AUC Estimates Research Articles

Long Non-Coding RNA Signatures Associated with Ferroptosis Predict Prognosis in Colorectal Cancer.

BackgroundCurrently, colorectal cancer has become a common gastrointestinal malignancy that usually occurs in the colon and rectum, and ferroptosis plays a vital role in the pathology and progression of colorectal tumors.MethodsA total of 627 patients (51 normal and 644 tumor samples) from The Cancer Genome Atlas (TCGA)-COAD and TCGA-READ were included in the study. Lasso and Cox’s regression was employed to analyze the characteristic lncRNAs in colorectal cancer samples, and a distinctive prognostic model of ferroptosis-related lncRNAs was established. By analyzing the divergence between the high and low-risk groups of ferroptosis-related lncRNAs, 15 characteristic lncRNAs related to the prognosis of colorectal cancer were evaluated. Kaplan–Meier analysis, operation characteristic curve (ROC), nomogram, and gene set enrichment analyses (GSEA) further confirmed the validity of the characteristic prognostic model with ferroptosis-related lncRNAs.ResultsKaplan–Meier analysis confirmed a high-risk group of ferroptosis-related lncRNA interrelated with a poor prognosis in colorectal cancer. AUC estimates of 1 -, 3 -, and 5-year survival rates for ferroptosis-related lncRNA characteristic models were 0.745, 0.767 and 0.789. GSEA analysis showed that immune and malignancy-related pathways were active in the high-risk score group. In addition, differential analyses of immune function, including Checkpoint, cytolytic, HLA, and T cell co-inhibition, differed significantly betwixt low - and high-risk groups.CD160, TNFRSF18, CD27, PDCD1, CD200R1, ADORA2A, TNFRSF14, LAIR1, CD244, CD40, TNFRSF4, CD70, TNFSF14, TNFRSF25, CD276, HHLA2, VTCN1, LAG3, TNFSF18, and other immune checkpoints had different expressions betwixt the high- and low-risk group.ConclusionFifteen kinds of lncRNAs with different expressions (AP003555.1, AC099850.3, AL031985.3, LINC01857, STPG3-AS1, AL137782.1, AC124067.4, AC012313.5, AC083900.1, AC010973.2, ALMS1-IT1, AC013652.1, AC133540.1, AP006621.2, AC018653.3) were closely associated with poor prognosis of colorectal cancer. These indicators were significantly correlated with the overall survival (OS) rate and could be used as prognostic evaluation criteria.

Limited sampling strategies for estimation of tacrolimus exposure in kidney transplant recipients receiving extended-release tacrolimus preparation.

Tacrolimus is the key component of most contemporary immunosuppressive drug regimens for the prevention of transplant rejection. Area under the concentration time curve over 24 h (AUC0–24) predicts efficacy, but predose (trough) tacrolimus blood concentration (C0) is currently used to guide dosing. In clinical or research situations where an estimate of AUC is required, collection of a full 24 h pharmacokinetic (PK) profile is cumbersome. Limited sampling strategies (LSSs) have been developed for some tacrolimus preparations but not for the new, extended‐release, once‐daily formulation of tacrolimus, ENVARSUS XR. Twenty‐four kidney transplant recipients were enrolled in this study. Twenty‐four tacrolimus PK profiles were obtained over 24 h. Multiple linear regression was used to generate LSSs with the best subset selection for accurate estimation of tacrolimus AUC0–24. The predictive performance of each model was assessed in the evaluation group. The correlation between actual and predicted AUC0–24 was evaluated and mean percentage prediction error (MPE%), mean absolute percentage prediction error (MAE%), and root mean squared error (RMSE) were calculated for each prediction model to assess bias and precision. The selected LSSs were highly correlated to AUC0–24 compared with the correlation between C0 and AUC0‐24. Two and three sampling points limited sampling strategies: C0, C2, and C10 provide the most reliable and effective LSS for estimation of tacrolimus AUC0–24 in routine clinic use. These limited sampling models can be applied in therapeutic drug monitoring schemes to personalize tacrolimus dosing for kidney transplant recipients on treatment with extended‐release tacrolimus.

Model Based Identification of Linezolid Exposure-toxicity Thresholds in Hospitalized Patients.

Evidence supports linezolid therapeutic drug monitoring as the exposure–response relationship has been identified for toxicity among patients receiving linezolid, but the data to establish the upper limit are limited and the published toxicity thresholds range widely. The purpose of this study was to determine the linezolid exposure–toxicity thresholds to improve the safety of linezolid. This is a multicenter retrospective study of adult patients treated with linezolid from 2018 to 2019. The population pharmacokinetic model of linezolid was established based on 270 plasma concentrations in 152 patients, which showed creatinine clearance and white cell count are covariates affecting the clearance of linezolid, and serum albumin is the covariate affecting the volume of distribution. Classification and regression tree analysis was used to determine the linezolid exposure thresholds associated with an increased probability of toxicity. Among 141 patients included for toxicity analysis, the rate of occurring toxicity was significantly higher among patients with an AUC0-24, d1 ≥163 mg h/L, AUC0-24, d2 ≥207 mg h/L, AUC0-24, ss ≥210 mg h/L, and Cmin,d2 ≥6.9 mg/L, Cmin,ss ≥6.9 mg/L, while no threshold was discovered for Cmin, d1. Those exposure thresholds and duration of linezolid treatment were independently associated with linezolid-related toxicity in the logistic regression analyses. In addition, the predictive performance of the AUC0-24 and Cmin thresholds at day 2 and steady state were close. Considering that the AUC estimation is cumbersome, Cmin threshold at 48 h and steady state with a value of ≥6.9 mg/L is recommended to improve safety, especially for patients with renal insufficiency and patients with low serum albumin.

Prediction of imminent fracture risk in Canadian women and men aged 45years or older: external validation of the Fracture Risk Evaluation Model (FREM).

The FREM is a tool developed from Danish public health registers (hospital diagnoses) to identify individuals over age 45years at high imminent risk of major osteoporotic fractures (MOF) and hip fracture (HF). In this study, our aim was to examine the ability of FREM to identify individuals at high imminent fracture risk in women and men from Manitoba, Canada. We used the population-based Manitoba Bone Mineral Density (BMD) Program registry, and identified women and men aged 45years or older undergoing baseline BMD assessment with 2years of follow-up data. From linked population-based data sources, we constructed FREM scores using up to 10years of prior healthcare information. The study population comprised 74,828 subjects, and during the 2years of observation, 1612 incident MOF and 299 incident HF occurred. We found significant fracture risk stratification for all FREM scores, with AUC estimates of 0.63-0.66 for MOF for both sexes and 0.84 for women and 0.65-0.67 for men for HF. FREM performed better than age alone but not as well as FRAX® with BMD. The inclusion of physician claims data gave slightly better performance than hospitalization data alone. Overall calibration for 1-year MOF prediction was reasonable, but HF prediction was overestimated. In conclusion, the FREM algorithm shows significant fracture risk stratification when applied to an independent clinical population from Manitoba, Canada. Overall calibration for MOF prediction was good, but hip fracture risk was systematically overestimated indicating the need for recalibration.

Dose individualization of intravenous busulfan in pediatric patients undergoing bone marrow transplantation: impact and in vitro evaluation of infusion lag-time.

To apply therapeutic drug monitoring and dose-individualization of intravenous Busulfan to paediatric patients and evaluate the impact of syringe-pump induced Busulfan infusion lag-time after in vitro estimation. 76 children and adolescents were administered 2 h intravenous Busulfan infusion every 6 h (16 doses). Busulfan plasma levels, withdrawn by an optimized sampling scheme and measured by a validated HPLC-PDA method, were used to estimate basic PK parameters, AUC, Cmax, kel, t1/2, applying Non-Compartmental Analysis. In vivo infusion lag-time was simulated in vitro and used to evaluate its impact on AUC estimation. Mean (%CV) Busulfan AUC, Cmax, clearance and t1/2 for pediatric population were found 962.3 μm × min (33.1), 0.95 mg/L (41.4), 0.27 L/h/kg (33.3), 2.2 h (27.8), respectively. TDM applied to 76 children revealed 6 (7.9%) being above and 25 (32.9%) below therapeutic-range (AUC: 900-1350 μm × min). After dose correction, all patients were measured below toxic levels (AUC < 1500 μm × min), no patient below 900 μm × min. Incorporation of infusion lag-time revealed lower AUCs with 17.1% more patients and 23.1% more younger patients, with body weight <16 kg, being below the therapeutic-range. TDM, applied successfully to 76 children, confirmed the need for Busulfan dose-individualization in paediatric patients. Infusion lag-time was proved clinically significant for younger, low body-weight patients and those close to the lower therapeutic-range limit.

Between-group comparison of area under the curve in clinical trials with censored follow-up: Application to HIV therapeutic vaccines.

In clinical trials, longitudinal data are commonly analyzed and compared between groups using a single summary statistic such as area under the outcome versus time curve (AUC). However, incomplete data, arising from censoring due to a limit of detection or missing data, can bias these analyses. In this article, we present a statistical test based on splines-based mixed-model accounting for both the censoring and missingness mechanisms in the AUC estimation. Inferential properties of the proposed method were evaluated and compared to ad hoc approaches and to a non-parametric method through a simulation study based on two-armed trial where trajectories and the proportion of missing data were varied. Simulation results highlight that our approach has significant advantages over the other methods. A real working example from two HIV therapeutic vaccine trials is presented to illustrate the applicability of our approach.

Population pharmacokinetics and genetics of oral meltdose tacrolimus (Envarsus) in stable adult liver transplant recipients.

AimsMeltdose tacrolimus (Envarsus) is marketed as a formulation with a more consistent exposure. Due to the narrow therapeutic window, therapeutic drug monitoring is essential to maintain adequate exposure. The primary objective of this study was to develop a population pharmacokinetic (PK) model of Envarsus among liver transplant patients and select a limited sampling strategy (LSS) for AUC estimation. The secondary objective was to investigate potential covariates including CYP3A/IL genotype suitable for initial dose optimization when converting to Envarsus.MethodsAdult liver transplant patients were converted from prolonged release tacrolimus (Advagraf) to Envarsus and blood samples were obtained using whole blood and dried blood spot sampling. Subsequently the population PK parameters were estimated using nonlinear‐mixed effect modelling. Demographic factors, and recipient and donor CYP3A4, CYP3A5, IL‐6, ‐10 and ‐18 genotype were tested as potential covariates to explain interindividual variability.ResultsFifty‐five patients were included. A 2‐compartment model with delayed absorption was the most suitable to describe population PK parameters. The population PK parameters were as follows: clearance, 3.27 L/h; intercompartmental clearance, 9.6 L/h; volume of distribution of compartments 1 and 2, 95 and 500 L, respectively. No covariates were found to significantly decrease interindividual variability. The best 3‐point LSS was t = 0,4,8 with a median bias of 1.8% (−12.5–12.5).ConclusionsThe LSS can be used to adequately predict the AUC. No clinically relevant covariates known to influence the PK of Envarsus, including CYP3A status, were identified and therefore do not seem useful for initial dose optimization.

Evaluation of predictive model performance of an existing model in the presence of missing data.

In medical research, the Brier score (BS) and the area under the receiver operating characteristic (ROC) curves (AUC) are two common metrics used to evaluate prediction models of a binary outcome, such as using biomarkers to predict the risk of developing a disease in the future. The assessment of an existing prediction models using data with missing covariate values is challenging. In this article, we propose inverse probability weighted (IPW) and augmented inverse probability weighted (AIPW) estimates of AUC and BS to handle the missing data. An alternative approach uses multiple imputation (MI), which requires a model for the distribution of the missing variable. We evaluated the performance of IPW and AIPW in comparison with MI in simulation studies under missing completely at random, missing at random, and missing not at random scenarios. When there are missing observations in the data, MI and IPW can be used to obtain unbiased estimates of BS and AUC if the imputation model for the missing variable or the model for the missingness is correctly specified. MI is more efficient than IPW. Our simulation results suggest that AIPW can be more efficient than IPW, and also achieves double robustness from miss-specification of either the missingness model or the imputation model. The outcome variable should be included in the model for the missing variable under all scenarios, while it only needs to be included in missingness model if the missingness depends on the outcome. We illustrate these methods using an example from prostate cancer.

Artificial intelligence for the classification of fractures around the knee in adults according to the 2018 AO/OTA classification system.

Fractures around the knee joint are inherently complex in terms of treatment; complication rates are high, and they are difficult to diagnose on a plain radiograph. An automated way of classifying radiographic images could improve diagnostic accuracy and would enable production of uniformly classified records of fractures to be used in researching treatment strategies for different fracture types. Recently deep learning, a form of artificial intelligence (AI), has shown promising results for interpreting radiographs. In this study, we aim to evaluate how well an AI can classify knee fractures according to the detailed 2018 AO-OTA fracture classification system. We selected 6003 radiograph exams taken at Danderyd University Hospital between the years 2002-2016, and manually categorized them according to the AO/OTA classification system and by custom classifiers. We then trained a ResNet-based neural network on this data. We evaluated the performance against a test set of 600 exams. Two senior orthopedic surgeons had reviewed these exams independently where we settled exams with disagreement through a consensus session. We captured a total of 49 nested fracture classes. Weighted mean AUC was 0.87 for proximal tibia fractures, 0.89 for patella fractures and 0.89 for distal femur fractures. Almost ¾ of AUC estimates were above 0.8, out of which more than half reached an AUC of 0.9 or above indicating excellent performance. Our study shows that neural networks can be used not only for fracture identification but also for more detailed classification of fractures around the knee joint.

Estimating the AUC with a graphical lasso method for high-dimensional biomarkers with LOD

This manuscript estimates the area under the receiver operating characteristic curve (AUC) of combined biomarkers in a high-dimensional setting. We propose a penalization approach to the inference of precision matrices in the presence of the limit of detection. A new version of expectation-maximization algorithm is then proposed for the penalized likelihood, with the use of numerical integration and the graphical lasso method. The estimated precision matrix is then applied to the inference of AUCs. The proposed method outperforms the existing methods in numerical studies. We apply the proposed method to a data set of brain tumor study. The results show a higher accuracy on the estimation of AUC compared with the existing methods.

Area under the Curve-Based Dosing of Vancomycin in Critically Ill Patients Using 6-Hour Urine Creatinine Clearance Measurement.

Background The area under the curve- (AUC-) guided vancomycin dosing is the best strategy for individualized therapy in critical illnesses. Since AUC can be calculated directly using drug clearance (CLvan), any parameter estimating CLvan will be able to achieve the goal of 24-hour AUC (AUC24 h). The present study was aimed to determine CLvan based on 6-hour urine creatinine clearance measurement in critically ill patients with normal renal function. Method 23 adult critically ill patients with an estimated glomerular filtration rate (eGFR) ≥60 mL/min who received vancomycin infusion were enrolled in this pilot study. Vancomycin pharmacokinetic parameters were determined for each patient using serum concentration data and a one-compartment model provided by MONOLIX software using stochastic approximation expectation-maximization (SAEM) algorithm. Correlation of CLvan with the measured creatinine clearance in 6-hour urine collection (CL6 h) and estimated creatinine clearance by the Cockcroft–Gault formula (CLCG) was investigated. Results Data analysis revealed that CL6 h had a stronger correlation with CLvan rather than CLCG (r = 0.823 vs. 0.594; p < 0.001 vs. 0.003). The relationship between CLvan and CL6 h was utilized to develop the following equation for estimating CLvan: CLvan (mL/min) = ─137.4 + CL6 h (mL/min) + 2.5 IBW (kg) (R2 = 0.826, p < 0.001). Regarding the described model, the following equation can be used to calculate the empirical dose of vancomycin for achieving the therapeutic goals in critically ill patients without renal impairment: total daily dose of vancomycin (mg) = (─137.4CL6-h (mL/min) + 2.5 IBW (kg)) × 0.06 AUC24 h (mg.hr/L). Conclusion For AUC estimation, CLvan can be obtained by collecting urine in a 6-hour period with good approximation in critically ill patients with normal renal function.

Optimal two-phase sampling for estimating the area under the receiver operating characteristic curve.

Statistical methods are well developed for estimating the area under the receiver operating characteristic curve (AUC) based on a random sample where the gold standard is available for every subject in the sample, or a two-phase sample where the gold standard is ascertained only at the second phase for a subset of subjects sampled using fixed sampling probabilities. However, the methods based on a two-phase sample do not attempt to optimize the sampling probabilities to minimize the variance of AUC estimator. In this paper, we consider the optimal two-phase sampling design for evaluating the performance of an ordinal test in classifying disease status. We derived the analytic variance formula for the AUC estimator and used it to obtain the optimal sampling probabilities. The efficiency of the two-phase sampling under the optimal sampling probabilities (OA) is evaluated by a simulation study, which indicates that two-phase sampling under OA achieves a substantial amount of variance reduction with an over-sample of subjects with low and high ordinal levels, compared with two-phase sampling under proportional allocation (PA). Furthermore, in comparison with an one-phase random sampling, two-phase sampling under OA or PA have a clear advantage in reducing the variance of AUC estimator when the variance of diagnostic test results in the disease population is small relative to its counterpart in nondisease population. Finally, we applied the optimal two-phase sampling design to a real-world example to evaluate the performance of a questionnaire score in screening for childhood asthma.

Comparison of Individualized Versus MAP-Bayesian Predicted AUC of Busulfan in FluBu4 Treated Patients Undergoing Allogeneic Transplant

Although myeloablative fludarabine/busulfan (FluBu4) has been widely adopted in clinical practice, considerable interpatient variability exists in systemic busulfan exposure (AUC) when using body weight or body surface area based-dosing, leading to decreased efficacy (i.e. relapse) or increased toxicity (i.e., mucositis, veno-occlusive disease). This well-defined dose-exposure-outcome relationship has led to the widespread clinical implementation of therapeutic drug monitoring (TDM). However, individualized TDM can be time and labor intensive as well as potentially biased due to the lack of incorporation of any previously established PK data (Bayesian prior). In contrast, Bayesian maximum a posteriori (MAP) PK models consider the Bayesian prior and individualized TDM to generate a revised probability distribution (Bayes conditional posterior) to more accurately and rapidly estimate the AUC with reduced bias. There are a paucity of data comparing busulfan AUCs using individualized PK versus MAP-Bayesian-based models in adults although these more sophisticated approaches may assist in optimizing dosing of busulfan in this vulnerable population. This was a retrospective, single-center study of patients who received FluBu4 with busulfan TDM between January 1999 and September 2019. 109 patients diagnosed with a hematologic malignancy who received either sequential (n=46) or concurrent (n=63) FluBu4 were analyzed. The median age was 48 (range: 18-66), and were Hispanic (n=38), White (n=46) or African-American (n=14). TDM was performed either after a test dose of 0.8 mg/kg (n=52) or after the first dose (3.2 mg/kg) of busulfan administered during the preparative regimen (n=71), with dosing was based on actual or adjusted body weight. For PK analysis, plasma busulfan concentrations were analyzed via gas chromatography with mass selective detection. Individualized PK data were generated using WinNonlin while the MAP-Bayesian approach utilized the Bayesian prior developed from McCune et al (Clin Cancer Res, 2014). An AUC of 4800 µM˖min/24 hours was targeted based on previous literature. Based on individualized PK data, total recommended busulfan doses ranged from 9.3-21.3 mg/kg (-27.1% to +66.7% compared to FDA labeled dose of 12.8 mg/kg). When first-dose busulfan PK was compared between busulfan given sequentially versus concurrently with fludarabine there was a trend towards a higher AUC with concomitant administration (4651 vs. 4988 µM˖min; p=0.13). A strong correlation between the AUC generated from both the individualized PK an MAP-Bayesian models was observed with both the test dose (R2=0.91) and first dose (R2=0.86) of busulfan. Using the MAP Bayesian model, AUC predictions were on average higher (mean AUC 5069 versus 4886 µM˖min, p&amp;lt;0.0001) compared to the patient-specific individualized PK estimates. Figure 1 shows the Bland Altman plots for comparison of the individualized AUC vs. MAP-Bayesian estimates for test dose and first dose. Our individualized busulfan PK approach generated relatively similar AUC values compared to MAP-Bayesian estimates, although the higher AUC generated via MAP-Bayesian predictions may allow for lower doses of busulfan to be administered thereby potentially reducing toxicity while maintaining efficacy. Further, use of the MAP-Bayesian method may allow for more rapid dose optimization and a decreased number of serum concentrations. Further prospective studies including more patients are warranted to confirm these findings. Figure 1 Disclosures Calip: Flatiron Health: Current Employment. Patel:Janssen: Consultancy; Amgen: Consultancy; Celgene: Consultancy.

A Systematic Review and Meta-analysis of Isoniazid Pharmacokinetics in Healthy Volunteers and Patients with Tuberculosis

PurposeThis systematic review and meta-analysis assesses the pharmacokinetic (PK) summary estimates of isoniazid (INH) between healthy volunteers and patients with tuberculosis (TB), evaluates whether the current INH dose regimen is appropriate in patients with TB, and evaluates the impact of N-acetyl-transferase-2 (NAT2) status on the PK properties of INH. MethodsA systematic approach was conducted to find studies with relevant INH PK data published in the English language up to February 2018. The PK properties of INH were extracted with their respective INH dosages and were dose normalized to allow a fair comparison between healthy volunteers and patients with TB. Meta-analysis was then performed for the Cmax and AUC estimates for all INH dosages. FindingsNinety studies were included in this systematic review. TB status significantly affected the INH Cmax and AUC estimates. In healthy volunteers, the dose-normalized INH Cmax and AUC were statistically higher than those of patients with TB. No significant differences were found in dose-normalized Cmax and AUC between adults with TB and adults with TB/HIV; however, the AUC in pediatric patients was significantly different between patients with TB and patients with TB/HIV. In addition, no significance was observed comparing the dose-normalized Cmax and AUC of pediatric patients with TB and TB/HIV with their respective adult counterparts. Dose-normalized INH Cmax and AUC in patients with fast and intermediate NAT2 were significantly lower than in patients with slow NAT2. ImplicationsThe current recommended dosages of INH were found to produce less drug exposure in patients with TB when compared with healthy volunteers. NAT2 polymorphism greatly impacts the PK properties of INH; hence, testing for acetylator status is highly recommended, and therapeutic drug monitoring would help reduce INH toxicity.

An efficient variance estimator of AUC and its applications to binary classification.

The area under the ROC (receiver operating characteristic) curve, AUC, is one of the most commonly used measures to evaluate the performance of a binary classifier. Due to sampling variation, the model with the largest observed AUC score is not necessarily optimal, so it is crucial to assess the variation of AUC estimate. We extend the proposal by Wang and Lindsay and devise an unbiased variance estimator of AUC estimate that is of a two-sample U-statistic form. The proposal can be easily generalized to estimate the variance of a K-sample U-statistic (K ≥ 2). To make our developed variance estimator more applicable, we employ a partition-resampling scheme that is computationally efficient. Simulation studies suggest that the developed AUC variance estimator yields much better or comparable performance to jackknife and bootstrap variance estimators, and computational times that are about 10 to 30 times faster than the times of its counterparts. In practice, the proposal can be used in the one-standard-error rule for model selection, or to construct an asymptotic confidence interval of AUC in binary classification. In addition to conducting simulation studies, we illustrate its practical applications using two real datasets in medical sciences.

Predicting Progression to Advanced Age-Related Macular Degeneration from Clinical, Genetic, and Lifestyle Factors Using Machine Learning

Current prediction models for advanced age-related macular degeneration (AMD) are based on a restrictive set of risk factors. The objective of this study was to develop a comprehensive prediction model applying a machine learning algorithm allowing selection of the most predictive risk factors automatically. Two population-based cohort studies. The Rotterdam Study I (RS-I; training set) included 3838 participants 55 years of age or older, with a median follow-up period of 10.8 years, and 108 incident cases of advanced AMD. The Antioxydants, Lipids Essentiels, Nutrition et Maladies Oculaires (ALIENOR) study (test set) included 362 participants 73 years of age or older, with a median follow-up period of 6.5 years, and 33 incident cases of advanced AMD. The prediction model used the bootstrap least absolute shrinkage and selection operator (LASSO) method for survival analysis to select the best predictors of incident advanced AMD in the training set. Predictive performance of the model was assessed using the area under the receiver operating characteristic curve (AUC). Incident advanced AMD (atrophic, neovascular, or both), based on standardized interpretation of retinal photographs. The prediction model retained (1) age, (2) a combination of phenotypic predictors (based on the presence of intermediate drusen, hyperpigmentation in one or both eyes, and Age-Related Eye Disease Study simplified score), (3) a summary genetic risk score based on 49 single nucleotide polymorphisms, (4) smoking, (5) diet quality, (6) education, and (7) pulse pressure. The cross-validated AUC estimation in RS-I was 0.92 (95% confidence interval [CI], 0.88-0.97) at 5 years, 0.92 (95% CI, 0.90-0.95) at 10 years, and 0.91 (95% CI, 0.88-0.94) at 15 years. In ALIENOR, the AUC reached 0.92 at 5 years (95% CI, 0.87-0.98). In terms of calibration, the model tended to underestimate the cumulative incidence of advanced AMD for the high-risk groups, especially in ALIENOR. This prediction model reached high discrimination abilities, paving the way toward making precision medicine for AMD patients a reality in the near future.

Kinetic modelling of myocardial necrosis biomarkers offers an easier, reliable and more acceptable assessment of infarct size

Infarct size is a major prognostic factor in ST-segment elevation myocardial infarction (STEMI). It is often assessed using repeated blood sampling and the estimation of biomarker area under the concentration versus time curve (AUC) in translational research. We aimed at developing limited sampling strategies (LSS) to accurately estimate biomarker AUC using only a limited number of blood samples in STEMI patients. This retrospective study was carried out on pooled data from five clinical trials of STEMI patients (TIMI blood flow 0/1) studies where repeated blood samples were collected within 72 h after admission to assess creatine kinase (CK), cardiac troponin I (cTnI) and muscle-brain CK (CK-MB). Biomarker kinetics was assessed using previously described biomarker kinetic models. A number of LSS models including combinations of 1 to 3 samples were developed to identify sampling times leading to the best estimation of AUC. Patients were randomly assigned to either learning (2/3) or validation (1/3) subsets. Descriptive and predictive performances of LSS models were compared using learning and validation subsets, respectively. An external validation cohort was used to validate the model and its applicability to different cTnI assays, including high-sensitive (hs) cTnI. 132 patients had full CK and cTnI dataset, 49 patients had CK-MB. For each biomarker, 180 LSS models were tested. Best LSS models were obtained for the following sampling times: T4–16 for CK, T8–T20 for cTnI and T8–T16 for CK-MB for 2-sample LSS; and T4–T16–T24 for CK, T4–T12–T20 for cTnI and T8–T16–T20 for CK-MB for 3-sample LSS. External validation was achieved on 103 anterior STEMI patients (TIMI flow 0/1), and the cTnI model applicability to recommended hs cTnI confirmed. Biomarker kinetics can be assessed with a limited number of samples using kinetic modelling. This opens the way for substantial simplification of future cardioprotection studies, more acceptable for the patients.

True verification probabilities should not be used in estimating the area under receiver operating characteristic curve.

In medical research, a two-phase study is often used for the estimation of the area under the receiver operating characteristic curve (AUC) of a diagnostic test. However, such a design introduces verification bias. One of the methods to correct verification bias is inverse probability weighting (IPW). Since the probability a subject is selected into phase 2 of the study for disease verification is known, both true and estimated verification probabilities can be used to form an IPW estimator for AUC. In this article, we derive explicit variance formula for both IPW AUC estimators and show that the IPW AUC estimator using the true values of verification probabilities even when they are known are less efficient than its counterpart using the estimated values. Our simulation results show that the efficiency loss can be substantial especially when the variance of test result in disease population is small relative to its counterpart in nondiseased population.

Assessment of a Model-Informed Precision Dosing Platform Use in Routine Clinical Care for Personalized Busulfan Therapy in the Pediatric Hematopoietic Cell Transplantation (HCT) Population.

IntroductionPopulation pharmacokinetic (PK) studies demonstrate model-based dosing for busulfan that incorporates body size and age improve clinical target attainment as compared to weight-based regimens. Recently, for clinical dosing of busulfan and TDM, our institution transitioned to a cloud-based clinical decision support tool (www.insight-rx.com). The goal of this study was to assess the dose decision tool for the achievement of target exposure of busulfan in children undergoing hematopoietic cell transplantation (HCT).Patients and MethodsPatients (N = 188) were grouped into cohorts A, B, or C based on the method for initial dose calculation and estimation of AUC: Cohort A: Initial doses were based on the conventional dosing algorithm (as outlined in the manufacturers' package insert) and non-compartmental analysis (NCA) estimation using the trapezoidal rule for estimation of AUC following TDM. Cohort B: Initial doses for busulfan were estimated by a first-generation PK model and NCA estimation of AUC following TDM. Cohort C: Initial doses were calculated by an updated, second-generation PK model available in the dose decision tool with an estimation of AUC following TDM.ResultsThe percent of individuals achieving the exposure target at the time of first PK collection was higher in subjects receiving initial doses provided by the model-informed precision dosing platform (cohort C, 75%) versus subjects receiving initial doses based on either of the two other approaches (conventional guidelines/cohort A, 25%; previous population PK model and NCA parameter estimation, cohort B, 50%). Similarly, the percent of subjects achieving the targeted cumulative busulfan exposure (cAUC) in cohort C was 100% vs. 66% and 88% for cohort A and B, respectively. For cAUC, the variability in the spread of target attainment (%CV) was low at 4.1% for cohort C as compared to cohort A (14.8%) and cohort B (17.1%).ConclusionAchievement of goal exposure early on in treatment was improved with the updated model for busulfan and the Bayesian platform. Model-informed dosing and TDM utilizing a Bayesian-based platform provides a significant advantage over conventional guidelines for the achievement of goal cAUC exposure.

Population Pharmacokinetics Analysis of Quetiapine Extended-release Formulation in Japanese Patients with Bipolar Depression

PurposeThe objectives of this study were to explore covariates of plasma quetiapine concentrations after oral administration of quetiapine extended-release formulation (XR), and to examine the exposure–response relationship in Japanese patients with bipolar depression, using population pharmacokinetics (PopPK) modeling. MethodsIn a multicenter, randomized, double-blind, placebo-controlled study of quetiapine XR in patients with bipolar depression, plasma for the measurement of quetiapine concentration was collected at weeks 2, 4, 8, 12, 20, 28, and 52 during oral administration of 150 or 300 mg once daily of quetiapine XR before bedtime. A PopPK model of quetiapine XR was developed using nonlinear mixed-effects modeling with first-order conditional estimation with interactions. The exposure–response relationship was examined using post-hoc exposures. The post-hoc AUC estimate was plotted against the change in the Montgomery-Åsberg Depression Rating Scale (ΔMADRS) total score from baseline to 8 weeks following once-daily doses at 300 mg. FindingsThe final PopPK analysis dataset contained 322 patients and 1162 observations (cutoff data at week 28; cutoff date, February 2016). The plasma quetiapine concentration–time profile in patients with bipolar depression after oral administration of quetiapine XR was represented well using a 1-compartment with first-order absorption model. Covariate analysis led to the selection of γ-glutamyl transpeptidase on apparent oral clearance and body weight on apparent volume of distribution as covariates. The final population mean values of apparent oral clearance and apparent volume of distribution were 87.7 L/h and 277 L, respectively, and the interindividual %CVs were 32.6% and 75.0%, respectively. ImplicationsThe effects of covariates on PK parameters were not large compared with the interindividual variability. In addition, there was no clear relationship between the AUC and ΔMADRS. ClinicalTrials.gov identifier: NCT01725308.