Non-monotone Exponential Time (NEXT) Model for the Longitudinal Trend of a Continuous Outcome in Clinical Trials.

The dose-response curve has been studied extensively for decades. However, most of these methods ignore intermediate measurements of the response variable and only use the measurement at the endpoint. In early phase trials, it is crucial to utilize all available data due to the smaller sample size. Simulation studies have shown that the longitudinal dose-response surface model provides a more precise parameter estimation compared to the traditional dose response using only data from the primary time point. However, the current longitudinal models with parametric assumptions assume the treatment effect increases monotonically over time, which may be controversial to reality. We propose a parametric non-monotone exponential time (NEXT) model, an enhanced longitudinal dose-response model with greater flexibility, capable of accommodating non-monotonic treatment effects and making predictions for longer-term efficacy. In addition, the estimator for the time to maximum treatment effect and its asymptotic distribution have been derived from NEXT. Extensive simulation studies using known data-generating models and using real clinical data showed the NEXT model outperformed the existing monotone longitudinal models.

Essential team science skills for biostatisticians on collaborative research teams.

Introduction: Despite the critical role that quantitative scientists play in biomedical research, graduate programs in quantitative fields often focus on technical and methodological skills, not on collaborative and leadership skills. In this study, we evaluate the importance of team science skills among collaborative biostatisticians for the purpose of identifying training opportunities to build a skilled workforce of quantitative team scientists. Methods: Our workgroup described 16 essential skills for collaborative biostatisticians. Collaborative biostatisticians were surveyed to assess the relative importance of these skills in their current work. The importance of each skill is summarized overall and compared across career stages, highest degrees earned, and job sectors. Results: Survey respondents were 343 collaborative biostatisticians spanning career stages (early: 24.2%, mid: 33.8%, late: 42.0%) and job sectors (academia: 69.4%, industry: 22.2%, government: 4.4%, self-employed: 4.1%). All 16 skills were rated as at least somewhat important by > 89.0% of respondents. Significant heterogeneity in importance by career stage and by highest degree earned was identified for several skills. Two skills ("regulatory requirements" and "databases, data sources, and data collection tools") were more likely to be rated as absolutely essential by those working in industry (36.5%, 65.8%, respectively) than by those in academia (19.6%, 51.3%, respectively). Three additional skills were identified as important by survey respondents, for a total of 19 collaborative skills. Conclusions: We identified 19 team science skills that are important to the work of collaborative biostatisticians, laying the groundwork for enhancing graduate programs and establishing effective on-the-job training initiatives to meet workforce needs.

Neurosurgery and coronavirus: impact and challenges-lessons learnt from the first wave of a global pandemic.

INTRODUCTION AND OBJECTIVES: The novel severe acute respiratory syndrome coronavirus 2 (COVID-19) pandemic has had drastic effects on global healthcare with the UK amongst the countries most severely impacted. The aim of this study was to examine how COVID-19 challenged the neurosurgical delivery of care in a busy tertiary unit serving a socio-economically diverse population. METHODS: A prospective single-centre cohort study including all patients referred to the acute neurosurgical service or the subspecialty multidisciplinary teams (MDT) as well as all emergency and elective admissions during COVID-19 (18th March 2020-15th May 2020) compared to pre-COVID-19 (18th of January 2020-17th March 2020). Data on demographics, diagnosis, operation, and treatment recommendation/outcome were collected and analysed. RESULTS: Overall, there was a reduction in neurosurgical emergency referrals by 33.6% and operations by 55.6% during the course of COVID-19. There was a significant increase in the proportion of emergency operations performed during COVID-19 (75.2% of total, n=155) when compared to pre-COVID-19 (n = 198, 43.7% of total, p < 0.00001). In contrast to other published series, the 30-day perioperative mortality remained low (2.0%) with the majority of post-operative COVID-19-infected patients (n = 13) having underlying medical co-morbidities and/or suffering from post-operative complications. CONCLUSION: The capacity to safely treat patients requiring urgent or emergency neurosurgical care was maintained at all times. Strategies adopted to enable this included proactively approaching the referrers to maintain lines of communications, incorporating modern technology to run clinics and MDTs, restructuring patient pathways/facilities, and initiating the delivery of NHS care within private sector hospitals. Through this multi-modal approach we were able to minimize service disruptions, the complications, and mortality.

Methods for training collaborative biostatisticians.

The emphasis on team science in clinical and translational research increases the importance of collaborative biostatisticians (CBs) in healthcare. Adequate training and development of CBs ensure appropriate conduct of robust and meaningful research and, therefore, should be considered as a high-priority focus for biostatistics groups. Comprehensive training enhances clinical and translational research by facilitating more productive and efficient collaborations. While many graduate programs in Biostatistics and Epidemiology include training in research collaboration, it is often limited in scope and duration. Therefore, additional training is often required once a CB is hired into a full-time position. This article presents a comprehensive CB training strategy that can be adapted to any collaborative biostatistics group. This strategy follows a roadmap of the biostatistics collaboration process, which is also presented. A TIE approach (Teach the necessary skills, monitor the Implementation of these skills, and Evaluate the proficiency of these skills) was developed to support the adoption of key principles. The training strategy also incorporates a "train the trainer" approach to enable CBs who have successfully completed training to train new staff or faculty.

Modeling the impact of preplanned dose titration on delayed response.

Dose titration becomes more and more common in improving drug tolerability as well as determining individualized treatment doses, thereby maximizing the benefit to patients. Dose titration starting from a lower dose and gradually increasing to a higher dose enables improved tolerability in patients as the human body may gradually adapt to adverse gastrointestinal effects. Current statistical analyses mostly focus on the outcome at the end-of-study follow-up without considering the longitudinal impact of dose titration on the outcome. Better understanding of the dynamic effect of dose titration over time is important in early-phase clinical development as it could allow to model the longitudinal trend and predict the longer term outcome more accurately. We propose a parametric model with two empirical methods of modeling the error terms for a continuous outcome with dose titrations. Simulations show that both approaches of modeling the error terms work well. We applied this method to analyze data from a few clinical studies and achieved satisfactory results.

Extending the classification approach for comparing two active treatment arms to binary and time-to-event outcomes.

For regulatory purposes, in a trial comparing two active treatments, a hypothesis such as noninferiority or superiority must be prespecified even when there is little known about how they compare against each other or when the objective is simply to identify the best. In this article, we extend an alternative classification methodology, the classification approach of Qu et al. (Statistics in Medicine, 30:3488-3495), to compare two active treatments when outcomes are binary and time-to-event variables. This method based on estimation approach instead of hypothesis testing can be useful when little prior information is available on which treatment has better efficacy. The entire decision space is divided into eight distinct possible outcomes based on predefined lower and upper non-inferiority margins, and the conclusion will be drawn according to the location of the confidence interval for relative risk or hazard ratio (or its logarithm transformation). We demonstrate theoretically that this method controls the misclassification rate at the specified level. We also illustrate the method by simulations and using data from a Phase 3 first-line nonsmall cell lung cancer study.

Nonparametric covariate adjustment in estimating hazard ratios.

In randomized clinical trials with time-to-event outcomes, the hazard ratio is commonly used to quantify the treatment effect relative to a control. The Cox regression model is commonly used to adjust for relevant covariates to obtain more accurate estimates of the hazard ratio between treatment groups. However, it is well known that the treatment hazard ratio based on a covariate-adjusted Cox regression model is conditional on the specific covariates and differs from the unconditional hazard ratio that is an average across the population. Therefore, covariate-adjusted Cox models cannot be used when the unconditional inference is desired. In addition, the covariate-adjusted Cox model requires the relatively strong assumption of proportional hazards for each covariate. To overcome these challenges, a nonparametric randomization-based analysis of covariance method was proposed to estimate the covariate-adjusted hazard ratios for multivariate time-to-event outcomes. However, empirical evaluations of the performance (power and type I error rate) of the method have not been studied. Although the method is derived for multivariate situations, for most registration trials, the primary endpoint is a univariate outcome. Therefore, this approach is applied to univariate outcomes, and performance is evaluated through a simulation study in this paper. Stratified analysis is also investigated. As an illustration of the method, we also apply the covariate-adjusted and unadjusted analyses to an oncology trial.

To adjust or not to adjust for baseline when analyzing repeated binary responses? The case of complete data when treatment comparison at study end is of interest.

The benefits of adjusting for baseline covariates are not as straightforward with repeated binary responses as with continuous response variables. Therefore, in this study, we compared different methods for analyzing repeated binary data through simulations when the outcome at the study endpoint is of interest. Methods compared included chi-square, Fisher's exact test, covariate adjusted/unadjusted logistic regression (Adj.logit/Unadj.logit), covariate adjusted/unadjusted generalized estimating equations (Adj.GEE/Unadj.GEE), covariate adjusted/unadjusted generalized linear mixed model (Adj.GLMM/Unadj.GLMM). All these methods preserved the type I error close to the nominal level. Covariate adjusted methods improved power compared with the unadjusted methods because of the increased treatment effect estimates, especially when the correlation between the baseline and outcome was strong, even though there was an apparent increase in standard errors. Results of the Chi-squared test were identical to those for the unadjusted logistic regression. Fisher's exact test was the most conservative test regarding the type I error rate and also with the lowest power. Without missing data, there was no gain in using a repeated measures approach over a simple logistic regression at the final time point. Analysis of results from five phase III diabetes trials of the same compound was consistent with the simulation findings. Therefore, covariate adjusted analysis is recommended for repeated binary data when the study endpoint is of interest.

Combined CRVO with CRAO in a patient with protein C deficiency.

PURPOSE: To study the underlying factors in a young patient having central retinal venous obstruction with central retinal arterial obstruction and its effects on visual outcome. METHODS: A 33-year-old man presented with a complaint of sudden loss of vision in his right eye. A complete ophthalmic evaluation with fundus angiography showed combined central retinal venous obstruction with central retinal arterial obstruction. Detailed investigations revealed reduced (20%) functional assay of protein C suggestive of the diagnosis of severe Type II protein C deficiency with Factor V Leiden mutation. This ocular event was the first sign of the underlying disorder. RESULTS: The ophthalmic event rapidly progressed to no perception of light. Anticoagulant therapy was started to prevent life-threatening systemic complications. Despite antiglaucoma medications, intraocular pressure was high and cyclocryotherapy was advised. CONCLUSION: Retinal vascular disorders in the younger population have different underlying risk factors than in the older age group. Clinical presentation, severity, and management also depend on causative factors. Combined retinal artery and vein occlusion is very uncommon. Combined cases have mostly been attributed to rheological causes like thrombophilia, vessel wall inflammation, and mechanical compression. Protein C deficiency has mainly been linked to vein occlusions and there is no reference to protein C deficiency as a cause of combined central retinal venous obstruction and central retinal arterial obstruction. Combined cases usually present with severe visual loss and have rapid progression. Young patients having such presentation should be thoroughly evaluated to diagnose underlying factors and initiate appropriate management at the earliest.

Aditya Jyot-Diabetic Retinopathy in Urban Mumbai Slums Study (AJ-DRUMSS): study design and methodology - report 1.

PURPOSE: To describe the methodology of Aditya Jyot-Diabetic Retinopathy in Urban Mumbai Slums Study (AJ-DRUMSS), which was designed (i) to estimate the prevalence of diabetic retinopathy (DR) in a general population, (ii) to study the risk factors associated with DR in those with type 2 diabetes mellitus (DM), and (iii) to create awareness for early detection and develop timely interventional management for DR. METHODS: AJ-DRUMSS is an ongoing population-based cross sectional study conducted in seven wards of slums in Mumbai, India, wherein eligible subjects from the general population were screened for DR and profiled for their demographic, social and biochemical parameters to study the associations of these factors. RESULTS: To date, nearly 54,000 households have been enumerated for both awareness and DR prevalence in five study areas (out of seven) during 17 awareness campaigns and 78 DR screening camps. Of these, 4295 households were included in AJ-DRUMSS. Nearly 15,000 camp subjects (including subjects from awareness-focused areas who also turned up for the screening camps) were screened from the total enumerated households, of which 16.1% were diagnosed with type 2 DM. A total of 14.5% of these had evidence of DR and 3.5% had sight-threatening DR. CONCLUSIONS: A detailed study design of AJ-DRUMSS is described. In the screening camps nearly 3.5% of the diabetic population had sight-threatening DR, which needed an active interventional strategy. This study will help in formulating efficient eye care policies, making optimum use of available resources, reorienting healthcare providers and the ignorant within the population regarding the need for periodic ophthalmic surveillance and timely intervention.

A test of systematic coarse-graining of molecular dynamics simulations: Transport properties.

To what extent can a "bottom-up" mesoscale fluid model developed through systematic coarse-graining techniques recover the physical properties of a molecular scale system? In a previous paper [C.-C. Fu, P. M. Kulkarni, M. S. Shell, and L. G. Leal, J. Chem. Phys. 137, 164106 (2012)], we addressed this question for thermodynamic properties through the development of coarse-grained (CG) fluid models using modified iterative Boltzmann inversion methods that reproduce correct pair structure and pressure. In the present work we focus on the dynamic behavior. Unlike the radial distribution function and the pressure, dynamical properties such as the self-diffusion coefficient and viscosity in a CG model cannot be matched during coarse-graining by modifying the pair interaction. Instead, removed degrees of freedom require a modification of the equations of motion to simulate their implicit effects on dynamics. A simple but approximate approach is to introduce a friction coefficient, γ, and random forces for the remaining degrees of freedom, in which case γ becomes an additional parameter in the coarse-grained model that can be tuned. We consider the non-Galilean-invariant Langevin and the Galilean-invariant dissipative particle dynamics (DPD) thermostats with CG systems in which we can systematically tune the fraction φ of removed degrees of freedom. Between these two choices, only DPD allows both the viscosity and diffusivity to match a reference Lennard-Jones liquid with a single value of γ for each degree of coarse-graining φ. This friction constant is robust to the pressure correction imposed on the effective CG potential, increases approximately linearly with φ, and also depends on the interaction cutoff length, rcut, of the pair interaction potential. Importantly, we show that the diffusion constant and viscosity are constrained by a simple scaling law that leads to a specific choice of DPD friction coefficient for a given degree of coarse-graining. Moreover, we find that the pair interaction distance cutoffs used for DPD random and dissipative forces should be considered separately from that of the conservative interaction potential.

Intraocular cysts of toxoplasma gondii in patients with necrotizing retinitis following periocular/intraocular triamcinolone injection.

PURPOSE: To report the detection of Toxoplasma gondii cysts in intraocular aspirates of patients with necrotizing retinitis following periocular/intraocular corticosteroid injection. DESIGN: Case report. METHODS: Two patients (2 eyes) with widespread necrotizing retinitis in a steroid-exposed eye posed a diagnostic challenge and underwent pars plana vitrectomy (PPV). Intraocular samples (vitreous fluid, retinal tissue, and subretinal aspirate in case 1, and vitreous fluid in case 2) were subjected to cytological examination. RESULTS: The subretinal aspirate (case 1) revealed encysted bradyzoites of Toxoplasma gondii. Vitreous fluid (case 2) tested positive for anti-toxoplasma antibodies and the smear showed encysted forms of Toxoplasma gondii on cytology. CONCLUSION. Toxoplasma gondii cysts were detected in eyes with necrotizing retinitis that developed secondary to injudicious use of corticosteroids.

Multiscale modeling with smoothed dissipative particle dynamics.

In this work, we consider two issues related to the use of Smoothed Dissipative Particle Dynamics (SDPD) as an intermediate mesoscale model in a multiscale scheme for solution of flow problems when there are local parts of a macroscopic domain that require molecular resolution. The first is to demonstrate that SDPD with different levels of resolution can accurately represent the fluid properties from the continuum scale all the way to the molecular scale. Specifically, while the thermodynamic quantities such as temperature, pressure, and average density remain scale-invariant, we demonstrate that the dynamic properties are quantitatively consistent with an all-atom Lennard-Jones reference system when the SDPD resolution approaches the atomistic scale. This supports the idea that SDPD can serve as a natural bridge between molecular and continuum descriptions. In the second part, a simple multiscale methodology is proposed within the SDPD framework that allows several levels of resolution within a single domain. Each particle is characterized by a unique physical length scale called the smoothing length, which is inversely related to the local number density and can change on-the-fly. This multiscale methodology is shown to accurately reproduce fluid properties for the simple problem of steady and transient shear flow.

Sensitivity in statistical evaluation of biomarkers.

In clinical research and practice, biomarkers help in understanding disease progression and are useful when monitoring patients and making treatment decisions. Correct quantification of biomarkers in predicting the clinical outcome is important for better treatment of individual patients. Despite the rich literature in statistical validation, application of these validation methods in real data is not well studied. Specifically, the question is whether the change in a biomarker or the actual assessed value of the biomarker should be used. Contrary to most published papers in which the actual value of the biomarker is used, we showed through theory, simulation, and an example that it is more appropriate to use the change or the actual value in the marker with adjustment for the baseline value in evaluating a marker.

Evaluating and utilizing probability of study success in clinical development.

BACKGROUND: Drug development has become increasingly costly, lengthy, and risky. The call for better decision making in research and development has never been stronger. Analytic tools that utilize available data can inform decision makers of the risks and benefits of various decisions, which could lead to better and more informed decisions. PURPOSE: Through some real oncology examples, we will demonstrate how using available data to analytically evaluate probability of study success (PrSS) can lead to better decisions in clinical development. METHODS: The predictive power, or average conditional power, is used to quantify the PrSS. To calculate the probability, we follow a general two-step process: (1) use Bayesian modeling and appropriate assumptions to synthesize relevant data to derive the distribution of treatment effect and (2) evaluate the PrSS analytically or via trial simulation. RESULTS: We applied the procedure to several compounds in our oncology pipeline. The analysis informed decision making where PrSS was an important factor to consider. LIMITATIONS: When modeling the treatment effect, we made certain assumptions, including how two drugs work together and exchangeable treatment effects across studies. Those assumptions are reasonable for our specific situations but may not generalize well. CONCLUSIONS: From our experience, PrSS based on available data can help decision making in drug development, particularly the Go/No-Go decision after the proof of concept trial is completed. When applicable, we recommend this evaluation be regularly done in addition to the routine data analysis for clinical trials.

Joint modeling of progression-free survival and overall survival by a Bayesian normal induced copula estimation model.

In cancer clinical trials, in addition to time to death (i.e., overall survival), progression-related measurements such as progression-free survival and time to progression are also commonly used to evaluate treatment efficacy. It is of scientific interest and importance to understand the correlations among these measurements. In this paper, we propose a Bayesian semi-competing risks approach to jointly model progression-related measurements and overall survival. This new model is referred to as the NICE model, which stands for the normal induced copula estimation model. Correlation among these variables can be directly derived from the joint model. In addition, when correlation exists, simulation shows that the joint model is able to borrow strength from correlated measurements, and as a consequence the NICE model improves inference on both variables. The proposed model is in a Bayesian framework that enables us to use it in various Bayesian contexts, such as Bayesian adaptive design and using posterior predictive samples to simulate future trials. We conducted simulation studies to demonstrate properties of the NICE model and applied this method to a data set from chemotherapy-naive patients with extensive-stage small-cell lung cancer.

A test of systematic coarse-graining of molecular dynamics simulations: thermodynamic properties.

Coarse-graining (CG) techniques have recently attracted great interest for providing descriptions at a mesoscopic level of resolution that preserve fluid thermodynamic and transport behaviors with a reduced number of degrees of freedom and hence less computational effort. One fundamental question arises: how well and to what extent can a "bottom-up" developed mesoscale model recover the physical properties of a molecular scale system? To answer this question, we explore systematically the properties of a CG model that is developed to represent an intermediate mesoscale model between the atomistic and continuum scales. This CG model aims to reduce the computational cost relative to a full atomistic simulation, and we assess to what extent it is possible to preserve both the thermodynamic and transport properties of an underlying reference all-atom Lennard-Jones (LJ) system. In this paper, only the thermodynamic properties are considered in detail. The transport properties will be examined in subsequent work. To coarse-grain, we first use the iterative Boltzmann inversion (IBI) to determine a CG potential for a (1-φ)N mesoscale particle system, where φ is the degree of coarse-graining, so as to reproduce the radial distribution function (RDF) of an N atomic particle system. Even though the uniqueness theorem guarantees a one to one relationship between the RDF and an effective pairwise potential, we find that RDFs are insensitive to the long-range part of the IBI-determined potentials, which provides some significant flexibility in further matching other properties. We then propose a reformulation of IBI as a robust minimization procedure that enables simultaneous matching of the RDF and the fluid pressure. We find that this new method mainly changes the attractive tail region of the CG potentials, and it improves the isothermal compressibility relative to pure IBI. We also find that there are optimal interaction cutoff lengths for the CG system, as a function of φ, that are required to attain an adequate potential while maintaining computational speedup. To demonstrate the universality of the method, we test a range of state points for the LJ liquid as well as several LJ chain fluids.

Choroidal metastasis as a presenting manifestation of lung cancer: a report of 3 cases and systematic review of the literature.

Symptomatic choroidal metastasis is a rare presenting manifestation of lung cancer. We describe here 3 patients with non-squamous non-small cell lung cancer who presented with choroidal metastasis and who were diagnosed and treated by the authors. We performed a systematic literature review of the previously reported patients with choroidal metastasis from lung cancer in the English-language literature. We excluded case series lacking individual patient data and identified 75 patients. In 23 of these patients, choroidal metastasis was not the presenting manifestation of lung cancer. Therefore, we included 55 patients (3 index and 52 previously reported) in the analysis. We present the demographic profile, histology, disease stage, ocular and lung lesions, diagnostic and treatment (systemic and ocular) modalities, and treatment outcomes. The majority of patients were male (67.3%) and were current or ex-smokers (78.3%); the mean age was 55.1 (standard deviation 11.2) years. Adenocarcinoma (n = 23) was the most common histologic type followed by squamous (n = 11) and small cell (n = 8). Left eye (n = 32) involvement was more common than right eye (n = 19) or bilateral (n = 4). Among patients for whom the location of primary lesion was specified, the left upper lobe (n = 13) was the most common site. The most common diagnostic modalities were bronchoscopic lung biopsy (n = 15) and enucleation (n = 13), while the liver (30.9%) was the most common extraocular metastatic site identified. Systemic chemotherapy was given in 56.4% of cases, and disease progression was the most common outcome among evaluable patients. Ocular treatment modalities included radiation (n = 23), enucleation (n = 14), and systemic steroids (n = 8). Regression of choroidal metastases with treatment was observed in 66.7% of patients who did not undergo enucleation as the primary treatment modality. Of the 3 index patients, 2 each received pemetrexed-cisplatin (as first-line therapy), gefitinib or erlotinib (as second- or third-line therapy), and intravitreal bevacizumab; and 1 patient received systemic bevacizumab. Two patients had partial response radiologically with systemic treatment, and all 3 patients had regression of choroidal metastases with ocular treatment. Recommendations regarding systemic and local (ocular) management of patients with choroidal metastasis as the presenting manifestation of lung cancer are provided.

Usefulness of F-18 FDG PET/CT as one-stop-shop imaging modality for diagnosis of occult primary and estimation of disease burden in patients with intraocular masses.

Metastases are the most common intraocular malignancy. Determination of the primary site and entire burden of disease as a part of initial staging is important in choosing the most appropriate management strategy. F-18 FDG PET/CT can demonstrate primary neoplasms and the whole-body total burden of disease. We present 2 cases of intraocular metastases from lung carcinoma where F-18 FDG PET/CT was useful as one-stop-shop imaging modality in detection of the primary and in assessment of total disease burden.

Wandering Ozurdex(®) implant.

PURPOSE: To report the behavior of intravitreal Ozurdex(®) implant in eyes with post-lensectomy-vitrectomy (PLV) aphakia. METHODS: Retrospective chart review of three eyes with PLV aphakia (three patients with uveitis) who received intravitreal injection of Ozurdex(®) for cystoid macular edema (one eye), persistent inflammation (one eye), and ocular hypotony (one eye). Final outcome was assessed in terms of effectiveness, stability, and tolerance of the implant. RESULTS: Following the implant, an initial improvement was seen in all the three eyes. However, the implant migrated into the anterior chamber (AC) at 1 week in two eyes and at 5 weeks in one eye, and wandered between the AC and vitreous cavity with changing postures of the patient. Two eyes developed corneal edema, of which one eye underwent implant removal from the AC. CONCLUSION: Ozurdex(®) implant should be contraindicated in eyes with PLV aphakia to avoid its deleterious effect on the corneal endothelium.