Results
Subject Characteristics
Baseline characteristics of the study groups are presented in Table 1. All obese subjects and healthy controls were Caucasian females with a mean age of 49·4 ± 0·6 years. We included 32 subjects with T2DM and 30 NGT obese individuals. Eight subjects dropped out during the course of the study either because they were unable to comply with the VLCD (n = 2), or because of mild postoperative complications after RYGB (n = 3, of which 2 had T2DM) or logistical issues (n = 3).
Prior to intervention, 50% of T2DM patients used antihypertensives against 33% in NGT patients (P = 0·15, Table S1 http://onlinelibrary.wiley.com/store/10.1111/cen.12496/asset/supinfo/cen12496-sup-0001-suppmat.doc?v=1&s=34462a94371903d4bb1a9315c6123ef09a5fa9e1). Pre-intervention use of oral antidiabetics was comparable between both groups of diabetic subjects. At the day of intervention, all blood glucose lowering agents were discontinued to avoid hypoglycaemia. Only Metformin treatment was reinstalled if fasting blood glucose levels remained above 7 mm after intervention (27% of subjects after RYGB vs 17% of subjects after VLCD, P = 0·32).
Effects of Intervention on Weight Loss and Glucose Homeostasis
Effects of dietary or surgical intervention on weight loss and glucose homeostasis in this cohort were recently reported. In brief, relative weight loss after 3 weeks was similar (4·8–7·3%) in NGT and T2DM groups with patients remaining markedly obese (BMI 37·7–40·9 kg/m). This was accompanied by decreased fasting glycaemia and postprandial glucose excursion in T2DM patients. After 3 months, GB was less effective (10·2%) in inducing weight loss than RYGB or dietary intervention (14·6–16·9%). Fasting hyperglycaemia and hyperinsulinemia were largely normalized or reversed at this later timepoint, indicating restored insulin sensitivity.
Baseline Fasting Bile Salt, FGF19 and FGF21 Levels
Total bile salt levels prior to intervention were not different between lean controls and NGT or T2DM obese subjects (Table 1, Fig. 1a). Baseline FGF19 levels were reduced in NGT obese subjects as compared to lean controls and obese T2DM subjects who had similar levels (Table 1, Fig. 1b). Baseline fasted FGF21 levels were higher in T2DM obese subjects compared to healthy lean controls or NGT obese subjects (Table 1, Fig. 1c).
(Enlarge Image)
Figure 1.
Calorie restriction and Roux-en-Y gastric bypass have opposite effects on bile salt and FGF21 levels in NGT and T2DM obese subjects. (a–c) Baseline fasted levels of total bile salts, FGF19 and FGF21 in lean controls (n = 12) and obese subjects with normal glucose tolerance (NGT,n = 27) or diabetes (DM,n = 27). (d–f) Fasted levels of afore mentioned analytes at baseline, and at 3 weeks (3 w) and 3 months (3 m) after intervention. (g–i) Prandial responses in lean controls (open symbols), and obese subjects with normal glucose tolerance (NGT, grey symbols) or diabetes (DM, black symbols) at baseline. Asterisk denotes a significant difference (P < 0·05) between groups at baseline, or within groups after intervention. n.s., nonsignificant; NGT, normal glucose tolerance; DM, diabetes; GB, gastric banding; RYGB, Roux-en-Y-gastric bypass; VLCD, very-low-calorie diet; min, minutes.
Postprandial Bile Salt, FGF19 and FGF21 Response
Bile salt levels increased after intake of a mixed-meal in all groups with levels peaking at 30–60 min (Fig. 1g). The postprandial bile salt response at baseline was similar in all groups (time × group effect: P = 0·07). As expected, the postprandial increase in FGF19 levels lagged behind the postprandial rise of bile salts, and was first apparent after 2 h (Fig. 1h). The postprandial FGF19 response at baseline was similar in all groups (P = 0·12). A standard mixed-meal test resulted in a significant postprandial decline of FGF21 level in all groups, which was more pronounced in T2DM obese subjects as compared to NGT obese subjects (group × time: P = 0·042) and lean controls (group × time: P < 0·001; Fig. 1i).
Effects of Intervention on Fasting Bile Salt, FGF19 and FGF21 Levels
Bile salt levels were not significantly changed 3 weeks after GB, but after 3 months, a reduction of serum bile salts was seen (−20%, P = 0·034; Table 2, Fig. 1d). Likewise, VLCD resulted in decreased bile salt levels that reached significance at the 3 weeks and 3 months timepoint (respectively −39%, P = 0·03 and −35%, P = 0·046; Table 2, Fig. 1d). Fasting bile salt levels were not affected by RYGB in either NGT or T2DM subjects (Table 2, Fig. 1d). Analysis of pooled data of all subjects undergoing RYGB (n = 31), however, showed a significant increase in bile salt levels after 3 weeks (4·1 ± 0·4–5·7 ± 0·8 μm;P = 0·049).
Only RYGB in NGT subjects caused a slight increase in fasting FGF19 levels (+52%, P = 0·045) after 3 weeks (Fig. 1e, Table 2). None of the interventions, however, caused a lasting change in fasting FGF19 levels. When data from RYGB subjects was pooled, there was a trend towards an increase in FGF19 levels after 3 weeks (0·10 ± 0·01–0·14 ± 0·03 ng/ml, +41%, P = 0·063).
Gastric Banding did not affect FGF21 levels after 3 weeks (P = 0·15), but resulted in a decrease after 3 months (−33%, P = 0·012; Table 2, Fig. 1f). The other restrictive intervention (VLCD) led to sustained decline (2·2–2·5-fold) of FGF21 levels after 3 weeks (P = 0·002) and 3 months (P = 0·004, Table 2, Fig. 1f). In contrast, RYGB induced an increase in FGF21 level after 3 weeks in both NGT (P = 0·002) and T2DM (P = 0·007) subjects with levels remaining significantly elevated after 3 months in NGT subjects (P = 0·004, Table 2, Fig. 1f). Analysis of the entire group of subjects undergoing RYGB, revealed increased FGF21 levels 3 weeks (P < 0·001) and 3 months (P = 0·052) after surgery (data not shown). Between-group analysis showed a significant different effect of VLCD as compared to RYGB at both timepoints and of GB as compared to RYGB after 3 months on FGF21 levels (P < 0·001).
Effects of Intervention on Postprandial Bile Salt, FGF19 and FGF21 Response
None of the interventions had an effect on the postprandial bile salt response (Ptime × occasion=n.s., Fig. 2a–d). Nevertheless, the altered small intestinal anatomy after RYGB appears to underlie a decline in time to peak, at least in the T2DM subjects (T2DM: 110 ± 11–72 ± 14 min; P = 0·034, NGT: 78 ± 13–84 ± 14 min; P = 0·74; Fig. 2c).
(Enlarge Image)
Figure 2.
The prandial response of FGF21 is altered after Roux-en-Y gastric bypass. Prandial responses of bile salts (a–d), FGF19 (e–h) and FGF21 (i–l) at baseline (open symbols) and 3 weeks after intervention (black symbols). Values are depicted as mean ± SEM. Significant effects (P < 0·05) of time (during meal tolerance test) or occasion (difference in response between occasions,) as analysed by repeated measures anova are displayed in the figures. NGT, normal glucose tolerance; DM, diabetes; GB, gastric banding; RYGB, Roux-en-Y-gastric bypass; VLCD, very-low-calorie diet; min, minutes.
The restrictive interventions had no effect on the postprandial FGF19 response (Ptime × occasion=n.s., Fig. 2e and h). RYGB, however, led to a leftward shift (i.e. an earlier time to peak: 145 ± 17–128 ± 11 min; P = 0·41) in the FGF19 response curve in T2DM subjects (Ptime × occasion = 0·042, Fig. 2g). A similar trend was observed in NGT subjects (Ptime × occasion = 0·095, time to peak: 139 ± 8–92 ± 10 min, P = 0·002, Fig. 2f).
GB did not affect the postprandial FGF21 response following a mixed-meal test (Ptime × occasion = 0·21, Fig. 2i). The postprandial decline of FGF21 levels was, however, less pronounced after VLCD (Ptime × occasion = 0·008, Fig. 2l). RYGB resulted in an altered postprandial FGF21 response in both NGT and T2DM subjects. In both subject groups, a transient drop in FGF21 level was apparent after RYGB at the 2 h timepoint. Moreover, both NGT and T2DM subjects showed a less pronounced postprandial FGF21 response after RYGB (NGT: Ptime × occasion = 0·041, T2DM: Ptime × occasion < 0·01, Fig. 2j and k).