Predicting subgroup treatment effects for a new study: Motivations, results and learnings from running a data challenge in a pharmaceutical corporation.

We present the motivation, experience, and learnings from a data challenge conducted at a large pharmaceutical corporation on the topic of subgroup identification. The data challenge aimed at exploring approaches to subgroup identification for future clinical trials. To mimic a realistic setting, participants had access to 4 Phase III clinical trials to derive a subgroup and predict its treatment effect on a future study not accessible to challenge participants. A total of 30 teams registered for the challenge with around 100 participants, primarily from Biostatistics organization. We outline the motivation for running the challenge, the challenge rules, and logistics. Finally, we present the results of the challenge, the participant feedback as well as the learnings. We also present our view on the implications of the results on exploratory analyses related to treatment effect heterogeneity.

Perceived Barriers to Equitable Participation in the Learning Assistant Program.

To build a more diverse STEM workforce, institutions seek to increase the representation of diverse groups in faculty and mentoring positions. The Learning Assistant (LA) near-peer student support program has the potential to bring diverse students into highly visible and impactful mentoring roles early in their college careers, benefiting both LAs and students in LA-supported courses. However, the demographic characteristics of potential students interested in the LA Program and the subsequent barriers to entry have yet to be investigated. This short-term longitudinal case study revealed that even though students from historically underserved groups (HUGs) started the semester equally as likely as non-HUGs to see themselves as future chemistry LAs, an inequity developed later in the semester. Similar trends were not detected based on students' gender or age group (traditional/nontraditional). Qualitative data indicated that regardless of demographic group, the most prominent barriers to students seeing themselves as future LAs include a lack of time and self-efficacy in chemistry content knowledge/social skills. The trends observed at this diverse metropolitan research institution indicate that further research is needed to find and lower barriers for students to have the opportunity to become LAs, especially those from HUGs.

How can we make better graphs? An initiative to increase the graphical expertise and productivity of quantitative scientists.

Graphics are at the core of exploring and understanding data, communicating results and conclusions, and supporting decision-making. Increasing our graphical expertise can significantly strengthen our impact as professional statisticians and quantitative scientists. In this article, we present a concerted effort to improve the way we create graphics at Novartis. We provide our vision and guiding principles, before describing seven work packages in more detail. The actions, principles, and experiences laid out in this paper are applicable generally, also beyond drug development, which is our field of work. The purpose of this article is to share our experiences and help foster the use of good graphs in pharmaceutical statistics and beyond. A Graphics Principles "Cheat Sheet" is available online at https://graphicsprinciples.github.io/.

All that Glitters Is not Gold: Type-I Error Controlled Variable Selection from Clinical Trial Data.

Clinical trials are primarily conducted to estimate causal effects, but the data collected can also be invaluable for additional research, such as identifying prognostic measures of disease or biomarkers that predict treatment efficacy. However, these exploratory settings are prone to false discoveries (type-I errors) due to the multiple comparisons they entail. Unfortunately, many methods fail to address this issue, in part because the algorithms used are generally designed to optimize predictions and often only provide the measures used for variable selection, such as machine learning model importance scores, as a byproduct. To address the resulting unclear uncertainty in the selection sets, the knockoff framework offers a model-agnostic, robust approach to variable selection with guaranteed type-I error control. Here, we review the knockoff framework in the setting of clinical data, highlighting main considerations using simulation studies. We also extend the framework by introducing a novel knockoff generation method that addresses two main limitations of previously suggested methods relevant for clinical development settings. With this new method, we empirically obtain tighter bounds on type-I error control and gain an order of magnitude in computational efficiency in mixed data settings. We demonstrate comparable selections to those of the competing method for identifying prognostic biomarkers for C-reactive protein levels in patients with psoriatic arthritis in four clinical trials. Our work increases access to the knockoff framework for variable selection from clinical trial data. Hereby, this paper helps to address the current replicability crisis which can result in unnecessary research efforts, increased patient burden, and avoidable costs.

Initial data analysis for longitudinal studies to build a solid foundation for reproducible analysis.

Initial data analysis (IDA) is the part of the data pipeline that takes place between the end of data retrieval and the beginning of data analysis that addresses the research question. Systematic IDA and clear reporting of the IDA findings is an important step towards reproducible research. A general framework of IDA for observational studies includes data cleaning, data screening, and possible updates of pre-planned statistical analyses. Longitudinal studies, where participants are observed repeatedly over time, pose additional challenges, as they have special features that should be taken into account in the IDA steps before addressing the research question. We propose a systematic approach in longitudinal studies to examine data properties prior to conducting planned statistical analyses. In this paper we focus on the data screening element of IDA, assuming that the research aims are accompanied by an analysis plan, meta-data are well documented, and data cleaning has already been performed. IDA data screening comprises five types of explorations, covering the analysis of participation profiles over time, evaluation of missing data, presentation of univariate and multivariate descriptions, and the depiction of longitudinal aspects. Executing the IDA plan will result in an IDA report to inform data analysts about data properties and possible implications for the analysis plan-another element of the IDA framework. Our framework is illustrated focusing on hand grip strength outcome data from a data collection across several waves in a complex survey. We provide reproducible R code on a public repository, presenting a detailed data screening plan for the investigation of the average rate of age-associated decline of grip strength. With our checklist and reproducible R code we provide data analysts a framework to work with longitudinal data in an informed way, enhancing the reproducibility and validity of their work.

DMC reports in the 21st century: towards better tools for decision-making.

Data Monitoring Committees (DMCs) have the important task to protect the safety of current and future patients during the conduct of a clinical study. Unfortunately, their work is often made difficult by voluminous DMC reports that are poorly structured and difficult to digest. In this article, we suggest improved solutions. Starting from a principled approach and building upon previous proposals, we offer concrete and easily understood displays, including related computer code. While leveraging modern tools, the most important is that these displays support the DMC's workflow in answering the relevant questions of interest. We hope that the adoption of these proposals can ease the task of DMCs, and importantly, lead to better decision-making for the benefit of patients.

A deep learning approach to private data sharing of medical images using conditional generative adversarial networks (GANs).

Clinical data sharing can facilitate data-driven scientific research, allowing a broader range of questions to be addressed and thereby leading to greater understanding and innovation. However, sharing biomedical data can put sensitive personal information at risk. This is usually addressed by data anonymization, which is a slow and expensive process. An alternative to anonymization is construction of a synthetic dataset that behaves similar to the real clinical data but preserves patient privacy. As part of a collaboration between Novartis and the Oxford Big Data Institute, a synthetic dataset was generated based on images from COSENTYX® (secukinumab) ankylosing spondylitis (AS) clinical studies. An auxiliary classifier Generative Adversarial Network (ac-GAN) was trained to generate synthetic magnetic resonance images (MRIs) of vertebral units (VUs), conditioned on the VU location (cervical, thoracic and lumbar). Here, we present a method for generating a synthetic dataset and conduct an in-depth analysis on its properties along three key metrics: image fidelity, sample diversity and dataset privacy.

Rethinking clinical study data: why we should respect analysis results as data.

The development and approval of new treatments generates large volumes of results, such as summaries of efficacy and safety. However, it is commonly overlooked that analyzing clinical study data also produces data in the form of results. For example, descriptive statistics and model predictions are data. Although integrating and putting findings into context is a cornerstone of scientific work, analysis results are often neglected as a data source. Results end up stored as "data products" such as PDF documents that are not machine readable or amenable to future analyses. We propose a solution to "calculate once, use many times" by combining analysis results standards with a common data model. This analysis results data model re-frames the target of analyses from static representations of the results (e.g., tables and figures) to a data model with applications in various contexts, including knowledge discovery. Further, we provide a working proof of concept detailing how to approach standardization and construct a schema to store and query analysis results.

Effective Visual Communication for the Quantitative Scientist.

Effective visual communication is a core competency for pharmacometricians, statisticians, and, more generally, any quantitative scientist. It is essential in every step of a quantitative workflow, from scoping to execution and communicating results and conclusions. With this competency, we can better understand data and influence decisions toward appropriate actions. Without it, we can fool ourselves and others and pave the way to wrong conclusions and actions. The goal of this tutorial is to convey this competency. We posit three laws of effective visual communication for the quantitative scientist: have a clear purpose, show the data clearly, and make the message obvious. A concise "Cheat Sheet," available on https://graphicsprinciples.github.io, distills more granular recommendations for everyday practical use. Finally, these laws and recommendations are illustrated in four case studies.

Discovery of a Fluorinated Enigmol Analog with Enhanced in Vivo Pharmacokinetic and Anti-Tumor Properties.

The orally bioavailable 1-deoxy-sphingosine analog, Enigmol, has demonstrated anticancer activity in numerous in vivo settings. However, as no Enigmol analog with enhanced potency in vitro has been identified, a new strategy to improve efficacy in vivo by increasing tumor uptake was adopted. Herein, synthesis and biological evaluation of two novel fluorinated Enigmol analogs, CF3-Enigmol and CF2-Enigmol, are reported. Each analog was equipotent to Enigmol in vitro, but achieved higher plasma and tissue levels than Enigmol in vivo. Although plasma and tissue exposures were anticipated to trend with fluorine content, CF2-Enigmol absorbed into tissue at strikingly higher concentrations than CF3-Enigmol. Using mouse xenograft models of prostate cancer, we also show that CF3-Enigmol underperformed Enigmol-mediated inhibition of tumor growth and elicited systemic toxicity. By contrast, CF2-Enigmol was not systemically toxic and demonstrated significantly enhanced antitumor activity as compared to Enigmol.

Novel synthesis and biological evaluation of enigmols as therapeutic agents for treating prostate cancer.

Enigmol is a synthetic, orally active 1-deoxysphingoid base analogue that has demonstrated promising activity against prostate cancer. In these studies, the pharmacologic roles of stereochemistry and N-methylation in the structure of enigmols were examined. A novel enantioselective synthesis of all four possible 2S-diastereoisomers of enigmol (2-aminooctadecane-3,5-diols) from l-alanine is reported, which features a Liebeskind-Srogl cross-coupling reaction between l-alanine thiol ester and (E)-pentadec-1-enylboronic acid as the key step. In vitro biological evaluation of the four enigmol diastereoisomers and 2S,3S,5S-N-methylenigmol against two prostate cancer cell lines (PC-3 and LNCaP) indicates that all but one diastereomer demonstrate potent oncolytic activity. In nude mouse xenograft models of human prostate cancer, enigmol was equally effective as standard prostate cancer therapies (androgen deprivation or docetaxel), and two of the enigmol diastereomers, 2S,3S,5R-enigmol and 2S,3R,5S-enigmol, also caused statistically significant inhibition of tumor growth. A pharmacokinetic profile of enigmol and N-methylenigmol is also presented.

The efficacy of emamectin benzoate against infestations of Lepeophtheirus salmonis on farmed Atlantic salmon (Salmo salar L) in Scotland, 2002-2006.

BACKGROUND: Infestations of the parasitic copepod Lepeophtheirus salmonis, commonly referred to as sea lice, represent a major challenge to commercial salmon aquaculture. Dependence on a limited number of theraputants to control such infestations has led to concerns of reduced sensitivity in some sea lice populations. This study investigates trends in the efficacy of the in-feed treatment emamectin benzoate in Scotland, the active ingredient most widely used across all salmon producing regions. METHODOLOGY/PRINCIPAL FINDINGS: Study data were drawn from over 50 commercial Atlantic salmon farms on the west coast of Scotland between 2002 and 2006. An epi-informatics approach was adopted whereby available farm records, descriptive epidemiological summaries and statistical linear modelling methods were used to identify factors that significantly affect sea lice abundance following treatment with emamectin benzoate (SLICE(R), Schering Plough Animal Health). The results show that although sea lice infestations are reduced following the application of emamectin benzoate, not all treatments are effective. Specifically there is evidence of variation across geographical regions and a reduction in efficacy over time. CONCLUSIONS/SIGNIFICANCE: Reduced sensitivity and potential resistance to currently available medicines are constant threats to maintaining control of sea lice populations on Atlantic salmon farms. There is a need for on-going monitoring of emamectin benzoate treatment efficacy together with reasons for any apparent reduction in performance. In addition, strategic rotation of medicines should be encouraged and empirical evidence for the benefit of such strategies more fully evaluated.

Determination of free and total cortisol in plasma and urine by liquid chromatography-tandem mass spectrometry.

Cortisol is an important adrenal steroid hormone involved in the regulation of metabolic homeostasis. A new liquid chromatography-mass spectrometry/mass spectrometry (LC-MS/MS) multiple reactant monitoring (MRM) procedure for the measurement of cortisol concentration in plasma ultrafiltrate, whole plasma, and urine was developed and validated. Plasma, plasma ultrafiltrate, or urine was extracted by ethyl acetate. The extract was subjected to liquid chromatography with an Inertsil ODS-3 column with an aqueous NH4Cl (1 mM, pH 9.0):methanol mobile phase. The presence of NH4Cl in the mobile phase induced the formation of [M+Cl] in the first quadrupole at m/z 397 and 409 for cortisol and 6alpha-methylprednisolone (internal standard), respectively. In the collision cell, the complex dissociated to the neutral parent and the chloride ion at m/z 35; the latter ion was used for quantification. The calibration curve was linear from 0.5 to 100 ng/mL. The lower limit of quantification was 0.50 ng/mL and the limit of detection was 0.25 ng/mL. For quality control samples prepared in water, the intrabatch assay precision was 5.6%, 9.6%, and 9.9% at 50, 10, and 1 ng/mL, respectively. The interbatch assay precision was 4.2%, 6.3%, and 7.5% at 50, 10, and 1 ng/mL, respectively. For measurement of endogenous cortisol in plasma and urine samples, the intra-assay and interassay precision was 10.8% and 4.8% for total plasma cortisol, 13.1% and 5.2% for free plasma cortisol, 10.9% and 13.1% for cortisol protein-binding free fraction, and 8.9% and 14.4% for urine cortisol, respectively. A simple procedure of ultrafiltration coupled with the highly sensitive LC-MS/MS quantification offered a rapid and reproducible assay for plasma free cortisol, which may be useful in the assessment of adrenal function in patients, especially critically ill patients with abnormal protein binding. It may also be useful for plasma and urinary cortisol measurements in pharmacodynamic studies of adrenocorticoid response.