A Commentary on the Process of Peer Review and Pathology Data Locking

A detailed report on pathology peer review and data locking, titled “Panel Discussion: Regulatory Perspective for Pathology Data,” was published recently in the Journal of Toxicologic Pathology (Panel Discussion 2009). The content was generated on January 28, 2009, during the 25th annual meeting of the Japanese Society of Toxicologic Pathology (JSTP) at Hamamatsu, Japan. The panelists who facilitated the discussion represented multiple societies of toxicologic pathology including the JSTP, the Society of Toxicologic Pathology (STP), and the European Society of Toxicologic Pathology (ESTP) together with the Japanese Society of Quality Assurance (JSQA). The meeting represented the endpoint of discussions begun by JSTP, after which the International Federation of Societies of Toxicologic Pathologists (IFSTP) and its Regulatory Interaction Committee (RIC) were asked for assistance in pursuing a globally acceptable approach to peer review. The purposes of the Hamamatsu panel discussion were twofold. The first was to clarify different peer-review practices occurring in Japan, the European Union, and the United States with regard to pathology data collection (including digital images). The second was to debate the harmonization of such processes, in particular that of the timing for pathology data locking with respect to sponsor review (a term used specifically in Japan in reference to the peer review carried out by a pharmaceutical or agrochemical company at the contract research organization [CRO]). For the purposes of this commentary, we focus on data locking and sponsor review.

The major difference between the Japanese regulatory requirement and pathology peer-review practices in other countries is the timing of the peer review relative to data locking. In Japan, sponsor peer review is performed after the pathology findings are fixed or locked. A recent survey carried out in Japan by the JSTP in cooperation with the IFSTP/RIC indicated that the majority of companies conduct sponsor peer review after the findings are fixed,¹ which is in line with the current Japanese Pharmaceuticals and Medical Devices Agency (PMDA) guidelines requiring the production of a detailed audit trail of changes to the data interpretation and the peer-review report (Yakuji Nipposha 2008). Interestingly, internal peer reviews within the same company are commonly conducted in Japan prior to fixing of findings, with the approval of PMDA. In contrast, in other nations, peer review typically takes place before locking of pathology data. A recent peer-review survey performed by the ESTP showed that 80% of companies do not fix pathology data prior to peer review, an approach supported by the fact that United States and European regulatory agencies almost never request that data be locked before peer review or that an audit trail of changes in the pathology report or in the interpretation of data be produced (Barale-Thomas and Bradley 2009).

Following the debate at the Hamamatsu meeting, the final view from JSTP is that sponsor review should be acceptable before pathology data are locked, so that the CRO study pathologist and sponsor peer-review pathologist can reach a consensus on the findings and their interpretation without having to produce an audit trail of their negotiations. The JSQA view is in agreement with that of JSTP in that there should be no problem in conducting a sponsor peer review prior to locking the data. However, JSQA believes that the working notes associated with the sponsor’s peer review should be saved. The PMDA continues to mandate that pathology data should be locked before a sponsor review is conducted. This position creates a double standard in that they waive the early locking requirement for studies conducted outside of Japan but not for those conducted in Japan.

We are writing to support the JSTP as they seek PMDA agreement that data need not be locked prior to sponsor peer review. Our opinion is that the JSTP position is appropriate, for a number of reasons.

The most important is that locking of pathology data prior to peer review is not suggested or required by the guidelines set forth by many other regulatory agencies around the world for evaluation of preclinical studies, as it is the signed and dated final pathology report that is considered to represent the histopathology raw data (Code of Federal Regulations, Volume 21 Part 58 and Part 11; U.S. Food and Drug Administration 1987; Committee on the Carcinogenicity of Chemicals in Food, Consumer Products and the Environment 2009). The peer review as performed in nations outside Japan thus represents the international standard for pathology peer review, and the JSTP position that pathology data be locked after sponsor peer review complies with that global practice.

The second reason is that peer review (internal or external) should not be considered an adversarial audit of one pathologist’s product by another but rather as a tool to facilitate delivery of the highest quality final data set and interpretation by the study pathologist. The need to perform a peer review is not disputed. Within the global toxicologic pathology community, there is a well-recognized need for pathology diagnoses made by the study pathologist to be examined by a peer reviewer to confirm accuracy of these findings, ensure that all target tissues have been identified, verify standardization of nomenclature and diagnostic criteria, and correctly identify the no observed adverse effect level (Morton et al. 2006; Society of Toxicologic Pathologists 1997, 1991; Ward et al. 1995; Crissman et al. 2004). Peer review is required because the proficiency with which anatomic pathology data are collected depends on pathology training, personal experience, and knowledge of specific lesions induced by a given compound class. The peer reviewer adds credibility to the original interpretation and provides more confidence on the final interpretation in reports submitted to regulatory agencies. In the United States, the U.S. Food and Drug Administration (FDA) does not specifically have a guideline for peer review. However, studies submitted with peer review are encouraged because it gives extra confidence in the reliability and quality of the data. Indeed, it is likely that a properly documented peer review in a report will actually expedite the review process at regulatory agencies (Ward et al. 1995). In Europe, the situation is similar, although the European Medicines Agency guideline asks specifically for peer review of carcinogenicity studies performed with pharmaceuticals (Committee for Proprietary Medicinal Products 2002; Mann 1996). Regardless, the intent of peer review in the United States and Europe is clearly to produce the highest quality data set with the best possible interpretation.

The third reason addresses the issue of sponsor pressure. The potential for manipulation of the pathology peer-review process was a concern expressed by the JSQA in the panel discussion and is one of the main reasons quoted by PMDA in their contention that data should be locked prior to peer review. Their view is that retention and submission of interim worksheets would deter this pressure. To address this proposition, we must first agree on the definition of pathology raw data. The view expressed by the U.S. FDA and the U.S. Environmental Protection Agency (EPA) and as defined by Good Laboratory Practices regulations is that the signed and dated final report of the study pathologist comprises the pathology raw data (Code of Federal Regulations Volume 21, Part 58 and Part 11; U.S. FDA 1987; Tuomari et al. 2004). Reconstruction and evaluation of the final report could be accomplished by a pathologist reassessing the retained slides and evaluating the signed final report (Lepore 1996; Frantz 1997). This would mitigate against the necessity for retention of interim worksheets (McCullough et al. 1997). In support of this, both the U.S. FDA and the U.S. EPA consider that interim notes are not essential for the reconstruction and evaluation of the pathology report and that study pathologists can modify their data without documentation until finalization (i.e., pathology data are not locked until the final report is signed and dated). Following study locking, all subsequent changes to pathology data result in the production of an audit trail. The interim worksheets are accepted as pathology work files and not raw data. We also suggest that the subtleties of a record of changes made in the course of serial interim worksheets may be difficult to understand by nonpathologists. Indeed, the additional diagnoses or variations from the original data could be interpreted at the most extreme as discrepancies between the two pathologists' conclusions. In fact, there may be no difference in the basic interpretation but only the willingness to present most accurately the nuances of each pathologist’s diagnoses. In our opinion, all regulatory documents (e.g., New Drug Application and Investigational New Drug submissions) carry data from a wide variety of inputting scientists and undergo review and discussion by various committees both internally and externally, so it is in no one’s interests to provide suboptimal data to the regulatory agencies.

In conclusion, for the reasons outlined above and in accordance with current U.S. and European guideline recommendations on pathology peer review, we support the JSTP in their pursuit of a revised Japanese pathology peer-review guideline (1) allowing sponsor peer review prior to locking of pathology data and (2) acknowledging that interim worksheets are not raw data and therefore do not need to be retained or submitted.