Combining mathematical modeling and deep learning to make rapid and explainable predictions of the patient-specific response to anticoagulant therapy under venous flow.

Anticoagulant drugs are commonly prescribed to prevent hypercoagulable states in patients with venous thromboembolism. The choice of the most efficient anticoagulant and the appropriate dosage regimen remain a complex problem because of the intersubject variability in the coagulation kinetics and the effect of blood flow. The rapid assessment of the patient-specific response to anticoagulant regimens would assist clinical decision-making and ensure efficient management of coagulopathy. In this work, we introduce a novel approach that combines computational modeling and deep learning for the fast prediction of the patient-specific response to anticoagulant regimens. We extend a previously developed model to explore the spatio-temporal dynamics of thrombin generation and thrombus formation under anticoagulation therapy. Using a 1D version of the model, we generate a dataset of thrombus formation for thousands of virtual patients by varying key parameters in their physiological range. We use this dataset to train an artificial neural network (ANN) and we use it to predict patient's response to anticoagulant therapy under flow. The algorithm is available and can be accessed through the link: https://github.com/MPS7/ML_coag. It yields an accuracy of 96 % which suggests that its usefulness can be assessed in a randomized clinical trial. The exploration of the model dynamics explains the decisions taken by the algorithm.

Needs for re-intervention on restored teeth in adults: a practice-based study.

OBJECTIVES: Evaluate the need for re-intervention on dental coronal restorations in adults seen in a network of general dental practitioners (ReCOL).  MATERIALS AND METHODS: This observational, cross-sectional, multicenter study involved 40 practitioners and 400 patients. Coronal restoration failures (needing re-intervention for unsatisfactory outcomes) were assessed with a simplified rating scale of seven criteria from the FDI World Dental Federation. The oral health status, the risk factors, and Oral Health Impact Profile-14 were also examined. Previous restoration characteristics (extent, technique, material) were analyzed according to the need for re-intervention (yes/no), the age group, and the risk profile. Qualitative variables were compared between "re-intervention" and "no re-intervention" group using Fisher exact test. RESULTS: The need for re-intervention was estimated at 74% (95% CI: 70; 79); it increased with age (49 to 90%), unfavorable risk profile (82 vs. 62%), and extent of the filling (32, 39, 44, and 44% on 1, 2, 3 surfaces, and crowns, respectively). More posterior than anterior teeth were restored (median per patient: 6 vs. 1) or needed re-intervention (median per patient: 1 vs. 0). CONCLUSIONS: The needs for re-intervention in adults are still high within a context of ever-changing materials and techniques, simplified and rationalized decision-makings, and demands for patient involvement. CLINICAL RELEVANCE: Meeting these needs requires the following: (i) consensus definitions and assessment methods for "failure" and (ii) reliable feedbacks on materials, procedures, and satisfaction. Building large and detailed databases fed by networks of motivated practitioners will help analyzing complex success/failure data by artificial intelligence and guiding treatment and research.

Effect of periodontal treatment on the glomerular filtration rate, reduction of inflammatory markers and mortality in patients with chronic kidney disease: A systematic review.

AIM: To assess the effect of periodontal treatment (PT) on glomerular filtration rate (GFR), systemic inflammation, or mortality in patients with chronic kidney disease (CKD). METHODS: A literature search was performed on PubMed and Web of Science databases on articles published until December 2019. The PRISMA guidelines were used throughout the manuscript. RESULTS: Of the total studies found, only 18 met the inclusion criteria; four retrospective and 14 prospective studies (including 3 randomized controlled trials-RCT). After PT, 3 studies investigated GFR, 2 found significant improvement; 11 (including 2 RCTs) investigated C-reactive protein levels, 9 found a significant improvement (including the 2 RCTs); 5 (including 3 RCTs) investigated Interleukine-6 level, 4 found a significant improvement (including 2 RCTs) and 2 studies evaluated mortality, one (retrospective study) found a significant difference. CONCLUSIONS: Within the limitations of the present study, PT seems to improve CKD status, especially by reducing the systemic inflammation. Further RCTs are needed to confirm the results and specifically assess the influence of different types of PT in CKD patients. Taking into consideration the ability of PT to prevent further tooth loss and denutrition, early management of periodontitis is extremely important in patients with impaired renal function.

Adherence to cysteamine in nephropathic cystinosis: A unique electronic monitoring experience for a better understanding. A prospective cohort study: CrYSTobs.

INTRODUCTION: In nephropathic cystinosis (NC), adherence to cysteamine remains challenging; poor adherence is worsening the disease progression with a decline of kidney function and increase of extrarenal morbidities. Our objective was to describe adherence to cysteamine in NC patients, using electronic monitoring systems. METHODS: Patients with confirmed NC, aged > 4 years and receiving oral cysteamine (short acting or delayed release formulation as standard of care) from 3 French reference centers, were included. Adherence to treatment was primarily assessed as the percentage of days with a good adherence score, adherence score rating from 0 (poor) to 2 (good). A descriptive analysis was performed after 1-year follow-up. RESULTS: Seventeen patients (10 girls, median age: 13.9 (5.4-33.0) years) were included. Median age at diagnosis was 17.0 (3.0-76.9) months and age at start of cysteamine was 21.0 (15.5-116.3) months. Median daily dose of cysteamine was 1.05 (0.55-1.63) g/m2/day. Over the year, the median percentage of days with a good adherence score was 80 (1-99)% decreasing to 68 (1-99)% in patients > 11 years old. The median of average number of hours covered by treatment in a day was 22.5 (6.1-23.9) versus 14.9 (9.2-20.5) hours for delayed release versus short acting cysteamine. CONCLUSION: Our data are the first describing a rather good adherence to cysteamine, decreasing in adolescents and adults. We described a potential interest of the delayed release formulation. Our data highlight the need for a multidisciplinary approach including therapeutic education and individualized approaches in NC patients transitioning to adulthood. Graphical abstract.

A mathematical model to quantify the effects of platelet count, shear rate, and injury size on the initiation of blood coagulation under venous flow conditions.

Platelets upregulate the generation of thrombin and reinforce the fibrin clot which increases the incidence risk of venous thromboembolism (VTE). However, the role of platelets in the pathogenesis of venous cardiovascular diseases remains hard to quantify. An experimentally validated model of thrombin generation dynamics is formulated. The model predicts that a high platelet count increases the peak value of generated thrombin as well as the endogenous thrombin potential (ETP) as reported in experimental data. To investigate the effects of platelets density, shear rate, and wound size on the initiation of blood coagulation, we calibrate a previously developed model of venous thrombus formation and implement it in 3D using a novel cell-centered finite-volume solver. We conduct numerical simulations to reproduce in vitro experiments of blood coagulation in microfluidic capillaries. Then, we derive a reduced one-equation model of thrombin distribution from the previous model under simplifying hypotheses and we use it to determine the conditions of clotting initiation on the platelet count, the shear rate, and the plasma composition. The initiation of clotting also exhibits a threshold response to the size of the wounded region in good agreement with the reported experimental findings.

Effect of Non-Steroidal Anti-Inflammatory Drugs on Sport Performance Indices in Healthy People: a Meta-Analysis of Randomized Controlled Trials.

BACKGROUND: Non-steroidal anti-inflammatory drugs (NSAIDs) are medications that are frequently used by athletes. There may also be some abuse of these substances, although it is unclear whether NSAIDs in fact enhance performance. We performed a systematic review and meta-analysis to evaluate the effect of NSAIDs on sport performance indices. METHODS: We selected randomized trials from the PubMed and Cochrane Library databases investigating the effects of NSAIDs on sport performance. Volunteers could be healthy adult men and women. Any NSAID, administered by any route, taken prior to any type of exercise, and for any duration could be used. The control intervention could be a placebo, an active substance, or no intervention. We included double-blind, single-blind, and open-label studies. The primary outcome was the maximum performance in exercises as defined in each study. The secondary outcomes were the time until self-reported exhaustion and the self-reported pain. RESULTS: Among 1631 records, we retained thirteen parallel-group and ten crossover studies, totaling 366 and 148 subjects, respectively. They were disparate regarding treatments, dose and duration, and the type of exercise. There was neither significant difference in the maximum performance between NSAIDs and control groups nor in the time until exhaustion nor in self-perceived pain. CONCLUSIONS: The existence of an ergogenic effect of NSAIDs on sport performance indices was unable to be concluded, since the level of evidence of the studies is low, the doses tested, and the exercises performed are very heterogeneous and far from those observed in real-life practices. More studies are required.

Collaboration between academics, small pharmaceutical company and patient organizations in the development of a new formulation of cysteamine in nephropathic cystinosis: A successful story.

Rare diseases usually concern small and disseminated population. Implementing clinical research with the right design, outcomes measures and the recruitment of patients are challenges. Collaborations, training and multidisciplinary approach are often required. In this article, we provide an overview of a successful collaboration in nephropathic cystinosis (NC), focusing on what was the key of success, the interactions between academics, the pharmaceutical company and patients organizations. NC is considered as a very rare disease. In 2010, a new formulation of cysteamine, the only available treatment to improve renal outcome of the disease, was proposed by a small American company. Studies were implemented in France under the coordination of an expert of the disease and the clinical investigation center of Lyon. The collaboration resulted in a good recruitment and retention of the patients in the study and most of all in the availability of the new formulation in France. Patients could have facilitated the research by being involved in the early stages of the studies. Involving patients and public early in the process is particularly important in rare diseases as the patient is a great source of knowledge and has his own expectations. Priorities of research, design, conduct and reporting of clinical trials can be defined in collaboration with adults but also with young patients or public, the first concerned in rare diseases. This concept is still to be developed and improved especially with paediatric patients.

A methodological framework for drug development in rare diseases. The CRESim program: Epilogue and perspectives.

Based on the 'European Child-Rare-Euro-Simulation' (CRESim) project, this article proposes a generalizable strategy utilizing datasets analysis in combination with modeling and simulation, in order to optimize the clinical drug development applied in the field of rare diseases. The global process includes: (i) the simulation of a realistic virtual population of patients (modeled from a real dataset of patients), (ii) the modeling of disease pathophysiological components and of pharmacokinetic-pharmacodynamic relations of the drug(s) of interest, (iii) the modeling of several randomized controlled clinical trials (RCTs) designs and (iv) the analysis of the results (multi-dimensional approach for RCTs durations and precision of the estimation of the treatment effect). However, whereas modeling and numerical simulation may provide supplementary tools for drug development, they cannot be considered as a substitute for RCTs performed in 'real' patients.

Blockade of the renin-angiotensin-aldosterone system in patients with arrhythmogenic right ventricular dysplasia: A double-blind, multicenter, prospective, randomized, genotype-driven study (BRAVE study).

Arrhythmogenic right ventricular dysplasia (ARVD) is a rare cardiomyopathy characterized by the progressive replacement of cardiomyocytes by fatty and fibrous tissue in the right ventricle (RV). These infiltrations lead to cardiac electrical instability and ventricular arrhythmia. Current treatment for ARVD is empirical and essentially based on treatment of arrhythmia. Thus, there is no validated treatment that will prevent the deterioration of RV function in patients with ARVD. The aim of the BRAVE study is to evaluate the effect of ramipril, an angiotensin-converting enzyme inhibitor, on ventricular myocardial remodeling and arrhythmia burden in patients with ARVD. Despite the fact that myocardial fibrosis is one of the structural hallmarks of ARVD, no study has tested an antifibrotic drug in ARVD patients. The trial is a double-blind, parallel, multicenter, prospective, randomized, phase 4 drug study. Patients will be randomized into 2 groups, ramipril or placebo. The 120 patients (60 per group) will be enrolled by 26 centers in France. Patients will be followed up every 6 months for 3 years. The 2 co-primary endpoints are defined as the difference of telediastolic RV volume measured by magnetic resonance imaging between baseline and 3 years of follow-up, and the change in arrhythmia burden during the 3 years of follow-up. A decrease in RV and/or left ventricular deterioration and in arrhythmia burden are expected in ARVD patients treated with ramipril. This reduction will improve quality of life of patients and will reduce the number of hospitalizations and the risk of terminal heart failure.

A double-blind placebo-controlled randomised trial of omega-3 supplementation in children with moderate ADHD symptoms.

OBJECTIVE: Clinical trials and inconclusive meta-analyses have investigated the effects of omega-3 supplements in children with Attention-Deficit Hyperactivity Disorder (ADHD). We performed a randomised placebo-controlled trial to evaluate the efficacy of omega-3 fatty acids. METHODS: Children aged 6-15 years with established diagnosis of ADHD were randomised 1:1 to receive either supplements containing docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA) or a placebo for 3 months. Psychotropic or omega-3-containing treatments were not authorised during the study. The primary outcome was the change in the Attention-Deficit Hyperactivity Disorder Rating Scale version 4 (ADHD-RS-IV). Other outcomes included safety, lexical level (Alouette test), attention (Test of Attentional Performance for Children-KiTAP), anxiety (48-item Conners Parent Rating Scale-Revised-CPRS-R), and depression (Children's Depression Inventory-CDI). RESULTS: Between 2009 and 2011, 162 children were included in five French child psychiatry centres. The mean age was 9.90 (SD 2.62) years and 78.4% were boys. The inclusion ADHD-RS-IV at was 37.31 (SD 8.40). The total ADHD-RS-IV score reduction was greater in the placebo group than in the DHA-EPA group: -19 (-26, -12)  % and -9.7 (-16.6, -2.9) %, respectively, p = 0.039. The other components of the Conners score had a similar variation but the differences between groups were not significant. Two patients in the DHA-EPA group and none in the placebo group experienced a severe adverse event (hospitalisation for worsening ADHD symptoms). CONCLUSION: This study did not show any beneficial effect of omega-3 supplement in children with mild ADHD symptoms.

One-step partial or complete caries removal and bonding with antibacterial or traditional self-etch adhesives: study protocol for a randomized controlled trial.

BACKGROUND: Current concepts in conservative dentistry advocate minimally invasive dentistry and pulp vitality preservation. Moreover, complete removal of carious dentin in deep carious lesions often leads to pulp exposure and root canal treatment, despite the absence of irreversible pulp inflammation. For years, partial caries removal has been performed on primary teeth, but little evidence supports its effectiveness for permanent teeth. Furthermore, the recent development of new antibacterial adhesive systems could be interesting in the treatment of such lesions. The objectives of this study are to compare the effectiveness of partial versus complete carious dentin removal in deep lesions (primary objective) and the use of an antibacterial versus a traditional two-step self-etch adhesive system (main secondary objective). METHODS/DESIGN: The DEep CAries Treatment (DECAT) study protocol is a multicenter, randomized, controlled superiority trial comparing partial versus complete caries removal followed by adhesive restoration. The minimum sample size required is 464 patients. Two successive randomizations will be performed (allocation ratio 1:1): the first for the type of excavation (partial versus complete) and the second (if no root canal treatment is required) for the type of adhesive (antibacterial versus traditional). For the two objectives, the outcome is the success of the treatment after 1 year, measured according to a composite outcome of five FDI criteria: material fracture and retention, marginal adaptation, radiographic examination (including apical pathologies), postoperative sensitivity and tooth vitality, and carious lesion recurrence. DISCUSSION: The study will investigate the interest of a conservative approach for the management of deep carious lesions in terms of dentin excavation and bioactive adhesive systems. The results may help practitioners achieve the most efficient restorative procedure to maintain pulp vitality and increase the restoration longevity. TRIAL REGISTRATION: ClinicalTrials.gov Identifier NCT02286388 . Registered in November 2014.

An in silico approach helped to identify the best experimental design, population, and outcome for future randomized clinical trials.

OBJECTIVES: The main objective of our work was to compare different randomized clinical trial (RCT) experimental designs in terms of power, accuracy of the estimation of treatment effect, and number of patients receiving active treatment using in silico simulations. STUDY DESIGN AND SETTING: A virtual population of patients was simulated and randomized in potential clinical trials. Treatment effect was modeled using a dose-effect relation for quantitative or qualitative outcomes. Different experimental designs were considered, and performances between designs were compared. One thousand clinical trials were simulated for each design based on an example of modeled disease. RESULTS: According to simulation results, the number of patients needed to reach 80% power was 50 for crossover, 60 for parallel or randomized withdrawal, 65 for drop the loser (DL), and 70 for early escape or play the winner (PW). For a given sample size, each design had its own advantage: low duration (parallel, early escape), high statistical power and precision (crossover), and higher number of patients receiving the active treatment (PW and DL). CONCLUSION: Our approach can help to identify the best experimental design, population, and outcome for future RCTs. This may be particularly useful for drug development in rare diseases, theragnostic approaches, or personalized medicine.

High Risk versus Proportional Benefit: Modelling Equitable Strategies in Cardiovascular Prevention.

OBJECTIVE: To examine the performances of an alternative strategy to decide initiating BP-lowering drugs called Proportional Benefit (PB). It selects candidates addressing the inequity induced by the high-risk approach since it distributes the gains proportionally to the burden of disease by genders and ages. STUDY DESIGN AND SETTING: Mild hypertensives from a Realistic Virtual Population by genders and 10-year age classes (range 35-64 years) received simulated treatment over 10 years according to the PB strategy or the 2007 ESH/ESC guidelines (ESH/ESC). Primary outcomes were the relative life-year gain (life-years gained-to-years of potential life lost ratio) and the number needed to treat to gain a life-year. A sensitivity analysis was performed to assess the impact of changes introduced by the ESH/ESC guidelines appeared in 2013 on these outcomes. RESULTS: The 2007 ESH/ESC relative life-year gains by ages were 2%; 10%; 14% in men, and 0%; 2%; 11% in women, this gradient being abolished by the PB (relative gain in all categories = 10%), while preserving the same overall gain in life-years. The redistribution of benefits improved the profile of residual events in younger individuals compared to the 2007 ESH/ESC guidelines. The PB strategy was more efficient (NNT = 131) than the 2013 ESH/ESC guidelines, whatever the level of evidence of the scenario adopted (NNT = 139 and NNT = 179 with the evidence-based scenario and the opinion-based scenario, respectively), although the 2007 ESH/ESC guidelines remained the most efficient strategy (NNT = 114). CONCLUSION: The Proportional Benefit strategy provides the first response ever proposed against the inequity of resource use when treating highest risk people. It occupies an intermediate position with regards to the efficiency expected from the application of historical and current ESH/ESC hypertension guidelines. Our approach allows adapting recommendations to the risk and resources of a particular country.

Methodological choices for the clinical development of medical devices.

Clinical evidence available for the assessment of medical devices (MDs) is frequently insufficient. New MDs should be subjected to high quality clinical studies to demonstrate their benefit to patients. The randomized controlled trial (RCT) is the study design reaching the highest level of evidence in order to demonstrate the efficacy of a new MD. However, the clinical context of some MDs makes it difficult to carry out a conventional RCT. The objectives of this review are to present problems related to conducting conventional RCTs and to identify other experimental designs, their limitations, and their applications. A systematic literature search was conducted for the period January 2000 to July 2012 by searching medical bibliographic databases. Problems related to conducting conventional RCTs of MDs were identified: timing the assessment, eligible population and recruitment, acceptability, blinding, choice of comparator group, and learning curve. Other types of experimental designs have been described. Zelen's design trials and randomized consent design trials facilitate the recruitment of patients, but can cause ethical problems to arise. Expertise-based RCTs involve randomization to a team that specializes in a given intervention. Sometimes, the feasibility of an expertise-based randomized trial may be greater than that of a conventional trial. Cross-over trials reduce the number of patients, but are not applicable when a learning curve is required. Sequential trials have the advantage of allowing a trial to be stopped early depending on the results of first inclusions, but they require an independent committee. Bayesian methods combine existing information with information from the ongoing trial. These methods are particularly useful in situations where the number of subjects is small. The disadvantage is the risk of including erroneous prior information. Other types of experimental designs exist when conventional trials cannot always be applied to the clinical development of MDs.

Determinants of torsades de pointes in older patients with drug-associated long QT syndrome: a case-control study.

OBJECTIVE: Many elderly patients are routinely exposed to drugs that may prolong the cardiac QT interval and cause Torsades de pointes (TdP). However, predictors of TdP in patients with drug-associated long QT syndrome (LQTS) are not fully understood, especially in the geriatric population. The objective of this study was to identify risk factors of TdP in elderly patients with drug-associated LQTS. METHODS: In this retrospective, case-control study, documented reports of drug-associated LQTS plus TdP (n = 125) and LQTS without TdP (n = 81) in patients ≥65 years of age were retrieved from the French Pharmacovigilance Database over a 10-year period. Available clinical, biological, and drug therapy data were compared in the two groups and logistic regression was performed to identify significant predictors of TdP. RESULTS: The uncorrected QT interval was significantly longer in patients with TdP than in patients without TdP (577 ± 79 vs. 519 ± 68 ms; p = 0.0001). The number of drugs with a known risk of TdP administered to each patient was not a predictor of arrhythmia, nor was female gender. Logistic regression analysis identified the uncorrected QT interval as the only significant predictor of TdP. The receiver operating characteristic curve analysis was characterized by an area under the curve of 0.77 (95 % confidence interval 0.64-0.88) and a QT cutoff of 550 ms. CONCLUSION: The uncorrected QT interval was significantly associated with the probability of TdP in elderly patients with acquired, drug-associated LQTS.

A methodological framework for drug development in rare diseases.

INTRODUCTION: Developing orphan drugs is challenging because of their severity and the requisite for effective drugs. The small number of patients does not allow conducting adequately powered randomized controlled trials (RCTs). There is a need to develop high quality, ethically investigated, and appropriately authorized medicines, without subjecting patients to unnecessary trials. AIMS AND OBJECTIVES: The main aim is to develop generalizable framework for choosing the best-performing drug/endpoint/design combinations in orphan drug development using an in silico modeling and trial simulation approach. The two main objectives were (i) to provide a global strategy for each disease to identify the most relevant drugs to be evaluated in specific patients during phase III RCTs, (ii) and select the best design for each drug to be used in future RCTs. METHODOLOGICAL APPROACH: In silico phase III RCT simulation will be used to find the optimal trial design and was carried out in two steps: (i) statistical analysis of available clinical databases and (ii) integrative modeling that combines mathematical models for diseases with pharmacokinetic-pharmacodynamics models for the selected drug candidates. CONCLUSION: There is a need to speed up the process of orphan drug development, develop new methods for translational research and personalized medicine, and contribute to European Medicines Agency guidelines. The approach presented here offers many perspectives in clinical trial conception.

Experimental designs for small randomised clinical trials: an algorithm for choice.

BACKGROUND: Small clinical trials are necessary when there are difficulties in recruiting enough patients for conventional frequentist statistical analyses to provide an appropriate answer. These trials are often necessary for the study of rare diseases as well as specific study populations e.g. children. It has been estimated that there are between 6,000 and 8,000 rare diseases that cover a broad range of diseases and patients. In the European Union these diseases affect up to 30 million people, with about 50% of those affected being children. Therapies for treating these rare diseases need their efficacy and safety evaluated but due to the small number of potential trial participants, a standard randomised controlled trial is often not feasible. There are a number of alternative trial designs to the usual parallel group design, each of which offers specific advantages, but they also have specific limitations. Thus the choice of the most appropriate design is not simple. METHODS: PubMed was searched to identify publications about the characteristics of different trial designs that can be used in randomised, comparative small clinical trials. In addition, the contents tables from 11 journals were hand-searched. An algorithm was developed using decision nodes based on the characteristics of the identified trial designs. RESULTS: We identified 75 publications that reported the characteristics of 12 randomised, comparative trial designs that can be used in for the evaluation of therapies in orphan diseases. The main characteristics and the advantages and limitations of these designs were summarised and used to develop an algorithm that may be used to help select an appropriate design for a given clinical situation. We used examples from publications of given disease-treatment-outcome situations, in which the investigators had used a particular trial design, to illustrate the use of the algorithm for the identification of possible alternative designs. CONCLUSIONS: The algorithm that we propose could be a useful tool for the choice of an appropriate trial design in the development of orphan drugs for a given disease-treatment-outcome situation.

Quantitative prediction of the impact of drug interactions and genetic polymorphisms on cytochrome P450 2C9 substrate exposure.

BACKGROUND AND OBJECTIVE: Cytochrome P450 (CYP) 2C9 is the most common CYP2C enzyme and makes up approximately onethird of total CYP protein content in the liver. It metabolises more than 100 drugs. The exposure of drugs mainly eliminated by CYP2C9 may be dramatically modified by drug-drug interactions (DDIs) and genetic variations. The objective of this study was to develop a modelling approach to predict the impact of genetic polymorphisms and DDIs on drug exposure in drugs metabolised by CYP2C9. We then developed dosing recommendations based on genotypes and compared them to current Epar/Vidal dosing guidelines. METHODS: We created two models. The genetic model was designed to predict the impact of CYP2C9 polymorphisms on drug exposure. It links the area under the concentration-time curve (AUC) ratio (mutant to wild-type patients) to two parameters: the fractional contribution of CYP2C9 to oral clearance in vivo (i.e. CR or contribution ratio), and the fractional activity of the allele combination with respect to the homozygous wild type (i.e. FA or fraction of activity). Data were available for 77 couples (substrate, genotype). We used a three-step approach: (1) initial estimates of CRs and FAs were calculated using a first bibliographic dataset; (2) external validation of these estimates was then performed through the comparison between the AUC ratios predicted by the model and the observed values, using a second published dataset; and (3) refined estimates of CRs and FAs were obtained using Bayesian orthogonal regression involving the whole dataset and initial estimates of CRs and FAs. Posterior distributions of AUC ratios, CRs and FAs were estimated using Monte-Carlo Markov chain simulation. The drug interaction model was designed to predict the impact of DDIs on drug exposure. It links the AUC ratio (ratio of drug given in combination to drug given alone) to several parameters: the CR, the inhibition ratio (IR) of an inhibitor, and the increase in clearance (IC) due to an inducer. Data were available for 80 DDIs. IRs and ICs were calculated using the interaction model and an external validation was performed. Doses adjustments were calculated in order to obtain equal values for drug exposure in extensive and poor metabolisers and then compared to Epar/Vidal dosing guidelines. RESULTS: CRs were assessed for 26 substrates, FAs for five genotype classes including CYP2C9*2 and *3 allelic variants, IRs for 27 inhibitors and ICs for two inducers. For the genetic model, the mean prediction error of AUC ratios was -0.01, while the mean prediction absolute error was 0.36. For the drug interaction model, the mean prediction error of AUC ratios was 0.01, while the mean prediction absolute error was 0.22. Of the 26 substrates and CYP2C9*2 and *3 variants investigated, 30 couples (substrate, genotype) lead to a dose adjustment, as opposed to only ten couples identified in the Epar/Vidal recommendations. CONCLUSION: These models were already used for CYP2D6. They are accurate at predicting the impact of drug interactions and genetic polymorphisms on CYP2C9 substrate exposure. This approach will contribute to the development of personalized medicine, i.e. individualized drug therapy with specific dosing recommendations based on CYP genotype or drug associations.