Islet Amyloid Polypeptide in Patients with Pancreatic Cancer and Diabetes

Patients and Tissue Studies

Table 1. Table 1. Characteristics of the Normal Subjects, Patients with Different Types of Cancer, and Patients with Diabetes.

We studied 30 patients with histologically or cytologically confirmed carcinoma of the pancreas. Seven of the patients were studied again three months after subtotal pancreatectomy that removed 80 to 85 percent of the organ⁷. We also studied 25 normal subjects of similar age and sex, none of whom had a family history of diabetes; 5 patients with insulin-dependent diabetes mellitus (IDDM); 18 patients with NIDDM; and 46 patients with cancers other than pancreatic cancer (Table 1). All the patients with cancer were candidates for radical curative surgery, and none had a large tumor burden. In all patients with pancreatic cancer who had diabetes, diabetes mellitus had been diagnosed no more than two years before the diagnosis of cancer. All the patients with pancreatic cancer were cachectic, with weight loss in the three months before diagnosis ranging from 6.8 to 8.7 percent. For tissue studies, we obtained samples of tumor tissue and adjacent pancreatic tissue from the operative specimens of 7 patients with pancreatic cancer and samples of pancreatic tissue from 15 cadaveric organ donors.

We obtained any history of diabetes and performed an oral glucose-tolerance test in all patients with pancreatic cancer. The patients were divided into subgroups according to the criteria of the World Health Organization for impaired glucose tolerance and diabetes¹³.

Blood samples were obtained from all subjects after an overnight fast and were collected into ice-cold tubes containing aprotinin (400 kallikrein inhibitory units per milliliter of blood) and EDTA (5 mg per milliliter of blood). The plasma was immediately separated in a refrigerated centrifuge. The samples of tumor and adjacent pancreatic tissue from the patients with pancreatic cancer and the samples of pancreatic tissue from the organ donors were immediately frozen at -70 °C for subsequent extraction and radioimmunoassay. Samples for immunohistochemical evaluation and in situ hybridization were taken from the pancreas of each organ donor and from the tumor and adjacent pancreatic tissue immediately after resection in the patients undergoing pancreatectomy. These tissue samples were fixed for about six hours in buffered 10 percent formalin and embedded in paraffin.

These studies were approved by the review committee of the University of Linkoping, and informed consent was obtained from all study subjects.

Radioimmunoassay for IAPP

IAPP concentrations were measured in extracts of plasma by radioimmunoassay in duplicate, with use of a commercial antiserum (Peninsula Laboratories, Belmont, Calif.) and IAPP labeled with the Bolton and Hunter reagent (Amersham, Arlington Heights, Ill.) and purified by reverse-phase high-performance liquid chromatography. The sensitivity of the assay was 0.4 fmol per tube (equivalent to 2.0 pmol per liter of plasma). The intraassay coefficient of variation was less than 6 percent, and the interassay coefficient of variation was less than 10 percent.

Plasma samples for the measurement of IAPP were extracted individually on C-18 reverse-phase Sep-Pak cartridges (Waters, Milford, Mass.) by means of multichannel syringe ram pumps (Harvard Instruments, Cambridge, Mass.). The cartridges were washed and then subjected to elution with 3 ml of 60 percent acetonitrile containing 0.1 percent trifluoroacetic acid. The eluates were lyophilized and reconstituted in buffer (60 mM phosphate, pH 7.4, with 0.1 percent Triton-X 100 and 0.1 percent bovine albumin) for assay. Extraction recovery was determined in each assay by adding 0, 5, or 20 fmol of IAPP per milliliter to plasma samples from 10 normal subjects before extraction. The measured values were compared with the expected values to determine the percentage of recovery, which ranged from 60 to 68 percent. The IAPP concentrations in the experimental samples in each assay were corrected in relation to the mean percentage of recovery for that assay. The samples from all patients with cancer and the normal subjects were analyzed simultaneously, along with the samples in which recovery was determined.

Pieces of tumor and pancreatic tissue were weighed while still frozen and then placed in boiling 0.5 M acetic acid (10 ml per gram, wet weight) for 10 minutes. The acid extracts were centrifuged and assayed in duplicate at each of three dilutions in a single assay. The percentage of recovery, calculated after the addition of 10 pmol of IAPP per milliliter to aliquots of extracts of pancreatic tissue from eight organ donors, ranged from 88 to 103 percent.

Plasma C-peptide concentrations were determined as previously described¹⁴.

Immunohistochemical Evaluation

Staining for both IAPP and insulin was carried out in all tissue samples, with rabbit antiserum AA116 raised against human IAPP and guinea pig antiserum Ma37 raised against porcine insulin¹⁵ and with peroxidase-conjugated avidin, biotinylated porcine antirabbit immunoglobulins, porcine antirabbit immunoglobulins, and horseradish peroxidase-antiperoxidase complex (Dakopatts, Copenhagen, Denmark). IAPP immunoreactivity was determined with the avidin-biotin method, and insulin immunoreactivity with the peroxidase-antiperoxidase method¹⁶. In each case, deparaffinized sections were incubated overnight at room temperature with the primary antiserum, diluted 1:200 to 1:1600, followed by biotinylated antibody or the secondary antiserum-peroxidase complex as appropriate. For IAPP staining, the sections were developed with diaminobenzidine. Control procedures included the use of normal nonimmune serum and absorption of antibody with antigen.

In Situ Hybridization

In situ hybridization of IAPP mRNA to a digoxin-labeled riboprobe containing the full-length coding region of IAPP was performed on deparaffinized sections of all tissue samples, as previously described¹⁵. The hybridization was visualized by an immunohistochemical method¹⁵. Control procedures were run with an IAPP sense probe in all tissues.

Statistical Analysis

The results of the plasma and tissue IAPP measurements in the study groups were compared by Wilcoxon's signed rank test. Values obtained before pancreatectomy were compared with those obtained after surgery with the use of paired analysis (Wilcoxon's test). All statistical tests used were two-tailed. Results are expressed as means ±SD unless otherwise indicated.

Table 2. Table 2. Characteristics of the Patients with Pancreatic Cancer, According to Their Status for Glucose Tolerance and Diabetes.

Eighteen of the 30 patients with pancreatic cancer (60 percent) had diabetes, and 3 others (10 percent) had abnormal glucose tolerance. The patients with pancreatic cancer were divided into four groups: those with normal glucose tolerance, those with impaired glucose tolerance, those with non-insulin-requiring diabetes, and those with insulin-requiring diabetes (Table 2). Of the seven patients with resectable tumors, six had diabetes before surgery (four required insulin). Postoperatively, glucose tolerance improved in these six patients: two patients had normal glucose tolerance, two no longer required insulin, and two required less than half the amount of insulin needed preoperatively. These improvements occurred in spite of a reduction in plasma C-peptide concentrations in all six patients. The seventh patient had normal glucose tolerance both before and after operation.

Table 3. Table 3. Fasting Plasma IAPP and C-Peptide Concentrations and the IAPP:C-Peptide Ratio in Normal Subjects and Subgroups of Patients with Pancreatic Cancer.

Figure 1. Figure 1. Fasting Plasma Concentrations of IAPP in Seven Patients with Pancreatic Cancer, before and Three Months after Subtotal Pancreatectomy.

IAPP concentrations were significantly decreased after the operation (P = 0.01). The solid circles denote patients with diabetes, the open circles patients with normal glucose tolerance, and the diamonds means ±SE.

Fasting plasma IAPP concentrations were higher in the 30 patients with pancreatic cancer than in the normal subjects (Table 1). In contrast, the plasma IAPP concentrations in the patients with other cancers were similar to those in normal subjects. The higher values in the patients with pancreatic cancer were due to increased concentrations in the two subgroups with diabetes (31.4 ±12.6 pmol per liter in patients with non-insulin-requiring diabetes and 25.0 ±8.7 pmol per liter in patients with insulin-requiring diabetes) (Table 3). The plasma IAPP concentrations were also elevated in the nondiabetic patients with pancreatic cancer. The plasma IAPP concentrations became normal in all seven patients studied before and after surgery (from 28.9 ±16.4 to 5.6 ±3.4 pmol per liter) (Figure 1). The postoperative values were similar in the patients who had normal glucose tolerance and those who had diabetes preoperatively.

The ratio of IAPP to C peptide in plasma from patients with pancreatic cancer was significantly higher than the ratio in plasma from the normal subjects, especially in the subgroups with diabetes (Table 3). In patients with other cancers, this ratio was similar to that in normal subjects (data not shown). In the seven patients with pancreatic cancer who were studied after surgery, the ratio decreased to a value similar to that in the normal subjects (Table 3).

The plasma IAPP concentrations in the patients who had IDDM but not pancreatic cancer were significantly lower than those in the normal subjects (Table 1). Among the patients with NIDDM only, those receiving insulin had lower plasma IAPP concentrations, whereas those receiving an oral hypoglycemic drug had normal concentrations.

Figure 2. Figure 2. Concentrations of IAPP in Pancreatic Tumor and Adjacent Pancreatic Tissue in Operative Specimens from 7 Patients and Normal Pancreatic Tissue from 15 Cadaveric Organ Donors.

IAPP concentrations in normal pancreatic tissue were significantly higher than those in pancreatic tumor (P<0.001) or the tissue adjacent to the cancer (P = 0.002). The diamonds denote means ±SE.

The extracts from pancreatic tumors contained immunoreactive IAPP, but the content varied among samples. In most cancers, the IAPP content was lower than in pancreatic tissue adjacent to the cancer (Figure 2). The IAPP content of the normal pancreatic tissue from the organ donors was significantly higher than that of the cancerous tissue (183 ±129 vs. 17 ±16 pmol per gram, P<0.001) and pancreatic tissue adjacent to the cancer (43 ±42 pmol per gram, P = 0.002).

Figure 3. Figure 3. Immunochemical Studies of Pancreatic Tissue in a Patient with Pancreatic Cancer and Diabetes.

Consecutive sections of a normal pancreatic islet contained beta cells immunostaining (granular brown material) for insulin (Panel A) and IAPP (Panel B) (both panels × 250). Sections of another normal pancreatic islet showed immunostaining for IAPP (Panel C) and in situ hybridization of IAPP mRNA (Panel D) (both panels × 210); note the similar distribution of reaction products. An islet adjacent to the pancreatic cancer showed very weak immunostaining for IAPP (Panel E) in spite of strong labeling for IAPP mRNA (Panel F) (both panels × 290).

No specific staining for IAPP was seen in tumor cells or cells closely related to tumor epithelium in any patient. However, the islets surrounded by tumor tissue or immediately adjacent to the tumors did not appear normal. Although the intensity of insulin staining of islets in these two locations was usually normal or only slightly reduced, staining for IAPP was greatly reduced or even absent in most instances (Figure 3). In contrast, the intensity of IAPP staining of islets distant from the tumors in all the patients with pancreatic cancer was normal or only slightly reduced.

Strong labeling for IAPP mRNA on in situ hybridization was seen both in islets in normal pancreatic tissue and in islets adjacent to tumor tissue that were negative for IAPP on immunohistochemical evaluation (Figure 3). These IAPP mRNA-labeled cells and the cells that stained with antibodies to insulin had a similar distribution. The islets in the control sections incubated with the IAPP sense probe showed no labeling.

The plasma IAPP concentrations in the normal subjects and the patients with IDDM or NIDDM were similar to those measured by other investigators using the same antibody^17,18. In contrast, the values in the patients with both pancreatic cancer and insulin-requiring or non-insulin-requiring diabetes were increased. The elevated ratio of IAPP to C peptide in these two groups indicated that, as compared with the normal subjects or the patients with IDDM or NIDDM alone, these patients were secreting relatively more IAPP than insulin. This increase was most pronounced in the patients with pancreatic cancer and insulin-requiring diabetes. An increased ratio of IAPP to C peptide has not been reported in other groups of diabetic patients^17,18. Insulin and IAPP are normally secreted simultaneously at a relatively fixed ratio, but IAPP production may be controlled independently of insulin production in transformed cells and in normal cells under certain experimental conditions^12,19,20.

Endocrine cells have been found in ductal pancreatic adenocarcinomas,⁸ indicating that tumors may produce polypeptide hormones. However, we found no cells containing IAPP or insulin in tumor tissue. This finding, together with the low concentrations of IAPP measured by radioimmunoassay in tumor extracts, suggests that tumor production of IAPP does not explain the increased plasma concentrations of IAPP in patients with pancreatic cancer. The weak IAPP immunoreactivity or absence of immunoreactivity in islets adjacent to the tumors resembled immunoreactivity in the islets of patients with NIDDM²¹ and in age-related diabetes in cats,²² conditions associated with the deposition of large amounts of IAPP as amyloid in islets. This low immunoreactivity may reflect increased IAPP secretion like that in diabetic obese mice (unpublished data), because the in situ hybridization studies suggest that IAPP transcription is normal in islets adjacent to the tumor. An increase in secretion would also account for the elevated plasma IAPP concentrations. Although the high plasma IAPP concentrations could have resulted from decreased degradation of IAPP, the results of the in situ hybridization studies and the decline after subtotal pancreatectomy make increased secretion the more likely explanation. The normalization of circulating levels of IAPP after surgery suggests normal secretion by the remaining beta cells.

The finding of a high frequency of diabetes and impaired glucose tolerance among patients with pancreatic cancer confirms the results of other investigations^1,2. The diabetes of patients with pancreatic cancer is in some respects similar to NIDDM in obese patients because both types of diabetes are characterized by hyperinsulinemia and peripheral insulin resistance²³. Although the plasma IAPP concentrations were elevated in almost all diabetic patients with pancreatic cancer whom we studied, obese patients with NIDDM have IAPP concentrations similar to those of the patients with pancreatic cancer and normal glucose tolerance -- i.e., within the upper part of the normal range¹⁷. However, most patients with pancreatic cancer, like those in the present study, are underweight. Therefore, some factor other than obesity must cause the insulin resistance and diabetes in these patients. Because both glucose tolerance and insulin sensitivity improve after tumor removal, in spite of a decrease in insulin production, it appears that insulin resistance is closely related to the tumor⁷.

IAPP inhibits glucose uptake and glycogen synthesis in skeletal muscle in vitro and in vivo^9,11 and in the liver in vivo²⁴. In animals, IAPP can cause impaired glucose tolerance¹¹. The plasma concentrations of IAPP that have diabetogenic effects in vivo and in vitro are a matter of controversy,^11,25-28 and there is disagreement about which tissue responds most to IAPP^11,24,27,28. Although the effects of IAPP on peripheral metabolism usually require unphysiologically high concentrations,^25,26 concentrations as low as 1 to 10 pmol per liter have some metabolic effects^27,28.

Another biologic effect of IAPP is the inhibition of food intake²⁹. The anorexia induced by IAPP is accompanied by alterations in neurotransmitter metabolism similar to those in anorexic animals with tumors²⁹. We found that infusing IAPP into rats resulted in a 48 percent decrease in food intake in a 72-hour period; at the end of this time the animals given IAPP were 7.2 percent lighter than control rats³⁰. It is tempting, therefore, to speculate that the high plasma IAPP concentrations in patients with pancreatic cancer may contribute to their cachectic state, as well as to the insulin resistance that causes their diabetes.

Measurements of plasma IAPP could prove valuable in the diagnosis of pancreatic cancer in the small subgroup of patients with newly diagnosed NIDDM who present with weight loss rather than obesity. The usefulness of the test needs to be documented by studies of patients with newly diagnosed NIDDM, because plasma IAPP concentrations may be higher in this group than in patients with NIDDM of longer duration like those we studied. In the majority of patients with pancreatic cancer who have diabetes, the diabetes is diagnosed within the two years preceding the diagnosis of cancer. No early screening tests are currently available for detecting pancreatic cancer. Although not all patients with pancreatic cancer have diabetes, a test that would help identify even a fraction of patients with this cancer could be useful.