Toxicologic Pathology Forum*: Commentary on “Opinion on Designation of Adverse and Nonadverse Histopathological Findings in Toxicity Studies: The Pathologist’s Dilemma”

There has been long-standing interest in aligning the designation of adversity by toxicologic pathologists. Recently, working groups from the Society of Toxicologic Pathology (STP) and European Society of Toxicologic Pathology (ESTP) have produced outstanding and well-received articles addressing these topics. ^1,2 The efforts of these groups were focused around establishing a common framework for such assessments but avoided any attempt to create a defined list or doctrinaire methodology. Although the concept of prospectively categorizing histopathology findings by adversity is convenient and appealing, the numbers of very clear examples that can be easily adjudicated based on histopathology alone are quickly exhausted since the vast majority require additional information and context for accurate decision-making. The designation of adversity is often complex and requires a careful and integrated approach that considers all available data. A key conclusion summarized from a recent ESTP workshop ¹ was that “there is no one formula or method that can be applied to all adversity decisions.”

In their recent article titled “Opinion on designation of adverse and nonadverse histopathological findings in toxicity studies: the pathologist’s dilemma,” the authors Gopinath and Mowat have shared their framework for designating adversity supplemented with photomicrographic examples. The term “pathologist’s dilemma” in the title immediately conjures the challenging and complex process that pathologists encounter when ascribing adversity. Unfortunately, the examples provided bypass a large component of this complexity by focusing almost entirely on histopathology with little to no inclusion of contextual parameters such as mechanism of action, clinical signs, clinical pathology, or related histopathology results. Additionally, the authors’ proposed categorizations and principles risk “muddying the waters” after the publication of the two very clear recent articles put forward by the STP and ESTP. Although Gopinath and Mowat state that the determination of adversity should be based on a weight of evidence approach, we felt that many of their proposed principles, designations, categorizations, and lists give the impression of a much different, overly simplistic approach that does not always consider all of the available data, discussions with other study team members, or study nuances which are paramount to an accurate adversity determination. Given the important implications of adversity designation on study conclusions such as the no observed adverse effect level (NOAEL) and clinical trial design, we felt compelled to provide additional commentary on some of the authors’ statements and proposed concepts.

The pathologist is faced with many instances where context beyond histopathology has the potential to play a critical role in adversity designation, and inclusion of some of these scenarios in the publication would have provided the reader with a more nuanced and realistic view into this process. Among the histomorphologic findings listed by the authors as adverse, some of which are even described as “clearly adverse,” we felt that several have the potential to be designated nonadverse when associated with a different weight of evidence. Similarly, we felt that several examples from the authors’ list of “nonadverse findings” (eg, adaptive changes, secondary changes, and minor pharmacologic changes) could easily be taken out of context and/or labeled adverse under certain circumstances. A portrayal of true complexity of adversity designations might have been better achieved through inclusion of examples where a histomorphologic finding under one set of conditions might be considered nonadverse and considered adverse when present as part of a different constellation of findings or additional contextual data.

While it is understandable to include neoplasia as a category of adverse findings, some additional framing would have been useful to include in the discussion. Test article–related tumors in short-term repeat dose toxicity studies would indeed represent a major safety concern, but test article–related oncogenicity is an extremely rare phenomenon in short-term studies. Oncogenicity risk is generally studied and determined much later in a dedicated carcinogenicity study, and while technically an NOAEL can come into play for neoplastic findings, these types of studies are focused on an assessment of carcinogenicity risk at various dose levels rather than determination of adversity.

The authors’ proposal to create a new “unclassifiable” adversity category is, in part, understandable since adversity assessments are often very challenging. But it is important to acknowledge that these are precisely the types of assessments that can benefit the most from a pathologist using their experience, training, and all information available. While the degree of certainty may vary, some level of assessment of adversity is often possible from a synthesis of all available data. Knowledge of the target, mechanism, and behavior of other drugs that impact similar targets can also be valuable in this assessment. It is not clear how using the proposed category of “unclassifiable” would be helpful in safety assessment, and we feel it would likely be relegated to a category of “potentially adverse” which many regulatory reviewers equate with adverse to err on the conservative side of hazard identification. Ultimately, the adversity designation may be proven wrong in subsequent studies of longer duration and, although rare, the designation may need to be changed based on new data.

Lastly, we felt compelled to highlight some of the more overly prescriptive statements made by the authors:

“Primary test substance–related necrosis of the myocardium, renal tubules, mucous membranes, skin, adrenal, pituitary, testes, lymphoid organs, and central nervous system might also be considered adverse, regardless of severity grading, in many cases.”

This assertion would better highlight the pathologist’s dilemma if framed in the context of a weight of evidence discussion considering all information. For example, many pathologists would not agree that necrosis of lymphocytes in a lymphoid organ such as the thymus should be unequivocally considered adverse. Additional evidence such as concurrent opportunistic infection(s) suggestive of immunosuppression would also be important to include in such an assessment.

“Focal and nodular hyperplasia could also be considered adverse, in our opinion, due to their potential for progression.”

Findings clearly on a spectrum associated with neoplasia based on literature evidence may be one example where potential progression could come into play, but this is a very rare exception and not the rule. In our opinion, the adversity designation for the examples provided by the authors would greatly benefit from information such as specific organ(s) impacted, degree of cellular atypia, and any literature evidence regarding progression based on the target/mechanism. Most lesions have the potential to progress to the point of adversity if they have the time and capacity to, such as vacuolation, hypertrophy, and inflammation to name a few, but common practice is to limit the adversity determination to the confines of the duration of the study, which is also consistent with the recent STP and ESTP manuscripts. If pathologists speculate on what lesions might become if dosing continues, it becomes difficult to know where to draw the line.

“Mild hormonally mediated changes in the female reproductive tract, accompanied by changes in the estrous cycle, might be considered nonadverse in standard toxicity studies as these effects are generally reversible, do not reflect harm to the animal in the context of the study, and fall into the category of physiological responses. Keratinisation or mucification of the vaginal epithelium, dilation of the uterine lumen, and a decrease in corpora lutea are among some of the nonadverse findings that can occur due to hormonal alterations in general toxicity studies. However, these might be considered adverse in studies designed to assess reproductive performance (Palazzi et al., 2016), highlighting the subjectivity of making adversity calls.”

Reproductive effects can be especially difficult to designate as adverse or nonadverse and deserve special mention when discussing dilemmas in adversity determinations. “Keratinisation or mucification of the vaginal epithelium, dilation of the uterine lumen, and a decrease in corpora lutea” can all occur normally in animals, so one has to be careful designating them as adverse or nonadverse and always assess the overall effect on the reproductive system. Additionally, International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) M3 and S6 both use histologic end points as surrogates for effects on the male and female reproductive system and there are times when histopathology is used in place of a traditional developmental and reproductive toxicology (DART) study, so it is unclear to us why the adversity of a particular reproductive tract finding would be designated one way in a DART study but another in a standard general toxicology study. We strongly feel that the type of study should not play a role in the adversity determination.

“Test substance–related altered foci in the liver are often also considered nonadverse when they occur at a low incidence and/or severity within an animal or dose group or are within background levels.”

As mentioned by the authors earlier in the paper, the first step when assessing a finding is to determine whether it is test item induced or not, and an adversity determination is not performed on findings that are considered to represent normal background variability. The example of liver foci should be clarified, as it provides criteria that are used to determine if a finding is test item related, such as incidence within a dose group, presence of a dose-related increase in severity grade, comparison to background levels, and so on; however, these criteria are typically not used to assess adversity.

Given its subject matter, the article by Gopinath and Mowat is certain to be thought-provoking and to stimulate discussion among toxicologic pathologists. At the same time, it could have provided a more comprehensive depiction of the pathologist’s dilemma by incorporating into the case examples sufficient context and detail to provide a true reflection of the complexity involved in such assessments. The authors’ proposed principles for designations, categorizations, lists, and illustrative snapshots of histopathologic lesions have the potential to be misunderstood, misrepresented, and could be paradoxically detrimental to the careful interpretative process that pathologists regularly utilize. We strongly support the guidelines outlined in previous publications on this topic, particularly the 2016 articles by Kerlin et al and Palazzi et al, which were published with extensive input and review from members across the global pathology societies. Consistent with these articles, we do not recommend attempting to assign adversity based on histopathology in isolation or production of an “atlas of nonadverse and adverse microscopic findings,” but instead favor the use of a case-by-case approach, even if this is an arduous process that doesn’t completely relieve the dilemma that pathologists often find themselves in.