Nichole E. Carlson

Purpose of review: Pulmonary sarcoidosis is characterized by marked radiologic heterogeneity and limited reproducibility of visual high-resolution computed tomography (HRCT) assessment, which together restrict standardization, phenotyping, and prognostication. This review examines the potential that quantitative and artificial intelligence-based approaches offer in solving these challenges. It then offers key insights into the solutions needed to fully realize the value of HRCT imaging in pulmonary sarcoidosis. Recent findings: Early quantitative and artificial intelligence-driven studies demonstrate that HRCT images can be transformed into objective, reproducible numerical representations that capture disease patterning beyond conventional visual interpretation. These approaches show promise for clinically relevant and standardized assessment of pulmonary involvement. However, in sarcoidosis, existing work remains largely preliminary and limited by small sample sizes, single center designs, technical heterogeneity, and unreliable ground truth imaging labels. Recent studies also highlight opportunities for alternative quantitative strategies that may be better suited to the data constraints of rare diseases. Summary: Quantitative HRCT analysis offers a compelling framework for advancing imaging-based assessment in pulmonary sarcoidosis, but meaningful progress will require large multicenter cohorts and more objective nonimaging outcomes that move beyond subjective visual assessment. Quantitative imaging may then help reposition HRCT as a reproducible biomarker platform for research and clinical care.

Introduction: Active post‐marketing surveillance of prescribing behavior of high‐risk drugs may provide early warning of unforeseen issues in a population, yet analysis approaches for surveillance using real‐world data are underdeveloped. This paper evaluates a modified statistical process control (SPC) method for surveillance of risk minimization measures derived from administrative claims data. The approach detects changes in population‐level prescribing behaviors and informs investigators of the timing and nature of any detected changes. Methods: We investigated prescription drug claims for tapentadol extended release (ER) from the Colorado All‐Payers Claims Database (2012–2019). The cohort was 2702 unique patients receiving their first prescriptions of tapentadol ER (an opiate with an FDA Risk Evaluation and Mitigation Strategy). The risk minimization measures were the prescribing rate and the proportion of prescriptions with appropriate dosing. A statistical model was fitted to data from January–December 2012 to establish a stable baseline and then updated biweekly through June 2019. We conducted surveillance with our modified SPC method and a classical SPC method, controlling the false alarm rate to 0.005 for each, and compared how detections aligned with external policy actions. Results: Our method detected two periods of unusual prescribing behavior beginning in March 2015 ( p < 0.005) and April 2016 ( p < 0.005). The classical method detected a single period in October 2016 ( p < 0.005). Our detections aligned with risk minimization activities in Colorado; the classical method aligned with only the second activity, 6 months later. Conclusion: Our modified SPC method, which monitors model misspecification rather than raw prescribing data, detects more periods of changing behavior that better temporally align with risk minimization actions. This tool may be useful for regulatory agencies to independently monitor for emerging risks and prescribing trends to complement REMS assessments.

Background: Although cognitive decline in Parkinson disease (PD) is common, there are currently no easily accessible clinical tools to predict the risk of cognitive decline in PD. Objective: To refine a tool to predict cognitive decline in PD and identify potentially modifiable factors. Methods: We used neuropsychological scores from the Parkinson’s Progression Markers Initiative de novo PD cohort to calculate a composite cognitive score (CCS) for each participant. Participants with a decrease of ≥0.5 SD in CCS between baseline and Year 4 were considered to have cognitive decline (PD-Decline), as predicted by conditional linear mixed modeling to account for nonignorable dropouts. We identified risk factors for PD-Decline using logistic regression and developed the PD Risk Estimator for Decline In Cognition Tool (PREDICT) scoring system using the regression β coefficients. The selected risk factors and the discriminative ability of the PREDICT scoring system were evaluated using bootstrapped samples. Results: Participants with excessive daytime sleepiness (odds ratio [OR]=2.8; 95% CI [1.3, 6.0]), moderate-severe motor symptoms (OR=2.6; 95% CI [1.2, 5.4]), and fewer years of formal education (OR=2.2; 95% CI [1.1, 4.1]) had significantly increased odds of being categorized as PD-decline. Possible PREDICT scores ranged from 0 to 3.5. The average bootstrap sensitivity and specificity were 59.9% and 43.5%, respectively. Conclusion: We identified excessive daytime sleepiness and moderate-severe motor symptoms as potentially modifiable factors. In the future, PREDICT may offer wider opportunities for personalized risk assessment and potential risk reduction.

Chest high resolution computed tomography (HRCT) is increasingly used in clinical practice for sarcoidosis. Visual assessment (VAS) remains standard even with its high inter- and intra-rater variation. Radiomics may be a less variable assessment. We developed radiomic profiles of HRCT and mapped them to radiologic, functional, and patient reported outcomes. Radiomics were calculated on HRCT for both lungs from sarcoidosis cases enrolled in the Genomic Research in Alpha-1 Antitrypsin Deficiency and Sarcoidosis (GRADS) study (N = 320). Cases were clustered using radiomics profiles via robust and sparse K-means. Differences in VAS by cluster were identified using chi-squared tests. Linear regression investigated how pulmonary function tests (PFT) and patient reported outcomes (PRO) differed between clusters with and without adjustment for other radiologic quantification. Sensitivity analyses were used for validation. Clustering produced four patient groups that were associated with, but did not mimic, Scadding stage and Oberstein score (p < 0.001). One cluster had markedly few abnormalities. Another cluster had consistently more abnormalities and fibrosis. Clusters explained 10–15% of PFT variation and significantly predicted PFT even after accounting for other radiologic findings (p < 0.001). Associations with PRO were less consistent. These findings provide early-stage evidence that radiomics may be a useful component in identifying disease phenotypes.

Objective. Sarcoidosis is a granulomatous disease affecting the lungs in over 90% of patients. Qualitative assessment of chest CT by radiologists is standard clinical practice and reliable quantification of disease from CT would support ongoing efforts to identify sarcoidosis phenotypes. Standard imaging feature engineering techniques such as radiomics suffer from extreme sensitivity to image acquisition and processing, potentially impeding generalizability of research to clinical populations. In this work, we instead investigate approaches to engineering variogram-based features with the intent to identify a robust, generalizable pipeline for image quantification in the study of sarcoidosis. Approach. For a cohort of more than 300 individuals with sarcoidosis, we investigated 24 feature engineering pipelines differing by decisions for image registration to a template lung, empirical and model variogram estimation methods, and feature harmonization for CT scanner model, and subsequently 48 sets of phenotypes produced through unsupervised clustering. We then assessed sensitivity of engineered features, phenotypes produced through unsupervised clustering, and sarcoidosis disease signal strength to pipeline. Main results. We found that variogram features had low to mild association with scanner model and associations were reduced by image registration. For each feature type, features were also typically robust to all pipeline decisions except image registration. Strength of disease signal as measured by association with pulmonary function testing and some radiologist visual assessments was strong (optimistic AUC ≈ 0.9, p ≪ 0.0001 in models for architectural distortion, conglomerate mass, fibrotic abnormality, and traction bronchiectasis) and fairly consistent across engineering approaches regardless of registration and harmonization for CT scanner. Significance. Variogram-based features appear to be a suitable approach to image quantification in support of generalizable research in pulmonary sarcoidosis.

Objective: Sarcoidosis is a granulomatous disease affecting the lungs in over 90% of patients. Qualitative assessment of chest CT by radiologists is standard clinical practice and reliable quantification of disease from CT would support ongoing efforts to identify sarcoidosis phenotypes. Standard imaging feature engineering techniques such as radiomics suffer from extreme sensitivity to image acquisition and processing, potentially impeding generalizability of research to clinical populations. In this work, we instead investigate approaches to engineering variogram-based features with the intent to identify a robust, generalizable pipeline for image quantification in the study of sarcoidosis. Approach: For a cohort of more than 300 individuals with sarcoidosis, we investigated 24 feature engineering pipelines differing by decisions for image registration to a template lung, empirical and model variogram estimation methods, and feature harmonization for CT scanner model, and subsequently 48 sets of phenotypes produced through unsupervised clustering. We then assessed sensitivity of engineered features, phenotypes produced through unsupervised clustering, and sarcoidosis disease signal strength to pipeline. Main results: We found that variogram features had low to mild association with scanner model and associations were reduced by image registration. For each feature type, features were also typically robust to all pipeline decisions except image registration. Strength of disease signal as measured by association with pulmonary function testing and some radiologist visual assessments was strong (optimistic AUC ≈ 0.9, p ≪ 0.0001 in models for architectural distortion, conglomerate mass, fibrotic abnormality, and traction bronchiectasis) and fairly consistent across engineering approaches regardless of registration and harmonization for CT scanner. Significance: Variogram-based features appear to be a suitable approach to image quantification in support of generalizable research in pulmonary sarcoidosis.

Data‐intensive research continues to expand with the goal of improving healthcare delivery, clinical decision‐making, and patient outcomes. Quantitative scientists, such as biostatisticians, epidemiologists, and informaticists, are tasked with turning data into health knowledge. In academic health centres, quantitative scientists are critical to the missions of biomedical discovery and improvement of health. Many academic health centres have developed centralized Quantitative Science Units which foster dual goals of professional development of quantitative scientists and producing high quality, reproducible domain research. Such units then develop teams of quantitative scientists who can collaborate with researchers. However, existing literature does not provide guidance on how such teams are formed or how to manage and sustain them. Leaders of Quantitative Science Units across six institutions formed a working group to examine common practices and tools that can serve as best practices for Quantitative Science Units that wish to achieve these dual goals through building long‐term partnerships with researchers. The results of this working group are presented to provide tools and guidance for Quantitative Science Units challenged with developing, managing, and evaluating Quantitative Science Teams. This guidance aims to help Quantitative Science Units effectively participate in and enhance the research that is conducted throughout the academic health centre—shaping their resources to fit evolving research needs.

Background: Bayesian statistical approaches are extensively used in new statistical methods but have not been adopted at the same rate in clinical and translational (C&T) research. The goal of this paper is to accelerate the transition of new methods into practice by improving the C&T researcher’s ability to gain confidence in interpreting and implementing Bayesian analyses. Methods: We developed a Bayesian data analysis plan and implemented that plan for a two-arm clinical trial comparing the effectiveness of a new opioid in reducing time to discharge from the post-operative anesthesia unit and nerve block usage in surgery. Through this application, we offer a brief tutorial on Bayesian methods and exhibit how to apply four Bayesian statistical packages from STATA, SAS, and RStan to conduct linear and logistic regression analyses in clinical research. Results: The analysis results in our application were robust to statistical package and consistent across a wide range of prior distributions. STATA was the most approachable package for linear regression but was more limited in the models that could be fitted and easily summarized. SAS and R offered more straightforward documentation and data management for the posteriors. They also offered direct programming of the likelihood making them more easily extendable to complex problems. Conclusion: Bayesian analysis is now accessible to a broad range of data analysts and should be considered in more C&T research analyses. This will allow C&T research teams the ability to adopt and interpret Bayesian methodology in more complex problems where Bayesian approaches are often needed.

Background: Neutralizing monoclonal antibodies (mAbs) are highly effective in reducing hospitalization and mortality among early symptomatic COVID-19 patients in clinical trials and real-world data. While resistance to some mAbs has since emerged among new variants, characteristics associated with treatment failure of mAbs remain unknown.Methods: This multicenter, observational cohort study included patients with COVID-19 who received mAb treatment between November 20, 2020, and December 9, 2021. We utilized electronic health records from a statewide health system plus state-level vaccine and mortality data. The primary outcome was mAb treatment failure, defined as hospitalization or death within 28 days of a positive SARS-CoV-2 test.Results: COVID-19 mAb was administered to 7406 patients. Hospitalization within 28 days of positive SARS-CoV-2 test occurred in 258 (3.5%) of all patients who received mAb treatment. Ten patients (0.1%) died within 28 days, and all but one were hospitalized prior to death. Characteristics associated with treatment failure included having two or more comorbidities excluding obesity and immunocompromised status (adjusted odds ratio [OR] 3.71, 95% confidence interval [CI] 2.52–5.56), lack of SARS-CoV-2 vaccination (OR 2.73, 95% CI 2.01–3.77), non-Hispanic black race/ethnicity (OR 2.21, 95% CI 1.20–3.82), obesity (OR 1.79, 95% CI 1.36–2.34), one comorbidity (OR 1.68, 95% CI 1.11–2.57), age ≥ 65 years (OR 1.62, 95% CI 1.13–2.35), and male sex (OR 1.56, 95% CI 1.21–2.02). Immunocompromised status (none, mild, or moderate/severe), pandemic phase, and type of mAb received were not associated with treatment failure (all p > 0.05).Conclusions: Comorbidities, lack of prior SARS-CoV-2 vaccination, non-Hispanic black race/ethnicity, obesity, age ≥ 65 years, and male sex are associated with treatment failure of mAbs.