spacer
home > ebr > autumn 2012 > picture perfect
PUBLICATIONS
European Biopharmaceutical Review

Picture Perfect


Digital pathology is revolutionising the field of bioimaging by reducing operational costs, improving quality and enabling researchers to fast-track the discovery of novel tissue-based biomarkers for companion diagnostics through automated tissue imaging.

The evaluation of tissue on a glass slide by experienced diagnostic pathologists using a microscope is central to many aspects of drug discovery, clinical trials and the development of diagnostic tests. This is particularly important in oncology and the evaluation of solid tumours. In the earliest stages of target identification, understanding cancer pathobiology, studying signalling pathways and evaluating new sequences and proteins as potential therapeutic targets, human tissue samples are widely used. Fresh, frozen or fixed archived tissue samples are analysed to examine the differential expression of gene products in normal tissues and cancer subtypes, as well as the relationship between tissue expression and clinical outcome. With increased momentum towards the development of companion diagnostics, tumour tissue has become a focus for biomarker discovery, again underpinning the importance of pathology in both biomarker discovery and clinical adoption.

Digital Pathology

Digital pathology describes the creation, viewing and analysis of a high resolution image of the glass slide, allowing the pathologist to review the ‘digital slide’ on a computer monitor rather than using a microscope. It follows closely on the heels of the digital radiology revolution, which has completely changed the face of medical imaging. However, one of the biggest benefits of digital pathology is that the pathologist and the physical glass slide no longer need to be in the same room. Digital slides can be viewed remotely via the web, in their entirety at any magnification, allowing pathologists to review tissue samples anywhere in the world, at any time.

Digital slides are large. With a resolution of 0.24μm per pixel and 25 by 20mm in size, average scans can easily be 100,000 by 80,000 pixels in size, resulting in compressed files in the order of 1Gb each. This requires significant storage capacity to digitally archive slides. However, with appropriate backup, the digital slide archive represents a permanent record of the trial or experiment, where slides can be immediately retrieved and viewed online at any point during or after the trial. There is no risk of fading or slide breakage, as is common with glass slides.

Remote review of digital slides does not require the entire image to be downloaded to the remote site. Instead, the image is served ‘on demand’, allowing the user to rapidly navigate the image freely with only the relevant regions of the image being downloaded to the client-side digital slide viewer, replicating how pathologists use standard microscopy.

Software is essential to bring all of the benefits of digital pathology to the drug discovery and diagnostics industry. It is the software and the architecture on which it sits that drives the improved efficiency, enhanced workflows and reduced operational costs that can be experienced when using digital pathology. In addition, since different scanner manufacturers generate own proprietary image format, it is important that digital pathology software can read and display various formats – particularly important in an industry where collaboration between organisations involves scanned images from many scanner brands and in many formats.

Improving Efficiency, Workflow and Interoperability

Digital pathology, as the name suggests, should also involve the management, distribution and analysis of the data associated with or derived from the images. In patient evaluation, in whatever context, the images are just part of a more complex profile and should not be studied in isolation. It is therefore essential that digital pathology software facilitates the integration and visualisation of the multivariate data associated with tissue diagnostics and biomarker discovery and the ability to report on, or score the slide, electronically. This can be achieved with a single user interface that displays all of the relevant digital slide and patient or sample data, and a reporting interface which allows pathological review and biomarker assessment. This provides a novel means of managing pathology studies entirely electronically, distributing digital slides to pathologists for remote review and collecting the results of their review centrally. This, in turn, can provide increased efficiency in managing pathology-centric studies and provides access to a greater pool of pathologists, potentially speeding up review.

Security and regulatory compliance is essential when digital pathology is used to handle sensitive and proprietary images and patient related data. Digital pathology software, including cloud-based software as a service and storage solutions, must ensure restricted, authorised access to data and demonstrate GLP, GCP and 21 CFR part 11 compliancy.

LIMS Integration

As digital pathology increasingly encompasses wider information management and facilitates workflow requirements, there is also a need to ensure interoperability with other laboratory systems that already store and process relevant information. This is particularly important in organisations that have existing laboratory information management systems (LIMS) that manage sample processing and associated information. Extending existing workflows to include the management of digital slides should be seamless, allowing effective data exchange between information platforms, and providing a means for digital pathology to be layered on top of existing information systems. This provides organisations with the enhanced advantages that digital pathology brings to a research and development programme without the need to strip out and completely replace existing systems.

Biobanking

Digital pathology is also playing a significant role as an integrated capability in tissue biobanks. Biobanks are organised collections of high quality biological samples linked with well-defined data sets, and are a critical platform for high impact diagnostics and therapeutics research, underpinning the development of molecular biomarkers, companion diagnostics and personalised medicine (1). Digital pathology provides the infrastructure to digitally store tissue histology and biomarker images, both as part of targeted tissue collections or clinical trials, and allowing these resources to be used for subsequent sample selection and retrieval for research. Together with integrated tissue microarray imaging and image analysis, this provides biobanks with added functionality to underpin their importance to biopharmaceutical research.

Multinational Integration

Digital pathology and its ability to virtually manage tissue samples for pathological review is a key benefit for large multi-national, multi-site clinical trials or biomarker studies. It reduces the cost associated with physical distribution of slides and fast-tracks data collection by allowing immediate real-time review and reporting of slides by pathologists offering the most competitive rates and having the right expertise. Digital pathology can also overcome the regulatory requirements around moving tissue samples outside of the country. Several CRO organisations are now facilitating access to drug and biomarker trials in China and India by using digital pathology to access the images, avoiding the need to relocate glass slides. Global clinical trials have the potential to shorten drug development times, lower operating costs and expand participant numbers. Digital pathology will not only be an enabling technology in these situations, but will also underpin the cost-benefits in geographically dispersed and tightly regulated trials.

Biomarker Discovery and Companion Diagnostics

Digital pathology is emerging as an essential platform on which to investigate, identify, validate and translate new tissue biomarkers. The convenience of using digital images for biomarker scoring and centralised collection of data better ensures data integrity and provides a reliable audit trail for biomarker trials.

Tissue Microarray Analysis

Tissue microarrays (TMAs) represent a highly parallelised way of assessing cellular biomarker expression in tissues and are widely used in tissue biomarker discovery and validation. By having hundreds of patient tissue samples (cores) on a single slide, with different clinical outcomes or treatment responses, individual cores can be evaluated to determine the relationship between the presence of a particular protein or sequence and clinical outcome. The increasing trend to develop companion biomarkers for new drugs and therapeutic regimes is already seeing the expanded use of TMAs in biomarker research and validation of companion diagnostics.

Digital pathology can add enormous value to tissue microarray (TMA) experiments by overcoming many of the manual, labour-intensive aspects of traditional TMA evaluation. By using a dedicated software interface to manage entire TMA experiments, configure TMA maps, define scoring criteria, view tissue cores, collate scores and provide experimental biomarker reports, digital pathology can significantly reduce the manual task involved in traditional TMA evaluation. This again can be carried out entirely on-line allowing experienced pathologists to be scoring a TMA from anywhere in world. Virtual TMA analysis using digital pathology can speed up conventional TMA-based biomarker analysis by up to five times, freeing up pathologists time, reducing costs and fast-tracking biomarker development (see Figure 1).

Image Analysis and Companion Algorithms

Visual interpretation of biomarker expression is inherently subjective and prone to error, even when carried out by experienced pathologists. Computerised image analysis provides the ability to use the information inherent in a digital image to extract quantitative data on tissue pattern and biomarker expression. Used appropriately, image analysis can provide important objectivity and repeatability in biomarker studies, allowing the identification of subtle changes in biomarker expression that might be missed by the naked eye. While the value of quantitative pathology has been advocated for many years, the advent of whole slide imaging has resulted in a resurgence of new methods and new applications for image analysis in tissue pathology, the development of biomarkers and stratified medicine.

A variety of image analysis toolboxes and target specific algorithms are commercially available that allow the quantitative measurement of morphological features, immunohistochemical (IHC) of specific proteins and in situ hybridisation (ISH) detection of nucleic acid sequences in cells. Few image analysis algorithms work ‘off the shelf’, and there is invariably the need to configure algorithms for the specific tissue type and biomarker that requires measurement. However, in experienced hands, algorithms can be developed which can reliably segment cellular compartments and extract quantitative data on the level of biomarker expression (see Figure 2, page 34). Due to its quantitative basis and improved objectivity and reliability, image analysis is now being recognised as an essential prerequisite to biomarker discovery and validation.

Image analysis should not be considered as a black box to biomarker assessment that can entirely replace experienced pathologists, except in well-validated and tested situations. Tissue imagery is extremely complex and image analysis should only be validated and used by those familiar with the underlying image content and tissue morphology. However, image analysis can provide a means to automate biomarker evaluation in tissues and provide numerical data which can be more easily verified than subjective visual interpretation. This is of particular importance in TMA biomarker studies, where hundreds of samples have to be evaluated within a single digital image. Building imaging software to automate the identification of tissue cores and algorithms to score biomarkers can significantly speed up multiplex biomarker assessment across numerous samples (2). High performance computing can further speed up tissue analysis by parallelising image processing and measurement algorithms (2). For diagnostics companies wishing to validate candidate tissue biomarkers for patient stratification, image analysis is going to become an essential tool.

As the drive for personalised medicine continues, drug development and companion diagnostics will increasingly be carried out in tandem. Once a companion tissue diagnostic has been identified, validated and approved for use, image analysis will also play an important role in the test laboratory as an objective, reliable assay for the tissue biomarker. This is already happening. While the FDA has not yet approved digital pathology for primary diagnosis, they have approved image analysis algorithms for the clinical evaluation of PR, ER and Her2 IHC in breast cancer. Her2 is a good example. Recognising that up to 20 per cent of Her2 evaluations by pathologists could be inaccurate, the American Society of Clinical Oncology and College of American Pathologists have recommended the use of image analysis for Her2 assessment, provided it is validated internally within a lab (3). The utility of Her2 image analysis has been extensively validated by some laboratories (4). This and other emerging examples represent a step-change in how new companion diagnostics are going to be derived in the diagnostics laboratory, and may become the standard, forcing the concurrent development of algorithms as part of the biomarker discovery process.

Interestingly, in addition to overall biomarker expression in tissues, tumour heterogeneity is emerging as an important clinical indicator of response to treatment. As an example, Her2 heterogeneity in breast cancer may account for the 30 per cent of patients that fail to respond to, or acquire resistance to Herceptin therapy. Image analysis provides the perfect tool to measure levels of Her2 expression in breast cancer and could act as a more efficient means of stratifying and selecting patients for herceptin therapy (5).

In addition to biomarker measurement, automated imaging and tissue analysis have many new emerging applications that will significantly reduce the time and cost involved in drug and biomarker development. Pathologists are heavily involved in reviewing tissue samples for subsequent mutational analysis, sequence evaluation and genomic profiling, both for discovery and in established molecular diagnostics. This and other manual tasks represent a significant bottleneck to discovery and the translation of new molecular tests based on tissue samples. Automation of these manual tasks using digital pathology will radically alter how tissue and molecular diagnostic tests will be developed and delivered in the future.

Conclusion

Digital pathology is now an essential technology for drug discovery and the development of new tissue diagnostic tests. The cost of hardware is falling and software is now meeting the regulatory requirements, quality standards and security levels necessary for use in private and public sector research. At its core is the ability to share images, data and reports, freeing up access to skilled pathologists who can be anywhere in the world. However, the deeper applications of digital pathology will actually transform discovery and the delivery of tissue-centric tests in the future.

References
  1. Hewitt RE, Biobanking: the foundation of personalised medicine, Curr Opin Oncol 23: pp112-119, 2011
  2. 2. Wang Y, Savage K, Grills C, McCavigan A, James JA and Hamilton PW, A TMA De-Arraying Method for High Throughput Biomarker Discovery in Tissue Research, PLoS ONE 6(10): e26007, 2011 doi:10.1371/journal. pone.0026007
  3. Wolff AC, Hammond ME, Schwartz JN et al, American Society of Clinical Oncology/College of American Pathologists guideline recommendations for human epidermal growth factor receptor 2 testing in breast cancer, Arch Pathol Lab Med 131: pp18-43, 2007
  4. Gustavson MD, Bourke-Martin B, Reilly D et al, Standardization of HER2 immunohistochemistry in breast cancer by automated quantitative analysis, Arch Pathol Lab Me 133: pp1,413-1,419, 2009
  5. Potts SJ, Krueger JS, Landid ND, Eberhard DA, Young GD, Schmechel SC and Lange H, Evaluatinfg tumour heterogeneity in immunohistochemistry-stained breast cancer tissue, Laboratory Investigation 92: pp1,342-1,357, 2012

Read full article from PDF >>

Rate this article You must be a member of the site to make a vote.  
Average rating:
0
     

There are no comments in regards to this article.

spacer
Peter Hamilton is VP for Research and Development at PathXL Ltd, a digital pathology software company. He is also Professor of Bioimaging and Bioinformatics at Queen’s University, Belfast. For the past 25 years, he has been leading research on computer vision and decision support in diagnostic cancer pathology and the identification of novel digital tissue and cell markers for diagnostics, prognostics and for predicting response to therapy in cancer. He has published over 100 peer reviewed journal articles on the subject of quantitative digital pathology and image analysis and has been the recipient of numerous national and international grants for his work in digital pathology.
spacer
Peter Hamilton
spacer
spacer
Print this page
Send to a friend
Privacy statement
News and Press Releases

Argonaut Manufacturing Services Set to Expand Drug Product Fill/Finish Capabilities with Significant Capital Investment

Carlsbad, CA, 10th January 2023 - Argonaut Manufacturing Services (‘Argonaut’), a contract manufacturing organization (CMO) serving the life sciences, molecular diagnostics, and biopharma industries, today announced a capital investment led by majority shareholder Telegraph Hill Partners that will be focused on expansion at their Carlsbad, CA, manufacturing facilities.
More info >>

White Papers

New Temperature Profiles from a Global Study

Laminar Medica

Recently there has been an influx of temperature mapping studies to collect sufficient ambient data to develop temperature profiles.  Solid temperature profiles on specific shipping routes are required to ensure streamlined packaging and robust lane qualification. The most reliable method to develop temperature profiles is to capture the actual reading from shipments, as well as using the historical and meteorological data to increase the confidence level.
More info >>

 

 

 

©2000-2011 Samedan Ltd.
Add to favourites

Print this page

Send to a friend
Privacy statement