Design considerations for two-stage enrichment clinical trials

When there is a predictive biomarker, enrichment can focus the clinical trial on a benefiting subpopulation. We describe a two-stage enrichment design, in which the first stage is designed to efficiently estimate a threshold and the second stage is a “phase III-like” trial on the enriched population. The goal of this paper is to explore design issues: sample size in Stages 1 and 2, and re-estimation of the Stage 2 sample size following Stage 1. By treating these as separate trials, we can gain insight into how the predictive nature of the biomarker specifically impacts the sample size. We also show that failure to adequately estimate the threshold can have disastrous consequences in the second stage. While any bivariate model could be used, we assume a continuous outcome and continuous biomarker, described by a bivariate normal model. The correlation coefficient between the outcome and biomarker is the key to understanding the behavior of the design, both for predictive and prognostic biomarkers. Through a series of simulations we illustrate the impact of model misspecification, consequences of poor threshold estimation, and requisite sample sizes that depend on the predictive nature of the biomarker. Such insight should be helpful in understanding and designing enrichment trials.     

Reinforcement learning strategies for clinical trials in nonsmall cell lung cancer

Typical regimens for advanced metastatic stage IIIB/IV nonsmall cell lung cancer (NSCLC) consist of multiple lines of treatment. We present an adaptive reinforcement learning approach to discover optimal individualized treatment regimens from a specially designed clinical trial (a "clinical reinforcement trial") of an experimental treatment for patients with advanced NSCLC who have not been treated previously with systemic therapy. In addition to the complexity of the problem of selecting optimal compounds for first- and second-line treatments based on prognostic factors, another primary goal is to determine the optimal time to initiate second-line therapy, either immediately or delayed after induction therapy, yielding the longest overall survival time. A reinforcement learning method called Q-learning is utilized, which involves learning an optimal regimen from patient data generated from the clinical reinforcement trial. Approximating the Q-function with time-indexed parameters can be achieved by using a modification of support vector regression that can utilize censored data. Within this framework, a simulation study shows that the procedure can extract optimal regimens for two lines of treatment directly from clinical data without prior knowledge of the treatment effect mechanism. In addition, we demonstrate that the design reliably selects the best initial time for second-line therapy while taking into account the heterogeneity of NSCLC across patients.     

Screening for lung cancer.

The survival from bronchogenic carcinoma is highly dependent upon stage at the time of treatment. This is particularly true for squamous cell carcinoma, adenocarcinoma, and large cell carcinoma, but holds true for small cell carcinoma as well. The problem presented to the medical profession has been to find a practical means of detecting lung cancer while it is still at an early stage. Three studies in progress have indicated that a larger proportion of the patients may be found to have early stage lung cancer when screened with a combination of chest X-rays and sputum cytology. However, the detection of these early stage cases has not yet been translated into an improvement in the overall mortality rate from lung cancer.     

Screening for occult lung cancer.

A pilot screening program for the early detection of lung cancer was carried out in Saskatchewan in 1968 using chest roentgenography and cytologic examination of sputum samples. The yield from 23 000 men aged 40 years and over was only 10 cases. Nine of the men had advanced disease. One had occult lung cancer. A period of 31 months elapsed between the discovery of malignant cells in this patient''s sputum and roentgenographic localization of the tumour. Following pneumonectomy he has survived with no discernible residual or metastatic tumour for 12 years. The morphologic changes in the resected lung provided a basis for discussing the preclinical phase of squamous cancer of the lung, the treatment of occult cancer and multicentric primary pulmonary tumours. The survey would have been more successful with a narrower target group and more frequent examination.     

Mouse models of Huntington's disease and methodological considerations for therapeutic trials

Huntington's disease (HD) is an autosomal dominant, progressive, and fatal neurodegenerative disorder caused by an expanded polyglutamine cytosine–adenine–guanine repeat in the gene coding for the protein huntingtin. Despite great progress, a direct causative pathway from the HD gene mutation to neuronal dysfunction and death has not yet been established. One important advance in understanding the pathogenic mechanisms of this disease has been the development of multiple murine models that replicate many of the clinical, neuropathological, and molecular events in HD patients. These models have played an important role in providing accurate and experimentally accessible systems to study multiple aspects of disease pathogenesis and to test potential therapeutic treatment strategies. Understanding how disease processes interrelate has become important in identifying a pharmacotherapy in HD and in the design of clinical trials. A review of the current state of HD mouse models and their successes in elucidating disease pathogenesis are discussed. There is no clinically proven treatment for HD that can halt or ameliorate the inexorable disease progression. As such, a guide to assessing studies in mouse models and salient issues related to translation from mice to humans are included.