Understanding individual drug response variation
Kroonen, Marjolein
DOI:
10.33612/diss.127010643
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2020
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Kroonen, M. (2020). Understanding individual drug response variation: Pharmacokinetic analysis of diabetes trials. University of Groningen. https://doi.org/10.33612/diss.127010643
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
4
Population pharmacokinetics and individual UACR
exposure-response analysis for Empagliflozin, Linagliptin
and Telmisartan in patients with type 2 diabetes
M.Y.A.M. Kroonen J.V. Koomen V. Rotbain Curovic F. Persson P. Rossing A. Kooy G.D. Laverman D. de Zeeuw H.J. Lambers-Heerspink J. Stevens
Submitted
Abstract
Introduction: Sodium-glucose co-transporter 2 inhibitors (SGLT2i), dipeptidyl
peptidase-4 inhibitors (DPP4i) and angiotensin-receptor blockers (ARB) have been shown to lower urinary albumin:creatinine ratio (UACR) in patients with diabetic kidney disease, although with large inter-individual variability in response. Our aim was to study inter-individual variability to these drugs and associate individual exposure with changes in UACR during treatment.
Methods: Twenty-nine patients with type 2 diabetes and elevated UACR (>30mg/g)
were randomized to 4 weeks of treatment with empagliflozin (10 mg/day, n=9), linagliptin (5 mg/day, n=8) or telmisartan (80 mg/day, n=12). Nine pharmacokinetic samples were collected over 24h. Change from baseline UACR was calculated at start and end of the treatment period. Population pharmacokinetic models were developed to
estimate the area under the curve from time 0 to infinity (AUC0-∞) as measure for
individual exposure.
Results: Mean AUC0-∞ (± 95% CI) for empagliflozin was 885.5 ng·h/mL (468.6 to
1239.9 ng·h/mL), for linagliptin 203.3 ng·h/mL (173.7 to 239.6 ng·h/mL) and for telmisartan 4204.0 ng·h/mL (747.1 to 9575.5 ng·h/mL). For every 100 ng·hr/ml increment in area under the plasma concentration curve, the percentage change in log transformed UACR was 2.7% (p=0.21) for empagliflozin, 48.1% (p=0.51) for linagliptin and -0.3% (p=0.08) for telmisartan.
Conclusions: The individual pharmacokinetic profiles of empagliflozin, linagliptin and
telmisartan were adequately described by their respective population pharmacokinetic models. We observed a large variation in individual exposure to these drugs. The individual exposure did not correlate with UACR changes which may be due to the relatively small sample size.
Introduction
Optimal glycemic- and blood pressure control, preferably with angiotensin converting enzyme inhibitors (ACEi) or angiotensin-receptor blockers (ARBs), is the cornerstone of the treatment of diabetic kidney disease. Drugs that improve glycemic control, such as sodium-glucose co-transporter 2 inhibitors (SGLT2i), glucagon like protein 1 (GLP1) agonists and dipeptidyl peptidase-4 inhibitors (DDP4i), as well as ACEi and ARBs have been shown to lower UACR. (1) The degree of UACR lowering has been associated with the degree of long-term renal protection. (2-4)
Prior studies have shown that the degree of UACR lowering varies among individual patients. In clinical trials with SGLT2i, DPP4i and ARBs, a large inter-individual variation in UACR response has been observed with approximately 30% of patients not achieving a reduction in UACR. This inter-individual variation in UACR response is reproducible after exposing the same individual to the same drug at the same dose indicating that the inter-individual variability is not random but reflects true pharmacological response variability. (5,6)
It is currently unknown which factors determine the inter-individual variability in response and to what extent individual exposures to these agents determine response variation. A better understanding of the relationship inter-individual exposure and UACR response may shed light on the underlying determinants of individual response and may help in designing strategies to overcome therapy resistance. (6-8)
This study investigated the inter-individual variability in exposure to the SGLT2i empagliflozin, the DPP4i linagliptin and the ARB telmisartan, and secondly, to associate the individual exposure with changes in UACR during treatment with these drugs.
Methods Clinical trial design
Data from 27 subjects that participated in a pharmacokinetic substudy of the ROTATE-2 trial were used for this analysis. The ROTATE-ROTATE-2 trial is a double-blind, 48-week, multicenter, cross-over study to determine the individual albuminuria lowering response of four different albuminuria lowering drug classes in patients with type 2 diabetes and micro- and macroalbuminuria. The study protocol of the ROTATE-2 trial is registered on www.trialregister.nl (ROTATE-2: NL5459 (NTR5603)). In short, diabetes mellitus
type 2 patients ≥ 18 years of age and elevated UACR (>30 mg/g and ≤500 mg/g) were eligible for enrolment. Exclusion criteria relevant for the pharmacokinetic study were, but not limited to, any medication surgical or medical condition which might significantly alter the absorption, distribution, metabolism or excretion of drugs.
During the run-in period, patient’s ACEi, ARB, direct renin inhibitor (DRI) or mineralocorticoid treatment was discontinued and blood pressure was stabilized if necessary, with a calcium antagonist, clonidine or metoprolol, at the discretion of the investigator. During the trial, patients received empagliflozin 10 mg/day, linagliptin 5 mg/day and telmisartan 80 mg/day in random order. Every study period consisted of 4 weeks treatment, with a 4 week wash-out period in between. The primary endpoint of the trial was the change in urinary albumin to creatinine ratio (UACR) from baseline over 4 weeks of treatment. Patients were invited to participate in a substudy to collect pharmacokinetic blood samples after the first dosing in each of the active treatment periods. The trial was performed in accordance with the Declaration of Helsinki and Good Clinical Practice. The study protocol was approved by the local medical ethics committee of the University Medical Center Groningen (UMCG), Groningen, the Netherlands. All participants signed written informed consent before any study-specific procedure commenced.
Pharmacokinetic measurements
In the pharmacokinetic sub-study, blood samples were obtained at pre-dose and 0.25, 0.5, 0.75, 1, 1.5, 2, 4 or 6 and 24 hours after dosing of empagliflozin, linagliptin or telmisartan. Plasma samples were stored at -80˚C before shipping (empagliflozin samples were shipped to BasI, West Lafayette, IN, USA, linagliptin samples were shipped to Covance, Harrogate, North Yorkshire, UK and telmisartan samples were shipped to Nuvisan, Neu-Ulm, Germany). Plasma concentrations of empagliflozin, linagliptin and telmisartan were quantitatively measured using a validated liquid chromatography with tandem mass spectroscopy (LC-MS/MS). All samples were analyzed in accordance with GLP guidelines. Data below the lower limit of quantification but before time to maximal plasma concentration were set to zero prior to data analysis. Data that were below the lower limit of quantification and after time to maximal plasma concentration were excluded from the analysis.
Patients collected three consecutive first morning void urine samples at day -2, -1 and 0 for measurement of urinary albumin and creatinine. UACR was calculated as the geometric mean of the three first morning void urine collections. Change from baseline in UACR was defined as the log ratio of the week 4 value divided by the baseline UACR value. The log transformation was applied to correct for the skewed distribution of UACR. Blood samples were collected every four weeks under fasted conditions for measuring a clinical chemistry panel. Every visit, office blood pressure measurements were taken by an automated device after 5 minutes rest. The average of three consecutive readings, spaced 1 to 2 minutes apart, was used for analysis. A clinical chemistry panel,
including HbA1c, albumin, bilirubin, creatinine, blood urea nitrogen and a lipid panel
consisting of LDL, HDL, and total cholesterol, was performed at the start and end of each treatment period.
Population pharmacokinetic model development
Population pharmacokinetic non-linear mixed effect models were developed to describe the individual pharmacokinetic profile and estimate individual exposure to the drug.
Models were estimated using a first-order conditional estimation method with interaction for all models. One- and two compartmental models with linear elimination and various absorption models were fitted to the data to determine the optimal structural model. Models were parameterized in physiological constants, e.g. apparent clearances (CL/F), apparent volumes of distribution (V/F) and were assumed to be log-normally distributed (Equation 1).
𝜃𝑖= 𝜃𝑝𝑜𝑝∙ 𝑒𝜂𝑖 𝜂 ~ 𝑁(0, 𝜔2) (1)
Where θi is the parameter estimate for the ith individual, θpop is the population parameter
estimate and ηi is the inter-individual variability for the ith individual, where η is normal
distributed with a mean of zero and a variance of ω2.
Proportional, additive or combined residual error structures (ɛ) were explored. After identifying the optimal structural model with inter-individual variability and optimizing the residual error model, a covariate analysis was performed to explain variability between patients. Covariates were formally tested during model development
Particular focus was put on previously published covariates, i.e. age, height, sex, BMI and eGFR (calculated with the Modification of Diet in Renal Disease equation) (9-12) as well as clinical parameters as blood pressure, albumin, bilirubin, creatinine, blood urea nitrogen and a lipid panel consisting of LDL, HDL, and total cholesterol. Also, allometric scaling was explored by normalizing clearances and distribution volumes to a 70 kg subject with exponents of 0.75 and 1 for CL and V respectively. (13)
Continuous covariates were implemented using a median-normalized (Equation 2) covariate structure. 𝜃𝑖= 𝜃𝑝𝑜𝑝 ∙ ( 𝐶𝑂𝑉𝑖 𝐶𝑂𝑉𝑚𝑒𝑑 ) 𝜃𝐶𝑂𝑉 ∙ 𝑒𝜂𝑖 (2)
Where COViis the covariate value for the ith individual, COVmed is the median of the
population and θcov is a parameter estimate for the covariate effect. Categorical
covariates, such as sex, were estimated proportionally (Equation 3).
𝜃𝑖= 𝜃𝑝𝑜𝑝∙ (1 + 𝜃𝐶𝑂𝑉) ∙ 𝑒𝜂𝑖 (3)
Population pharmacokinetic model evaluation
The minimum value of the objective function (MVOF) was used for model comparison during model development under the assumption that the difference in MVOF is following a Chi-Square distribution with degrees of freedom equal to the number of parameters added to the model. Therefore, with 1 additional degree of freedom, a model was statistically improved (p<0.05) if the decrease in MVOF was more than 3.84 points compared to its parent model.
Models were graphically evaluated using standard goodness-of-fit plots consisting of the individual predictions (IPREDs) and population predictions (PPRED) of the model vs. the observations and the conditional weighted residuals with interaction (CWRESI) vs. PPRED and TIME. (14)
Numerical evaluation comprised of percentage relative standard error (RSE%) for population parameter estimates and coefficient of variation (CV%) for inter-individual variation parameter values as shown
in Equation 4.
𝐶𝑉(%) = √exp (𝜔2) − 1 ∙ 100 (4)
RSE was considered acceptable when <50%. Physiological plausibility of the parameter estimates was also evaluated.
Exposure-response analysis
Area under the curve from zero to infinity (AUC0-∞) was calculated per patient by
dividing the administered dose over the individual clearance from the population pharmacokinetic model for each drug. After estimating individual exposure, the relationship with UACR, FPG and SBP was investigated. The relationship was investigated using linear regression analysis. Two-sided p-values <0.05 in a linear regression were used to indicate statistical significance. Pearson correlation coefficient was calculated to estimate the association.
Software
All data preparation, statistical analysis and graphical presentation was performed in R Version 3.6.1 and R Studio version 3.4.3 (Copyright © 2017 The R Foundation for Statistical Computing, Vienna, Austria). Population pharmacokinetic model development was performed using NONMEM version 7.3.0, ICON Development Solutions, Ellicott City, MD, USA, using the subroutine ADVAN13 with TOL9 for empagliflozin and linagliptin and ADVAN4 TRANS4 for telmisartan.
Results
Study population of the pharmacokinetic substudy
The pharmacokinetic substudy enrolled 29 patients: 9 patients for empagliflozin, 8 patients for linagliptin and 12 patients for telmisartan. One patient developed severe diarrhea during telmisartan treatment which resulted in early study medication discontinuation. Another patient receiving telmisartan showed extremely high plasma
concentrations and reported gastrointestinal side effects. As a consequence, these patients were excluded from the analysis. Overall, this resulted in the inclusion of 9 patients for empagliflozin, 8 patients for linagliptin and 10 patients for telmisartan. Demographics are shown in Table 1.
Population pharmacokinetic model development
The population pharmacokinetic dataset consisted of 235 concentration measurements (78 for empagliflozin, 68 for linagliptin and 89 for telmisartan). Parameter estimates for the final population pharmacokinetic models are reported in Table 2.
Table 1. Demographic characteristics at baseline per treatment period of the population
included in the pharmacokinetic sub-study. Data are presented as mean (SD) or median [IQR].
Empagliflozin 10 mg/day Linagliptin 5 mg/day Telmisartan 80 mg/day n 9 8 10 Age (years) 65.3 (9.0) 63.9 (8.6) 66.1 (9.3) Gender (male) (%) 6 (66.7) 4 (50.0) 7 (70.0) Bodyweight (kg) 98.2 (10.8) 90.5 (18.1) 102.1 (12.5)
Body mass index (kg/m2) 31.0 (1.8) 31.0 (2.7) 32.7 (2.7)
Systolic BP (mm Hg) 141.9 (14.8) 134.3 (16.6) 138.5 (10.7) Diastolic BP (mm Hg) 81.9 (7.3) 75.9 (7.3) 73.9 (10.0) eGFR (ml/min/1.73m2)* 79.2 (22.4) 85.1 (12.0) 76 (25.5) HbA1c (mmol/mol) 56.9 (10.8) 61.4 (8.8) 62.3 (12.6) Urinary albumin to creatinine ratio (mg/g) 404.8 [378.7 - 701.2] 296.7 [146.4 - 330.5] 296.5 [139.4 - 741.7] *estimated by MDRD formula
T ab le 2 . P ha rma co ki ne tic mo de l p op ul at io n pa ra me te r e st ima tio ns o f e mp agl ifl oz in , l in agl ip tin a nd t el mis ar ta n r ep or te d a s po pu la tio n p ar am et er s ( RS E %) a nd c oe ff ic ie nt o f v ar ia tio n ( C V% ). N /E : n ot e st im at ed. E m pa gl ifl oz in L ina gl ipt in Te lm is ar ta n Pa ra m et er Pa ra m et er D es cr ipt io n Pa ra m et er Es ti m at e RS E (% ) CV (% ) Pa ra m et er Es ti m at e RS E (% ) CV (% ) Pa ra m et er Es ti m at e RS E (% ) CV (% ) C l/F Ap pa re nt c le ar an ce fr om c en tr al co m pa rtm en t (L . h -1 ) 9.4 9.5 29.7 22.5 11.0 2.4 37.5 25.0 99.4 V2 /F Ap pa re nt v ol um e of dis tr ib ut io n fo r cen tr al co m pa rt m en t ( L) 50.0 13.5 12.7 858.0 18.3 58.2 570 24.6 101.0 Q /F Ap pa re nt in te rc om pa rt m en ta l cl ea ra nc e ( L . h -1 ) 14.9 25.5 N /E N /E N /E N /E 80.3 43.9 N /E V3 /F Ap pa re nt v ol um e of dis tr ib ut io n fo r pe riphe ra l co m pa rt m en t ( L) 31.8 29.6 N /E N /E N /E N /E 164 28.9 N /E MMT M ea n t ra ns it t im e 0.5 26.8 95.3 0.66 14.5 42.4 0.32 20.3 64.2 NN N um be r o f co m pa rtm en ts 4.4 22.4 60.4 8.58 39.2 133.0 1.58 37.4 N /E P ro po rt io na l r es idu al e rr or / addi tiv e r es idu al e rr or 0. 02 /0 .0 01 0. 01 6/0 .0 00 2 0. 09 2/ N /E
Population pharmacokinetic model development
Empagliflozin
Observed plasma concentrations of empagliflozin were best described by a two-compartment model with first order elimination and a transit absorption model. (15) Inter-individual variability (IIV) was identified for apparent clearance (CL/F), apparent volume of distribution (V2/F), mean transit time (MTT) and number of transit
compartments (NN). Also, the covariance could be estimated between CL/F and V2/F. The residual error was best described by a combined error model. In order to improve model stability, model parameters were µ-referenced.
The central and individual trend of the data is adequately captured by the model as the population predictions and individual predictions are evenly distributed around the line of unity (Figure 1.I A and B). In CWRES all data were normally distributed around zero over the entire TIME and PRED range (Figure 1-I C and D). Population parameters were accurately estimated, as reflected by RSE values of <30%. The covariate
bodyweight was implemented in the model using allometric scaling as this improved the goodness-of-fit plots. Other covariates could not be identified. Overall, our model described the individual data accurately (Supplementary figure 1) and the variability in the data was also captured adequately (Supplementary figure 2). Therefore, the model
was considered adequate to estimate individual exposure. The individual AUC0-∞ for
empagliflozin ranged from 411.73 to 1249.8 g·h/mL with a mean value of 885.1 g·h/mL and (95% CI: 468.6 to 1239.9).
Linagliptin
The observed plasma concentrations of Linagliptin were best described by a one-compartment model with first-order elimination and a transit absorption model. IIV could be identified on CL/F, V/F MTT and NN. The residual error was best described by a combined error model.
The central and individual trend of the data is well captured by the model as the population predictions and individual predictions are evenly distributed around the line of unity (Figure 1.II A and B). In CWRES all data were normally distributed around zero over the entire TIME and PRED range (Figure 1.II C and D). Population parameters were accurately estimated as reflected by RSE values of <40%. Covariate weight was implemented in the model using allometric scaling as the goodness-of-fit plots improved.
Other covariates could not be identified. Overall, our model described the individual data accurately (Supplementary figure 1) and the variability in the data was also captured adequately (Supplementary figure 2). Therefore, the model was considered adequate to
estimate individual exposure. The individual AUC0-∞ for linagliptin ranged from 173.4 to
243.1 g·h/mL with a mean AUC0-∞ of 203.3 g·h/mL (95% CI: 173.7 to 239.6).
Telmisartan
Telmisartan observed plasma concentrations were best described by a two-compartment model with first order elimination and first order absorption with a lag time. Inter-individual variability was identified on CL/F, V2/F and MTT. The residual error was best described by a proportional error model.
The central and individual trend of the data is well captured by the model as the population predictions and individual predictions are evenly distributed around the line of unity (Figure 1.III A and B). In CWRES all data were normally distributed around zero over the entire TIME and PRED range (Figure 1.III C and D). Population parameters were accurately estimated as reflected by RSE values of <45%. Sex was identified as a significant covariate on CL/F. Other covariates could not be identified. Overall, our model described the individual data accurately (Supplementary figure 1) and the variability in the data was also captured adequately (Supplementary figure 2). Therefore, the model considered adequate to estimate individual exposure. The
individual AUC0-∞ for telmisartan ranged from 736.0 to 9954.5 g·h/mL with a mean
AUC0-∞ of 4204.0 g·h/mL (95% CI: 747.1 to 9575.5).
Pharmacodynamic response analysis
The pharmacodynamic dataset was comprised of 150 first morning void albumin to creatinine ratio measurements (52 for empagliflozin, 39 for linagliptin and 59 for telmisartan).
Figure 2 shows the inter-individual variation in UACR response, fasting plasma glucose (FPG) response and systolic blood pressure response for the three different
drugs. Median percentage change from baseline (25th to 75th percentile) after 4 weeks of
treatment with empagliflozin in UACR response was -17.8% (-33.9 to -7.9%) for linagliptin 8.7% (-14.6 to 17.3%) and for telmisartan -40.0% ( -57.1 to -34.6%). The
glucose after treatment of 4 weeks with empagliflozin was -6.6% (-24.5 to 8.2) for linagliptin -5.4% (-6.8 to 9.1) and for telmisartan -7.5% (-15.0 to 4.3). The median
percentage change from baseline in systolic blood pressure (25th to 75th percentile) after
treatment of 4 weeks with empagliflozin was -2.9% (-6.4 to 4.2) for linagliptin 8.6% (3.6 to 11.6) and for telmisartan 0.4% (-4.1 to 7.6) (Figure 2).
The association between individual AUC0-∞ to UACR change, change in fasting plasma
glucose and change in systolic blood pressure for empagliflozin, linagliptin and telmisartan is shown in figure 3. For every 100 ng·hr/ml increment in area under the plasma concentration curve, the percentage change in log transformed UACR was 2.7% for empagliflozin (p=0.21; Pearson correlation coefficient (PCC) = 0.46), 48.1% for linagliptin (p=0.51; PCC=0.30) and -0.3% for telmisartan (p=0.08; PCC=0.58). The percentage change in fasting plasma glucose was 0.8% for empagliflozin (p=0.82), 60.6% for linagliptin (p=0.54) and 0.19% for telmisartan (p=0.27). The percentage change in systolic blood pressure was 0.62% (p=0.54) for empagliflozin, 4.2% (p=0.85) for linagliptin and 0.12% (p=0.44) for telmisartan.
Discussion
To identify whether individual differences in exposure can explain inter-individual variability in response to telmisartan, linagliptin, and empagliflozin, we successfully developed population pharmacokinetic models for these drugs and analyzed the exposure-response relationships for the pharmacodynamic markers UACR, fasting plasma glucose and systolic blood pressure. The individual exposure over time was well described by the model, but there was no clear indication that the individual exposure to these drugs explained the between-individual variability in response in these markers. The developed PK models described the data adequately with population PK parameters being estimated with acceptable accuracy and precision, and with limited remaining residual error. For the covariate analysis, we tested previous published covariates for their influence on the pharmacokinetic profiles of the individual patients. The limited number of subjects and thus subsequent low power may explain why previously reported covariates, such as eGFR (16), could not be identified in this study. However, in keeping with previous studies, body weight using allometric scaling was identified as important covariate in the empagliflozin and linagliptin model. (9,17) Sex was a covariate in the
telmisartan model in accordance with literature. (18) Despite the relatively small number of patients, the pharmacokinetic models accurately described the individual
pharmacokinetic profiles for empagliflozin, linagliptin and telmisartan which allowed us to obtain individual exposure in terms of AUC0-∞ for the exposure-response analysis.
The individual exposures to the three drugs did not correlate with responses in UACR, fasting plasma glucose and SBP. Various explanations may exist for the lack of correlation. First of all, it is possible that systemic exposure to these drugs truly do not associate with pharmacodynamic response. The relatively small sample size and inherent limited statistical power may be an explanation. The large day-to-day variability in UACR, fasting plasma glucose and systolic blood pressure may be another explanation. (19-21) Despite the fact that usage of multiple urine collections to define baseline and end of treatment UACR values, assessed fasting plasma glucose in fasted condition and used multiple blood pressure readings to minimize biological variability, the large day-to-day variations in the pharmacodynamics parameters may have obscured detection of a true correlation with individual exposure. Furthermore, it should be noted that systolic blood pressure data at baseline were obtained at the end of the morning, whereas the systolic blood pressure data at the subsequent visits were obtained in the early morning. The reduction in systolic blood pressure over the morning period, as a consequence of circadian rhythm may be subject dependent. (22) This may have introduced another source of variability and hampered assessment of individual exposure-response relation for this outcome. (23) Finally, we note that all measurements were performed in the systemic circulation which may not reflect the exposure at the drug target, such as the angiotensin-1-receptor or SGLT2 transporter in the kidney.
This study has limitations. First, while the sample size was sufficient to adequately develop PK models, there was limited power to assess the correlation between individual exposure and responses in pharmacodynamic parameters. Secondly, we enrolled a Caucasian population of type 2 diabetes patients with elevated albuminuria who were predominantly male. As a consequence, the developed PK models cannot be readily generalized to other populations with type 2 diabetes. Finally, the optimal effect of the drugs is influenced by patient adherence. To limit its influence on drug response, we have performed adherence checks by pill count.
I. GOF plots − Empagliflozin ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0 50 100 150 0 40 80 120 PRED O b s e rv e d A ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●●●● ● ● ● ● ●● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0 50 100 150 0 50 100 150 IPRED O b s e rv e d B ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −3 −2 −1 0 1 2 3 0 40 80 120 PRED C W R E S C ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −3 −2 −1 0 1 2 3 0 5 10 15 20 25 TIME C W R E S D
II. GOF plots − Linagliptin
●● ● ● ●●● ● ●● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● 0 5 10 15 0 2 4 6 PRED O b s e rv e d A ●● ● ● ●●● ● ● ●● ● ●●● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●●● ● ● ●● ● ● ●●●● ● ● ● ● ●●● ● ● ● ● ● ● ● ●● ● ● ● ● ● 0 5 10 15 0 5 10 15 IPRED O b s e rv e d B ● ● ● ● ●● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● −2 −1 0 1 2 0 2 4 6 PRED C W R E S C ● ● ●●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ●●● ● ● ● ● ● ● −2 −1 0 1 2 0 5 10 15 20 25 TIME C W R E S D
III. GOF plots − Telmisartan
● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● 0 100 200 300 400 0 25 50 75 100 125 PRED O b s e rv e d A ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ●●● ● ● ● ● ● ● 0 100 200 300 400 0 100 200 300 IPRED O b s e rv e d B ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −2 −1 0 1 2 3 0 25 50 75 100 125 C W R E S C ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ●● ● ● ●● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● −2 −1 0 1 2 3 0 5 10 15 20 25 C W R E S D
Figure 1. Goodness-of-fit plots for the
Population pharmacokinetic analysis of
Empagliflozin (I), Linagliptin (II) and Telmisartan (III). (a) Observed vs. population-predicted plasma concentrations (PRED); (b) Observed vs. individual predicted plasma concentrations (IPRED); (c) conditional- weighted residual (CWRES) vs. PRED; and (d) CWRES vs. time.
I. GOF plots − Empagliflozin
●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0 50 100 150 0 40 80 120 PRED O b s e rv e d A ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●●●● ● ● ● ● ●● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● 0 50 100 150 0 50 100 150 IPRED O b s e rv e d B ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −3 −2 −1 0 1 2 3 0 40 80 120 PRED C W R E S C ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −3 −2 −1 0 1 2 3 0 5 10 15 20 25 TIME C W R E S D
II. GOF plots − Linagliptin
●● ● ● ●●● ● ●● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● 0 5 10 15 0 2 4 6 PRED O b s e rv e d A ●● ● ● ●●● ● ● ●● ● ●●● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●●● ● ● ●● ● ● ●●●● ● ● ● ● ●●● ● ● ● ● ● ● ● ●● ● ● ● ● ● 0 5 10 15 0 5 10 15 IPRED O b s e rv e d B ● ● ● ● ●● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● −2 −1 0 1 2 0 2 4 6 PRED C W R E S C ● ● ●●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ●●● ● ● ● ● ● ● −2 −1 0 1 2 0 5 10 15 20 25 TIME C W R E S D
III. GOF plots − Telmisartan
● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● 0 100 200 300 400 0 25 50 75 100 125 PRED O b s e rv e d A ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●●●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ●●● ● ● ● ● ● ● 0 100 200 300 400 0 100 200 300 IPRED O b s e rv e d B ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● −2 −1 0 1 2 3 0 25 50 75 100 125 C W R E S C ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ●● ● ● ●● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● −2 −1 0 1 2 3 0 5 10 15 20 25 C W R E S D
Figure 2. Boxplot representing the
inter-individual variability in response in urine albumin to creatinine ratio (Delta UACR %), fasting plasma glucose (FPG) and systolic blood pressure (SBP) after treatment with empagliflozin 10 mg, linagliptin 5 mg or telmisartan 80 mg.
Association of exposure-response
Figure 3. The association between exposure (AUC0-∞) and change from baseline in urine albumin creatinine
ratio (UACR), fasting plasma glucose (FPG) and systolic blood pressure (SBP). Circles; observations, grey areas; 95% confidence interval.
Conclusion
We successfully described the individual pharmacokinetic profiles in patients with type 2 diabetes and diabetic nephropathy for empagliflozin, linagliptin and telmisartan.
Individual variation in AUC0-∞ of empagliflozin, linagliptin and telmisartan did not
explain the variation in UACR, blood pressure or plasma glucose changes during treatment with these drugs.
References
1. Rosenstock J, Perkovic V, Johansen OE, et al. Effect of linagliptin vs placebo on major cardiovascular events in adults with type 2 diabetes and high cardiovascular and renal risk: The CARMELINA randomized clinical trial. JAMA. 2019;321(1):69-79.
2. Coresh J, Heerspink HJL, Sang Y, et al. Change in albuminuria and subsequent risk of end-stage kidney disease: An individual participant-level consortium meta-analysis of observational studies. Lancet Diabetes Endocrinol. 2019;7(2):115-127.
3. Heerspink HJL, Greene T, Tighiouart H, et al. Change in albuminuria as a surrogate endpoint for progression of kidney disease: A meta-analysis of treatment effects in randomised clinical trials. Lancet Diabetes Endocrinol. 2019;7(2):128-139.
4. Levey AS, Gansevoort RT, Coresh J, et al. Change in albuminuria and GFR as end points for clinical trials in early stages of CKD: A scientific workshop sponsored by the national kidney foundation in collaboration with the US food and drug administration and european medicines agency. Am J Kidney Dis. 2019.
5. Groop PH, Cooper ME, Perkovic V, Emser A, Woerle HJ, von Eynatten M. Linagliptin lowers albuminuria on top of recommended standard treatment in patients with type 2 diabetes and renal dysfunction. Diabetes Care. 2013;36(11):3460-3468.
6. Heerspink HJL, Sjostrom CD, Inzucchi SE, et al. Reduction in albuminuria with dapagliflozin cannot be predicted by baseline clinical characteristics or changes in most other risk markers. Diabetes Obes Metab. 2019;21(3):720-725.
7. Bos H, Henning RH, De Boer E, et al. Addition of AT1 blocker fails to overcome resistance to ACE inhibition in adriamycin nephrosis. Kidney Int. 2002;61(2):473-480.
8. Petrykiv S, Laverman GD, de Zeeuw D, Heerspink HJL. Does SGLT2 inhibition with dapagliflozin overcome individual therapy resistance to RAAS inhibition? Diabetes Obes Metab. 2017.
9. Baron KT, Macha S, Broedl UC, Nock V, Retlich S, Riggs M. Population pharmacokinetics and exposure-response (efficacy and safety/tolerability) of empagliflozin in patients with type 2 diabetes. Diabetes Ther. 2016;7(3):455-471.
10. Retlich S, Duval V, Graefe-Mody U, et al. Population pharmacokinetics and pharmacodynamics of linagliptin in patients with type 2 diabetes mellitus. Clin Pharmacokinet. 2015;54(7):737-750.
11. Tatami S, Yamamura N, Sarashina A, Yong CL, Igarashi T, Tanigawara Y. Pharmacokinetic comparison of an angiotensin II receptor antagonist, telmisartan, in japanese and western hypertensive patients using population pharmacokinetic method. Drug Metab Pharmacokinet. 2004;19(1):15-23.
12. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: A new prediction equation. modification of diet in renal disease study group. Ann Intern Med. 1999;130(6):461-470.
13. West GB, Brown JH, Enquist BJ. A general model for the origin of allometric scaling laws in biology. Science. 1997;276(5309):122-126.
14. Nguyen TH, Mouksassi MS, Holford N, et al. Model evaluation of continuous data pharmacometric models: Metrics and graphics. CPT Pharmacometrics Syst Pharmacol. 2017;6(2):87-109.
15. Savic RM, Jonker DM, Kerbusch T, Karlsson MO. Implementation of a transit compartment model for describing drug absorption in pharmacokinetic studies. J Pharmacokinet Pharmacodyn. 2007;34(5):711-726. 16. Macha S, Mattheus M, Halabi A, Pinnetti S, Woerle HJ, Broedl UC. Pharmacokinetics,
pharmacodynamics and safety of empagliflozin, a sodium glucose cotransporter 2 (SGLT2) inhibitor, in subjects with renal impairment. Diabetes Obes Metab. 2014;16(3):215-222.
17. Retlich S, Duval V, Graefe-Mody U, et al. Population pharmacokinetics and pharmacodynamics of linagliptin in patients with type 2 diabetes mellitus. Clin Pharmacokinet. 2015;54(7):737-750.
18. Tatami S, Yamamura N, Sarashina A, Yong CL, Igarashi T, Tanigawara Y. Pharmacokinetic comparison of an angiotensin II receptor antagonist, telmisartan, in japanese and western hypertensive patients using population pharmacokinetic method. Drug Metab Pharmacokinet. 2004;19(1):15-23.
19. Chau NP, Bouhanick B, Mestivier D, Taki M, Marre M. Normal and abnormal day-to-day variability of urinary albumin excretion in control and diabetic subjects. Diabetes Metab. 2000;26(1):36-41.
20. Zawadzki MJ, Small AK, Gerin W. Ambulatory blood pressure variability: A conceptual review. Blood Press Monit. 2017;22(2):53-58.
21. Ollerton RL, Playle R, Ahmed K, Dunstan FD, Luzio SD, Owens DR. Day-to-day variability of fasting plasma glucose in newly diagnosed type 2 diabetic subjects. Diabetes Care. 1999;22(3):394-398. 22. Giles TD. Circadian rhythm of blood pressure and the relation to cardiovascular events. J Hypertens
Suppl. 2006;24(2):S11-6.
23. Bell KJ, Hayen A, Macaskill P, et al. Monitoring initial response to angiotensin-converting enzyme inhibitor-based regimens: An individual patient data meta-analysis from randomized, placebo-controlled trials. Hypertension. 2010;56(3):533-539.
Supplementary materials
Supplementary figure 1. Individual predicted plasma concentration time profiles per
patient (light grey). Open circles; observations, black line; population predicted plasma concentration time profiles.
Supplementary figure 2.
Visual predictive checks (VPCs) of empagliflozin, linagliptin and telmisartan. The lines represent the 10th, 50th and 90th percentiles of the observations. The shaded areas represent the 95% confidence intervals of the simulated concentrations of the 10th, 50th and 90th percentiles.
Linagliptin
