Introduction

Top Abstract Introduction Methods Results Discussion References

Although it is now common practice to treat insulin-dependent diabetic patients with human insulin, normoglycemia has rarely been achieved by subcutaneous administration (Capaldo et al., 1984; Skyler, 1986). Hyperglycemia has often occurred during meals as a result of slow absorption of insulin; and hypoglycemia has often occurred between meals as a result of prolonged duration of absorption. Subcutaneous insulin absorption is influenced by many factors (Brange et al., 1990), among which the association state of human insulin (in hexamer) in pharmaceutical formulation (100 IU/ml ~ 0.6 mM) may be of importance (Brange et al., 1990). A series of human insulin analogues with reduced tendency to self-associate have been developed by recombinant DNA technology (Brange et al., 1987, 1988) and recently marketed (e.g., Humalog) (Howey et al., 1994; Torlone et al., 1994; Trautmann, 1994). These analogues promote rapid adsorption, which is better suited to meal-related therapy (Kang et al., 1990; Vora et al., 1988). However, some of insulin analogues (Table 1), such as Asp^B9Glu^B27 and Asp^B9, which have improved absorption properties but less potent than human insulin in receptor-binding affinity (human hepatoma HepG2 cell line) and free-fat cell assay (Drejer et al., 1988; Brange et al., 1988). Nevertheless, these analogues exhibit biological activity similar to or the same as human insulin in mouse blood glucose assay (Drejer et al., 1988; Brange et al., 1988). Analogue Asp^B9Glu^B27 was also observed to have full bioactivity in healthy pig assay by euglycemic clamp technique (Ribel et al., 1990). In addition to animal data, Asp^B9Glu^B27 and Asp^B28 were also observed to have similar bioactivities to that of insulin in normal (Kang et al., 1991a, b) and diabetic (Kang et al., 1990, 1991c) human subjects in clinical studies. However, no suitable diseased model has been developed for studying the pharmacokinetic and pharmacodynamic properties of various insulin/analogues formulations before human studies.

Swine's physiological similarity to humans (Panepinto and Phillips, 1986), and the development of effective and gentle animal-handling techniques (Panepinto et al., 1983; Lin and Chien, 1997), have led to increased use of swine as a reliable large-animal model for studying pharmacokinetic and phar- macodynamic properties of various investigational drugs (Oberle et al., 1994; Kaltenbach et al., 1996; Xing et al., 1998). The swine model has potential for providing a good prediction of clinical performance. In addition, the feasibility of using a continuous blood glucose monitoring system to continuously monitor the glycemic state in the conscious animal has been demonstrated in this laboratory (Lin et al., 1993). The use of the continuous glucose monitoring system has made possible an accurate determination of the nadir value, which often occurs at unpredictable times and lasts only briefly. In this investigation, a streptozotocin-induced chronic diabetic Yucatan minipig model (Marshall, 1979; Wang and Chien, 1996; Stanley et al., 1997) was used and evaluated for studying the pharmacokinetics and pharmacodynamics of four human insulin analogues (Asp^B9Glu^B27, Asp^B9, Glu^B27 and Asp^B28) for comparison with human insulin in acid solutions.

	Methods

Top Abstract Introduction Methods Results Discussion References

Materials. Glucose oxidase membrane, glucose standards and buffer solution used in the Glucose Analyzer (YSI model 27) were purchased from Yellow Springs Instrument (Yellow Springs, OH). Peristaltic pump and tubings (for the pump) were obtained from Cole-Parmer Instrument Co. (Chicago, IL). PE-10 (nonradiopaque polyethylene micro-tubing) was from Clay Adams (Division of Becton Dickinson, Parsippany, NJ). Tridodecylmethylammonium chloride-heparin complex (TDMAC-heparin) was obtained from Polysciences, Inc. (Warrington, PA). STZ, citrate acid monohydrate and sodium phosphate were purchased from Sigma Chemical Co. (St. Louis, MO). Human insulin and insulin analogues were gifts from Novo Research Institute (Bagsuaerd, Denmark).

Instrumentation. The continuous blood glucose monitoring system used for this investigation was assembled by connecting the sensor chamber of the glucose analyzer to a peristaltic pump, a specially designed mixing chamber and a data-acquisition station (Lin et al., 1993). The system is composed of three stations in sequence: one for blood sampling and mixing with buffer solution, one for blood glucose measurement and one for data acquisition.

Animals. Male Yucatan miniature swine were purchased from Buckshire Corporation (Perkasie, PA). The animals were housed individually in a pig pen (approximately 3.5 by 7.0 ft) and had free access to fresh water. The animals were fed a standard, commercial pig diet twice daily ad libitum and exposed to automated 12-hr lighting cycles. All synchronizers, including the feeding schedule, temperature (68-70°F) and relative humidity (50%), were fixed. Initially, considerable time was spent hand feeding and handling each pig to acclimated it to its surroundings.

Chronic diabetic minipig model. The animals had a mean (±S.E.) body weight of 57 (±6) kg at the time of inducing diabetes. They were then fasted for 12 to 24 hr before induction of diabetes mellitus. Fresh solution of STZ (120 mg/ml) was prepared in citrate phosphate buffer (0.1 M, pH 4.5) and used within 1 hr (Marshall, 1979; Wang and Chien, 1996). After the baseline blood glucose determination, an initial bolus dose of STZ (60 mg/kg) was injected i.v. Blood glucose levels were then monitored for at least 10 hr by a continuous glucose-monitoring system (Lin et al., 1993), on a continuous on-line basis, and plasma concentrations of insulin were measured by radioimmunoassay at predetermined time intervals to study the diabetogenic effects of STZ. During the experimental period, dextrose solution (25% w/v) was administered i.v., when the blood glucose level declined to drop down to 20 mg/dl, to offset the transient fatal hypoglycemia induced by STZ. One week after the initial dose of STZ, a second dose of STZ (60 mg/kg) was administered i.v. to each of the minipigs. During the course of studies, blood glucose levels and plasma insulin concentrations in the minipigs were regularly measured to determine the degree of hyperglycemia induced. The i.v. insulin tolerance, i.v. tolbutamide tolerance and i.v. glucose tolerance tests were also performed to confirm the diabetic state (Wang and Chien, 1996). Animals with a fasting blood glucose concentration, without any hyperglycemic treatment, maintained at a level higher than 120 mg/dl [compared to normal fasting glucose level, i.e., 49 ± 2 mg/dL (mean ± S.E.M., n = 12)] and lower than 200 mg/dl were selected and used in this investigation.

Preparation of animals. On the day of experiment, minipig was prepared as described in elsewhere (Lin and Chien, 1997). In brief, after a 16- to 24-hr fast, each minipig was put into an upright position and lightly restrained in a sling (Charles River Laboratories, Wilmington, MA), which gives minipig a comfortable support and minimized stress (Panepinto et al., 1983). Two sections of nonthrombogenic PE-10 tubing, coated on internal surface with TDMAC-heparin complex, were cannulated into the veins of both ears (one tube for each ear). The cannulated tubings allowed easy serial/continuous blood sampling during the study.

Animal studies. Before injections, the blood glucose level in each test minipig was continuously monitored. After a relatively stable baseline was attained and maintained for a period of at least 30 min, a s.c. dose (0.6 nmol/kg) of human insulin or one of the various analogue solutions was administered at inner-upper region of back leg. Insulin/analogues solutions (0.6 mM) were prepared by dissolving insulin/analogues in normal saline containing phenol (0.2%) and albumin (0.1%), and then adjusted to pH ~3.0 with 1N HCl. The blood glucose levels in the STZ-diabetic minipig were continuously monitored for a period of 12 hr. A recovery period of 1 to 2 wk was allowed between experiments with the same minipig. Each minipig received all insulin/analogue administrations and one placebo, and the sequence of insulin/analogue administrations was randomized. Eighteen experiments were performed, 6 in each minipig, and all were carried out between the 5th and the 6th mo after the initiation of diabetes induction. The mean (±S.E.) body weights of minipigs were 60 (±5) and 62 (±4) kg, respectively, at time of the first and last experiment.

Insulin radioimmunoassay. During the course of continuous blood glucose monitoring through the first PE-10 tubing, blood samples were also withdrawn through the second PE-10 tubing at predetermined intervals for radioimmunoassay of insulin/analogues. The blood samples were each collected in a chilled microtube and immersed in an ice bath. The blood samples and containers were maintained at 2 to 8°C throughout the entire process of blood collection and handling. The plasma was separated by centrifugation in a refrigerated centrifuge and the plasma was then aspirated and transferred into a microtube and immediately frozen until assayed. Assay of insulin/analogues was performed using Coat-A-Count Insulin kits (Diagnostic Products Co., Los Angeles, CA) with the respective insulin/analogue standards. The standard curves (range from 0-2400 pM) of four human insulin analogues were also evaluated and compared with human insulin.

Pharmacokinetic and pharmacodynamic analysis. For purposes of quantitative comparison of the pharmacokinetics between human insulin and various analogues, the values of C₀ (fasting basal plasma insulin level), C_max (peak concentration observed during the dosing period), t_max (time to C_max) and MRT after s.c. administration were determined from the plasma insulin (or analogue) concentration-time profiles. Moreover, the quantitative comparison of the pharmacodynamics between human insulin and various analogues, the values of E₀ (fasting blood glucose level), nadir (concentration at maximum glycemic reduction observed during the dosing period), t_nadir (time to nadir), and E_12-hr (blood glucose level at 12-hr after dosing) after s.c. administration were determined from the blood glucose concentration-time profiles.

Statistical analysis. The Student's paired t test was used to determine the statistical significance of the difference between pharmacokinetic and pharmacodynamic parameters of each of the various analogues and human insulin.

	Results

Top Abstract Introduction Methods Results Discussion References

Chronic diabetic minipig model. A typical set of blood glucose and plasma immunoreactive insulin profiles after a rapid, single i.v. injection of STZ (60 mg/kg) are graphically shown in figure 1. Yucatan minipigs have a baseline fasted glucose level of ~50 mg/dl [49 ± 2 (mean ± S.E.M., n = 12)] and baseline plasma insulin level of ~9 µIU/ml [9 ± 1 (mean ± S.E.M., n = 12)]. The results in figure 1 indicate that, after the injection of STZ, hyperglycemia was induced within 1 to 2 hr, and sustained several hours, and then hypoglycemia subsequently appeared. During the period of hypoglycemia, it is critically important to maintain the blood glucose level above 20 mg/dl. The transient fatal hypoglycemia induced by STZ was offset by intravenous administrations of dextrose (25%) solution (10 ml/hr).

View larger version (22K): [in this window] [in a new window]   Fig. 1.   Blood glucose and plasma insulin profiles in one Yucatan minipig, under induction of diabetes, after a single i.v. administration of streptozotocin (60 mg/kg) and hourly bolus injections (10 ml each) of dextrose (25%) solution.

View larger version (22K): [in this window] [in a new window]   Fig. 2.   Long-term blood glucose and plasma insulin profiles in the streptozotocin-induced chronic diabetic Yucatan minipigs (n = 3). The basal blood glucose and plasma insulin concentrations in normal Yucatan minipigs (n = 12) are displayed as mean (±S.E.).

Insulin radioimmunoassay. A linear logit-log relationship exits between the percentage of insulin bound and the concentration of human insulin or its various analogues added into all standard solutions for radioimmunoassay using the Coat-A-Count Insulin kits (fig. 3). The attainment of linearity indicates that the commercially available insulin kits can be used for the measurement and computation of various insulin analogues studied in this investigation. The mean (±S.D.) of sensitivities (95% intercept acquired from logit-log line) calculated from eight repeated RIAs were 27 (±19), 30 (±14), 34 (±26), 26 (±17) and 30 (±19) pM, respectively, for Asp^B9Glu^B27, Asp^B9, Glu^B27, Asp^B28 and insulin. The coefficient variations of inter-assays were 7, 4, 9, 5 and 7%, respectively, for correspondent analogues and insulin.

View larger version (30K): [in this window] [in a new window]   Fig. 3.   Comparison between human insulin (open circles) and its monomeric and dimeric analogues (solid circles) in their linear logit-log relationship measured by radioimmunoassay (n = 8).

Animal studies. The basal levels normalized plasma immunoreactive insulin/analogues and blood glucose profiles, after the s.c. administration of human insulin or its various analogues (0.6 nmol/kg) in the same group of diabetic minipig (n = 3) are compared graphically in figure 4A-D. The results indicate that the plasma concentrations of insulin/analogues all increase rapidly and reach their respective C_max within 1 hr after the s.c. injection. It was then maintained at a relatively steady level for at least another 4 hr, with the exception of Asp^B9Glu^B27 (fig. 4A) and Asp^B28 (fig. 4D), which gradually decline to baseline after reaching peak. However, the hypoglycemic response profiles of all analogues were very similar to that of human insulin: the blood glucose concentration declines gradually, with an onset time of about 30 min (from the baseline level), after s.c. injection, and all reach the maximum level of glycemic reduction (i.e., nadir) within 6 hr. After reaching nadir, the glucose concentrations rise gradually, but the original baseline level is not recovered within the observation period. Moreover, the plasma insulin and blood glucose profiles from the placebo were observed to be maintained at a relatively stable throughout the test period (fig. 4A) that indicates that the normalizations of pharmacokinetic and pharmacodynamic parameters by the basal insulin and glucose levels are possible.

View larger version (39K): [in this window] [in a new window]   Fig. 4.   Upper panel, Comparison in the plasma profiles of insulin and analogues in the streptozotocin-induced chronic diabetic Yucatan minipigs (n = 3) after the s.c. administration of human insulin and one of its analogues [AspB9GluB27 (A), AspB9 (B), GluB27 (C), and AspB28 (D)]. Lower panel, Comparison in the reduction profiles of blood glucose (in which the S.E.s are displayed at 1-hr intervals along the course of continuous blood glucose profiles).

Pharmacokinetic and pharmacodynamic analysis. The comparisons of the pharmacokinetics and pharmacodynamics between human insulin and its various analogues, after s.c. administration, were outlined in table 2. The mean (±S.E.) values of C_max were 598 (±21), 528 (±44), 176 (±21), 325 (±60) and 228 (±33) pM for Asp^B9Glu^B27, Asp^B9, Glu^B27, Asp^B28 and human insulin, respectively. The differences in C_max values were statistically significant between Asp^B9Glu^B27 and human insulin [P = .016, power = 0.95, 95% CI = 576 to 164], and between Asp^B9 and human insulin (P < 0.01, power > 0.99, 95% CI = 354 to 246), but not between Glu^B27 (P = .306, power = 0.13, 95% CI = 112 to 216) or Asp^B28 (P = .133, power = 0.29, 95% CI = 268 to 73) and human insulin. However, there is no difference between the values of time to C_max between various analogues and human insulin. Moreover, the mean (±S.E.) values of nadir were 67 (±9), 74 (±5), 65 (±5), 57 (±15) and 72 (±5)%, respectively, for Asp^B9Glu^B27, Asp^B9, Glu^B27, Asp^B28 and human insulin. There were no differences in the values of nadir and time to nadir between various analogues and human insulin. The mean (±S.E.) values of MRT were 148 (±4), 176 (±12), 165 (±9), 140 (±8) and 171 (±2) min for Asp^B9Glu^B27, Asp^B9, Glu^B27, Asp^B28 and human insulin, respectively. The differences in MRT values were statistically significant between Asp^B9Glu^B27 (P = .048, power = 0.63, 95% CI = 0.6 to 46) and human insulin, and between Asp^B28 (P = .032, power = 0.78, 95% CI = 7 to 55) and human insulin, but not between Asp^B9 (P = .700, power = 0.06, 95% CI = 53 to 43) or Glu^B27 (P = .504, power = 0.08, 95% CI = 26 to 38) and human insulin.

06

06

012

012

	Discussion

Top Abstract Introduction Methods Results Discussion References

The transient hyperglycemic state (fig. 1), following the i.v. administration of STZ in minipig, could be attributed to the diabetogenic effect of STZ; whereas the hypoglycemia observed resulted possibly from either liver damage or excessive insulin secretion by the injured Langerham islet. Therefore, it is critically important to offset the transient fatal hypoglycemia induced by STZ by administrations of dextrose to minipig. And, it is not possible without the using of the continuous glucose-monitoring system, which the blood glucose levels can be monitored on a continuous on-line basis.

Although hyperglycemia, after induction of diabetes, was maintained without any administration of antidiabetic treatment except for the various insulin analogues administered during the study, a lower basal insulin level than that of normal minipig were observed in STZ-induced diabetic minipigs up to 8 mo (fig. 2). The observation of basal endogenous insulin level suggests that the long-term maintenance of chronic diabetic minipigs becomes possible without any daily administration of exogenous insulin. Moreover, severe hyperglycemia can be prevented and then metabolic fluctuations can be minimized by the low basal endogenous insulin. Therefore, the life in these chronic diabetic minipigs can be prolonged, although the successful rate for this diseased model was around 50%. A total of six chronic diabetic minipigs, which had a fasting glucose level in the range of 120 to 200 mg/dl without any hyperglycemic treatment at all time, were involved in the study from its beginning. One of these diabetic minipigs was dropped out from the study, due to the severe of hyperglycemia within the first month. Although the remaining five completed the study, two of them showed some reduction in their diabetes state during the study and so excluded from the final analysis. Therefore, only the data from the three chronically stable diabetic minipigs were reported.

Complete cross-reactivity of antibody with human insulin and its various analogues were observed in figure 3 using the commercially available insulin kits. The antibody binding affinities, calculated from insulin standard curves, in comparison with human insulin are 74% for Asp^B9Glu^B27, 78% for Glu^B27, 69% for Asp^B28 and 97% for Asp^B9. The observations suggested that antibody binding affinities of human insulin have not been altered by the replacement of Asp^B9, but decreased by Glu^B27 and Asp^B28.

The finding of a significant difference in the plasma insulin profiles between the analogues and human insulin in diabetic minipig model agrees well with the results that reported in healthy swine (Ribel et al., 1990), healthy human (Kang et al., 1991a, b) and diabetic human (Kang et al., 1990, 1991c). They indicate that plasma insulin concentrations have been observed higher for the analogues with low-affinity to the insulin receptor (e.g., Asp^B9Glu^B27 and Asp^B9) than those for human insulin, although plasma insulin levels for the high-affinity analogues (e.g., Glu^B27 and Asp^B28) were lower than those for human insulin (fig. 4A-D; table 2). In addition, the blood glucose profiles between the analogues and human insulin in diabetic minipig model have similar results as in healthy swine (Ribel et al., 1990) and suggest that in vivo biological activities between human insulin and analogues with low and high affinity to the insulin receptor are equivalent regardless of the difference in pharmacokinetics.

The pharmacokinetics (C_max, AUC and MRT) and pharmacodynamics (nadir and ABGC) of various insulin analogues, after equimolar amount of insulin administered s.c. in chronic diabetic minipigs, in comparison with human insulin are outlined in table 3. Although the values of C_max and AUC in table 3 for Asp^B9Glu^B27 and Asp^B9 were observed 3- to 4-fold higher than that for Glu^B27 and Asp^B28, there was no difference in pharmacodynamics. It indicates that the biological potencies for Asp^B9Glu^B27 and Asp^B9 were only one-fourth to one-third of Glu^B27 and Asp^B28. The agreement of less biological potency for Asp^B9Glu^B27 and Asp^B9 in diabetic minipigs was reached with the results, shown in table 1, obtained from in vitro receptor-binding affinity (human hepatoma HepG2 cell line) and free-fat cell assay (Drejer et al., 1988; Brange et al., 1988). In addition, the similar in vivo biological activity between these analogues and human insulin was observed in diabetic minipigs, which was the same as reported in mouse blood glucose assay (Drejer et al., 1988; Brange et al., 1988) and in healthy pig assay by euglycemic clamp technique (Ribel et al., 1990). It indicates that the agreement could suggest that the mechanism of these analogues and human insulin in the chronic diabetic minipig could be supported by that in receptor-binding affinity, free-fat cell assay, mouse blood glucose assay and in healthy pig assay with euglycemic clamp technique. Furthermore, it is interesting to observe that there was a direct relationship between MRT and hypoglycemic potency for various analogues in chronic diabetic minipigs. In summary, although the insulin analogues are different from human insulin in pharmacokinetics, they exhibit similar biological activity to human insulin in the streptozotocin-induced chronic diabetic minipigs.

However, the observations of pharmacodynamics are different from that observed clinically in the normal and diabetic human (Kang et al., 1991a, c). Results indicate that the magnitude of the hypoglycemic nadirs is greater for Asp^B9Glu^B27 (2.7 mM) and Asp^B28 (2.9 mM) than that for human insulin (1.9 mM). Moreover, the time of the hypoglycemic nadirs is shorter for Asp^B9Glu^B27 (62 min) and Asp^B28 (65 min) than that for human insulin (201 min). The different observations could be due to either the use of a small number of animal in the study or the difference in the hypoglycemic profiles of insulins between minipigs and humans. In summary, although the pharmacokinetics of Asp^B9Glu^B27 and Asp^B28 in chronic diabetic minipigs have been observed similar to that in healthy and diabetic human subjects, the hypoglycemic effects of these analogues in chronic diabetic minipigs have been shown to be smaller than that in normal and human subjects. Thus, the difference of the hypoglycemic effect between minipigs and humans needs to be further investigated.