RESUMEN
In clinical trials for diseases with very small patient populations, trial investigators may encounter recruitment difficulties. It can be challenging to conduct clinical trials with enough power to detect a treatment effect, and randomization may not be feasible due to timeline, budget, and ethical concerns. To bring breakthrough therapies to the market quickly, it is important to come up with efficient approaches to utilizing individual patient data through improved study design and sound statistical methods. Emerging topics in this area include the use of Bayesian approaches to flexibly incorporate prior information into the current clinical trials, the use of historical controls to efficiently conduct trials that will reduce the number of subjects recruited and ease ethical considerations, and the use of innovative study designs, such as a platform design, to improve the efficiency and speed of the medical therapy development progress. In this paper, we describe three scenarios which highlight some of the challenges encountered in small-sized clinical trial development and provide potential statistical approaches to overcome the aforementioned challenges.
Asunto(s)
Proyectos de Investigación , Teorema de Bayes , HumanosRESUMEN
Recruitment of patients in concurrent control arms can be very challenging for clinical trials for pediatric and rare diseases. Innovative approaches, such as platform trial designs, including shared internal control arm(s), can potentially reduce the needed sample size, improving the efficiency and speed of the drug development program. Furthermore, historical borrowing, which involves leveraging information from control arms in previous relevant clinical trials, may further enhance a clinical trial's efficiency. In this paper, we discuss platform trials highlighting their advantages and limitations. We then compare various strategies that borrow historical data or information, such as pooling data from different studies, analyzing data from studies separately, test-then-pool, dynamic pooling, and Bayesian hierarchical modeling, which focuses on the meta-analytic-predictive (MAP) prior. We further propose a procedure to illustrate the feasibility of utilizing historical controls under a platform setting and describe the statistical performance of our method via simulations.
Asunto(s)
Bases de Datos Factuales/estadística & datos numéricos , Ensayos Clínicos Controlados Aleatorios como Asunto/métodos , Ensayos Clínicos Controlados Aleatorios como Asunto/estadística & datos numéricos , Teorema de Bayes , Ensayos Clínicos como Asunto/métodos , Ensayos Clínicos como Asunto/estadística & datos numéricos , Humanos , Modelos Estadísticos , Tamaño de la MuestraRESUMEN
BACKGROUND: The current clinical workflow for esophageal gross tumor volume (GTV) contouring relies on manual delineation with high labor costs and inter-user variability. PURPOSE: To validate the clinical applicability of a deep learning multimodality esophageal GTV contouring model, developed at one institution whereas tested at multiple institutions. MATERIALS AND METHODS: We collected 606 patients with esophageal cancer retrospectively from four institutions. Among them, 252 patients from institution 1 contained both a treatment planning CT (pCT) and a pair of diagnostic FDG-PET/CT; 354 patients from three other institutions had only pCT scans under different staging protocols or lacking PET scanners. A two-streamed deep learning model for GTV segmentation was developed using pCT and PET/CT scans of a subset (148 patients) from institution 1. This built model had the flexibility of segmenting GTVs via only pCT or pCT+PET/CT combined when available. For independent evaluation, the remaining 104 patients from institution 1 behaved as an unseen internal testing, and 354 patients from the other three institutions were used for external testing. Degrees of manual revision were further evaluated by human experts to assess the contour-editing effort. Furthermore, the deep model's performance was compared against four radiation oncologists in a multi-user study using 20 randomly chosen external patients. Contouring accuracy and time were recorded for the pre- and post-deep learning-assisted delineation process.
RESUMEN
The human lung airway is a complex inverted tree-like structure. Detailed airway measurements can be extracted from MDCT-scanned lung images, such as segmental wall thickness, airway diameter, parent-child branch angles, etc. The wealth of lung airway data provides a unique opportunity for advancing our understanding of the fundamental structure-function relationships within the lung. An important problem is to construct and identify important lung airway features in normal subjects and connect these to standardized pulmonary function test results such as FEV1%. Among other things, the problem is complicated by the fact that a particular airway feature may be an important (relevant) predictor only when it pertains to segments of certain generations. Thus, the key is an efficient, consistent method for simultaneously conducting group selection (lung airway feature types) and within-group variable selection (airway generations), i.e., bi-level selection. Here we streamline a comprehensive procedure to process the lung airway data via imputation, normalization, transformation and groupwise principal component analysis, and then adopt a new composite penalized regression approach for conducting bi-level feature selection. As a prototype of composite penalization, the proposed composite bridge regression method is shown to admit an efficient algorithm, enjoy bi-level oracle properties, and outperform several existing methods. We analyze the MDCT lung image data from a cohort of 132 subjects with normal lung function. Our results show that, lung function in terms of FEV1% is promoted by having a less dense and more homogeneous lung comprising an airway whose segments enjoy more heterogeneity in wall thicknesses, larger mean diameters, lumen areas and branch angles. These data hold the potential of defining more accurately the "normal" subject population with borderline atypical lung functions that are clearly influenced by many genetic and environmental factors.
RESUMEN
RATIONALE AND OBJECTIVES: We evaluated the role of automated quantitative computed tomography (CT) scan interpretation algorithm in detecting interstitial lung disease (ILD) and/or emphysema in a sample of elderly subjects with mild lung disease. We hypothesized that the quantification and distributions of CT attenuation values on lung CT, over a subset of Hounsfield units (HUs) range (-1000 HU, 0 HU), can differentiate early or mild disease from normal lung. MATERIALS AND METHODS: We compared the results of quantitative spiral rapid end-exhalation (functional residual capacity, FRC) and end-inhalation (total lung capacity, TLC) CT scan analyses of 52 subjects with radiographic evidence of mild fibrotic lung disease to the results of 17 normal subjects. Several CT value distributions were explored, including (1) that from the peripheral lung taken at TLC (with peels at 15 or 65 mm), (2) the ratio of (1) to that from the core of lung, and (3) the ratio of (2) to its FRC counterpart. We developed a fused-lasso logistic regression model that can automatically identify sub-intervals of -1000 HU and 0 HU over which a CT value distribution provides optimal discrimination between abnormal and normal scans. RESULTS: The fused-lasso logistic regression model based on (2) with 15-mm peel identified the relative frequency of CT values of over -1000 HU and -900 and those over -450 HU and -200 HU as a means of discriminating abnormal versus normal lung, resulting in a zero out-sample false-positive rate, and 15% false-negative rate of that was lowered to 12% by pooling information. CONCLUSIONS: We demonstrated the potential usefulness of this novel quantitative imaging analysis method in discriminating ILD and/or emphysema from normal lungs.