Search

Interpretation of kidney graft biopsies using the Banff classification is still heterogeneous. In this study, extreme gradient boosting classifiers learned from two large training datasets (n = 631 and 304 cases) where the "reference diagnoses" were not strictly defined following the Banff rules but from central reading by expert pathologists and further interpreted consensually by experienced transplant nephrologists, in light of the clinical context. In three external validation datasets (n = 3744, 589, and 360), the classifiers yielded a mean ROC curve AUC (95%CI) of: 0.97 (0.92-1.00), 0.97 (0.96-0.97), and 0.95 (0.93-0.97) for antibody-mediated rejection (ABMR); 0.94 (0.91-0.96), 0.94 (0.92-0.95), and 0.91 (0.88-0.95) for T cell-mediated rejection;0.96 (0.90-1.00) with all three for interstitial fibrosis-tubular atrophy. We also developed a classifier to discriminate active and chronic active ABMR with 95% accuracy. In conclusion, we built highly sensitive and specific artificial intelligence classifiers able to interpret kidney graft scoring together with a few clinical data and automatically diagnose rejection, with excellent concordance with the Banff rules and reference diagnoses made by a group of experts. Some discrepancies may point toward possible improvements that could be made to the Banff classification.

Slow-flow vascular malformations (VMs), especially those with venous components, can be complicated by localized intravascular coagulopathy (LIC), responsible for pain and impaired quality of life. Several studies have shown the effectiveness of mTOR inhibitors (especially sirolimus) on slow-flow VMs but its effect on coagulation has been poorly studied, especially in children. Our study shows that venous and combined VMs are associated with coagulation abnormalities and provides novel evidence that sirolimus improves coagulopathy in venous malformations. However we did not clearly evidence predictive biomarkers of response to sirolimus but this is the first study attempting to highlight predictive markers of response to sirolimus.

Maximum a posteriori Bayesian estimation (MAP-BE) based on a limited sampling strategy and a population pharmacokinetic model is frequently used to estimate pharmacokinetic parameters in individuals, however with some uncertainty (bias). Recent works have shown that the performance in individual estimation or pharmacokinetic parameters can be improved by combining population pharmacokinetic and machine learning algorithms.The objective of this work was to investigate the use of a hybrid machine learning/population pharmacokinetic approach to improve individual iohexol clearance estimation.The reference iohexol clearance values were derived from 500 simulated profiles (samples collected between 0.1 and 24.7 h) using a population pharmacokinetic model we recently developed in Monolix and obtained using all the concentration timepoints available. Xgboost and glmnet algorithms able to predict the error of MAP-BE clearance estimates based on a limited sampling strategy (0.1 h, 1 h, and 9 h) versus reference values were developed in a training subset (75%) and were evaluated in a testing subset (25%) and in 36 real patients.The MAP-BE limited sampling strategy estimated clearance was corrected by the machine learning predicted error leading to a decrease in root mean squared error by 29% and 24%, and in the percentage of profiles with the mean prediction error out of the ± 20% bias by 60% and 40% in the external validation dataset for the glmnet and Xgboost machine learning algorithms, respectively. These results were attributable to a decrease in the eta-shrinkage (shrinkage for a MAP-BE limited sampling strategy = 32.4%, glmnet = 18.2%, and Xgboost = 19.4% in the external dataset).In conclusion, this hybrid algorithm represents a significant improvement in comparison to MAP-BE estimation alone.

Estimation of vancomycin area under the curve (AUC) is challenging in the case of discontinuous administration. Machine learning approaches are increasingly used and can be an alternative to population pharmacokinetic (POPPK) approaches for AUC estimation. The objectives were to train XGBoost algorithms based on simulations performed in a previous POPPK study to predict vancomycin AUC from early concentrations and a few features (i.e. patient information) and to evaluate them in a real-life external dataset in comparison to POPPK.Six thousand simulations performed from 6 different POPPK models were split into training and test sets. XGBoost algorithms were trained to predict trapezoidal rule AUC a priori or based on 2, 4 or 6 samples and were evaluated by resampling in the training set and validated in the test set. Finally, the 2-sample algorithm was externally evaluated on 28 real patients and compared to a state-of-the-art POPPK model-based averaging approach.The trained algorithms showed excellent performances in the test set with relative mean prediction error (MPE)/ imprecision (RMSE) of the reference AUC = 3.3/18.9, 2.8/17.4, 1.3/13.7% for the 2, 4 and 6 samples algorithms respectively. Validation in real patient showed flexibility in sampling time post-treatment initiation and excellent performances MPE/RMSE1.5/12% for the 2 samples algorithm in comparison to different POPPK approaches.The Xgboost algorithm trained from simulation and evaluated in real patients allow accurate and precise prediction of vancomycin AUC. It can be used in combination with POPPK models to increase the confidence in AUC estimation.

Iohexol clearance has been proposed to estimate the glomerular filtration rate (GFR). A population pharmacokinetics (popPK) model was developed from heterogeneous patients. A Bayesian estimator (MAP-BE) based on a limited sampling strategy (LSS) was derived and evaluated in external patients.Full pharmacokinetic data (7-12 samples) from 172 patients receiving iohexol for measurement of their GFR (unstable and stable ICU patients, liver failure patients and kidney transplant patients) were split into development (n = 136) and validation (n = 36) datasets. A PopPK model was developed in Monolix and was used to develop MAP-BE based on LSS. Its performance for GFR estimation was evaluated in the validation set.A two-compartment model with first-order elimination best described the data. The final model included the type of patients on volume of distribution (Vd), clearance and intercompartmental constants, serum creatinine on clearance and body weight on Vd. The best LSS included samples at 0.1-1-9 h exhibiting a relative mean prediction error (MPE) (RMSE) = -3.7% (14.3%) and better performance than the Bröchner-Mortensen formula (-3.0%/17%). Split by type of patients, the highest interindividual variability and imprecision was observed in unstable ICU patients (MPE (RMSE) = 3.7% (18.8%)) while the best performances were obtained for renal transplant patients (MPE (RMSE) = 1.0% (5.8%)). All LSS that included samples before 9 hours for the third sample were associated with an increased imprecision.A single MAP-BE of iohexol based on a three-sample LSS for four heterogeneous populations was developed and allowed accurate estimation of GFR in kidney transplant patients, slightly biased in stable ICU patients and slightly imprecise in unstable ICU patients.

The aim of this work was to evaluate, in a large data set of renal transplant recipients, the intraindividual variability of the area under the curve (AUC)/predose concentration (C0) ratio in comparison with that of AUC, C0, AUC/dose, and C0/dose.Patients with at least 2 tacrolimus AUC estimation requests were extracted from the Immunosuppressant Bayesian dose Adjustment website, and relative variations between 2 consecutive visits for the different metrics were calculated and compared.Data from 1325 patients on tacrolimus (3827 measured C0 and estimated AUC) showed that the lowest mean relative variation between 2 consecutives visits was for the AUC/C0 ratio (95% confidence interval [CI] relative fold change = -43% to 44% for AUC/C0; 95% CI, -77% to 72% for AUC; 95% CI, -82% to 98% for AUC/dose; 95% CI, -81% to 80% for C0 and 95% CI, -94% to 117% for C0/dose. The correlation between 2 consecutive requests, whether close or far apart, was also best for the AUC/C0 ratio ( r = 0.33 and r = 0.34, respectively) in comparison with C0 ( r = 0.21 and r = 0.22, respectively) and AUC ( r = 0.19 and 0.28, respectively). Regression analysis between AUC0-24 and C0 showed that for some patients, the usual C0 targets translated into some very unusual AUC values. As the AUC/C0 ratio is quite stable during large periods, individualized C0 targets can be derived from the AUC targets, and an algorithm that estimates the individualized C0 was developed for situations in which prior AUC estimates are available or not.In this study, we confirmed in a large data set that the AUC/C0 ratio yields low intraindividual variability, whereas C0 shows the largest, and we propose to calculate individualized C0 targets based on this ratio.

BACKGROUND Immunosuppressant Bayesian Dose Adjustment (ISBA) is an online expert system that estimates the area under the curve (AUC) of immunosuppressive drugs through pharmacokinetic modelling and Bayesian estimation to propose dose adjustments to reach predefined exposure targets. The ISBA database was retrospectively analyzed to describe tacrolimus pharmacokinetics and exposure, evaluate the efficiency of ISBA dose recommendations, and propose tacrolimus AUC0-12h target ranges for pediatric renal allograft recipients treated with immediate release tacrolimus. METHODS The database included 1935 tacrolimus dose adjustment requests from 419 patients

BACKGROUND In children with cystic fibrosis (CF), the currently recommended amikacin dose ranges between 30 and 35 mg/kg/d; however, data supporting this dosing efficacy are lacking. In this article, the objectives were to develop a nonparametric pharmacokinetic population model for amikacin in children with CF and investigate the efficacy and toxicity at different dose rates for distinct minimum inhibitory concentration (MIC) clinical breakpoints using Monte Carlo simulations. METHODS Data from 94 children with CF (613 serum concentrations) from the Bordeaux University Hospital's CF-centre were analyzed. After determination of nonparametric pharmacokinetic population model parameters and associated influent covariates in Pmetrics, 1000 Monte Carlo simulations were performed for 7 different dose rates between 30 and 60 mg/kg/d, to predict the probability of obtaining peak serum amikacin ≥10 × MIC and trough level ≤2.5 mg/L, for MIC values between 1 and 16 mg/L. RESULTS The median (min-max) age and weight were 10 (0.3-17) years and 29 (6-71) kg, respectively, with only 2 children younger than 1 year of age. Body weight and creatinine clearance significantly impacted the amikacin volume of distribution and clearance. The mean relative bias/root mean squared error between observed and individual predicted concentrations was -0.68%/8.1%. Monte Carlo simulations showed that for sensitive bacteria with MICs ≤ 4, 30 mg/kg/d was most appropriate for a 100% success rate; for bacteria with MICs ≥ 8 [eg, Pseudomonas aeruginosa (MICamikacin = 8)], a dose of at least 40 mg/kg/d allowed a high success probability (90%), with a trough level below 2.5 mg/L. CONCLUSIONS For intermediate pathogens, a dose of at least 40 mg/kg/d can improve efficacy, with an acceptable calculated residual trough level in cases of normal or hyperfiltration. Because amikacin undergoes renal clearance, which is immature until 1 year of age, dosing recommendations for this age group may be markedly high, warranting cautious interpretation.

Given a high pharmacokinetic inter-individual variability and a low exposure target achievement, ganciclovir (GCV) therapeutic drug monitoring is sometimes used in children. We aimed to develop and validate Bayesian estimators based on limited sampling strategies for the estimation of GCV area under the concentration–time curve from 0 to 24 h in pediatric transplant recipients treated with valganciclovir (VGCV) or GCV. Solid organ transplant or stem-cell transplant recipients who received GCV or VGCV and had available GCV concentrations per standard of care were retrospectively included in this study for pharmacokinetic modeling and development of Bayesian estimators using the iterative two-stage Bayesian method. Validation datasets included additional child recipients of a solid organ transplant or stem-cell transplant, and child recipients of a kidney or liver transplant enrolled in a previous study. Various combinations of three or two sampling times, applicable in clinical practice, were assessed based on the relative mean bias, standard deviation, and the root mean square error in a development dataset and three independent validation datasets. In the development dataset, the mean bias/standard deviation/root mean square error for the 1 h/2 h/3 h and 1 h/3 h limited sampling strategies were − 1.4%/9.3%/9.1% and − 3.5%/12.2%/12.3%, respectively for GCV, while for VGCV, the mean bias/standard deviation/root mean square error for the 1 h/2 h/6 h and 1 h/6 h limited sampling strategies were 0.7%/13.5%/13.3% and − 0.1%/12.1%/11.8%, respectively. In the independent validation datasets, seven (13%) and five (14%) children would have had misclassifications of their exposure using these Bayesian estimators and limited sampling strategies for VGCV and GCV, respectively. Three plasma samples collected at 1 h/2 h/3 h and 1 h/2 h/6 h post-dose for GCV and VGCV respectively, are sufficient to accurately determine GCV area under the concentration–time curve from 0 to 24 h for pharmacokinetic-enhanced therapeutic drug monitoring.

Vancomycin is one of the antibiotics most used in neonates. Continuous infusion has many advantages over intermittent infusions, but no consensus has been achieved regarding the optimal initial dose. The objectives of this study were: to develop a Machine learning (ML) algorithm based on pharmacokinetic profiles obtained by Monte Carlo simulations using a population pharmacokinetic model (POPPK) from the literature, in order to derive the best vancomycin initial dose in preterm and term neonates, and to compare ML performances with those of an literature equation (LE) derived from a POPPK previously published.The parameters of a previously published POPPK model of vancomycin in children and neonates were used in the mrgsolve R package to simulate 1900 PK profiles. ML algorithms were developed from these simulations using Xgboost, GLMNET and MARS in parallel, benchmarked and used to calculate the ML first dose. Performances were evaluated in a second simulation set and in an external set of 82 real patients and compared to those of a LE.The Xgboost algorithm yielded numerically best performances and target attainment rates: 46.9% in the second simulation set of 400-600 AUC/MIC ratio vs. 41.4% for the LE model (p = 0.0018); and 35.3% vs. 28% in real patients (p = 0.401), respectively). The Xgboost model resulted in less AUC/MIC 600, thus decreasing the risk of nephrotoxicity.The Xgboost algorithm developed to estimate the initial dose of vancomycin in term or preterm infants has better performances than a previous validated LE and should be evaluated prospectively.

Aims To develop a pharmacokinetic model of isoniazid (INH) concentration taking into account demographic factors and genetic variables [N-acetyltransferase 2 (NAT2) genotype], and to propose an initial INH dosage that could maximize the probability of achieving the desired INH concentration. Methods A retrospective analysis was undertaken of INH concentration data collected from patients with tuberculosis in Tunisia. Results In total, 118 patients were included in this study. The one-compartment model [volume of distribution (V), elimination rate (Ke)] was found to have good predictive performance. Multi-variate analysis showed that NAT2 affected both V and Ke significantly, but age, gender and weight did not. Internal validation of the final model showed correlation of 0.95 between individual predicted INH concentration 3 h after drug intake (C3) and observed C3. External validation showed that percentage mean absolute prediction error and percentage root mean squared error were 9.11% (range 0.62–35.8%) and 11.6%, respectively. Monte-Carlo simulation showed that doses of at least 225 mg/24 h and at least 450 mg/24 h attained a therapeutic concentration in >80% of patients in the NAT2 slow acetylator group and the NAT2 rapid/intermediate acetylator group, respectively. Conclusion The pharmacokinetic model allowed optimization of individual dosing regimens of INH in patients with tuberculosis in Tunisia. This tool may facilitate improved efficacy of INH and prevent its toxicity in this population.

AIMS Ganciclovir (GCV) and its prodrug valganciclovir (VGCV) are first-line agents to prevent and treat cytomegalovirus in transplant recipients. There is high pharmacokinetic (PK) interindividual variability and PK data are scarce, especially in paediatric stem cell transplant (SCT) recipients. We sought to determine the optimal GCV and VGCV dosing in transplanted children. METHODS We conducted a single-centre retrospective population PK (POPPK) study of IV GCV and enteral VGCV in paediatric solid organ transplant (SOT) and SCT recipients. We included children who were transplanted and had available plasma GCV concentrations, done per standard of care. POPPK analysis was performed using a nonlinear mixed effects modelling approach with NONMEM. Optimal dosing was determined based on the achievement of the surrogate efficacy target: GCV 24 h area under the concentration-time curve (AUC0-24h ) of 40-60 mg.h.L-1 . RESULTS Fifty children with a median [range] age of 7.5 years [0.5-17.4] contributed 580 PK samples. A two-compartment model with first-order absorption with a lag time and first-order elimination fit the data well. Creatinine clearance and body weight (WT) were significant covariates for GCV clearance (CL); and WT for the volumes of distribution. IV GCV 15-20 mg.kg-1 .day-1 divided every 12 hours achieved the highest probability of target achievement (PTA) (33.0-33.8%). Enteral VGCV 30 and 40 mg.kg-1 .day-1 divided every 12 hours in children 0

The aim of this work is to estimate the area-under the blood concentration curve of tacrolimus (TAC) following b.i.d. or q.d. dosing in organ transplant patients, using Xgboost machine learning (ML) models. A total of 4,997 and 1,452 TAC interdose area under the curves (AUCs) from patients on b.i.d. and q.d. TAC, sent to our Immunosuppressant Bayesian Dose Adjustment expert system (www.pharmaco.chu-limoges.fr/) for AUC estimation and dose recommendation based on TAC concentrations measured at least at 3 sampling times (predose, ~ 1 and 3 hours after dosing) were used to develop 4 ML models based on 2 or 3 concentrations. For each model, data splitting was performed to obtain a training set (75%) and a test set (25%). The Xgboost models in the training set with the lowest root mean square error (RMSE) in a 10-fold cross-validation experiment were evaluated in the test set and in 6 independent full-pharmacokinetic (PK) datasets from renal, liver, and heart transplant patients. ML models based on two or three concentrations, differences between these concentrations, relative deviations from theoretical times of sampling, and four covariates (dose, type of transplantation, age, and time between transplantation and sampling) yielded excellent AUC estimation performance in the test datasets (relative bias < 5% and relative RMSE < 10%) and better performance than maximum a posteriori Bayesian estimation in the six independent full-PK datasets. The Xgboost ML models described allow accurate estimation of TAC interdose AUC and can be used for routine TAC exposure estimation and dose adjustment. They will soon be implemented in a dedicated web interface.

Tacrolimus has a narrow therapeutic range and requires dose adjustment, usually based on the trough blood concentration but preferably on the area under the concentration–time curve over 12 h post-dose (AUC0–12h). The single-arm, multicentre, clinical study IMPAKT aimed: (i) to develop, in de novo kidney transplant recipients, pharmacokinetic models and maximum a-posteriori Bayesian estimators for a generic, immediate-release, oral formulation of tacrolimus to estimate tacrolimus AUC0–12h at different post-transplant periods using a limited sampling strategy, and considering the CYP3A5*3 polymorphism as a covariate and (ii) to compare the performance of these Bayesian estimators to those previously developed for the original formulation. Thirty patients were enrolled and 29 provided nine blood samples over 9 h at day 7 and months 1 and 3 post-transplant. Tacrolimus blood profiles measured with liquid chromatography-tandem mass spectrometry were modelled using one-compartment, double gamma absorption, linear elimination models developed in-house. Different limited sampling strategies of three time-points within 4 h post-dose were tested for the maximum a-posteriori Bayesian estimator of tacrolimus AUC0–12h. The models and estimators were validated internally and their performance compared to that of models previously developed for the original formulation. The concentration–time curves, AUC0–12h/dose and trough blood concentration/dose exhibited wide inter-individual variability. The covariate-free pharmacokinetic models developed for the three post-transplant periods closely fitted the individual profiles. Maximum a-posteriori Bayesian estimators based on three different limited sampling strategies and no covariate yielded accurate AUC0–12h estimates, including for the five cytochrome P450 3A5 expressers and for the four patients without corticosteroids. The 0–1 h–3 h strategy finally chosen had very low bias (− 4.0 to − 2.5%) and imprecision (root mean square error 5.5–9.2%). The maximum a-posteriori Bayesian estimators previously developed for the reference product fitted the generic profiles with similar performance. We developed original pharmacokinetic models and accurate maximum a-posteriori Bayesian estimators to estimate patient exposure and adjust the dose of generic tacrolimus, and confirmed that the robust tools previously developed for the original formulation can be applied to this generic.

National audience; Recently, the International Association of Therapeutic Drug Monitoring (IATDMCT), that is the learning society for biological pharmacology and toxicology, issued recommendations on the therapeutic drug monitoring (TDM) of tacrolimus. This is the second consensus after the one issued in 2009. In this document, the role of tacrolimus TDM for the four principal transplanted organs is discussed. The analytical aspects, pharmacogenetics, TDM alternative approaches and the positioning of biomarkers are also presented. Stronger recommendations are about trough concentration targets in kidney and liver transplantation and for other indication of tacrolimus use. For the first time, an area under the curve of tacrolimus concentrations target is recommended for recipients management. Eventually, another set of recommendations are proposed for pharmacodynamic biomarkers used in patients’ follow-up.

Mycophenolic acid (MPA), the active compound of mycophenolate mofetil (MMF), is widely used in lupus nephritis treatment. Therapeutic drug monitoring of adults suggests that area under the concentration-time curve (AUC) of MPA (MPA-AUC) is associated with clinical outcomes, but childhood data are scarce.Retrospective study of 27 children with biopsy-proven lupus nephritis treated with MMF between 2008 and 2016. In 25 children, MPA-AUC was performed within 6 months after kidney biopsy and MMF initiation. Treatment response at 6 months was defined as normal or improved GFR by 25% compared with baseline, 50% reduction of proteinuria to 0.5 g/day or 50 mg/mmol, and no hematuria.A total of 62 MPA-AUC were analyzed in 27 patients. Overall median was 44 mg h/L (interquartile range [IQR] 33-54). Individual dose adaptation was required in 32 cases (52%) to achieve target AUC of 30-60 mg h/L. At 6 months, 14/25 patients were defined as responders (56%, median MPA-AUC 49 mg h/L (40-59)) and 11/25 as non-responders (44%, 29 mg h/L (24-38)). Patients with MPA-AUC levels 45, 30-45, and 30 mg h/L had 6-month response rates of 89% (8/9), 60% (6/10), and 0% (0/6), respectively. In a logistic regression model adjusted for age, sex, lupus nephritis classification, and time since MMF initiation, an MPA-AUC 45 mg h/L was significantly associated with therapeutic response (OR 3.6, 95% CI 2.4-9.5, p = 0.03).Therapeutic drug monitoring leading to individualized dosing may improve efficacy of MMF. MPA-AUC 45 mg h/L is associated with better response rate and may be considered as a target value in pediatric lupus nephritis.

This work aims to evaluate whether a machine learning approach is appropriate to estimate the glomerular filtration rate in intensive care unit patients based on sparse iohexol pharmacokinetic data and a limited number of predictors. Eighty-six unstable patients received 3250 mg of iohexol intravenously and had nine blood samples collected 5, 30, 60, 180, 360, 540, 720, 1080, and 1440 min thereafter. Data splitting was performed to obtain a training (75%) and a test set (25%). To estimate the glomerular filtration rate, 37 candidate potential predictors were considered and the best machine learning approach among multivariate-adaptive regression spline and extreme gradient boosting (Xgboost) was selected based on the root-mean-square error. The approach associated with the best results in a ten-fold cross-validation experiment was then used to select the best limited combination of predictors in the training set, which was finally evaluated in the test set. The Xgboost approach yielded the best performance in the training set. The best combination of covariates comprised iohexol concentrations at times 180 and 720 min; the relative deviation from these theoretical times; the difference between these two concentrations; the Simplified Acute Physiology Score II; serum creatinine; and the fluid balance. It resulted in a root-mean-square error of 6.2 mL/min and an r2 of 0.866 in the test set. Interestingly, the eight patients in the test set with a glomerular filtration rate

Therapeutic drug monitoring (TDM) is mandatory for the immunosuppressive drug tacrolimus (Tac). For clinical applicability, TDM is performed using morning trough concentrations. With recent developments making tacrolimus concentration determination possible in capillary microsamples and Bayesian estimator predicted area under the concentration curve (AUC), AUC-guided TDM may now be clinically applicable. Tac circadian variation has, however, been reported, with lower systemic exposure following the evening dose. The aim of the present study was to investigate tacrolimus pharmacokinetic (PK) after morning and evening administrations of twice-daily tacrolimus in a real-life setting without restrictions regarding food and concomitant drug timing. Two 12 hour tacrolimus investigations were performed; after the morning dose and the following evening dose, respectively, in 31 renal transplant recipients early after transplantation both in a fasting-state and under real-life nonfasting conditions (14 patients repeated the investigation). We observed circadian variation under fasting-conditions: 45% higher peak-concentration and 20% higher AUC following the morning dose. In the real-life nonfasting setting, the PK-profiles were flat but comparable after the morning and evening doses, showing slower absorption rate and lower AUC compared with the fasting-state. Limited sampling strategies using concentrations at 0, 1, and 3 hours predicted AUC after fasting morning administration, and samples obtained at 1, 3, and 6 hours predicted AUC for the other conditions (evening and real-life nonfasting). In conclusion, circadian variation of tacrolimus is present when performed in patients who are in the fasting-state, whereas flatter PK-profiles and no circadian variation was present in a real-life, nonfasting setting.

Aims Intravenous mycophenolate mofetil (IV MMF), a prodrug of mycophenolic acid (MPA), is used during nonmyeloablative and reduced-intensity conditioning haematopoetic stem cell transplantation (HCT) to improve engraftment and reduce graft-versus-host disease. The aims of this study were to develop population pharmacokinetic models and Bayesian estimators based on limited sampling strategies to allow for individual dose adjustment of intravenous mycophenolate mofetil administered by infusion in haematopoietic stem cell transplant patients. Methods Sixty-three MPA concentration-time profiles (median [min-max] = 6 [4-7] samples) were collected from 34 HCT recipients transplanted for 14 (1-45) days and administered IV MMF every 8 hours, concomitantly with cyclosporine. The database was split into development (75%) and validation (25%) datasets. Pharmacokinetic models characterized by a single compartment with first-order elimination, combined with two gamma distributions to describe the transformation of MMF into mycophenolic acid, were developed using in parallel nonparametric (Pmetrics) and parametric (ITSIM) approaches. The performances of the models and the derived Bayesian estimators were evaluated in the validation set. Results The best limited sampling strategy led to a bias (min, max), root mean square error between observed and modeled interdose areas under the curve in the validation dataset of -11.72% (-31.08%, 5.00%), 14.9% for ITSIM and -2.21% (-23.40%, 30.01%), 12.4% for Pmetrics with three samples collected at 0.33, 2 and 3 hours post dosing. Conclusion Population pharmacokinetic models and Bayesian estimators for IV MMF in HCT have been developed and are now available online (https://pharmaco.chu-limoges.fr) for individual dose adjustment based on the interdose area under the curve.

Genetic variations in CYP3A4, CYP3A5, and m-TOR could contribute to interpatient variability regarding m-TOR inhibitors pharmacokinetics or cellular effects. The purpose of this study was to evaluate the influence of selected candidate variations in these genes on everolimus pharmacokinetics, efficacy, and toxicity in cancer patients. Thirty-four patients receiving everolimus for breast (n = 22) or renal (n = 10) cancers, or neuroendocrine tumors of pancreatic origin (n = 2) were included in the study. Six variants in genes related to everolimus pharmacokinetics (CYP3A4*22 and CYP3A5*3) or pharmacodynamics (m-TOR rs2295079, rs2295080, rs2024627 and rs1057079) were genotyped. Associations with trough concentrations (C0), dose reductions, or treatment interruptions due to toxicity and progression-free survival were investigated using generalized estimating equations and Cox models. CYP3A5 nonexpressers had significantly higher C0 as compared with expressers (βGG vs AG = + 6.32 ± 2.22 ng/mL, p = 0.004). m-TOR rs2024627 was significantly associated with an increased risk of cancer progression studied alone or as part of an haplotype (T vs C: HR = 2.60, 95% CI [1.16–5.80], p = 0.020; CTCG vs other haplotypes HR = 2.29, 95% CI [1.06–4.95], p = 0.035, respectively). This study showed that CYP3A5 expression impacts everolimus pharmacokinetics in cancer patients and identified a genetic variation in m-TOR associated with the risk of cancer progression.

Mycophenolate mofetil (MMF), a pro-drug of mycophenolic acid (MPA), has become a major therapeutic option in juvenile systemic lupus erythematosus (jSLE). Monitoring MPA exposure using area under curve (AUC) has proved its value to increase efficacy and safety in solid organ transplantation both in children and adults, but additional data are required in patients with autoimmune diseases. In order to facilitate MMF therapeutic drug monitoring (TDM) in children, Bayesian estimators (BE) of MPA AUCPharmacokinetic blood samples were collected from jSLE treated by MMF and MPA plasma concentrations were determined using high-performance liquid chromatography system with ultraviolet detection (HPLC-UV). Individual AUCWe received 41 rich pediatric PK at steady state from jSLE and calculated individual AUCAccording to our external validation of two LSS of drug exposure, the ISBA model is recommended for Bayesian estimation of MPA AUC

Background: Mycophenolate mofetil (MMF) is the most widely used second-line agent in auto-immune hepatitis (AIH). It is generally titrated up to patient response and continued for at least two years following complete liver enzyme normalization. However, in this maintenance phase individual dose adjustment to reach mycophenolic acid (MPA) exposure with the best benefit-risk probability may avoid adverse outcomes. The aim of the present study was to develop population pharmacokinetic (popPK) models and Maximum A-Posteriori Bayesian estimators (MAP-BEs) to estimate MPA inter-dose area under the curve (AUC0-12h) in AIH patients administered MMF using nonlinear mixed effect modelling. Methods: We analysed 50 MPA PK profiles from 34 different patients, together with some demographic, clinical, and laboratory test data. The median number of samples per profile, immediately preceding and following the morning MMF dose, was 7 [4 – 10]. PopPK modeling was performed using parametric, top-down, nonlinear mixed effect modelling with NONMEM 7.3. MAP-BEs were developed based on the the best popPK model and the best limited sampling strategy (LSS) selected among several. Results: The pharmacokinetic data were best described by a 2-compartment model, Erlang distribution to describe the absorption phase, and a proportional error. The best MAP-BE relied on the LSS at 0.33, 1 and 3 hours after mycophenolate mofetil dose administration and was very accurate (bias=5.6%) and precise (RMSE

Population pharmacokinetics consists in analyzing pharmacokinetic (PK) data collected in groups of individuals. Population PK is widely used to guide drug development and to inform dose adjustment via therapeutic drug monitoring (TDM) and model-informed precision dosing (MIPD). There are two main types of population PK methods: parametric (P) and nonparametric (NP). The characteristics of P and NP population methods have been previously reviewed. The aim of this article is to answer some frequently asked questions that are often raised by scholars, clinicians and researchers about P and NP population PK methods. The strengths and limitations of both approaches are explained, and the characteristics of the main software programs are presented. We also review the results of studies that compared the results of both approaches in the analysis of real data. This opinion paper may be informative for potential users of population methods in PK and guide them in the selection and use of those tools. It also provides insights on future research in this area. This article is protected by copyright. All rights reserved.

BackgroundThe Banff classification standardizes the diagnoses of kidney transplant rejection based on histological criteria. Clinical decisions are generally made after integration of the Banff diagnoses in the clinical context. However, interpretation of the biopsy cases is still heterogeneous among pathologists or clinicians. Machine Learning (ML) algorithms may be trained from expertly assessed cases to provide clinical decision support.MethodsThe ML technique of Extreme Gradient Boosting learned from two large training datasets from the European programs BIOMARGIN and ROCKET (n= 631 and 304), in which biopsies were read centrally and consensually interpreted by a group of experts and used as a reference for untargeted biomarker screenings. The model was then externally validated in three independent datasets (n= 3744, 589 and 360).ResultsIn the three validation datasets, the algorithm yielded a ROC curve AUC of mean (95% CI) 0.97 (0.92-1.00), 0.97 (0.96-0.97) and 0.95 (0.93-0.97) for antibody-mediated rejection (ABMR); 0.94 (0.91-0.96), 0.94 (0.92-0.95) and 0.91 (0.88-0.95) for T cell-mediated rejection; >0.96 (0.90-1.00) in all three for interstitial fibrosis - tubular atrophy (IFTA). Finally, using the largest validation cohort, we developed an additional algorithm to discriminate active and chronic active ABMR with an accuracy of 0.95.ConclusionWe built an Artificial Intelligence algorithm able to interpret histological lesions together with a few routine clinical data with very high sensitivity and specificity. This algorithm should be useful in routine or clinical trials to help pathologists and clinicians and increase biopsy interpretation homogeneity.

Importance Sirolimus is increasingly being used to treat various vascular anomalies, although evidence of its efficacy is lacking. Objective To assess the efficacy and safety of sirolimus for children with slow-flow vascular malformations to better delineate the indications for treatment. Design, setting and participants This multicenter, open-label, observational-phase randomized clinical trial included 59 children aged 6 to 18 years with a slow-flow vascular malformation who were recruited between September 28, 2015, and March 22, 2018, in 11 French tertiary hospital centers. Statistical analysis was performed on an intent-to-treat basis from December 4, 2019, to November 10, 2020. Interventions Patients underwent an observational period, then switched to an interventional period when they received oral sirolimus (target serum levels, 4-12 ng/mL). The switch time was randomized from month 4 to month 8, and the whole study period lasted 12 months for each patient. Main outcomes and measures The primary outcome was change in the volume of vascular malformations detected on magnetic resonance imaging scan (with centralized interpretation) per unit of time (ie, between the interventional period and the observational period). Secondary outcomes included subjective end points: pain, bleeding, oozing, quality of life, and safety. Results Among the participants (35 girls [59.3%]; mean [SD] age, 11.6 [3.8] years), 22 (37.3%) had a pure venous malformation, 18 (30.5%) had a cystic lymphatic malformation, and 19 (32.2%) had a combined malformation, including syndromic forms. Variations in the volume of vascular malformations detected on magnetic resonance imaging scans associated with the duration period were not overall significantly different between the interventional period and the observational period (all vascular malformations: mean [SD] difference, -0.001 [0.007]; venous malformations: mean [SD] difference, 0.001 [0.004]; combined malformations: mean [SD] difference, 0.001 [0.009]). However, a significant decrease in volume was observed for children with pure lymphatic malformations (mean [SD] difference, -0.005 [0.005]). Overall, sirolimus had positive effects on pain, especially for combined malformations, and on bleeding, oozing, self-assessed efficacy, and quality of life. During sirolimus treatment, 56 patients experienced 231 adverse events (5 serious adverse events, none life-threatening). The most frequent adverse event was an oral ulcer (29 patients [49.2%]). Conclusions and relevance This observational-phase randomized clinical trial allows for clarifying the goals of patients and families when starting sirolimus therapy for children older than 6 years. Pure lymphatic malformations seem to be the best indication for sirolimus therapy because evidence of decreasing lymphatic malformation volume per unit of time, oozing, and bleeding and increasing quality of life was found. In combined malformations, sirolimus significantly reduced pain, oozing, and bleeding. Benefits seemed lower for pure venous malformations than for the 2 other subgroups, also based on symptoms. Trial registration ClinicalTrials.gov Identifier: NCT02509468; clinicaltrialsregister.eu Identifier: 2015-001096-43.

Everolimus is a mammalian target of rapamycin (m-TOR) inhibitor that has been approved for the treatment of hormone receptor-positive advanced breast cancer, metastatic renal cancer, and neuroendocrine tumors. Although therapeutic drug monitoring (TDM) of everolimus is well established in the transplantation field, it is not currently performed in oncology. The last consensus conference about the TDM of everolimus states that for the use of everolimus in oncology, "further studies are required to determine the clinical utility of TDM for everolimus in oncology settings." In this review, the authors will discuss the current evidences and perspectives, based on observational studies available, in favor of the TDM of everolimus in oncology focusing on (1) the management of everolimus in routine practice, (2) the prerequisites for TDM of everolimus in oncology, (3) the pharmacodynamics (including a description of the biomarker of resistance and mutations in m-TOR), and (4) a general outlook.

In critically ill patients treated with meropenem, the proposed pharmacokinetics/pharmacodynamics (PK/PD) efficacy index is to keep the free drug concentration 4–5 times above the minimum inhibitory concentration (MIC) of the germ isolated, for 100% of the interval regimen. The objectives were to design a population pharmacokinetics model for meropenem in critically ill patients and to evaluate different dosage schemes that achieve the optimal PK/PD objectives. This retrospective, observational, single-centre study included 80 critically ill patients (154 samples) treated with meropenem between May 2011 and December 2017. Patient data, concentrations, treatment and bacteriological variables were collected from electronic medical records. Total and free concentrations of meropenem were modelled in Pmetrics. Monte Carlo simulations were performed to assess the probability of achieving the PK/PD target for different dosage regimens. For patients with available data, the number of patients with a free concentration 4 times higher or lower than the observed MIC for the P. aeruginosa and E. coli was investigated. A one-compartment model with first-order elimination adequately described serum total and free meropenem concentrations. The only variable that significantly influenced the elimination constant of meropenem was the creatinine clearance (CLcr) calculated using the CKD-EPI formula. The highest probability of achieving the pharmacodynamic objective was with 3-h infusion dosage regimens. Sixty percent and 89% of patients attained a free drug concentration 4 times above the MIC for P. aeruginosa and E. coli respectively. This study proposed different dosing regimens depending on renal clearance strata and the MIC of the germ targeted.

Aims Tacrolimus is a narrow therapeutic range drug that requires fine dose adjustment, for which pharmacokinetic (PK) models have been amply proposed in renal, but not in liver, transplant recipients. This study aimed to build population PK models and Bayesian estimators (BEs) in adult de novo liver transplant patients receiving either the immediate-release (Prograf, twice daily, TD) or prolonged-release (Advagraf, once daily, OD) forms to help PK-guided dose individualization. Methods In total, 160 tacrolimus concentration-time profiles (1654 samples) were collected from 80 patients, at day 7 (D7) and week 6 (W6) post-transplant. Four population PK models were developed using in-parallel parametric and nonparametric approaches for each formulation and period post-transplant. The best limited sampling strategies for estimating the area-under-the-curve (AUC) were selected by comparing predicted values to an independent dataset. Finally, the doses required to reach AUC targets were estimated using each BE and compared to the doses obtained using the trapezoidal AUC. Results Tacrolimus PK was best described using a 1-compartmental model with first-order elimination and 2 γ-distributions to describe the absorption. In the validation datasets, Bayesian AUC estimates yielded mean bias/root mean squared prediction error of −5.06%/13.43% (OD D7), 2.25%/8.51% (OD W6), −2.36%/7.27% (TD D7) and 0.87%/9.07% (TD W6) for the in-parallel parametric approach; and 8.95%/17.84% (OD D7), −0.11%/10.13% (OD W6), 3.57%/18.40% (TD D7) and 4.48%/12.59% (TD W6) for the nonparametric approach. Conclusion The BEs and limited sampling strategies proposed here are able to predict accurately and precisely tacrolimus AUC in liver patients using only 3 plasma concentrations. The dosing methods are available on our ImmunoSuppressive Bayesian dose Adjustment website (www.pharmaco.chu-limoges.fr).

The efficiency of new generation sequencing methods and the reduction of their cost has led pharmacogenomics to gradually supplant pharmacogenetics, leading to new applications in personalized medicine along with new perspectives in drug design or identification of drug response factors. The amount of data generated in genomics fits the definition of big data, and need a specific bioinformatics processing following standard steps: data collection, processing, analysis and interpretation. Pitfalls of pharmacogenomics studies are directly related to these steps. This review aims to describe these steps from a pharmacogenomic point of view, focusing on bioinformatics aspects.

Given a high pharmacokinetic inter-individual variability and a low exposure target achievement, ganciclovir (GCV) therapeutic drug monitoring is sometimes used in children. We aimed to develop and validate Bayesian estimators based on limited sampling strategies for the estimation of GCV area under the concentration-time curve from 0 to 24 h in pediatric transplant recipients treated with valganciclovir (VGCV) or GCV.Solid organ transplant or stem-cell transplant recipients who received GCV or VGCV and had available GCV concentrations per standard of care were retrospectively included in this study for pharmacokinetic modeling and development of Bayesian estimators using the iterative two-stage Bayesian method. Validation datasets included additional child recipients of a solid organ transplant or stem-cell transplant, and child recipients of a kidney or liver transplant enrolled in a previous study. Various combinations of three or two sampling times, applicable in clinical practice, were assessed based on the relative mean bias, standard deviation, and the root mean square error in a development dataset and three independent validation datasets.In the development dataset, the mean bias/standard deviation/root mean square error for the 1 h/2 h/3 h and 1 h/3 h limited sampling strategies were - 1.4%/9.3%/9.1% and - 3.5%/12.2%/12.3%, respectively for GCV, while for VGCV, the mean bias/standard deviation/root mean square error for the 1 h/2 h/6 h and 1 h/6 h limited sampling strategies were 0.7%/13.5%/13.3% and - 0.1%/12.1%/11.8%, respectively. In the independent validation datasets, seven (13%) and five (14%) children would have had misclassifications of their exposure using these Bayesian estimators and limited sampling strategies for VGCV and GCV, respectively.Three plasma samples collected at 1 h/2 h/3 h and 1 h/2 h/6 h post-dose for GCV and VGCV respectively, are sufficient to accurately determine GCV area under the concentration-time curve from 0 to 24 h for pharmacokinetic-enhanced therapeutic drug monitoring.

Ganciclovir and valganciclovir are first choice drugs for the prevention and treatment of cytomegalovirus infection and disease in solid organ and stem cell transplant recipients. Only a few studies on the pharmacokinetics and exposure/efficacy or exposure/safety relationships of ganciclovir and valganciclovir in transplant recipients have been published so far, and there are still controversies about the exposure parameter to use for therapeutic drug monitoring (TDM). We performed an extensive literature review of the clinical pharmacokinetics data, the exposure/effect relationships in terms of efficacy and safety, and the available tools for valganciclovir and ganciclovir TDM in adults and pediatrics transplant recipients. The pharmacokinetics of ganciclovir and valganciclovir is well described in adults and children, and a high interindividual variability is commonly observed. In contrast, the drug pharmacodynamics has been poorly described in adults and barely in children. The average 24-hour area under the concentration-time curve (AUC0-24h ) seems to be the best predictor of efficacy and toxicity. The benefit of TDM remains controversial in adult patients but should be considered in children due to higher interindividual variability and lower probability of target attainment. Several bayesian estimators based on limited sampling strategies have been developed with this aim and may be used in clinical practice for the AUC-based individual dose adjustment of ganciclovir and valganciclovir.

The demand for organs to be transplanted increases pressure on procurement centers, to the detriment of organ quality, increasing complications. New preservation protocols are urgently needed, requiring an in-depth understanding of ischemia-reperfusion mechanisms. We performed a proteomic analysis using LC-MS/MS-TOF data analyzed through R software and Cytoscape's ClueGO application, comparing the proteome of kidney endothelial cells, key cell type, subjected to 3, 6, 12, 19, and 24 h of cold ischemia and 6 h reperfusion. Critical pathways such as energy metabolism, cytoskeleton structure/transport system, and gene transcription/translation were modulated. Important time windows were revealed: a—during the first 3 h, central proteins were upregulated within these pathways, b—the majority of these upregulations were maintained until 12 h cold ischemia time (CIT), c—after that time, the overall decrease in protein expression was observed, d—at reperfusion, proteins expressed in response to cold ischemia were all downregulated. This shows that cold ischemia is not a simple slowing down of metabolism, as deep changes take place within the proteome on major pathways. Time-sensitive expression of key protein reveals possible quality biomarkers as well as potential targets for new strategies to maintain or optimize organ quality.

We previously demonstrated that Machine learning (ML) algorithms can accurately estimate drug area under the curve (AUC) of tacrolimus or mycophenolate mofetil (MMF) based on limited information, as well as or even better than maximum a posteriori Bayesian estimation (MAP-BE). However, the major limitation in the development of such ML algorithms is the limited availability of large databases of concentration vs. time profiles for such drugs. The objectives of this study were: (i) to develop a Xgboost model to estimate tacrolimus inter-dose AUC based on concentration-time profiles obtained from a literature population pharmacokinetic (POPPK) model using Monte Carlo simulation; and (ii) to compare its performance with that of MAP-BE in external datasets of rich concentration-time profiles. The population parameters of a previously published PK model were used in the mrgsolve R package to simulate 9000 rich interdose tacrolimus profiles (one concentration simulated every 30 min) at steady-state. Data splitting was performed to obtain a training set (75%) and a test set (25%). Xgboost algorithms able to estimate tacrolimus AUC based on 2 or 3 concentrations were developed in the training set and the model with the lowest RMSE in a ten-fold cross-validation experiment was evaluated in the test set, as well as in 4 independent, rich PK datasets from transplant patients. ML algorithms based on 2 or 3 concentrations and a few covariates yielded excellent AUC estimation in the external validation datasets (relative bias 5% and relative RMSE 10%), comparable to those obtained with MAP-BE. In conclusion, Xgboost machine learning models trained on concentration-time profiles simulated using literature POPPK models allow accurate tacrolimus AUC estimation based on sparse concentration data. This study paves the way to the development of artificial intelligence at the service of precision therapeutic drug monitoring in different therapeutic areas.

When mycophenolic acid (MPA) was originally marketed for immunosuppressive therapy, fixed doses were recommended by the manufacturer. Awareness of the potential for a more personalized dosing has led to development of methods to estimate MPA area under the curve based on the measurement of drug concentrations in only a few samples. This approach is feasible in the clinical routine and has proven successful in terms of correlation with outcome. However, the search for superior correlates has continued, and numerous studies in search of biomarkers that could better predict the perfect dosage for the individual patient have been published. As it was considered timely for an updated and comprehensive presentation of consensus on the status for personalized treatment with MPA, this report was prepared following an initiative from members of the International Association of Therapeutic Drug Monitoring and Clinical Toxicology (IATDMCT). Topics included are the criteria for analytics, methods to estimate exposure including pharmacometrics, the potential influence of pharmacogenetics, development of biomarkers, and the practical aspects of implementation of target concentration intervention. For selected topics with sufficient evidence, such as the application of limited sampling strategies for MPA area under the curve, graded recommendations on target ranges are presented. To provide a comprehensive review, this report also includes updates on the status of potential biomarkers including those which may be promising but with a low level of evidence. In view of the fact that there are very few new immunosuppressive drugs under development for the transplant field, it is likely that MPA will continue to be prescribed on a large scale in the upcoming years. Discontinuation of therapy due to adverse effects is relatively common, increasing the risk for late rejections, which may contribute to graft loss. Therefore, the continued search for innovative methods to better personalize MPA dosage is warranted. Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved

Therapeutic drug monitoring of mycophenolic acid (MPA) based on area under the curve (AUC) is well-established and machine learning (ML) approaches could help to estimate AUC. The aim of this work is to estimate the AUC of MPA in organ transplant patients using extreme gradient boosting (Xgboost R package) ML models. A total of 12,877 MPA AUC from 0 to 12 hours (AUC0-12 h ) requests from 6,884 patients sent to our Immunosuppressant Bayesian Dose Adjustment expert system (https://abis.chu-limoges.fr) for AUC estimation and dose recommendation based on MPA concentrations measured at least at three sampling times (~ 20 minutes, 1 and 3 hours after dosing) were used to develop two ML models based on two or three concentrations. Data were split into a training set (75%) and a test set (25%) and the Xgboost models in the training set with the lowest root mean squared error (RMSE) in a 10-fold cross-validation experiment were evaluated in the test set and in 4 independent full-pharmacokinetic (PK) datasets from renal or heart transplant recipients. ML models based on two or three concentrations, differences between these concentrations, relative deviations from theoretical times of sampling, presence of a delayed absorption peak, and five covariates (dose, type of transplantation, associated immunosuppressant, age, and time between transplantation and sampling) yielded accurate AUC estimation performances in the test datasets (relative bias < 5% and relative RMSE < 20%) and better performance than MAP Bayesian estimation in the four independent full-PK datasets. The Xgboost ML models described allow accurate estimation of MPA AUC0-12 h and can be used for routine exposure estimation and dose adjustment and will soon be implemented in a dedicated web interface.