Toxicologic pathology is transitioning from analog to digital methods. This transition seems inevitable due to a host of ongoing social and medical technological forces. Of these, artificial intelligence (AI) and in particular machine learning (ML) are globally disruptive, rapidly growing sectors of technology whose impact on the long-established field of histopathology is quickly being realized. The development of increasing numbers of algorithms, peering ever deeper into the histopathological space, has demonstrated to the scientific community that AI pathology platforms are now poised to truly impact the future of precision and personalized medicine. However, as with all great technological advances, there are implementation and adoption challenges. This review aims to define common and relevant AI and ML terminology, describe data generation and interpretation, outline current and potential future business cases, discuss validation and regulatory hurdles, and most importantly, propose how overcoming the challenges of this burgeoning technology may shape toxicologic pathology for years to come, enabling pathologists to contribute even more effectively to answering scientific questions and solving global health issues. [Box: see text]
We introduce HistoNet, a deep neural network trained on normal tissue. On 1690 slides with rat tissue samples from 6 preclinical toxicology studies, tissue regions were outlined and annotated by pathologists into 46 different tissue classes. From these annotated regions, we sampled small 224 × 224 pixels images (patches) at 6 different levels of magnification. Using 4 studies as training set and 2 studies as test set, we trained VGG-16, ResNet-50, and Inception-v3 networks separately at each magnification level. Among these model architectures, Inception-v3 and ResNet-50 outperformed VGG-16. Inception-v3 identified the tissue from query images, with an accuracy up to 83.4%. Most misclassifications occurred between histologically similar tissues. Investigation of the features learned by the model (embedding layer) using Uniform Manifold Approximation and Projection revealed not only coherent clusters associated with the individual tissues but also subclusters corresponding to histologically meaningful structures that had not been annotated or trained for. This suggests that the histological representation learned by HistoNet could be useful as the basis of other machine learning algorithms and data mining. Finally, we found that models trained on rat tissues can be used on non-human primate and minipig tissues with minimal retraining.
In October 2019, Novartis launched brolucizumab, a single-chain variable fragment molecule targeting vascular endothelial growth factor A, for the treatment of neovascular age-related macular degeneration. In 2020, rare cases of retinal vasculitis and/or retinal vascular occlusion (RV/RO) were reported, often during the first few months after treatment initiation, consistent with a possible immunologic pathobiology. This finding was inconsistent with preclinical studies in cynomolgus monkeys that demonstrated no drug-related intraocular inflammation, or RV/RO, despite the presence of preexisting and treatment-emergent antidrug antibodies (ADAs) in some animals. In this study, the immune response against brolucizumab in humans was assessed using samples from clinical trials and clinical practice. In the brolucizumab-naïve population, anti-brolucizumab ADA responses were detected before any treatment, which was supported by the finding that healthy donors can harbor brolucizumab-specific B cells. This suggested prior exposure of the immune system to proteins with structural similarity. Experiments on samples showed that naïve and brolucizumab-treated ADA-positive patients developed a class-switched, high-affinity immune response, with several linear epitopes being recognized by ADAs. Only patients with RV/RO showed a meaningful T cell response upon recall with brolucizumab. Further studies in cynomolgus monkeys preimmunized against brolucizumab with adjuvant followed by intravitreal brolucizumab challenge demonstrated that high ADA titers were required to generate ocular inflammation and vasculitis/vascular thrombosis, comparable to RV/RO in humans. Immunogenicity therefore seems to be a prerequisite to develop RV/RO. However, because only 2.1% of patients with ADA develop RV/RO, additional factors must play a role in the development of RV/RO.
A Working Group of the Society of Toxicologic Pathology’s Scientific and Regulatory Policy Committee conducted a technical and scientific review of current practices relating to the fixation, trimming, and sectioning of the nonrodent eye to identify key points and species-specific anatomical landmarks to consider when preparing and evaluating eyes of rabbits, dogs, minipigs, and nonhuman primates from ocular and general toxicity studies. The topics addressed in this Points to Consider article include determination of situations when more comprehensive evaluation of the globe and/or associated extraocular tissues should be implemented (expanded ocular sampling), and what constitutes expanded ocular sampling. In addition, this manuscript highlights the practical aspects of fixing, trimming, and sectioning the eye to ensure adequate histopathological evaluation of all major ocular structures, including the cone-dense areas (visual streak/macula/fovea) of the retina for rabbits, dogs, minipigs, and nonhuman primates, which is a current regulatory expectation for ocular toxicity studies.
Biomedical research involving animal models continues to provide important insights into disease pathogenesis and treatment of diseases that impact human health. In particular, nonhuman primates (NHPs) have been used extensively in translational research due to their phylogenetic proximity to humans and similarities to disease pathogenesis and treatment responses as assessed in clinical trials. Microscopic changes in tissues remain a significant endpoint in studies involving these models. Spontaneous, expected (ie, incidental or background) histopathologic changes are commonly encountered and influenced by species, genetic variations, age, and geographical origin of animals, including exposure to infectious or parasitic agents. Often, the background findings confound study-related changes, because numbers of NHPs used in research are limited by animal welfare and other considerations. Moreover, background findings in NHPs can be exacerbated by experimental conditions such as treatment with xenobiotics (eg, infectious morphological changes related to immunosuppressive therapy). This review and summary of research-relevant conditions and pathology in rhesus and cynomolgus macaques, baboons, African green monkeys, common marmosets, tamarins, and squirrel and owl monkeys aims to improve the interpretation and validity of NHP studies.
The 2019 manuscript by the Special Interest Group on Digital Pathology and Image Analysis of the Society of Toxicologic pathology suggested that a synergism between artificial intelligence (AI) and machine learning (ML) technologies and digital toxicologic pathology would improve the daily workflow and future impact of toxicologic pathologists globally. Now 2 years later, the authors of this review consider whether, in their opinion, there is any evidence that supports that thesis. Specifically, we consider the opportunities and challenges for applying ML (the study of computer algorithms that are able to learn from example data and extrapolate the learned information to unseen data) algorithms in toxicologic pathology and how regulatory bodies are navigating this rapidly evolving field. Although we see similarities with the “Last Mile” metaphor, the weight of evidence suggests that toxicologic pathologists should approach ML with an equal dose of skepticism and enthusiasm. There are increasing opportunities for impact in our field that leave the authors cautiously excited and optimistic. Toxicologic pathologists have the opportunity to critically evaluate ML applications with a “call-to-arms” mentality. Why should we be late adopters? There is ample evidence to encourage engagement, growth, and leadership in this field.