Regulatory Forum Opinion Piece*

Pathology peer review is one of those pesky perennial issues that continue to haunt industrial pathologists. Just when you think it has been nailed once and for all by the issuance of a “Best Practice” document from the Society of Toxicological Pathology (STP 1997; Crissman et al. 2004; Morton et al. 2010), it rises, phoenix like from the ashes, to consume us for another few months of letter writing and fervent hallway discussion. Nevertheless, as new generations join the professions of toxicology and pathology, it will likely continue to be a recurring theme and so maybe will merit continued discussion, in which we are happy to engage.

The opinion article by Engelhardt, Hardisty, and Mann published in this issue has a noble goal, which we applaud, of trying to increase transparency and thus reduce anxiety among regulators about the “black box” of pathology peer review. But, we submit that the means they believe will accomplish this will fail, and indeed, we hope to convince you that it will actually create the opposite effect.

Their proposal is simply to retain the interim tables created by the peer review pathologist (PRP), thus apparently demystifying the process of arriving at the final consensually agreed set of diagnoses. And where there is general agreement between the study pathologist (SP) and the PRP, or where disagreement centers around nontreatment-related findings, there is no problem, although what value this material may be is questionable. But when there are differences of opinion in treatment-related findings in the initial tables, it creates an issue and a potential set of dilemmas where none should exist. If an SP diagnoses hyperplasia and a PRP diagnoses adenoma, but the final tables say hyperplasia, what is a regulator supposed to think? Has there been undue influence on the PRP pressuring them to move to a less impactful diagnosis? Or, did the SP convince the PRP that he or she was overcalling the diagnosis in a legitimate educative process of coequal learning and intellectual give and take? What about when an inexperienced SP breaks out components of chronic progressive nephropathy (CPN) in a rat study resulting in tables showing an apparent increase in incidence of some components with dose, but a decrease of others with increasing dose. The experienced PRP convinces the SP that this is a purely serendipitous effect and that lumping the components together shows no treatment effect for CPN and an incidence that is not increased above the historical background rate for the rat strain. An inexperienced regulator may surmise that the PRP inappropriately influenced the SP to suppress some treatment-related renal findings.

Fundamentally, there is no value to retention of the interim tables. Where there is disagreement between tables created by SP and PRP, one or both of these interim data will be wrong—mutually agreed as wrong by the SP and PRP—and the final tables will have the consensually agreed correct diagnosis. There is no support by either pathologist for a diagnosis in any interim table which they both have agreed is now incorrect, and therefore no point in reporting this wrong diagnosis. To be continually challenged to defend why they changed diagnoses between interim and final tables would be an enormous burden, and frankly a fruitless exercise if what we really want is the best possible set of tables to describe the effect of the test article on the target species. Wrong is wrong; it is only the final, correct interpretation and diagnoses that matter, attested to by the SP and the PRP by their signatures. We find it very hard to support the routine preservation of these interim notes containing potential peer review disagreements. The aim of pathology peer review, as has been expressed many times, is to assist the SP with creating the best, most accurate set of tables of diagnoses, and help them with formulating the interpretation of those diagnoses to describe the toxicity of the test article in the test species. The critical piece of this is the notion of the PRP assisting the SP to formulate an opinion based on their combined experience and thinking, with the only necessary record being a short document indicating what was evaluated by the PRP to come to that agreement. It was not designed as another read with de novo diagnoses, but as a constructive way for a reviewing pathologist to agree with, or suggest modifications and improvements to, the SP’s work. But, what has been described by Engelhardt, Hardisty, and Mann effectively creates the appearance of a second primary read of the study with a subsequent reconciliation step, retaining multiple sets of unnecessary tables that may be used punitively, but are not raw data according to current interpretation, and are of little value for the final purpose of producing the most accurate depiction of the effects of the test article.

The work that veterinary pathologists do in industry is a medical diagnostic process that takes observations made on tissues and imparts a subjective name and interpretation to them. Indeed, this is a “black box,” and these “data” that we create are nothing more than our opinions, based on our respective experience and training. Peer review, whether formal or informal, is a valuable way to make this better by getting another pathologist’s opinion about our thoughts as an SP on how we name and interpret study-related toxicities. If the process of two pathologists working together to get to a consensus diagnosis is erroneously perceived as being easily deconstructed, scrutinized, and analyzed to provide some type of advantage, we should work to educate our customers that this is not the case. The last thing we should do is to provide copious material that can be misunderstood, and potentially misused, and which will not ultimately help regulators understand what it is that we really do. The current way we do peer review is a time-tested process that works well and should not be burdened by retention of interim information that is not required (i.e., is not raw data), is not useful, and in fact may be used improperly for a purpose for which it was not created. Keeping interim tables will not demystify the process of peer review. Rather, it will obscure and potentially impede and delay the real goal of producing the most accurate tables and interpretations.