In the last decade, numerous initiatives have emerged worldwide to reduce the use of animals in drug development, including more recently the introduction of Virtual Control Groups (VCGs) concept for nonclinical toxicity studies. Although replacement of concurrent controls (CCs) by virtual controls (VCs) represents an exciting opportunity, there are associated challenges that will be discussed in this paper with a more specific focus on anatomic pathology. Coordinated efforts will be needed from toxicologists, clinical and anatomic pathologists, and regulators to support approaches that will facilitate a staggered implementation of VCGs in nonclinical toxicity studies. Notably, the authors believe that a validated database for VC animals will need to include histopathology (digital) slides for microscopic assessment. Ultimately, the most important step lies in the validation of the concept by performing VCG and the full control group in parallel for studies of varying duration over a reasonable timespan to confirm there are no differences in outcomes (dual study design). The authors also discuss a hybrid approach, whereby control groups comprised both concurrent and VCs to demonstrate proof-of-concept. Once confidence is established by sponsors and regulators, VCs have the potential to replace some or all CC animals.
Advancements in digital pathology and artificial intelligence (AI) have enormous transformative potential for nonclinical toxicologic pathology and are already changing the ways in which pathologists work. However, due to the rapid evolution of digital pathology and AI, the toxicologic pathology community would benefit from an update on these advancements, which can be used to aid drug development. Here we identify key articles published on the use of digital pathology and AI in the field and provide current regulatory statuses and guidelines. For digital pathology, we outline the requirements for equipment, validation processes, workflows, and archiving. Challenges to achieve system interoperability and to establish harmonization through Digital Imaging and Communications in Medicine compatibility are also discussed. For AI, we highlight considerations for model development, including the determination of ground truth, problems that may arise due to bias, and how the accuracy and precision of AI algorithms can be assessed. Finally, we discuss the challenges and potential for AI-assisted toxicologic pathology, picturing a future where technology and scientific expertise work hand-in-hand to improve the quality and efficiency of nonclinical drug safety evaluation. This publication is a deliverable of the European Innovative Medicines Initiative 2 Joint Undertaking, “Bigpicture.”
Digital toxicologic histopathology has been broadly adopted in preclinical compound development for informal consultation and peer review. There is now increased interest in implementing the technology for good laboratory practice–regulated study evaluations. However, the implementation is not straightforward because systems and work processes require qualification and validation, with consideration also given to security. As a result of the high-throughput, high-volume nature of safety evaluations, computer performance, ergonomics, efficiency, and integration with laboratory information management systems are further key considerations. The European Society of Toxicologic Pathology organized an international expert workshop with participation by toxicologic pathologists, quality assurance/regulatory experts, and information technology experts to discuss qualification and validation of digital histopathology systems in a good laboratory practice environment, and to share the resulting conclusions broadly in the toxicologic pathology community.
Deep learning models have been applied on various tissues in order to recognize malignancies. However, these models focus on relatively narrow tissue context or well-defined pathologies. Here, instead of focusing on pathologies, we introduce models characterizing the diversity of normal tissues. We obtained 1,690 slides with rat tissue samples from the control groups of six preclinical toxicology studies, on which tissue regions were outlined and annotated by pathologists into 46 different tissue classes. From these annotated regions, we sampled small patches of 224 × 224 pixels at six different levels of magnification. Using four studies as training set and two studies as test set, we trained VGG-16, ResNet-50, and Inception-v3 networks separately at each of these magnification levels. Among these models, Inception-v3 consistently outperformed the other networks and attained accuracies up to 83.4% (top-3 accuracy: 96.3%). Further analysis showed that most tissue confusions occurred within clusters of histologically similar tissues. Investigation of the embedding layer using the UMAP method revealed not only pronounced clusters corresponding to the individual tissues, but also subclusters corresponding to histologically meaningful structures that had neither been annotated nor trained for. This suggests that the histological representation learned by the normal histology network could also be used to flag abnormal tissue as outliers in the embedding space without a need to explicitly train for specific types of abnormalities. Finally, we found that models trained on rat tissues can be used on non-human primate and minipig tissues with minimal retraining.Author contribution: T.S. and H.H. contributed equally to this work.Significance statement: Like many other scientific disciplines, histopathology has been profoundly impacted by recent advances in machine learning with deep neural networks. In this field, most deep learning models reported in the literature are trained on pathologies in specific tissues/contexts. Here, we aim to establish a model of normal tissues as a foundation for future models of histopathology. We build models that are specific to histopathology images and we show that their embeddings are better feature vectors for describing the underlying images than those of off-the shelf CNN models. Therefore, our models could be useful for transfer learning to improve the accuracy of other histopathology models.
The last session of the 2024 European Society of Toxicologic Pathology Congress was dedicated to recent neuro concepts, encompassing various topics including artificial intelligence (AI) applied to toxicologic pathology, new technologies for visualization and/or analysis of tissues, as well as specific case reports. Four presentations detailed the role of AI and computational pathology for toxicologic pathologists, usage of cryofluorescent tomography in neuropathology, blood-brain barrier organoids in early drug screening, and spatial transcriptomics in toxicologic pathology. In addition, there were two short presentations about a spontaneous brain lesion in Beagle dogs as well as central nervous system lesions related to adeno-associated virus vectors administration in various preclinical species. This session underscored the current and prospective impact of emerging technologies on drug screening and development.