0021-972X/86/6201 -0077$02.00/0 Journal of Clinical Endocrinology and Metabolism Copyright © 1986 by The Endocrine Society

Vol. 62, No. 1 Printed in U.S.A.

Effects of Insulin on Fasting and Meal-Stimulated Somatostatin-Like Immunoreactivity in NoninsulinDependent Diabetes Mellitus: Evidence for More than One Mechanism of Action* M. GUTNIAK, V. GRILL, AND S. EFENDIC Department of Endocrinology, Karolinska Hospital, S-10401 Stockholm, Sweden

(glucose clamping). Basal SLI levels decreased significantly only during the normoglycemic clamp. The SLI response to a meal was more pronounced during the normoglycemic than the hyperglycemic clamp. The patterns of glucagon and GIP were similar during the two clamp conditions, while both basal and stimulated C-peptide levels were lower during the normoglycemic clamp. To investigate the temporal relationship between changes in blood glucose and SLI levels, patients were studied during a prolonged (270-min) period of normoglycemic clamp and fasting. After attaining normoglycemia, SLI levels continued to decline for 150 min, whereas glucagon and GIP levels did not change. We conclude that in patients with NIDDM, 1) insulin significantly lowers basal SLI levels if normoglycemia is concomitantly attained; this action of insulin was partially dissociated from its hypoglycemic action; 2) hyperglycemia per se inhibits a meal-induced SLI response, and 3) insulin effects on SLI are not secondary to changes in glucagon or GIP levels. {J Clin Endocrinol Metab 62: 77,1986)

ABSTRACT. We assessed the effects of insulin and normalization of blood glucose on plasma levels of somatostatin-like immunoreactivity (SLI) in patients with noninsulin-dependent diabetes mellitus (NIDDM). In one series of experiments, normalization of blood glucose was achieved by Biostator-controlled feedback infusion of insulin. This procedure reduced plasma SLI levels by 34% [from 17.1 ± 2.1 (±SEM) to 11.3 ± 1.9 pg/ml; P < 0.05], concomitant with a significant reduction in plasma glucagon and C-peptide and an evanescent decrease in plasma gastric inhibitory peptide (GIP) levels. An ensuing mixed meal elicited a rise in SLI that reached the same levels during infusion of insulin as during uncontrolled hyperglycemia; the incremental increase was, however, 45% higher (P < 0.005) during insulin infusion. Furthermore feedback insulin infusion enhanced GIP and decreased C-peptide responses, but did not affect the glucagon response to the meal. To further evaluate the influence of insulin on SLI levels, we compared the effects of normo- and hyperglycemia during constant hyperinsulinemia by varying the rate of glucose infusion

S

EVERAL features of somatostatin release from the pancreas and gastrointestinal tract are abnormal in a diabetic state. Basal levels of somatostatin-like immunoreactivity (SLI) are increased and the response to amino acids is enhanced in rats (1) and dogs (2), whereas the normal response to glucose is abolished or attenuated in animals (2-5) and in diabetic man (6). Administration of insulin in vivo lowers plasma SLI levels in diabetic dogs (7) and restores responsiveness to glucose in diabetic rats (4, 5) whereas insulin in vitro has no effect (3, 5). The latter results suggest that indirect rather than direct effects of insulinopenia underlie the abnormalities in somatostatin secretion. Indirect effects of insulinopenia could, in turn, be related to hyperglycemia, to other

metabolic abnormalities, or to changes in levels of hormones that influence somatostatin secretion. The present study was undertaken to investigate the short term effects of insulin infusion and blood glucose normalization by an artificial pancreas (Biostator) on basal and meal-stimulated plasma levels of SLI in patients with noninsulin-dependent diabetes mellitus (NIDDM). After we found that insulin infusion influenced SLI levels in NIDDM patients, we investigated whether these effects were due solely to the insulin administered or were due to the normoglycemia that was achieved. Levels of glucagon and gastric inhibitory peptide (GIP) were also measured to assess the possible influence of these stimulators of somatostatin release (8, 9) on plasma SLI concentrations.

Received October 17,1984. Address requests for reprints to: Dr. Mark Gutniak, Department of Endocrinology, Karolinska Hospital, Box 60500, Stockholm S-10401, Sweden. * This work was supported by grants from the Swedish Medical Research Council (no. 00034 and 04540), the Nordic Insulin Foundation, and the Swedish Diabetes Association.

Materials and Methods Patients Twenty NIDDM patients were studied (13 men and 7 women). The mean ideal weight (according to Metropolitan 77

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 22 January 2017. at 11:32 For personal use only. No other uses without permission. . All rights reserved.

78

GUTNIAK, GRILL, AND EFENDlC

Life Insurance Tables) was 110.9%, with a range of 90-138%. The mean age was 49.0 yr (range, 29-67 yr). The mean duration of diabetes was 5.5 yr (range, 1-15 yr). The mean haemoglobin Aic was 10.6% [range, 7.9-13%; normal values, 5.95 ± 0.7 (±2 SD)]. The patients satisfied the criteria for NIDDM of the USA National Diabetes Data Group (10). No patient had renal impairment, gastrointestinal disturbances, or clinical evidence of neuropathy. A standard diet was recommended for all patients, with restrictions on simple carbohydrates. No patient had taken insulin in the past. Four patients had taken oral hypoglycemic agents; they stopped drug treatment for at least 3 weeks preceding the investigation. None was taking any medication at the time of study. None of the parameters mentioned differed significantly among the subgroups participating in different protocols.

egg or 20 g hard cheese. This breakfast was carbohydrate poor, containing, in energy equivalents, 25% carbohydrates, 24% protein, and 51% fat. The total energy content was 300 Cal. When insulin was infused, 5.1 ± 1.1 IU were given for 120 min after breakfast. In a second series of experiments, seven other patients (four men and three women) were studied after an overnight fast. Hyperinsulinemia was induced by insulin infusion, and blood glucose was clamped at either normoglycemic or hyperglycemic levels. The insulin was infused at a constant rate of 1.6 mU/ kg-min throughout the experiments. A normoglycemic or hyperglycemic clamp was maintained by a Biostator-directed variable infusion of glucose. The amount of insulin infused during the postprandial 120 min was about 3 times greater than the amount infused during the insulin feedback experiments. No insulin bolus doses were given. The following standard /BD

Protocols

The protocols of the study were approved by the Ethical Committee of the Karolinska Hospital. All patients were studied as out-patients after an overnight fast. Four to 14 days elapsed between experiments in individual patients. At 07.30 h on the morning of the study, three cannulae were inserted. One cannula, inserted in^n antecubital vein, was used for intermittent blood sampling for hormone levels and was flushed with saline after use. A second cannula was inserted retrogradely in a dorsal hand vein and was used for continuous blood withdrawal. Arterialization of venous blood was achieved by heating the forearm and hand in a thermoregulated sleeve at 50 C. The third cannula was inserted in a contralateral antecubital vein and was used for all infusions. In the first series of experiments, six patients (four men and two women) were studied in the fasted state and for 120 min after a meal. The patients were studied on two occasions, i.e. during poor metabolic control as well as after normalization of blood glucose by feedback insulin infusion delivered by a Biostator. The patients were connected to the Biostator in random order on both occasions. On one occasion the Biostator was used only for monitoring blood glucose levels; on the other occasion the Biostator was also used for delivering insulin. The standard Biostator controller algorithm1 was employed during feedback insulin infusion. This algorithm uses the following constants: VAR, 100-200; BI, 80; FI, 376; RD, 50; BD, 60; and FD, 296. Insulin infusion was started immediately after the first blood sample had been obtained. Normoglycemia was then achieved within a 20- to 60-min period. The time point of stable normalization was decided during the experiments from the blood glucose curves, an example of which is presented in Fig. 1. This time point was designated -30. Thirty minutes later, the patients ate, within 10 min, a standard breakfast which consisted of 0.2 liter whole milk, one piece of crispy bread with 5 g margarine or butter, 20 g cold ham, and either one boiled 1

JCE & M • 1986 Vol 62 • No 1

VAR, a control variable for modifying insulin infusion rate (%); BI, the target blood glucose level (mg%); FI, the maximum insulin infusion rate (mU/min); RD, the maintenance rate of glucose infusion at safety level (mg/min); BD, a safety level of blood glucose (mg%); FD, the maximum glucose infusion rate (mg/min); DR, the calculated glucose infusion rate; G, the last minutes' average blood glucose reading.

— GV

algorithm was used: DR = RD I — 1 , with the following \ 25 / constants: RD, 6.23 ± 0.34 (±SEM) mg/kgmin for the normoglycemic clamp and 14.3 ± 0.5 mg/kg-min for the hyperglycemic clamp. The patients received the same standard breakfast described above 30 min after normalization of blood glucose. In a third series of experiments, seven other patients (five men and two women) participated in a single 270-min study. Normoglycemia was maintained by the normoglycemic clamp procedure, and the patients fasted throughout the experiment. Assays Blood samples were collected in plastic tubes containing EDTA (0.048 ml; 0.34 M) and Trasylol (1000 kallikrein inhibitor units; Bayer, Leverkusen, West Germany) and immediately placed on ice. Samples were centrifugated at 4 C, and plasma was frozen at -20 C. Blood glucose was measured by a glucose oxidase method (11). Haemoglobin Alc was measured by isoelectric focusing (12). Plasma insulin was determined by RIA (13), C-peptide by RIA using a commercially available kit (Novo Research Institute, Bagsvaerd, Denmark), glucagon by RIA based on the method of Faloona and Unger (14) using the antibody 30K, and GIP as described previously (15). Unlabeled and 125I-labeled GIP as well as the antibody R 65 were kind gifts from Dr. A. J. Moody, Novo Research Institute. The antibody measures primarily the 5000-dalton GIP (15). All samples from an individual patient were analyzed at the same time in each assay. SLI was extracted from plasma and concentrated using Vycor glass (mesh 140, code 7930, Corning Glass International, Corning, New York), as previously described in detail (6, 16). After centrifugation, the supernatant was removed, and SLI was eluted from the glass using 1 ml 80% methanol. After evaporation, the residue was dissolved in 250 y\ assay buffer. Aliquots of 100 n\ were assayed in duplicate. A standard amount of somatostatin-14 (200 pg in 2 ml plasma) was extracted in parallel to the samples and served as standards in the assay after appropriate dilutions. This procedure corrected for losses of somatostatin during the extraction procedure. Tyrosine-1somatostatin (a kind gift from Dr. A. Arimura, Tulane University, New Orleans, LA) was labeled with 125I and used as tracer. The antibody used (R 141E) was raised in our laboratory

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 22 January 2017. at 11:32 For personal use only. No other uses without permission. . All rights reserved.

79

INSULIN EFFECTS ON SOMATOSTATIN IN NIDDM

Biostator

feed-back regulation

blood glucose (= mmol/l

Insulin infusidn(=i mU/min

FIG. 1. An example of Biostator regulation by feedback insulin infusion.

against somatostatin-14. Its specificity was validated previously (17). The limit of detection of somatostatin-14 was 0.5 pg/ assay tube. The interassay coefficient of variation was 13.5% (n = 8), and the intraassay coefficient of variation was 6% (n = 12). Statistical methods Results are given as.the mean ± SEM. Significance testing was carried out using Student's t test for paired data. A mealstimulated response was calculated as the integrated increase above the basal levels of hormone. These basal levels were calculated and expressed as the mean of measurements during the 30-min period preceding the meal, except when otherwise specifically indicated.

Results No treatment or feedback insulin before and during breakfast The feedback insulin infusion normalized the blood glucose level, which in untreated patients was 10.4 ± 0.6 mmol/liter (Table 1). An example of the infusion procedure is shown in Fig. 1. The mean insulin levels recorded during -30 to 0 min were 31 ± 9 mU/liter in the untreated group and 33 ± 8 mU/liter in the group receiving feedback insulin infusion. Biostator feedback insulin lowered basal SLI levels by 34% (Fig. 2 and Table 1). Glucagon and C-peptide levels

were also significantly decreased during this procedure (Fig. 2 and Table 1). During the feedback infusion, GIP levels declined evanescently, i.e. from 423 ± 66 before insulin infusion to 197 ± 50 pg/ml 30 min before breakfast (P < 0.05; Fig. 2). Ingestion of breakfast increased blood glucose more in the untreated patients than during feedback insulin infusion (Table 1). The meal raised SLI to similar levels whether patients were untreated or receiving insulin; however, the incremental response was 45% higher during the latter condition (Fig. 2 and Table 1). The glucagon response to breakfast was not significantly greater when patients were normoglycemic, whereas the GIP response was enhanced by 94% (Table 1). The rise in Cpeptide levels after the meal was depressed during feedback insulin (Table 1). Normoglycemic and hyperglycemic clamping before and during breakfast

To evaluate the influence of insulin infusion and changes in blood glucose vis a vis changes in insulin levels per se, we compared basal and meal-stimulated levels of hormones during noririoglycemia and hyperglycemia using a constant rate of insulin and a variable rate of glucose infusion. Two concentrations of blood glucose were chosen to approximate those achieved during the experiments with and without feedback insulin, as de-

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 22 January 2017. at 11:32 For personal use only. No other uses without permission. . All rights reserved.

80

JCE & M • 1986 Vol 62 • No 1

GUTNIAK, GRILL, AND EFENDIC

TABLE i. Effects of Biostator feedback insulin infusion on basal and meal-stimulated changes in blood glucose and hormonal parameters Basal

Blood glucose (mM) SLI (pg/ml) Glucagon (pg/ml) GIP (pg/ml) C-Peptide (pmol/liter)

Increment during meal (min 0-120)

Untreated

Feedback

P value"

10.4 ± 0.6 17.1 ± 2.1 134 ± 23 314 ± 60 0.4 ± 0.1

4.5 ± 0.4 11.3 ± 1.9 103 ± 18 261 ± 41 0.2 ± 0