論文

© 2019 the Japanese Society of Clinical Pharmacology and Therapeutics(JSCPT) Clinical trials using adaptive designs(AD)are more flexible in study design and implementation compared to conventional fixed designs. In the US, the draft guidance documentËAdaptive Design Clinical Trials for Drugs and BiologicsÌwas issued by the FDA in 2010, and many studies using AD have been performed. On the other hand, the situation of practical application of AD to clinical trials in Japan has not been reported and no guidance has been issued. We investigated the use of AD in both Japan and the US by searching public clinical trial registration databases in both countries. Four databases were used for searching clinical trials carried out in Japan, and one database for the US. We included clinical trials using AD registered between January 1, 2013 and December 31, 2017. Twenty-four AD trials conducted in Japan and 97 conducted in the US were identified. The AD were used mainly in the early phase of clinical development. Of the 24 trials conducted in Japan, 23 were sponsored by pharmaceutical companies. In particular, AD were frequently used in clinical trials led by global pharmaceutical companies and overseas affiliate of Japanese companies. In the US, the numbers of domestic AD trials and multinational AD trials were almost the same, and some of them were led by academic institutions. These findings suggest that Japanese companies have limited experience in using AD, and that AD are not yet popular in Japan. Efforts to enhance understanding of AD, such as issuing guidance by regulatory authorities in Japan, are needed to promote diffusion of efficient clinical development using AD.

Background Disease/patient registries are underutilized despite their multiple advantages over clinical trials in the clinical evaluation of drugs, such as the capacity for long-term curation, provision of patient outcome data in routine clinical practice, and provision of benchmark data for comparison. Ensuring the fit-for-purpose quality of data generated from such registries is important to informing regulatory decision making. Here, we report the construction of a quality management system aiming to ensure regulatory-grade data quality for a registry of Japanese patients with glaucoma to evaluate long-term patient outcomes. Methods The quality management system was established by reference to the risk-based approach in the ICH-E6 (R2) recommendations. The following three-component approach was taken: establishment of governance, computerized system validation (CSV), and implementation of risk assessment and control. Compliance of the system with the recommendations of regulatory guidelines relevant to use of the registry was assessed. Results Governance by academic collaboration was established. This was followed by the development of a total of 15 standard operating procedures, including CSV, data management, monitoring, audit, and management of imaging data. The data management system was constructed based on a data management plan, which specified data/paper flow and data management procedures. The electronic data capture (EDC) system was audited by an external vendor, and configured and validated using the V-model framework as recommended in the GAMP5 guideline. Informed consent, eligibility assessment and major ophthalmology measurements were determined as Critical to Quality (CTQ) factors. A total of 22 risk items were identified and classified into three categories, and operationalized in the form of a risk control plan, which included training sessions and risk-based monitoring. The glaucoma registry addressed most quality recommendations in official guidelines issued by multiple health authorities, although two recommendations were not met. Conclusions We established and configured a quality management system for a glaucoma registry to ensure fit-for-purpose data quality for regulatory use, and to curate long-term follow-up data of glaucoma patients in a prospective manner.

This paper proposes a data augmentation method for imbalanced healthcare datasets. This method was inspired by a data augmentation method in natural language processing (NLP) that generates synthetic sentences for training by replacing some words with similar words. The proposed method generates synthetic patient records by replacing patient backgrounds with similar backgrounds. In this paper, the cosine similarity of the distributed representations was used as the similarity metric between patient backgrounds. The distributed representations of the patient backgrounds were generated by the skip-gram model. To confirm the performance improvement with the proposed data augmentation method, the prediction performance of adverse events (AEs) caused by drug administration was experimentally evaluated on a real-world medical dataset with 1,510,137 records. The combination of the proposed data augmentation method and a conventional undersampling method resulted in an 80.0% improvement in accuracy and a 40.0% improvement in the precision and F1-score. The multifaceted evaluation demonstrated that the proposed method is effective, especially for predicting AEs with positive ratios ranging from 1.0% to 2.1%, which are difficult to predict with conventional machine learning methods but should be predictable in the medical field.