Glucagon Treatment for Post-Gastric Bypass Hypoglycemia

Abstract

Hyperinsulinemic hypoglycemia is a recently described complication of Roux-en-Y gastric bypass (RYGB). We hypothesized that glucagon administration would help maintain normal postprandial plasma glucose concentrations by stimulating hepatic glucose output, and if so, represent a new therapeutic option for postbypass hypoglycemia. In this study, we compared the insulin and glycemic response to a mixed meal with and without concomitant glucagon infusion in a patient with severe recurrent hypoglycemia after RYGB. Although effective in transiently raising postprandial plasma glucose values, glucagon infusion was also associated with higher insulin concentrations, and failed to prevent symptomatic hypoglycemia. This case demonstrates that glucagon may have limited clinical utility in the treatment of post-RYGB hyperinsulinemic hypoglycemia.

A 59-year-old woman with prior history of hypertension, hyperlipidemia, and hypothyroidism underwent uncomplicated Roux-en-Y gastric bypass (RYGB) as a treatment for morbid obesity. Her preoperative BMI was 62.6 kg/m²; 1 year postoperatively her BMI was 40.5 kg/m² and subsequently remained stable. She developed vitamin B₁₂ and D deficiencies, and osteopenia. Her medications included levothyroxine, simvastatin, enalapril, metoprolol, B complex and multivitamins, vitamin D, calcium, and aspirin.

Two years following bariatric surgery the patient developed episodes of light-headedness and diaphoresis, occurring predictably 30–60 min after meals. In the context of these symptoms, capillary blood glucose monitoring revealed values as low as 30 mg/dl on multiple occasions. The episodes became progressively more frequent and debilitating, including neuroglycopenia with loss of consciousness and a motor vehicle accident.

The patient was initially treated with dietary modification, including frequent small meals of low glycemic index and cornstarch supplementation. Her symptoms improved but did not resolve. α-Glucosidase inhibitor (acarbose) therapy to slow carbohydrate absorption and thus diminish insulin secretion resulted in initial improvement, but neuroglycopenia recurred and diazoxide was added to reduce insulin secretion. Selective arterial catheterization with dynamic calcium infusion (1) demonstrated increased calcium-stimulated insulin secretion from both the head and tail of the pancreas. Partial pancreatectomy was considered but deferred as education and compliance with nutritional and pharmacologic therapy resulted in improvement.

We hypothesized that glucagon administration would stimulate hepatic glucose output, and thereby help maintain postprandial blood glucose concentrations. Thus, we compared the insulin and glycemic response to a mixed meal with and without concomitant glucagon infusion.

METHODS AND PROCEDURES

The Joslin Diabetes Center Institutional Review Board approved the study, and the patient provided written informed consent. Paired studies were performed 1 month apart. All medications were held for 12 h before the studies. Following overnight fast a liquid mixed meal tolerance test using Ensure (240 kcal, fat 6 g, carbohydrate 40 g, protein 9 g) was performed. During the first visit, blood samples were obtained before and 20, 30, 60, 90, and 120 min after the mixed meal. During the second visit, baseline blood samples were drawn (time −60 min), and then a primed continuous infusion of glucagon (300 μg/h) was initiated. At time 0, blood sampling was repeated and the mixed meal was administered. Glucagon was administered for 4 h (until time 180 min, 2 h beyond the patient’s usual window for development of hypoglycemia). Samples were obtained at 20, 30, 60, 90, 120, 180, and 210 min (see Figure 1a). Samples were assayed for glucose, insulin, C-peptide, and glucagon concentrations. Plasma glucose was measured by glucose oxidation. Radioimmunoassay was used to measure serum insulin, C-peptide (Diagnostic Systems Laboratories, Webster, TX) and glucagon (Millipore, Billerica, MA).

Figure 1.

(a) Glucose and (b) insulin concentrations and (c) insulin to glucose molar ratios during two mixed meal tolerance tests, one with concomitant glucagon infusion (Glucagon) the other without (Baseline) in a patient status-post gastric bypass with hyperinsulinemic hypoglycemia. MMTT, mixed meal tolerance test.

RESULTS

Fasting glucose concentrations were comparable on the two study days (86 and 83 mg/dl). In response to glucagon infusion (before consumption of the mixed meal), glucose rose, as expected, to 123 mg/dl. Following consumption of the mixed meal, glucose increased during both studies, but with glucagon administration, the glucose concentrations were higher at all time points between 20 and 90 min, and the peak glucose was higher (188 vs. 136 mg/dl). By 120 min, the glucose concentrations were equivalent (64 vs. 61 mg/dl) (Figure 1a).

Upon completion of the mixed meal tolerance test on the first day, the patient experienced no symptoms of hypoglycemia. In contrast, on the second study day, upon discontinuation of the glucagon infusion she complained of light-headedness, lethargy, nausea, flushing, and sweating—her typical hypoglycemic symptoms. Her glucose subsequently decreased to 34 mg/dl at 210 min (Figure 1a), and she required infusion of dextrose to raise her glucose (to 80 mg/dl), which led to resolution of symptoms.

In parallel with the rise in glucose, insulin concentrations rose during glucagon administration, even before consumption of the mixed meal: fasting insulin concentration was 6.4 μU/ml at time −60 min and increased to 69.9 μU/ml after glucagon infusion for 1 h (time 0). Furthermore, postmeal insulin concentrations were higher at all time points during the glucagon study compared to the baseline study (peak insulin concentration 964 vs. 351 μU/ml) (Figure 1b). C-peptide concentrations were also higher during the glucagon study (not shown). The molar ratios of insulin to glucose were markedly higher at times 30–120 min during glucagon administration (Figure 1c).

Fasting glucagon concentrations were comparable on the two study days (75.3 and 73.2 pg/ml). On the first study day, postprandial glucagon concentrations ranged from 83.8 to 97.8 pg/ml. During the glucagon study, the postprandial glucagon concentrations ranged from 2,956 to 3,431 pg/ml. These levels are consistent with published data on glucagon levels achieved in normal subjects during similar glucagon infusion protocols (2).

DISCUSSION

Hyperinsulinemic hypoglycemia is a rare complication of RYGB; its pathophysiology remains incompletely understood (1,3,4). These patients exhibit inappropriately high insulin and C-peptide concentrations during hypoglycemia (5,6). In addition, exaggerated insulin and incretin responses are observed during mixed meal tolerance test (5) and may contribute to hypoglycemia, potentially mediated in part by islet cell hyperplasia and/or altered function (1,4). Although most cases are mild and managed with dietary modification, symptoms can be profound and result in serious adverse consequences, and clinical management can be challenging. Treatment options include intensive dietary modification (6,7), α-glucosidase inhibitors, octreotide, and diazoxide (5,7). However, some patients remain refractory to treatment, and partial pancreatectomy has been undertaken, with varying success (4). Given the lack of efficacy of available interventions in some patients, and the goal of averting pancreatectomy, additional pharmacologic options are needed.

Glucagon plays a critical role in glucose homeostasis by increasing glycogenolysis and gluconeogenesis. Therapeutically, it is used in patients with diabetes to treat severe hypoglycemia due to excess exogenous insulin administration. Its use has also been described in neonatal hyperinsulinism (8) and tumor-induced hypoglycemia (9). We therefore tested glucagon as a potential therapy in a patient with severe post-RYGB hypoglycemia. Our hypothesis was that it would stimulate hepatic glucose output and thereby help maintain normoglycemia. The dose chosen corresponded to published recommendations for glucagon use in other hyperinsulinemic hypoglycemic syndromes (7).

Our case study revealed that glucagon infusion resulted in higher glucose concentrations, as anticipated. Insulin (and C-peptide) levels were also higher during glucagon administration, consistent with the higher glucose concentrations observed. However, the calculated insulin to glucose ratios suggest that during the glucagon study, insulin secretion was augmented beyond that expected for the observed increase in glucose. One possibility is that the relationship between glucose and insulin secretion is not linear in this dynamic context, and that the higher plasma glucose levels resulting from glucagon’s effects on the liver are solely responsible for potentiating insulin secretion. Alternatively, glucagon may exert a direct effect to potentiate glucose-stimulated insulin secretion from pancreatic β-cells. Indeed, in isolated human islet cells, glucose-stimulated insulin release is potentiated by glucagon, and this effect is abolished by a glucagon receptor antagonist (10); similarly, human physiology studies suggest that insulin concentrations rise in response to glucagon administration independently of glucose values (11). Moreover, glucagon values increase rapidly during mixed meal tolerance test in patients post-RYGB (5), suggesting that elevations in postprandial glucagon and other intestinal peptides may actually contribute to excessive insulin secretion in this setting.

Glucagon may also have contributed to postprandial hypoglycemia in our study subject because of its inhibitory effect on intestinal motility, which prolongs meal transit time (2). Such direct gastrointestinal effects may modulate nutrient absorption, and contribute to, rather than prevent, postprandial hypoglycemia.

It is noteworthy that our subject developed hypoglycemic symptoms during the glucagon study when the glucose concentration was greater than the lowest glucose during the baseline study. In patients with diabetes, glycemic thresholds for symptom and counter-regulatory responses to hypoglycemia are dynamic and vary within an individual; one contributing factor is the frequency of recent antecedent hypoglycemia. Our observations suggest that glycemic thresholds for hypoglycemic symptoms may similarly be variable in post-RYGB hyperinsulinemic hypoglycemia.

In conclusion, although our observations are limited to a single patient, these findings suggest that glucagon administration is not an effective treatment of post-RYGB hyperinsulinemic hypoglycemia. The reasons for lack of clinical utility may include a prolonged potentiation of glucose-stimulated insulin secretion, resulting in substantially higher circulating insulin concentrations and failure to prevent hypoglycemia, as well as direct gastrointestinal effects modulating nutrient absorption.