Pancreatectomy

Diabetes mellitus that develops after surgical removal of the pancreas is truly an insulin-dependent diabetes mellitus. Metabolically, it is characterized by insulin and glucagon deficiency.⁵ The magnitude of the hyperglycemia and of its characteristics depends on the quantity of pancreas removed (Table 17-3). Total or near-total pancreatectomy results in severe hyperglycemia, decreased plasma insulin, virtually absent plasma glucagon, and elevated plasma levels of gluconeogenic precursors (alanine, lactate, glycerol).^5–11

The effect of removal of 50% of the pancreas was studied in 28 normal transplant donors 1 year after surgery.¹² The donors lost a mean of 3.4 kg of body weight, and their mean fasting plasma glucose level had risen by 9 mg/dL (88 ± 7 to 97 ± 16 mg/dL). Similarly, their mean serum glucose concentration 2 hours after an oral glucose load was higher (117 ± 18 to 156 ± 53 mg/dL), and the area under the 5 hour plasma glucose curve after the oral glucose was 19.5% higher. Both the mean fasting plasma insulin concentration and the area under the 5 hour plasma insulin curve were significantly lower than preoperatively (−14% and −31%, respectively). None of the donors had any evidence of deficient pancreatic exocrine function. On further analysis, the investigators noted that 21 of the donors had no significant postoperative change in plasma glucose or insulin, whereas seven showed a marked increase in the entire 5 hour plasma glucose curve (either IGT nor diabetes) with no concomitant increase in 5 hour plasma insulin curves. The seven donors in whom some degree of hypoinsulinemia and hyperglycemia had developed did not have fasting hyperglycemia 1 year postoperatively. Two of the seven were studied from 2 to 7 years after surgery and had not had a further increase in fasting plasma glucose. A recent report of eight donor/recipient pairs evaluated 9 to 18 years after the original surgery indicated that the residual pancreatic mass is a significant determinant of long-term glucose homeostasis.¹³ However, other variables were implicated in the discordancy for the development of diabetes. In particular, obesity seemed to be a major factor in that all of the individuals (four donors and two recipients) who had developed diabetes were among the eight patients who were obese.¹³ The investigators interpreted their data as “suggesting that obesity should be a contraindication to donation of pancreatic segments and that donors should assiduously avoid becoming obese.”¹³

In a more recent follow-up of 15 normal individuals of 21 who had donated 50% of their pancreas between 1997 and 2003, two donors were taking oral diabetic medications, two had impaired fasting plasma glucose (IFG), one had impaired glucose tolerance (IGT), three had both IFG and IGT, and one met the criteria for the diagnosis of diabetes.¹⁴ Six of the 21 patients were lost to follow-up. Only 6 of 15 patients had normal glucose values. Despite the use of stringent criteria to exclude those at risk for developing abnormalities in glucose metabolism, 43% of the total population of healthy humans who underwent a 50% pancreatectomy developed some abnormality of glucose metabolism within 3 to 10 years.

Sun et al. studied the metabolic effects of removing 20% to 88% of the pancreas in dogs and found that no significant metabolic changes occurred until approximately 50% had been removed.¹⁵

From the data available, it seems reasonable to conclude that metabolic abnormalities that arise after pancreatectomy are likely to be clinically relevant at removal of 50% or more, and that a progressively greater number of metabolic abnormalities occur as the extent of pancreatectomy increases. The concomitant presence of insulin resistance further increases the likelihood of metabolic abnormalities.

Major characteristics of the development of diabetes mellitus after extensive pancreatectomy include an absence of glucagon secretion and marked impairment in insulin secretion. Absence of glucagon slows, but does not interfere with, the development of hyperglycemia and ketonemia after insulin withdrawal.^5,7 This observation indicates that glucagon is not necessary for development of the metabolic abnormalities of insulin-dependent diabetes mellitus. However, the absence of glucagon secretion does leave a pancreatectomized individual with diabetes at high risk for severe hypoglycemia during insulin treatment.^16–19 This situation is exaggerated by the associated nutritional deficiencies and weight loss that ordinarily accompany exocrine pancreatic insufficiency. A concomitant deficiency of pancreatic polypeptide may contribute to persistent hyperglycemia caused by impaired hepatic insulin action.²⁰ Treatment of an individual with pancreatic diabetes requires insulin, is associated with marked lability in glucose regulation, and is linked with an increased rate of both ketoacidosis and death from hypoglycemia. The development of autonomic neuropathy in a patient with pancreatic diabetes greatly adds to the risk of severe hypoglycemia with insulin treatment.²¹

Jethwa et al. have reported in a recent review that patients with total pancreatectomy can maintain reasonably good glycemic control with few serious hypoglycemic or ketoacidotic events provided they use self–blood glucose monitoring and are followed closely by their physicians.^22,23 Among the 33 patients available for a median follow-up of 50 months, the median hemoglobin A1c (HbA1C) was 8.2%, which was comparable with that in type 1 diabetic patients followed by their institution. Patients resected for cancer maintained better glycemic control than those treated for chronic pancreatitis. The median insulin dose for the entire group was 46 units per day.

Chronic Pancreatitis

Chronic pancreatitis accounts for a little less than 1% of cases of diabetes mellitus in Western countries and Japan.20,21,24–27 In tropical countries, where nonalcoholic calcific pancreatitis is common, the incidence may be somewhat higher, but reliable data are not available. Although long-term ingestion of alcohol is the most common cause of chronic pancreatitis (particularly in Western cultures), other conditions such as genetic mutations, pancreatic duct obstruction, hypertriglyceridemia, hypercalcemia, autoimmunity, calcific tropical pancreatitis, and idiopathic pancreatitis are not uncommon.²⁸ The development of diabetes mellitus in patients with chronic pancreatitis is most frequent in those with calcific disease (55% to 70%) and occurs less often in those with noncalcific disease (30%).²⁹ The prevalence of diabetes in patients with chronic pancreatitis increases with increasing duration of pancreatitis and with increasing exocrine deficiency.^26,30–34

The inflammatory response causes loss of exocrine tissue, extensive fibrosis, and distorted and blocked ducts. The islets of Langerhans are relatively resistant and undergo pathologic changes only late in the disease. Chronic pancreatitis is associated with loss of functioning β cells and a somewhat lesser loss of α cells.17,35–41 Hormonal alterations seen include a decrease in insulin secretion in response to nutrients, followed later by a decrease in fasting C peptide levels. Plasma C peptide levels rather than insulin levels may be a better means of assessment of insulin secretion because associated liver disease may change hepatic extraction rates of insulin. With progressive chronic pancreatitis, insulin secretion falls even lower. Glucagon secretion is impaired in moderate to severe chronic pancreatitis. Insulin resistance frequently develops in such individuals.

Diabetes mellitus is seen after several years of chronic pancreatitis. In an unselected series of patients with chronic pancreatitis, 35% had type 1 diabetes, 31% had type 2 diabetes or IGT, and 34% had normal glucose tolerance.⁴² The nature of the diabetes reflects the severity of the chronic pancreatitis. Mild pancreatitis may be associated only with IGT, whereas severe pancreatitis is associated primarily with insulin-dependent (type 1) diabetes. Patients with chronic pancreatitis and diabetes mellitus fail to secrete glucagon in response to hypoglycemia. If patients have concomitant autonomic neuropathy, they are extremely susceptible to severe and prolonged hypoglycemia.

Treatment of diabetes mellitus in patients with chronic pancreatitis should entail the use of small doses of short-acting or rapid-acting insulins to manage the hyperglycemia, replacement enzymes for the malnutrition and malabsorption, and elimination of the use of alcoholic beverages.29,39,43,44 Surgery⁴⁴ with subtotal resection or near-total resection may be necessary to relieve severe pain.^45,46 Following pancreatic surgery for pain or other complications, a significant incidence of diabetes mellitus occurs; however, whether diabetes develops is dependent on the type of surgical intervention used. Whipple procedures usually are associated with an incidence of 25% to 40%, a Frey or Berne procedure 8% to 22%, and a distal pancreatectomy approximately 60%.⁴⁷ In such patients, successful islet allotransplants and autotransplants in the liver have been able to maintain near normoglycemia.⁴⁸ The hepatic islet cell transplants were able to secrete insulin in response to nutrients but were unable to secrete glucagon in response to hypoglycemia.^18,19 A recent report of the results of pancreas allotransplants in patients who had undergone total pancreatectomy in the past showed a 70% survival rate and successful correction of both exocrine and endocrine deficiencies.⁴⁹

Pancreatic exocrine function has been reported to be reduced in some type 1 and type 2 diabetic patients. This has been explained as a complication of the diabetes. A retrospective analysis of pancreatograms of patients with known diabetes has suggested that chronic pancreatitis may be much more common as a cause of diabetes than was previously thought.⁴⁸ A total of 38 type 1 and 118 type 2 diabetic patients underwent endoscopic retrograde cholangiopancreatography (ERCP) studies for varying reasons. Pancreatic ducts were classified as normal in 23.3% and as exhibiting chronic pancreatitis degree I, II, and III in 22.7%, 32.7%, and 21.3%, respectively.⁵⁰ The investigators suggested that a substantial number of patients with primary diabetes mellitus may have a concomitant chronic pancreatitis, or perhaps that many cases of primary diabetes mellitus may be related to chronic pancreatitis.

Pancreatic Cancer

Diabetes mellitus is known to occur more frequently in patients with pancreatic cancer than in the general population.51–53 Published data indicate that as many as 70% of patients with pancreatic carcinoma have impaired glucose tolerance or frank diabetes mellitus, and that 60% demonstrate improved glucose metabolism after surgery.⁵⁴ Wakasugi et al. reported that 53.1% of patients with invasive ductal pancreatic carcinoma had diabetes mellitus, and in 45.9% it was thought to be secondary to the carcinoma.⁵⁵ The reasons for this association have been the subject of much speculation. Some studies show that diabetes mellitus is associated with increased risk for susceptibility to pancreatic cancer (2.15 to 4.9 in men).^56,57 Other studies have indicated that in most patients with pancreatic carcinoma, the diabetes is secondary to some effect of the cancer, which causes insulin resistance and impairs the function of normal β cells.^58–60 A multicenter case-control study of 720 patients with pancreatic cancer addressed this issue.⁵¹ The prevalence of diabetes in the patients with pancreatic cancer was 22.8%, whereas that in the matched control population was 8.3%. The patients with pancreatic cancer were characterized as having type 2 diabetes. A recent diagnosis of diabetes had been made in 40.2% of the patients with pancreatic cancer with diabetes, as contrasted with only 3.3% of the control population with diabetes (Table 17-4). A higher percentage of the control population with diabetes than of the pancreatic cancer population with diabetes had had their diabetes for longer than 15 years (see Table 17-4). These data support the idea that in a small number of patients, diabetes predisposes to the development of diabetes mellitus.

Possible mechanisms by which pancreatic cancer could contribute to the development of type 2 diabetes include the following: (1) destruction of islets, (2) impairment of the insulin secretory mechanism, (3) development of insulin resistance, and (4) tumor-related pancreatitis. Insulin and C-peptide measurements during oral glucose tolerance testing in patients with pancreatic cancer have shown abnormal β cell function with reduced plasma C-peptide responses in 50% of patients and increased plasma proinsulin-to-C-peptide ratios^61,62 Insulin resistance has been demonstrated in most patients with pancreatic carcinoma.^63–65 Morphometric studies of tumor-free regions of the pancreas have shown reduced β cell populations.⁶⁶ An inverse correlation was noted between the number of β cells and the fasting plasma glucose concentration. These data can be interpreted as indicating that pancreatic cancers produce substances or responses that destroy normal β cells.^64–67 The presence of diabetes in patients with pancreatic cancer predicts that the tumor is less likely to be respectable and the patient has a poorer prognosis than if diabetes is not present.⁵³ Patients with pancreatic cancer typically have type 2 diabetes; most have been treated with oral antihyperglycemic agents.^52,53

Several very intriguing observations that were reported recently make this relationship between pancreatic cancer and diabetes mellitus even more complex and confusing. Li et al. reported that metformin treatment of diabetic patients was associated with a decreased risk of developing pancreatic carcinoma, while insulin or insulin secretogogue treatment was associated with an increased risk.⁶⁸ In a large, multicenter clinical trial that compared the treatment of type 2 diabetic patients with rosiglitazone-based therapy versus metformin plus sulfonylurea therapy, the development of pancreatic cancer during the study was statistically less frequent in the rosiglitazone arm than in the active comparator arm (2 cases versus 13 cases; P < .007).⁶⁹

Hemochromatosis

Hemochromatosis is an autosomal recessive genetic disorder that results in excessive deposition of iron in parenchymal cells of the liver, pancreas, muscle, heart, anterior pituitary, and other organs.^70,71 The clinical diagnosis in the past was made by the findings of diabetes mellitus, hepatomegaly, and skin pigmentation. More recently, it has been recognized through biochemical and genetic testing.⁷² The first phenotypic expression of the disease is an elevation in serum transferrin saturation. This abnormality is followed by iron accumulation in the tissues and an elevation in the serum ferritin concentration. Early clinical findings are related to hepatic dysfunction and joint symptoms. Clinical diabetes mellitus and skin pigmentation occur relatively late in the course of the disease.⁷¹ A candidate gene for human leukocyte antigen (HLA)-linked hemochromatosis has been cloned and a mutation (C282Y) of the HFE gene identified that may account for 60% or more of cases of hereditary hemochromatosis. Another mutation (H63D) has been identified more recently.⁷³

Diabetes mellitus has been reported in 50% to 60% of patients with hemochromatosis.^71,74 Another 20% to 30% had glucose intolerance. These figures represent data from older series in which the diagnosis was made late in the course of the disease. Diabetes mellitus is more frequent in patients who have a family history of diabetes mellitus.

Follow-up of 237 patients who were given a diagnosis of hemochromatosis and treated at a single center from 1983 through 2005 has diabetes mellitus associated with the disease.⁷³ Before 1996, hemochromatosis was diagnosed by the classic clinical and laboratory features of the disease. Since 1996, genetic testing, which is now available commercially, made it possible for them to diagnose hemochromatosis in asymptomatic patients. Of 45 patients diagnosed before 1996, 30.2% had cirrhosis of the liver and 35.6% had diabetes mellitus. After 1996, 192 patients were given the diagnosis of hemochromatosis (34% by family screening and 40% by clinical suspicion) and only 7.5% had cirrhosis of the liver and 17.7% had diabetes mellitus at the time of diagnosis. It is obvious that early through genetic testing and effective treatment can significantly reduce development of the complications of hemochromatosis. Of note was the observation that performing phlebotomy and reducing tissue iron in well-established clinical disease did not improve diabetes or lessen its management requirements. Thus early diagnosis and treatment is primarily preventative.⁷³

Before 1996, hemochromatosis was diagnosed by the classic clinical and laboratory features of the disease. Since 1996, genetic testing, which is now available commercially, has made it possible to diagnose hemochromatosis in asymptomatic patients. Of the 45 patients diagnosed before 1996, 30.2% had cirrhosis of the liver and 35.6% had diabetes mellitus. After 1996, 192 patients were given the diagnosis of hemochromatosis (34% by family screening and 40% by clinical suspicion); only 7.5% had cirrhosis of the liver and 17.7% had diabetes at the time of diagnosis. It is obvious that early diagnosis through genetic testing and effective treatment have significantly reduced development of the complications. Of note is the observation that performing phlebotomy and reducing tissue iron did not improve diabetes or lessen its management requirements. Thus early diagnosis and treatment is primarily preventative.⁷³

The metabolic studies that have been done show that patients with hemochromatosis have marked insulin resistance. Histologic study of the pancreas shows iron deposits that are greatest in the acinar cells but do involve islet cells. Insulin secretion in response to glucose or arginine is decreased; however, glucagon secretory responses to arginine are increased and unaffected by glucose.^75,76 The data are compatible with a marked reduction in β cell function and no disturbance in α cell function. The hyperglycemia is a result of insulin resistance and decreased β cell function. The prevalence of diabetes mellitus probably could be reduced by early diagnosis of hemochromatosis and initiation of phlebotomy therapy.

Therapy for patients with hemochromatosis and clinical diabetes frequently requires insulin (40% to 50% of patients), although no systematic studies of therapy have been done.⁷² Reduction of tissue iron stores, although most beneficial in the early stages of disease, nonetheless can help improve glycemic control in 35% to 45% of patients.^70,71

Hemosiderosis

Excessive iron deposition occurs in a variety of conditions other than primary hemochromatosis. In thalassemia major, frequent blood transfusions are necessary and may lead to massive iron overload. The reported prevalence of diabetes mellitus in treated thalassemia major is about 16%. This figure is highly correlated with the number of blood transfusions given and the duration of the disease. The incidence of IGT is reported to be 60%.⁷⁷

Further evidence that deposits of excess tissue iron themselves are responsible for many of the metabolic abnormalities seen in hemochromatosis and thalassemia major comes from studies in rural male Bantus.⁷⁸ Many Bantus drink alcoholic beverages that are brewed in iron containers and ingest in excess of 100 mg of iron per day. In those individuals, the prevalence of diabetes mellitus is 10-fold higher than in non–alcoholic beverage–consuming males.

Mechanistic studies in patients with thalassemia major with normal, impaired, and diabetic glucose tolerance tests show that increased iron stores are associated with the development of insulin resistance and a delay in early insulin secretion.⁷⁹ A correlation between increased iron stores in normal women and the development of type 2 diabetes was demonstrated recently in the Nurses Health Study.⁸⁰

Cystic Fibrosis

Cystic fibrosis (CF) is a monogenetic disorder with abnormal cyclic adenosine monophosphate–regulated chloride channel activity. It is an autosomal recessive genetic disease with an incidence of 1 in 2500 live births in Caucasian populations. More than 1000 gene mutations have been identified in the CF gene.⁸¹ Organs as diverse as the lung, exocrine pancreas, large and small intestine, hepatobiliary system, and sweat glands are involved. Failure to secrete Na⁺, HCO₃⁻, and water leads to retention of enzymes in the pancreas and ultimately to destruction of pancreatic tissue.^82–84 Histologic examination of the pancreas in patients with CF shows fatty infiltration, necrosis, and fibrosis of the exocrine pancreas. Islet cell architecture is disrupted and the absolute number of pancreatic islets diminished. Islets that are present show significant decreases in β cells, α cells, and pancreatic polypeptide–producing cells and increases in δ (somatostatin-producing) cells. Islet amyloid deposits have been found in 69% of diabetic CF cases examined.

Diabetes mellitus requiring medical therapy (usually insulin) has been reported in 4.9% of patients of all ages with CF in a large European study of 1348 patients⁸⁵ and in 5.1% of 18,627 patients of all ages monitored at CF centers in the United States and Canada.⁸² Diabetes occurs more often in individuals who are homozygous for the most common CF mutation, ΔF508.^82,86,87 Diabetes also occurs with greater frequency with increasing age; it has been reported in 32% of Danish patients who were older than 25 years. Routine oral glucose tolerance testing suggests that of the total CF population aged 5 years or older, 35% have normal glucose tolerance, 37% have IGT, 17% have CF-related diabetes without fasting hyperglycemia, and 11% have CF-related diabetes with fasting hyperglycemia.

Several features of CF-related diabetes are noteworthy. Autoantibodies to pancreatic heat shock protein 60 have been found to precede the development of glucose intolerance (IGT and diabetes) and to decline subsequently with the onset of glucose intolerance.⁸⁸ The development of diabetes in patients with CF is characterized initially by abnormal oral glucose tolerance and a delay in oral glucose–stimulated insulin secretion, followed later by a decrease in total insulin, glucagon, and pancreatic polypeptide secretion.⁸⁹ First-phase insulin secretion after intravenous glucose is reduced markedly in patients with CF compared with matched controls. Patients with CF with IGT have normal plasma free fatty acid levels as compared with matched controls.⁹⁰ Their plasma tumor necrosis factor-α levels are elevated, and they have insulin resistance as measured by the hyperinsulinemic euglycemic clamp.⁹⁰ Decreased translocation of Glut-4 glucose transporters in muscle was observed during the peak insulin effect. Patients with CF have an increase in hepatic glucose production and are resistant to suppression of hepatic glucose production by insulin even in the nondiabetic state.⁹¹ Peripheral insulin sensitivity is increased in healthy nondiabetic individuals with CF, but insulin resistance occurs later as IGT and diabetes develop, and patients experience additional complications related to their CF.⁹² The development of diabetes worsens pulmonary function and other clinical manifestations of CF and may increase mortality by up to sixfold.⁹³ Insulin treatment appears to reduce the extent of this deterioration. Hyperglycemia in patients with CF can be intermittent or permanent. Intermittent hyperglycemia occurs with glucocorticoid therapy, infection, or stress and must be treated with insulin until it resolves. Permanent hyperglycemia is always treated with insulin.^82,84 It is likely that patients with CF progress from intermittent hyperglycemia to permanent hyperglycemia as pancreatic destruction continues to occur.

A small short-term study (12 weeks) suggests that treatment of CF-related diabetes with glargine insulin may provide some advantages over treatment with NPH insulin.⁹⁴

Microvascular complications have been observed in patients with CF-related diabetes with fasting hyperglycemia for longer than 10 years. Fourteen percent had microalbuminuria and 16% had retinopathy.⁹⁵ Macrovascular disease appears to be uncommon.

Acromegaly

Acromegaly is characterized by excessive and autonomous secretion of growth hormone and IGF-1.^96,97 The prevalence of overt diabetes mellitus reported in different series of acromegalic patients ranges from 30% to 56%.^96,98 IGT may be present in as many as 36% of acromegalic patients.⁹⁶ In a specific population, the percentage of acromegalic patients in whom diabetes mellitus will develop depends on the prevalence of predisposition to type 2 diabetes in the population and the magnitude of elevation of serum IGF-1 levels.

Elevated GH and IGF-1 levels cause excessive hepatic glucose production and impaired insulin-mediated muscle glucose uptake.99–101 This insulin resistance is correlated with circulating IGF-1 levels and has been demonstrated by the euglycemic hyperinsulinemic clamp and the minimal model techniques.

Reduction in circulating GH and IGF-1 levels by successful surgical removal of the tumor producing growth hormone or GH-releasing factor results in significant improvement in glycemic control in acromegalic patients with diabetes mellitus.^100,102 Recent data suggest that circulating GH must be lowered to 2 ng/L and IGF-1 lowered to the normal range to be considered curative.^97,103,104 Transsphenoidal surgery achieves growth hormone levels less than 5 ng/L in approximately 60% of patients. Curative levels are attained in about 70% of patients with microadenomas (<10 mm in diameter) but are attained considerably less often in those with macroadenomas of the pituitary.^97,103

Use of the somatostatin analogue octreotide to treat acromegaly as the primary medical therapy or to supplement prior inadequate surgical treatment or radiotherapy has allowed attainment of greater and more consistent reductions in circulating GH and IGF-1 levels (GH levels ≤5 ng/L in 65% and ≤2 ng/L in 40%, and IGF-1 levels in the normal range in 64% of patients).103–105

Treatment of acromegalic patients with octreotide presents several issues with respect to glucose metabolism.103–106 Reductions in circulating GH and IGF-1 levels will decrease insulin resistance and should lead to improvement in glycemic control in subjects with diabetes mellitus or IGT. However, pharmacologic doses of a somatostatin analogue also reduce insulin secretion (decreased insulinogenic index), and such a reduction should cause a deterioration in glucose tolerance. Thus in any particular patient, octreotide therapy will modify glucose metabolism in accordance with these competing effects. Approximately two thirds of acromegalic patients with diabetes mellitus are treated with insulin and one third with oral hypoglycemic agents. Octreotide treatment in patients with diabetes mellitus and acromegaly frequently leads to improvement in glycemic control as measured by a reduction in the insulin dose, conversion from insulin therapy to oral hypoglycemic agent therapy, or conversion from oral hypoglycemic agent therapy to dietary management.^103,106 Some patients (those with more severe insulin deficiency), however, will have significant deterioration in glycemic control.^104,106 When higher doses of octreotide are given, IGT and even frank diabetes mellitus (as high as 20% and 29%, respectively) may develop in acromegalic patients with normal glucose tolerance before octreotide treatment.¹⁰⁶ An alternative to treatment with somatostatin analogues is treatment with the GH receptor blocker Pegvisomant. This agent will decrease IGF-1 levels and can improve glycemic control.^107,108 However, it does cause an increase in visceral fat and occasionally is associated with increased liver enzymes.¹⁰⁹

Appropriate treatment for acromegaly is necessary to reduce the increased mortality that has been seen in the past.¹¹⁰