Covariate selection in pharmacometric analyses: a review of methods

Hutmacher, Matthew M.; Kowalski, Kenneth G.

doi:10.1111/bcp.12451

Cited by 65 publications

(81 citation statements)

References 74 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…All LASSO models were fit using the glmnet v2.0-5 and hdnom v4.6 packages (cv.glmnet and hdcox.lasso functions) (Friedman et al, 2010; Xiao et al, 2016). For comparison, we also used a stepwise selection algorithm for model selection (stepAIC function in the R MASS package); stepwise model selection, while widely used, has poor predictive performance compared to modern approaches like LASSO penalized regression (Hutmacher and Kowalski, 2015; Walter and Tiemeier, 2009). We selected the LASSO-penalized Cox model that resulted in minimal prediction error, using Leave-One-Out Cross-Validation (LOOCV), and assessed the stability of the results by bootstrap analysis (n=1000 times).…”

Section: Star Methodsmentioning

confidence: 99%

“…We removed 7 with many missing cases, leaving 13 for multivariate Cox regression analysis (Figure S7A). Results of nine candidate penalized methods were approximately equivalent (Figure S7B), we chose LASSO regression (Hutmacher and Kowalski, 2015; Walter and Tiemeier, 2009) to fit a multivariate model. For mRNA, lncRNA, miRNA and MSig subtypes, we set the best-survival subtype as the reference variable.…”

Section: Univariate and Multivariate Survivalmentioning

confidence: 99%

See 1 more Smart Citation

Comprehensive Molecular Characterization of Muscle-Invasive Bladder Cancer

Robertson¹,

Kim²,

Al‐Ahmadie³

et al. 2017

Cell

1,745

1,624

View full text Add to dashboard Cite

Summary We report a comprehensive analysis of 412 muscle-invasive bladder cancers characterized by multiple TCGA analytical platforms. Fifty-eight genes were significantly mutated, and the overall mutational load was associated with APOBEC-signature mutagenesis. Clustering by mutation signature identified a high-mutation subset with 75% 5-year survival. mRNA expression clustering refined prior clustering analyses and identified a poor-survival ‘neuronal’ subtype in which the majority of tumors lacked small cell or neuroendocrine histology. Clustering by mRNA, lncRNA, and miRNA expression converged to identify subsets with differential epithelial-mesenchymal transition status, carcinoma-in-situ scores, histologic features, and survival. Our analyses identified 5 expression subtypes that may stratify response to different treatments.

show abstract

Section: Star Methodsmentioning

confidence: 99%

Section: Univariate and Multivariate Survivalmentioning

confidence: 99%

Comprehensive Molecular Characterization of Muscle-Invasive Bladder Cancer

Robertson¹,

Kim²,

Al‐Ahmadie³

et al. 2017

Cell

1,745

1,624

View full text Add to dashboard Cite

show abstract

“…For continuous variables such as estimated glomerular filtration rate (eGFR), age, weight, and body mass index (BMI), subject variables were centered at medians. Then, linear, natural log, and power models were evaluated [46]. The covariates were assessed by likelihood ratio test (LRT), with a forward selection at α = 0.01 (OFV decreased by 6.64), then a backward elimination at α = 0.001 (OFV increased by 10.8).…”

Section: Methodsmentioning

confidence: 99%

Model Linking Plasma and Intracellular Tenofovir/Emtricitabine with Deoxynucleoside Triphosphates

et al. 2016

View full text Add to dashboard Cite

The coformulation of the nucleos(t)ide analogs (NA) tenofovir (TFV) disoproxil fumarate (TDF) and emtricitabine (FTC) is approved for HIV-infection treatment and prevention. Plasma TFV and FTC undergo complicated hybrid processes to form, accumulate, and retain as their active intracellular anabolites: TFV-diphosphate (TFV-DP) and FTC-triphosphate (FTC-TP). Such complexities manifest in nonlinear intracellular pharmacokinetics (PK). In target cells, TFV-DP/FTC-TP compete with endogenous deoxynucleoside triphosphates (dNTP) at the active site of HIV reverse transcriptase, underscoring the importance of analog:dNTP ratios for antiviral efficacy. However, NA such as TFV and FTC have the potential to disturb the dNTP pool, which could augment or reduce their efficacies. We conducted a pharmacokinetics-pharmacodynamics (PKPD) study among forty subjects receiving daily TDF/FTC (300 mg/200 mg) from the first-dose to pharmacological intracellular steady-state (30 days). TFV/FTC in plasma, TFV-DP/FTC-TP and dNTPs in peripheral blood mononuclear cells (PBMC) were quantified using validated LC/MS/MS methodologies. Concentration-time data were analyzed using nonlinear mixed effects modeling (NONMEM). Formations and the accumulation of intracellular TFV-DP/FTC-TP was driven by plasma TFV/FTC, which was described by a hybrid of first-order formation and saturation. An indirect response link model described the interplay between TFV-DP/FTC-TP and the dNTP pool change. The EC50 (interindividual variability, (%CV)) of TFV-DP and FTC-TP on the inhibition of deoxyadenosine triphosphate (dATP) and deoxycytidine triphosphate (dCTP) production were 1020 fmol/106 cells (130%) and 44.4 pmol/106 cells (82.5%), resulting in (90% prediction interval) 11% (0.45%, 53%) and 14% (2.6%, 35%) reductions. Model simulations of analog:dNTP molar ratios using IPERGAY dosing suggested that FTC significantly contributes to the protective effect of preexposure prophylaxis (PrEP). Simulation-based intracellular operational multiple dosing half-lives of TFV-DP and FTC-TP were 6.7 days and 33 hours. This model described the formation of intracellular TFV-DP/FTC-TP and the interaction with dNTPs, and can be used to simulate analog:dNTP time course for various dosing strategies.

show abstract

“…However, this would also increase the number of preselected false positives, thereby increasing computational time. It is also important to note that the use of preselection techniques does not reduce the importance of considering the scientific plausibility of the tested covariate relationships a priori, as this remains crucial to limit the amount of spurious covariates included (6,17).…”

Section: Discussionmentioning

confidence: 99%

“…Whether or not the inclusion of a particular covariate significantly improves the RTTE model is commonly tested with the likelihood ratio test (LRT), which tests the difference in likelihood of a model with and without inclusion of a covariate relationship for statistical significance (5,6). However, the performance of LRT as a method for covariate selection in RTTE models has only been evaluated for binary covariates (2).…”

Section: Introductionmentioning

confidence: 99%

Covariates in Pharmacometric Repeated Time-to-Event Models: Old and New (Pre)Selection Tools

Goulooze

Krekels

Hankemeier

et al. 2018

AAPS J

View full text Add to dashboard Cite

During covariate modeling in pharmacometrics, computational time can be reduced by using a fast preselection tool to identify a subset of promising covariates that are to be tested with the more computationally demanding likelihood ratio test (LRT), which is considered to be the standard for covariate selection. There is however a lack of knowledge on best practices for covariate (pre)selection in pharmacometric repeated time-to-event (RTTE) models. Therefore, we aimed to systematically evaluate the performance of three covariate (pre)selection tools for RTTE models: the likelihood ratio test (LRT), the empirical Bayes estimates (EBE) test, and a novel Schoenfeld-like residual test. This was done in simulated datasets with and without a Btrue^time-constant covariate, and both in the presence and absence of high EBE shrinkage. In scenarios with a Btrue^covariate effect, all tools had comparable power to detect this effect. In scenarios without a Btrue^covariate effect, the false positive rates of the LRT and the Schoenfeld-like residual test were slightly inflated to 5.7% and 7.2% respectively, while the EBE test had no inflated false positive rate. The presence of high EBE shrinkage (> 40%) did not affect the performance of any of the covariate (pre)selection tools. We found the EBE test to be a fast and accurate tool for covariate preselection in RTTE models. The novel Schoenfeld-like residual test proposed here had a similar performance in the tested scenarios and might be applied more readily to time-varying covariates, such as drug concentration and dynamic biomarkers.

show abstract

Covariate selection in pharmacometric analyses: a review of methods

Cited by 65 publications

References 74 publications

Comprehensive Molecular Characterization of Muscle-Invasive Bladder Cancer

Comprehensive Molecular Characterization of Muscle-Invasive Bladder Cancer

Model Linking Plasma and Intracellular Tenofovir/Emtricitabine with Deoxynucleoside Triphosphates

Covariates in Pharmacometric Repeated Time-to-Event Models: Old and New (Pre)Selection Tools

Contact Info

Product

Resources

About