE, systolic and diastolic BP, heart rate, blood glucose, serum insulin, and GLP-1 and GIP concentrations. The maximum fall in systolic and diastolic BP and maximum rise in heart rate were defined as the greatest change from baseline in each subject at any given time point for each treatment. For blood glucose, serum insulin, and GLP-1 and GIP concentrations, the peak absolute value was analysed in each subject at any given time point for each treatment. Areas under the curve (AUC), between t = 0 and 120 min, were calculated using the trapezoidal rule and analysed by one-way ANOVA to evaluate a ‘treatment’ effect for GE, systolic and diastolic BP and heart rate and between t = − 2 and 120 min for blood glucose, serum insulin, and GLP-1 and GIP concentrations. All analyses were performed using SPSS version 16.0.2 (SPSS, Inc., Chicago, IL, USA). Data are shown as changes from baseline and means with their standard errors, unless otherwise stated. The number of subjects studied was based on power calculations derived from our previous work; the sample size of eight subjects was calculated to have 80 % power at the P = 0·05 significance level to detect a difference in maximum fall in systolic BP between glucose and xylose of 7·3 mmHg(Reference Visvanathan, Chen and Garcia36). A P value < 0·05 was considered significant in all analyses.
Results
The studies were well tolerated, and there were no adverse events. No subject had definite autonomic neuropathy (mean score 0·9, range 0–2), or had PPH.
Blood pressure and heart rate
There was no difference in baseline (t = 0 min) BP or heart rate between the 3 d: systolic BP (‘W’ 118·4 (sem 6·0) mmHg v. ‘G’ 120·4 (sem 7·0) mmHg v. ‘X’ 118·9 (sem 5·6) mmHg; P = 0·44); diastolic BP (‘W’ 69·8 (sem 2·6) mmHg v. ‘G’ 71·1 (sem 2·7) mmHg v. ‘X’ 70·3 (sem 3·0) mmHg; P = 0·40); heart rate (‘W’ 57·9 (sem 2·2) beats per min (bpm) v. ‘G’ 58·8 (sem 3·0) bpm v. ‘X’ 59·1 (sem 2·6) bpm; P = 0·79).
Systolic blood pressure
Between t = 0 and 120 min, there was a fall in systolic BP during ‘G’ (P = 0·02) and no change during ‘W’ (P = 0·71) or ‘X’ (P = 0·63) (Fig. 1(a)). There was a treatment effect (P < 0·001) for the AUC of the change in systolic BP between t = 0 and 120 min, so that systolic BP was less during ‘G’ when compared with ‘W’ and ‘X’ (P = 0·003 for both), without any difference between ‘W’ and ‘X’ (P = 0·19). During ‘G’, the maximum fall in BP was 15·1 (sem 2·8) mmHg occurring at 64 (sem 9) min. At t = 120 min, systolic BP was not different from baseline after ‘W’ (120·6 (sem 6·0) mmHg; P = 0·23), ‘G’ (119·4 (sem 5·7) mmHg; P = 0·69) or ‘X’ (117·6·4 (sem 4·6) mmHg; P = 0·43).
Fig. 1 Change in (a) systolic blood pressure, (b) diastolic blood pressure and (c) heart rate (in beats per min; bpm) from baseline in response to oral water (W, ●), glucose (G, ○) and xylose (X, △). Values are means, with their standard errors represented by vertical bars (n 8). Mean values were significantly different for ‘G’ when compared with ‘W’ and ‘X’ for the systolic (*P = 0·003) and diastolic (**P ≤ 0·005) blood pressure treatment effects.
Diastolic blood pressure
Between t = 0 and 120 min, there was a fall in diastolic BP during ‘G’ (P = 0·003), and no change during ‘W’ (P = 0·88) or ‘X’ (P = 0·26) (Fig. 1(b)). There was a treatment effect (P < 0·001) for the AUC of the change in diastolic BP between t = 0 and 120 min, so that diastolic BP was less during ‘G’ when compared with ‘W’ (P = 0·002) and ‘X’ (P = 0·005), without any significant difference between ‘W’ and ‘X’ (P = 0·92). During ‘G’, the maximum fall in BP was 12·9 (sem 1·6) mmHg occurring at 56 (sem 11) min. At t = 120 min, diastolic BP was not different from baseline after ‘W’ (70·3 (sem 2·4) mmHg; P = 0·58), ‘G’ (69·6 (sem 3·3) mmHg; P = 0·41) or ‘X’ (71·9 (sem 2·6) mmHg; P = 0·27).
Heart rate
Between t = 0 and 120 min, there was no significant change in heart rate during ‘W’ (P = 0·22), ‘G’ (P = 0·28) or ‘X’ (P = 0·19) (Fig. 1(c)). At t = 120 min, heart rate was not significantly different from baseline after ‘W’ (57·5 (sem 2·5) bpm; P = 0·77), ‘G’ (59·5 (sem 2·8) bpm; P = 0·63) or ‘X’ (64·8 (sem 5·1) bpm; P = 0·13).
Gastric emptying
There was a significant treatment effect (P < 0·001) for the AUC for GE between t = 0 and 120 min (Fig. 2) . ‘W’ emptied in an overall exponential, and more rapid, fashion when compared with ‘G’ and ‘X’, which emptied linearly and more slowly (P < 0·001 for both), with no significant difference between ‘G’ and ‘X’ (P = 0·47). The 50 % GE time of ‘W’ (t = 19 (sem 3) min) was less than ‘G’ (t = 75 (sem 7) min) and ‘X’ (t = 75 (sem 8) min) (P < 0·001).
Fig. 2 Gastric emptying of water (W, ●), glucose (G, ○) and xylose (X, △). Values are means, with their standard errors represented by vertical bars (n 8). Mean values were significantly different for ‘W’ when compared with ‘G’ and ‘X’ in the treatment effect of the AUC (***P < 0·001).
Blood glucose
There was no difference in baseline (t = − 2 min) blood glucose between the 3 d (‘W’ v. ‘G’ v. ‘X’): 6·2 (sem 0·2) mmol/l v. 6·2 (sem 0·2) mmol/l v. 6·1 (sem 0·2) mmol/l; P = 0·89. Between t = − 2 and 120 min, there was a rise in blood glucose during ‘G’ (P < 0·001), and a slight rise following ‘X’ (P = 0·03), but no change during ‘W’ (P = 0·50) (Fig. 3(a)). There was a significant treatment effect (P < 0·001) for the AUC of the blood glucose concentration between t = − 2 and 120 min, so that the magnitude of the rise in blood glucose was much greater during ‘G’ compared with both ‘W’ (P ≤ 0·001) and ‘X’ (P ≤ 0·001). During ‘G’, peak blood glucose was 10·2 (sem 0·6) mmol/l at 53 (sem 8) min. At t = 120 min, blood glucose concentrations were not different from baseline after ‘W’ (6·1 (sem 0·1) mmol/l; P = 0·58), ‘G’ (6·8 (sem 0·5) mmol/l; P = 0·33), but were slightly higher after ‘X’ (6·5 (sem 0·2) mmol/l; P = 0·03).
Fig. 3 Change in (a) blood glucose, (b) serum insulin, (c) serum glucagon-like peptide-1 (GLP-1) and (d) serum glucose-dependent insulinotropic polypeptide (GIP) in response to oral water (W, ●), glucose (G, ○) and xylose (X, △). Values are means, with their standard errors represented by vertical bars (n 8). Mean values were significantly different for ‘G’ when compared with ‘W’ and ‘X’ for the blood glucose, serum insulin and serum GIP effects (***P ≤ 0·001). Mean values were significantly different for ‘X’ when compared with ‘W’ and ‘G’ for the serum GLP-1 treatment effect (**P ≤ 0·01).
Serum insulin
There was no difference in baseline (t = − 2 min) serum insulin between the 3 d (‘W’ v. ‘G’ v. ‘X’): 8·7 (sem 1·3) v. 8·5 (sem 1·1) v. 8·4 (sem 1·6) mU/l; P = 0·88. Between t = − 2 and 120 min, there was a rise in serum insulin during ‘G’ (P < 0·001), a trend for a fall during ‘W’ (P = 0·06) and no change during ‘X’ (P = 0·18) (Fig. 3(b)). There was a significant treatment effect (P < 0·001) for the AUC of serum insulin between t = − 2 and 120 min, so that the magnitude of the rise in serum insulin was much greater during ‘G’ compared with ‘W’ and ‘X’ (P < 0·001 for both), without any significant difference between ‘W’ compared with ‘X’ (P = 0·13). At t = 120 min, serum insulin concentrations were not different from baseline after ‘X’ (8·0 (sem 1·6) mU/l; P = 0·63), slightly lower following ‘W’ (7·3 (sem 1·0) mU/l; P = 0·03) and substantially higher after ‘G’ (42·8 (sem 10·1) mU/l; P = 0·009).
Serum glucagon-like peptide-1
There was no significant difference in baseline (t = − 2 min) serum GLP-1 between the 3 d (‘W’ v. ‘G’ v. ‘X’): 16·6 (sem 2·3) v. 13·8 (sem 1·4) v. 18·9 (sem 3·3) pmol/l; P = 0·08. Between t = − 2 and 120 min, there was a rise in serum GLP-1 during ‘G’ (P = 0·01) and ‘X’ (P < 0·001), but no change during ‘W’ (P = 0·39) (Fig. 3(c)). There was a significant treatment effect (P ≤ 0·001) for the AUC of serum GLP-1 concentration between t = − 2 and 120 min, so that the magnitude of the rise in serum GLP-1 was much greater during ‘X’ compared with ‘W’ (P ≤ 0·001) and ‘G’ (P = 0·002), with a trend for a difference between ‘G’ compared with ‘W’ (P = 0·07). During ‘G’, peak GLP-1 was 30·5 (sem 4·6) pmol/l at 26 (sem 5) min, and during ‘X’, peak GLP-1 was 42·0 (sem 4·0) pmol/l at 48 (sem 5) min (P < 0·05 for peak and P < 0·01 for time to peak). At t = 120 min, serum GLP-1 concentrations were not different from baseline after ‘W’ (15·7 (sem 1·2) pmol/l; P = 0·15) and ‘G’ (12·3 (sem 1·0) pmol/l; P < 0·001), but higher following ‘X’ (27·2 (sem 1·8) pmol/l; P = 0·002).
Serum glucose-dependent insulinotropic polypeptide
There was no significant difference in baseline (t = − 2 min) serum GIP between the 3 d (‘W’ v. ‘G’ v. ‘X’): 17·3 (sem 1·3) v. 18·4 (sem 1·6) pmol/l v. 18·9 (sem 1·6) pmol/l; P = 0·30. Between t = − 2 and 120 min, there was a prompt rise in serum GIP during ‘G’ (P < 0·001), and a fall, albeit minor, during ‘W’ and ‘X’ (P < 0·001 for both) (Fig. 3(d)). There was a significant treatment effect (P ≤ 0·001) of the AUC for serum GIP concentration between t = − 2 and 120 min, so that the magnitude of the rise in serum GIP was much greater during ‘G’ compared with ‘W’ and ‘X’ (P ≤ 0·001 for both), without any difference between ‘W’ compared with ‘X’ (P = 0·41). During ‘G’, peak GIP was 61·0 (sem 8·0) pmol/l at 56 (sem 11) min. At t = 120 min, serum GIP concentrations were not different from baseline after ‘W’ (15·7 (sem 2·0) pmol/l; P = 0·15), less following ‘X’ (16·2 (sem 1·1) pmol/l; P = 0·02) and greater after ‘G’ (48·9 (sem 4·7) pmol/l; P < 0·001).
Discussion
The present study indicates that oral xylose (50 g), unlike glucose, has no effect on BP in healthy older subjects despite emptying from the stomach at a comparable rate. Xylose is also more potent than glucose in stimulating GLP-1, but has no effect on GIP and has minimal effect on glycaemia and insulinaemia, at least during euglycaemia.
The present study confirms that oral glucose induces a substantial fall (15·1 (sem 2·8) mmHg) in systolic BP in healthy older subjects, studied under resting conditions. Previous studies relating to the effects of xylose on BP have been inconsistent(Reference Robinson, Stowell and Purdie15–Reference Mathias18), but GE was not measured in any of these studies, and may have potentially accounted for the observations, given that the rate of nutrient delivery into the small intestine affects the fall in BP both as a result of gastric distension(Reference Gentilcore, Meyer and Rayner10, Reference Jones, Tonkin and Horowitz37) and as a result of the exposure of the small intestine to nutrients(Reference O'Donovan, Feinle and Tonkin7).
The present study establishes that glucose and xylose empty from the stomach at a comparable rate with an overall linear pattern that is substantially slower than water, which empties exponentially, consistent with a previous animal (primate) study(Reference Moran and McHugh19). Hence, GE does not account for the different effects of glucose and xylose on BP. The regulation of the GE of nutrients arises predominantly as a result of inhibitory feedback from receptors in the small intestine, the magnitude of which is dependent on the length and, possibly, region(Reference Lin, Doty and Reedy38) of the small intestine exposed, as influenced by the energy load. Accordingly, it appears that the magnitude of this inhibitory feedback is comparable for xylose and glucose, although the mechanism(s) which account for this feedback may differ(Reference Moran and McHugh19). In humans, a study in healthy adult males reported that xylose in a dose of 25 g in 50 ml water, given immediately after the consumption of a scrambled egg meal, markedly prolonged GE, when compared with the same amount of glucose(Reference Shafer, Levine and Marlette20). Differences in the rate of the GE of xylose between these studies, possibly influenced by the xylose dose, may account for the discrepant observations.
While it is clear that differences in GE do not account for the substantial, differential effects of xylose and glucose on BP, the two sugars had discrepant effects on glycaemia, insulinaemia and the secretion of the incretin hormones, GIP and GLP-1, which, accordingly, warrant consideration. It is well documented, and confirmed in the present study, that xylose has minimal, if any, effect on plasma glucose or insulin(Reference Robinson and Potter16–Reference Mathias18). However, both hyperglycaemia and hyperinsulinaemia are unlikely to play a major role in PPH, e.g. intravenous glucose has little, if any, effect on BP(Reference Maule, Tredici and Dematteis39). The comparative effects of xylose and glucose on splanchnic blood flow remain to be determined, and it is possible that the relatively poorly absorbed xylose induces a lesser increase. This is the first evaluation of the effect of xylose on the release of GLP-1 and GIP – that xylose had no effect on GIP is predictable, given that the secretion of GIP occurs predominantly in the proximal small intestine and, in the case of carbohydrate, appears to be dependent on an affinity for the transporter, sodium-dependent glucose cotransporter-1(Reference Baggio and Drucker27). There is also no evidence that GIP affects BP. It has been reported that xylose has no effect on GLP-1 secretion in the dog(Reference Shima, Suda and Nishimoto26), although xylose apparently stimulates the release of glucagon-like immunoreactivity in the canine intestine(Reference Marco, Valverde and Faloona40). The present study establishes that xylose is a potent stimulant of GLP-1 in humans – the sustained stimulation is likely to reflect the delay in intestinal absorption when compared with glucose, so that the distal small intestine is exposed; the initial stimulation appeared similar to that induced by glucose. It is not surprising that the stimulation of GLP-1 by xylose was not associated with a substantial increase in serum insulin in the present study, as the insulinotropic property of GLP-1 is known to be glucose-dependent, i.e. GLP-1 has little, if any, effect on insulin during euglycaemia(Reference Holst and Gromada21). It is accordingly probable that xylose will stimulate insulin in type 2 patients during hyperglycaemia by increasing GLP-1. The stimulation of GLP-1 secretion by xylose may also be of relevance to the use of dipeptidyl peptidase-IV inhibitors and GLP-1 analogues in the management of type 2 diabetes(Reference Khoo, Rayner and Jones41). As discussed, this stimulation of GLP-1 may account for the absence of any fall in BP. We studied a small number of subjects precluding assessment of meaningful correlations. Further studies are required to address this issue, including the effects of different xylose loads. Given that GLP-1 plays a physiological role to slow GE(Reference Deane, Nguyen and Stevens42), it is perhaps surprising that xylose did not empty from the stomach slower than glucose. However, it should also be recognised that glucose ingestion increased the blood glucose concentration substantially, whereas xylose did not, and elevations of blood glucose, even within a normal postprandial range, slow GE(Reference Schvarcz, Palmer and Aman43). It is also not known whether the presence of xylose in a glucose drink could attenuate the fall in BP. Furthermore, the effects of xylose in patients with PPH remain to be determined. In considering the potential dietary use of xylose, it should be recognised that while xylose is palatable, it is relatively expensive. In view of our observations, it would be of interest to evaluate the effects of the related pentose sugar, xylitol(Reference Shafer, Levine and Marlette44), which is considerably cheaper.
In summary, in healthy older subjects, oral xylose, unlike glucose in a dose of 50 g, has no effect on BP, despite emptying from the stomach at a comparable rate with glucose, and is a potent stimulant of GLP-1 secretion. These observations suggest that xylose may represent an alternative sweetener to glucose in the management of PPH.
Acknowledgements
The authors have no conflicts of interest or financial interests to declare. The present study was supported by the National Health and Medical Research Council (NHMRC) of Australia. Professor Jones' and Professor Feinle-Bisset's salaries are funded by the NHMRC of Australia, and Dr Gentilcore was supported by a Postdoctoral Fellowship from the National Heart Foundation of Australia (PR 07A 3309). The purchase of the Logiq™ 9 ultrasonography system was supported by an Equipment Grant from the NHMRC of Australia and funds from the University of Adelaide and GE Medical Systems Australia. The authors' contributions are as follows: L. V. performed the acquisition of subjects, data collection, analysis and interpretation, and preparation of the manuscript; T. H. and D. G. contributed to the data collection and preparation of the manuscript; R. S. R. performed the data collection; C. K. R. and C. F.-B. contributed to the preparation of the manuscript; M. H. performed the data analysis and interpretation, and contributed to the preparation of the manuscript; K. L. J. contributed to the concept and design of the study, data analysis and interpretation, and preparation of the manuscript.
