Richard Millstein, DO

Bariatric surgery, though beneficial, is associated with complications, one of which is post-gastric bypass hypoglycemia (PGBH).¹ The mean time from gastric bypass to documented hypoglycemia is about 28 months.²

PGBH is probably more common than initially thought. In older reports, the prevalence was only 0.1% to 0.36%.^1,3 In contrast, in a mail survey in 2015,⁴ one-third of bariatric surgery patients reported symptoms that raised the suspicion of hypoglycemia. Those with suspicious symptoms were more likely to have undergone Roux-en-Y surgery, to have had no preoperative diabetes, to have had a longer interval since surgery, and to be female. Restricting the suspicion of postprandial hypoglycemia to those who reported more serious symptoms, including needing third-party assistance, the prevalence was 11.6%.

Kefurt et al⁵ followed Roux-en-Y patients who wore a continuous glucose monitor for 86 months after surgery and found that 38% had hypoglycemia; however, symptoms of hypoglycemia were not discussed.

Thus, the exact prevalence is currently unknown. But as time goes by and more procedures are performed, the incidence will likely rise.

OBESITY IS ON THE RISE, AND SO IS WEIGHT-LOSS SURGERY

Obesity is rampant, and its prevalence continues to rise. In 2011–2012, more than two-thirds of adults in the United States were reported as obese.⁶ Complications of obesity such as cardiac disease, diabetes, and cancer lead to increased mortality risk.⁷ Obesity is difficult to reverse, as many people fail to lose weight with diet, exercise, and pharmacotherapy.

Given the difficulty of losing weight and the complications that arise from obesity, bariatric surgery has become increasingly popular. Not only do patients lose significantly more weight with bariatric surgery than with conventional measures, but surgery also reduces and often cures conditions associated with obesity.⁸

Nguyen et al⁹ reported that 671,959 patients underwent gastric bypass procedures in the United States from 2003 to 2008. In a registry maintained by the American Society for Metabolic and Bariatric Surgery¹⁰ from June 2007 to May 2009, the most common bariatric procedure in the United States was Roux-en-Y gastric bypass, followed by sleeve gastrectomy.

DIFFERENTIAL DIAGNOSIS AND DEFINITIONS

The differential diagnosis for hyperinsulinemic hypoglycemia after gastric bypass surgery includes exogenous and endogenous causes (Table 1). Exogenous causes include abuse of insulin secretagogues such as sulfonylureas or meglitinides and abuse of insulin, which may occur in patients with Munchausen syndrome, Munchausen syndrome by proxy, or malingering. Endogenous causes include insulinoma, early and late dumping syndromes, and PGBH.

When differentiating endogenous from exogenous hypoglycemia, insulin and C-peptide levels are useful (Table 2). The pancreas produces proinsulin, which is broken down into insulin and C-peptide. Since exogenous insulin does not have a C-peptide component, people abusing insulin have elevated insulin levels with a low C-peptide level.¹¹ Insulin secretagogues cause endogenous insulin secretion, resulting in elevated levels of both insulin and C-peptide. Thus, a screen for these medications is necessary to determine this as the cause.

Differentiating endogenous causes of hypoglycemia

Differentiating the endogenous causes (insulinoma, early or late dumping syndrome, and PGBH) can be challenging, as all 3 have similar biochemical profiles (Table 2).

Insulinoma is a tumor of pancreatic beta cells that produces excessive amounts of insulin. Unlike dumping syndrome, which only occurs postprandially, insulinoma primarily causes fasting hypoglycemia, although postprandial hypoglycemia can occur less commonly. Insulinoma after Roux-en-Y is rare. Only 7 cases have been reported.¹²

Dumping syndrome is classified as either early or late.

Early dumping syndrome usually occurs within 20 minutes of eating. The rapid transit of carbohydrates into the small intestine results in a fluid shift and a sympathetic response characterized by tachycardia, nausea, and diarrhea. Hypoglycemia is not present. Early dumping syndrome usually arises during the first few months after surgery.¹³

Late dumping syndrome usually occurs 1 to 4 hours after ingestion of a carbohydrate load, with symptoms of diaphoresis, dizziness, and fatigue caused by hypoglycemia from an excessive insulin release in response to the carbohydrates.¹³ It does not tend to cause neuroglycopenic symptoms.¹⁴ We define late dumping syndrome as postprandial hypoglycemic symptoms that occur after eating simple sugars and that resolve with dietary changes alone.

Differentiating late dumping syndrome from PGBH is difficult, as the line between the 2 processes is blurred.¹³

PGBH is defined as postprandial hypoglycemia (although it can be fasting in severe cases), often with neuroglycopenic symptoms, that occurs despite adherence to an acceptable bariatric diet (outlined in Table 3). We categorize PGBH as mild, moderate, or severe. Mild PGBH resolves with dietary changes with or without an alpha-glucosidase inhibitor. Moderate PGBH does not respond to an alpha-glucosidase inhibitor and dietary changes, and alternative or additional medication or medications are needed for resolution. Severe PGBH does not respond to dietary or medical interventions, and patients experience persistent episodes of neuroglycopenia.

THE EXACT MECHANISM IS UNCERTAIN

Patients with PGBH have a significant postprandial rise in glucose (often with levels > 200 mg/dL), leading to a robust insulin response and a subsequent drop in blood glucose.¹⁵

The exact mechanisms causing hypoglycemia are unknown, but excessive release of the incretin hormones glucagon-like peptide 1 (GLP-1) and gastric inhibitory polypeptide (GIP) are thought to contribute. GLP-1 is primarily secreted in the gut in response to nutrients, causing a glucose-dependent release of insulin and suppression of glucagon, as well as a delay in gastric emptying and motility. Salehi et al¹⁶ demonstrated excessive GLP-1 and insulin release after glucose administration in postbypass patients, with a more exaggerated response in those experiencing postprandial hypoglycemia.

Excessive incretin hormones may also contribute to pancreatic islet cell hyperplasia, leading to hyperinsulinism.¹⁷ Other proposed mechanisms of PGBH are the lack of a decrease in beta cell mass after gastric bypass, a postoperative increase in insulin sensitivity, a decrease in ghrelin (an insulin counterregulatory hormone), and an abnormal glucagon response.^13,17

Pathologic changes vary widely

PGBH is a challenging diagnosis to make pathologically. On review of pancreatic tissue from 36 patients undergoing partial pancreatectomy for PGBH, the pancreatic islet cells of the PGBH group were larger and more irregular compared with controls.^18,19 This histologic condition with islet-cell hypertrophy, hyperplasia, and other changes has been termed nesidioblastosis.^11,14,20 However, the pancreatic tissue appears grossly normal. The histopathologic findings can vary greatly in individual cases and in one-third of cases the pancreatic changes can be minimal, so that “normal” and PGBH cells can be nearly impossible to distinguish from each other.²¹

DIAGNOSIS AND TREATMENT

We recommend a stepwise approach to evaluating and treating PGBH (Figures 1 and 2).

Step 1: Evaluate blood glucose and Whipple triad

The first step is a thorough history, including food consumption and timing of hypoglycemic symptoms. Give the patient a glucometer to take home, with instructions to check blood glucose levels when hypoglycemic symptoms occur. The patient should keep a log documenting time tested, food consumed, symptoms, and blood glucose data.

Hypoglycemic symptoms are categorized as autonomic and neuroglycopenic. Autonomic symptoms include anxiety, palpitations, tremulousness, and diaphoresis. Neuroglycopenic symptoms include confusion, falls, seizures, and loss of consciousness.¹²

There are degrees of hypoglycemia and hypoglycemic symptoms. Clinical hypoglycemia—a blood glucose level low enough to cause signs or symptoms—can be confirmed by the Whipple triad:

• Measured low blood glucose
• Symptoms of low blood glucose
• Relief of symptoms when low blood glucose is corrected.

Hypoglycemic symptoms can occur when the blood glucose level falls to less than 55 mg/dL in healthy people, but this cutoff can shift lower in someone who has recurrent hypoglycemia.

When the Whipple triad is documented, rule out nonhyperinsulinemic causes of hypoglycemia such as hypothyroidism, adrenal insufficiency, underlying organ dysfunction (ie, liver disease), and medications that cause hypoglycemia.

Step 2: Modify the diet

If postprandial hypoglycemia is occurring, the next step is dietary modification. Two studies showed that a low-carbohydrate diet prevented hypoglycemia; however, these diets contained nearly no carbohydrates (with meals consisting of eggs, sausage, cheese, and black coffee or tea).^15,22

Instruct patients to never eat pure carbohydrates without fat or protein, as this can result in a more severe hypoglycemic response.²² In addition, foods with a high glycemic index (a measure of how a carbohydrate-containing food raises blood sugar) should be avoided, and a low glycemic index diet is recommended.²³ High glycemic index foods include white bread, bagels, pretzels, and pineapple. Low glycemic index foods include 100% stone-ground whole wheat or pumpernickel bread, lima beans, butter beans, peas, legumes, lentils, and nonstarchy vegetables.

Our bariatric surgeons provide all postbariatric surgery patients with the dietary guidelines shown in Table 3.²⁴ We also ask our patients with PGBH to limit carbohydrates to 15 to 30 g per meal and to limit added sugars to less than 4 g per meal, including regular and sugar alcohols (polyols). Snacks should contain only protein and fat. In severe cases, we further limit the diet to 15 g of carbohydrate per meal, with no added sugars.

The hypoglycemia occurring with PGBH is treated differently than the hypoglycemia that occurs in diabetic patients. Advise patients with PGBH to treat their hypoglycemic episodes with a simple sugar combined with a protein or fat (eg, a small handful of candy with a spoonful of peanut butter), as they will often have recurrent hypoglycemia if a simple sugar is used alone. If patients regain weight, ask them about frequent eating, which would be related to self-treatment of hypoglycemia.

Step 3: Start an alpha-glucosidase inhibitor

If postprandial hypoglycemia persists despite dietary modification, then start an alpha-glucosidase inhibitor such as acarbose. Acarbose inhibits carbohydrate absorption, resulting in a decreased insulin response; thus, it blunts the decline in postprandial blood glucose.

Unfortunately, gastrointestinal side effects such as flatulence, diarrhea, and abdominal pain occur in up to 20% of patients who take acarbose, often leading to its discontinuation.²⁵ To minimize gastrointestinal side effects, we usually start with 25 mg of acarbose with 1 meal daily for 1 week, then increase the dosage weekly to 25 mg with the other 2 meals. If tolerated, acarbose can be increased to 50 to 100 mg with 3 meals daily.

Step 4: Obtain a mixed meal tolerance test or a provocation meal test

If dietary changes and an alpha-glucosidase inhibitor do not prevent postprandial hypoglycemia from recurring, then confirmation of PGBH is needed, using a mixed meal tolerance test or a provocation meal test.

In a mixed meal tolerance test, the meal consists of 55% carbohydrate, 30% fat, and 15% protein. Patients with hyperinsulinemic hypoglycemia have a rapid rise in blood glucose (> 200 mg/dL) with a robust insulin response that is often followed by hypoglycemia after ingesting a meal containing carbohydrates in this test. Insulin levels that remain elevated after the plasma glucose level falls to less than 55 mg/dL indicate hyperinsulinism.¹¹

Nevertheless, a mixed meal tolerance test will not always induce hypoglycemia. In a study of 51 patients with PGBH, all wore a continuous glucose monitor, were instructed to follow their normal diet for 5 days, and then underwent a mixed meal tolerance test on day 6. The glucose monitor revealed hypoglycemia in 75% of patients, while the mixed meal tolerance test was positive in only 29%.⁵

Moreover, to date, there is no standardized mixed meal.^5,15 This might also explain the difference in prevalence of hypoglycemia detected by this test.

Based on these conflicting findings, we recommend a provocation meal test—ie, the patient is given foods that have induced hypoglycemia earlier.

Of note, the Endocrine Society guidelines on hypoglycemia state that an oral glucose tolerance test should never be used to document postprandial hypoglycemia.²⁶ Lev-Ran and Anderson²⁷ found that an oral glucose tolerance test could be positive in at least 10% of normal people.

Step 5: Consider other pharmacotherapy

For moderate to severe PGBH in which dietary modification and acarbose have failed, additional medical therapy is the next step. Medical therapies include calcium channel blockers, somatostatin analogues (eg, octreotide), and diazoxide.

Calcium channel blockers inhibit insulin release from beta cells²⁸ but at the risk of hypotension. Mordes and Alonso²⁹ treated 6 PGBH patients with nifedipine or verapamil with or without acarbose, and symptoms resolved in 5 of the 6 patients.

When we treat PGBH, we often add a calcium channel blocker as the next step in therapy if the patient has hypertension or if the blood pressure can tolerate this. If the patient’s blood pressure is low, then avoiding calcium channel blocker therapy may be necessary. The next step would be octreotide and then diazoxide.

Somatostatin analogues such as octreotide inhibit GLP-1 and insulin release.³⁰ The most common side effects of octreotide are diarrhea and abdominal pain. Bile stone formation can also occur, but this is not common.

Diazoxide opens adenosine triphosphate-sensitive potassium channels and reduces the opening of calcium channels, inhibiting insulin release and raising blood glucose. In a study of 6 Japanese patients with inoperable insulinoma, diazoxide was used to treat hypoglycemia.³¹ Unfortunately, the doses required to control the low blood sugars also led to adverse reactions, most of which involved edema secondary to volume overload and other heart failure symptoms. Diazoxide also commonly causes hypotension and hirsutism.

Step 6: 72-hour fast

A 72-hour fast is recommended in severe cases of PGBH in patients for whom dietary modification and the additional pharmacotherapy outlined in step 5 have failed. A 72-hour fast is always indicated in evaluating confirmed fasting hypoglycemia. People with insulinoma usually have fasting hypoglycemia, while patients with dumping syndrome do not. Patients with PGBH usually do not have fasting hypoglycemia, but they can in severe cases.¹¹

For safety, this test should be done in the hospital. Baseline plasma levels of insulin, C-peptide, proinsulin, beta-hydroxybutyrate, and glucose should be obtained. The patient then fasts, consuming only noncaloric and noncaffeinated beverages for 72 hours. During this time, capillary glucose checks are performed every 6 hours. If the capillary glucose level falls below 55 mg/dL,^11,26 then the baseline tests are redrawn along with a sulfonylurea screen. To reduce costs and unnecessary testing, the tests are not sent for laboratory processing unless the plasma glucose is less than 55 mg/dL.

When the plasma glucose is less than 55 mg/dL, insulin production should cease. Elevated insulin levels and insulin byproducts raise concern for hyperinsulinism. These values confirm hyperinsulinemic hypoglycemia²⁶:

• Glucose < 55 mg/dL
• Insulin ≥ 3 µU/mL
• C-peptide ≥ 0.2 nmol/L
• Proinsulin ≥ 5.0 pmol/L.

After hypoglycemia is confirmed, 1 mg of glucagon is given intravenously, and plasma glucose levels are obtained at 10, 20, and 30 minutes.^11,26 A rise in plasma glucose of at least 25 mg/dL after intravenous glucagon injection indicates hypoglycemia due to hyperinsulinemia. Two-thirds of patients with insulinoma experience hypoglycemia within the first 24 hours, and nearly all experience hypoglycemia within 48 hours.²⁶

Step 7: Obtain pancreatic imaging

If fasting hypoglycemia is present and hyperinsulinemic hypoglycemia is confirmed during a 72-hour fast, then pancreatic imaging should be obtained to evaluate for an insulinoma. We also recommend pancreatic imaging to rule out insulinoma when severe PGBH has not responded to dietary modification or pharmacotherapy.

Imaging is not recommended in PGBH that has been successfully treated with dietary modification with or without pharmacotherapy.

Endoscopic ultrasonography alone has 80% to 92% sensitivity for localizing a pancreatic mass as small as 5 mm. However, when coupled with computed tomography or magnetic resonance imaging, the sensitivity increases to nearly 100%.¹²

Step 8: Selective arterial calcium stimulation test

If a patient is found to have hyperinsulinemic hypoglycemia during a 72-hour fast but pancreatic imaging is negative, then selective arterial calcium stimulation testing (SACST) and hepatic vein sampling should be performed. Also, for severe PGBH, in which hypoglycemia has persisted despite dietary modification and pharmacotherapy, SACST can be performed to evaluate for possible localization of hyperinsulinism in patients considering surgery. For mild and moderate cases of PGBH, in which the hypoglycemia has been successfully treated with dietary changes with or without pharmacotherapy, SACST is not necessary.

This test can localize the area of excess insulin production in the pancreas in patients with an insulinoma. Patients with severe PGBH usually have diffuse hyperinsulinism without localization on SACST.^32,33

When SACST is performed, a sampling catheter is placed in the femoral vein. Calcium gluconate is injected into the major arteries of the pancreas (superior mesenteric, gastroduodenal, and splenic arteries). Calcium stimulates release of insulin from an insulinoma or hyperplastic beta cells. Resultant insulin levels are measured in the hepatic vein. If there is a greater than twofold increase in insulin release from 2 segments, then the test is considered positive.

Thompson et al³⁴ documented that insulin release from insulinoma is almost 4 times higher than in diffuse nesidioblastosis. SACST has a sensitivity of 96% for detecting insulinomas.³⁵

Step 9: Other alternatives and surgery

In patients with severe PGBH for whom dietary modification and all pharmacotherapy have failed and who continue to have debilitating neuroglycopenia, there are options before proceeding with surgery, the last resort in this condition.

Continuous glucose monitoring is helpful in many patients with severe PGBH. Many of them have hypoglycemia unawareness, and the monitor alerts them when their blood sugar is low. In addition, the monitor indicates when the blood sugar is dropping, so that intervention can occur before hypoglycemia occurs.

Unfortunately, insurance coverage for continuous monitors in this patient population is limited. We argue that insurance should cover the cost for these severe cases.

Pasireotide, a somatostatin analogue that is longer-acting than octreotide, is approved for use in Cushing disease and acromegaly and actually causes hyperglycemia. In a case report of a 50-year-old woman, pasireotide resulted in less hypoglycemia and higher glucagon levels then octreotide.³⁶ Pasireotide is available from Novartis for compassionate use in patients with severe PGBH.

Glucocorticoids are another off-label option. However, in excess, they can lead to iatrogenic Cushing syndrome, which has its own complications. Prednisone and diazoxide have been used together to help prevent hypoglycemia in a patients with inoperable insulinoma.³¹

Tube feeding. Some researchers have studied altering nutrition access through surgical means. McLaughlin et al³⁷ discussed a case of gastric tube insertion into the remnant stomach of a patient with PGBH, with resolution of hypoglycemic symptoms and hypoglycemia; however, this does not always provide complete resolution of symptoms.^37,38 If gastric bypass reversal is being considered, a trial of solely remnant stomach tube feeds (with no oral intake) should be pursued first. If this ameliorates the hypoglycemia, then gastric bypass reversal may be of benefit.

Surgery is the last resort if all of the above treatments have failed and severe debilitating neuroglycopenia persists. However, surgery poses risks, and the success rate in correcting hypoglycemia is not ideal. Surgical options include Roux-en-Y reversal, gastric pouch resection, and pancreatic resection.

In a review by Mala,² 75 patients with documented PGBH underwent surgical therapy. Hypoglycemic symptoms resolved in 34 of 51 pancreatic resections, 13 of 17 Roux-en-Y reversals, and 9 of 11 gastric pouch resections. However, the follow-up period was short.

As noted above, we recommend calcium stimulation testing only for severe cases of PGBH when surgery is being considered to evaluate for possible localization of hyperinsulinism for which partial pancreatectomy would be of benefit. Since there is no localization in many PGBH cases and the success rates are slightly higher in gastric bypass reversal, bypass reversal is usually preferred over partial or complete pancreatectomy.^2,32,33

POTENTIAL FUTURE THERAPIES

Given the elevated GLP-1 levels and robust insulin response to glucose observed in PGBH, blocking GLP-1 may provide clinical benefit. Salehi et al¹⁶ found that a GLP-1 antagonist prevented surges in GLP-1 and reduced hypoglycemic episodes in patients with PGBH. Unfortunately, the medication they used was given as a continuous infusion and is not currently available.

Conversely, a GLP-1 agonist showed benefit in a series of 5 cases of PGBH.³⁹ In addition, an insulin receptor antibody is undergoing phase 2 trials and has been shown to reverse insulin-induced hypoglycemia in rodents and humans; it may be a novel therapy in the future for hyperinsulinemic hypoglycemia.⁴⁰

MORE STUDY NEEDED

As the prevalence of obesity continues to rise and more people opt for bariatric surgery for weight loss, we will likely continue to see an increase in PGBH, since the onset of PGBH can be delayed for many years after surgery.²⁸

Unfortunately, the disease process involved in PGBH is not well understood. For example, we do not know why GLP-1 elevations or a robust insulin response causing hypoglycemia occurs in some but not all gastric bypass patients. Study is needed to elucidate the pathophysiology to further understand why most patients have no hypoglycemia after gastric bypass, some have mild to moderate PGBH, and a small percentage have severe PGBH with debilitating neuroglycopenia unresponsive to dietary changes and medications.     

Bariatric surgery, though beneficial, is associated with complications, one of which is post-gastric bypass hypoglycemia (PGBH).¹ The mean time from gastric bypass to documented hypoglycemia is about 28 months.²

PGBH is probably more common than initially thought. In older reports, the prevalence was only 0.1% to 0.36%.^1,3 In contrast, in a mail survey in 2015,⁴ one-third of bariatric surgery patients reported symptoms that raised the suspicion of hypoglycemia. Those with suspicious symptoms were more likely to have undergone Roux-en-Y surgery, to have had no preoperative diabetes, to have had a longer interval since surgery, and to be female. Restricting the suspicion of postprandial hypoglycemia to those who reported more serious symptoms, including needing third-party assistance, the prevalence was 11.6%.

Kefurt et al⁵ followed Roux-en-Y patients who wore a continuous glucose monitor for 86 months after surgery and found that 38% had hypoglycemia; however, symptoms of hypoglycemia were not discussed.

Thus, the exact prevalence is currently unknown. But as time goes by and more procedures are performed, the incidence will likely rise.

OBESITY IS ON THE RISE, AND SO IS WEIGHT-LOSS SURGERY

Obesity is rampant, and its prevalence continues to rise. In 2011–2012, more than two-thirds of adults in the United States were reported as obese.⁶ Complications of obesity such as cardiac disease, diabetes, and cancer lead to increased mortality risk.⁷ Obesity is difficult to reverse, as many people fail to lose weight with diet, exercise, and pharmacotherapy.

Given the difficulty of losing weight and the complications that arise from obesity, bariatric surgery has become increasingly popular. Not only do patients lose significantly more weight with bariatric surgery than with conventional measures, but surgery also reduces and often cures conditions associated with obesity.⁸

Nguyen et al⁹ reported that 671,959 patients underwent gastric bypass procedures in the United States from 2003 to 2008. In a registry maintained by the American Society for Metabolic and Bariatric Surgery¹⁰ from June 2007 to May 2009, the most common bariatric procedure in the United States was Roux-en-Y gastric bypass, followed by sleeve gastrectomy.

DIFFERENTIAL DIAGNOSIS AND DEFINITIONS

The differential diagnosis for hyperinsulinemic hypoglycemia after gastric bypass surgery includes exogenous and endogenous causes (Table 1). Exogenous causes include abuse of insulin secretagogues such as sulfonylureas or meglitinides and abuse of insulin, which may occur in patients with Munchausen syndrome, Munchausen syndrome by proxy, or malingering. Endogenous causes include insulinoma, early and late dumping syndromes, and PGBH.

When differentiating endogenous from exogenous hypoglycemia, insulin and C-peptide levels are useful (Table 2). The pancreas produces proinsulin, which is broken down into insulin and C-peptide. Since exogenous insulin does not have a C-peptide component, people abusing insulin have elevated insulin levels with a low C-peptide level.¹¹ Insulin secretagogues cause endogenous insulin secretion, resulting in elevated levels of both insulin and C-peptide. Thus, a screen for these medications is necessary to determine this as the cause.

Differentiating endogenous causes of hypoglycemia

Differentiating the endogenous causes (insulinoma, early or late dumping syndrome, and PGBH) can be challenging, as all 3 have similar biochemical profiles (Table 2).

Insulinoma is a tumor of pancreatic beta cells that produces excessive amounts of insulin. Unlike dumping syndrome, which only occurs postprandially, insulinoma primarily causes fasting hypoglycemia, although postprandial hypoglycemia can occur less commonly. Insulinoma after Roux-en-Y is rare. Only 7 cases have been reported.¹²

Dumping syndrome is classified as either early or late.

Early dumping syndrome usually occurs within 20 minutes of eating. The rapid transit of carbohydrates into the small intestine results in a fluid shift and a sympathetic response characterized by tachycardia, nausea, and diarrhea. Hypoglycemia is not present. Early dumping syndrome usually arises during the first few months after surgery.¹³

Late dumping syndrome usually occurs 1 to 4 hours after ingestion of a carbohydrate load, with symptoms of diaphoresis, dizziness, and fatigue caused by hypoglycemia from an excessive insulin release in response to the carbohydrates.¹³ It does not tend to cause neuroglycopenic symptoms.¹⁴ We define late dumping syndrome as postprandial hypoglycemic symptoms that occur after eating simple sugars and that resolve with dietary changes alone.

Differentiating late dumping syndrome from PGBH is difficult, as the line between the 2 processes is blurred.¹³

PGBH is defined as postprandial hypoglycemia (although it can be fasting in severe cases), often with neuroglycopenic symptoms, that occurs despite adherence to an acceptable bariatric diet (outlined in Table 3). We categorize PGBH as mild, moderate, or severe. Mild PGBH resolves with dietary changes with or without an alpha-glucosidase inhibitor. Moderate PGBH does not respond to an alpha-glucosidase inhibitor and dietary changes, and alternative or additional medication or medications are needed for resolution. Severe PGBH does not respond to dietary or medical interventions, and patients experience persistent episodes of neuroglycopenia.

THE EXACT MECHANISM IS UNCERTAIN

Patients with PGBH have a significant postprandial rise in glucose (often with levels > 200 mg/dL), leading to a robust insulin response and a subsequent drop in blood glucose.¹⁵

The exact mechanisms causing hypoglycemia are unknown, but excessive release of the incretin hormones glucagon-like peptide 1 (GLP-1) and gastric inhibitory polypeptide (GIP) are thought to contribute. GLP-1 is primarily secreted in the gut in response to nutrients, causing a glucose-dependent release of insulin and suppression of glucagon, as well as a delay in gastric emptying and motility. Salehi et al¹⁶ demonstrated excessive GLP-1 and insulin release after glucose administration in postbypass patients, with a more exaggerated response in those experiencing postprandial hypoglycemia.

Excessive incretin hormones may also contribute to pancreatic islet cell hyperplasia, leading to hyperinsulinism.¹⁷ Other proposed mechanisms of PGBH are the lack of a decrease in beta cell mass after gastric bypass, a postoperative increase in insulin sensitivity, a decrease in ghrelin (an insulin counterregulatory hormone), and an abnormal glucagon response.^13,17

Pathologic changes vary widely

PGBH is a challenging diagnosis to make pathologically. On review of pancreatic tissue from 36 patients undergoing partial pancreatectomy for PGBH, the pancreatic islet cells of the PGBH group were larger and more irregular compared with controls.^18,19 This histologic condition with islet-cell hypertrophy, hyperplasia, and other changes has been termed nesidioblastosis.^11,14,20 However, the pancreatic tissue appears grossly normal. The histopathologic findings can vary greatly in individual cases and in one-third of cases the pancreatic changes can be minimal, so that “normal” and PGBH cells can be nearly impossible to distinguish from each other.²¹

DIAGNOSIS AND TREATMENT

We recommend a stepwise approach to evaluating and treating PGBH (Figures 1 and 2).

Step 1: Evaluate blood glucose and Whipple triad

The first step is a thorough history, including food consumption and timing of hypoglycemic symptoms. Give the patient a glucometer to take home, with instructions to check blood glucose levels when hypoglycemic symptoms occur. The patient should keep a log documenting time tested, food consumed, symptoms, and blood glucose data.

Hypoglycemic symptoms are categorized as autonomic and neuroglycopenic. Autonomic symptoms include anxiety, palpitations, tremulousness, and diaphoresis. Neuroglycopenic symptoms include confusion, falls, seizures, and loss of consciousness.¹²

There are degrees of hypoglycemia and hypoglycemic symptoms. Clinical hypoglycemia—a blood glucose level low enough to cause signs or symptoms—can be confirmed by the Whipple triad:

• Measured low blood glucose
• Symptoms of low blood glucose
• Relief of symptoms when low blood glucose is corrected.

Hypoglycemic symptoms can occur when the blood glucose level falls to less than 55 mg/dL in healthy people, but this cutoff can shift lower in someone who has recurrent hypoglycemia.

When the Whipple triad is documented, rule out nonhyperinsulinemic causes of hypoglycemia such as hypothyroidism, adrenal insufficiency, underlying organ dysfunction (ie, liver disease), and medications that cause hypoglycemia.

Step 2: Modify the diet

If postprandial hypoglycemia is occurring, the next step is dietary modification. Two studies showed that a low-carbohydrate diet prevented hypoglycemia; however, these diets contained nearly no carbohydrates (with meals consisting of eggs, sausage, cheese, and black coffee or tea).^15,22

Instruct patients to never eat pure carbohydrates without fat or protein, as this can result in a more severe hypoglycemic response.²² In addition, foods with a high glycemic index (a measure of how a carbohydrate-containing food raises blood sugar) should be avoided, and a low glycemic index diet is recommended.²³ High glycemic index foods include white bread, bagels, pretzels, and pineapple. Low glycemic index foods include 100% stone-ground whole wheat or pumpernickel bread, lima beans, butter beans, peas, legumes, lentils, and nonstarchy vegetables.

Our bariatric surgeons provide all postbariatric surgery patients with the dietary guidelines shown in Table 3.²⁴ We also ask our patients with PGBH to limit carbohydrates to 15 to 30 g per meal and to limit added sugars to less than 4 g per meal, including regular and sugar alcohols (polyols). Snacks should contain only protein and fat. In severe cases, we further limit the diet to 15 g of carbohydrate per meal, with no added sugars.

The hypoglycemia occurring with PGBH is treated differently than the hypoglycemia that occurs in diabetic patients. Advise patients with PGBH to treat their hypoglycemic episodes with a simple sugar combined with a protein or fat (eg, a small handful of candy with a spoonful of peanut butter), as they will often have recurrent hypoglycemia if a simple sugar is used alone. If patients regain weight, ask them about frequent eating, which would be related to self-treatment of hypoglycemia.

Step 3: Start an alpha-glucosidase inhibitor

If postprandial hypoglycemia persists despite dietary modification, then start an alpha-glucosidase inhibitor such as acarbose. Acarbose inhibits carbohydrate absorption, resulting in a decreased insulin response; thus, it blunts the decline in postprandial blood glucose.

Unfortunately, gastrointestinal side effects such as flatulence, diarrhea, and abdominal pain occur in up to 20% of patients who take acarbose, often leading to its discontinuation.²⁵ To minimize gastrointestinal side effects, we usually start with 25 mg of acarbose with 1 meal daily for 1 week, then increase the dosage weekly to 25 mg with the other 2 meals. If tolerated, acarbose can be increased to 50 to 100 mg with 3 meals daily.

Step 4: Obtain a mixed meal tolerance test or a provocation meal test

If dietary changes and an alpha-glucosidase inhibitor do not prevent postprandial hypoglycemia from recurring, then confirmation of PGBH is needed, using a mixed meal tolerance test or a provocation meal test.

In a mixed meal tolerance test, the meal consists of 55% carbohydrate, 30% fat, and 15% protein. Patients with hyperinsulinemic hypoglycemia have a rapid rise in blood glucose (> 200 mg/dL) with a robust insulin response that is often followed by hypoglycemia after ingesting a meal containing carbohydrates in this test. Insulin levels that remain elevated after the plasma glucose level falls to less than 55 mg/dL indicate hyperinsulinism.¹¹

Nevertheless, a mixed meal tolerance test will not always induce hypoglycemia. In a study of 51 patients with PGBH, all wore a continuous glucose monitor, were instructed to follow their normal diet for 5 days, and then underwent a mixed meal tolerance test on day 6. The glucose monitor revealed hypoglycemia in 75% of patients, while the mixed meal tolerance test was positive in only 29%.⁵

Moreover, to date, there is no standardized mixed meal.^5,15 This might also explain the difference in prevalence of hypoglycemia detected by this test.

Based on these conflicting findings, we recommend a provocation meal test—ie, the patient is given foods that have induced hypoglycemia earlier.

Of note, the Endocrine Society guidelines on hypoglycemia state that an oral glucose tolerance test should never be used to document postprandial hypoglycemia.²⁶ Lev-Ran and Anderson²⁷ found that an oral glucose tolerance test could be positive in at least 10% of normal people.

Step 5: Consider other pharmacotherapy

For moderate to severe PGBH in which dietary modification and acarbose have failed, additional medical therapy is the next step. Medical therapies include calcium channel blockers, somatostatin analogues (eg, octreotide), and diazoxide.

Calcium channel blockers inhibit insulin release from beta cells²⁸ but at the risk of hypotension. Mordes and Alonso²⁹ treated 6 PGBH patients with nifedipine or verapamil with or without acarbose, and symptoms resolved in 5 of the 6 patients.

When we treat PGBH, we often add a calcium channel blocker as the next step in therapy if the patient has hypertension or if the blood pressure can tolerate this. If the patient’s blood pressure is low, then avoiding calcium channel blocker therapy may be necessary. The next step would be octreotide and then diazoxide.

Somatostatin analogues such as octreotide inhibit GLP-1 and insulin release.³⁰ The most common side effects of octreotide are diarrhea and abdominal pain. Bile stone formation can also occur, but this is not common.

Diazoxide opens adenosine triphosphate-sensitive potassium channels and reduces the opening of calcium channels, inhibiting insulin release and raising blood glucose. In a study of 6 Japanese patients with inoperable insulinoma, diazoxide was used to treat hypoglycemia.³¹ Unfortunately, the doses required to control the low blood sugars also led to adverse reactions, most of which involved edema secondary to volume overload and other heart failure symptoms. Diazoxide also commonly causes hypotension and hirsutism.

Step 6: 72-hour fast

A 72-hour fast is recommended in severe cases of PGBH in patients for whom dietary modification and the additional pharmacotherapy outlined in step 5 have failed. A 72-hour fast is always indicated in evaluating confirmed fasting hypoglycemia. People with insulinoma usually have fasting hypoglycemia, while patients with dumping syndrome do not. Patients with PGBH usually do not have fasting hypoglycemia, but they can in severe cases.¹¹

For safety, this test should be done in the hospital. Baseline plasma levels of insulin, C-peptide, proinsulin, beta-hydroxybutyrate, and glucose should be obtained. The patient then fasts, consuming only noncaloric and noncaffeinated beverages for 72 hours. During this time, capillary glucose checks are performed every 6 hours. If the capillary glucose level falls below 55 mg/dL,^11,26 then the baseline tests are redrawn along with a sulfonylurea screen. To reduce costs and unnecessary testing, the tests are not sent for laboratory processing unless the plasma glucose is less than 55 mg/dL.

When the plasma glucose is less than 55 mg/dL, insulin production should cease. Elevated insulin levels and insulin byproducts raise concern for hyperinsulinism. These values confirm hyperinsulinemic hypoglycemia²⁶:

• Glucose < 55 mg/dL
• Insulin ≥ 3 µU/mL
• C-peptide ≥ 0.2 nmol/L
• Proinsulin ≥ 5.0 pmol/L.

After hypoglycemia is confirmed, 1 mg of glucagon is given intravenously, and plasma glucose levels are obtained at 10, 20, and 30 minutes.^11,26 A rise in plasma glucose of at least 25 mg/dL after intravenous glucagon injection indicates hypoglycemia due to hyperinsulinemia. Two-thirds of patients with insulinoma experience hypoglycemia within the first 24 hours, and nearly all experience hypoglycemia within 48 hours.²⁶

Step 7: Obtain pancreatic imaging

If fasting hypoglycemia is present and hyperinsulinemic hypoglycemia is confirmed during a 72-hour fast, then pancreatic imaging should be obtained to evaluate for an insulinoma. We also recommend pancreatic imaging to rule out insulinoma when severe PGBH has not responded to dietary modification or pharmacotherapy.

Imaging is not recommended in PGBH that has been successfully treated with dietary modification with or without pharmacotherapy.

Endoscopic ultrasonography alone has 80% to 92% sensitivity for localizing a pancreatic mass as small as 5 mm. However, when coupled with computed tomography or magnetic resonance imaging, the sensitivity increases to nearly 100%.¹²

Step 8: Selective arterial calcium stimulation test

If a patient is found to have hyperinsulinemic hypoglycemia during a 72-hour fast but pancreatic imaging is negative, then selective arterial calcium stimulation testing (SACST) and hepatic vein sampling should be performed. Also, for severe PGBH, in which hypoglycemia has persisted despite dietary modification and pharmacotherapy, SACST can be performed to evaluate for possible localization of hyperinsulinism in patients considering surgery. For mild and moderate cases of PGBH, in which the hypoglycemia has been successfully treated with dietary changes with or without pharmacotherapy, SACST is not necessary.

This test can localize the area of excess insulin production in the pancreas in patients with an insulinoma. Patients with severe PGBH usually have diffuse hyperinsulinism without localization on SACST.^32,33

When SACST is performed, a sampling catheter is placed in the femoral vein. Calcium gluconate is injected into the major arteries of the pancreas (superior mesenteric, gastroduodenal, and splenic arteries). Calcium stimulates release of insulin from an insulinoma or hyperplastic beta cells. Resultant insulin levels are measured in the hepatic vein. If there is a greater than twofold increase in insulin release from 2 segments, then the test is considered positive.

Thompson et al³⁴ documented that insulin release from insulinoma is almost 4 times higher than in diffuse nesidioblastosis. SACST has a sensitivity of 96% for detecting insulinomas.³⁵

Step 9: Other alternatives and surgery

In patients with severe PGBH for whom dietary modification and all pharmacotherapy have failed and who continue to have debilitating neuroglycopenia, there are options before proceeding with surgery, the last resort in this condition.

Continuous glucose monitoring is helpful in many patients with severe PGBH. Many of them have hypoglycemia unawareness, and the monitor alerts them when their blood sugar is low. In addition, the monitor indicates when the blood sugar is dropping, so that intervention can occur before hypoglycemia occurs.

Unfortunately, insurance coverage for continuous monitors in this patient population is limited. We argue that insurance should cover the cost for these severe cases.

Pasireotide, a somatostatin analogue that is longer-acting than octreotide, is approved for use in Cushing disease and acromegaly and actually causes hyperglycemia. In a case report of a 50-year-old woman, pasireotide resulted in less hypoglycemia and higher glucagon levels then octreotide.³⁶ Pasireotide is available from Novartis for compassionate use in patients with severe PGBH.

Glucocorticoids are another off-label option. However, in excess, they can lead to iatrogenic Cushing syndrome, which has its own complications. Prednisone and diazoxide have been used together to help prevent hypoglycemia in a patients with inoperable insulinoma.³¹

Tube feeding. Some researchers have studied altering nutrition access through surgical means. McLaughlin et al³⁷ discussed a case of gastric tube insertion into the remnant stomach of a patient with PGBH, with resolution of hypoglycemic symptoms and hypoglycemia; however, this does not always provide complete resolution of symptoms.^37,38 If gastric bypass reversal is being considered, a trial of solely remnant stomach tube feeds (with no oral intake) should be pursued first. If this ameliorates the hypoglycemia, then gastric bypass reversal may be of benefit.

Surgery is the last resort if all of the above treatments have failed and severe debilitating neuroglycopenia persists. However, surgery poses risks, and the success rate in correcting hypoglycemia is not ideal. Surgical options include Roux-en-Y reversal, gastric pouch resection, and pancreatic resection.

In a review by Mala,² 75 patients with documented PGBH underwent surgical therapy. Hypoglycemic symptoms resolved in 34 of 51 pancreatic resections, 13 of 17 Roux-en-Y reversals, and 9 of 11 gastric pouch resections. However, the follow-up period was short.

As noted above, we recommend calcium stimulation testing only for severe cases of PGBH when surgery is being considered to evaluate for possible localization of hyperinsulinism for which partial pancreatectomy would be of benefit. Since there is no localization in many PGBH cases and the success rates are slightly higher in gastric bypass reversal, bypass reversal is usually preferred over partial or complete pancreatectomy.^2,32,33

POTENTIAL FUTURE THERAPIES

Given the elevated GLP-1 levels and robust insulin response to glucose observed in PGBH, blocking GLP-1 may provide clinical benefit. Salehi et al¹⁶ found that a GLP-1 antagonist prevented surges in GLP-1 and reduced hypoglycemic episodes in patients with PGBH. Unfortunately, the medication they used was given as a continuous infusion and is not currently available.

Conversely, a GLP-1 agonist showed benefit in a series of 5 cases of PGBH.³⁹ In addition, an insulin receptor antibody is undergoing phase 2 trials and has been shown to reverse insulin-induced hypoglycemia in rodents and humans; it may be a novel therapy in the future for hyperinsulinemic hypoglycemia.⁴⁰

MORE STUDY NEEDED

As the prevalence of obesity continues to rise and more people opt for bariatric surgery for weight loss, we will likely continue to see an increase in PGBH, since the onset of PGBH can be delayed for many years after surgery.²⁸

Unfortunately, the disease process involved in PGBH is not well understood. For example, we do not know why GLP-1 elevations or a robust insulin response causing hypoglycemia occurs in some but not all gastric bypass patients. Study is needed to elucidate the pathophysiology to further understand why most patients have no hypoglycemia after gastric bypass, some have mild to moderate PGBH, and a small percentage have severe PGBH with debilitating neuroglycopenia unresponsive to dietary changes and medications.     

Bariatric surgery, though beneficial, is associated with complications, one of which is post-gastric bypass hypoglycemia (PGBH).¹ The mean time from gastric bypass to documented hypoglycemia is about 28 months.²

PGBH is probably more common than initially thought. In older reports, the prevalence was only 0.1% to 0.36%.^1,3 In contrast, in a mail survey in 2015,⁴ one-third of bariatric surgery patients reported symptoms that raised the suspicion of hypoglycemia. Those with suspicious symptoms were more likely to have undergone Roux-en-Y surgery, to have had no preoperative diabetes, to have had a longer interval since surgery, and to be female. Restricting the suspicion of postprandial hypoglycemia to those who reported more serious symptoms, including needing third-party assistance, the prevalence was 11.6%.

Kefurt et al⁵ followed Roux-en-Y patients who wore a continuous glucose monitor for 86 months after surgery and found that 38% had hypoglycemia; however, symptoms of hypoglycemia were not discussed.

Thus, the exact prevalence is currently unknown. But as time goes by and more procedures are performed, the incidence will likely rise.

OBESITY IS ON THE RISE, AND SO IS WEIGHT-LOSS SURGERY

Obesity is rampant, and its prevalence continues to rise. In 2011–2012, more than two-thirds of adults in the United States were reported as obese.⁶ Complications of obesity such as cardiac disease, diabetes, and cancer lead to increased mortality risk.⁷ Obesity is difficult to reverse, as many people fail to lose weight with diet, exercise, and pharmacotherapy.

Given the difficulty of losing weight and the complications that arise from obesity, bariatric surgery has become increasingly popular. Not only do patients lose significantly more weight with bariatric surgery than with conventional measures, but surgery also reduces and often cures conditions associated with obesity.⁸

Nguyen et al⁹ reported that 671,959 patients underwent gastric bypass procedures in the United States from 2003 to 2008. In a registry maintained by the American Society for Metabolic and Bariatric Surgery¹⁰ from June 2007 to May 2009, the most common bariatric procedure in the United States was Roux-en-Y gastric bypass, followed by sleeve gastrectomy.

DIFFERENTIAL DIAGNOSIS AND DEFINITIONS

The differential diagnosis for hyperinsulinemic hypoglycemia after gastric bypass surgery includes exogenous and endogenous causes (Table 1). Exogenous causes include abuse of insulin secretagogues such as sulfonylureas or meglitinides and abuse of insulin, which may occur in patients with Munchausen syndrome, Munchausen syndrome by proxy, or malingering. Endogenous causes include insulinoma, early and late dumping syndromes, and PGBH.

When differentiating endogenous from exogenous hypoglycemia, insulin and C-peptide levels are useful (Table 2). The pancreas produces proinsulin, which is broken down into insulin and C-peptide. Since exogenous insulin does not have a C-peptide component, people abusing insulin have elevated insulin levels with a low C-peptide level.¹¹ Insulin secretagogues cause endogenous insulin secretion, resulting in elevated levels of both insulin and C-peptide. Thus, a screen for these medications is necessary to determine this as the cause.

Differentiating endogenous causes of hypoglycemia

Differentiating the endogenous causes (insulinoma, early or late dumping syndrome, and PGBH) can be challenging, as all 3 have similar biochemical profiles (Table 2).

Insulinoma is a tumor of pancreatic beta cells that produces excessive amounts of insulin. Unlike dumping syndrome, which only occurs postprandially, insulinoma primarily causes fasting hypoglycemia, although postprandial hypoglycemia can occur less commonly. Insulinoma after Roux-en-Y is rare. Only 7 cases have been reported.¹²

Dumping syndrome is classified as either early or late.

Early dumping syndrome usually occurs within 20 minutes of eating. The rapid transit of carbohydrates into the small intestine results in a fluid shift and a sympathetic response characterized by tachycardia, nausea, and diarrhea. Hypoglycemia is not present. Early dumping syndrome usually arises during the first few months after surgery.¹³

Late dumping syndrome usually occurs 1 to 4 hours after ingestion of a carbohydrate load, with symptoms of diaphoresis, dizziness, and fatigue caused by hypoglycemia from an excessive insulin release in response to the carbohydrates.¹³ It does not tend to cause neuroglycopenic symptoms.¹⁴ We define late dumping syndrome as postprandial hypoglycemic symptoms that occur after eating simple sugars and that resolve with dietary changes alone.

Differentiating late dumping syndrome from PGBH is difficult, as the line between the 2 processes is blurred.¹³

PGBH is defined as postprandial hypoglycemia (although it can be fasting in severe cases), often with neuroglycopenic symptoms, that occurs despite adherence to an acceptable bariatric diet (outlined in Table 3). We categorize PGBH as mild, moderate, or severe. Mild PGBH resolves with dietary changes with or without an alpha-glucosidase inhibitor. Moderate PGBH does not respond to an alpha-glucosidase inhibitor and dietary changes, and alternative or additional medication or medications are needed for resolution. Severe PGBH does not respond to dietary or medical interventions, and patients experience persistent episodes of neuroglycopenia.

THE EXACT MECHANISM IS UNCERTAIN

Patients with PGBH have a significant postprandial rise in glucose (often with levels > 200 mg/dL), leading to a robust insulin response and a subsequent drop in blood glucose.¹⁵

The exact mechanisms causing hypoglycemia are unknown, but excessive release of the incretin hormones glucagon-like peptide 1 (GLP-1) and gastric inhibitory polypeptide (GIP) are thought to contribute. GLP-1 is primarily secreted in the gut in response to nutrients, causing a glucose-dependent release of insulin and suppression of glucagon, as well as a delay in gastric emptying and motility. Salehi et al¹⁶ demonstrated excessive GLP-1 and insulin release after glucose administration in postbypass patients, with a more exaggerated response in those experiencing postprandial hypoglycemia.

Excessive incretin hormones may also contribute to pancreatic islet cell hyperplasia, leading to hyperinsulinism.¹⁷ Other proposed mechanisms of PGBH are the lack of a decrease in beta cell mass after gastric bypass, a postoperative increase in insulin sensitivity, a decrease in ghrelin (an insulin counterregulatory hormone), and an abnormal glucagon response.^13,17

Pathologic changes vary widely

PGBH is a challenging diagnosis to make pathologically. On review of pancreatic tissue from 36 patients undergoing partial pancreatectomy for PGBH, the pancreatic islet cells of the PGBH group were larger and more irregular compared with controls.^18,19 This histologic condition with islet-cell hypertrophy, hyperplasia, and other changes has been termed nesidioblastosis.^11,14,20 However, the pancreatic tissue appears grossly normal. The histopathologic findings can vary greatly in individual cases and in one-third of cases the pancreatic changes can be minimal, so that “normal” and PGBH cells can be nearly impossible to distinguish from each other.²¹

DIAGNOSIS AND TREATMENT

We recommend a stepwise approach to evaluating and treating PGBH (Figures 1 and 2).

Step 1: Evaluate blood glucose and Whipple triad

The first step is a thorough history, including food consumption and timing of hypoglycemic symptoms. Give the patient a glucometer to take home, with instructions to check blood glucose levels when hypoglycemic symptoms occur. The patient should keep a log documenting time tested, food consumed, symptoms, and blood glucose data.

Hypoglycemic symptoms are categorized as autonomic and neuroglycopenic. Autonomic symptoms include anxiety, palpitations, tremulousness, and diaphoresis. Neuroglycopenic symptoms include confusion, falls, seizures, and loss of consciousness.¹²

There are degrees of hypoglycemia and hypoglycemic symptoms. Clinical hypoglycemia—a blood glucose level low enough to cause signs or symptoms—can be confirmed by the Whipple triad:

• Measured low blood glucose
• Symptoms of low blood glucose
• Relief of symptoms when low blood glucose is corrected.

Hypoglycemic symptoms can occur when the blood glucose level falls to less than 55 mg/dL in healthy people, but this cutoff can shift lower in someone who has recurrent hypoglycemia.

When the Whipple triad is documented, rule out nonhyperinsulinemic causes of hypoglycemia such as hypothyroidism, adrenal insufficiency, underlying organ dysfunction (ie, liver disease), and medications that cause hypoglycemia.

Step 2: Modify the diet

If postprandial hypoglycemia is occurring, the next step is dietary modification. Two studies showed that a low-carbohydrate diet prevented hypoglycemia; however, these diets contained nearly no carbohydrates (with meals consisting of eggs, sausage, cheese, and black coffee or tea).^15,22

Instruct patients to never eat pure carbohydrates without fat or protein, as this can result in a more severe hypoglycemic response.²² In addition, foods with a high glycemic index (a measure of how a carbohydrate-containing food raises blood sugar) should be avoided, and a low glycemic index diet is recommended.²³ High glycemic index foods include white bread, bagels, pretzels, and pineapple. Low glycemic index foods include 100% stone-ground whole wheat or pumpernickel bread, lima beans, butter beans, peas, legumes, lentils, and nonstarchy vegetables.

Our bariatric surgeons provide all postbariatric surgery patients with the dietary guidelines shown in Table 3.²⁴ We also ask our patients with PGBH to limit carbohydrates to 15 to 30 g per meal and to limit added sugars to less than 4 g per meal, including regular and sugar alcohols (polyols). Snacks should contain only protein and fat. In severe cases, we further limit the diet to 15 g of carbohydrate per meal, with no added sugars.

The hypoglycemia occurring with PGBH is treated differently than the hypoglycemia that occurs in diabetic patients. Advise patients with PGBH to treat their hypoglycemic episodes with a simple sugar combined with a protein or fat (eg, a small handful of candy with a spoonful of peanut butter), as they will often have recurrent hypoglycemia if a simple sugar is used alone. If patients regain weight, ask them about frequent eating, which would be related to self-treatment of hypoglycemia.

Step 3: Start an alpha-glucosidase inhibitor

If postprandial hypoglycemia persists despite dietary modification, then start an alpha-glucosidase inhibitor such as acarbose. Acarbose inhibits carbohydrate absorption, resulting in a decreased insulin response; thus, it blunts the decline in postprandial blood glucose.

Unfortunately, gastrointestinal side effects such as flatulence, diarrhea, and abdominal pain occur in up to 20% of patients who take acarbose, often leading to its discontinuation.²⁵ To minimize gastrointestinal side effects, we usually start with 25 mg of acarbose with 1 meal daily for 1 week, then increase the dosage weekly to 25 mg with the other 2 meals. If tolerated, acarbose can be increased to 50 to 100 mg with 3 meals daily.

Step 4: Obtain a mixed meal tolerance test or a provocation meal test

If dietary changes and an alpha-glucosidase inhibitor do not prevent postprandial hypoglycemia from recurring, then confirmation of PGBH is needed, using a mixed meal tolerance test or a provocation meal test.

In a mixed meal tolerance test, the meal consists of 55% carbohydrate, 30% fat, and 15% protein. Patients with hyperinsulinemic hypoglycemia have a rapid rise in blood glucose (> 200 mg/dL) with a robust insulin response that is often followed by hypoglycemia after ingesting a meal containing carbohydrates in this test. Insulin levels that remain elevated after the plasma glucose level falls to less than 55 mg/dL indicate hyperinsulinism.¹¹

Nevertheless, a mixed meal tolerance test will not always induce hypoglycemia. In a study of 51 patients with PGBH, all wore a continuous glucose monitor, were instructed to follow their normal diet for 5 days, and then underwent a mixed meal tolerance test on day 6. The glucose monitor revealed hypoglycemia in 75% of patients, while the mixed meal tolerance test was positive in only 29%.⁵

Moreover, to date, there is no standardized mixed meal.^5,15 This might also explain the difference in prevalence of hypoglycemia detected by this test.

Based on these conflicting findings, we recommend a provocation meal test—ie, the patient is given foods that have induced hypoglycemia earlier.

Of note, the Endocrine Society guidelines on hypoglycemia state that an oral glucose tolerance test should never be used to document postprandial hypoglycemia.²⁶ Lev-Ran and Anderson²⁷ found that an oral glucose tolerance test could be positive in at least 10% of normal people.

Step 5: Consider other pharmacotherapy

For moderate to severe PGBH in which dietary modification and acarbose have failed, additional medical therapy is the next step. Medical therapies include calcium channel blockers, somatostatin analogues (eg, octreotide), and diazoxide.

Calcium channel blockers inhibit insulin release from beta cells²⁸ but at the risk of hypotension. Mordes and Alonso²⁹ treated 6 PGBH patients with nifedipine or verapamil with or without acarbose, and symptoms resolved in 5 of the 6 patients.

When we treat PGBH, we often add a calcium channel blocker as the next step in therapy if the patient has hypertension or if the blood pressure can tolerate this. If the patient’s blood pressure is low, then avoiding calcium channel blocker therapy may be necessary. The next step would be octreotide and then diazoxide.

Somatostatin analogues such as octreotide inhibit GLP-1 and insulin release.³⁰ The most common side effects of octreotide are diarrhea and abdominal pain. Bile stone formation can also occur, but this is not common.

Diazoxide opens adenosine triphosphate-sensitive potassium channels and reduces the opening of calcium channels, inhibiting insulin release and raising blood glucose. In a study of 6 Japanese patients with inoperable insulinoma, diazoxide was used to treat hypoglycemia.³¹ Unfortunately, the doses required to control the low blood sugars also led to adverse reactions, most of which involved edema secondary to volume overload and other heart failure symptoms. Diazoxide also commonly causes hypotension and hirsutism.

Step 6: 72-hour fast

A 72-hour fast is recommended in severe cases of PGBH in patients for whom dietary modification and the additional pharmacotherapy outlined in step 5 have failed. A 72-hour fast is always indicated in evaluating confirmed fasting hypoglycemia. People with insulinoma usually have fasting hypoglycemia, while patients with dumping syndrome do not. Patients with PGBH usually do not have fasting hypoglycemia, but they can in severe cases.¹¹

For safety, this test should be done in the hospital. Baseline plasma levels of insulin, C-peptide, proinsulin, beta-hydroxybutyrate, and glucose should be obtained. The patient then fasts, consuming only noncaloric and noncaffeinated beverages for 72 hours. During this time, capillary glucose checks are performed every 6 hours. If the capillary glucose level falls below 55 mg/dL,^11,26 then the baseline tests are redrawn along with a sulfonylurea screen. To reduce costs and unnecessary testing, the tests are not sent for laboratory processing unless the plasma glucose is less than 55 mg/dL.

When the plasma glucose is less than 55 mg/dL, insulin production should cease. Elevated insulin levels and insulin byproducts raise concern for hyperinsulinism. These values confirm hyperinsulinemic hypoglycemia²⁶:

• Glucose < 55 mg/dL
• Insulin ≥ 3 µU/mL
• C-peptide ≥ 0.2 nmol/L
• Proinsulin ≥ 5.0 pmol/L.

After hypoglycemia is confirmed, 1 mg of glucagon is given intravenously, and plasma glucose levels are obtained at 10, 20, and 30 minutes.^11,26 A rise in plasma glucose of at least 25 mg/dL after intravenous glucagon injection indicates hypoglycemia due to hyperinsulinemia. Two-thirds of patients with insulinoma experience hypoglycemia within the first 24 hours, and nearly all experience hypoglycemia within 48 hours.²⁶

Step 7: Obtain pancreatic imaging

If fasting hypoglycemia is present and hyperinsulinemic hypoglycemia is confirmed during a 72-hour fast, then pancreatic imaging should be obtained to evaluate for an insulinoma. We also recommend pancreatic imaging to rule out insulinoma when severe PGBH has not responded to dietary modification or pharmacotherapy.

Imaging is not recommended in PGBH that has been successfully treated with dietary modification with or without pharmacotherapy.

Endoscopic ultrasonography alone has 80% to 92% sensitivity for localizing a pancreatic mass as small as 5 mm. However, when coupled with computed tomography or magnetic resonance imaging, the sensitivity increases to nearly 100%.¹²

Step 8: Selective arterial calcium stimulation test

If a patient is found to have hyperinsulinemic hypoglycemia during a 72-hour fast but pancreatic imaging is negative, then selective arterial calcium stimulation testing (SACST) and hepatic vein sampling should be performed. Also, for severe PGBH, in which hypoglycemia has persisted despite dietary modification and pharmacotherapy, SACST can be performed to evaluate for possible localization of hyperinsulinism in patients considering surgery. For mild and moderate cases of PGBH, in which the hypoglycemia has been successfully treated with dietary changes with or without pharmacotherapy, SACST is not necessary.

This test can localize the area of excess insulin production in the pancreas in patients with an insulinoma. Patients with severe PGBH usually have diffuse hyperinsulinism without localization on SACST.^32,33

When SACST is performed, a sampling catheter is placed in the femoral vein. Calcium gluconate is injected into the major arteries of the pancreas (superior mesenteric, gastroduodenal, and splenic arteries). Calcium stimulates release of insulin from an insulinoma or hyperplastic beta cells. Resultant insulin levels are measured in the hepatic vein. If there is a greater than twofold increase in insulin release from 2 segments, then the test is considered positive.

Thompson et al³⁴ documented that insulin release from insulinoma is almost 4 times higher than in diffuse nesidioblastosis. SACST has a sensitivity of 96% for detecting insulinomas.³⁵

Step 9: Other alternatives and surgery

In patients with severe PGBH for whom dietary modification and all pharmacotherapy have failed and who continue to have debilitating neuroglycopenia, there are options before proceeding with surgery, the last resort in this condition.

Continuous glucose monitoring is helpful in many patients with severe PGBH. Many of them have hypoglycemia unawareness, and the monitor alerts them when their blood sugar is low. In addition, the monitor indicates when the blood sugar is dropping, so that intervention can occur before hypoglycemia occurs.

Unfortunately, insurance coverage for continuous monitors in this patient population is limited. We argue that insurance should cover the cost for these severe cases.

Pasireotide, a somatostatin analogue that is longer-acting than octreotide, is approved for use in Cushing disease and acromegaly and actually causes hyperglycemia. In a case report of a 50-year-old woman, pasireotide resulted in less hypoglycemia and higher glucagon levels then octreotide.³⁶ Pasireotide is available from Novartis for compassionate use in patients with severe PGBH.

Glucocorticoids are another off-label option. However, in excess, they can lead to iatrogenic Cushing syndrome, which has its own complications. Prednisone and diazoxide have been used together to help prevent hypoglycemia in a patients with inoperable insulinoma.³¹

Tube feeding. Some researchers have studied altering nutrition access through surgical means. McLaughlin et al³⁷ discussed a case of gastric tube insertion into the remnant stomach of a patient with PGBH, with resolution of hypoglycemic symptoms and hypoglycemia; however, this does not always provide complete resolution of symptoms.^37,38 If gastric bypass reversal is being considered, a trial of solely remnant stomach tube feeds (with no oral intake) should be pursued first. If this ameliorates the hypoglycemia, then gastric bypass reversal may be of benefit.

Surgery is the last resort if all of the above treatments have failed and severe debilitating neuroglycopenia persists. However, surgery poses risks, and the success rate in correcting hypoglycemia is not ideal. Surgical options include Roux-en-Y reversal, gastric pouch resection, and pancreatic resection.

In a review by Mala,² 75 patients with documented PGBH underwent surgical therapy. Hypoglycemic symptoms resolved in 34 of 51 pancreatic resections, 13 of 17 Roux-en-Y reversals, and 9 of 11 gastric pouch resections. However, the follow-up period was short.

As noted above, we recommend calcium stimulation testing only for severe cases of PGBH when surgery is being considered to evaluate for possible localization of hyperinsulinism for which partial pancreatectomy would be of benefit. Since there is no localization in many PGBH cases and the success rates are slightly higher in gastric bypass reversal, bypass reversal is usually preferred over partial or complete pancreatectomy.^2,32,33

POTENTIAL FUTURE THERAPIES

Given the elevated GLP-1 levels and robust insulin response to glucose observed in PGBH, blocking GLP-1 may provide clinical benefit. Salehi et al¹⁶ found that a GLP-1 antagonist prevented surges in GLP-1 and reduced hypoglycemic episodes in patients with PGBH. Unfortunately, the medication they used was given as a continuous infusion and is not currently available.

Conversely, a GLP-1 agonist showed benefit in a series of 5 cases of PGBH.³⁹ In addition, an insulin receptor antibody is undergoing phase 2 trials and has been shown to reverse insulin-induced hypoglycemia in rodents and humans; it may be a novel therapy in the future for hyperinsulinemic hypoglycemia.⁴⁰

MORE STUDY NEEDED

As the prevalence of obesity continues to rise and more people opt for bariatric surgery for weight loss, we will likely continue to see an increase in PGBH, since the onset of PGBH can be delayed for many years after surgery.²⁸

Unfortunately, the disease process involved in PGBH is not well understood. For example, we do not know why GLP-1 elevations or a robust insulin response causing hypoglycemia occurs in some but not all gastric bypass patients. Study is needed to elucidate the pathophysiology to further understand why most patients have no hypoglycemia after gastric bypass, some have mild to moderate PGBH, and a small percentage have severe PGBH with debilitating neuroglycopenia unresponsive to dietary changes and medications.     

The advent of the insulin pump in the late 1970s was a step forward in diabetes treatment,¹ and recent improvements make these devices easier to use in intensive insulin management. Today, more than 400,000 people in the United States are thought to be using an insulin pump.²

See related editorial

With a pump, patients can adjust the dosage and discreetly give themselves boluses by simply pushing a button instead of giving themselves multiple daily injections. Also, pump therapy can be tailored to correct for hepatic glucose production in a way that injections cannot.

This article reviews the clinical application of continuous subcutaneous insulin therapy—ie, the insulin pump—and provides recommendations for patient selection and management.

INDICATIONS FOR AN INSULIN PUMP

The American Association of Clinical Endocrinologists³ recommends considering an insulin pump for patients with type 1 or 2 diabetes mellitus who have a clear indication:

• Suboptimal control on basal-bolus injections, ie, not achieving glycemic goals despite maximal adherence to multiple daily injections
• Wide and erratic glycemic excursions
• Frequent severe hypoglycemia, or hypoglycemia unawareness
• A marked “dawn phenomenon” (spike in blood glucose level early in the morning)
• Pregnancy or planning for pregnancy
• Erratic lifestyle
• Personal preference.

WHO IS A GOOD CANDIDATE FOR AN INSULIN PUMP?

Good candidates for a pump are patients with type 1 diabetes (and some with type 2) who are well versed in taking multiple daily injections, are already checking their glucose four or more times daily, “counting carbs” (estimating or, preferably, measuring how much carbohydrate they are eating, and limiting their intake accordingly), and demonstrate the ability to adjust their dosing appropriately (Table 1).

A pump is not a shortcut to checking glucose less frequently or making fewer decisions. However, for those who actively manage their diabetes, it provides more real-time flexibility and some important safety features, as discussed below.

IS A PUMP BETTER THAN INJECTIONS?

Several studies have compared insulin pump therapy and multiple daily injections.^4–7 While some found no difference in glucose control in terms of hemoglobin A_1c or hypoglycemia, others showed improved glucose control with pumps in patients who had higher baseline hemoglobin A_1c levels (> 10%).⁶ In this subgroup, a pump lowered hemoglobin A_1c an additional estimated 0.65% compared with multiple daily injections.⁶ Fructosamine levels also improved in pump users.⁶

Using continuous glucose monitoring for 3 days in a study in children with type 1 diabetes, Schreiver et al⁸ found lower insulin requirements and less-severe glycemic excursions with a pump than with multiple daily injections.

A 2013 study⁹ of 57 patients ages 13 to 71 with type 2 diabetes who were struggling to control their blood sugar with multiple daily injections found that they achieved better control with less insulin using a pump.

A meta-analysis found pump therapy to be more effective than multiple daily injections for those who used it more than 1 year.¹⁰

ADVANTAGES AND DISADVANTAGES OF INSULIN PUMP THERAPY

Intensive glucose control reduces microvascular complications in type 1 diabetes.^11–14 The advantages of using a pump include better adherence, more accurate dosing, greater lifestyle flexibility, control of the dawn phenomenon without induction of nocturnal hypoglycemia, and the ability to suspend or temporarily reduce basal insulin to compensate for increased physical activity.¹⁵

Disadvantages include the high degree of technical aptitude required, the need for high-level engagement, skin reactions to tape, a higher risk of diabetic ketoacidosis from pump malfunction, infusion-site problems such as “tunneling” of insulin (leakage of insulin along the outside of the cannula and back to the skin surface) and clogging of the infusion set, and a risk of inactivation of insulin from exposure to heat, which can lead to ketoacidosis in a few hours if not addressed promptly.¹⁵

IS IT COST-EFFECTIVE?

There is evidence that continuous subcutaneous insulin infusion is cost-effective, both in general and compared with multiple daily injections for children and adults with type 1 diabetes mellitus. Cohen and Shaw¹⁶ found that life expectancy and quality-adjusted life-years increased in pump users, although the price per life-year gained varied greatly depending on the model used.

And this therapy is expensive. Most pumps cost more than $6,000, and supplies cost about $300 per month. Most insurance providers cover this therapy for patients with type 1 diabetes (Table 2) but less often for those with type 2. Further, many insurance policies have copayments, and patients may find a 20% co-payment a significant financial burden. Physicians need to obtain preapproval for insulin pumps from the insurance company. Typically, prescriptions for supplies are written annually. Despite these significant costs, most patients with type 1 diabetes who use an insulin pump find that the benefits of improved control and greater independence justify the cost.

An annual review of currently available insulin pumps and other diabetes-related equipment is published in Diabetes Forecast.¹⁷

PATIENT PERSPECTIVE ON INSULIN PUMP USE

Many patients who use a pump find that it gives them greater flexibility to adjust to day-to-day changes in schedules and routines. For example, consuming an extra serving at a meal could necessitate another injection for a patient on multiple daily injections, but a pump user would need only to push a few buttons. With cell phone apps available to control some pumps, many people find that an insulin pump is more discreet and easier to manage than carrying around injection supplies. Further, the complex calculations of carbohydrate ratios and correction factors are easier and more accurate with a pump.

In an open-label randomized study,¹⁸ 29 of 41 patients with type 1 diabetes said they preferred a pump to multiple daily injections.

Conversely, some people do not want a pump because it is attached all the time and identifies them to others as having an illness. Other patients do not trust a machine and want control in their own hands. (Actually, machines typically are much more reliable and less mistake-prone than humans.)

HOW DOES A PUMP WORK COMPARED WITH MULTIPLE DAILY INJECTIONS?

Patients taking multiple daily injections must use two types of insulin: a long-acting one that reaches a steady level in the blood without a peak and lasts from 12 to 24 hours, and a rapid-acting one taken with meals, usually having a peak of action and an effect lasting 3 to 5 hours. The idea is to approximate normal insulin patterns, with a basal level in the background and peaks (boluses) of insulin with carbohydrate intake.

Insulin pumps use only one kind of insulin—a rapid-acting one, ie, lispro, aspart, or glulisine. They preserve the basal-bolus concept, but with many refinements (discussed below).¹⁵

Most pumps are attached to the patient by plastic tubing that connects the reservoir to a subcutaneous cannula or steel needle. However, some pumps have a reservoir directly attached to a subcutaneous cannula without the tubing. This type of pump is controlled with a remote device.

The infusion set and the site should be changed every 3 days

The infusion set (cannula or needle and tubing) and the site should be changed every third day to minimize the risk of infection and abnormal delivery due to protein buildup on the cannula os, epithelial healing, and irritation around the site. Failure to do so often results in higher blood glucose concentrations.¹⁹

The patient and healthcare team work together to calculate the patient’s daily insulin needs, and the pump is programmed based on the patient’s requirements, lifestyle, and sensitivity to insulin. Once the pump is started, the patient operates it to deliver the insulin dose according to carbohydrate intake and blood glucose level.

PUMP SETTINGS

Basal rate

The basal rate is programmed by the physician and is intended to mimic physiologic insulin release. The pump can be set to a number of basal rates within any 24-hour period. This provides more physiologic matching of insulin delivery to hourly insulin needs based on the patient’s daily schedule.

If the patient has been taking multiple daily injections, the hourly basal rate can be calculated by dividing the daily basal dose by 24. However, lower rates are usually used after midnight, and rates are increased early in the morning to counteract the dawn phenomenon.

The rates can also be adjusted temporarily (for up to 24 hours), with a feature called the temporary basal rate. People tend to have higher blood glucose levels when they have a respiratory illness, are under significant stress, or are menstruating. Thus, a person with influenza could increase the basal rate by 25%, or a student could run a temporary basal rate of 150% for 4 hours before taking a final exam.

Conversely, exercising increases insulin’s effectiveness at the muscle level, and insulin requirements drop. To counteract this, one would temporarily decrease the basal rate in the pump before exercising.

Many factors affect the bolus dose

A pump is not a shortcut to checking glucose less frequently, or to making fewer decisions

A bolus of insulin is given for meals and to correct hyperglycemia, as with multiple daily injections. A pump calculates the bolus based on the carbohydrate ratio, correction factor, or both. These ratios are programmed into the pump by the physician. A benefit of the insulin pump is that the patient just has to input the amount of carbohydrates to be eaten or record a blood glucose level and the pump will calculate the bolus dose of insulin to be given.

The carbohydrate ratio is the amount of insulin that should be taken per amount of carbohydrate. A typical ratio is 1:15, meaning that the patient should take 1 unit of insulin for every 15 g of carbohydrates to be eaten. This varies by patient depending on insulin sensitivity.

The correction factor describes how much the glucose level is expected to drop per unit of insulin given. For example, if the target glucose level is 100 mg/dL and the correction factor is 25, then the patient will get 1 unit of correction of insulin if his or her glucose level is 125 mg/dL, 2 units if it is 150 mg/dL, and so on. A pump can dispense fractions of a unit.

The target glucose level or range is set by the physician and patient and is one of the factors the pump uses in calculating a bolus dose. Insulin pumps allow for multiple target glucose levels. Commonly, to minimize the risk of hypoglycemia, a higher (less strict) target is set for bedtime and overnight than for daytime.

Active insulin time defines how soon the patient can take another bolus.

Often, people eat more than they thought they would. They may also find that the glucose level did not increase or decrease as much as expected. Many patients who actively manage their glucose take additional boluses of insulin after a meal if their glucose is higher than they thought it would be. A patient taking injections cannot know how much of the insulin from the before-meal bolus is still working and has to guess.

Insulin pumps use a logarithmic formula to calculate this and prevent the user from “stacking” insulin boluses and lowering the glucose level too much. For example, if the active insulin time is 4 hours and the patient took a bolus for lunch at noon, he or she would be unable to take a full insulin correction dose until 4:00 pm. The patient can override this feature. Although the active insulin time varies from patient to patient, it is rarely more than 4 hours.

Additional safety features

Suspend. When a person who is taking insulin injections starts to experience hypoglycemia, he or she has one option—to eat something to treat the low blood glucose. The insulin injection has already been taken and cannot be reversed. However, with an insulin pump the patient can first suspend the pump so that no additional insulin is infused until it is safe again, and then eat to treat the low sugar level. This allows the patient to eat less, prevent overtreating, and, hopefully, prevent rebound hyperglycemia.

Reverse correction. When patients take insulin for an upcoming meal, they estimate the amount needed for the carbohydrates that they are about to eat as well as how much correction is needed. If their glucose level is below the target range, they may or may not subtract insulin from the dose to achieve the glucose target. The pump does this automatically, resulting in a lower dose of insulin for that bolus. This allows the patient to take a bolus for a meal even if he or she is below the target, and thus prevent hyperglycemia.

CAN INSULIN PUMPS BE USED IN THE HOSPITAL?

Patients can keep using their insulin pump in the hospital under the right conditions.

Inpatient hypoglycemia increases the risk of death, and although not all patients require tight glycemic control, there is still benefit in avoiding extremes in blood sugar levels,²⁰ including at night.^20–22 Insulin pump therapy, when used in the hospital, results in fewer episodes of severe hyperglycemia (glucose levels > 300 mg/dL) and hypoglycemia (levels < 40 mg/dL) than multiple daily injections.²² Moreover, most pump users feel more comfortable when they can manage their own therapy. Using the pump in the hospital has the additional benefit that patients can treat themselves before and after meals easily with less staff time and effort.

Bailon et al²³ retrospectively studied 35 patients with insulin pumps in 50 hospitalizations. More than half of the patients were allowed to continue using their pump in the hospital. Reasons for discontinuing the pump included lack of access to supplies, unfamiliarity with the pump, attempted suicide, malfunctioning hardware, diabetic ketoacidosis, and altered mental status. Patients using their pump had fewer episodes of hypoglycemia (glucose levels < 70 mg/dL) than patients who removed their pump. In patients who continued using the pump throughout their hospitalization, no adverse events (eg, site infection or mechanical failure) were noted.

Leonhardi et al²⁴ reviewed 25 hospital admissions, and the outcomes were similar to those reported by Bailon et al,²³ with no adverse outcomes related to the pumps.

When using an insulin pump in the hospital

Most insulin pumps cost more than $6,000, plus about $300 per month for supplies

When a physician wants a patient to continue using an insulin pump in the hospital, a number of things must happen. The nursing staff must be informed that the patient is wearing a pump and can self-administer insulin. Most facilities will still follow routine protocols for checking blood glucose but will document that the patient is administering his or her own insulin. The patient must be well enough to manage the pump. If the infusion site needs to be changed, the patient would be expected to do so with his or her own supplies.

Imaging and insulin pumps

Advice differs on what to do if a patient with an insulin pump needs to undergo radiographic imaging. For example, the University of Wisconsin radiology department says it is safe to keep an insulin pump in place if the x-ray beam will be on for less than 3 seconds at a time and if the device is covered by a lead apron.²⁵ However, radiation can induce electrical currents in the circuitry, which can alter the function of the pump. For this reason, some manufacturers recommend removing the device before the patient enters any room in which radiation or magnetic resonance imaging will be used.^26–31

Insulin pumps and surgery

Insulin pumps have been used in the perioperative and intraoperative periods, with positive outcomes.³² An analysis of 20 patients on pumps undergoing a total of 23 surgeries (mostly orthopedic procedures) found that 13 of the 20 patients wore their pump during surgery. No adverse events were noted in any of these cases, although the sample size was small.³³

Corney et al³⁴ retrospectively compared insulin pumps with alternative methods of perioperative glucose management. Multiple surgical specialties were included. No significant difference in mean blood glucose levels was found between those who continued to use their pump and those who used other methods. In those who continued to use their pump, there were no episodes of intraoperative technical difficulties related to the pump.

Any patient who may be undergoing a procedure or surgery must let the surgeon and anesthesiologist know that he or she has a pump. If the infusion site is too close to the site of the surgery or procedure, it must be moved.

Concerns during surgery include catheter or site disconnection or loss, crystallization within the tubing (a potential problem not limited to surgery), and pump malfunction. If the procedure involves imaging, the pump should probably be disconnected or covered by lead shielding as directed in the pump manufacturer’s manual. The surgeon and anesthesiologist must decide whether to continue use of a pump during a surgical procedure. However, the study by Corney et al³⁴ shows it is possible.

Most office-based procedures can be done with the insulin pump in place, as the patient is not under general anesthesia and so can adjust the insulin regimen as needed.

Abdelmalak et al,³⁵ in a comprehensive review of insulin pump use in noncardiac surgery, commented that the type of surgery may play a role in determining the best approach to perioperative glucose management. Major surgery causes a large inflammatory response that makes it difficult to control blood sugar, especially when steroids or beta agonists are given, whereas minor surgery does not affect blood glucose nearly as much. The authors offered recommendations on pump use during various surgical procedures depending on the length of the procedure:

• If surgery is anticipated to last less than 1 hour, then keep the insulin pump on, and have the patient manage corrections preoperatively and postoperatively.
• For surgery of intermediate length (1–3 hours), have the patient take a bolus of 1 hour’s worth of insulin (based on the basal rate for that time period) before the procedure, then remove the insulin pump. Do this only if blood sugar is normal or close to normal. If the patient is severely hyperglycemic, remove the insulin pump and start an intravenous insulin infusion.
• If the procedure will take more than 3 hours, remove the pump and start an insulin infusion regardless of the blood sugar level.³⁵

AIR TRAVEL AND INSULIN PUMPS

Insulin pumps can be easy to manage during airline travel if the user is prepared (Table 3).

First, it is important to have a letter from the treating physician stating that the pump is a necessary medical device. All supplies should be carried on and in a separate bag for easy inspection. The more forthcoming the user is at the security checkpoint, the easier the process.

According to the Transportation Security Administration, insulin pump users can keep their pump on during screening, and the metal detectors and full-body scanners will not harm the device.³⁶

However, manufacturer recommendations differ. Medtronic recommends that patients not expose their insulin pump to x-rays, and that instead of going through a full-body scanner the patient should request a pat-down.³⁷ Animas recommends the same.³⁸ OmniPod states that their system can be worn through airport imaging, making it the only approved continuous insulin delivery system that can be taken through airport imaging.³⁹

Another potential problem is the change in atmospheric pressure during takeoff and landing. Bubbles can form in the insulin reservoir as air pressure decreases with ascent, thereby displacing insulin from the pump to the patient. The opposite happens during descent. King et al⁴⁰ corroborated this phenomenon with Animas and Medtronic pumps. Asante recommends removing their pump tubing during takeoff and landing.³⁰

If PROBLEMS ARISE

Like any machine, an insulin pump can fail. Most failures result in lack of insulin delivery—the patient does not get excess insulin from insulin pump failure. Excess insulin delivery is most often due to operator error. All insulin is either preprogrammed (basal by provider or patient) or must be confirmed by the patient at the time of delivery (meal or correction boluses).

Pump manufacturers have 24-hour support programs and hotlines, with experts who will either walk the patient through the problem or send a replacement pump—often within 24 hours.

EVOLVING TECHNOLOGY

Pump technology is evolving quickly. On the way are “smart” pumps that interact with other systems, smaller pumps with advanced touch-screen features, and patch pumps that do not have tubing but operate similarly to pumps with tubing (ie, a cannula is still required for insulin delivery).

Some insulin pumps can be linked to an external glucose sensor. These systems provide a great amount of information to the patient and provider. Often, there is increased awareness of fluctuations in glucose, allowing earlier intervention to prevent high and low glucose excursions. Sensor-augmented pumps may further improve safety by suspending infusion during hypoglycemia.^41,42

Researchers continue to strive for closed-loop systems that would allow the pump to automatically respond to circulating glucose and thus provide truly physiologic control.⁴³ A recent study showed the effectiveness of the outpatient use of a bihormonal (insulin and glucagon) “bionic pancreas,” which provided improved glucose control and similar or less hypoglycemia in adults and adolescents who had been using a traditional insulin pump.⁴⁴

The advent of the insulin pump in the late 1970s was a step forward in diabetes treatment,¹ and recent improvements make these devices easier to use in intensive insulin management. Today, more than 400,000 people in the United States are thought to be using an insulin pump.²

See related editorial

With a pump, patients can adjust the dosage and discreetly give themselves boluses by simply pushing a button instead of giving themselves multiple daily injections. Also, pump therapy can be tailored to correct for hepatic glucose production in a way that injections cannot.

This article reviews the clinical application of continuous subcutaneous insulin therapy—ie, the insulin pump—and provides recommendations for patient selection and management.

INDICATIONS FOR AN INSULIN PUMP

The American Association of Clinical Endocrinologists³ recommends considering an insulin pump for patients with type 1 or 2 diabetes mellitus who have a clear indication:

• Suboptimal control on basal-bolus injections, ie, not achieving glycemic goals despite maximal adherence to multiple daily injections
• Wide and erratic glycemic excursions
• Frequent severe hypoglycemia, or hypoglycemia unawareness
• A marked “dawn phenomenon” (spike in blood glucose level early in the morning)
• Pregnancy or planning for pregnancy
• Erratic lifestyle
• Personal preference.

WHO IS A GOOD CANDIDATE FOR AN INSULIN PUMP?

Good candidates for a pump are patients with type 1 diabetes (and some with type 2) who are well versed in taking multiple daily injections, are already checking their glucose four or more times daily, “counting carbs” (estimating or, preferably, measuring how much carbohydrate they are eating, and limiting their intake accordingly), and demonstrate the ability to adjust their dosing appropriately (Table 1).

A pump is not a shortcut to checking glucose less frequently or making fewer decisions. However, for those who actively manage their diabetes, it provides more real-time flexibility and some important safety features, as discussed below.

IS A PUMP BETTER THAN INJECTIONS?

Several studies have compared insulin pump therapy and multiple daily injections.^4–7 While some found no difference in glucose control in terms of hemoglobin A_1c or hypoglycemia, others showed improved glucose control with pumps in patients who had higher baseline hemoglobin A_1c levels (> 10%).⁶ In this subgroup, a pump lowered hemoglobin A_1c an additional estimated 0.65% compared with multiple daily injections.⁶ Fructosamine levels also improved in pump users.⁶

Using continuous glucose monitoring for 3 days in a study in children with type 1 diabetes, Schreiver et al⁸ found lower insulin requirements and less-severe glycemic excursions with a pump than with multiple daily injections.

A 2013 study⁹ of 57 patients ages 13 to 71 with type 2 diabetes who were struggling to control their blood sugar with multiple daily injections found that they achieved better control with less insulin using a pump.

A meta-analysis found pump therapy to be more effective than multiple daily injections for those who used it more than 1 year.¹⁰

ADVANTAGES AND DISADVANTAGES OF INSULIN PUMP THERAPY

Intensive glucose control reduces microvascular complications in type 1 diabetes.^11–14 The advantages of using a pump include better adherence, more accurate dosing, greater lifestyle flexibility, control of the dawn phenomenon without induction of nocturnal hypoglycemia, and the ability to suspend or temporarily reduce basal insulin to compensate for increased physical activity.¹⁵

Disadvantages include the high degree of technical aptitude required, the need for high-level engagement, skin reactions to tape, a higher risk of diabetic ketoacidosis from pump malfunction, infusion-site problems such as “tunneling” of insulin (leakage of insulin along the outside of the cannula and back to the skin surface) and clogging of the infusion set, and a risk of inactivation of insulin from exposure to heat, which can lead to ketoacidosis in a few hours if not addressed promptly.¹⁵

IS IT COST-EFFECTIVE?

There is evidence that continuous subcutaneous insulin infusion is cost-effective, both in general and compared with multiple daily injections for children and adults with type 1 diabetes mellitus. Cohen and Shaw¹⁶ found that life expectancy and quality-adjusted life-years increased in pump users, although the price per life-year gained varied greatly depending on the model used.

And this therapy is expensive. Most pumps cost more than $6,000, and supplies cost about $300 per month. Most insurance providers cover this therapy for patients with type 1 diabetes (Table 2) but less often for those with type 2. Further, many insurance policies have copayments, and patients may find a 20% co-payment a significant financial burden. Physicians need to obtain preapproval for insulin pumps from the insurance company. Typically, prescriptions for supplies are written annually. Despite these significant costs, most patients with type 1 diabetes who use an insulin pump find that the benefits of improved control and greater independence justify the cost.

An annual review of currently available insulin pumps and other diabetes-related equipment is published in Diabetes Forecast.¹⁷

PATIENT PERSPECTIVE ON INSULIN PUMP USE

Many patients who use a pump find that it gives them greater flexibility to adjust to day-to-day changes in schedules and routines. For example, consuming an extra serving at a meal could necessitate another injection for a patient on multiple daily injections, but a pump user would need only to push a few buttons. With cell phone apps available to control some pumps, many people find that an insulin pump is more discreet and easier to manage than carrying around injection supplies. Further, the complex calculations of carbohydrate ratios and correction factors are easier and more accurate with a pump.

In an open-label randomized study,¹⁸ 29 of 41 patients with type 1 diabetes said they preferred a pump to multiple daily injections.

Conversely, some people do not want a pump because it is attached all the time and identifies them to others as having an illness. Other patients do not trust a machine and want control in their own hands. (Actually, machines typically are much more reliable and less mistake-prone than humans.)

HOW DOES A PUMP WORK COMPARED WITH MULTIPLE DAILY INJECTIONS?

Patients taking multiple daily injections must use two types of insulin: a long-acting one that reaches a steady level in the blood without a peak and lasts from 12 to 24 hours, and a rapid-acting one taken with meals, usually having a peak of action and an effect lasting 3 to 5 hours. The idea is to approximate normal insulin patterns, with a basal level in the background and peaks (boluses) of insulin with carbohydrate intake.

Insulin pumps use only one kind of insulin—a rapid-acting one, ie, lispro, aspart, or glulisine. They preserve the basal-bolus concept, but with many refinements (discussed below).¹⁵

Most pumps are attached to the patient by plastic tubing that connects the reservoir to a subcutaneous cannula or steel needle. However, some pumps have a reservoir directly attached to a subcutaneous cannula without the tubing. This type of pump is controlled with a remote device.

The infusion set and the site should be changed every 3 days

The infusion set (cannula or needle and tubing) and the site should be changed every third day to minimize the risk of infection and abnormal delivery due to protein buildup on the cannula os, epithelial healing, and irritation around the site. Failure to do so often results in higher blood glucose concentrations.¹⁹

The patient and healthcare team work together to calculate the patient’s daily insulin needs, and the pump is programmed based on the patient’s requirements, lifestyle, and sensitivity to insulin. Once the pump is started, the patient operates it to deliver the insulin dose according to carbohydrate intake and blood glucose level.

PUMP SETTINGS

Basal rate

The basal rate is programmed by the physician and is intended to mimic physiologic insulin release. The pump can be set to a number of basal rates within any 24-hour period. This provides more physiologic matching of insulin delivery to hourly insulin needs based on the patient’s daily schedule.

If the patient has been taking multiple daily injections, the hourly basal rate can be calculated by dividing the daily basal dose by 24. However, lower rates are usually used after midnight, and rates are increased early in the morning to counteract the dawn phenomenon.

The rates can also be adjusted temporarily (for up to 24 hours), with a feature called the temporary basal rate. People tend to have higher blood glucose levels when they have a respiratory illness, are under significant stress, or are menstruating. Thus, a person with influenza could increase the basal rate by 25%, or a student could run a temporary basal rate of 150% for 4 hours before taking a final exam.

Conversely, exercising increases insulin’s effectiveness at the muscle level, and insulin requirements drop. To counteract this, one would temporarily decrease the basal rate in the pump before exercising.

Many factors affect the bolus dose

A pump is not a shortcut to checking glucose less frequently, or to making fewer decisions

A bolus of insulin is given for meals and to correct hyperglycemia, as with multiple daily injections. A pump calculates the bolus based on the carbohydrate ratio, correction factor, or both. These ratios are programmed into the pump by the physician. A benefit of the insulin pump is that the patient just has to input the amount of carbohydrates to be eaten or record a blood glucose level and the pump will calculate the bolus dose of insulin to be given.

The carbohydrate ratio is the amount of insulin that should be taken per amount of carbohydrate. A typical ratio is 1:15, meaning that the patient should take 1 unit of insulin for every 15 g of carbohydrates to be eaten. This varies by patient depending on insulin sensitivity.

The correction factor describes how much the glucose level is expected to drop per unit of insulin given. For example, if the target glucose level is 100 mg/dL and the correction factor is 25, then the patient will get 1 unit of correction of insulin if his or her glucose level is 125 mg/dL, 2 units if it is 150 mg/dL, and so on. A pump can dispense fractions of a unit.

The target glucose level or range is set by the physician and patient and is one of the factors the pump uses in calculating a bolus dose. Insulin pumps allow for multiple target glucose levels. Commonly, to minimize the risk of hypoglycemia, a higher (less strict) target is set for bedtime and overnight than for daytime.

Active insulin time defines how soon the patient can take another bolus.

Often, people eat more than they thought they would. They may also find that the glucose level did not increase or decrease as much as expected. Many patients who actively manage their glucose take additional boluses of insulin after a meal if their glucose is higher than they thought it would be. A patient taking injections cannot know how much of the insulin from the before-meal bolus is still working and has to guess.

Insulin pumps use a logarithmic formula to calculate this and prevent the user from “stacking” insulin boluses and lowering the glucose level too much. For example, if the active insulin time is 4 hours and the patient took a bolus for lunch at noon, he or she would be unable to take a full insulin correction dose until 4:00 pm. The patient can override this feature. Although the active insulin time varies from patient to patient, it is rarely more than 4 hours.

Additional safety features

Suspend. When a person who is taking insulin injections starts to experience hypoglycemia, he or she has one option—to eat something to treat the low blood glucose. The insulin injection has already been taken and cannot be reversed. However, with an insulin pump the patient can first suspend the pump so that no additional insulin is infused until it is safe again, and then eat to treat the low sugar level. This allows the patient to eat less, prevent overtreating, and, hopefully, prevent rebound hyperglycemia.

Reverse correction. When patients take insulin for an upcoming meal, they estimate the amount needed for the carbohydrates that they are about to eat as well as how much correction is needed. If their glucose level is below the target range, they may or may not subtract insulin from the dose to achieve the glucose target. The pump does this automatically, resulting in a lower dose of insulin for that bolus. This allows the patient to take a bolus for a meal even if he or she is below the target, and thus prevent hyperglycemia.

CAN INSULIN PUMPS BE USED IN THE HOSPITAL?

Patients can keep using their insulin pump in the hospital under the right conditions.

Inpatient hypoglycemia increases the risk of death, and although not all patients require tight glycemic control, there is still benefit in avoiding extremes in blood sugar levels,²⁰ including at night.^20–22 Insulin pump therapy, when used in the hospital, results in fewer episodes of severe hyperglycemia (glucose levels > 300 mg/dL) and hypoglycemia (levels < 40 mg/dL) than multiple daily injections.²² Moreover, most pump users feel more comfortable when they can manage their own therapy. Using the pump in the hospital has the additional benefit that patients can treat themselves before and after meals easily with less staff time and effort.

Bailon et al²³ retrospectively studied 35 patients with insulin pumps in 50 hospitalizations. More than half of the patients were allowed to continue using their pump in the hospital. Reasons for discontinuing the pump included lack of access to supplies, unfamiliarity with the pump, attempted suicide, malfunctioning hardware, diabetic ketoacidosis, and altered mental status. Patients using their pump had fewer episodes of hypoglycemia (glucose levels < 70 mg/dL) than patients who removed their pump. In patients who continued using the pump throughout their hospitalization, no adverse events (eg, site infection or mechanical failure) were noted.

Leonhardi et al²⁴ reviewed 25 hospital admissions, and the outcomes were similar to those reported by Bailon et al,²³ with no adverse outcomes related to the pumps.

When using an insulin pump in the hospital

Most insulin pumps cost more than $6,000, plus about $300 per month for supplies

When a physician wants a patient to continue using an insulin pump in the hospital, a number of things must happen. The nursing staff must be informed that the patient is wearing a pump and can self-administer insulin. Most facilities will still follow routine protocols for checking blood glucose but will document that the patient is administering his or her own insulin. The patient must be well enough to manage the pump. If the infusion site needs to be changed, the patient would be expected to do so with his or her own supplies.

Imaging and insulin pumps

Advice differs on what to do if a patient with an insulin pump needs to undergo radiographic imaging. For example, the University of Wisconsin radiology department says it is safe to keep an insulin pump in place if the x-ray beam will be on for less than 3 seconds at a time and if the device is covered by a lead apron.²⁵ However, radiation can induce electrical currents in the circuitry, which can alter the function of the pump. For this reason, some manufacturers recommend removing the device before the patient enters any room in which radiation or magnetic resonance imaging will be used.^26–31

Insulin pumps and surgery

Insulin pumps have been used in the perioperative and intraoperative periods, with positive outcomes.³² An analysis of 20 patients on pumps undergoing a total of 23 surgeries (mostly orthopedic procedures) found that 13 of the 20 patients wore their pump during surgery. No adverse events were noted in any of these cases, although the sample size was small.³³

Corney et al³⁴ retrospectively compared insulin pumps with alternative methods of perioperative glucose management. Multiple surgical specialties were included. No significant difference in mean blood glucose levels was found between those who continued to use their pump and those who used other methods. In those who continued to use their pump, there were no episodes of intraoperative technical difficulties related to the pump.

Any patient who may be undergoing a procedure or surgery must let the surgeon and anesthesiologist know that he or she has a pump. If the infusion site is too close to the site of the surgery or procedure, it must be moved.

Concerns during surgery include catheter or site disconnection or loss, crystallization within the tubing (a potential problem not limited to surgery), and pump malfunction. If the procedure involves imaging, the pump should probably be disconnected or covered by lead shielding as directed in the pump manufacturer’s manual. The surgeon and anesthesiologist must decide whether to continue use of a pump during a surgical procedure. However, the study by Corney et al³⁴ shows it is possible.

Most office-based procedures can be done with the insulin pump in place, as the patient is not under general anesthesia and so can adjust the insulin regimen as needed.

Abdelmalak et al,³⁵ in a comprehensive review of insulin pump use in noncardiac surgery, commented that the type of surgery may play a role in determining the best approach to perioperative glucose management. Major surgery causes a large inflammatory response that makes it difficult to control blood sugar, especially when steroids or beta agonists are given, whereas minor surgery does not affect blood glucose nearly as much. The authors offered recommendations on pump use during various surgical procedures depending on the length of the procedure:

• If surgery is anticipated to last less than 1 hour, then keep the insulin pump on, and have the patient manage corrections preoperatively and postoperatively.
• For surgery of intermediate length (1–3 hours), have the patient take a bolus of 1 hour’s worth of insulin (based on the basal rate for that time period) before the procedure, then remove the insulin pump. Do this only if blood sugar is normal or close to normal. If the patient is severely hyperglycemic, remove the insulin pump and start an intravenous insulin infusion.
• If the procedure will take more than 3 hours, remove the pump and start an insulin infusion regardless of the blood sugar level.³⁵

AIR TRAVEL AND INSULIN PUMPS

Insulin pumps can be easy to manage during airline travel if the user is prepared (Table 3).

First, it is important to have a letter from the treating physician stating that the pump is a necessary medical device. All supplies should be carried on and in a separate bag for easy inspection. The more forthcoming the user is at the security checkpoint, the easier the process.

According to the Transportation Security Administration, insulin pump users can keep their pump on during screening, and the metal detectors and full-body scanners will not harm the device.³⁶

However, manufacturer recommendations differ. Medtronic recommends that patients not expose their insulin pump to x-rays, and that instead of going through a full-body scanner the patient should request a pat-down.³⁷ Animas recommends the same.³⁸ OmniPod states that their system can be worn through airport imaging, making it the only approved continuous insulin delivery system that can be taken through airport imaging.³⁹

Another potential problem is the change in atmospheric pressure during takeoff and landing. Bubbles can form in the insulin reservoir as air pressure decreases with ascent, thereby displacing insulin from the pump to the patient. The opposite happens during descent. King et al⁴⁰ corroborated this phenomenon with Animas and Medtronic pumps. Asante recommends removing their pump tubing during takeoff and landing.³⁰

If PROBLEMS ARISE

Like any machine, an insulin pump can fail. Most failures result in lack of insulin delivery—the patient does not get excess insulin from insulin pump failure. Excess insulin delivery is most often due to operator error. All insulin is either preprogrammed (basal by provider or patient) or must be confirmed by the patient at the time of delivery (meal or correction boluses).

Pump manufacturers have 24-hour support programs and hotlines, with experts who will either walk the patient through the problem or send a replacement pump—often within 24 hours.

EVOLVING TECHNOLOGY

Pump technology is evolving quickly. On the way are “smart” pumps that interact with other systems, smaller pumps with advanced touch-screen features, and patch pumps that do not have tubing but operate similarly to pumps with tubing (ie, a cannula is still required for insulin delivery).

Some insulin pumps can be linked to an external glucose sensor. These systems provide a great amount of information to the patient and provider. Often, there is increased awareness of fluctuations in glucose, allowing earlier intervention to prevent high and low glucose excursions. Sensor-augmented pumps may further improve safety by suspending infusion during hypoglycemia.^41,42

Researchers continue to strive for closed-loop systems that would allow the pump to automatically respond to circulating glucose and thus provide truly physiologic control.⁴³ A recent study showed the effectiveness of the outpatient use of a bihormonal (insulin and glucagon) “bionic pancreas,” which provided improved glucose control and similar or less hypoglycemia in adults and adolescents who had been using a traditional insulin pump.⁴⁴

The advent of the insulin pump in the late 1970s was a step forward in diabetes treatment,¹ and recent improvements make these devices easier to use in intensive insulin management. Today, more than 400,000 people in the United States are thought to be using an insulin pump.²

See related editorial

With a pump, patients can adjust the dosage and discreetly give themselves boluses by simply pushing a button instead of giving themselves multiple daily injections. Also, pump therapy can be tailored to correct for hepatic glucose production in a way that injections cannot.

This article reviews the clinical application of continuous subcutaneous insulin therapy—ie, the insulin pump—and provides recommendations for patient selection and management.

INDICATIONS FOR AN INSULIN PUMP

The American Association of Clinical Endocrinologists³ recommends considering an insulin pump for patients with type 1 or 2 diabetes mellitus who have a clear indication:

• Suboptimal control on basal-bolus injections, ie, not achieving glycemic goals despite maximal adherence to multiple daily injections
• Wide and erratic glycemic excursions
• Frequent severe hypoglycemia, or hypoglycemia unawareness
• A marked “dawn phenomenon” (spike in blood glucose level early in the morning)
• Pregnancy or planning for pregnancy
• Erratic lifestyle
• Personal preference.

WHO IS A GOOD CANDIDATE FOR AN INSULIN PUMP?

Good candidates for a pump are patients with type 1 diabetes (and some with type 2) who are well versed in taking multiple daily injections, are already checking their glucose four or more times daily, “counting carbs” (estimating or, preferably, measuring how much carbohydrate they are eating, and limiting their intake accordingly), and demonstrate the ability to adjust their dosing appropriately (Table 1).

A pump is not a shortcut to checking glucose less frequently or making fewer decisions. However, for those who actively manage their diabetes, it provides more real-time flexibility and some important safety features, as discussed below.

IS A PUMP BETTER THAN INJECTIONS?

Several studies have compared insulin pump therapy and multiple daily injections.^4–7 While some found no difference in glucose control in terms of hemoglobin A_1c or hypoglycemia, others showed improved glucose control with pumps in patients who had higher baseline hemoglobin A_1c levels (> 10%).⁶ In this subgroup, a pump lowered hemoglobin A_1c an additional estimated 0.65% compared with multiple daily injections.⁶ Fructosamine levels also improved in pump users.⁶

Using continuous glucose monitoring for 3 days in a study in children with type 1 diabetes, Schreiver et al⁸ found lower insulin requirements and less-severe glycemic excursions with a pump than with multiple daily injections.

A 2013 study⁹ of 57 patients ages 13 to 71 with type 2 diabetes who were struggling to control their blood sugar with multiple daily injections found that they achieved better control with less insulin using a pump.

A meta-analysis found pump therapy to be more effective than multiple daily injections for those who used it more than 1 year.¹⁰

ADVANTAGES AND DISADVANTAGES OF INSULIN PUMP THERAPY

Intensive glucose control reduces microvascular complications in type 1 diabetes.^11–14 The advantages of using a pump include better adherence, more accurate dosing, greater lifestyle flexibility, control of the dawn phenomenon without induction of nocturnal hypoglycemia, and the ability to suspend or temporarily reduce basal insulin to compensate for increased physical activity.¹⁵

Disadvantages include the high degree of technical aptitude required, the need for high-level engagement, skin reactions to tape, a higher risk of diabetic ketoacidosis from pump malfunction, infusion-site problems such as “tunneling” of insulin (leakage of insulin along the outside of the cannula and back to the skin surface) and clogging of the infusion set, and a risk of inactivation of insulin from exposure to heat, which can lead to ketoacidosis in a few hours if not addressed promptly.¹⁵

IS IT COST-EFFECTIVE?

There is evidence that continuous subcutaneous insulin infusion is cost-effective, both in general and compared with multiple daily injections for children and adults with type 1 diabetes mellitus. Cohen and Shaw¹⁶ found that life expectancy and quality-adjusted life-years increased in pump users, although the price per life-year gained varied greatly depending on the model used.

And this therapy is expensive. Most pumps cost more than $6,000, and supplies cost about $300 per month. Most insurance providers cover this therapy for patients with type 1 diabetes (Table 2) but less often for those with type 2. Further, many insurance policies have copayments, and patients may find a 20% co-payment a significant financial burden. Physicians need to obtain preapproval for insulin pumps from the insurance company. Typically, prescriptions for supplies are written annually. Despite these significant costs, most patients with type 1 diabetes who use an insulin pump find that the benefits of improved control and greater independence justify the cost.

An annual review of currently available insulin pumps and other diabetes-related equipment is published in Diabetes Forecast.¹⁷

PATIENT PERSPECTIVE ON INSULIN PUMP USE

Many patients who use a pump find that it gives them greater flexibility to adjust to day-to-day changes in schedules and routines. For example, consuming an extra serving at a meal could necessitate another injection for a patient on multiple daily injections, but a pump user would need only to push a few buttons. With cell phone apps available to control some pumps, many people find that an insulin pump is more discreet and easier to manage than carrying around injection supplies. Further, the complex calculations of carbohydrate ratios and correction factors are easier and more accurate with a pump.

In an open-label randomized study,¹⁸ 29 of 41 patients with type 1 diabetes said they preferred a pump to multiple daily injections.

Conversely, some people do not want a pump because it is attached all the time and identifies them to others as having an illness. Other patients do not trust a machine and want control in their own hands. (Actually, machines typically are much more reliable and less mistake-prone than humans.)

HOW DOES A PUMP WORK COMPARED WITH MULTIPLE DAILY INJECTIONS?

Patients taking multiple daily injections must use two types of insulin: a long-acting one that reaches a steady level in the blood without a peak and lasts from 12 to 24 hours, and a rapid-acting one taken with meals, usually having a peak of action and an effect lasting 3 to 5 hours. The idea is to approximate normal insulin patterns, with a basal level in the background and peaks (boluses) of insulin with carbohydrate intake.

Insulin pumps use only one kind of insulin—a rapid-acting one, ie, lispro, aspart, or glulisine. They preserve the basal-bolus concept, but with many refinements (discussed below).¹⁵

Most pumps are attached to the patient by plastic tubing that connects the reservoir to a subcutaneous cannula or steel needle. However, some pumps have a reservoir directly attached to a subcutaneous cannula without the tubing. This type of pump is controlled with a remote device.

The infusion set and the site should be changed every 3 days

The infusion set (cannula or needle and tubing) and the site should be changed every third day to minimize the risk of infection and abnormal delivery due to protein buildup on the cannula os, epithelial healing, and irritation around the site. Failure to do so often results in higher blood glucose concentrations.¹⁹

The patient and healthcare team work together to calculate the patient’s daily insulin needs, and the pump is programmed based on the patient’s requirements, lifestyle, and sensitivity to insulin. Once the pump is started, the patient operates it to deliver the insulin dose according to carbohydrate intake and blood glucose level.

PUMP SETTINGS

Basal rate

The basal rate is programmed by the physician and is intended to mimic physiologic insulin release. The pump can be set to a number of basal rates within any 24-hour period. This provides more physiologic matching of insulin delivery to hourly insulin needs based on the patient’s daily schedule.

If the patient has been taking multiple daily injections, the hourly basal rate can be calculated by dividing the daily basal dose by 24. However, lower rates are usually used after midnight, and rates are increased early in the morning to counteract the dawn phenomenon.

The rates can also be adjusted temporarily (for up to 24 hours), with a feature called the temporary basal rate. People tend to have higher blood glucose levels when they have a respiratory illness, are under significant stress, or are menstruating. Thus, a person with influenza could increase the basal rate by 25%, or a student could run a temporary basal rate of 150% for 4 hours before taking a final exam.

Conversely, exercising increases insulin’s effectiveness at the muscle level, and insulin requirements drop. To counteract this, one would temporarily decrease the basal rate in the pump before exercising.

Many factors affect the bolus dose

A pump is not a shortcut to checking glucose less frequently, or to making fewer decisions

A bolus of insulin is given for meals and to correct hyperglycemia, as with multiple daily injections. A pump calculates the bolus based on the carbohydrate ratio, correction factor, or both. These ratios are programmed into the pump by the physician. A benefit of the insulin pump is that the patient just has to input the amount of carbohydrates to be eaten or record a blood glucose level and the pump will calculate the bolus dose of insulin to be given.

The carbohydrate ratio is the amount of insulin that should be taken per amount of carbohydrate. A typical ratio is 1:15, meaning that the patient should take 1 unit of insulin for every 15 g of carbohydrates to be eaten. This varies by patient depending on insulin sensitivity.

The correction factor describes how much the glucose level is expected to drop per unit of insulin given. For example, if the target glucose level is 100 mg/dL and the correction factor is 25, then the patient will get 1 unit of correction of insulin if his or her glucose level is 125 mg/dL, 2 units if it is 150 mg/dL, and so on. A pump can dispense fractions of a unit.

The target glucose level or range is set by the physician and patient and is one of the factors the pump uses in calculating a bolus dose. Insulin pumps allow for multiple target glucose levels. Commonly, to minimize the risk of hypoglycemia, a higher (less strict) target is set for bedtime and overnight than for daytime.

Active insulin time defines how soon the patient can take another bolus.

Often, people eat more than they thought they would. They may also find that the glucose level did not increase or decrease as much as expected. Many patients who actively manage their glucose take additional boluses of insulin after a meal if their glucose is higher than they thought it would be. A patient taking injections cannot know how much of the insulin from the before-meal bolus is still working and has to guess.

Insulin pumps use a logarithmic formula to calculate this and prevent the user from “stacking” insulin boluses and lowering the glucose level too much. For example, if the active insulin time is 4 hours and the patient took a bolus for lunch at noon, he or she would be unable to take a full insulin correction dose until 4:00 pm. The patient can override this feature. Although the active insulin time varies from patient to patient, it is rarely more than 4 hours.

Additional safety features

Suspend. When a person who is taking insulin injections starts to experience hypoglycemia, he or she has one option—to eat something to treat the low blood glucose. The insulin injection has already been taken and cannot be reversed. However, with an insulin pump the patient can first suspend the pump so that no additional insulin is infused until it is safe again, and then eat to treat the low sugar level. This allows the patient to eat less, prevent overtreating, and, hopefully, prevent rebound hyperglycemia.

Reverse correction. When patients take insulin for an upcoming meal, they estimate the amount needed for the carbohydrates that they are about to eat as well as how much correction is needed. If their glucose level is below the target range, they may or may not subtract insulin from the dose to achieve the glucose target. The pump does this automatically, resulting in a lower dose of insulin for that bolus. This allows the patient to take a bolus for a meal even if he or she is below the target, and thus prevent hyperglycemia.

CAN INSULIN PUMPS BE USED IN THE HOSPITAL?

Patients can keep using their insulin pump in the hospital under the right conditions.

Inpatient hypoglycemia increases the risk of death, and although not all patients require tight glycemic control, there is still benefit in avoiding extremes in blood sugar levels,²⁰ including at night.^20–22 Insulin pump therapy, when used in the hospital, results in fewer episodes of severe hyperglycemia (glucose levels > 300 mg/dL) and hypoglycemia (levels < 40 mg/dL) than multiple daily injections.²² Moreover, most pump users feel more comfortable when they can manage their own therapy. Using the pump in the hospital has the additional benefit that patients can treat themselves before and after meals easily with less staff time and effort.

Bailon et al²³ retrospectively studied 35 patients with insulin pumps in 50 hospitalizations. More than half of the patients were allowed to continue using their pump in the hospital. Reasons for discontinuing the pump included lack of access to supplies, unfamiliarity with the pump, attempted suicide, malfunctioning hardware, diabetic ketoacidosis, and altered mental status. Patients using their pump had fewer episodes of hypoglycemia (glucose levels < 70 mg/dL) than patients who removed their pump. In patients who continued using the pump throughout their hospitalization, no adverse events (eg, site infection or mechanical failure) were noted.

Leonhardi et al²⁴ reviewed 25 hospital admissions, and the outcomes were similar to those reported by Bailon et al,²³ with no adverse outcomes related to the pumps.

When using an insulin pump in the hospital

Most insulin pumps cost more than $6,000, plus about $300 per month for supplies

When a physician wants a patient to continue using an insulin pump in the hospital, a number of things must happen. The nursing staff must be informed that the patient is wearing a pump and can self-administer insulin. Most facilities will still follow routine protocols for checking blood glucose but will document that the patient is administering his or her own insulin. The patient must be well enough to manage the pump. If the infusion site needs to be changed, the patient would be expected to do so with his or her own supplies.

Imaging and insulin pumps

Advice differs on what to do if a patient with an insulin pump needs to undergo radiographic imaging. For example, the University of Wisconsin radiology department says it is safe to keep an insulin pump in place if the x-ray beam will be on for less than 3 seconds at a time and if the device is covered by a lead apron.²⁵ However, radiation can induce electrical currents in the circuitry, which can alter the function of the pump. For this reason, some manufacturers recommend removing the device before the patient enters any room in which radiation or magnetic resonance imaging will be used.^26–31

Insulin pumps and surgery

Insulin pumps have been used in the perioperative and intraoperative periods, with positive outcomes.³² An analysis of 20 patients on pumps undergoing a total of 23 surgeries (mostly orthopedic procedures) found that 13 of the 20 patients wore their pump during surgery. No adverse events were noted in any of these cases, although the sample size was small.³³

Corney et al³⁴ retrospectively compared insulin pumps with alternative methods of perioperative glucose management. Multiple surgical specialties were included. No significant difference in mean blood glucose levels was found between those who continued to use their pump and those who used other methods. In those who continued to use their pump, there were no episodes of intraoperative technical difficulties related to the pump.

Any patient who may be undergoing a procedure or surgery must let the surgeon and anesthesiologist know that he or she has a pump. If the infusion site is too close to the site of the surgery or procedure, it must be moved.

Concerns during surgery include catheter or site disconnection or loss, crystallization within the tubing (a potential problem not limited to surgery), and pump malfunction. If the procedure involves imaging, the pump should probably be disconnected or covered by lead shielding as directed in the pump manufacturer’s manual. The surgeon and anesthesiologist must decide whether to continue use of a pump during a surgical procedure. However, the study by Corney et al³⁴ shows it is possible.

Most office-based procedures can be done with the insulin pump in place, as the patient is not under general anesthesia and so can adjust the insulin regimen as needed.

Abdelmalak et al,³⁵ in a comprehensive review of insulin pump use in noncardiac surgery, commented that the type of surgery may play a role in determining the best approach to perioperative glucose management. Major surgery causes a large inflammatory response that makes it difficult to control blood sugar, especially when steroids or beta agonists are given, whereas minor surgery does not affect blood glucose nearly as much. The authors offered recommendations on pump use during various surgical procedures depending on the length of the procedure:

• If surgery is anticipated to last less than 1 hour, then keep the insulin pump on, and have the patient manage corrections preoperatively and postoperatively.
• For surgery of intermediate length (1–3 hours), have the patient take a bolus of 1 hour’s worth of insulin (based on the basal rate for that time period) before the procedure, then remove the insulin pump. Do this only if blood sugar is normal or close to normal. If the patient is severely hyperglycemic, remove the insulin pump and start an intravenous insulin infusion.
• If the procedure will take more than 3 hours, remove the pump and start an insulin infusion regardless of the blood sugar level.³⁵

AIR TRAVEL AND INSULIN PUMPS

Insulin pumps can be easy to manage during airline travel if the user is prepared (Table 3).

First, it is important to have a letter from the treating physician stating that the pump is a necessary medical device. All supplies should be carried on and in a separate bag for easy inspection. The more forthcoming the user is at the security checkpoint, the easier the process.

According to the Transportation Security Administration, insulin pump users can keep their pump on during screening, and the metal detectors and full-body scanners will not harm the device.³⁶

However, manufacturer recommendations differ. Medtronic recommends that patients not expose their insulin pump to x-rays, and that instead of going through a full-body scanner the patient should request a pat-down.³⁷ Animas recommends the same.³⁸ OmniPod states that their system can be worn through airport imaging, making it the only approved continuous insulin delivery system that can be taken through airport imaging.³⁹

Another potential problem is the change in atmospheric pressure during takeoff and landing. Bubbles can form in the insulin reservoir as air pressure decreases with ascent, thereby displacing insulin from the pump to the patient. The opposite happens during descent. King et al⁴⁰ corroborated this phenomenon with Animas and Medtronic pumps. Asante recommends removing their pump tubing during takeoff and landing.³⁰

If PROBLEMS ARISE

Like any machine, an insulin pump can fail. Most failures result in lack of insulin delivery—the patient does not get excess insulin from insulin pump failure. Excess insulin delivery is most often due to operator error. All insulin is either preprogrammed (basal by provider or patient) or must be confirmed by the patient at the time of delivery (meal or correction boluses).

Pump manufacturers have 24-hour support programs and hotlines, with experts who will either walk the patient through the problem or send a replacement pump—often within 24 hours.

EVOLVING TECHNOLOGY

Pump technology is evolving quickly. On the way are “smart” pumps that interact with other systems, smaller pumps with advanced touch-screen features, and patch pumps that do not have tubing but operate similarly to pumps with tubing (ie, a cannula is still required for insulin delivery).

Some insulin pumps can be linked to an external glucose sensor. These systems provide a great amount of information to the patient and provider. Often, there is increased awareness of fluctuations in glucose, allowing earlier intervention to prevent high and low glucose excursions. Sensor-augmented pumps may further improve safety by suspending infusion during hypoglycemia.^41,42

Researchers continue to strive for closed-loop systems that would allow the pump to automatically respond to circulating glucose and thus provide truly physiologic control.⁴³ A recent study showed the effectiveness of the outpatient use of a bihormonal (insulin and glucagon) “bionic pancreas,” which provided improved glucose control and similar or less hypoglycemia in adults and adolescents who had been using a traditional insulin pump.⁴⁴