Population Pharmacodynamic Dose-Response Analysis of Serum Potassium Following Dosing with Sodium Zirconium Cyclosilicate.

BACKGROUND: Sodium zirconium cyclosilicate (SZC) is an approved oral treatment for hyperkalemia that selectively binds potassium (K+) in the gastrointestinal tract and removes K+ from the body through increased fecal excretion. Here, we describe the population pharmacodynamic (PopPD) response of serum K+ concentration in patients with hyperkalemia who are treated with SZC, estimate the impact of patients' intrinsic and extrinsic factors, and compare predicted serum K+ responses between 5 g alternate daily (QOD) and 2.5 g once daily (QD) maintenance doses. METHODS: PopPD analysis was based on pooled data from seven phase II and III clinical trials for SZC. A semi-mechanistic longitudinal mixed-effects (base) model was used to characterize serum K+ concentration after SZC dosing. Indirect-response, virtual pharmacokinetics-pharmacodynamics (PK-PD) modeling was used to mimic the drug exposure compartment. Full covariate modeling was used to assess covariate impact on the half-maximal effective concentration of drug (EC50), placebo response, and Kout. Models were evaluated using goodness-of-fit plots, relative standard errors, and visual predictive checks, and data were stratified to optimize model performance across subgroups. Covariate effects were evaluated based on the magnitude of change in serum K+ between baseline and end of correction phase dosing (48 h, SZC 10 g three times a day) and maintenance phase dosing (28 days, SZC 10 g QD) using a reference subject. RESULTS: The analysis data set included 2369 patients and 25,764 serum K+ observations. The mean (standard deviation) patient age was 66.0 (12) years, 61% were male, 68% were White, 34% had congestive heart failure, and 62% had diabetes. Mean (standard deviation) serum K+ at baseline was 5.49 (0.43) mmol/L. Both the base and full covariance models adequately described observed data. In the final model, there was a sigmoid exposure response on Kin, with EC50 of 32.8 g and a Hill coefficient of 1.36. The predicted placebo-adjusted dose-responses of serum K+ change appeared nearly linear in the correction and maintenance phases. No clinically meaningful difference in placebo-adjusted serum K+ change from baseline at 28 days was observed between maintenance regimens of SZC 5 g QOD and 2.5 g QD. A greater SZC treatment response was associated with high serum K+ at baseline, advanced age, lower body weight, lower estimated glomerular filtration rate, and Black/African American and Asian race, compared with the reference patient. The impact of heart failure status and diabetes status was only minor. CONCLUSIONS: The PopPD model of SZC adequately described changes in serum K+ concentration during correction and maintenance phase dosing. A greater treatment response was associated with various covariates, but the impact of each was modest. Overall, these findings suggest that no adjustment in SZC dose is needed for any of the covariates evaluated.

Population Pharmacokinetic Modeling of Cotadutide: A Dual Agonist Peptide of Glucagon-Like Peptide and Glucagon Receptors Administered to Participants with Type II Diabetes Mellitus, Chronic Kidney Disease, Obesity and Non-Alcoholic Steatohepatitis.

BACKGROUND: Cotadutide is a dual glucagon-like peptide-1 (GLP-1) and glucagon (GCG) receptor agonist peptide. The objective of this analysis was to develop a population pharmacokinetic (popPK) model of cotadutide, and to identify any potential effect on the PK from intrinsic and extrinsic covariates. METHODS: The popPK analysis utilized a non-linear mixed-effects modeling approach using the data from 10 clinical studies in different participant categories following once-daily subcutaneous dose administration ranging from 20 to 600 µg. Additionally, the covariates affecting cotadutide exposure were quantified, and the model performance was evaluated through the prediction-corrected visual predictive checks. RESULTS: A one-compartment model with first-order absorption and elimination adequately described the data as confirmed via visual predictive check plots and parameter plausibility. The mean values for cotadutide apparent clearance (CL/F), apparent volume of distribution (V/F), absorption rate constant (Ka), and half-life were 1.05 L/h, 20.0 L, 0.38 h-1, and 13.3 hours, respectively. Covariate modeling identified body weight, alanine transaminase, albumin, anti-drug antibody (ADA) titer values, formulation strength and injection device, and participant categories as significant covariates on PK parameters, where ADAs have been identified to decrease cotadutide clearance. The model demonstrated that a 150-kg participant was estimated to have 30% lower for both AUC and Cmax and a 66 kg participant was estimated to have 35% higher for both AUC and Cmax relative to a reference individual with a median weight of 96 kg. CONCLUSIONS: A popPK model was developed for cotadutide with cotadutide clinical data, and the impact of the statistically significant covariates identified was not considered clinically meaningful. The popPK model will be used to evaluate exposure-response relationships for cotadutide clinical data.

Zibotentan in combination with dapagliflozin compared with dapagliflozin in patients with chronic kidney disease (ZENITH-CKD): a multicentre, randomised, active-controlled, phase 2b, clinical trial.

BACKGROUND: In patients with chronic kidney disease, SGLT2 inhibitors and endothelin A receptor antagonists (ERAs) can reduce albuminuria and glomerular filtration rate (GFR) decline. We assessed the albuminuria-lowering efficacy and safety of the ERA zibotentan combined with the SGLT2 inhibitor dapagliflozin. METHODS: ZENITH-CKD was a multicentre, randomised, double-blind, active-controlled clinical trial, done in 170 clinical practice sites in 18 countries. Adults (≥18 to ≤90 years) with an estimated GFR (eGFR) of 20 mL/min per 1·73 m2 or greater and a urinary albumin-to-creatinine ratio (UACR) of 150-5000 mg/g were randomly assigned (2:1:2) to 12 weeks of daily treatment with zibotentan 1·5 mg plus dapagliflozin 10 mg, zibotentan 0·25 mg plus dapagliflozin 10 mg, or dapagliflozin 10 mg plus placebo, as adjunct to angiotensin-converting enzyme inhibitors or angiotensin receptor blockers if tolerated. The primary endpoint was a change from baseline in log-transformed UACR (zibotentan 1·5 mg plus dapagliflozin vs dapagliflozin plus placebo) at week 12. Fluid retention was an event of special interest, defined as an increase in bodyweight of at least 3% (at least 2·5% must have been from total body water) from baseline or an increase of at least 100% in B-type natriuretic peptide (BNP) and either a BNP concentration greater than 200 pg/mL if without atrial fibrillation or BNP greater than 400 pg/mL if with atrial fibrillation. This trial is registered with ClinicalTrials.gov, NCT04724837, and is completed. FINDINGS: Between April 28, 2021, and Jan 17, 2023, we assessed 1492 participants for eligibility. For the main analysis, we randomly assigned 449 (30%) participants, 447 (99%) of whom (mean age 62·8 years [SD 12·1], 138 [31%] female, 309 [69%] male, 305 [68%] White, mean eGFR 46·7 mL/min per 1·73 m2 [SD 22·4], and median UACR 565·5 mg/g [IQR 243·0-1212·6]) received treatment with zibotentan 1·5 mg plus dapagliflozin (n=179 [40%]), zibotentan 0·25 mg plus dapagliflozin (n=91 [20%]), or dapagliflozin plus placebo (n=177 [40%]). Zibotentan 1·5 mg plus dapagliflozin and zibotentan 0·25 mg plus dapagliflozin reduced UACR versus dapagliflozin plus placebo throughout the treatment period of the study. At week 12, the difference in UACR versus dapagliflozin plus placebo was -33·7% (90% CI -42·5 to -23·5; p<0·0001) for zibotentan 1·5 mg plus dapagliflozin and -27·0% (90% CI -38·4 to -13·6; p=0·0022) for zibotentan 0·25 mg plus dapagliflozin. Fluid-retention events were observed in 33 (18%) of 179 participants in the zibotentan 1·5 mg plus dapagliflozin group, eight (9%) of 91 in the zibotentan 0·25 mg plus dapagliflozin group, and 14 (8%) of 177 in the dapagliflozin plus placebo group. INTERPRETATION: Zibotentan combined with dapagliflozin reduced albuminuria with an acceptable tolerability and safety profile and is an option to reduce chronic kidney disease progression in patients already receiving currently recommended therapy. FUNDING: AstraZeneca.

Ticagrelor vs placebo for the reduction of vaso-occlusive crises in pediatric sickle cell disease: the HESTIA3 study.

The phase 3 HESTIA3 study assessed the efficacy and safety of the reversible P2Y12 inhibitor ticagrelor vs placebo in preventing vaso-occlusive crises in pediatric patients with sickle cell disease (SCD). Patients aged 2 to 17 years were randomly assigned 1:1 to receive weight-based doses of ticagrelor or matching placebo. The primary end point was the rate of vaso-occlusive crises, a composite of painful crises and/or acute chest syndrome (ACS). Key secondary end points included number and duration of painful crises, number of ACS events, and number of vaso-occlusive crises requiring hospitalization or emergency department visits. Exploratory end points included the effect of ticagrelor on platelet activation. In total, 193 patients (ticagrelor, n = 101; placebo, n = 92) underwent randomization at 53 sites across 16 countries. The study was terminated 4 months before planned completion for lack of efficacy. Median ticagrelor exposure duration was 296.5 days. The primary end point was not met: estimated yearly incidence of vaso-occlusive crises was 2.74 in the ticagrelor group and 2.60 in the placebo group (rate ratio, 1.06; 95% confidence interval, 0.75-1.50; P = .7597). There was no evidence of efficacy for ticagrelor vs placebo across secondary end points. Median platelet inhibition with ticagrelor at 6 months was 34.9% predose and 55.7% at 2 hours' postdose. Nine patients (9%) in the ticagrelor group and eight patients (9%) in the placebo group had at least one bleeding event. In conclusion, no reduction of vaso-occlusive crises was seen with ticagrelor vs placebo in these pediatric patients with SCD. This trial was registered at www.clinicaltrials.gov as #NCT03615924.

Pharmacokinetics and safety of ticagrelor in infants and toddlers with sickle cell disease aged <24 months.

Inhibition of platelet activation may reduce vaso-occlusion rates in patients with sickle cell disease (SCD). In the HESTIA4 (NCT03492931) study, 21 children with SCD received a single oral dose of the antiplatelet agent ticagrelor (0.1 mg/kg <6 months; 0.2 mg/kg ≥6 to <24 months). All patients had measurable ticagrelor plasma concentrations. Ticagrelor and active metabolite (AR-C124910XX) exposure were comparable across all groups (<6 months, ≥6 to <12 months and ≥12 to <24 months). Ticagrelor was well tolerated. Palatability was generally acceptable. These data will be used to enable dose selection for further investigations of ticagrelor efficacy and safety in children with SCD.

Pharmacokinetics of Roxadustat: A Population Analysis of 2855 Dialysis- and Non-Dialysis-Dependent Patients with Chronic Kidney Disease.

BACKGROUND: Roxadustat is a novel, small-molecule, first-in-class therapeutic that stimulates erythropoiesis by inhibiting hypoxia-inducible factor prolyl hydroxylase enzymes. This agent (roxadustat) is in clinical development for the treatment of anemia in patients with non-dialysis-dependent (NDD) and dialysis-dependent (DD) chronic kidney disease. A population pharmacokinetic analysis was undertaken to evaluate the effect of intrinsic and extrinsic factors on roxadustat pharmacokinetics. METHODS: Non-linear mixed-effects models implemented in NONMEM software were fitted to 8209 pharmacokinetic samples from 2855 DD and NDD subjects enrolled in four phase III studies with roxadustat dose concentrations of 20-400 mg as orally administered tablets. Effects of intrinsic and extrinsic factors were evaluated using a stepwise covariate modeling procedure in combination with the full covariate approach, and defined no-effect boundaries for exposure were based on the difference in exposure between 70 and 100 mg of roxadustat (i.e., - 30%, + 43%). RESULTS: A two-compartment model with first-order absorption adequately described roxadustat pharmacokinetics, with parameter estimates (relative standard error) for apparent clearance of 1.1 (0.0223) L/h in NDD subjects, and apparent central and peripheral volumes of distribution of 14.9 (0.0278) L and 9.5 (0.0872) L, respectively. Stepwise covariate modeling identified bodyweight, dialysis status, race, and dose as statistically significant covariates on apparent clearance, and bodyweight, sex, and albumin as statistically significant covariates on apparent central volume of distribution. However, the effects of these covariates did not result in roxadustat area under the curve or maximum plasma concentration changes outside of the defined no-effect boundaries. The effects of concomitant oral iron, clopidogrel, and staggered sevelamer, calcium carbonate, or calcium acetate were investigated using a full covariate approach but did not result in roxadustat area under the curve or maximum plasma concentration changes outside of the defined no-effect boundaries. CONCLUSIONS: A population pharmacokinetic model was developed for the pharmacokinetics of roxadustat in the target population. None of the investigated intrinsic or extrinsic factors resulted in a significant change in roxadustat exposure outside of the defined no-effect boundaries.

Dose-Response Mixed Models for Repeated Measures - a New Method for Assessment of Dose-Response.

PURPOSE: In this paper we investigated a new method for dose-response analysis of longitudinal data in terms of precision and accuracy using simulations. METHODS: The new method, called Dose-Response Mixed Models for Repeated Measures (DR-MMRM), combines conventional Mixed Models for Repeated Measures (MMRM) and dose-response modeling. Conventional MMRM can be applied for highly variable repeated measure data and is a way to estimate the drug effect at each visit and dose, however without any assumptions regarding the dose-response shape. Dose-response modeling, on the other hand, utilizes information across dose arms and describes the drug effect as a function of dose. Drug development in chronic kidney disease (CKD) is complicated by many factors, primarily by the slow progression of the disease and lack of predictive biomarkers. Recently, new approaches and biomarkers are being explored to improve efficiency in CKD drug development. Proteinuria, i.e. urinary albumin-to-creatinine ratio (UACR) is increasingly used in dose finding trials in patients with CKD. We use proteinuria to illustrate the benefits of DR-MMRM. RESULTS: The DR-MMRM had higher precision than conventional MMRM and less bias than a dose-response model on UACR change from baseline to end-of-study (DR-EOS). CONCLUSIONS: DR-MMRM is a promising method for dose-response analysis.

Population Pharmacokinetics/Pharmacodynamics of Ticagrelor in Children with Sickle Cell Disease.

BACKGROUND AND OBJECTIVE: Ticagrelor, a reversible P2Y12 platelet inhibitor, is under investigation as a sickle cell disease (SCD) therapy in children. HESTIA1 (NCT02214121) was the first ticagrelor study generating pharmacokinetic (PK), pharmacodynamic (PD, P2Y12 reactivity units [PRU]), and safety data in 45 pediatric SCD patients. Population PK and PK/PD relationships for ticagrelor were quantified using a PK approach. METHODS: An adult population PK model was refined to describe ticagrelor and AR-C124910XX (active metabolite) plasma concentration and time data over a wide range of single/repeated ticagrelor doses (0.125-2.25 mg/kg). Population PK/PD modeling was used to describe the time course and extent of platelet inhibition. Demographic covariate relationships were investigated. RESULTS: The final population PK model adequately described ticagrelor and AR-C124910XX plasma concentrations over time. An allometric body weight relationship between ticagrelor and AR-C124910XX clearances and volumes of distribution was used. Significant covariates for ticagrelor were sex (relative bioavailability) and cholecystectomy (central volume of distribution). Estimated oral clearances (35 kg patient; median bodyweight) were 22.8 L/h (ticagrelor) and 9.97 L/h (AR-C124910XX). The final population PK/PD model well-described the time course and extent of platelet inhibition. Estimated baseline PRU was 283, maximum PRU effect was fixed at 1, and the ticagrelor concentration for half-maximum PRU effect was 233 nmol/L. CONCLUSIONS: These analyses offer the first quantitative characterization of the dose-exposure-response relationship for ticagrelor in pediatric SCD patients. This model-based approach may be used to inform dose selection and design of subsequent studies that aim to define ticagrelor safety and efficacy in pediatric SCD patients.

Model-based characterization of the relationship between dapagliflozin systemic exposure and HbA1c response in patients with type 1 diabetes mellitus.

AIMS: To quantitatively describe the relationship between dapagliflozin systemic exposure and HbA1c response among patients with type 1 diabetes mellitus (T1DM) and assess the potential impact of covariate effects. MATERIALS AND METHODS: Individual longitudinal HbA1c data from two phase 3 studies in patients with T1DM (24-week treatment with once-daily dapagliflozin 5 or 10 mg or placebo, with adjustable insulin) were analyzed using a non-linear mixed effect modeling approach. Area under the concentration curve was used to measure dapagliflozin systemic exposure. Baseline HbA1c, estimated glomerular filtration rate, reduction in total insulin dose, baseline glucose concentrations, age, sex, race (Asian vs. non-Asian), and insulin administration method (multiple daily injections vs. insulin pump) were assessed as covariates. RESULTS: A maximum effect (Emax ) model identified a positive exposure-response relationship. Model-predicted placebo-corrected HbA1c reductions after 24 weeks for dapagliflozin 5- and 10-mg doses were - 0.42% [95% confidence interval (CI) -0.47 to -0.36) and - 0.45% (95% CI -0.50 to -0.40), respectively; baseline HbA1c was ~8.4%. This was in good agreement with actual observations from both studies. Baseline HbA1c was a significant covariate: patients with higher baseline HbA1c were predicted to have greater HbA1c reductions. CONCLUSIONS: The relationship between dapagliflozin systemic exposure and HbA1c response was successfully described in patients with T1DM. None of the tested covariates affected the efficacy of dapagliflozin to a clinically relevant extent. Therefore, no dose adjustment of dapagliflozin is required in patients with T1DM based on the tested covariates. ClinicalTrials.gov, NCT02268214; NCT02460978.

Pharmacokinetic-pharmacodynamic modelling of platelet response to ticagrelor in stable coronary artery disease and prior myocardial infarction patients.

AIMS: To characterize ticagrelor exposure-response relationship for platelet inhibition in patients with stable coronary artery disease (CAD) and a history of myocardial infarction (MI), using nonlinear mixed effects modelling and simulation. METHODS: Platelet function data were integrated with plasma concentration data of ticagrelor and its active metabolite AR-C1249010XX in a population pharmacokinetic (PK) and pharmacodynamic (PD) model, based on two clinical studies. In the ONSET/OFFSET study, PK and platelet function were assessed in 123 CAD patients receiving placebo, ticagrelor (180 mg followed by 90 mg twice daily) or clopidogrel (600 mg followed by 75 mg once daily). In the PEGASUS-TIMI 54 platelet function substudy, PK and platelet function were assessed during maintenance dosing in 180 prior MI patients receiving placebo, ticagrelor 60 mg or ticagrelor 90 mg twice daily. RESULTS: Platelet inhibition by ticagrelor was described by a sigmoidal Emax model. On average, half maximal inhibition was reached at ticagrelor concentrations of 116 (RSE: 5.3%) nmol l-1 . Simulations showed that near maximal platelet inhibition is achieved with both ticagrelor 60 and 90 mg twice daily. At simulated lower doses, platelet inhibition is overall reduced, more variable between patients, and show greater peak-to-trough variability. Ticagrelor antiplatelet response was similar between the studied patient populations. CONCLUSIONS: In patients with stable CAD or a history of MI, near maximal platelet inhibition is achieved with both ticagrelor 60 and 90 mg twice daily. At modelled doses <60 mg, the response is reduced overall, more variable between patients, and patients will display greater peak-to-trough variability.

Effect of Tenapanor on Serum Phosphate in Patients Receiving Hemodialysis.

Hyperphosphatemia is common among patients with CKD stage 5D and is associated with morbidity and mortality. Current guidelines recommend lowering serum phosphate concentrations toward normal. Tenapanor is a minimally absorbed small molecule inhibitor of the sodium/hydrogen exchanger isoform 3 that functions in the gut to reduce sodium and phosphate absorption. This randomized, double-blind, placebo-controlled trial assessed the effects of tenapanor on serum phosphate concentration in patients with hyperphosphatemia receiving hemodialysis. After a 1- to 3-week washout of phosphate binders, we randomly assigned 162 eligible patients (serum phosphate =6.0 to <10.0 mg/dl and a 1.5-mg/dl increase from before washout) to one of six tenapanor regimens (3 or 30 mg once daily or 1, 3, 10, or 30 mg twice daily) or placebo for 4 weeks. The primary efficacy end point was change in serum phosphate concentration from baseline (randomization) to end of treatment. In total, 115 patients (71%) completed the study. Mean serum phosphate concentrations at baseline (after washout) were 7.32-7.92 mg/dl for tenapanor groups and 7.87 mg/dl for the placebo group. Tenapanor provided dose-dependent reductions in serum phosphate level from baseline (least squares mean change: tenapanor =0.47-1.98 mg/dl; placebo =0.54 mg/dl; P=0.01). Diarrhea was the most common adverse event (tenapanor =18%-68%; placebo =12%) and frequent at the highest tenapanor doses. In conclusion, tenapanor treatment resulted in statistically significant, dose-dependent reductions in serum phosphate concentrations in patients with hyperphosphatemia receiving hemodialysis. Additional studies are required to clarify the optimal dosing of tenapanor in patients with CKD-related hyperphosphatemia.

Activity of neutrophil elastase reflects the progression of acute pancreatitis.

OBJECTIVE: Neutrophil elastase (NE) concentration is associated with progression of acute pancreatitis (AP), but measuring total NE concentration includes biologically inactive NE. This study aims to investigate the relationship between NE activity and the aetiology and severity of AP and associated organ failure. METHODS: Seventy-five patients admitted to our surgery department with a first episode of AP during 2004-2005 were age- and sex-matched to 20 healthy volunteers (controls). NE activity was assessed using venous blood samples obtained on patient admission and after 1, 2 and 14 days. One sample was also taken from each control. ANOVA was used for statistical comparison between groups. RESULTS: Baseline NE activity (geometric mean; 95% confidence intervals) differed between patients (58.6 nM of substrate 7-amino-4-methylcoumarin [AMC]/hour; 48.52-70.72) and controls (31.5 nM AMC/hour; 25.5-39.0) (p = 0.0003), and did not correlate with time between symptom onset and admission. Patients with alcohol-induced AP demonstrated higher mean activity (59.1 nM AMC/h; 44.7-78.2) than those with gallstone-induced AP (41.7 nM AMC/h; 33.9-51.4) (p = 0.0496). NE activity was higher overall in patients with predicted severe AP (60.9 nM AMC/h; 48.0-77.2) than in those with predicted mild AP (42.1 nM AMC/h; 34.9-50.8) (p = 0.027). Patients with respiratory failure had higher NE activity (82.5 nM AMC/h; 57.5-118.4) than those without (43.9 nM AMC/h; 37.6-51.3) (p = 0.0024). CONCLUSIONS: NE activity was associated with predicted severity of AP and AP-associated respiratory failure. Specific NE inhibitors may have therapeutic potential in acute pancreatitis.

Prevention of relapse of healed reflux esophagitis is related to the duration of intragastric pH > 4.

Proton pump inhibitors (PPIs) are the preferred treatment for maintenance of healed reflux esophagitis (RE). However, little is known regarding the relationship between prevention of RE relapse and degree of gastric acid suppression. The aim of this review was to examine this relationship in further detail. Data from four comparative studies on maintenance PPI therapy for prevention of relapse of RE were combined with data from two pharmacodynamic studies of duration of intragastric pH >4 during the 24-hour period on day 5 of PPI dosing in healthy subjects. A log-linear model was fitted to the data using the method of maximum likelihood. Variability in relapse rates and pH data was taken into account using a binomial and normal likelihood function, respectively. Pharmacodynamic studies resulted in a wide range of acid-suppressive effect, and based on corresponding maintenance of RE healing rates with different PPIs and doses, an inverse (non-linear) statistically significant relationship between percentage of time with pH >4 and maintenance of RE healing was identified (P<0.001). These findings indicate that long-term maintenance of healed RE is related to the extent of acid suppression in a 24-hour period.

Empirical Bayes models for multiple probe type microarrays at the probe level.

BACKGROUND: When analyzing microarray data a primary objective is often to find differentially expressed genes. With empirical Bayes and penalized t-tests the sample variances are adjusted towards a global estimate, producing more stable results compared to ordinary t-tests. However, for Affymetrix type data a clear dependency between variability and intensity-level generally exists, even for logged intensities, most clearly for data at the probe level but also for probe-set summarizes such as the MAS5 expression index. As a consequence, adjustment towards a global estimate results in an intensity-level dependent false positive rate. RESULTS: We propose two new methods for finding differentially expressed genes, Probe level Locally moderated Weighted median-t (PLW) and Locally Moderated Weighted-t (LMW). Both methods use an empirical Bayes model taking the dependency between variability and intensity-level into account. A global covariance matrix is also used allowing for differing variances between arrays as well as array-to-array correlations. PLW is specially designed for Affymetrix type arrays (or other multiple-probe arrays). Instead of making inference on probe-set summaries, comparisons are made separately for each perfect-match probe and are then summarized into one score for the probe-set. CONCLUSION: The proposed methods are compared to 14 existing methods using five spike-in data sets. For RMA and GCRMA processed data, PLW has the most accurate ranking of regulated genes in four out of the five data sets, and LMW consistently performs better than all examined moderated t-tests when used on RMA, GCRMA, and MAS5 expression indexes.

Improved covariance matrix estimators for weighted analysis of microarray data.

Empirical Bayes models have been shown to be powerful tools for identifying differentially expressed genes from gene expression microarray data. An example is the WAME model, where a global covariance matrix accounts for array-to-array correlations as well as differing variances between arrays. However, the existing method for estimating the covariance matrix is very computationally intensive and the estimator is biased when data contains many regulated genes. In this paper, two new methods for estimating the covariance matrix are proposed. The first method is a direct application of the EM algorithm for fitting the multivariate t-distribution of the WAME model. In the second method, a prior distribution for the log fold-change is added to the WAME model, and a discrete approximation is used for this prior. Both methods are evaluated using simulated and real data. The first method shows equal performance compared to the existing method in terms of bias and variability, but is superior in terms of computer time. For large data sets (>15 arrays), the second method also shows superior computer run time. Moreover, for simulated data with regulated genes the second method greatly reduces the bias. With the proposed methods it is possible to apply the WAME model to large data sets with reasonable computer run times. The second method shows a small bias for simulated data, but appears to have a larger bias for real data with many regulated genes.

Contrast normalization of oligonucleotide arrays.

Affymetrix high-density oligonucleotide array is a tool that has the capacity to simultaneously measure the abundance of thousands of mRNA sequences in biological samples. In order to allow direct array-to-array comparisons, normalization is a necessity. When deciding on an appropriate normalization procedure there are a couple questions that need to be addressed, e.g., on which level should the normalization be performed: On the level of feature intensities or on the level of expression indexes? Should all features/expression indexes be used or can we choose a subset of features likely to be unregulated? Another question is how to actually perform the normalization: normalize using the overall mean intensity or use a smooth normalization curve? Most of the currently used normalization methods are linear; e.g., the normalization method implemented in the Affymetrix software GeneChip is based on the overall mean intensity. However, along with alternative methods of summarizing feature intensities into an expression index, nonlinear methods have recently started to appear. For many of these alternative methods, the natural choice is to normalize on the level of feature intensities, either using all feature intensities or only perfect match intensities. In this report, a nonlinear normalization procedure aimed for normalizing feature intensities is proposed.