Introduction

Dipeptidyl peptidase-4 (DPP-4) inhibitors improve glycemic control in patients with type 2 diabetes mellitus (T2DM) by increasing circulating levels of incretins, endogenous gut-derived peptide hormones that enhance insulin secretion and suppress glucagon release in a glucose-dependent manner [1–4]. Several DPP-4 inhibitors are widely approved for use as oral antihyperglycemic agents in treating patients with T2DM. All have been shown to provide significant improvements in standard indices of glycemic control [5]. There may be important differences between them, however, in the extent to which they provide sustained pharmacodynamic (PD) action (DPP-4 inhibition) throughout their recommended dosing cycles. The existing PD data on these agents have come from separate studies in which each agent was evaluated individually, with different study populations and dosing regimens (single versus multiple dose administration) [6–11]. There is a need, therefore, for comparison of levels of DPP-4 inhibition between different agents within a single study in which conditions are standardized. There also is a need to minimize the impact of dilution and substrate competition artifacts that may be introduced in the process of assaying DPP-4 inhibition ex vivo [9, 11–14] and prevent meaningful comparison between agents even when tested in parallel within the same study.

Reagent solutions must be added to plasma in order to assay DPP-4 activity. As a result, plasma drug concentrations will be lower in an ex vivo assay than they are in vivo and this dilution effect may lead to underestimation of levels of DPP-4 inhibition. The impact of this effect will vary depending on the rate at which a DPP-4 inhibitor dissociates from its binding site on the enzyme, the amount of plasma dilution, and the length of time between the addition of reagents and the measurement of enzyme activity. In addition, there is the potential for ex vivo underestimation of DPP-4 inhibition due to substrate competition. If inhibition is assessed with substrate present at a level that saturates the enzyme, then competitive DPP-4 inhibitors are likely to appear to be less effective than they are under physiological conditions in vivo, where endogenous substrates are likely to be present at subsaturating concentrations.

Measurement of DPP-4 inhibition by sitagliptin is expected to be especially sensitive to these two artifacts because binding between sitagliptin and DPP-4 is competitive and rapidly reversible. When plasma containing sitagliptin is diluted, the half-life (t 1/2) for readjustment to the new equilibrium for DPP-4 binding is <2 min [14], an interval that is brief compared with the length of time over which assays of plasma DPP-4 are typically run. Measurements of DPP-4 inhibition by vildagliptin may also be sensitive to the dilution artifact because the t 1/2 for dissociation of vildagliptin from DPP-4 is in the order of 3.5 min [14]. Measurements of DPP-4 inhibition by saxagliptin and an active metabolite, 5-hydroxy-saxagliptin, are likely to be relatively insensitive to this dilution effect because these molecular species dissociate slowly from DPP-4, with t 1/2 values of 50 and 23 min, respectively [14]. Given these differences in binding kinetics, it follows that valid comparisons between these agents can only be made when DPP-4 inhibition is measured using an assay condition that minimizes the impact of plasma dilution. In the present study, we have directly compared profiles of DPP-4 inhibition over time in patients with T2DM treated with saxagliptin, sitagliptin, and vildagliptin using an assay in which plasma was diluted by only 11%, a level that is likely to have only a minimal effect on outcome, regardless of binding kinetics. Using a randomized, placebo-controlled, open-label, crossover design, patients in a single cohort were treated in successive study periods for 5 days (so that all agents could reach steady-state conditions) with the highest recommended doses for each agent. Pharmacokinetic (PK) and DPP-4 inhibition data were obtained at various times throughout these treatments and for an additional 4 days after the last doses were given. The prespecified primary endpoint was the ‘trough’ level of DPP-4 inhibition measured after patients received their final doses in each study period. It was assessed at 24 h postdose in subjects being treated with study drugs q.d. and at 12 h postdose in one study period in which patients were treated b.i.d. (with vildagliptin).

Methods

This study was performed in accordance with all local laws and regulations, the Declaration of Helsinki of 1975, as revised in 2000, and guidelines established by the International Conference on Harmonization Good Clinical Practice. The protocol was approved by the National Ethics Committee Clinical Research of Medicines and New Methods of Treatment (Chisinau, Moldova). Prior to participating, all patients provided written informed consent. The study was performed at the Clinical Republican Hospital (Chisinau, Moldova) between June and December, 2012.

Patients

Eligible patients were male or female, 18–65 years of age, and had T2DM with HbA1C values ≥6.5% and ≤10.0% either while treatment-naïve or following washout of prior oral antihyperglycemic medications for ≥6 weeks (12 weeks for patients previously treated with peroxisome proliferator-activated receptor-γ agonists, insulin, DPP-4 inhibitors, or glucagon-like peptide-1 analogs). Patients with fasting plasma glucose >200 mg/dL (11.1 mmol/L) or estimated creatinine clearance ≤60 mL/min were excluded. Concomitant, stable use of medications for pre-existing hypertension and dyslipidemia was permitted. Use of medications, nutritional supplements, and dietary products known to be cytochrome P3A4/5 (CYP3A4/5) inhibitors, CYP3A4/5 inducers, P-glycoprotein-1 inhibitors, or human organic anion transporter 3 inhibitors was prohibited.

Study Design

This was an open-label, placebo-controlled, five-period crossover study in which each patient received in separate periods 5 mg saxagliptin q.d., 100 mg sitagliptin q.d., 50 mg vildagliptin q.d., 50 mg vildagliptin b.i.d. (50 mg every 12 h), and placebo q.d. These specific doses were tested because each is a highest recommended dose. With vildagliptin, 50 mg b.i.d. is a highest dose approved for use as monotherapy and 50 mg q.d. is the recommended dose approved for use in combination with a sulfonylurea. Assignment to treatment sequences was randomized using a computer-generated schedule. Treatments were administered for 5 days and separated by ≥10 days.

Procedures

Patients received nutritional and exercise counseling and were provided with glucose meters for home blood glucose monitoring. They were instructed to perform fingerstick glucose measurements 5 min before and 2 h after each meal, and to contact the study-site staff in the event that any blood glucose value was <70 mg/dL (3.9 mmol/L) or any fasting blood glucose value was >200 mg/dL (11.1 mmol/L). It was prespecified that patients would be discontinued from the study in the event of having more than one occurrence of fasting blood glucose >200 mg/dL or a single occurrence of any blood glucose measurement >250 mg/dL (14 mmol/L).

Patients were domiciled in the clinic in each study period from Day −1 (the day before the first dose) through the morning of Day 1. They were also domiciled from the evening of Day 4 through the morning of Day 6 (and provided with standard meals and snacks during this time). Approximately 14 days after receiving final doses of study drug, patients returned for a follow-up visit. All doses were administered in the clinic, either during outpatient visits or while patients were domiciled. The first dose was preceded by an overnight fast of at least 8 h and study medications were administered with approximately 240 mL of water. Evening doses of vildagliptin were given without regard to timing of consumption of food or drink.

Blood samples for PD analysis were collected into Vacutainer™ EDTA tubes (Becton Dickinson, Franklin Lakes, NJ, USA) predose on Days 1–5 and 12, 24, 36, 48, 96 h after the final morning dose on Day 5. Plasma was then isolated by centrifugation, divided into aliquots, and stored at −80 °C until analysis. Blood samples for PK analysis were collected predose on Days 1–5 and 0.5, 1, 2, 4, 8, 12, 16, 24, 36, 48, and 96 h following the final morning dose on Day 5. During treatment with vildagliptin b.i.d., additional samples were collected 13 and 14 h after the final morning dose on Day 5. For analysis of saxagliptin, 5-hydroxy-saxagliptin, and sitagliptin concentrations, samples were collected into Vacutainer™ K2EDTA tubes (Becton Dickinson) and plasma was stored at −20 °C. For analysis of vildagliptin, samples were collected into Vacutainer™ P800 tubes (Becton Dickinson) and plasma was stored at −80 °C.

Measurement of Plasma DPP-4 Activity

Assay of plasma DPP-4 activity was performed at PPD Bioanalytical Lab (Richmond, VA, USA). In this assay, 90 μL of plasma was combined with 10 μL of 100 μM glycyl-prolyl-amino-methylcoumarin hydrobromide (Bachem, Torrance, CA, USA). The final concentration of substrate was therefore 10 μM, a level much lower than the K M of human DPP-4 for this substrate (approximately 50 μM). Inhibition of DPP-4 was thus assayed under conditions designed to minimize the effects of plasma dilution and substrate competition.

The assay solutions were incubated at 37 °C in a black Immuno Plate (Part # 7205, Thermo Scientific, Waltham, MA, USA) on a SpectraMax Gemini EM™ Fluorescence Microplate Reader (Molecular Devices, Sunnyvale, CA, USA). Data were analyzed using SoftmaxPro® applications (Molecular Devices, Sunnyvale, CA, USA). Formation of reaction product was monitored by measuring the increase in relative fluorescence at 460 nm at 45-s intervals (using 360 nm as the excitation wavelength). Enzyme activity (A) was defined as the slope of this increase [in relative fluorescent units per seconds (RFU/s)] measured over the first 5 min of incubation. In a preliminary analysis, it was determined that 1 RFU/s was equivalent to product formation at a rate of approximately 0.2 pmol/s.

Percent inhibition of DPP-4 was calculated as 100 × (1 − A t /A 0), where A 0 was the enzyme activity measured predose and A t was the activity measured postdose at time t in the same treatment period. The lower limit of quantification (LOQ) for measurement of A was defined as the lowest level at which the imprecision (coefficient of variation) of measurements remained <20%. The LOQ was regarded as constant across all subjects and study periods. When A t did not exceed the LOQ, it was assigned the value of the LOQ (0.366 RFU/s).

Pharmacokinetic Analyses

Plasma concentrations of saxagliptin, 5-hydroxy-saxagliptin, sitagliptin, and vildagliptin were analyzed by liquid chromatography–tandem mass spectrometry at Pharmanet Canada, Inc. (now Inventive Health Clinical, Québec City, Québec, Canada). The analysis of sitagliptin was performed as previously described [15]. For saxagliptin and 5-hydroxy-saxagliptin, the validated calibration ranges were 100–50,000 pg/mL and 200–100,000 pg/mL, respectively. For sitagliptin, the validated range was 1.00–1,000 ng/mL. For vildagliptin, it was 1.00–2,000 ng/mL.

Non-compartmental PK analysis was performed with actual sampling times using the WinNonlin™ program (Professional, version 5.0.1; Pharsight®, Mountain View, CA, USA). Areas under the curve (AUCs) for concentration versus time were calculated using the linear trapezoidal method for ascending concentrations and the log trapezoidal method for descending concentrations. The terminal elimination rate constant (λz) was calculated as the negative of the slope of the terminal phase of the log-linear concentration versus time curve. These analyses included at least three terminal points. Apparent terminal t 1/2 was calculated as ln(2)/λz.

Statistical Methods

All available data were included in the analysis (including data from patients who did not complete the study but had at least one measurement). Levels of DPP-4 inhibition were analyzed using a linear mixed-effects model with fixed-effects terms for treatment and period. Data were transformed to a log-percent scale for analysis and then back-transformed to a linear scale for reporting. Baseline was defined in each period as the predose value on Day 1. Inference testing was one-sided and significance was defined as p < 0.05.

A heterogeneous compound symmetry covariance matrix was used to allow for unequal treatment variances and to model correlation between different treatment effects within the same subject via the REPEATED Statement® in SAS PROC MIXED® (SAS Institute, Inc., Cary, NC, USA). Kenward and Roger’s method [16] was used to calculate the denominator degrees of freedom for the fixed effects. It was assumed that time and carryover effects were negligible. One-sided 95% confidence intervals (CIs) for between-treatment differences in trough DPP-4 inhibition were calculated using the linear mixed-effects model mentioned above.

The primary hypothesis was that treatment of patients with T2DM with 100 mg sitagliptin q.d. provides significantly greater trough DPP-4 inhibition than treatment with 5 mg saxagliptin or 50 mg vildagliptin q.d. It was prespecified that if the lower bound of the one-sided 95% CI for the difference in trough DPP-4 inhibition between sitagliptin treatment and a comparator was >0, then this hypothesis would be confirmed. In addition, 95% CIs were calculated for the differences between all other pairs of treatments in this study and summary statistics were calculated for each treatment and timepoint in the study.

Based on historical observations of variability, it was calculated that a sample comprising 15 completing patients would provide at least 83% power (α = 0.05, one-sided) to detect a 20% difference (absolute percentage points) between treatments in trough DPP-4 inhibition.

Results

The randomized cohort included 12 women and 10 men (all Caucasian) with a mean (range) age of 55 (43–64) years and body mass index of 32 (26–39) kg/m2. Mean (range) baseline HbA1C was 7.4% (6.4–9.0%).

Five patients discontinued prior to study completion. One withdrew consent, one was determined after randomization not to have met eligibility criteria, and three patients had elevated blood glucose concentrations that met the prespecified criteria for discontinuation. These incidents of elevated blood glucose occurred primarily prior to dosing with active treatments (i.e., during washout periods). The investigator did not consider these to be adverse events (AEs).

Predose levels of DPP-4 activity (A 0) were 3.70–6.36 RFU/s (10.1- to 17.4-fold greater than the LOQ). Thus, the upper limits for quantitation of DPP-4 inhibition in individual measurements were between 90% and 94%. Actual values varied for each subject and treatment period based on the baseline DPP-4 activity measured in that period.

Plasma DPP-4 Inhibition

Trough levels of DPP-4 inhibition were generally stable by the morning of Day 3 during all active treatments (Fig. 1a) and each of these provided nearly maximal inhibition of DPP-4 at 12 h postdose as assessed on Day 5 (Fig. 1b). Two of the treatments, sitagliptin and vildagliptin b.i.d., provided nearly maximal inhibition at 24 h postdose as determined predose on Days 1–5 and also on the morning of Day 6, when the primary endpoint was assessed. With these treatments, LS-mean trough DPP-4 inhibition on Day 6 was 91.7% and 90.6%, respectively (Table 1). The difference between them was not significant (p = 0.128).

Fig. 1
figure 1

Mean (±standard error) percent dipeptidyl peptidase-4 (DPP-4) inhibition over time in plasma collected a predose on Days 1–5 of treatment with 5 mg saxagliptin q.d., 100 mg sitagliptin q.d., 50 mg vildagliptin q.d., 50 mg vildagliptin b.i.d., and placebo, and b at the indicated times on Days 5–9. The measurement at 12 h postdose on Day 5 was obtained before the second vildagliptin b.i.d. dose was administered. The dashed horizontal lines indicate 94% inhibition, the maximum level of DPP-4 inhibition that was quantifiable under the assay conditions used in this study. SE standard error

Full size image
Table 1 Percent inhibition of dipeptidyl peptidase-4 activity 24 h after final morning doses in patients with type 2 diabetes mellitus treated for 5 days with 5 mg saxagliptin q.d., 100 mg sitagliptin q.d., 50 mg vildagliptin q.d., 50 mg vildagliptin b.i.d., or placebo
Full size table

Treatment with either saxagliptin or vildagliptin q.d. provided significantly (p < 0.001) less trough DPP-4 inhibition than treatment with either sitagliptin or vildagliptin b.i.d. With saxagliptin, the LS-mean trough DPP-4 inhibition on Day 6 was 73.5% (18.2% and 17.1% less than with sitagliptin and vildagliptin b.i.d., respectively). With vildagliptin q.d., it was 28.9% (62.9% and 61.8% less than with sitagliptin and vildagliptin b.i.d., respectively). Treatment with saxagliptin resulted in 44.6% greater trough DPP-4 inhibition on Day 6 than treatment with vildagliptin q.d. (p < 0.001).

At 36 and 48 h after morning doses were given on Day 5, the mean levels of DPP-4 inhibition following sitagliptin treatment were 83% and 72%, respectively (Fig. 1b). Following vildagliptin b.i.d., saxagliptin, and vildagliptin q.d., the mean levels of DPP-4 inhibition at these times were 50% and 17%, 41% and 29%, and 7% and 21%, respectively. At the final timepoint (96 h after the morning dose on Day 5), mean DPP-4 inhibition was <25% with all treatments.

Pharmacokinetic Observations

Observed values for AUC0–24 h, maximal concentration (C max), concentration at 24 h (C 24 h), and other PK parameters are shown in Table 2. All dosing regimens appeared to result in steady-state plasma concentrations of parent drugs by Day 3 (data not shown). Plasma concentrations of 5-hydroxy-saxagliptin appeared to reach steady state by Day 5.

Table 2 Plasma pharmacokinetic parameters for saxagliptin, 5-hydroxy-saxagliptin (5-OH-saxagliptin), sitagliptin, and vildagliptin following treatment for 5 days with saxagliptin 5 mg q.d., sitagliptin 100 mg q.d., vildagliptin 50 mg q.d., and 50 mg vildagliptin b.i.d. in patients with type 2 diabetes mellitus
Full size table

Safety

The study treatments were generally well tolerated. In total, 10 AEs were reported in 7 of the 22 patients enrolled. Headache was the most commonly reported AE (six events reported in three patients). Five AEs were considered to be study related. These included four events of headache (two that were attributed to saxagliptin treatment and two that were attributed to sitagliptin treatment) and one event of influenza that was attributed to saxagliptin treatment. All AEs were reported to be mild (n = 6) or moderate (n = 4) in intensity and transient in nature. There were no deaths, no serious AEs, and no AEs that led to withdrawal from the study.

Other Observations

In any period in which a patient was administered placebo, plasma concentrations of saxagliptin, 5-hydroxy-saxagliptin, sitagliptin, and vildagliptin were measured predose on Day 1 and these were observed to be below the LOQ in all cases. These findings suggest that there was no carryover of drug concentration from one period to the next. Similarly, predose levels of DPP-4 inhibition on Day 1 were similar among patients in all treatment sequences suggesting no PD carryover, a conclusion that was further supported by the results of a sensitivity analysis in which predose measurements of DPP-4 activity in Period 1, Day 1 were used as the baseline for all subsequent periods (data not shown).

Discussion

The findings from this study show that saxagliptin, sitagliptin, and vildagliptin differ significantly in the trough levels of DPP-4 inhibition that they provide when administered to patients with T2DM following recommended q.d. and b.i.d. dosing regimens. Sitagliptin and twice-daily vildagliptin each provided trough inhibition levels of approximately 90%, the maximum inhibition level that was observed at any time in this study. In contrast, saxagliptin and once-daily vildagliptin provided trough inhibition levels of approximately 75% and 30%, respectively. Thus, it was observed that sitagliptin and twice-daily vildagliptin provided sustained, high levels of DPP-4 inhibition throughout 24 h, whereas saxagliptin and once-daily vildagliptin did not.

The trough level of DPP-4 inhibition observed in this study following saxagliptin treatment corresponded well with the trough level of inhibition (~70%) observed previously in a 2-week study in which patients were treated with saxagliptin 5 mg q.d. [17]. To our knowledge, there are no published DPP-4 inhibition data from patients treated with multiple doses of 50 mg vildagliptin b.i.d.; direct comparison between present and historical data is thus not possible. In a study in which single doses of 50 and 100 mg vildagliptin were given to patients with T2DM, the levels of DPP-4 inhibition observed at 24 h postdose were approximately 14% and 35%, respectively [7]. In the present study, the mean (±standard error) level of DPP-4 inhibition 24 h after the first dose of vildagliptin 50 mg q.d. was 9.1 ± 5.0% (Fig. 1). This value is reasonably consistent with the prior result.

In the present study, sitagliptin provided a greater trough level of DPP-4 inhibition than was previously observed in a Phase I trial in which patients with T2DM were administered a single dose of 200 mg sitagliptin. In that study, plasma samples were diluted 2.5-fold prior to ex vivo assay and the level of DPP-4 inhibition measured at 24 h postdose was 80.1% [6]. The greater level of DPP-4 inhibition observed in the present study could be accounted for by the difference in plasma dilution. A statistical method has been described to correct for this dilution, and based on this method one would expect to observe 80% DPP-4 inhibition by sitagliptin after 2.5-fold dilution of plasma when the true level of inhibition in vivo is ~96% [9]. We have not applied this correction in the present analysis because the extent of plasma dilution in our assay was sufficiently small (11%) that its effect could reasonably be regarded as negligible.

The PK findings in the present study were generally consistent with those reported previously [6, 8, 10]. It was confirmed that all plasma drug concentrations were at steady state when the primary endpoint was measured and that PD effects were stable from Day 3 onwards. Hence, differences in drug exposure did not appear to play a role in the observed differences in PD effect.

One limitation of this study was that antihyperglycemic efficacy was not evaluated. Hence, uncertainty remains as to whether the observed differences in DPP-4 inhibition have meaningful implications for glycemic efficacy. However, a recent, model-based meta-analysis of published data found evidence for greater reductions in HbA1C with more sustained DPP-4 inhibition [18]. Increasing reductions in HbA1C were found in this meta-analysis even as the weighted average DPP-4 inhibition increased beyond 90%, supporting the concept that maintaining nearly complete inhibition of DPP-4 over entire dosing cycles may be important in order to maximize antihyperglycemic efficacy with this class of drugs. In addition, we note that the finding of similar DPP-4 inhibition during treatment in our comparison of once-daily sitagliptin with twice-daily vildagliptin is consistent with prior reports suggesting generally similar glycemic efficacy when these regimens are compared [19].

In a study comparing the efficacy of sitagliptin 100 mg q.d. with saxagliptin 5 mg q.d. as an add-on to metformin for 18 weeks, the adjusted mean change from baseline in HbA1C was −0.62% versus −0.52% for sitagliptin and saxagliptin, respectively [20]. Although this difference was within the study’s prespecified limit for noninferiority for HbA1C, the sitagliptin group had a numerically greater adjusted mean change from baseline in fasting plasma glucose than the saxagliptin group [−16.2 mg/dL (−0.90 mmol/L) versus −10.8 mg/dL (−0.60 mmol/L), respectively]. The results of the present study suggest a possible mechanism for this glycemic outcome, in that sitagliptin 100 mg provided significantly greater DPP-4 inhibition at 24 h postdose than saxagliptin 5 mg.

Conclusion

This study found that three approved DPP-4 inhibitors provided significantly different trough levels of DPP-4 inhibition when administered to patients with T2DM following recommended q.d. and b.i.d. dosing regimens. Maximal inhibition was sustained throughout 24 h in patients treated with sitagliptin 100 mg once daily and vildagliptin 50 mg twice daily. Trough DPP-4 inhibition was submaximal in patients treated with saxagliptin 5 mg and vildagliptin 50 mg once daily.