Pharmacokinetics and pharmacodynamics of gliclazide in Caucasians and Australian Aborigines with type 2 diabetes

Abstract

Aims

Gliclazide pharmacokinetics and pharmacodynamics were assessed in 9 Caucasians and 10 Australian Aborigines with uncomplicated type 2 diabetes.

Methods

Subjects were on a stable dose of 80 mg gliclazide twice daily, took 160 mg on the morning of study and had a standard breakfast. No further gliclazide was given over the next 48 h. Regular blood samples were drawn for serum glucose, insulin and gliclazide assay. Gliclazide was measured using h.p.l.c. Noncompartmental analysis was used to describe primary data. A multicompartment model incorporating entero-hepatic recirculation was fitted to group mean serum gliclazide profiles.

Results

The Caucasians were older than the Aborigines (mean ± s.d. age 53.4 ± 12.2 vs 40.3 ± 6.9 years, P < 0.05) but had similar diabetes duration, body mass index and glycated haemoglobin. Noncompartmental analysis revealed no between-group differences in gliclazide kinetics. Post-breakfast serum glucose and insulin responses were also similar apart from a longer time to maximum concentration (t_max) for glucose amongst the Aborigines (2.6 ± 0.4 vs 2.2 ± 0.3 h in Caucasians; P = 0.024). Gliclazide t_max exhibited a skewed unimodal distribution and was not associated with gliclazide maximum concentration, or glucose or insulin responses. Most patients had a serum gliclazide profile suggestive of enterohepatic recirculation and/or biphasic absorption. Model-derived estimates of the extent of putative enterohepatic recirculation were 30% and 20% of dose in Caucasians and Aborigines, respectively.

Conclusions

Gliclazide is equally effective in Caucasian and Aboriginal diabetic patients. The pharmacokinetics of oral gliclazide appear more complex than previously thought. Gliclazide pharmacodynamics are unrelated to rate and extent of absorption, consistent with a threshold concentration for hypoglycaemic effect.

Introduction

Gliclazide is a second-generation sulphonylurea drug used in the treatment of type 2 diabetes. It is generally well tolerated, is associated with a relatively low incidence of hypoglycaemia and may have beneficial effects beyond the reduction of serum glucose [1]. There have been several studies of the pharmacokinetics and efficacy of gliclazide in diabetic patients [2–5]. In one of these studies [2], there was evidence that a serum gliclazide concentration of 1.5 mg l⁻¹ represents the threshold for maximal hypoglycaemic effect. If there were such an ‘all or none’ response, the timing of administration of gliclazide in relation to meals and the fact that ‘slow absorbers’ (time to maximum concentration (t_max) > 4 h) have been identified in studies of healthy volunteers [2, 3] would be of limited importance as long as the threshold concentration was exceeded.

Metformin has been strongly recommended as first-line treatment for type 2 diabetes in Australian Aboriginal patients [6]. However, Aborigines have high rates of cardiovascular disease and renal impairment [7] that are relative contraindications to metformin use. Sulphonylurea drugs may therefore be a more appropriate first-line treatment in this context. Gliclazide would be preferable to longer-acting sulphonylureas such as glibenclamide because it is less likely to cause hypoglycaemia, an important issue in patients with a high ethanol intake [6]. Although there are no clear differences in gliclazide pharmacokinetic parameters between Japanese and Caucasian subjects in published studies [1], racial differences in the disposition and effect of a variety of commonly used drugs have been reported [8–11]. There are, however, no such studies in Australian Aborigines.

We have assessed the pharmacokinetics and pharmacodynamics of oral gliclazide in Caucasian and Aboriginal patients with uncomplicated type 2 diabetes. The results show that there are no clinically important differences in gliclazide disposition and effect between the two racial groups. Nevertheless, our data suggest that (i) there is either enterohepatic recirculation or absorption of gliclazide from two sites in type 2 diabetes (ii) that the t_max for gliclazide exhibits a skewed distribution rather than having a bimodal distribution of ‘early’ and ‘late’ absorbers, and (iii) that delay in absorption of gliclazide has no relationship to postprandial serum insulin and glucose responses, consistent with a threshold concentration for maximal hypoglycaemic effect.

Methods

Patients

Ten Aboriginal and 9 Caucasian patients with type 2 diabetes were studied (see Table 1). The Aboriginal subjects were classified as such by both identifying themselves, and being identified by the Aboriginal community, as Aborigine, and having a substantial proportion of Aboriginal ancestors. Subjects were recruited either from outpatient clinics at Fremantle Hospital or through community nurses involved in the management of Aboriginal diabetic patients in the Perth metropolitan area.

Table 1.

Demographic and laboratory details of the 9 Caucasian and 10 Aboriginal patients studied. Data are, unless otherwise stated, mean ± s.d.

	Caucasian	Aboriginal	95% CI on difference
Age (years)	53.4 ± 12.2	40.3 ± 6.9*	−3.7, −22.6
Gender (M/F)	3/6	4/6	–
Duration of diabetes (years)	5.8 ± 5.4	4.5 ± 2.9	2.8, −5.4
Body mass index (kg m−2)	33.8 ± 8.8	33.1 ± 11.6	9.4, −10.8
Glycated haemoglobin (%)¶	8.4 ± 1.8	8.8 ± 1.7	2.0, −1.3
Number of current smokers	2	4	–

^¶normal range < 6%

^*P < 0.05 vs Caucasian group.

All subjects were aged between 30 and 65 years, had been diabetic for at least 12 months and had been taking gliclazide 80 mg twice daily as sole treatment for diabetes for at least 4 weeks before recruitment. None of the patients was taking other medications known to alter glucose tolerance, had established macrovascular or microvascular complications of diabetes (including autonomic neuropathy) or other severe concomitant illnesses, or had experienced an episode of hypoglycaemia within the week before study. All patients had serum transaminase concentrations less than twice the upper limit of the laboratory reference range and a normal serum creatinine (< 120 µmol l⁻¹). The Aboriginal patients were recruited first and the Caucasian patients were matched as closely as possible to the Aborigines for age, gender and body mass index (BMI). Each patient gave informed consent to study procedures that were approved by the Human Rights Committee, Fremantle Hospital.

Study design

All subjects were required to fast from 20.00 h on the evening before study. After admission to the Metabolic Ward, Fremantle Hospital at 08.00 h the next day, subjects were weighed, their height was measured and they were placed on bed rest. At 08.15 h, an intravenous cannula was inserted into a forearm vein and samples for serum gliclazide, glucose and insulin assay were taken. The cannula was kept patent by flushing with small volumes of sterile normal saline. After baseline blood sampling, a bedside blood glucose estimation was performed to exclude hypoglycaemia and 160 mg gliclazide (Servier Laboratories Australia Pty Ltd, Melbourne, Australia) was administered orally with water. A further set of blood samples for gliclazide, glucose and insulin assay were taken at 09.00 h, followed immediately by a standard 2228 kJ breakfast (66 g carbohydrate, 21 g fat and 19 g protein).

Further blood samples were taken at 30 min intervals for 4.5 h until 13.30 h, when patients had a standard 1618 kJ lunch, and then at 16.30 h and 20.30 h, the latter sample followed by standard dinner (2614 kJ). Blood sampling continued the next day at 02.30 h and 08.30 h, after which a 2195 kJ breakfast was taken. The patients were then discharged from hospital to return at 20.30 h on day 2 and 08.30 h before breakfast on day 3 for further blood samples. Patients took no gliclazide after the initial 160 mg dose but were offered regular subcutaneous sliding-scale insulin if the blood glucose concentrations after 20.30 h on day 1 rose to levels either associated with symptoms or viewed as unacceptable by the patients themselves or the study team. After the final sample on the morning of day 3, patients were instructed to restart their usual doses of gliclazide. Patients were monitored closely throughout and adverse events, including hypoglycaemia, were recorded on standard forms.

Assay methodology

All blood samples were centrifuged and sera separated within 1 h of venepuncture. Sera were stored at −80 °C until assayed. Serum glucose estimations were performed using the glucose oxidase technique and serum insulin was assayed by immunoenzyzmometric methods (Tosoh, Tokyo, Japan).

Gliclazide concentrations in serum were measured by h.p.l.c. Serum (0.25 ml) and internal standard (glibenclamide 1.5 µg) were acidified and extracted with 5 ml of 2% v/v isoamyl alcohol in hexane. The samples were centrifuged and the organic layer transferred to a clean polypropylene test tube containing 0.2 ml of 0.02 m sodium hydroxide. The contents were mixed well, centrifuged and a 0.02-ml aliquot injected on to the column. A Merck Lichrospher 60 RP Select B 5 µm column (25 cm × 4 mm internal diameter) was used, with a mobile phase of 50% v/v acetonitrile in 0.6% phosphate buffer adjusted to pH 3.8 with phosphoric acid. The solvent was pumped at a flow rate of 1.5 ml min⁻¹ and the eluting peaks monitored at a wavelength of 235 nm. Serum concentrations of gliclazide were interpolated from a standard curve (1–16 mg l⁻¹) run with each batch of samples. The within-day coefficient of variation for the assay ranged from 2.4% at 0.25 mg l⁻¹ to ≤ 1.0% at ≥ 5.0 mg l^−1. Free serum gliclazide concentrations in 2–6 samples from each patient which covered a broad serum total gliclazide concentration range were assayed after ultrafiltration using Amicon Centrifree YM-30 centrifugal filter devices (Millipore Corp, MA, USA).

Pharmacokinetic modelling and statistical analysis

Initial analysis of serum gliclazide concentration-time data was by noncompartmental pharmacokinetic methods. The elimination rate constant (k_el) was first estimated by log-linear least squares regression analysis of the last 4–5 concentration-time data pairs (12–48 h). The contribution from previous doses was subtracted by use of the superposition principle. The resulting serum gliclazide concentration-time profiles were analysed by single-dose noncompartmental methods [12].

In all but one of the 19 subjects, there was at least one nadir in serum gliclazide during the first 6 h postdose, consistent with enterohepatic recirculation, biphasic gastric emptying or absorption of drug from two distinct sites within the upper gastrointestinal tract. As studies in animals have suggested enterohepatic recirculation [13, 14], this was included in a linear multicompartmental model (Figure 1) developed to describe gliclazide kinetics following oral administration. As individual data sets did not always allow for good parameter estimation, in part because samples were taken relatively less frequently between 6 and 12 h postdose, the model was fitted to mean linearly interpolated data with between-subject variance as the error model. The SAAM II program (SAAM Institute, Seattle, WA, USA) was used to develop the model.

Figure 1.

Multicompartmental model, incorporating enterohepatic recirculation, which was fitted to group mean serum gliclazide-time co-ordinates. GI = gastrointestinal tract and EV = extravascular exchange pool.

The model includes a two-pool system that describes serum kinetics and an enterohepatic recirculation loop, and is a variation on previously described models [15, 16]. Gliclazide absorption was assumed to start in compartment 1 (gastrointestinal tract), from which drug can be transported to the liver (compartment 2; k(1,2)) or lost from the system (k(1,0)) via the faeces. Compartment 2 is not a single compartment but a series of compartments with equal residence time that provide for the delay prior to appearance of gliclazide in the circulation. From compartment 2, gliclazide is transported (k(2,3)) to the serum (compartment 3) or enters the enterohepatic circulation (compartment 5) via k(2,5). Like compartment 2, compartment 5 is a series of compartments with equal residence time that delays the release of gliclazide back into the gastrointestinal tract for reabsorption. This compartment provides a continuous time lag that may represent transit of drug through the gallbladder. Compartment 3 represents serum gliclazide which can be transported (k(3,4)) to an extravascular exchange pool (compartment 4) or cleared from the circulation (k(3,0)). In addition to group mean gliclazide concentrations, previously published data from a number of studies demonstrating that 60–70% and 10–20% of gliclazide is excreted via urine and faeces, respectively [1] were used to constrain the ratio of fluxes from compartments 1 and 3, the sites of drug elimination.

Basal beta cell function (%B) and insulin sensitivity (%S) for each patient relative to a mean of 100% for each measure for healthy young nondiabetic Caucasian males were estimated from fasting serum glucose and insulin concentrations using homeostasis model assessment (HOMA [17]). Changes in serum glucose and insulin in response to the standard breakfast were assessed from the incremental area under the curve (AUC) from baseline to 13.30 h, as well as the maximum concentration and time to maximum concentration in each case.

Statistical analysis was by parametric methods using SPSS for Windows (SPSS Inc., Chicago, Illinois, USA). Two-sample comparisons were by Student's t–tests. Associations between variables were assessed using linear regression. Two-tailed levels of significance were used and data are, unless otherwise stated, reported as mean± 1 standard deviation (s.d. mean) or geometric mean and s.d. range.

Results

Gliclazide pharmacokinetics and tolerability

All 19 patients completed study procedures without any adverse effects. There were no episodes of hypoglycaemia, either detected at the bedside or on the results of subsequent formal serum glucose assay. None of the patients required subcutaneous insulin during the last 36 h of the 48 h study period and none experienced any problems resuming regular gliclazide treatment subsequently.

Mean concentration profiles of total serum gliclazide in the two groups are shown in Figure 2 and pharmaco-kinetic parameters derived from noncompartmental analysis are summarized in Table 2. There were no significant differences between the two groups for any parameter (P > 0.2 in each case). All but four patients (79%) had a day 1 baseline serum gliclazide of > 1.5 mg l^−1. After the 160 mg dose, the peak serum gliclazide concentration (C_max) ranged from 7.4 to 24.5 mg l^−1. The distribution of gliclazide t_max values in the two groups combined is shown in Figure 3. Log-transformation of individual t_max values normalized the distribution by Kolmogorov-Smirnov one-sample test (P < 0.05 for untransformed and P = 0.33 for log-transformed data vs normal distribution). The percentage of free gliclazide ranged from 3.3% to 7.1% and was independent of the total serum gliclazide concentration (data not shown).

Figure 2.

Mean serum gliclazide concentrations in Caucasian (^) and Aboriginal (•) patients. Vertical bars represent + 1 s.d. or - 1 s.d.

Table 2.

Gliclazide pharmacokinetic parameters in Caucasian and Aboriginal subjects derived from noncompartmental analysis. Data are mean ± s.d.

	Caucasian	Aboriginal	95% CI on difference
Baseline serum concentration (mg l−1)	5.7 ± 4.7	3.4 ± 3.0	1.5, −6.1
Area under the curve (AUC(0,∞) (mg l−1 h)	171.1 ± 59.9	143.3 ± 84.7	44.0, −99.6
Maximum serum concentration (Cmax; mg l−1)	15.0 ± 3.7	14.1 ± 5.1	3.5, −5.2
Time to maximum concentration (tmax; h)	2.8 ± 1.6	2.1 ± 0.7	0.5, −1.8
Volume (l kg−1)	0.23 ± 0.11	0.29 ± 0.15	0.19, −0.07
Mean residence time (h)	18.8 ± 4.3	18.8 ± 4.1	4.1, −4.1
Elimination half-time (t½,z; h)	12.5 ± 2.3	14.2 ± 4.1	5.1, −1.6
Systemic clearance (CL/F; l h−1 kg−1)	0.012 ± 0.006	0.015 ± 0.008	0.010, −0.003
Free gliclazide in serum (%)	4.7 ± 0.8	4.6 ± 1.2	1.0, −1.1

Figure 3.

Frequency histogram of 19 pooled values of the time to maximum serum gliclazide concentration (t_max). The distribution was unimodal but non-Gaussian.

The multicompartmental model fitted to group mean serum gliclazide concentration data provided an estimate of the amount of drug which undergoes enterohepatic recirculation (30% and 20% in the Caucasian and Aboriginal groups, respectively; see Table 3). Other parameters of potential clinical relevance were derived including elimination half time (t_½; 9.2 and 11.0 h), delay in absorption (1.3 and 1.0 h) and delay in reappearance of gliclazide in the gastrointestinal tract after accumulation in, and excretion from, compartment 5 (1.1 and 1.6 h in the Caucasian and Aboriginal groups, respectively; see Table 3). Figure 4 shows the model fits for mean concentration-time co-ordinates in the two groups, together with the amount of drug in compartment 5 (the gallbladder). Model goodness-of-fit was assessed using the Akaike information criterion (AIC). AIC values were lower for the model that included enterohepatic compartments (0.005 and 0.10 in Caucasians and Aborigines, respectively) compared to those derived from the same model without such compartments (0.65 and 0.38, respectively).

Table 3.

Pharmacokinetic parameters derived from amulti-compartmental analysis of mean serum gliclazide profiles in the Caucasian and Aboriginal groups incorporating enterohepatic recirculation. Data are parameter estimates and (95% confidence intervals).

	Caucasian	Aboriginal
Bioavailability (%)	86 (83–90)	89 (81–96)
Enterohepatic recycling (%)	30 (8–53)	20 (8–31)
Volume of distribution (l)	9.8 (5.1–14.5)	11.9 (9.6–14.1)
k(3,0) (h−1)	0.08 (0.04–0.11)	0.06 (0.01–0.11)
k(4,3) (h−1)	0.30 (0.00–0.57)	0.01 (0.00–0.02)
k(3,4) (h−1)	0.07 (0.02–0.12)	0.06 (0.02–0.10)
k(1,0) (h−1)	1.5 (0.9–2.0)	1.5 (0.5–2.4)
k(2,3) (h−1)	7.7 (4.7–10.8)	10.5 (8.8–12.3)
k(1,2) (h−1)	9.6 (8.9–10.3)	11.7 (10.2–13.1)
k(5,1) (h−1)	25.4 (13.1–37.6)	17.6 (10.9–24.2)
k(2,5) (h−1)	3.4 (1.0–5.7)	2.6 (1.1–4.1)
q2 delay (h)	1.26 (1.14–1.38)	1.07 (0.98–1.15)
q5 delay (h)	1.10 (0.57–1.63)	1.59 (0.99–2.19)

Figure 4.

Fitted curves from application of a multicompartmental model incorporating enterohepatic recirculation (Figure 1) fitted to mean serum gliclazide concentrations in Caucasian (^) and Aboriginal (•) patients. The model-derived time course for gliclazide in compartment 5 for both patient groups is shown in the upper right panel.

Basal and postprandial serum glucose and insulin

Serum glucose and insulin profiles are shown in Figure 5. Baseline serum glucose and insulin concentrations, as well as %B and %S, were similar in the two patient groups (see Table 4). The postprandial changes in serum glucose and insulin were also similar, apart from a significantly longer time to peak serum glucose concentration in the Aboriginal group (P = 0.024), representing a mean difference between the two groups of 25 min.

Figure 5.

Mean postbreakfast serum glucose (upper panel) and serum insulin (lower panel) concentrations in Caucasian (^) and Aboriginal (•) patients. Vertical bars represent + 1 s.d. or −1 s.d.

Table 4.

Baseline serum glucose and insulin concentrations and postprandial responses in Caucasian and Aboriginal subjects given a standard breakfast after gliclazide administration. Data are mean ± s.d. or geometric mean [s.d. range].

	Caucasian	Aboriginal	95% CI on difference
Baseline serum glucose (mmol l−1)	12.1 ± 3.9	10.9 ± 3.8	2.5, −4.9
Baseline serum insulin (mU l−1)	20.3[8.4–49.0]	14.5 [6.9–30.3]	10, −32¶
HOMA-derived beta cell function (%)	50 [18–137]	46 [21–100]	36, −64¶
HOMA-derived insulin sensitivity (%)	15[7–36]	22[10–46]	25, −8¶
Glucose incremental area under the curve (AUCglucose; mmol l−1 h)	2.0 ± 6.0	9.5 ± 11.2	16.3, −1.3
Insulin incremental area under the curve (AUCinsulin; mU l−1 h)	84 ± 111	157 ± 130	190, −43
Maximum glucose concentration (Cmax, glucose; mmoll−1h)	15.8 ± 5.2	16.1 ± 4.9	5.2, −4.6
Maximum insulin concentration (Cmax, insulin; mUl−1h)	73.4 [40.2–133.9]	65.0 [31.3–134.9]	50.7, −58.4¶
Time to maximum glucose concentration (tmax, glucose; h)	2.2 ± 0.3	2.6 ± 0.4*	0.78, 0.06
Time to maximum insulin concentration (tmax, insulin; h)	2.4 ± 0.5	2.1 ± 0.6	0.3, −0.3

^¶derived from untransformed data

^*P = 0.024 vs Caucasian group.

In order to assess whether delay in absorption of gliclazide influenced either the insulin or glucose response to the standard breakfast, univariate regression analysis was performed. There was no association between either insulin t_max or glucose t_max and ln[t_max] for gliclazide in the total series (P > 0.1 in each case). Conversely, in linear regression analysis with the ln[t_max] as dependent variable and gender, body mass index, baseline serum glucose and the logarithm of the baseline serum insulin (ln[ins]) as independent variables, ln[ins] was the only variable retained in the model (r² = 0.23, P = 0.036).

Discussion

Our results suggest that the pharmacokinetics of gliclazide are similar in Caucasian and Aboriginal patients with type 2 diabetes. There has been one previous study in which formal analysis of gliclazide kinetics was performed in diabetic patients taking regular doses [3]. In these four subjects, and in other single-dose studies involving a further 35 Caucasian and Asian patients [4, 5], reported mean values for t_max, volume of distribution (V/F) and t_½ were similar to those in our two patient groups. The percentage of free drug in our patients was also similar to previously published data, confirming that gliclazide is extensively protein-bound [2, 3].

We used a large dose of gliclazide (160 mg), sampled frequently and extended the study duration to 48 h. This protocol enabled novel comparisons between our pharmacokinetic data and those of others [1]. Firstly, our patients had proportionately higher C_max and AUC values than those reported previously for 40 mg and 80 mg doses [3–5], suggesting that pharmacokinetic properties such as t_max, V/F and t_½ are not concentration-dependent. Secondly, we found a skewed distribution of t_max that questions the classification of patients as ‘fast’ (t_max 0–4 h) or ‘slow’ (t_max 4–8 h) absorbers [1–3]. Were t_max largely a reflection of delayed gastric emptying, gender and body mass index would be prime determinants [18, 19]. We did not find an association between either gender or BMI and t_max. Nevertheless, consistent with the results of a previous study in which hyperinsulinaemia slowed gastric emptying [20], the basal serum insulin in our patients was a significant positive predictor of t_max, explaining approximately one quarter of the variability in this parameter. Thirdly, gliclazide protein binding was independent of serum total drug concentration across a range from basal to peak after the maximum recommended single dose (160 mg). Finally, we were able to show that gliclazide undergoes sufficient enterohepatic recirculation and/or absorption from a second more distal site in the gastrointestinal tract to have a significant effect on the shape of the early part of the serum concentration-time curve in both groups. There was a clear nadir in mean serum gliclazide at 3–4 h in almost all patients. A similar secondary absorption peak suggestive of enterohepatic recirculation has also been observed in studies of glipizide [21].

Excretion of drugs into bile depends mainly on their molecular weight and structure [22]. A molecular weight above 300 is associated with appreciable biliary excretion in rats but higher values (400–500) favour biliary excretion in humans [22]. Gliclazide has a molecular weight of 323. Nevertheless, compounds with molecular weights similar to that of gliclazide have been shown to undergo significant enterohepatic recycling in humans [23], while enterohepatic recirculation of gliclazide has been reported previously in animal studies [13, 14]. Although there is no direct evidence for biliary excretion of gliclazide in humans, faecal radioactivity has been reported up to 96 h after a single dose of [¹⁴C]-gliclazide in healthy volunteers [2].

Given this background, and since biliary excretion of other sulphonylurea drugs may occur [21, 24], we developed a multicompartmental model based on previous studies [15, 16] that enabled preliminary estimation of the extent of a putative enterohepatic recycling process. Although we used group mean data and estimated 8 kinetic parameters, there were 19 mean data points per group and the 95% confidence intervals were all below 50% of the individual parameter estimates. In addition, AIC values supported the use of a model that included enterohepatic compartments. The extent of mean enterohepatic recirculation in the two groups (30% and 20%) was consistent with data from animals in which 20% of the dose was estimated to recycle [14]. Our compartmental model also suggested that there was a significant delay in absorption of > 1 h in most patients. The model-derived t_½ of serum gliclazide in both groups was several hours shorter than the respective noncompartmental estimates, consistent with the structure of the model.

Type 2 diabetes in Aborigines presents typically at a young age and is associated strongly with insulin resistance [25]. Despite access to a large, community-based type 2 diabetes database, we were unable to find sufficient young Caucasians who fulfilled the study entry criteria and could be matched with the Aboriginal patients. Thus, the mean age of the Caucasians was 13 years greater than that in the Aboriginal group. Nevertheless, the two groups were similar in other respects, with comparable degrees of insulin resistance and pancreatic beta-cell dysfunction assessed using the HOMA model. This was an important consideration in the comparison of pharmacodynamic responses to gliclazide.

The only significant pharmacodynamic difference between our Caucasian and Aboriginal patients was in glucose t_max which occurred a mean of 25 min later in the Aboriginal group. Given the similarity of other measures of glucose and insulin homoeostasis, this appears to be of limited clinical importance. There was, in fact, a narrow distribution of glucose t_max (95% confidence interval [CI] 2.2–2.6 h) and insulin t_max (95% CI 1.9–2.5 h) compared with gliclazide t_max (95% CI 1.8–3.0 h) and no significant correlations between these variables and with gliclazide C_max. This indicates that the rate and extent of gliclazide absorption have limited impact on glycaemia, consistent with a threshold concentration for maximal hypoglycaemic effect. The observation that the patient with the longest gliclazide t_max (6.3 h) had a glucose t_max value (2.2 h) that was the same as the group mean is in accord with this hypothesis. Most of our patients had a steady state basal serum gliclazide > 1.5 mg l⁻¹ [2] and the remainder exceeded this level within 1 h of the 160 mg dose being given. This suggests that an 80-mg twice daily regimen should be adequate for the majority of patients with type 2 diabetes.

Recent data from Caucasian patients confirm the role of the sulphonylurea drugs, together with metformin and insulin, as safe and effective agents in the management of type 2 diabetes [26]. Our results suggest that gliclazide also can be used first-line in ‘Westernised’ urban-dwelling Aborigines with type 2 diabetes. Given the mean elimination half-time of 12–14 h in our two patient groups and the fact that approximately 1 in 5 of our patients had a morning predose serum gliclazide concentration below the theoretical efficacy threshold, it would seem appropriate to recommend a twice daily regimen for most patients. Our data also suggest that the timing of gliclazide administration in relation to meals is unimportant, consistent with the results of a previous study which specifically addressed this issue in patients in whom the total serum gliclazide concentration did not fall below 2 mg l⁻¹ [27]. Given that gliclazide has proved an effective hypoglycaemic agent for several decades, the clinical significance of possible enterohepatic recirculation and/or dual-site absorption is uncertain but merits further study.