Designing Rigorous and Efficient Clinical Utility Studies for Early Detection Biomarkers.

Before implementing a biomarker in routine clinical care, it must demonstrate clinical utility by leading to clinical actions that positively affect patient-relevant outcomes. Randomly controlled early detection utility trials, especially those targeting mortality endpoint, are challenging due to their high costs and prolonged duration. Special design considerations are required to determine the clinical utility of early detection assays. This commentary reports on discussions among the National Cancer Institute's Early Detection Research Network investigators, outlining the recommended process for carrying out single-organ biomarker-driven clinical utility studies. We present the early detection utility studies in the context of phased biomarker development. We describe aspects of the studies related to the features of biomarker tests, the clinical context of endpoints, the performance criteria for later phase evaluation, and study size. We discuss novel adaptive design approaches for improving the efficiency and practicality of clinical utility trials. We recommend using multiple strategies, including adopting real-world evidence, emulated trials, and mathematical modeling to circumvent the challenges in conducting early detection utility trials.

Designing cancer screening trials for reduction in late-stage cancer incidence.

Before implementing a biomarker test for early cancer detection into routine clinical care, the test must demonstrate clinical utility, that is, the test results should lead to clinical actions that positively affect patient-relevant outcomes. Unlike therapeutical trials for patients diagnosed with cancer, designing a randomized controlled trial (RCT) to demonstrate the clinical utility of an early detection biomarker with mortality and related endpoints poses unique challenges. The hurdles stem from the prolonged natural progression of the disease and the lack of information regarding the time-varying screening effect on the target asymptomatic population. To facilitate the study design of screening trials, we propose using a generic multistate disease history model and derive model-based effect sizes. The model links key performance metrics of the test, such as sensitivity, to primary endpoints like the incidence of late-stage cancer. It also incorporates the practical implementation of the biomarker-testing program in real-world scenarios. Based on the chronological time scale aligned with RCT, our method allows the assessment of study powers based on key features of the new program, including the test sensitivity, the length of follow-up, and the number and frequency of repeated tests. The calculation tool from the proposed method will enable practitioners to perform realistic and quick evaluations when strategizing screening trials for specific diseases. We use numerical examples based on the National Lung Screening Trial to demonstrate the method.

Estimating individualized treatment rules by optimizing the adjusted probability of a longer survival.

Individualized treatment rules inform tailored treatment decisions based on the patient's information, where the goal is to optimize clinical benefit for the population. When the clinical outcome of interest is survival time, most of current approaches typically aim to maximize the expected time of survival. We propose a new criterion for constructing Individualized treatment rules that optimize the clinical benefit with survival outcomes, termed as the adjusted probability of a longer survival. This objective captures the likelihood of living longer with being on treatment, compared to the alternative, which provides an alternative and often straightforward interpretation to communicate with clinicians and patients. We view it as an alternative to the survival analysis standard of the hazard ratio and the increasingly used restricted mean survival time. We develop a new method to construct the optimal Individualized treatment rule by maximizing a nonparametric estimator of the adjusted probability of a longer survival for a decision rule. Simulation studies demonstrate the reliability of the proposed method across a range of different scenarios. We further perform data analysis using data collected from a randomized Phase III clinical trial (SWOG S0819).

Constructing time-invariant dynamic surveillance rules for optimal monitoring schedules.

Dynamic surveillance rules (DSRs) are sequential surveillance decision rules informing monitoring schedules in clinical practice, which can adapt over time according to a patient's evolving characteristics. In many clinical applications, it is desirable to identify and implement optimal time-invariant DSRs, where the parameters indexing the decision rules are shared across different decision points. We propose a new criterion for DSRs that accounts for benefit-cost tradeoff during the course of disease surveillance. We develop two methods to estimate the time-invariant DSRs optimizing the proposed criterion, and establish asymptotic properties for the estimated parameters of biomarkers indexing the DSRs. The first approach estimates the optimal decision rules for each individual at every stage via regression modeling, and then estimates the time-invariant DSRs via a classification procedure with the estimated time-varying decision rules as the response. The second approach proceeds by optimizing a relaxation of the empirical objective, where a surrogate function is utilized to facilitate computation. Extensive simulation studies are conducted to demonstrate the superior performances of the proposed methods. The methods are further applied to the Canary Prostate Active Surveillance Study (PASS).

Reinforced risk prediction with budget constraint using irregularly measured data from electronic health records.

Uncontrolled glycated hemoglobin (HbA1c) levels are associated with adverse events among complex diabetic patients. These adverse events present serious health risks to affected patients and are associated with significant financial costs. Thus, a high-quality predictive model that could identify high-risk patients so as to inform preventative treatment has the potential to improve patient outcomes while reducing healthcare costs. Because the biomarker information needed to predict risk is costly and burdensome, it is desirable that such a model collect only as much information as is needed on each patient so as to render an accurate prediction. We propose a sequential predictive model that uses accumulating patient longitudinal data to classify patients as: high-risk, low-risk, or uncertain. Patients classified as high-risk are then recommended to receive preventative treatment and those classified as low-risk are recommended to standard care. Patients classified as uncertain are monitored until a high-risk or low-risk determination is made. We construct the model using claims and enrollment files from Medicare, linked with patient Electronic Health Records (EHR) data. The proposed model uses functional principal components to accommodate noisy longitudinal data and weighting to deal with missingness and sampling bias. The proposed method demonstrates higher predictive accuracy and lower cost than competing methods in a series of simulation experiments and application to data on complex patients with diabetes.

Hypothalamic-Pituitary-Adrenal Hormones Impair Pig Fertilization and Preimplantation Embryo Development via Inducing Oviductal Epithelial Apoptosis: An In Vitro Study.

Previous studies show that stressful events after ovulation in sows significantly impaired the embryo cleavage with a significant elevation of blood cortisol. However, the effects of corticotropin-releasing hormone (CRH), adrenocorticotropic hormone (ACTH) and cortisol on fertilization and embryo development remain to be specified, and whether they damage pig embryos directly or indirectly is unclear. This study demonstrated that embryo development was unaffected when pig parthenotes were cultured with different concentrations of CRH/ACTH/cortisol. However, embryo development was significantly impaired when the embryos were cocultured with pig oviductal epithelial cells (OECs) in the presence of CRH/cortisol or cultured in medium that was conditioned with CRH/cortisol-pretreated OECs (CRH/cortisol-CM). Fertilization in CRH/cortisol-CM significantly increased the rates of polyspermy. CRH and cortisol induced apoptosis of OECs through FAS and TNFα signaling. The apoptotic OECs produced less growth factors but more FASL and TNFα, which induced apoptosis in embryos. Pig embryos were not sensitive to CRH because they expressed no CRH receptor but the CRH-binding protein, and they were tolerant to cortisol because they expressed more 11-beta hydroxysteroid dehydrogenase 2 (HSD11B2) than HSD11B1. When used at a stress-induced physiological concentration, while culture with either CRH or cortisol alone showed no effect, culture with both significantly increased apoptosis in OECs. In conclusion, CRH and cortisol impair pig fertilization and preimplantation embryo development indirectly by inducing OEC apoptosis via the activation of the FAS and TNFα systems. ACTH did not show any detrimental effect on pig embryos, nor OECs.

Early Detection Initiative: A randomized controlled trial of algorithm-based screening in patients with new onset hyperglycemia and diabetes for early detection of pancreatic ductal adenocarcinoma.

Pancreatic ductal adenocarcinoma (PDAC) is the only leading cause of cancer death without an early detection strategy. In retrospective studies, 0.5-1% of subjects >50 years of age who newly develop biochemically-defined diabetes have been diagnosed with PDAC within 3 years of meeting new onset hyperglycemia and diabetes (NOD) criteria. The Enriching New-onset Diabetes for Pancreatic Cancer (ENDPAC) algorithm further risk stratifies NOD subjects based on age and changes in weight and diabetes parameters. We present the methodology for the Early Detection Initiative (EDI), a randomized controlled trial of algorithm-based screening in patients with NOD for early detection of PDAC. We hypothesize that study interventions (risk stratification with ENDPAC and imaging with Computerized Tomography (CT) scan) in NOD will identify earlier stage PDAC. EDI uses a modified Zelen's design with post-randomization consent. Eligible subjects will be identified through passive surveillance of electronic medical records and eligible study participants randomized 1:1 to the Intervention or Observation arm. The sample size is 12,500 subjects. The ENDPAC score will be calculated only in those randomized to the Intervention arm, with 50% (n = 3125) expected to have a high ENDPAC score. Consenting subjects in the high ENDPAC group will undergo CT imaging for PDAC detection and an estimate of potential harm. The effectiveness and efficacy evaluation will compare proportions of late stage PDAC between Intervention and Observation arm per randomization assignment or per protocol, respectively, with a planned interim analysis. The study is designed to improve the detection of sporadic PDAC when surgical intervention is possible.

Simultaneous spatial smoothing and outlier detection using penalized regression, with application to childhood obesity surveillance from electronic health records.

Electronic health records (EHRs) have become a platform for data-driven granular-level surveillance in recent years. In this paper, we make use of EHRs for early prevention of childhood obesity. The proposed method simultaneously provides smooth disease mapping and outlier information for obesity prevalence that are useful for raising public awareness and facilitating targeted intervention. More precisely, we consider a penalized multilevel generalized linear model. We decompose regional contribution into smooth and sparse signals, which are automatically identified by a combination of fusion and sparse penalties imposed on the likelihood function. In addition, we weigh the proposed likelihood to account for the missingness and potential nonrepresentativeness arising from the EHR data. We develop a novel alternating minimization algorithm, which is computationally efficient, easy to implement, and guarantees convergence. Simulation studies demonstrate superior performance of the proposed method. Finally, we apply our method to the University of Wisconsin Population Health Information Exchange database.

Restricted survival benefit with right-censored data.

The hazard ratio is widely used to quantify treatment effects. However, it may be difficult to interpret for patients and practitioners, especially when the hazard ratio is not constant over time. Alternative measures of the treatment effects have been proposed such as the difference of the restricted mean survival times, the difference in survival proportions at some fixed follow-up time, or the net chance of a longer survival. In this paper, we propose the restricted survival benefit (RSB), a quantity that can incorporate multiple useful measurements of treatment effects. Hence, it provides a framework for a comprehensive assessment of the treatment effects. We provide estimation and inference procedures for the RSB that accommodate censored survival outcomes, using methods of the inverse-probability-censoring-weighted U$U$ -statistic and the jackknife empirical likelihood. We conduct extensive simulation studies to examine the numerical performance of the proposed method, and we analyze data from a randomized Phase III clinical trial (SWOG S0777) using the proposed method.

Estimation and inference on high-dimensional individualized treatment rule in observational data using split-and-pooled de-correlated score.

With the increasing adoption of electronic health records, there is an increasing interest in developing individualized treatment rules, which recommend treatments according to patients' characteristics, from large observational data. However, there is a lack of valid inference procedures for such rules developed from this type of data in the presence of high-dimensional covariates. In this work, we develop a penalized doubly robust method to estimate the optimal individualized treatment rule from high-dimensional data. We propose a split-and-pooled de-correlated score to construct hypothesis tests and confidence intervals. Our proposal adopts the data splitting to conquer the slow convergence rate of nuisance parameter estimations, such as non-parametric methods for outcome regression or propensity models. We establish the limiting distributions of the split-and-pooled de-correlated score test and the corresponding one-step estimator in high-dimensional setting. Simulation and real data analysis are conducted to demonstrate the superiority of the proposed method.

Discussion of Kallus (2020) and Mo et al (2020).

We discuss the results on improving the generalizability of individualized treatment rule following the work in Kallus [1] and Mo et al. [5]. We note that the advocated weights in Kallus [1] are connected to the efficient score of the contrast function. We further propose a likelihood-ratio-based method (LR-ITR) to accommodate covariate shifts, and compare it to the CTE-DR-ITR method proposed in Mo et al. [5]. We provide the upper-bound on the risk function of the target population when both the covariate shift and the contrast function shift are present. Numerical studies show that LR-ITR can outperform CTE-DR-ITR when there is only covariate shift.

Improved doubly robust estimation in learning optimal individualized treatment rules.

Individualized treatment rules (ITRs) recommend treatment according to patient characteristics. There is a growing interest in developing novel and efficient statistical methods in constructing ITRs. We propose an improved doubly robust estimator of the optimal ITRs. The proposed estimator is based on a direct optimization of an augmented inverse-probability weighted estimator (AIPWE) of the expected clinical outcome over a class of ITRs. The method enjoys two key properties. First, it is doubly robust, meaning that the proposed estimator is consistent when either the propensity score or the outcome model is correct. Second, it achieves the smallest variance among the class of doubly robust estimators when the propensity score model is correctly specified, regardless of the specification of the outcome model. Simulation studies show that the estimated ITRs obtained from our method yield better results than those obtained from current popular methods. Data from the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study is analyzed as an illustrative example.

Small area estimation and childhood obesity surveillance using electronic health records.

There is an urgent need for childhood surveillance systems to design, implement, and evaluate interventions at the local level. We estimated obesity prevalence for individuals aged 5-17 years using a southcentral Wisconsin EHR data repository, Public Health Information Exchange (PHINEX, 2007-2012). The prevalence estimates were calculated by aggregating the estimated probability of each individual being obese, which was obtained via a generalized linear mixed model. We incorporated the random effects at the area level into our model. A weighted procedure was employed to account for missingness in EHR data. A non-parametric kernel smoothing method was used to obtain the prevalence estimates for locations with no or little data (<20 individuals) from the EHR. These estimates were compared to results from newly available obesity atlas (2015-2016) developed from various EHRs with greater statewide representation. The mean of the zip code level obesity prevalence estimates for males and females aged 5-17 years is 16.2% (SD 2.72%); 17.9% (SD 2.14%) for males and 14.4% (SD 2.00%) for females. The results were comparable to the Wisconsin Health Atlas (WHA) estimates, a much larger dataset of local community EHRs in Wisconsin. On average, prevalence estimates were 2.12% lower in this process than the WHA estimates, with lower estimation occurring more frequently for zip codes without data in PHINEX. Using this approach, we can obtain estimates for local areas that lack EHRs data. Generally, lower prevalence estimates were produced for those locations not represented in the PHINEX database when compared to WHA estimates. This underscores the need to ensure that the reference EHRs database can be made sufficiently similar to the geographic areas where synthetic estimates are being created.

Corticosterone induced apoptosis of mouse oviduct epithelial cells independent of the TNF-α system.

It has been reported in recent studies that restraint stress on pregnant mice during the preimplantation stage elevated corticotrophin-releasing hormone (CRH) and glucocorticoid levels in the serum and oviducts; furthermore, CRH and corticosterone (CORT) impacted preimplantation embryos indirectly by triggering the apoptosis of oviductal epithelial cells (OECs) through activation of the Fas system. However, it remains unclear whether TNF-α signaling is involved in CRH- and/or glucocorticoid-induced apoptosis of OECs. In the present study, it was shown that culture with either CRH or CORT induced significant apoptosis of OECs. The culture of OECs with CRH augmented both FasL expression and TNF-α expression. However, culture with CORT increased FasL, but decreased TNF-α, expression significantly. Although knocking down/knocking out FasL expression in OECs significantly ameliorated the proapoptotic effects of both CRH and CORT, knocking down/knocking out TNF-α expression relieved only the proapoptotic effect of CRH but not that of CORT. Taken together, our results demonstrated that CRH-induced OEC apoptosis involved both Fas signaling and TNF-α signaling. Conversely, CORT-induced OEC apoptosis involved only the Fas, but not the TNF-α, signaling pathway. The data obtained are crucial for our understanding of the mechanisms by which various categories of stress imposed on pregnant females impair embryo development, as well as for the development of measures to protect the embryo from the adverse effects of stress.

Multicategory Outcome Weighted Margin-based Learning for Estimating Individualized Treatment Rules.

Due to heterogeneity for many chronic diseases, precise personalized medicine, also known as precision medicine, has drawn increasing attentions in the scientific community. One main goal of precision medicine is to develop the most effective tailored therapy for each individual patient. To that end, one needs to incorporate individual characteristics to detect a proper individual treatment rule (ITR), by which suitable decisions on treatment assignments can be made to optimize patients' clinical outcome. For binary treatment settings, outcome weighted learning (OWL) and several of its variations have been proposed recently to estimate the ITR by optimizing the conditional expected outcome given patients' information. However, for multiple treatment scenarios, it remains unclear how to use OWL effectively. It can be shown that some direct extensions of OWL for multiple treatments, such as one-versus-one and one-versus-rest methods, can yield suboptimal performance. In this paper, we propose a new learning method, named Multicategory Outcome weighted Margin-based Learning (MOML), for estimating ITR with multiple treatments. Our proposed method is very general and covers OWL as a special case. We show Fisher consistency for the estimated ITR, and establish convergence rate properties. Variable selection using the sparse l 1 penalty is also considered. Analysis of simulated examples and a type 2 diabetes mellitus observational study are used to demonstrate competitive performance of the proposed method.

Glucocorticoids impair oocyte competence and trigger apoptosis of ovarian cells via activating the TNF-α system.

Mechanisms by which female stress and particularly glucocorticoids impair oocyte competence are largely unclear. Although one study demonstrated that glucocorticoids triggered apoptosis in ovarian cells and oocytes by activating the FasL/Fas system, other studies suggested that they might induce apoptosis through activating other signaling pathways as well. In this study, both in vivo and in vitro experiments were conducted to test the hypothesis that glucocorticoids might trigger apoptosis in oocytes and ovarian cells through activating the TNF-α system. The results showed that cortisol injection of female mice (1.) impaired oocyte developmental potential and mitochondrial membrane potential with increased oxidative stress; (2.) induced apoptosis in mural granulosa cells (MGCs) with increased oxidative stress in the ovary; and (3.) activated the TNF-α system in both ovaries and oocytes. Culture with corticosterone induced apoptosis and activated the TNF-α system in MGCs. Knockdown or knockout of TNF-α significantly ameliorated the pro-apoptotic effects of glucocorticoids on oocytes and MGCs. However, culture with corticosterone downregulated TNF-α expression significantly in oviductal epithelial cells. Together, the results demonstrated that glucocorticoids impaired oocyte competence and triggered apoptosis in ovarian cells through activating the TNF-α system and that the effect of glucocorticoids on TNF-α expression might vary between cell types.

Constructing dynamic treatment regimes with shared parameters for censored data.

Dynamic treatment regimes are sequential decision rules that adapt throughout disease progression according to a patient's evolving characteristics. In many clinical applications, it is desirable that the format of the decision rules remains consistent over time. Unlike the estimation of dynamic treatment regimes in regular settings, where decision rules are formed without shared parameters, the derivation of the shared decision rules requires estimating shared parameters indexing the decision rules across different decision points. Estimation of such rules becomes more complicated when the clinical outcome of interest is a survival time subject to censoring. To address these challenges, we propose two novel methods: censored shared-Q-learning and censored shared-O-learning. Both methods incorporate clinical preferences into a qualitative rule, where the parameters indexing the decision rules are shared across different decision points and estimated simultaneously. We use simulation studies to demonstrate the superior performance of the proposed methods. The methods are further applied to the Framingham Heart Study to derive treatment rules for cardiovascular disease.

Designing precision medicine trials to yield a greater population impact.

Traditionally, a clinical trial is conducted comparing treatment to standard care for all patients. However, it could be inefficient given patients' heterogeneous responses to treatments, and rapid advances in the molecular understanding of diseases have made biomarker-based clinical trials increasingly popular. We propose a new targeted clinical trial design, termed as Max-Impact design, which selects the appropriate subpopulation for a clinical trial and aims to optimize population impact once the trial is completed. The proposed design not only gains insights on the patients who would be included in the trial but also considers the benefit to the excluded patients. We develop novel algorithms to construct enrollment rules for optimizing population impact, which are fairly general and can be applied to various types of outcomes. Simulation studies and a data example from the SWOG Cancer Research Network demonstrate the competitive performance of our proposed method compared to traditional untargeted and targeted designs.

Learning-based biomarker-assisted rules for optimized clinical benefit under a risk constraint.

Novel biomarkers, in combination with currently available clinical information, have been sought to improve clinical decision making in many branches of medicine, including screening, surveillance, and prognosis. Statistical methods are needed to integrate such diverse information to develop targeted interventions that balance benefit and harm. In the specific setting of disease detection, we propose novel approaches to construct a multiple-marker-based decision rule by directly optimizing a benefit function, while controlling harm at a maximally tolerable level. These new approaches include plug-in and direct-optimization-based algorithms, and they allow for the construction of both nonparametric and parametric rules. A study of asymptotic properties of the proposed estimators is provided. Simulation results demonstrate good clinical utilities for the resulting decision rules under various scenarios. The methods are applied to a biomarker study in prostate cancer surveillance.

Near-optimal Individualized Treatment Recommendations.

The individualized treatment recommendation (ITR) is an important analytic framework for precision medicine. The goal of ITR is to assign the best treatments to patients based on their individual characteristics. From the machine learning perspective, the solution to the ITR problem can be formulated as a weighted classification problem to maximize the mean benefit from the recommended treatments given patients' characteristics. Several ITR methods have been proposed in both the binary setting and the multicategory setting. In practice, one may prefer a more flexible recommendation that includes multiple treatment options. This motivates us to develop methods to obtain a set of near-optimal individualized treatment recommendations alternative to each other, called alternative individualized treatment recommendations (A-ITR). We propose two methods to estimate the optimal A-ITR within the outcome weighted learning (OWL) framework. Simulation studies and a real data analysis for Type 2 diabetic patients with injectable antidiabetic treatments are conducted to show the usefulness of the proposed A-ITR framework. We also show the consistency of these methods and obtain an upper bound for the risk between the theoretically optimal recommendation and the estimated one. An R package aitr has been developed, found at https://github.com/menghaomiao/aitr.