Consequences on β-Cell Function and Reserve After Long-Term Pancreas Transplantation

This work reports data from 102 individuals (63 pancreas recipients and 39 control subjects). The recipients all received their transplants at the University of Minnesota, but most lived in other states. The 153 successful recipients were asked to participate in these studies, and those who had failed transplants were not studied. Although the need to travel to one of the two study sites was a major disincentive, 63 recipients agreed to be studied on one occasion and 21 agreed to return for repeat studies. All patients had either received a whole pancreas from a brain-dead, heart-beating donor or a 50% segment of pancreas from living family members for at least 1 year and for as long as 22 years before the studies. Some patients had also received a kidney transplant simultaneously with or before the PTX. The studies were performed between 1994 and 2003. Recipients and control subjects were studied at the University of Minnesota General Clinical Research Center in Minneapolis and at the University of Washington General Clinical Research Center in Seattle. All clinical studies and laboratory assays were performed in the laboratory of the same investigator (R.P.R.). In some cases, the costs of transportation and lodging were paid for; however, no stipends were paid to the patients for participating. The experimental protocol was approved by the Institutional Review Boards of the University of Minnesota and the University of Washington. Insulin secretory reserve data from 16 patients have previously appeared in a small cross-sectional study (8), and acute insulin response data have appeared in other small cross-sectional studies (9,10); however, no cumulative analysis of the entire series of patients with emphasis on longitudinal studies (the focus of this work) has previously been published. The subgroup of 21 patients who were studied longitudinally successfully maintained a PTX for at least 2 and up to 22 years. These patients were studied at least twice and up to four times after transplantation. None of the patients were using insulin or other hypoglycemic agents for glucose control. The control subjects were sex-, age-, and BMI-matched to some of the pancreas recipients and were paid a modest stipend, which was approved by the two Institutional Review Boards.

All studies were preceded by an overnight fast from food and caloric drink and were performed fasting at bed rest the following day between 7:00 a.m. and noon. Intravenous cannulae were inserted into peripheral veins in both arms for the purposes of infusing fluids and drawing blood samples. Baseline blood samples were drawn for glucose, HbA_1c, and insulin. Arginine-induced insulin secretion was assessed by injecting at time 0 an intravenous pulse of arginine, 5 g in 40 ml saline, and samples were drawn 2, 3, 4, 5, 7, 10, 25, and 30 min following the injection (Fig. 1). Immediately thereafter, glucose-induced insulin secretion and glucose disappearance rates were assessed by injecting an intravenous pulse of glucose, 20 g in 40 ml saline, after which samples were collected at 3, 4, 5, 7, 10, 15, 20, 25, 30, 115, and 120 min. Insulin secretory reserve was assessed by the method of glucose potentiation of arginine-induced insulin secretion (1–3). This technique involves a 67-min infusion of a glucose solution (900 mg/min), a rate previously determined to maximally potentiate the insulin response to nonglucose stimuli, with arginine pulse given at 60 min and sample collections at 50, 55, 60, 62, 63, 64, 65, and 67 min (1–3). Glucose levels were determined by an automated glucose analyzer (11), HbA_1c levels were determined by high-performance liquid chromatography, and insulin and C-peptide levels were determined by radioimmunoassay (2).

AIR_arg was calculated as the mean of the three highest insulin levels obtained within 5 min after the arginine pulse less the prestimulus insulin level. AIR_gluc was calculated as the mean of the insulin levels obtained 3, 4, and 5 min after the glucose pulse less the prestimulus level. AIR_argmax, the measure of GPAIS, was calculated as the mean of the three highest insulin levels obtained within 5 min after the second arginine pulse less the prestimulus insulin level after 50–60 min of glucose infusion. Similar calculations were made for the acute C-peptide response to arginine (ACR_arg), the acute C-peptide response to glucose (ACR_gluc), and the acute C-peptide response to GPAIS (ACR_argmax). K_g was calculated as the slope of the linear correlation between time and the natural log of the glucose concentration obtained at 10, 15, 20, 25, and 30 min following the pulse of intravenous glucose. Although K_g is traditionally termed a glucose disappearance rate, it is recognized that it actually results from a combination of glucose disappearance and endogenous glucose production. All data are expressed as means ± SE. Statistical analyses were performed by Student’s t tests, by ANOVA, and by Pearson product linear correlations. Significance was ascribed to P < 0.05.

Of the 63 recipients studied, 13 were transplanted with segmental pancreata from living, related donors and 50 with whole pancreata from brain-dead, heart-beating donors (Table 1). On average the segmental recipients had been transplanted twice as long (P < 0.001), but there were no significant differences between the two groups of recipients in age, BMI, FPG, K_g, fasting insulin, or fasting C-peptide levels. The whole PTX group, but not the segmental PTX group, had significantly higher fasting insulin and fasting C-peptide levels than the control group (both P < 0.001), but no significant differences were found in BMI, FPG, or K_g.

An inverse relationship was observed within the segmental and whole PTX groups between the FPG level and the magnitude of the AIR_gluc (Table 2) (Fig. 2). To place into context the reciprocal relationship between AIR_gluc and FPG, the AIR_gluc responses in the two transplant groups were compared with those of the control group. This comparison required expression of the AIR_gluc data as a function of fasting insulin because the transplanted pancreata were drained by the systemic venous circuit, thereby bypassing first-pass hepatic degradation of insulin and causing hyperinsulinemia (12), which is in contrast to the control group, whose insulin secretion was delivered into the hepatic portal vein before appearance in the systemic venous circulation. This calculation is justified by the statistically significant correlations between AIR_gluc and fasting insulin for both groups (r = 0.43, P < 0.001 and r = 0.56, P < 0.001 for the transplant and control groups, respectively). The AIR_gluc in the control group fell within the distribution of values of the transplant recipients (Fig. 3). AIR_gluc values became 0 at FPG levels >115 mg/dl. However, those recipients with no AIR_gluc still demonstrated an AIR_arg, although the AIR_arg also diminished progressively as FPG values rose (Table 2). To assess whether differences existed between the segmental and the whole PTX groups, recipients within a narrower FPG range were segregated to avoid outliers (four in each group) and to create a more homogeneous group. In this analysis, the segmental PTX group had secretion values that were ∼50% smaller and statistically significantly different from the whole PTX group for AIR_gluc, AIR_arg, ACR_gluc, ACR_arg, and AIR_argmax (Table 3). The level of glycemia reached by the 60th minute of the 1-h glucose infusion during the GPAIS portion of the studies did not differ between the segmental and whole PTX groups (segmental, 330 ± 31, and whole, 292± 14 mg/dl; mean ± SE; P = NS).

For the segmental PTX group, highly statistically significant correlations were observed between AIR_arg and AIR_argmax, AIR_gluc and AIR_argmax, and AIR_arg and AIR_gluc (Fig. 4). Lesser degrees of statistical significance were observed for the relationships between ACR_arg and ACR_argmax, ACR_gluc and ACR_argmax, and ACR_arg and ACR_gluc in the segmental PTX group (Table 4). In the recipients of whole pancreata, highly statistical correlations were found between all relationships except ACR_arg and fasting C-peptide (Fig. 4) (Table 4).

Of the 21 recipients studied longitudinally, 11 were men and 10 were women (Table 5). Six received segmental PTX, and 15 received whole PTX. Six of the procedures involved pancreas transplantation alone, 11 involved simultaneous pancreas and kidney transplantation, and 4 involved pancreas after kidney transplantation. The average age of the recipients at the time of the procedure was 35 ± 1 years, and the average length of follow-up was 10 ± 1 years, with a maximum of 22 years. The most frequent immunosuppressive drug regimen was cyclosporine, azathioprine, and prednisone (15 of 21 patients). Fasting glucose levels remained generally stable and in the normal range, with only one recipient developing a level >126 mg/dl by 22 years after segmental PTX (Fig. 5A). No significant trend was observed when the data were expressed as change from initial value (Fig. 5B). AIR_gluc and AIR_arg tended to be lower in the recipients who had been transplanted the longest (Figs. 6 and 7). These patients were also the recipients who tended to have segmental rather than whole PTX. When these data were expressed as change in AIR_gluc and change in AIR_arg from the initial value obtained, the trends toward smaller values over time were not statistically significant for AIR_arg (P = 0.0546) (Fig. 6B) but were for AIR_gluc (P < 0.01) (Fig. 7B). A similar trend toward lesser values in AIR_argmax over time was observed (Fig. 8A), but when the data were examined as change from the initial value, no statistical difference was observed (Fig. 8B). The levels of glycemia in the final studies at the 60th minute of the 1-h glucose infusion were 296 ± 13 mg/dl and showed no significant trend upwards or downwards from the initial studies. Glucose disappearance rates after intravenous glucose injection generally were in the normal range, although seven recipients had levels slightly below normal on at least one occasion (Fig. 9A). When the K_g data were expressed as change from the initial values, no significant trend was observed (Fig. 9B). HbA_1c values were generally stable and within the normal range, although five recipients had slightly elevated values on at least one occasion (Fig. 10A). When the HbA_1c data were expressed as change from the initial values, no statistical trends toward higher values were observed (Fig. 10B).

This report provides both cross-sectional and long-term, longitudinal metabolic data from a large cohort of PTX patients studied for up to 22 years posttransplantation. The cross-sectional study provides novel information indicating that the physiologic reciprocal relationship between the magnitude of the AIR_gluc and the fasting glucose level normally observed in diabetic and nondiabetic humans (13) is maintained in successful recipients of PTXs. This is of interest because insulin from transplanted pancreata is delivered directly into the systemic venous circulation, thereby bypassing first-pass hepatic degradation, which leads to hyperinsulinemia in both the fasting and stimulated states (11). It might be anticipated that this could lead to hypoglycemia, and indeed mild postprandial hypoglycemia has been documented in some cases (14–16). However, this study demonstrates that the normal relationship between levels of glycemia and insulin secretion is generally maintained after pancreas transplantation and explains why hyperinsulinism in PTX recipients does not usually lead to hypoglycemia. Restored glucagon responses to hypoglycemia clearly also plays an important role in preventing hypoglycemia (17,18). This study also documents that the deterioration in insulin responses to glucose that accompanies development of higher fasting glucose levels is accompanied parri passu by deterioration in insulin responses to arginine, as is the case in patients with type 2 diabetes (1). The finding that both of these responses are diminished by ∼50% in recipients of segmental PTXs compared with recipients of whole organs is consistent with previous reports (4,5) suggesting that β-cell mass is an important determinant of the magnitude of glucose- and non–glucose-stimulated insulin responses. While seemingly intuitively self-evident, this need not have been the outcome because the former procedure uses fresh tissue obtained shortly before transplantation, whereas the latter involves procurement of a pancreas from a brain-dead donor on life-support measures and transportation of the pancreas from distant geographic sites. At a more pragmatic level, the highly statistically significant correlations between AIR_arg and AIR_argmax, as well as AIR_gluc and AIR_argmax, demonstrate that either measure can be used as accurate surrogates for GPAIS, the much more labor-intensive measurement of insulin secretory reserve. In certain clinical situations, such as the period immediately following pancreatic islet transplantation, when it may be desirable to avoid hyperglycemia, AIR_arg may be preferred to AIR_gluc as a test of insulin secretory reserve. In other cases, AIR_gluc will be preferred because intravenous preparations of glucose are less expensive and more readily available. Generally, C-peptide measurements provided the same information as insulin measurements, so they can be used instead of insulin measurements in partially successful recipients who receive small doses of insulin to manage glycemia.

The longitudinal study of segmental and whole PTX recipients was designed to ascertain whether the passage of time posttransplantation would be accompanied with steady deterioration in glycemic control and in β-cell function. The novelty and strength of this study lay in its exclusive use of paired data from 21 individual recipients studied for an average of 10 years and for up to 22 years posttransplant. A previous long-term study (8) reported paired data for FPG only, and not for AIR_arg, AIR_gluc, AIR_argmax, K_g, or HbA_1c and, therefore, did not provide an assessment of the stability of β-cell function over time posttransplantation. The results herein demonstrate remarkable stability in levels of FPG and intravenous glucose disappearance rates despite the diminution of AIR_gluc. This result emphasizes the sensitivity of AIR_gluc as a marker for altered β-cell responsivity because AIR_arg, AIR_argmax, intravenous glucose tolerance, and HbA_1c did not change significantly.

Figures 5–10 portray islet function in individual subjects studied more than once, but not every year, during the years portrayed. A limitation of these data is that they are therefore only longitudinal to a limited extent. Over the years studied, there were substantial changes in case selection, pancreas procurement, pancreas transplantation technique, immunosuppression, and graft survival. Successful subjects shown functioning 15 years after transplantation, for example, are not representative of cohort islet function 15 years after transplantation, some members of which have subsequently experienced loss of graft function through chronic rejection. Since subjects whose graft function was lost were excluded from this analysis, the rate of decline of islet function for a given year’s cohort shown in these studies is unavoidably higher than shown. This bias is inevitable because studies could not be carried out in subjects with graft failure. Moreover, islet function in subjects transplanted 15 years ago may not predict long-term islet function in subjects transplanted more recently. With improved organ procurement, surgical techniques and less toxic immunosuppression, it is reasonable to predict that the decline in islet function may not be as marked within individual subjects.

This information from PTX recipients can be used in a broader context to assess physiologic relationships among acute insulin secretory responses, insulin secretory reserve, and β-cell mass. In the cross-sectional studies, the parallel declines in AIR_gluc and AIR_arg as FPG levels rose may seem intuitively predictable, but previously reported studies have suggested otherwise. Preservation of acute insulin secretory responses to intravenous nonglucose secretagogues when the response to intravenous glucose is absent is well established from studies in type 2 diabetic patients using isoproterenol, secretin, and arginine as intravenous agonists (1,19,20). Yet, disagreement exists in the literature regarding the sensitivity with which responses to glucose and arginine reflect diminution of β-cell mass. Ward et al. (4) compared these two agonists in partially pancreatectomized dogs and concluded that the β-cell loss was underestimated by the acute insulin response to glucose. In that study, after partial pancreatectomy the responses to arginine, but not to glucose, declined dramatically when the dogs were normoglycemic and even more so during hyperglycemia induced by an intravenous glucose infusion. In contrast, McCulloch et al. (5) reported just the opposite from studies of streptozotocin-induced β-cell loss in baboons, i.e., a proportionately greater reduction in insulin responses to glucose compared with arginine with mild β-cell loss. At more severe levels of β-cell loss due to streptozotocin, i.e., when there was only 40–50% of the original β-cell mass detectable histologically, insulin responses to glucose were completely lost, whereas responses to arginine were clearly present. The present cross-sectional studies in humans, in which AIR_gluc reflected the degree of hyperglycemia in both segmental and whole PTX recipients just as sensitively as AIR_arg, agree more, although not entirely, with the findings of McCulloch et al. (5) as well as with published data from studies (1) of patients with type 2 diabetic patients. This is clearly demonstrable by the comparison of data from segmental versus whole PTX recipients because both AIR_gluc and AIR_arg were similarly reduced in the segmental PTX group. The latter comparisons also support the validity of AIR_argmax as a measure of insulin secretory reserve because the recipients of the 50% segments had ∼50% less AIR_argmax than the recipients of whole pancreata. Although a recent study (6) of GPAIS in minipigs given streptozocin has been published, comparisons with the data reported herein are not possible because AIR_arg was not reported in that publication.

A judgment as to whether or not these observations are necessarily in conflict with previous reports (4,5) requires several considerations. First, major species differences are involved, i.e., baboon versus dog versus humans with type 1 diabetes. Second, the animal studies involved surgery to remove 65% of the pancreas in one case and streptozotocin to kill β-cells in the other. While surgical excision is clearly a precise approach to diminishing β-cell mass, the use of streptozotocin is not necessarily as precise. Even though histology and immunochemistry are valuable approaches to assessing residual β-cell mass, they cannot assess the functional integrity of the surviving β-cells, some of which may have undergone functional damage but not death during exposure to the drug. Third, decline in the magnitude of acute insulin responses to glucose over time in PTX recipients may reflect events other than declining β-cell mass. Other variables include the decreasing doses of prednisone that patients use over time. This would be accompanied by lesser degrees of insulin resistance and therefore less secretory demand on and smaller insulin responses from the allografted organ’s β-cells. This situation is similar to that observed in previously obese individuals who have lost weight, have become more insulin sensitive, and have smaller glucose-stimulated insulin responses. Another variable is the use of cyclosporine in PTX recipients, an immunosuppressive drug that is well documented to decrease insulin content of the islet as well as inhibit glucose-stimulated insulin responses (21–24). However, a previous study by Teuscher et al. (3) demonstrated that this drug also decreases insulin secretory reserve in humans, so it seems unlikely to be an explanation for the results in the present study in which AIR_argmax was unchanged. Perhaps the most reasonable position to take is that the previously reported dog and baboon studies are sufficiently different from the current human studies that the markedly different results do not necessarily contradict one another. This position is reinforced by the fact that the dog and baboon studies are in disagreement with each other because in one case AIR_arg changed little when AIR_gluc decreased greatly and in the other AIR_arg changed greatly when AIR_gluc changed little.

In conclusion, these studies establish that in successful recipients of segmental and whole PTXs, the physiologic reciprocal relationship between AIR_gluc and FPG is maintained; that AIR_arg and AIR_gluc decline in parallel as FPG increases; that AIR_arg is present in instances where AIR_gluc has disappeared; that segmental PTX recipients have roughly 50% of the magnitudes of AIR_gluc, AIR_arg, and AIR_argmax observed in whole PTX recipients; that the simpler measurements of AIR_gluc and AIR_arg can both be used reliably to estimate insulin secretory reserve in PTX, and presumably islet transplant, recipients; and that successful PTX recipients studied long term and longitudinally, who maintain euglycemia without exogenous insulin treatment, experience time-dependent declines in β-cell responsivity as reflected by AIR_gluc. At the clinical level, the most important indicator of success for transplant recipients is their ability to maintain stable levels of fasting glucose, intravenous glucose tolerance, and HbA_1c over many years despite the use of immunosuppressive drugs known to diminish insulin sensitivity and β-cell function. While it seems fairly remarkable that some patients can maintain normoglycemia under these therapeutic conditions for >20 years, the answer to the question of whether the decrease in the magnitude of AIR_gluc in the longitudinal studies is a harbinger of eventual organ failure and a return to hyperglycemia will have to await continued follow-up and further metabolic studies.

This work was supported by National Institutes of Health Grant no. RO1 DK 39994 and by the National Institutes of Health General Clinical Research Centers at the University of Washington and the University of Minnesota.

The author gratefully acknowledges the superb assistance provided by the research staffs of the University of Minnesota and University of Washington General Clinical Research Centers as well as the excellent technical assistance of Elizabeth Oseid and valuable manuscript preparation assistance of Mike Toney.