Related Articles

The extensive interindividual variability in oral bioavailability of itraconazole prompted an assessment of the bioequivalence of two formulations marketed in Brazil, namely, Sporanox (reference) and Traconal (test). Eighteen healthy volunteers received single 200-mg oral doses of each formulation at 2-week intervals in a randomized, crossover protocol. The concentrations of itraconazole and hydroxy-itraconazole in plasma were measured by high-performance liquid chromatography, and the datum points (n = 396) were subsequently used to develop limited-sampling strategy models for estimation of the areas under the curve (AUCs) for both compounds. The 90% confidence intervals for individual percent ratios (test/reference formulations) of the maximum concentration of drug in serum, the AUC from 0 to 48 h and the AUC from time zero to infinity (AUC0–∞) for itraconazole and hydoxy-itraconazole were below the range of 80 to 125%, suggesting that these formulations are not bioequivalent. Linear regression analysis of the AUC0–∞ against time and a “jackknife” validation procedure revealed that models based on three sampling times accurately predict (R2, >0.98; bias, <3%; precision, 3 to 7%) the AUC0–∞ for each of the four formulation-compound pairs tested. Increasing the number of sampling points to more than three adds little to the accuracy of the estimates of AUC0–∞. The three-point models developed for the reference formulation were validated retrospectively and were found to predict within 2% the AUC0–∞ reported in previous studies performed under similar protocols. In conclusion, the data in this study indicate (i) that the tested formulations are not bioequivalent when single doses are compared and (ii) that limited-sampling strategy models based on three points predict accurately the AUC0–∞s for itraconazole and hydroxy-itraconazole and could be a valuable tool in pharmacokinetic and bioequivalence studies of single oral doses of itraconazole.

Antibiotic pharmacodynamic modeling allows variations in pathogen susceptibility and human pharmacokinetics to be accounted for when considering antibiotic doses, potential bacterial pathogen targets for therapy, and clinical susceptibility breakpoints. Variation in the pharmacodynamic index (area-under-the-concentration curve to 24 h [AUC24]/MIC; maximum serum concentration of drug in the serum/MIC; time the serum concentration remains higher than the MIC [T > MIC]) is not usually considered. In an in vitro pharmacokinetic model of infection using a dose-ranging design, we established the relationship between AUC24/MIC and the antibacterial effect for moxifloxacin against 10 strains of Staphylococcus aureus. The distributions of AUC24/MIC targets for 24-h bacteriostatic effect and 1-log, 2-log, and 3-log drops in bacterial counts were used to calculate potential clinical breakpoint values, and these were compared with those obtained by the more conventional approach of taking a single AUC24/MIC target. Consideration of the AUC24/MIC as a distribution rather than a single value resulted in a lower clinical breakpoint.

Motivation: The area under the receiver operating characteristic (ROC) curve (AUC), long regarded as a ‘golden’ measure for the predictiveness of a continuous score, has propelled the need to develop AUC-based predictors. However, the AUC-based ensemble methods are rather scant, largely due to the fact that the associated objective function is neither continuous nor concave. Indeed, there is no reliable numerical algorithm identifying optimal combination of a set of biomarkers to maximize the AUC, especially when the number of biomarkers is large.

Results: We have proposed a novel AUC-based statistical ensemble methods for combining multiple biomarkers to differentiate a binary response of interest. Specifically, we propose to replace the non-continuous and non-convex AUC objective function by a convex surrogate loss function, whose minimizer can be efficiently identified. With the established framework, the lasso and other regularization techniques enable feature selections. Extensive simulations have demonstrated the superiority of the new methods to the existing methods. The proposal has been applied to a gene expression dataset to construct gene expression scores to differentiate elderly women with low bone mineral density (BMD) and those with normal BMD. The AUCs of the resulting scores in the independent test dataset has been satisfactory.

Conclusion: Aiming for directly maximizing AUC, the proposed AUC-based ensemble method provides an efficient means of generating a stable combination of multiple biomarkers, which is especially useful under the high-dimensional settings.

Contact: lutian@stanford.edu

Supplementary Information: Supplementary data are available at Bioinformatics online.

Purpose

Standardized uptake values (SUV) are commonly used for quantification of whole-body [18F]fluoro-2-deoxy-D-glucose (FDG) positron emission tomography (PET) studies. Changes in SUV following therapy, however, only provide a proper measure of response in case of homogeneous FDG uptake in the tumour. The purpose of this study was therefore to implement and characterize a method that enables quantification of heterogeneity in tumour FDG uptake.

Methods

Cumulative SUV-volume histograms (CSH), describing % of total tumour volume above % threshold of maximum SUV (SUVmax), were calculated. The area under a CSH curve (AUC) is a quantitative index of tumour uptake heterogeneity, with lower AUC corresponding to higher degrees of heterogeneity. Simulations of homogeneous and heterogeneous responses were performed to assess the value of AUC-CSH for measuring uptake and/or response heterogeneity. In addition, partial volume correction and image denoising was applied prior to calculating AUC-CSH. Finally, the method was applied to a number of human FDG scans.

Results

Partial volume correction and noise reduction improved CSH curves. Both simulations and clinical examples showed that AUC-CSH values corresponded with level of tumour heterogeneity and/or heterogeneity in response. In contrast, this correspondence was not seen with SUVmax alone. The results indicate that the main advantage of AUC-CSH above other measures, such as 1/COV (coefficient of variation), is the possibility to measure or normalize AUC-CSH in different ways.

Conclusion

AUC-CSH might be used as a quantitative index of heterogeneity in tracer uptake. In response monitoring studies it can be used to address heterogeneity in response.

In order to comply with the requirements for a drug listed in China, the study was developed to compare the pharmacokinetics and relative bioavailability of two different enteric formulations of omeprazole (OPZ) in healthy Chinese subjects. A total of 32 volunteers participated in the study. Plasma concentrations were analyzed by nonstereospecific liquid chromatography/tandem mass spectrometric (LC-MS/MS) method. After administration of a single 40-mg dose of the two OPZ formulations, the comparative bioavailability was assessed by calculating individual AUC0‒t (the area under the concentration-time curve from time zero to the last measurable concentration), AUC0‒∞ (the area under the concentration-time curve extrapolated to infinity), C max (the maximum observed concentration), and T peak (the time to C max) values of OPZ, 5-hydroxyomeprazole (OH-OPZ), and omeprazole sulfone (OPZ-SFN), respectively. The 90% confidence intervals (CIs) of AUC0‒t, AUC0‒∞, and C max were 85.4%‒99.0%/88.8%‒98.6%/87.6%‒99.4%, 85.5%‒99.2%/89.0%‒98.6%/88.5%‒101.3%, and 72.3%‒87.6%/79.6%‒91.1%/88.4%‒99.1% for OPZ/OH-OPZ/OPZ-SFN, respectively, and T peak values did not differ significantly. In this study, the test formulation of OPZ in fasting healthy Chinese male volunteers met the Chinese bioequivalance standard to the reference formulation based on AUC, C max, and T peak.

Aims

To assess the relationship between genetic polymorphisms and indinavir pharmacokinetic variability and to study the link between concentrations and short-term response or metabolic safety.

Methods

Forty protease inhibitor naive-patients initiating indinavir/ritonavir containing HAART and enrolled in the COPHAR 2 - ANRS 111 trial were studied. At week 2, 4 blood samples were taken before and up to 6 hours following drug intake. A population pharmacokinetic analysis was performed using the Stochastic Approximation Expectation Maximization (SAEM) algorithm implemented in the MONOLIX software. Indinavir area under the concentration-time curve (AUC), maximal (Cmax) and trough concentrations (Ctrough) were derived from the population model and tested for correlation with short-term viral response and safety measurements, while ritonavir AUC, Cmax and Ctrough were tested for correlation with short-term biochemical safety.

Results

A one-compartment model with first-order absorption and elimination best described both indinavir and ritonavir concentrations. For indinavir, the estimated clearance and volume of distribution were 22.2 L/h and 97.3 L respectively. The eight patients *1B/*1B for CYP3A4 gene had an absorption decreased by 70% compared to *1A/*1B or *1A/*1A genotypes (0.5 versus 2.1, P=0.04, likelihood ratio test by permutation). Indinavir AUC and Ctrough were positively correlated with the HIV RNA decrease between week 0 and week 2 (r=−0.4, P = 0.03 and r=−0.4, P = 0.03, respectively). Patients with the *1B/*1B genotype had significantly lower indinavir Cmax (median 3.6 [range 2.1 – 5.2] ng/mL versus 4.4 [2.2 – 8.3] ng/mL, P=0.04) and a lower triglycerides increase during the first 4 weeks of treatment (0.1 [−0.7 – 1.4] versus 0.6 [−0.5 – 1.7] mmol/L, P = 0.02). For ritonavir, the estimated clearance and volume of distribution were 8.3 L/h and 60.7 L respectively and concentrations were not found to be correlated to biochemical safety. Indinavir and ritonavir absorption rate constants were found to be correlated, as well as their apparent volumes of distribution and clearances indicating correlated bioavailability of the two drugs.

Conclusion

CYP3A4*1B polymorphism was found to influence the pharmacokinetics of indinavir and in some extent the biochemical safety of indinavir.

Background:

Area under the concentration–time curve (AUC) is a pharmacokinetic parameter that represents overall exposure to a drug. For selected anti-infective agents, pharmacokinetic–pharmacodynamic parameters, such as AUC/MIC (where MIC is the minimal inhibitory concentration), have been correlated with outcome in a few studies. A limited-sampling strategy may be used to estimate pharmacokinetic parameters such as AUC, without the frequent, costly, and inconvenient blood sampling that would be required to directly calculate the AUC.

Objective:

To discuss, by means of a systematic review, the strengths, limitations, and clinical implications of published studies involving a limited-sampling strategy for anti-infective agents and to propose improvements in methodology for future studies.

Methods:

The PubMed and EMBASE databases were searched using the terms “anti-infective agents”, “limited sampling”, “optimal sampling”, “sparse sampling”, “AUC monitoring”, “abbreviated AUC”, “abbreviated sampling”, and “Bayesian”. The reference lists of retrieved articles were searched manually. Included studies were classified according to modified criteria from the US Preventive Services Task Force.

Results:

Twenty studies met the inclusion criteria. Six of the studies (involving didanosine, zidovudine, nevirapine, ciprofloxacin, efavirenz, and nelfinavir) were classified as providing level I evidence, 4 studies (involving vancomycin, didanosine, lamivudine, and lopinavir–ritonavir) provided level II-1 evidence, 2 studies (involving saquinavir and ceftazidime) provided level II-2 evidence, and 8 studies (involving ciprofloxacin, nelfinavir, vancomycin, ceftazidime, ganciclovir, pyrazinamide, meropenem, and alpha interferon) provided level III evidence. All of the studies providing level I evidence used prospectively collected data and proper validation procedures with separate, randomly selected index and validation groups. However, most of the included studies did not provide an adequate description of the methods or the characteristics of included patients, which limited their generalizability.

Conclusions:

Many limited-sampling strategies have been developed for anti-infective agents that do not have a clearly established link between AUC and clinical outcomes in humans. Future studies should first determine if there is an association between AUC monitoring and clinical outcomes. Thereafter, it may be worthwhile to prospectively develop and validate a limited-sampling strategy for the particular anti-infective agent in a similar population.

The area under the ROC curve (AUC) and partial area under the ROC curve (pAUC) are summary measures used to assess the accuracy of a biomarker in discriminating true disease status. The standard sampling approach used in biomarker validation studies is often inefficient and costly, especially when ascertaining the true disease status is costly and invasive. To improve efficiency and reduce the cost of biomarker validation studies, we consider a test-result-dependent sampling (TDS) scheme, in which subject selection for determining the disease state is dependent on the result of a biomarker assay. We first estimate the test-result distribution using data arising from the TDS design. With the estimated empirical test-result distribution, we propose consistent nonparametric estimators for AUC and pAUC and establish the asymptotic properties of the proposed estimators. Simulation studies show that the proposed estimators have good finite sample properties and that the TDS design yields more efficient AUC and pAUC estimates than a simple random sampling (SRS) design. A data example based on an ongoing cancer clinical trial is provided to illustrate the TDS design and the proposed estimators. This work can find broad applications in design and analysis of biomarker validation studies.

We previously reported the 5-fluorouracil (5-FU) sensitivity of cancer cells obtained from colorectal cancer (CRC) patients using the collagen gel droplet-embedded culture-drug sensitivity test (CD-DST). Multiple drug concentrations and contact durations, and the area under the concentration curve (AUC) and growth inhibition rate (IR) were combined, resulting in the AUC-IR curve, which was approximated to the logarithmic curve. Moreover, the individualized AUCIR50, the AUC value which gives 50% growth inhibition, was calculated using the AUC-IR curve. This study aimed to identify responders/non-responders to 5-FU based on the individual AUCIR50 obtained with CD-DST in order to establish individualized chemotherapy for CRC patients. The individual AUCIR50 was calculated from each AUC-inhibition rate regression curve in all patients using the CD-DST. The cumulative distribution of the individual AUCIR50 in CRC patients was evaluated. The cumulative distribution of the individual AUCIR50 was regressed over the sigmoid curve (logarithmic scale). The approximate expression was almost exactly y=ab^exp(−cx) (a=0.9739, b=1.7096E-21, c=0.8990, the sum of square residuals, 0.0279). In the 80 cases examined, no notable change was observed in the regression curve when the number of patients increased. A standard curve was obtained describing responders to 5-FU among all CRC patients. From this standard curve, we ascertained that non-responders accounted for approximately 5% of all patients. Moreover, we were able to classify responders into good or intermediate responders to 5-FU. The standard curve describing response to 5-FU in CRC patients offers a useful tool in the establishment of individualized chemotherapy.

Objectives: Bobath based (BB) and movement science based (MSB) physiotherapy interventions are widely used for patients after stroke. There is little evidence to suggest which is most effective. This single-blind randomised controlled trial evaluated the effect of these treatments on movement abilities and functional independence.

Methods: A total of 120 patients admitted to a stroke rehabilitation ward were randomised into two treatment groups to receive either BB or MSB treatment. Primary outcome measures were the Rivermead Motor Assessment and the Motor Assessment Scale. Secondary measures assessed functional independence, walking speed, arm function, muscle tone, and sensation. Measures were performed by a blinded assessor at baseline, and then at 1, 3, and 6 months after baseline. Analysis of serial measurements was performed to compare outcomes between the groups by calculating the area under the curve (AUC) and inserting AUC values into Mann-Whitney U tests.

Results: Comparison between groups showed no significant difference for any outcome measures. Significance values for the Rivermead Motor Assessment ranged from p = 0.23 to p = 0.97 and for the Motor Assessment Scale from p = 0.29 to p = 0.87.

Conclusions: There were no significant differences in movement abilities or functional independence between patients receiving a BB or an MSB intervention. Therefore the study did not show that one approach was more effective than the other in the treatment of stroke patients.

The bioavailability of iron in five ferrous sulfate preparations was studied in 10 healthy male volunteers. The preparations were an oral solution, two types of film-coated tablets and two types of enteric-coated tablets. Blood samples were drawn hourly from 8 am to 6 pm on the day before each study day to assess baseline serum iron concentrations and on the study day. Spectrophotometry was used to measure the serum iron concentrations. The area under the curve (AUC), the maximum concentration and the time to achieve the maximum concentration were compared by analysis of variance. The enteric-coated preparations resulted in AUCs less than 30% of the AUC for the oral solution. The two film-coated products produced AUCs essentially equivalent to that of the oral solution. We conclude that the bioavailability of iron in the enteric-coated preparations was low, relative to that of the film-coated products and the oral solution, and that these products should not be considered interchangeable.

Antimicrobial therapy of soft tissue infections in patients with sepsis sometimes lacks efficiency, despite the documented susceptibility of the causative pathogen to the administered antibiotic. In this context, impaired equilibration between the antibiotic concentrations in plasma and those in tissues in critically ill patients has been discussed. To characterize the impact of tissue penetration of anti-infective agents on antimicrobial killing, we used microdialysis to measure the concentration-versus-time profiles of levofloxacin in the interstitial space fluid of skeletal muscle in patients with sepsis. Subsequently, we applied an established dynamic in vivo pharmacokinetic-in vitro pharmacodynamic approach to simulate bacterial killing at the site of infection. The population mean areas under the concentration-time curves (AUCs) for levofloxacin showed that levofloxacin excellently penetrates soft tissues, as indicated by the ratio of the AUC from time zero to 8 h (AUC0-8) for muscle tissue (AUC0-8 muscle) to the AUC0-8 for free drug in plasma (AUC0-8 plasma free) (AUC0-8 muscle/AUC0-8 plasma free ratio) of 0.85. The individual values of tissue penetration and maximum concentration (Cmax) in muscle tissue were highly variable. No difference in bacterial killing of a select Staphylococcus aureus strain for which the MIC was 0.5 μg/ml was found between individuals after exposure to dynamically changing concentrations of levofloxacin in plasma and tissue in vitro. In contrast, the decrease in the bacterial counts of Pseudomonas aeruginosa (MIC = 2 μg/ml) varied extensively when the bacteria were exposed to levofloxacin at the concentrations determined from the individual concentration-versus-time profiles obtained in skeletal muscle. The extent of bacterial killing could be predicted by calculating individual Cmax/MIC and AUC0-8 muscle/AUC0-8 plasma free ratios (R = 0.96 and 0.93, respectively). We have therefore shown in the present study that individual differences in the tissue penetration of levofloxacin may markedly affect target site killing of bacteria for which MICs are close to 2 μg/ml.

Objectives:

To validate published prediction models for the presence of obstructive coronary artery disease (CAD) in patients with new onset stable typical or atypical angina pectoris and to assess the incremental value of the CT coronary calcium score (CTCS).

Methods:

We searched the literature for clinical prediction rules for the diagnosis of obstructive CAD, defined as ≥50% stenosis in at least one vessel on conventional coronary angiography. Significant variables were re-analysed in our dataset of 254 patients with logistic regression. CTCS was subsequently included in the models. The area under the receiver operating characteristic curve (AUC) was calculated to assess diagnostic performance.

Results:

Re-analysing the variables used by Diamond & Forrester yielded an AUC of 0.798, which increased to 0.890 by adding CTCS. For Pryor, Morise 1994, Morise 1997 and Shaw the AUC increased from 0.838 to 0.901, 0.831 to 0.899, 0.840 to 0.898 and 0.833 to 0.899. CTCS significantly improved model performance in each model.

Conclusions:

Validation demonstrated good diagnostic performance across all models. CTCS improves the prediction of the presence of obstructive CAD, independent of clinical predictors, and should be considered in its diagnostic work-up.

Aims

Genetic determinants of variability in response to β-blockers are poorly characterized. We defined changes in mRNA expression after a β-blocker to identify novel genes that could affect response and correlated these with inhibition of exercise-induced tachycardia, a measure of β-blocker sensitivity.

Methods

Nine subjects exercised before and after a single oral dose of 25mg atenolol and mRNA gene expression was measured using an Affymetrix GeneChip Human Gene 1.0 ST Array. The area under the heart rate-exercise intensity curve (AUC) was calculated for each subject; the difference between post- and pre-atenolol AUCs (Δ AUC), a measure of β-blocker response, was correlated with the fold-change in mRNA expression of the genes that changed more than 1.3-fold.

Results

Fifty genes showed more than 1.3-fold increase in expression; 9 of these reached statistical significance (P < 0.05). Thirty-six genes had more than 1.3-fold decrease in expression after atenolol; 6 of these reached statistical significance (P < 0.05). Change in mRNA expression of FGFBP2 and Probeset ID 8118979 was significantly correlated with atenolol response (P = 0.03 and 0.02, respectively).

Conclusion

The expression of several genes not previously identified as part of the adrenergic signaling pathway changed in response to a single oral dose of atenolol. Variation in these genes could contribute to unexplained differences in response to β-blockers.

Genome-wide association studies in human populations have facilitated the creation of genomic profiles which combine the effects of many associated genetic variants to predict risk of disease. The area under the receiver operator characteristic (ROC) curve is a well established measure for determining the efficacy of tests in correctly classifying diseased and non-diseased individuals. We use quantitative genetics theory to provide insight into the genetic interpretation of the area under the ROC curve (AUC) when the test classifier is a predictor of genetic risk. Even when the proportion of genetic variance explained by the test is 100%, there is a maximum value for AUC that depends on the genetic epidemiology of the disease, i.e. either the sibling recurrence risk or heritability and disease prevalence. We derive an equation relating maximum AUC to heritability and disease prevalence. The expression can be reversed to calculate the proportion of genetic variance explained given AUC, disease prevalence, and heritability. We use published estimates of disease prevalence and sibling recurrence risk for 17 complex genetic diseases to calculate the proportion of genetic variance that a test must explain to achieve AUC = 0.75; this varied from 0.10 to 0.74. We provide a genetic interpretation of AUC for use with predictors of genetic risk based on genomic profiles. We provide a strategy to estimate proportion of genetic variance explained on the liability scale from estimates of AUC, disease prevalence, and heritability (or sibling recurrence risk) available as an online calculator.

Author Summary

Genome-wide association studies in human populations have facilitated the creation of genomic profiles that combine the effects of many associated genetic variants to predict risk of disease. However, genomic profiles are inherently constrained in their ability to classify diseased from non-diseased individuals dictated by the genetic epidemiology of the disease. In this paper, we use a genetic interpretation to provide insight into the constraints on genomic profiles for risk prediction. We provide a strategy to estimate proportion of genetic variance explained on the liability scale from estimates of AUC, disease prevalence, and heritability available as an online calculator.

OBJECTIVE

To determine if the dipeptidyl peptidase-4 inhibitor vildagliptin more effectively inhibits glucagon levels than the sulfonylurea glimepiride during a meal.

RESEARCH DESIGN AND METHODS

Glucagon responses to a standard meal were measured at baseline and study end point (mean 1.8 years) in a trial evaluating add-on therapy to metformin with 50 mg vildagliptin b.i.d. compared with glimepiride up to 6 mg q.d. in type 2 diabetes (baseline A1C 7.3 ± 0.6%).

RESULTS

A1C and prandial glucose area under the curve (AUC)0–2 h were reduced similarly in both groups, whereas prandial insulin AUC0–2 h increased to a greater extent by glimepiride. Prandial glucagon AUC0–2 h (baseline 66.6 ± 2.3 pmol · h−1 · l−1) decreased by 3.4 ± 1.6 pmol · h−1 · l−1 by vildagliptin (n = 137) and increased by 3.8 ± 1.7 pmol · h−1 · l−1 by glimepiride (n = 121). The between-group difference was 7.3 ± 2.1 pmol · h−1 · l−1 (P < 0.001).

CONCLUSIONS

Vildagliptin therapy but not glimepiride improves postprandial α-cell function, which persists for at least 2 years.

Renal transplant patients prescribed mycophenolate mofetil (MMF) may require treatment for tuberculosis with a regimen including the tuberculocidal drug rifampicin. MMF is an ester prodrug which is rapidly hydrolysed to the active compound, mycophenolic acid (MPA). Therapeutic drug monitoring of mycophenolate involves the measurement of MPA area under the curve (MPA-AUC0-12). Rifampicin is known to increase the metabolism and decrease enterohepatic recirculation of mycophenolic acid, (MPA). When MPA is monitored after the discontinuation of rifampicin, an important factor is the time required for the MPA area under the curve to return to the pre-rifampicin value. At present this is not known. This report describes one such renal allograft patient, on long term MMF and prescribed rifampicin by a local physician. As expected there was a clinically significant decrease in MPA-AUC0-12 Three weeks after rifampicin was discontinued the MPA-AUC0-12 was still only 65% of the pre-rifampicin value and only 55% of the steady state MPA-AUC0-12 measured six months later.

Purpose

To assess efficacy of the Pentacam (PTC) and the anterior segment optical coherence tomography (AOCT) for detection of occludable angles.

Materials and Methods

Fourty-one eyes with gonioscopically diagnosed occludable angles and 32 normal open-angle eyes were included. Anterior chamber angle (ACA) and anterior chamber depth (ACD) were measured with PTC and AOCT. Receiver operating characteristic (ROC) curve was constructed for each parameter and the area under the ROC curve (AUC) was calculated.

Results

Values of ACA and ACD measured by PTC and AOCT were similar not only in normal open angle eyes but also in occludable angle eyes. For detection of occludable angle, the AUCs of PTC with ACA and ACD were 0.935 and 0.969, respectively. The AUCs of AOCT with ACA and ACD were 0.904 and 0.947, respectively.

Conclusion

Both PTC and AOCT allow accurate discrimination between open and occludable angle eyes, so that they may aid to screening the occludable angles.

Background

The receiver operating characteristic (ROC) curve is a fundamental tool to assess the discriminant performance for not only a single marker but also a score function combining multiple markers. The area under the ROC curve (AUC) for a score function measures the intrinsic ability for the score function to discriminate between the controls and cases. Recently, the partial AUC (pAUC) has been paid more attention than the AUC, because a suitable range of the false positive rate can be focused according to various clinical situations. However, existing pAUC-based methods only handle a few markers and do not take nonlinear combination of markers into consideration.

Results

We have developed a new statistical method that focuses on the pAUC based on a boosting technique. The markers are combined componentially for maximizing the pAUC in the boosting algorithm using natural cubic splines or decision stumps (single-level decision trees), according to the values of markers (continuous or discrete). We show that the resulting score plots are useful for understanding how each marker is associated with the outcome variable. We compare the performance of the proposed boosting method with those of other existing methods, and demonstrate the utility using real data sets. As a result, we have much better discrimination performances in the sense of the pAUC in both simulation studies and real data analysis.

Conclusions

The proposed method addresses how to combine the markers after a pAUC-based filtering procedure in high dimensional setting. Hence, it provides a consistent way of analyzing data based on the pAUC from maker selection to marker combination for discrimination problems. The method can capture not only linear but also nonlinear association between the outcome variable and the markers, about which the nonlinearity is known to be necessary in general for the maximization of the pAUC. The method also puts importance on the accuracy of classification performance as well as interpretability of the association, by offering simple and smooth resultant score plots for each marker.

The inflammatory response, and its subsequent resolution, are the result of a very complex cascade of events originating at the site of injury or infection. When the response is severe and persistent, Systemic Inflammatory Response Syndrome can set in, which is associated with a severely debilitating systemic hypercatabolic state. This complex behavior, mediated by cytokines and chemokines, needs to be further explored to better understand its systems properties and potentially identify multiple targets that could be addressed simultaneously. In this context, short term responses of serum cytokines and chemokines were analyzed in two types of insults: rats receiving a “sterile” cutaneous dorsal burn on 20% of the total body surface area (TBSA); rats receiving a cecum ligation and puncture treatment (CLP) to induce infection. Considering the temporal variability observed in the baseline corresponding to the control group, the concept of area under the curve (AUC) was explored to assess the dynamic responses of cytokines and chemokines. MCP-1, GROK/KC, IL-12, IL-18 and IL-10 were observed in both burn and CLP groups. While IL-10 concentration was only increased in the burn group, Eotaxin was only elevated in CLP group. It was also observed that Leptin and IP-1 concentrations were decreased in both CLP and sham-CLP groups. The link between the circulating protein mediators and putative transcription factors regulating the cytokine/chemokine gene expression was explored by searching the promoter regions of cytokine/chemokine genes in order to characterize and differentiate the inflammatory responses based on the dynamic data. Integrating multiple sources together with the bioinformatics tools identified mediators sensitive to type and extent of injury, and provided putative regulatory mechanisms. This is essential to gain a better understanding for the important regulatory points that can be used to modulate the inflammatory state at molecular level.

Introduction

Previous studies have shown that traditional risk factors such as hypercholesterolemia and hypertension account for only a small proportion of the dramatically increased risk of atherosclerotic coronary artery disease (CAD) in systemic lupus erythematosus (SLE). However, in these studies, exposure to risk factors was measured only at baseline. In this study, our objective was to compare measures of cumulative exposure with remote and recent values for each of total cholesterol (TC), systolic (SBP), and diastolic (DBP) blood pressure in terms of ability to quantify risk of atherosclerotic CAD in patients with SLE.

Methods

Patients in the Toronto lupus cohort had TC and BP measured at each clinic visit and were followed up prospectively for the occurrence of CAD. For each patient, arithmetic mean, time-adjusted mean (AM) and area-under-the-curve (AUC) were calculated for serial TC, SBP, and DBP measurements. Proportional hazards regression models were used to compare these summary measures with recent and first-available ("remote") measurements in terms of ability to quantify risk of CAD events, defined as myocardial infarction, angina, or sudden cardiac death.

Results

The 991 patients had a mean ± SD of 19 ± 19 TC measurements per patient. Over a follow-up of 6.7 ± 6.4 years, 86 CAD events occurred; although remote TC was not significantly predictive of CAD, mean and AM TC were more strongly predictive (hazard ratio (HR) 2.07; P = 0.003) than recent TC (HR 1.86, P = 0.001). AUC TC was not predictive of CAD. A similar pattern was seen for DBP and SBP. Older age, male sex, higher baseline and recent disease activity score, and corticosteroid use also increased CAD risk, whereas antimalarials were protective.

Conclusions

In contrast to the population-based Framingham model, first-available TC and BP are not predictive of CAD among patients with SLE, in whom measures reflecting cumulative exposure over time are better able to quantify CAD risk. This is an important consideration in future studies of dynamic risk factors for CAD in a chronic relapsing-remitting disease such as SLE. Our findings also underpin the importance of adequate control of SLE disease activity while minimizing corticosteroid use, and highlight the cardioprotective effect of antimalarials.

BACKGROUND

Genomewide association studies have identified multiple genetic variants associated with breast cancer. The extent to which these variants add to existing risk-assessment models is unknown.

METHODS

We used information on traditional risk factors and 10 common genetic variants associated with breast cancer in 5590 case subjects and 5998 control subjects, 50 to 79 years of age, from four U.S. cohort studies and one case–control study from Poland to fit models of the absolute risk of breast cancer. With the use of receiveroperating- characteristic curve analysis, we calculated the area under the curve (AUC) as a measure of discrimination. By definition, random classification of case and control subjects provides an AUC of 50%; perfect classification provides an AUC of 100%. We calculated the fraction of case subjects in quintiles of estimated absolute risk after the addition of genetic variants to the traditional risk model.

RESULTS

The AUC for a risk model with age, study and entry year, and four traditional risk factors was 58.0%; with the addition of 10 genetic variants, the AUC was 61.8%. About half the case subjects (47.2%) were in the same quintile of risk as in a model without genetic variants; 32.5% were in a higher quintile, and 20.4% were in a lower quintile.

CONCLUSIONS

The inclusion of newly discovered genetic factors modestly improved the performance of risk models for breast cancer. The level of predicted breast-cancer risk among most women changed little after the addition of currently available genetic information.

During the last decade some studies have shown that the area under the curve (AUC)/MIC ratio is the pharmacodynamic index that best predicts the efficacies of quinolones, while other studies suggest that the predictive value of the peak concentration/MIC (peak/MIC) ratio is superior to the AUC/MIC ratio in explaining clinical and microbiological outcomes. In classical fractionated dose-response studies with animals, it is difficult to differentiate between the AUC/MIC ratio and the peak/MIC ratio because of colinearity. Three different levofloxacin and ciprofloxacin dosing regimens were studied in a neutropenic mouse pneumonia model. The different regimens were used with the aim of increasing the AUC/MIC ratio without changing the peak/MIC ratio and vice versa. The first regimen (RC) consisted of daily doses of 5 up to 160 mg/kg of body weight divided into one, two, or four doses. In the second regimen (R0), mice were given 1.25 mg/kg every hour from 1 to 23 h, while the dose given at 0 h was 2.5, 5, 10, 20, 40, or 80 mg/kg. In the third regimen (R11), mice also received 1.25 mg/kg every hour from 0 to 23 h; but in addition, they also received 2.5, 5, 10, 20, 40, or 80 mg/kg at 11 h. The level of protein binding was also evaluated. The results indicate that the unbound fraction (fu) was concentration dependent for both levofloxacin and ciprofloxacin and ranged from approximately 0.67 to 0.88 for both drugs between concentrations of 0.5 and 80 mg/liter. The relationships between the AUC/MIC ratio and the number of CFU were slightly better than those between the peak/MIC ratio and the number of CFU. There was no clear relationship between the amount of time that the concentration remained above the MIC and effect (R2 < 0.1). For both drugs, the peak/MIC ratio that resulted in a 50% effective concentration was lower for the R0 and R11 dosing regimens, indicating the importance of the AUC/MIC ratio. The same was true for the static doses. Survival studies showed that for mice treated with the low doses the rate of survival was comparable to that for the controls, but with the higher doses the rate of survival was better for mice receiving the R0 regimen. We conclude that for quinolones the AUC/MIC ratio best correlates with efficacy against pneumococci and that the effect of the peak/MIC ratio found in some studies could be partly explained by concentration-dependent protein binding.

Summary

Receiver operating characteristic (ROC) curves can be used to assess the accuracy of tests measured on ordinal or continuous scales. The most commonly used measure for the overall diagnostic accuracy of diagnostic tests is the area under the ROC curve (AUC). A gold standard test on the true disease status is required to estimate the AUC. However, a gold standard test may sometimes be too expensive or infeasible. Therefore, in many medical research studies, the true disease status of the subjects may remain unknown. Under the normality assumption on test results from each disease group of subjects, using the expectation-maximization (EM) algorithm in conjunction with a bootstrap method, we propose a maximum likelihood based procedure for construction of confidence intervals for the difference in paired areas under ROC curves in the absence of a gold standard test. Simulation results show that the proposed interval estimation procedure yields satisfactory coverage probabilities and interval lengths. The proposed method is illustrated with two examples.

Aims

To assess performance of classifiers trained on Heidelberg Retina Tomograph 3 (HRT3) parameters for discriminating between healthy and glaucomatous eyes.

Methods

Classifiers were trained using HRT3 parameters from 60 healthy subjects and 140 glaucomatous subjects. The classifiers were trained on all 95 variables and smaller sets created with backward elimination. Seven types of classifiers, including Support Vector Machines with radial basis (SVM-radial), and Recursive Partitioning and Regression Trees (RPART), were trained on the parameters. The area under the ROC curve (AUC) was calculated for classifiers, individual parameters and HRT3 glaucoma probability scores (GPS). Classifier AUCs and leave-one-out accuracy were compared with the highest individual parameter and GPS AUCs and accuracies.

Results

The highest AUC and accuracy for an individual parameter were 0.848 and 0.79, for vertical cup/disc ratio (vC/D). For GPS, global GPS performed best with AUC 0.829 and accuracy 0.78. SVM-radial with all parameters showed significant improvement over global GPS and vC/ D with AUC 0.916 and accuracy 0.85. RPART with all parameters provided significant improvement over global GPS with AUC 0.899 and significant improvement over global GPS and vC/D with accuracy 0.875.

Conclusions

Machine learning classifiers of HRT3 data provide significant enhancement over current methods for detection of glaucoma.