Modeling basal body temperature data using horseshoe process regression.

Biomedical data often exhibit jumps or abrupt changes. For example, women's basal body temperature may jump at ovulation, menstruation, implantation, and miscarriage. These sudden changes make these data challenging to model: many methods will oversmooth the sharp changes or overfit in response to measurement error. We develop horseshoe process regression (HPR) to address this problem. We define a horseshoe process as a stochastic process in which each increment is horseshoe-distributed. We use the horseshoe process as a nonparametric Bayesian prior for modeling a potentially nonlinear association between an outcome and its continuous predictor, which we implement via Stan and in the R package HPR. We provide guidance and extensions to advance HPR's use in applied practice: we introduce a Bayesian imputation scheme to allow for interpolation at unobserved values of the predictor within the HPR; include additional covariates via a partial linear model framework; and allow for monotonicity constraints. We find that HPR performs well when fitting functions that have sharp changes. We apply HPR to model women's basal body temperatures over the course of the menstrual cycle.

Optimizing the doses of cancer drugs after usual dose finding.

Since the middle of the 20th century, oncology's dose-finding paradigm has been oriented toward identifying a drug's maximum tolerated dose, which is then carried forward into phase 2 and 3 trials and clinical practice. For most modern precision medicines, however, maximum tolerated dose is far greater than the minimum dose needed to achieve maximal benefit, leading to unnecessary side effects. Regulatory change may decrease maximum tolerated dose's predominance by enforcing dose optimization of new drugs. Dozens of already approved cancer drugs require re-evaluation, however, introducing a new methodologic and ethical challenge in cancer clinical trials. In this article, we assess the history and current landscape of cancer drug dose finding. We provide a set of strategic priorities for postapproval dose optimization trials of the future. We discuss ethical considerations for postapproval dose optimization trial design and review three major design strategies for these unique trials that would both adhere to ethical standards and benefit patients and funders. We close with a discussion of financial and reporting considerations in the realm of dose optimization. Taken together, we provide a comprehensive, bird's eye view of the postapproval dose optimization trial landscape and offer our thoughts on the next steps required of methodologies and regulatory and funding regimes.

Shrinkage priors for isotonic probability vectors and binary data modeling, with applications to dose-response modeling.

Motivated by the need to model dose-response or dose-toxicity curves in clinical trials, we develop a new horseshoe-based prior for Bayesian isotonic regression modeling a binary outcome against an ordered categorical predictor, where the probability of the outcome is assumed to be monotonically non-decreasing with the predictor. The set of differences between outcome probabilities in consecutive categories of the predictor is equipped with a multivariate prior having support over simplex. The Dirichlet distribution, which can be derived from a normalized sum of independent gamma-distributed random variables, is a natural choice of prior, but using mathematical and simulation-based arguments, we show that the resulting posterior is prone to underflow and other numerical instabilities, even under simple data configurations. We propose an alternative prior based on horseshoe-type shrinkage that is numerically more stable. We show that this horseshoe-based prior is not subject to the numerical instability seen in the Dirichlet/gamma-based prior and that the horseshoe-based posterior can estimate the underlying true curve more efficiently than the Dirichlet-based one. We demonstrate the use of this prior in a model predicting the occurrence of radiation-induced lung toxicity in lung cancer patients as a function of dose delivered to normal lung tissue. Our methodology is implemented in the R package isotonicBayes and therefore suitable for use in the design of dose-finding studies or other dose-response modeling contexts.

Extracorporeal membrane oxygenation for COVID-19: evolving outcomes from the international Extracorporeal Life Support Organization Registry.

BACKGROUND: Over the course of the COVID-19 pandemic, the care of patients with COVID-19 has changed and the use of extracorporeal membrane oxygenation (ECMO) has increased. We aimed to examine patient selection, treatments, outcomes, and ECMO centre characteristics over the course of the pandemic to date. METHODS: We retrospectively analysed the Extracorporeal Life Support Organization Registry and COVID-19 Addendum to compare three groups of ECMO-supported patients with COVID-19 (aged ≥16 years). At early-adopting centres-ie, those using ECMO support for COVID-19 throughout 2020-we compared patients who started ECMO on or before May 1, 2020 (group A1), and between May 2 and Dec 31, 2020 (group A2). Late-adopting centres were those that provided ECMO for COVID-19 only after May 1, 2020 (group B). The primary outcome was in-hospital mortality in a time-to-event analysis assessed 90 days after ECMO initiation. A Cox proportional hazards model was fit to compare the patient and centre-level adjusted relative risk of mortality among the groups. FINDINGS: In 2020, 4812 patients with COVID-19 received ECMO across 349 centres within 41 countries. For early-adopting centres, the cumulative incidence of in-hospital mortality 90 days after ECMO initiation was 36·9% (95% CI 34·1-39·7) in patients who started ECMO on or before May 1 (group A1) versus 51·9% (50·0-53·8) after May 1 (group A2); at late-adopting centres (group B), it was 58·9% (55·4-62·3). Relative to patients in group A2, group A1 patients had a lower adjusted relative risk of in-hospital mortality 90 days after ECMO (hazard ratio 0·82 [0·70-0·96]), whereas group B patients had a higher adjusted relative risk (1·42 [1·17-1·73]). INTERPRETATION: Mortality after ECMO for patients with COVID-19 worsened during 2020. These findings inform the role of ECMO in COVID-19 for patients, clinicians, and policy makers. FUNDING: None.

Tracheostomy Practices and Outcomes in Patients With COVID-19 Supported by Extracorporeal Membrane Oxygenation: An Analysis of the Extracorporeal Life Support Organization Registry.

OBJECTIVES: The use of extracorporeal membrane oxygenation (ECMO) in patients with COVID-19 has been supported by major healthcare organizations, yet the role of specific management strategies during ECMO requires further study. We sought to characterize tracheostomy practices, complications, and outcomes in ECMO-supported patients with acute respiratory failure related to COVID-19. DESIGN: Retrospective cohort study. SETTING: ECMO centers contributing to the Extracorporeal Life Support Organization Registry. PATIENTS: Patients 16 years or older receiving venovenous ECMO for respiratory support for: 1) COVID-19 in 2020 and 2021 (through October 2021) and 2) pre-COVID-19 viral pneumonia in 2019. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: We identified 7,047 patients who received ECMO support for acute respiratory failure related to COVID-19. A total of 32% of patients were recorded as having a tracheostomy procedure during ECMO, and 51% had a tracheostomy at some point during hospitalization. The frequency of tracheostomy was similar in pre-COVID-19 viral pneumonia, but tracheostomies were performed 3 days earlier compared with patients with COVID-19 (median 6.7 d [interquartile range [IQR], 3.0-12.0 d] vs 10.0 d [IQR, 5.0-16.5 d]; p < 0.001). More patients were mobilized with pre-COVID-19 viral pneumonia, but receipt of a tracheostomy during ECMO was associated with increased mobilization in both cohorts. More bleeding complications occurred in patients who received a tracheostomy, with 9% of patients with COVID-19 who received a tracheostomy reported as having surgical site bleeding. CONCLUSIONS: Tracheostomies are performed in COVID-19 patients receiving ECMO at rates similar to practices in pre-COVID-19 viral pneumonia, although later during the course of ECMO. Receipt of a tracheostomy was associated with increased patient mobilization. Overall mortality was similar between those who did and did not receive a tracheostomy.

Utility based approach in individualized optimal dose selection using machine learning methods.

The goal in personalized medicine is to individualize treatment using patient characteristics and improve health outcomes. Selection of optimal dose must balance the effect of dose on both treatment efficacy and toxicity outcomes. We consider a setting with one binary efficacy and one binary toxicity outcome. The goal is to find the optimal dose for each patient using clinical features and biomarkers from available dataset. We propose to use flexible machine learning methods such as random forest and Gaussian process models to build models for efficacy and toxicity depending on dose and biomarkers. A copula is used to model the joint distribution of the two outcomes and the estimates are constrained to have non-decreasing dose-efficacy and dose-toxicity relationships. Numerical utilities are elicited from clinicians for each potential bivariate outcome. For each patient, the optimal dose is chosen to maximize the posterior mean of the utility function. We also propose alternative approaches to optimal dose selection by adding additional toxicity based constraints and an approach taking into account the uncertainty in the estimation of the utility function. The proposed methods are evaluated in a simulation study to compare expected utility outcomes under various estimated optimal dose rules. Gaussian process models tended to have better performance than random forest. Enforcing monotonicity during modeling provided small benefits. Whether and how, correlation between efficacy and toxicity, was modeled, had little effect on performance. The proposed methods are illustrated with a study of patients with liver cancer treated with stereotactic body radiation therapy.

Extracorporeal membrane oxygenation support in COVID-19: an international cohort study of the Extracorporeal Life Support Organization registry.

BACKGROUND: Multiple major health organisations recommend the use of extracorporeal membrane oxygenation (ECMO) support for COVID-19-related acute hypoxaemic respiratory failure. However, initial reports of ECMO use in patients with COVID-19 described very high mortality and there have been no large, international cohort studies of ECMO for COVID-19 reported to date. METHODS: We used data from the Extracorporeal Life Support Organization (ELSO) Registry to characterise the epidemiology, hospital course, and outcomes of patients aged 16 years or older with confirmed COVID-19 who had ECMO support initiated between Jan 16 and May 1, 2020, at 213 hospitals in 36 countries. The primary outcome was in-hospital death in a time-to-event analysis assessed at 90 days after ECMO initiation. We applied a multivariable Cox model to examine whether patient and hospital factors were associated with in-hospital mortality. FINDINGS: Data for 1035 patients with COVID-19 who received ECMO support were included in this study. Of these, 67 (6%) remained hospitalised, 311 (30%) were discharged home or to an acute rehabilitation centre, 101 (10%) were discharged to a long-term acute care centre or unspecified location, 176 (17%) were discharged to another hospital, and 380 (37%) died. The estimated cumulative incidence of in-hospital mortality 90 days after the initiation of ECMO was 37·4% (95% CI 34·4-40·4). Mortality was 39% (380 of 968) in patients with a final disposition of death or hospital discharge. The use of ECMO for circulatory support was independently associated with higher in-hospital mortality (hazard ratio 1·89, 95% CI 1·20-2·97). In the subset of patients with COVID-19 receiving respiratory (venovenous) ECMO and characterised as having acute respiratory distress syndrome, the estimated cumulative incidence of in-hospital mortality 90 days after the initiation of ECMO was 38·0% (95% CI 34·6-41·5). INTERPRETATION: In patients with COVID-19 who received ECMO, both estimated mortality 90 days after ECMO and mortality in those with a final disposition of death or discharge were less than 40%. These data from 213 hospitals worldwide provide a generalisable estimate of ECMO mortality in the setting of COVID-19. FUNDING: None.

Incorporating historical models with adaptive Bayesian updates.

This article considers Bayesian approaches for incorporating information from a historical model into a current analysis when the historical model includes only a subset of covariates currently of interest. The statistical challenge is 2-fold. First, the parameters in the nested historical model are not generally equal to their counterparts in the larger current model, neither in value nor interpretation. Second, because the historical information will not be equally informative for all parameters in the current analysis, additional regularization may be required beyond that provided by the historical information. We propose several novel extensions of the so-called power prior that adaptively combine a prior based upon the historical information with a variance-reducing prior that shrinks parameter values toward zero. The ideas are directly motivated by our work building mortality risk prediction models for pediatric patients receiving extracorporeal membrane oxygenation (ECMO). We have developed a model on a registry-based cohort of ECMO patients and now seek to expand this model with additional biometric measurements, not available in the registry, collected on a small auxiliary cohort. Our adaptive priors are able to use the information in the original model and identify novel mortality risk factors. We support this with a simulation study, which demonstrates the potential for efficiency gains in estimation under a variety of scenarios.

Evaluating a Novel Prehospital Emergency Trauma Care Assessment Tool (PETCAT) for Low- and Middle-Income Countries in Sierra Leone.

BACKGROUND: WHO recommends training lay first responders (LFRs) as the first step toward formal emergency medical services development, yet no tool exists to evaluate LFR programs. METHODS: We developed Prehospital Emergency Trauma Care Assessment Tool (PETCAT), a seven-question survey administered to first-line hospital-based healthcare providers, to independently assess LFR prehospital intervention frequency and quality. PETCAT surveys were administered one month pre-LFR program launch (June 2019) in Makeni, Sierra Leone and again 14 months post-launch (August 2020). Using a difference-in-differences approach, PETCAT was also administered in a control city (Kenema) with no LFR training intervention during the study period at the same intervals to control for secular trends. PETCAT measured change in both the experimental and control locations. Cronbach's alpha, point bi-serial correlation, and inter-rater reliability using Cohen's Kappa assessed PETCAT reliability. RESULTS: PETCAT administration to 90 first-line, hospital-based healthcare providers found baseline prehospital intervention were rare in Makeni and Kenema prior to LFR program launch (1.2/10 vs. 1.8/10). Fourteen months post-LFR program implementation, PETCAT demonstrated prehospital interventions increased in Makeni with LFRs (5.2/10, p < 0.0001) and not in Kenema (1.2/10) by an adjusted difference of + 4.6 points/10 (p < 0.0001) ("never/rarely" to "half the time"), indicating negligible change due to secular trends. PETCAT demonstrated high reliability (Cronbach's α = 0.93, Cohen's K = 0.62). CONCLUSIONS: PETCAT measures changes in rates of prehospital care delivery by LFRs in a resource-limited African setting and may serve as a robust tool for independent EMS quality assessment.

A modular framework for early-phase seamless oncology trials.

BACKGROUND: As our understanding of the etiology and mechanisms of cancer becomes more sophisticated and the number of therapeutic options increases, phase I oncology trials today have multiple primary objectives. Many such designs are now "seamless," meaning that the trial estimates both the maximum tolerated dose and the efficacy at this dose level. Sponsors often proceed with further study only with this additional efficacy evidence. However, with this increasing complexity in trial design, it becomes challenging to articulate fundamental operating characteristics of these trials, such as (1) what is the probability that the design will identify an acceptable, that is., safe and efficacious, dose level? or (2) how many patients will be assigned to an acceptable dose level on average? METHODS: In this manuscript, we propose a new modular framework for designing and evaluating seamless oncology trials. Each module is comprised of either a dose assignment step or a dose-response evaluation, and multiple such modules can be implemented sequentially. We develop modules from existing phase I/II designs as well as a novel module for evaluating dose-response using a Bayesian isotonic regression scheme. RESULTS: We also demonstrate a freely available R package called seamlesssim to numerically estimate, by means of simulation, the operating characteristics of these modular trials. CONCLUSIONS: Together, this design framework and its accompanying simulator allow the clinical trialist to compare multiple different candidate designs, more rigorously assess performance, better justify sample sizes, and ultimately select a higher quality design.

Genetic anticipation in Swedish Lynch syndrome families.

Among hereditary colorectal cancer predisposing syndromes, Lynch syndrome (LS) caused by mutations in DNA mismatch repair genes MLH1, MSH2, MSH6 or PMS2 is the most common. Patients with LS have an increased risk of early onset colon and endometrial cancer, but also other tumors that generally have an earlier onset compared to the general population. However, age at first primary cancer varies within families and genetic anticipation, i.e. decreasing age at onset in successive generations, has been suggested in LS. Anticipation is a well-known phenomenon in e.g neurodegenerative diseases and several reports have studied anticipation in heritable cancer. The purpose of this study is to determine whether anticipation can be shown in a nationwide cohort of Swedish LS families referred to the regional departments of clinical genetics in Lund, Stockholm, Linköping, Uppsala and Umeå between the years 1990-2013. We analyzed a homogenous group of mutation carriers, utilizing information from both affected and non-affected family members. In total, 239 families with a mismatch repair gene mutation (96 MLH1 families, 90 MSH2 families including one family with an EPCAM-MSH2 deletion, 39 MSH6 families, 12 PMS2 families, and 2 MLH1+PMS2 families) comprising 1028 at-risk carriers were identified among the Swedish LS families, of which 1003 mutation carriers had available follow-up information and could be included in the study. Using a normal random effects model (NREM) we estimate a 2.1 year decrease in age of diagnosis per generation. An alternative analysis using a mixed-effects Cox proportional hazards model (COX-R) estimates a hazard ratio of exp(0.171), or about 1.19, for age of diagnosis between consecutive generations. LS-associated gene-specific anticipation effects are evident for MSH2 (2.6 years/generation for NREM and hazard ratio of 1.33 for COX-R) and PMS2 (7.3 years/generation and hazard ratio of 1.86). The estimated anticipation effects for MLH1 and MSH6 are smaller.

Multicentre retrospective study of intravascular large B-cell lymphoma treated at academic institutions within the United States.

Intravascular large B-cell lymphoma (IVLBCL) is a rare entity, with a generally aggressive course that may vary based on geographic presentation. While a United States (US) registry study showed relatively good outcomes with IVLBCL, clinicopathological and treatment data were unavailable. We performed a detailed retrospective review of cases identified at 8 US medical centres, to improve understanding of IVLBCL and inform management. We compiled data retrieved via an Institutional Review Board-approved review of IVLBCL cases identified from 1999 to 2015 at nine academic institutions across the US. We characterized the cohort's clinical status at time of diagnosis, presenting diagnostic and clinical features of the disease, treatment modalities used and overall prognostic data. Our cohort consisted of 54 patients with varying degrees of clinical features. Adjusting for age, better performance status at presentation was associated with increased survival time for the patients diagnosed in vivo (hazard ratio: 2·12, 95% confidence interval 1·28, 3·53). Based on the data we have collected, it would appear that the time interval to diagnosis is a significant contributor to outcomes of patients with IVLBCL.

Accounting for established predictors with the multistep elastic net.

Multivariable models for prediction or estimating associations with an outcome are rarely built in isolation. Instead, they are based upon a mixture of covariates that have been evaluated in earlier studies (eg, age, sex, or common biomarkers) and covariates that were collected specifically for the current study (eg, a panel of novel biomarkers or other hypothesized risk factors). For that context, we present the multistep elastic net (MSN), which considers penalized regression with variables that can be qualitatively grouped based upon their degree of prior research support: established predictors vs unestablished predictors. The MSN chooses between uniform penalization of all predictors (the standard elastic net) and weaker penalization of the established predictors in a cross-validated framework and includes the option to impose zero penalty on the established predictors. In simulation studies that reflect the motivating context, we show the comparability or superiority of the MSN over the standard elastic net, the Integrative LASSO with Penalty Factors, the sparse group lasso, and the group lasso, and we investigate the importance of not penalizing the established predictors at all. We demonstrate the MSN to update a prediction model for pediatric ECMO patient mortality.

Default Priors for the Intercept Parameter in Logistic Regressions.

In logistic regression, separation occurs when a linear combination of predictors perfectly discriminates the binary outcome. Because finite-valued maximum likelihood parameter estimates do not exist under separation, Bayesian regressions with informative shrinkage of the regression coefficients offer a suitable alternative. Classical studies of separation imply that efficiency in estimating regression coefficients may also depend upon the choice of intercept prior, yet relatively little focus has been given on whether and how to shrink the intercept parameter. Alternative prior distributions for the intercept are proposed that downweight implausibly extreme regions of the parameter space, rendering regression estimates that are less sensitive to separation. Through simulation and the analysis of exemplar datasets, differences across priors stratified by established statistics measuring the degree of separation are quantified. Relative to diffuse priors, these proposed priors generally yield more efficient estimation of the regression coefficients themselves when the data are nearly separated. They are equally efficient in non-separated datasets, making them suitable for default use. Modest differences were observed with respect to out-of-sample discrimination. These numerical studies also highlight the interplay between priors for the intercept and the regression coefficients: findings are more sensitive to the choice of intercept prior when using a weakly informative prior on the regression coefficients than an informative shrinkage prior.

A single center phase II study of ixazomib in patients with relapsed or refractory cutaneous or peripheral T-cell lymphomas.

The transcription factor GATA-3, highly expressed in many cutaneous T-cell lymphoma (CTCL) and peripheral T-cell lymphomas (PTCL), confers resistance to chemotherapy in a cell-autonomous manner. As GATA-3 is transcriptionally regulated by NF-κB, we sought to determine the extent to which proteasomal inhibition impairs NF-κB activation and GATA-3 expression and cell viability in malignant T cells. Proteasome inhibition, NF-κB activity, GATA-3 expression, and cell viability were examined in patient-derived cell lines and primary T-cell lymphoma specimens ex vivo treated with the oral proteasome inhibitor ixazomib. Significant reductions in cell viability, NF-κB activation, and GATA-3 expression were observed preclinically in ixazomib-treated cells. Therefore, an investigator-initiated, single-center, phase II study with this agent in patients with relapsed/refractory CTCL/PTCL was conducted. Concordant with our preclinical observations, a significant reduction in NF-κB activation and GATA-3 expression was observed in an exceptional responder following one month of treatment with ixazomib. While ixazomib had limited activity in this small and heterogeneous cohort of patients, inhibition of the NF-κB/GATA-3 axis in a single exceptional responder suggests that ixazomib may have utility in appropriately selected patients or in combination with other agents.

Hospital-level variation in inpatient cost among children receiving extracorporeal membrane oxygenation.

OBJECTIVE: Pediatric extracorporeal membrane oxygenation (ECMO) varies in the way care is provided from hospital to hospital. This variability in hospital ECMO care can be represented by the variation in ECMO costs. We hypothesized that hospitals will demonstrate large variations in case-mix-adjusted ECMO inpatient costs for children requiring ECMO and higher volume hospitals will have lower associated costs. METHODS: We retrospectively analyzed the inpatient cost of children receiving ECMO in 2006, 2009 and 2012, using the Healthcare Cost and Utilization Project Kids' Inpatient Database. We used a hierarchical linear regression model and the intraclass correlation coefficient to quantify how much of the difference in ECMO inpatient costs was associated with the hospital where a child received care. To do this, we adjusted for patient factors, hospital factors and potentially modifiable factors such as complications, procedures and length of stay. RESULTS: The median inflation-adjusted inpatient costs for children requiring ECMO were $183,000, $240,000 and $241,000 in years 2006, 2009 and 2012, respectively. The largest median cost for ECMO cases in a given hospital in a given year ($690,000) was more than 11 times that of the smallest median cost ($60,000). After case-mix adjustment, 27% of the variation in inpatient costs was associated with the hospital where ECMO care was provided. Average hospital costs were not associated with hospital ECMO volume. CONCLUSIONS: The large variation in ECMO inpatient costs between hospitals suggests great variation in care between hospitals, which is important because hospitals have a co-existing variation in ECMO survival rates.

Low-Cost Intervention to Increase Influenza Vaccination Rate at a Comprehensive Cancer Center.

Influenza morbidity and mortality can be severe and costly. Vaccination rates remain suboptimal in cancer patients due to provider- and patient-related factors. The objective of this study was to evaluate whether low-cost provider- and patient-focused interventions would increase influenza vaccination rates at the University of Michigan Comprehensive Cancer Center (UMCCC). This quality improvement project included all patients without documentation of influenza vaccination prior to their first outpatient appointment during the 2011-2012 and 2012-2013 influenza seasons. The multi-stepped intervention included provider and patient reminders. Influenza vaccination rates were compiled using CPT-4 codes. Same-day (with appointment) vaccination rates during the intervention seasons were compared to historical (2005-2011 seasons) controls; vaccination rates were also compared to contemporary control population at the University of Michigan Health System (UMHS). Reasons for non-adherence with vaccination were explored. The cumulative same-day vaccination rate in eligible adults was 10.1 % (2011-2012) and 9.4 % (2012-2013) compared to an average 6.9 % during influenza seasons 2005-2011. Based on logistic regression analysis, there was a 37.6 % (95 % CI 35-40.3 %) and 56.1 % (95 % CI 40.9-73 %) relative increase in the adult vaccination rate associated with the intervention, with 399 and 697 additional vaccinations, respectively, for each season. During the 2012-2013 season, the UMCCC adult vaccination rate was higher compared to the remainder of that of the UMHS. The intervention was well accepted by providers. Reasons for no vaccination were provider- and patient-related. Increasing provider and patient awareness with a simple, inexpensive intervention was associated with higher influenza vaccination rates at a large academic cancer center. The intervention is permanently implemented during influenza seasons.

Tests for Gene-Environment Interactions and Joint Effects With Exposure Misclassification.

The number of methods for genome-wide testing of gene-environment (G-E) interactions continues to increase, with the aim of discovering new genetic risk factors and obtaining insight into the disease-gene-environment relationship. The relative performance of these methods, assessed on the basis of family-wise type I error rate and power, depends on underlying disease-gene-environment associations, estimates of which may be biased in the presence of exposure misclassification. This simulation study expands on a previously published simulation study of methods for detecting G-E interactions by evaluating the impact of exposure misclassification. We consider 7 single-step and modular screening methods for identifying G-E interaction at a genome-wide level and 7 joint tests for genetic association and G-E interaction, for which the goal is to discover new genetic susceptibility loci by leveraging G-E interaction when present. In terms of statistical power, modular methods that screen on the basis of the marginal disease-gene relationship are more robust to exposure misclassification. Joint tests that include main/marginal effects of a gene display a similar robustness, which confirms results from earlier studies. Our results offer an increased understanding of the strengths and limitations of methods for genome-wide searches for G-E interaction and joint tests in the presence of exposure misclassification.

The impact of exposure-biased sampling designs on detection of gene-environment interactions in case-control studies with potential exposure misclassification.

With limited funding and biological specimen availability, choosing an optimal sampling design to maximize power for detecting gene-by-environment (G-E) interactions is critical. Exposure-enriched sampling is often used to select subjects with rare exposures for genotyping to enhance power for tests of G-E effects. However, exposure misclassification (MC) combined with biased sampling can affect characteristics of tests for G-E interaction and joint tests for marginal association and G-E interaction. Here, we characterize the impact of exposure-biased sampling under conditions of perfect exposure information and exposure MC on properties of several methods for conducting inference. We assess the Type I error, power, bias, and mean squared error properties of case-only, case-control, and empirical Bayes methods for testing/estimating G-E interaction and a joint test for marginal G (or E) effect and G-E interaction across three biased sampling schemes. Properties are evaluated via empirical simulation studies. With perfect exposure information, exposure-enriched sampling schemes enhance power as compared to random selection of subjects irrespective of exposure prevalence but yield bias in estimation of the G-E interaction and marginal E parameters. Exposure MC modifies the relative performance of sampling designs when compared to the case of perfect exposure information. Those conducting G-E interaction studies should be aware of exposure MC properties and the prevalence of exposure when choosing an ideal sampling scheme and method for characterizing G-E interactions and joint effects.