Gastric Bypass Reduces Symptoms and Hormonal Responses in Hypoglycemia

Gastric bypass (GBP) surgery for the treatment of morbid obesity induces marked weight loss and metabolic effects. Excess BMI loss is typically ∼70–80% (1), and almost instant changes occur in glucose homeostasis postsurgery, including frequent remission of diabetes up to 70% (1). After a meal, the rise in glucose and insulin is ∼50% higher, reflecting the faster absorption of glucose postsurgery. Incretins are also affected; prandial gastric inhibitory polypeptide (GIP) and glucagon-like peptide 1 (GLP-1) peak at ∼2 and 10 times higher levels after surgery, respectively (2).

The sustained weight loss achieved by GBP surgery, superior to that achieved by diet regimens, reflects lowered caloric intake and possibly altered secretion of gastrointestinal peptides (2). Comparing eating habits before and after the GBP surgery, patients lowered the intake with marked reductions in carbohydrate intake in the form of sweets, soda, and milk/ice cream. Fear of dumping was suggested as the main mechanism (3). Dumping can occur within 10–30 min of food intake when, in the absence of a pyloric function, it quickly enters the intestine and produces an osmotic effect. Fluid is shifted from the circulation into the intestine, resulting in a fall of blood pressure and tachycardia. In a rat model, similar food preference adaptations evolved with time, indicating that the preference shift was dependent on learning and dumping (4).

In a recent study with continuous glucose monitoring (5), we observed hypoglycemic episodes (glucose <3.3 mmol/L [60 mg/dL]) in 50% of patients who had undergone GBP surgery, of whom none had experienced symptoms of hypoglycemia. On average, the patients spent 40 min per 24-h period in a hypoglycemic glucose range, mostly (80%) without any symptoms. Goldfine et al. (6) observed a high frequency of asymptomatic hypoglycemia after a mixed-meal test in patients after GBP surgery, and another study (7) reported asymptomatic hyperinsulinemic hypoglycemia in a small number of patients who have undergone gastric banding. Furthermore, in a well-documented study of eight subjects with massive weight loss after vertical banded gastroplasty, Guldstrand et al. (8) have demonstrated a reduction in the hypoglycemic response of classic counterregulatory hormones.

The brain relies on glucose as the sole source of energy unless ketone bodies are provided, typically upon starvation. Gluco-sensing neurons are placed in the brain (e.g., in the hypothalamus) and in peripheral tissues, such as the carotid body, oral cavity, gut, and the hepatic portal vein, conveying signals to the brain (9). Upon perceived hypoglycemia, the brain acts on the anterior pituitary, pancreas, and adrenal medulla to increase the levels of counterregulatory hormones such as growth hormone (GH), glucagon, cortisol, epinephrine, and norepinephrine (9,10). In addition, endocrine cells in the pancreatic islets can respond directly to surrounding glucose levels. Autonomic nerve activation contributes to hormonal responses and also directly influences glucose turnover in liver, muscle, and adipose tissue (11). Furthermore, the secretion of the incretins GLP-1 and GIP may be involved, regulated by vagal signaling and the enteric nervous system (12). These systems provide a highly conserved and powerful integrated defense against severe hypoglycemia.

In view of previous findings by our group and others, we hypothesized that the counterregulatory response to hypoglycemia would be attenuated postsurgery. In the current study, we examined symptoms and counterregulatory responses during hyperinsulinemic-hypoglycemic clamps in patients before and after GBP surgery. Further, we analyzed GLP-1 and GIP in view of the upregulation of their prandial levels after undergoing GBP surgery (13), as well as their reported role in enhancing the glucagon response to hypoglycemia (14).

Morbidly obese (BMI >35 kg/m²) patients without diabetes were recruited at the Metabolic Outpatient Clinic of the Uppsala University Hospital. Patients accepted for bariatric surgery were consecutively invited to participate in the study. Fifteen patients were enrolled in the study, but three patients discontinued their participation after the first clamp, one due to pregnancy and two due to lack of time. The per-protocol cohort thus consisted of eight women and four men who were examined 3 months (1–6) before and 4–5 months after surgery using a hyperinsulinemic-hypoglycemic clamp. All patients underwent laparoscopic Roux-en-Y GBP surgery, including a 100-cm Roux limb connected to a small proximal gastric pouch and a 50-cm biliopancreatic limb. The GBP surgery was preceded by 4 weeks of eating a low-calorie diet, according to clinical routine, to reduce liver size and intestinal fat. Baseline characteristics are shown in Table 1.

A hyperinsulinemic-hypoglycemic clamp was performed after an overnight fast (water intake allowed). The clamp was modified according to the study by Norjavaara et al. (15). After a priming infusion, insulin was infused at a fixed rate, 80 mU/m² body surface/min together with a variable glucose (200 mg/mL) infusion to maintain plasma glucose at levels of 5 mmol/L (90 mg/dL) for 0–60 min, 4 mmol/L (72 mg/dL) for 60–90 min, 3.2 mmol/L (58 mg/dL) for 90–135 min, and 2.7 mmol/L (49 mg/dL) for 135–165 min.

At 165 min, the insulin infusion was stopped, with glucose infusion continued until a glucose level of >4.0 mmol/L (72 mg/dL) was achieved. Plasma glucose in arterialized venous blood was measured every 5 min during the clamp with the Contour Glucose Meter (Bayer Healthcare, Leverkusen, Germany). Levels of insulin, C-peptide, GH, glucagon, cortisol, epinephrine, norepinephrine, GLP-1, and GIP were measured at 0, 60, 90, 120, 135, 150, and 165 min during clamp, and free fatty acids (FFAs) and glycerol were measured at 0, 135, and 165 min. Participants were continuously monitored with electrocardiogram, and recordings were used for heart rate variability (HRV) assessment.

The Edinburgh Hypoglycemia Score, composed of the 11 symptoms statistically derived to most closely associate with hypoglycemia, was used (16). The 11 hypoglycemic symptoms contain four autonomic symptoms (sweating, palpitation, shaking, and hunger), five neuroglycopenic symptoms (confusion, drowsiness, odd behavior, speech difficulty, and incoordination), and two malaise symptoms (nausea and headache). The participants graded their symptoms between 1 (no symptoms) and 7 (maximum symptoms) at 0, 60, 90, 120, 135, 150, and 165 min, and scores for all 11 symptoms were added.

Blood samples were taken at 8:00 a.m. after 10 h of fasting. Routine blood chemistry and most hormonal analyses were performed at the Department of Clinical Chemistry at the University Hospital, Uppsala, Sweden. If not analyzed immediately, samples were frozen at −80°C. The following analyses were used: insulin (Cobas e; Roche), cortisol (Cobas e; Roche), GH (Immulite XP; Siemens Healthcare Global), and C-peptide (Cobas e; Roche). Catecholamine analyses (liquid chromatography) were performed at the Laboratory of Clinical Chemistry at the Karolinska Universitetssjukhuset, Stockholm, Sweden).

Specific analyses were performed at the Clinical Diabetes Research Laboratory: glucagon was measured with ELISA (Glucagon #10–1271–01; Mercodia, Uppsala, Sweden). Total GLP-1 and total GIP were measured with ELISA (Merck Millipore, Darmstadt, Germany), FFAs were measured with the Free Fatty Acid Fluorometric Assay Kit (Cayman Chemical, Ann Arbor, MI), and glycerol was measured with Free Glycerol reagent (Sigma-Aldrich, St. Louis, MO). HOMA of insulin resistance was calculated as follows: (fasting insulin [in mU/L] × glucose [in mmol/L])/22.5.

As a marker of efferent activity in the autonomic nervous system, HRV analyses were performed as described previously (17,18). Six of the 12 subjects had complete recordings both presurgery and postsurgery that were used for the statistical analysis. In the other subjects, at least one recording was incomplete because of technical issues. The total spectral power (P_TOT), the power of the low-frequency component (P_LF) (0.04–0.15 Hz), and the power of the high-frequency component (P_HF) (0.15–0.50 Hz), all log transformed, were calculated over consecutive 5-min periods from the complete recording. P_HF mainly reflects the parasympathetic activity, whereas P_LF reflects a combination of sympathetic and parasympathetic activity, and the P_LF/P_HF ratio is used as a marker of the balance between sympathetic and parasympathetic activity (18). The HRV analysis was performed using Matlab Software (MathWorks, Natick, MA).

Statistical calculations were performed in Statistica (Dell Statistica; StatSoft, Aliso Viejo, CA) except for the analysis of HRV indices, which were performed using R (version 3.1, 2014, Vienna, Austria). Shapiro-Wilks tests were performed to assess normal distributions, in addition to visual graph inspection. Anthropometric and fasting laboratory results presurgery and postsurgery were compared using Student t tests. Metabolite and hormone levels during hypoglycemic clamp were analyzed with repeated-measures ANOVA. Areas under the curve during the hypoglycemic period (AUC_hypo) (90–165 min) were analyzed by paired t tests. Symptom scores were compared using paired t tests.

For HRV analysis, the time-variant changes in HRV during the complete recording were modeled using generalized additive mixed-effects models using thin plate splines (19). Differences between presurgery and postsurgery recordings were modeled as a binominal categorical variable. Changes during the euglycemic phase (0–60 min) and hypoglycemic phase (90–165 min) were evaluated by ANOVA for repeated measurements.

The Regional Ethics Committee of Uppsala, Sweden, approved this study (Dnr 2013/480). Patients gave written informed consent, and the study was conducted according to the tenets of the Declaration of Helsinki.

GBP surgery induced marked metabolic effects, notably lower glucose, insulin, and HbA_1c levels (Table 1).

The glucose, insulin, and C-peptide levels during the clamp investigations are depicted in Fig. 1. The mean (SD) glucose levels achieved during the clamp investigations, presurgery versus postsurgery, were as follows: for the 5.0 mmol/L target period 5.0 (0.4) mmol/L vs. 5.0 (0.4) mmol/L; for the 4.0 mmol/L target period 4.3 (0.3) mmol/L vs. 4.1 (0.3) mmol/L; for the 3.2 mmol/L time period 3.2 (0.3) mmol/L vs. 3.0 (0.2) mmol/L; and for the 2.7 mmol/L target period: 2.9 (0.3) mmol/L vs. 2.8 (0.2) mmol/L (all differences were nonsignificant). The circulating levels of insulin during insulin infusion increased to ∼200 mU/L before surgery and 150 mU/L after surgery, suggesting an increase in insulin clearance after bariatric surgery, which is in line with what has been previously reported (8,20). The glucose infusion rate was significantly higher during the postsurgery clamp, indicating improvement in insulin sensitivity.

Patients experienced fewer symptoms during hypoglycemia in the postsurgical clamp. The total mean composite Edinburgh Hypoglycemia Score at the end of hypoglycemia (165 min) was 24.0 (SD 6.9) before surgery and 18.5 (5.2) after surgery. The hypoglycemia-induced increase in total symptom score was halved after surgery (Table 2).

Heart rate increased during both clamps, from 65 to 75 bpm presurgery and from 56 to 69 bpm postsurgery, respectively. The mean time to peak heart rate was longer postsurgery; peaks appeared at 130 min presurgery and at 150 min postsurgery. There were no presurgery versus postsurgery difference in hypoglycemia-induced increase in systolic or diastolic blood pressure or heart rate (Table 2).

Counterregulatory hormone levels are shown in Fig. 2 and Table 3. The hypoglycemia-induced glucagon response was markedly decreased postsurgery. During presurgery examination, the glucagon rise started at 120 min and during the postsurgery examination at 135 min. Cortisol levels were similar during the euglycemic phase of both clamps, and hypoglycemia responses became significant at 135 min both presurgery and postsurgery but were lower postsurgery than presurgery (Table 3). Epinephrine and norepinephrine levels rose at 120 min during the clamp presurgery and later (at 135 min) postsurgery. The epinephrine and norepinephrine responses during hypoglycemia were lower postsurgery. Levels of GH were similar up to 120 min and then increased during the presurgery clamp. Postsurgery, the GH levels responded later, at 135 min, and reached a threefold higher level postsurgery versus presurgery.

Plasma levels of incretin hormones GLP-1 and GIP are shown in Table 3 and Fig. 2. Both hormones rose during the hypoglycemic part of the clamps. The responses were significantly attenuated postsurgery versus presurgery.

FFAs and glycerol exhibited equal fasting levels presurgery and postsurgery. The levels fell, FFAs relatively more than glycerol, during the hyperinsulinemic-hypoglycemic clamps. The nadir levels were lower postsurgery versus presurgery (Fig. 3 and Table 3).

When comparing HRV presurgery and postsurgery, there were significant increases in P_TOT (total variability measure) and spectral components (P_LF, P_HF) during both euglycemia and hypoglycemia (Table 4). P_LF/P_HF ratio, a measure of sympathetic/parasympathetic balance, showed less of an increase from the euglycemic part of the clamp to the hypoglycemic part after surgery compared with before surgery (Fig. 2). The interbeat R-R interval was longer postsurgery versus presurgery, corresponding to the decrease in heart rate, both during euglycemia and hypoglycemia.

In this study, we found that GBP surgery is followed by reduced symptoms and responses in classic counterregulatory hormones and autonomic nervous outflow during hypoglycemia. GLP-1 and GIP secretion rose during hypoglycemia, implicating a role of the incretins in counterregulation.

The glucagon, cortisol, catecholamine, and incretin responses to hypoglycemia were all attenuated postsurgery and showed a similar pattern with later and lower responses to hypoglycemia at the post-GBP examinations. GH, like the other counterregulatory hormones, exhibited a later response postsurgery compared with presurgery. During the postsurgery clamp, however, the magnitude of the response was higher than during the presurgery clamp. This may nonetheless be attenuated, in light of the participants’ loss of weight after surgery. GH levels are suppressed in obesity (21), and Corneli et al. (22) found a mean GH response of ∼30 μg/L in subjects with a BMI of ∼30 kg/m² who were subjected to a GH-releasing hormone-arginine test compared with the rise to 15 μg/L during postsurgery hypoglycemia in the current study (BMI 30.1 kg/m²).

Insulin levels during clamps were ∼25% lower postsurgery versus presurgery, and the absolute insulin amount infused was ∼12% lower after surgery (data not shown), indicating increased insulin clearance, and more so if adjustments to body surface are performed. The difference in insulin concentration was not correlated with changes in hormonal responses (data not shown). The findings are in accordance with those of previous reports, and a more effective hepatic clearance after bariatric surgery has been proposed (8,20).

The nadir levels of FFAs and glycerol were lower postsurgery versus presurgery, reflecting the composite effects of increased insulin sensitivity; reduced levels of counterregulatory hormones, in particular catecholamines, glucagon, and cortisol; and attenuation of the sympathetic nervous response (see paragraph below). Sympathetic activity in the adipose tissue can stimulate lipolysis directly via norepinephrine release (reviewed in Bartness et al. [23]).

The HRV results are in accordance with a reduced sympathetic response to hypoglycemia postsurgery, which could contribute to the delayed and attenuated catecholamine levels and lipolysis. In a previous study (24), a rapid improvement in vasoreactivity and a reduction of heart rate were observed, indicating a reduced sympathetic activity after GBP.

Presurgery, GLP-1 levels doubled during the hypoglycemic phase, an increase of the same order as the response to hyperglycemia after a mixed meal in healthy control subjects (13). This is, however, less than the marked increase in GLP-1 levels typically seen during the postprandial hyperglycemic phase postsurgery (25). GLP-1 level has a role in glucagon secretion, and it may influence the α-cell either directly or via indirect mechanisms like insulin or autonomic nervous control (26). In fact, the effects on both insulin and glucagon secretion may be mediated primarily via nervous system circuits (27). There are data suggesting an effect of GLP-1 analogs on increasing glucagon secretion during hypoglycemia (28), an action similar to that of GIP (14). In a previous study (14), when GIP was infused in physiological doses during hypoglycemia, the glucagon level rose 1.5 times more than with saline infusion.

In patients with diabetes treated with insulin, hypoglycemia is known to lower the threshold for counterregulatory hormonal responses and symptoms (29). Resetting of the hypoglycemia threshold is also known to occur in pregnancy (30) and has been described in healthy subjects after exercise (31). Boyle et al. (32) kept healthy subjects in a hyperinsulinemic-hypoglycemic clamp (glucose 2.9 mmol/L [52 mg/dL]) for 4 consecutive days and reported that on day 1 the threshold for cognitive impairment was 3.05 mmol/L (55 mg/dL) and for symptoms 3.6 mmol/L (65 mg/dL), whereas it decreased to 2.5 mmol/L (45 mg/dL) for both on the final day. Collectively, previous reports and the present data illustrate a remarkable ability of the brain to adapt to lower glucose levels in different physiological settings. The potential adverse effects of an asymptomatic low glucose level have not been well established.

Our present data show an attenuated hypoglycemic response in multiple neuroendocrine pathways after GBP surgery. This indicates that glucose sensing and possibly glucose utilization in the brain have been augmented, but the underlying mechanisms need to be further characterized. Glucokinase activity in ventromedial hypothalamus modulates the counterregulatory responses, and an increased activity attenuates hormonal responses to hypoglycemia (33). An increase in gene expression for glucokinase, GLUT2, and GLP-1 receptor in the hypothalamus has been reported in food-restricted rats (34). Possibly, increased brain glucose uptake leads to an enhanced satiety signal in proopiomelanocortin neurons in the hypothalamus (35) and contributes to the weight maintenance seen in patients who have undergone GBP surgery.

The current study is limited by the relatively few participants. Moreover, the hypoglycemia was induced under experimental circumstances and not during real-life conditions, when the insulin concentration would drop during hypoglycemia. However, the participants in the current study closely resemble the average patient after GBP surgery (1,36) in that they were aged 43 years, two of three were women, and they had lost ∼10 BMI units. The activity level of participants was not known.

In conclusion, we found that GBP surgery is followed by attenuated symptoms, counterregulatory hormonal responses, and sympathetic nervous activation during hypoglycemia. We suggest that such adaptive mechanisms contribute to the occurrence of asymptomatic hypoglycemia and to the prevention of type 2 diabetes after GBP surgery.

Acknowledgments. The authors thank Jan Hall for invaluable work with the clamps and Prasad Kamble, Cherno Sidibeh, Caroline Moberg, Lovisa Nordlinder, and Maria Joao Pereira for invaluable assistance during clamps and laboratory work (all from the Department of Medical Sciences, Uppsala University). The authors also thank Marcus Karlsson (Department of Radiation Sciences, Biomedical Engineering, Umeå University) for analysis of the electrocardiogram recordings and calculation of the HRV indices.

Funding. This study was funded by grants from the Research Fund of the Swedish Diabetes Association, Exodiab, Ernfors Foundation, and ALF (Swedish Government Research Fund).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. N.A., F.A.K., and J.W.E. designed the study; researched the data; and wrote, revised, and approved the manuscript. J.L.B., M.S., and U.W. researched the data and revised and approved the manuscript. N.A. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in poster form at the 76th Scientific Sessions of the American Diabetes Association, New Orleans, LA, 10–14 June 2016.