The vast majority of drug compounds fail in the preclinical stage, and pharmaceutical and biotech companies are continuously looking for preclinical process improvements to better manage costs and identify candidate failures as soon as possible. Digital pathology is an emerging technology that provides an image-based environment for managing and interpreting information generated from a digitized glass slide.
The process starts with high-resolution whole slide images—pathology slides from a study that have been scanned at 20x, 40x, or even 100x resolution, and stored in a searchable database. Users can search for slides, organize and annotate slides, conduct virtual peer reviews, and run quantitative immunohistochemistry and morphology algorithms on digitized, virtual slides. Researchers access a secure internet-based environment for annotation, sharing, and analysis of the images.
Digital pathology provides a viable platform to move drug candidates through discovery and into clinical trials as quickly as possible. It offers process improvements over standard microscopy by facilitating biomarker studies on patient populations, using computer algorithms to flag abnormals, enabling data sharing via digitized slides, and making virtual peer review a reality. The volume of slides generated by late-stage discovery and preclinical trials can be automatically analyzed for correlations and interactions that might not be otherwise observable improving efficiency and throughput. A comparison between traditional microscopy and digital pathology is shown in Figure 1.
The Adoption Curve
As with all new technologies, pathologists at contract research organization and pharmaceutical companies are adopting digital pathology gradually, similar to how scientists migrated from the use of 35 mm photographic slides to PowerPoint for scientific presentations. Figure 2 shows the typical adoption: from initially scanning only abnormal cases or slides for peer reviews, to scanning slides for quantitative algorithms, to scanning all preclinical slides for a complete GLP study.
Timely Quantitative Pathology
While information from excreted fluids, plasma samples, clinical endpoints, and other toxicology data is normally obtained promptly and automatically entered into a library information system, the information from pathology takes far longer to complete—and the most valuable data, the histopathology of abnormal lesions, still sits trapped in glass slide boxes or sleeves.
There are several reasons that the toxicologic pathology data takes longer to complete. First, finding and scheduling a toxicologic pathologist can be difficult. Frequently in a study, there is an unusual lesion that has not been seen before, or a generalist pathologist requires help from a specialist for a rare event. Second, there is always time lost in flying slides to the pathologists, or flying pathologists to the slides. Digital pathology can improve turnaround time.
Digital pathology provides flexibility through its ability to distribute digital slide images simultaneously in a web-based setting independent of a microscope. One benefit of digital pathology is global access to information by pathologists who can interpret digital slide images wherever there is internet capability.
Digital pathology also includes the use of standard grading slides in quality assurance and proficiency testing, and pathologist-centric computer algorithms in immunohistochemistry and morphology. It isn’t difficult to envision a standardization approach across remote locations with digital pathology. Everyone views the same slide and scores their results, and through a digital slide conference the differences are discussed. This type of standardization ensures pharmaceutical or biotech preclinical managers that their study is getting fair treatment, regardless of which pathologist or pathology location conducts the study. It is particularly important when a company merges two groups of pathologists or adds and trains a new pathologist.
Standardization across studies is rare in the industry. It is difficult to retrieve from boxes of glass slides all the liver sections scored in the last four years as “marked glycogen suppression” or “minimal single-cell necrosis.” With digital pathology this is a query that takes under 10 seconds. Slides can be copied to folders for each lesion type and grade, and then sorted side-by-side with the slide viewer. This factor alone will likely improve the grading of toxicological results.
In most QA programs, one tumor or lesion is cut into sections and then sent to multiple sites for analysis. It is impossible to reproduce identical biology with this approach, and also impossible to reproduce the same testing across several years as slides deteriorate, get lost, or become damaged. With digital pathology, one digital slide can be stored securely and viewed infinite times.
Immunohistochemistry is a terrific technology for measuring protein biomarkers in tissue. It can be run at all stages of drug development. It offers the ability to delivery protein expression results, but on a platform and format comfortable to the pathologists and histologists who predominate in both preclinical and clinical trials. The algorithms employed for measuring membrane and protein expression have seen widespread use in the clinical environment and are expanding in preclinical testing.
The advantages of standardization across studies and pathologists can be utilized to determine and discuss immunohistochemistry scoring systems (e.g. +3, +2, +1, 0). Figure 3 demonstrates an analysis of cytoplasm protein expression of pS6 Ser235 immunostain of breast carcinoma.
Digital toxicologic pathology is the next step in the full electronic data capture and submission of preclinical studies. It is important that the data be integrated with existing LIMS and LIS. Electronic reports of IHC or H&E diagnoses can be offered to pharmaceutical and biotech companies. These can then be loaded into existing drug development customer electronic systems.
Many drug classes have decades of preclinical data, and the most valuable pathology data remains inaccessible in archives. However, when all the slides in a study are scanned digitally, a searchable database is available at study completion that can be linked to compound or target databases, merged with gene expression databases, or mined for toxicogenomic profiling. It is predicted that toxicologic pathologists will adapt to new technology, provided that it truly advances their mission—to protect the human public through preclinical research.