Regulatory Forum Opinion Piece*

Abstract

Electronic images of histopathological changes are commonly and increasingly used in toxicologic pathology for morphological evaluation, illustration, peer review, or reporting. Toxicity studies in which cell proliferation is an end point are also pivotal in determining the carcinogenic potential of new molecules. In this article, we describe the approach of the European Cell Proliferation and Apoptosis working group (CEPA) for performing cell proliferation studies and morphometry using electronic images. The Society of Toxicologic Pathology (STP) has published a position statement on handling of pathology image data in compliance with 21 Code of Federal Regulations (CFR) Parts 58 and 11. CEPA supports the STP position and shares the issues involved in the use of electronic images in pathology. However, considering the experience and current know-how of members, particularly in conducting cell proliferation studies, CEPA would like to recommend in this article that electronic images acquired using state-of-the-art slide imaging techniques, including whole slide scanning, need not be considered as raw data, and therefore are not subject to 21 CFR Parts 58 and 11 regulations for archiving. In this article, we detail the reasons why we come to this proposal and we describe the measures that are taken to ensure Good Laboratory Practice–compliant execution of cell proliferation studies that include acquisition and validation of imaging and image analysis systems, development and validation of methods for their intended use, formulation, and use of standard operating procedures.

European Cell Proliferation and Apoptosis Working Group’s (CEPA) Contribution to a Standardization for Performing Cell Proliferation Studies

Measuring cell proliferation in a specific organ is often needed for predicting tumor formation or demonstrating the mode of action of tumor induction in preclinical studies.

The increasing importance of cell proliferation and apoptosis studies and their standardization in toxicology prompted the formation of the CEPA working group by European pharmaceutical, agrochemical, and chemical companies in 1998, organized by Fraunhofer Institute für Toxikologie und Experimentelle Medizin (ITEM) as a subgroup of Registry of Toxicology Animal Data (RITA). The primary goal of CEPA has been to develop guidance on the conduct of cellular proliferation and apoptosis studies. Such standardized methodology allows generating reliable data and proper interpretation and comparison among different laboratories. Another focus area for the CEPA group is the exchange on special staining techniques and other new technologies used in those morphometric measurements. The different topics shared during the last 10 years by CEPA members include counting of proliferating or apoptotic cells and determination of indices of cell proliferation in a specific cell type and organ, counting of ovarian follicles, neuronal plaques. But there are also numerous other examples of morphometry measurement for which those tools can apply.

This common effort reduces significantly the time and cost of method development. The strategy of the CEPA group in performing cell proliferation and apoptosis studies using bromodeoxyuridine (BrdU), proliferation cell nuclear antigen (PCNA), Ki67, and terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL), including organ guides, has been described (Nolte et al. 2005). Organ guides are available for CEPA members through its web page (http://reni.item.fraunhofer.de/reni/public/cepa/#cepa3). CEPA represents a forum to share experience, knowledge, and solutions on the conduct and interpretation of such studies at a state of the art among its members.

Some of those studies performed during the development of a chemical compound are performed under Good Laboratory Practice (GLP) and are submitted to the different authorities in charge of the evaluation and authorization of a new product. The compliance to GLP for those new techniques is also a common issue shared among the group members.

The Use of Electronic Images: A Common but Challenging Practice

Quantification studies are performed nowadays on electronic images acquired with digital cameras attached to microscopes. Manual or semiautomatic counting has been used on individual images, but automated image analysis is the current tool for assessing cell proliferation. More recently, entire digital slides generated by whole slide scanning (WSS) have been utilized routinely (Persohn et al. 2007). WSS and subsequent automated image analysis allow assessment of entire tissue sections, offering new opportunities for the quantitative evaluation of compound effects in toxicity studies.

At the same time, electronic image data files generated using this approach represent a challenge for the logistics of acquisition, evaluation, storage, and archiving, in particular for GLP studies. Indeed, slide scanners generate image data files with vendor-specific file formats and huge file size. The specific file formats may create a significant issue when retrieving the data from an archive at a future date, due to rapid technological development of computer systems and imaging equipment as well as the difficulties to maintain the original systems used to create the data. In addition, the high resolution and very large size of these image files make the printing for archiving purposes unreasonable and this does not correspond to the philosophy of electronic technologies. These have to be stored as electronic files.

The use of electronic image data in compliance with 21 Code of Federal Regulations (CFR) Part 11 regulations is debated among pathologists and experts in quality assurance because, until recently, current technologies and implementation of processes still lag behind the practical fulfillment of the mandatory requirements concerning traceability, reconstruction, integrity, authentication, and archiving of such data. Authentication and archiving of electronic image data present considerable challenges, especially in view of the ever growing amount and size of data. Currently, within the CEPA group, fully GLP-compliant archives for electronic images are limited to very few companies.

The Society of Toxicologic Pathology (STP) Position Paper on Pathology Image Data (Tuomari et al. 2007) clearly describes the acquisition and use of pathology images using electronic devices for data generation, in compliance with the CFR Volume 21, Parts 58 (GLP) and 11 (Electronic Records/Signatures). In this article, the authors suggest consideration of electronic images as raw data when they are used to generate data (e.g., for morphometric analysis). As a consequence, they need to be acquired and handled using equipment, software, and procedures that ensure their integrity, authentication, and reliability. Thus, archiving of such electronic images is necessary. The required steps to validate digital pathology systems in such cases have recently been described in detail by the Scientific and Regulatory Policy Committee (SRPC) of the STP (Long et al. 2013).

CEPA Proposes That Electronic Images Are Not Raw Data in Cell Proliferation Studies

While we fully support the STP position, we want to highlight current know-how concerning image analysis using electronic images for cell proliferation studies. This includes the use of WSS, which allows reliable re-creation of slide images from the original durable specimen (Gilbertson et al. 2006). Good practice based on appropriate validation of the image-based analysis method for the intended use, detailed documentation of the analysis process, and carefully devised standard operating procedures (SOPs) in combination with proper training makes cell proliferation studies using electronic images possible and compliant with GLP (Persohn et al. 2007) without the necessity to archive the electronic images.

The CEPA group considers electronic images used in cell proliferation studies as temporary data and not raw data. They can be stored but do not need to be archived. In contrast, the descriptive data summarizing the obtained values and the related pathology report are the raw data of such studies, which can then be archived electronically or as printouts. The histological slide has to be archived as well. A complete validation and in-depth documentation of the analysis process are a prerequisite for performing those cell proliferation studies. The data generated during the validation process should also be archived and can be shown if needed.

Fully GLP-compliant Archives for Automated Image Analysis Software and Electronic Images and Their Implementation on a Larger Scale Remains a Challenge

If images are considered raw data, authentication and archiving become mandatory. Especially in the case of digital slide data, this becomes a major issue in part due to the vendor-specific file formats used. Fully GLP-compliant image archives are not yet readily available to the members of the CEPA group. Because they are continually decreasing, storage costs will allow the storage of big volumes of electronic image data including digital slides. But full implementation of 21 CFR Part 11 regulations (i.e., authentication and full audit trail) will remain a challenge. Maintaining computer systems that generate electronic images as raw data is difficult. Lifetime support for these systems is typically less than 10 years and is often outpaced by rapidly evolving technology. This requires that well-documented data formats are in place in order to allow re-creation of data from electronic records even if the original acquisition or processing systems are no longer available. The Digital Imaging and Communication in Medicine (DICOM) consortium has addressed this issue for radiographic (X-ray), magnetic resonance (MR), and computed tomography (CT) images. The DICOM working group 26 has recently published a supplement to the DICOM standard that addresses digital slides (DICOM Standards Committee WG 26 2010). Full implementation of a DICOM standard for digital slides in the future could allow large-scale Picture Archiving and Communication (PAC) systems to be used for storage and archiving of digital slides. This is a possible manner to resolve the issue.

GLP-compliant Validation of Systems Should Allow Researchers the Use of Novel Technologies and Electronic Images

An important step in the preparation of cell proliferation studies is the validation of the image analysis method for the intended use. For this purpose, an approach has been established, where the image analysis method is developed so that the automated cell counts match the results obtained through manual counts performed by several pathologists (4–5) on a number of randomly selected image samples, counting a sufficient number of cells (Persohn et al. 2007). An identical validation approach has been described in the paper published by the SRPC of the STP (Long et al. 2013). Many of the challenges described above relate to the promotion of data integrity and are addressed by carefully devised SOPs that detail the processes involved in performing cell proliferation studies.

A feature of whole slide imaging is that current state-of-the-art slide scanner technology permits re-creation of exact images from durable slide-based specimen at any time. Considering this feature of slide scanning technology in combination with the validation of the image analysis and carefully devised SOPs, CEPA members regard digital slides used to generate cell proliferation data as temporary data that do not need to be archived. The data sheet with the resulting cell counts and the respective pathology report are considered the raw data of a cell proliferation study, and these are archived together with the method validation report.

Examples of Image-based Assays Where Temporary Image Data Are Discarded

Our proposal is analogous to other measurement systems, where image data are acquired, counting is performed, and exported to data tables and where subsequently the temporary image data are discarded and not considered raw data. Examples for image-based assays that employ GLP-compliant images and analysis include flow cytometry, immunofluorescent or confocal microscopy, and the micronucleus test for genotoxicity (Frieauff and Romagna 1994). For technical reasons, the sample may not be available for re-creation of the image data, because it is destroyed in flow cytometry, or because fluorescent labeling fades in fluorescent or confocal microscopy. In contrast, appropriately stained cytosopin glass slides are available for image re-creation in the micronucleus test.

The Use of Digital Slides in Cell Counting Studies Has Advantages, but Gaps Remain for Efficient Use and Archiving of Digital Images

WSS, generating digital slides, is a recent and important advancement in automated microscopy. It has obvious advantages over traditional image sampling of tissue sections using a microscope and camera: a much larger region of interest provides significantly more cells assessed to determine indices, and manual selection bias is eliminated, thus resulting in more robust data.

The CEPA group shares the opinion of the STP that WSS is a big step forward toward a reliable re-creation of an image from the tissue specimen. Current hardware and software are capable of producing excellent color representation of the glass slide. However, there is some discussion whether an “exact” image can be re-created. This issue is best addressed in an SOP describing the use of such scanning systems to include a calibration step prior to application in GLP studies. Such an SOP should also specify a margin for accepting slide scans and subsequently generated results as “equal” to the reference method.

While the technologies for image acquisition (including WSS) and image analysis as well as storage of digital data are rapidly evolving, mid- and long-term archiving of such data are not yet readily available. A central issue remains the availability of the required processes for archiving, which include electronic signatures, long-term storage of the image data in a well-documented and accessible data format independent of any specific hardware, and (long-term) retrieval of data from the archives. As suggested by STP, currently the best approach to archiving is the use of printouts. Regardless of the availability of electronic archiving procedures and systems, a clear definition of what is the raw data for a study is a key step in reducing the need for archiving of unnecessary data.

The entire community of pathologists and other scientists, quality assurance authorities, information technology specialists, and suppliers of imaging equipment and software have to continue to work on commonly accepted tools and processes in order to support seamless and easy use of electronic images including digital slides in compliance with GLP.

Conclusion

The CEPA group recommends that digital images used in cell proliferation studies do not need to be considered as raw data and electronic archiving of these images therefore is not required. Only the data tables resulting from the analysis and the associated pathology report should be considered raw data for this type of studies.

State-of-the-art slide scanners produce whole slide image data, which dramatically increase the tissue area available for analysis and reduce any bias introduced by manual selection of image fields. This greatly improves the statistical evaluation and robustness of cell proliferation studies. In addition, current slide scanners allow researchers to reliably re-create the whole slide image data from the durable specimen at any desired time after the analysis, and this makes the archiving of such digital slides unnecessary. As indicated in the STP position paper, these novel technologies should be employed and lack of full GLP-compliant validation of systems should not prevent researchers from using them.

Specific measures need to be taken to make cell proliferation studies using electronic images (including digital slides) GLP compliant. The qualification of the image analysis tools, the method, and the analysis results have to be achieved and documented. This is achieved by performing a validation of the cell proliferation (or other) analysis method for the intended use, by implementing specific SOPs, and by training of personnel.

The archiving of electronic images and image analysis software when considered raw data remain a challenge. Off-the-shelf archiving solutions fully integrated with workflows utilizing electronic images—including digital slides—and image analysis are not readily available. Hopefully in the near future, such systems based on unified image file formats will be available.