Search

Studies of critically ill, hospitalized patients often follow participants and characterize daily health status using an ordinal outcome variable. Statistically, longitudinal proportional odds models are a natural choice in these settings since such models can parsimoniously summarize differences across patient groups and over time. However, when one or more of the outcome states is absorbing, the proportional odds assumption for the follow-up time parameter will likely be violated, and more flexible longitudinal models are needed. Motivated by the VIOLET Study(1), a parallel-arm, randomized clinical trial of Vitamin D(3) in critically ill patients, we discuss and contrast several treatment effect estimands based on time-dependent odds ratio parameters, and we detail contemporary modeling approaches. In VIOLET, the outcome is a four-level ordinal variable where the lowest ‘not alive’ state is absorbing and the highest ‘at-home’ state is nearly absorbing. We discuss flexible extensions of the proportional odds model for longitudinal data that can be used for either model-based inference, where the odds ratio estimator is taken directly from the model fit, or for model-assisted inferences, where heterogeneity across cumulative log odds dichotomizations is modeled and results are summarized to obtain an overall odds ratio estimator. We focus on direct estimation of cumulative probability model parameters using likelihood-based analysis procedures that naturally handle absorbing states. We illustrate the modeling procedures, the relative precision of model-based and model-assisted estimators, and the possible differences in the values for which the estimators are consistent through simulations and analysis of the VIOLET Study data.

We propose a general class of 2-phase epidemiologic study designs for quantitative, longitudinal data that are useful when phase 1 longitudinal outcome and covariate data are available but data on the exposure (e.g., a biomarker) can only be collected on a subset of subjects during phase 2. To conduct a study using a design in the class, one first summarizes the longitudinal outcomes by fitting a simple linear regression of the response on a time-varying covariate for each subject. Sampling strata are defined by splitting the estimated regression intercept or slope distributions into distinct (low, medium, and high) regions. Stratified sampling is then conducted from strata defined by the intercepts, by the slopes, or from a mixture. In general, samples selected with extreme intercept values will yield low variances for associations of time-fixed exposures with the outcome and samples enriched with extreme slope values will yield low variances for associations of time-varying exposures with the outcome (including interactions with time-varying exposures). We describe ascertainment-corrected maximum likelihood and multiple-imputation estimation procedures that permit valid and efficient inferences. We embed all methodological developments within the framework of conducting a substudy that seeks to examine genetic associations with lung function among continuous smokers in the Lung Health Study (United States and Canada, 1986–1994).

We discuss tradeoffs and errors associated with approaches to modeling health economic decisions. Through an application in pharmacogenomic (PGx) testing to guide drug selection for individuals with a genetic variant, we assessed model accuracy, optimal decisions and computation time for an identical decision scenario modeled four ways: using (1) coupled-time differential equations [DEQ]; (2) a cohort-based discrete-time state transition model [MARKOV]; (3) an individual discrete-time state transition microsimulation model [MICROSIM]; and (4) discrete event simulation [DES]. Relative to DEQ, the Net Monetary Benefit for PGx testing (vs. a reference strategy of no testing) based on MARKOV with rate-to-probability conversions using commonly used formulas resulted in different optimal decisions. MARKOV was nearly identical to DEQ when transition probabilities were embedded using a transition intensity matrix. Among stochastic models, DES model outputs converged to DEQ with substantially fewer simulated patients (1 million) vs. MICROSIM (1 billion). Overall, properly embedded Markov models provided the most favorable mix of accuracy and run-time, but introduced additional complexity for calculating cost and quality-adjusted life year outcomes due to the inclusion of “jumpover” states after proper embedding of transition probabilities. Among stochastic models, DES offered the most favorable mix of accuracy, reliability, and speed.

Mathematical models of biomolecular networks are commonly used to study cellular processes; however, their usefulness to explain and predict dynamic behaviors is often questioned due to parameter uncertainty. Here, we employ a novel dynamic network analysis within a Bayesian statistic context to probabilistically identify dominant reaction paths in signaling networks. We introduce our approach in an extrinsic apoptosis model using fluorescence microscopy data from HeLa cells for parameter inference. While Bayesian optimization yields several parameter vectors with equally good-fit to experimental data, three execution modes emerge upon analysis of the reaction flux throughout the network using a discretization theory approach. Each mode exhibits different behaviors to in silico perturbations through different mechanisms, thus underscoring the need for thorough characterization of model parameters and associated dynamics. We then apply our approach on the JAK2/STAT5 pathway in colony-forming unit-erythroid (CFU-E) cells. Our analysis provides mechanistic explanation for survival responses of CFU-E cell population that was impossible by considering protein concentration dynamics in isolation.

We discuss a decision-theoretic approach to building a panel-based, preemptive genotyping program. The method is based on findings that a large percentage of patients are prescribed medications that are known to have pharmacogenetic associations, and over time, a substantial proportion are prescribed additional such medication. Preemptive genotyping facilitates genotype-guided therapy at the time medications are prescribed; panel-based testing allows providers to reuse previously collected genetic data when a new indication arises. Because it is cost-prohibitive to conduct panel-based genotyping on all patients, we describe a three-step approach to identify patients with the highest anticipated benefit. First, we construct prediction models to estimate the risk of being prescribed one of the target medications using readily available clinical data. Second, we use literature-based estimates of adverse event rates, variant allele frequencies, secular death rates, and costs to construct a discrete event simulation that estimates the expected benefit of having an individual’s genetic data in the electronic health record after an indication has occurred. Finally, we combine medication prescription risk with expected benefit of genotyping once a medication is indicated to calculate the expected benefit of preemptive genotyping. For each patient-clinic visit, we calculate this expected benefit across a range of medications and select patients with the highest expected benefit overall. We build a proof of concept implementation using a cohort of patients from a single academic medical center observed from July 2010 through December 2012. We then apply the results of our modeling strategy to show the extent to which we can improve clinical and economic outcomes in a cohort observed from January 2013 through December 2015.

In vitro cell proliferation assays are widely used in pharmacology, molecular biology, and drug discovery. Using theoretical modeling and experimentation, we show that current antiproliferative drug effect metrics suffer from time-dependent bias, leading to inaccurate assessments of parameters such as drug potency and efficacy. We propose the drug-induced proliferation (DIP) rate, the slope of the line on a plot of cell population doublings versus time, as an alternative, time-independent metric., Comment: 44 pages: main and supplementary text, 3 main text figures, 9 supplementary figures, 2 supplementary tables, Nature Methods, 2016 (advanced online publication)

Key Points Question Is it cost-effective to implement population-wide genomic screening for hereditary breast and ovarian cancer (HBOC)? Findings This decision analytical model study found that genomic screening for HBOC among unselected women may be cost-effective depending on the age distribution of the women screened. Cascade testing of first-degree relatives added a modest improvement in clinical and economic value. Meaning Population-level genomic screening for HBOC targeting women aged 20 to 35 years could be considered in settings in which the outcomes of screening can be evaluated, particularly to avoid a reduction in mammography screening among patients with negative test results., Importance Genomic screening for hereditary breast and ovarian cancer (HBOC) in unselected women offers an opportunity to prevent cancer morbidity and mortality, but the potential clinical impact and cost-effectiveness of such screening have not been well studied. Objective To estimate the lifetime incremental incidence of HBOC and the quality-adjusted life-years (QALYs), costs, and cost-effectiveness of HBOC genomic screening in an unselected population vs family history–based testing. Design, Setting, and Participants In this study conducted from October 27, 2017, to May 3, 2020, a decision analytic Markov model was developed that included health states for precancer, for risk-reducing mastectomy (RRM) and risk-reducing salpingo-oophorectomy (RRSO), for earlier- and later-stage HBOC, after cancer, and for death. A complimentary cascade testing module was also developed to estimate outcomes in first-degree relatives. Age-specific RRM and RRSO uptake probabilities were estimated from the Geisinger MyCode Community Health Initiative and published sources. Parameters including RRM and RRSO effectiveness, variant-specific cancer risk, costs, and utilities were derived from published sources. Sensitivity and scenario analyses were conducted to evaluate model assumptions and uncertainty. Main Outcomes and Measures Lifetime cancer incidence, QALYs, life-years, and direct medical costs for genomic screening in an unselected population vs family history–based testing only were calculated. The incremental cost-effectiveness ratio (ICER) was calculated as the difference in cost between strategies divided by the difference in QALYs between strategies. Earlier-stage and later-stage cancer cases prevented and total cancer cases prevented were also calculated. Results The model found that population screening of 30-year-old women was associated with 75 (95% credible range [CR], 60-90) fewer overall cancer cases and 288 QALYs (95% CR, 212-373 QALYs) gained per 100 000 women screened, at an incremental cost of $25 million (95% CR, $21 millon to $30 million) vs family history–based testing; the ICER was $87 700 (78% probability of being cost-effective at a threshold of $100 000 per QALY). In contrast, population screening of 45-year-old women was associated with 24 (95% CR, 18-29) fewer cancer cases and 97 QALYs (95% CR, 66-130 QALYs) gained per 100 000 women screened, at an incremental cost of $26 million (95% CR, $22 million to $30 million); the ICER was $268 200 (0% probability of being cost-effective at a threshold of $100 000 per QALY). A scenario analysis without cascade testing increased the ICER to $92 600 for 30-year-old women and $354 500 for 45-year-old women. A scenario analysis assuming a 5% absolute decrease in mammography screening in women without a variant was associated with the potential for net harm (−90 QALYs per 100 000 women screened; 95% CR, −180 to 10 QALYs). Conclusions and Relevance The results of this study suggest that population HBOC screening may be cost-effective among younger women but not among older women. Cascade testing of first-degree relatives added a modest improvement in clinical and economic value. The potential for harm conferred by inappropriate reduction in mammography among noncarriers should be quantified., This decision analysis estimates the lifetime incremental incidence of hereditary breast and ovarian cancer and the quality-adjusted life-years, costs, and cost-effectiveness of hereditary breast and ovarian cancer genomic screening in an unselected population vs family history–based testing.

Pharmacogenomics, or the application of genetic testing to guide drug selection and/or dosing, is often cited as integral to the vision of how precision medicine can be integrated into routine clinical practice. Yet despite a growing base of scientific discovery on genetic variation that predicts drug response, reimbursement for genetic testing among health systems and payers remains uneven. In large measure this is because the cascading impacts of genetic testing on individual and provider incentives and behavior, as well as downstream health care spending and outcomes, remain poorly understood. In this study, we couple evidence from a real-world implementation of pharmacogenomic testing with a discrete event simulation model. We use this framework to evaluate the cost-effectiveness of various genetic testing strategies. We find that the cost-effectiveness of multiplexed genetic testing (e.g., whole genome sequencing) hinges on the ability of a health system to ensure that dense genotypic information is routinely utilized by physicians. Moreover, while much attention has been paid to lowering the cost of genetic tests, we demonstrate that in practice, other scientific and behavioral factors, focused on certain high-yield drug-gene pairs, are key to implementing precision medicine in ways that maximize its value.

Experiments have shown that, even in a homogeneous population of cells, the distribution of division times is highly variable. In addition, a homogeneous population of cells will exhibit a heterogeneous response to drug therapy. We present a simple stochastic model of the cell cycle as a multistep stochastic process. The model, which is based on our conception of the cell cycle checkpoint, is used to derive an analytical expression for the distribution of cell cycle times. We demonstrate that this distribution provides an accurate representation of cell cycle time variability and show how the model relates drug-induced changes in basic biological parameters to variability in the response to drug treatment.

Polydimethylsiloxane (PDMS) is a commonly used polymer in the fabrication of microfluidic devices due to such features as transparency, gas permeability, and ease of patterning with soft lithography. The surface characteristics of PDMS can also be easily changed with oxygen or low pressure air plasma converting it from a hydrophobic to a hydrophilic state. As part of such a transformation, surface methyl groups are removed and replaced with hydroxyl groups making the exposed surface to resemble silica, a gas impermeable substance. We have utilized Platinum(II)-tetrakis(pentaflourophenyl)porphyrin immobilized within a thin (~1.5 um thick) polystyrene matrix as an oxygen sensor, Stern-Volmer relationship, and Fick's Law of simple diffusion to measure the effects of PDMS composition, treatment, and storage on oxygen diffusion through PDMS. Results indicate that freshly oxidized PDMS showed a significantly smaller diffusion coefficient, indicating that the SiO2 layer formed on the PDMS surface created an impeding barrier. This barrier disappeared after a 3-day storage in air, but remained significant for up to 3 weeks if PDMS was maintained in contact with water. Additionally, higher density PDMS formulation (5:1 ratio) showed similar diffusion characteristics as normal (10:1 ratio) formulation, but showed 60 % smaller diffusion coefficient after plasma treatment that never recovered to pre-treatment levels even after a 3-week storage in air. Understanding how plasma surface treatments contribute to oxygen diffusion will be useful in exploiting the gas permeability of PDMS to establish defined normoxic and hypoxic oxygen conditions within microfluidic bioreactor systems.

We present an integrated method that uses extended time-lapse automated imaging to quantify the dynamics of cell proliferation. Cell counts are fit with a quiescence-growth model that estimates rates of cell division, entry into quiescence and death. The model is constrained with rates extracted experimentally from the behavior of tracked single cells over time. We visualize the output of the analysis in fractional proliferation graphs, which deconvolve dynamic proliferative responses to perturbations into the relative contributions of dividing, quiescent (nondividing) and dead cells. The method reveals that the response of 'oncogene-addicted' human cancer cells to tyrosine kinase inhibitors is a composite of altered rates of division, death and entry into quiescence, a finding that challenges the notion that such cells simply die in response to oncogene-targeted therapy.

Stochastic profiling, a method to rank heterogeneity of gene expression in a cell population, shows that quantifying cell-to-cell variability has come of age and leads to biological insight.

Substudies of the Childhood Asthma Management Program [Control. Clin. Trials 20 (1999) 91-120; N. Engl. J. Med. 343 (2000) 1054-1063] seek to identify patient characteristics associated with asthma symptoms and lung function. To determine if genetic measures are associated with trajectories of lung function as measured by forced vital capacity (FVC), children in the primary cohort study retrospectively had candidate loci evaluated. Given participant burden and constraints on financial resources, it is often desirable to target a subsample for ascertainment of costly measures. Methods that can leverage the longitudinal outcome on the full cohort to selectively measure informative individuals have been promising, but have been restricted in their use to analysis of the targeted subsample. In this paper we detail two multiple imputation analysis strategies that exploit outcome and partially observed covariate data on the nonsampled subjects, and we characterize alternative design and analysis combinations that could be used for future studies of pulmonary function and other outcomes. Candidate predictor (e.g., IL10 cytokine polymorphisms) associations obtained from targeted sampling designs can be estimated with very high efficiency compared to standard designs. Further, even though multiple imputation can dramatically improve estimation efficiency for covariates available on all subjects (e.g., gender and baseline age), relatively modest efficiency gains were observed in parameters associated with predictors that are exclusive to the targeted sample. Our results suggest that future studies of longitudinal trajectories can be efficiently conducted by use of outcome-dependent designs and associated full cohort analysis., Published at http://dx.doi.org/10.1214/15-AOAS826 in the Annals of Applied Statistics (http://www.imstat.org/aoas/) by the Institute of Mathematical Statistics (http://www.imstat.org)

International audience; Cells grown in culture act as a model system for analyzing the effects of anticancer compounds, which may affect cell behavior in a cell cycle position-dependent manner. Cell synchronization techniques have been generally employed to minimize the variation in cell cycle position. However, synchronization techniques are cumbersome and imprecise and the agents used to synchronize the cells potentially have other unknown effects on the cells. An alternative approach is to determine the age structure in the population and account for the cell cycle positional effects post hoc. Here we provide a formalism to use quantifiable age distributions from live cell microscopy experiments to parameterize an age-structured model of cell population response.

Metabolism, and especially glucose uptake, is a key quantitative cell trait that is closely linked to cancer initiation and progression. Therefore, developing high-throughput assays for measuring glucose uptake in cancer cells would be enviable for simultaneous comparisons of multiple cell lines and microenvironmental conditions. This study was designed with two specific aims in mind: the first was to develop and validate a high-throughput screening method for quantitative assessment of glucose uptake in "normal" and tumor cells using the fluorescent 2-deoxyglucose analog 2-[N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino]-2-deoxyglucose (2-NBDG), and the second was to develop an image-based, quantitative, single-cell assay for measuring glucose uptake using the same probe to dissect the full spectrum of metabolic variability within populations of tumor cells in vitro in higher resolution.The kinetics of population-based glucose uptake was evaluated for MCF10A mammary epithelial and CA1d breast cancer cell lines, using 2-NBDG and a fluorometric microplate reader. Glucose uptake for the same cell lines was also examined at the single-cell level using high-content automated microscopy coupled with semi-automated cell-cytometric image analysis approaches. Statistical treatments were also implemented to analyze intra-population variability.Our results demonstrate that the high-throughput fluorometric assay using 2-NBDG is a reliable method to assess population-level kinetics of glucose uptake in cell lines in vitro. Similarly, single-cell image-based assays and analyses of 2-NBDG fluorescence proved an effective and accurate means for assessing glucose uptake, which revealed that breast tumor cell lines display intra-population variability that is modulated by growth conditions.These studies indicate that 2-NBDG can be used to aid in the high-throughput analysis of the influence of chemotherapeutics on glucose uptake in cancer cells.

Mapping quantitative cell traits (QCT) to underlying molecular defects is a central challenge in cancer research because heterogeneity at all biological scales, from genes to cells to populations, is recognized as the main driver of cancer progression and treatment resistance. A major roadblock to a multiscale framework linking cell to signaling to genetic cancer heterogeneity is the dearth of large-scale, single-cell data on QCT-such as proliferation, death sensitivity, motility, metabolism, and other hallmarks of cancer. High-volume single-cell data can be used to represent cell-to-cell genetic and nongenetic QCT variability in cancer cell populations as averages, distributions, and statistical subpopulations. By matching the abundance of available data on cancer genetic and molecular variability, QCT data should enable quantitative mapping of phenotype to genotype in cancer. This challenge is being met by high-content automated microscopy (HCAM), based on the convergence of several technologies including computerized microscopy, image processing, computation, and heterogeneity science. In this chapter, we describe an HCAM workflow that can be set up in a medium size interdisciplinary laboratory, and its application to produce high-throughput QCT data for cancer cell motility and proliferation. This type of data is ideally suited to populate cell-scale computational and mathematical models of cancer progression for quantitatively and predictively evaluating cancer drug discovery and treatment.

Patients with activating epidermal growth factor receptor (EGFR) mutations display drastic initial responses to erlotinib, but tumor reduction and progression-free survival vary widely. An open question is whether response variation derives, in part, from tumor heterogeneity, i.e., differential composition of cells responding to erlotinib with distinct cell fates (death, division, or quiescence). Conventional assays obscure response heterogeneity by taking average measurements of cell populations. To address this question we developed high-throughput live-cell imaging assays that quantify both subpopulation growth dynamics and cell-to-cell fate variability in response to drug. The Dynamic Colony Growth Assay (DCGA) tracks simultaneously the erlotinib response of hundreds of single-cell-derived colonies from within a cell population. In the PC9 lung cancer cell line (widely used to model oncogene addiction, exon19del EGFR) the DCGA reveals that individual colonies span a wide range of erlotinib response, from continued cell division to massive apoptosis. Plotting net growth rates of 207 colonies shows that PC9 parental is comprised of a normally-distributed aggregate of individual single colonies responding to erlotinib with steady-state growth rates from positive to negative. Notably, the positive-growth colonies lie in the tail of this response distribution, suggesting they do not originate from rare genetic variants or cancer stem cells but, rather, are part of a continuous response distribution that is an attribute of parental PC9. To determine whether isolated PC9 colonies maintain unique rate responses, we expanded random-sampled single cells (without erlotinib selection) into 7 discrete sublines (DS), which remain highly sensitive to erlotinib (IC50 ∼50nM). In each DS, erlotinib induced an initial non-linear growth period (72h) followed by a distinct steady-state growth rate that predicts the long-term (10d) response. DS steady-state growth rates were resolved into individual cell fate composition by Fractional Proliferation Assays (FPA). As expected from random selection, the growth rates of treated DS clustered around the median of the DCGA rate distribution, missing the most extreme, less frequent erlotinib responses. Isolating the greatest variance in erlotinib response should facilitate identification of molecular mechanisms underlying drug response variation. Therefore, we isolated 96 new DS and selected the two with highest (DS-B03) and lowest (DS-C03) erlotinib steady state growth rate. We are currently measuring the activity of hundreds of proteins in these two DS by Microwestern Arrays to discover protein signatures and/or mechanistic models that quantitatively link FPA-resolved cell fates to underlying signaling network events. Thus tumor response variability and recurrence may be explained by a continuous distribution of erlotinib response at the level of single cells within a tumor. Citation Format: Peter L. Frick, Darren R. Tyson, Shawn P. Garbett, Carlos F. Lopez, Zach W. Jones, Vito Quaranta. Quantifying erlotinib response variability in EGFR-addicted cells. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 4626. doi:10.1158/1538-7445.AM2013-4626