Abstract
There is limited direction in the literature or regulatory guidance on determination of adversity for clinical pathology (CP) biomarkers in preclinical safety studies. Toxicologic clinical pathologists representing the American Society for Veterinary Clinical Pathology—Regulatory Affairs Committee and Society of Toxicologic Pathology—Clinical Pathology Interest Group identified principles, overall approach, and unique considerations for assessing adversity in CP data interpretation to provide a consensus opinion. Emphasized is the need for pathophysiologic context and a weight-of-evidence approach. Most CP biomarkers do not have the potential to be adverse in isolation, regardless of magnitude of change. Rather, they quantify or describe the impact of effects, provide adjunct or supportive information regarding a process or pathogenesis, and provide translational biomarkers of effect. Most often, CP changes are part of a constellation of findings that collectively are adverse. Thus, most CP changes must be interpreted in conjunction with other study findings and require contextual and integrative interpretation. Exceptions include critical CP changes without correlates that indicate a health risk in the tested species. Overall, CP changes should not be interpreted in isolation and their adversity is best addressed with an integrated approach.
Assessment of adversity is one of the most challenging aspects of toxicity study reporting and has been the subject of much deliberation. There is limited literature and regulatory guidance addressing the determination of adversity for clinical pathology (CP) end points in preclinical safety studies. The World Health Organization Core Assessment Group on Pesticide Residues (2015) provides recommendations for assessing test article-related CP changes. Although detailed discussion of these recommendations is beyond the scope of this article, the only change specifically referred to as adverse with these products is an increase in blood methemoglobin (>5% in dogs and >1.5% in rats). Best practice recommendations for assessing adversity have been recently published by the Society of Toxicologic Pathology (STP) Scientific and Regulatory Policy Committee (Kerlin et al. 2016). These recommendations are based on general principles that are well considered and broadly applicable, but are largely focused on anatomic pathology and provide limited guidance for CP. In a recent complementary effort, the European Society of Toxicological Pathology (ESTP) coordinated an international workshop to further characterize the concept of adversity and to frame discussions specific to particular organs or types of lesions (Palazzi et al. 2016). One workshop session of the ESTP working group concluded that CP changes should not be considered adverse in isolation without consideration for associated anatomic pathology findings and adverse in-life outcomes. The current article applies the fundamental principles identified in Kerlin et al. (2016) and Palazzi et al. (2016) to the specific challenges of adversity determination in CP. Here, toxicologic clinical pathologists representing the Regulatory Affairs Committee (RAC) of the American Society of Veterinary Clinical Pathology (ASVCP) and the Clinical Pathology Interest Group (CPIG) of the STP delineate overarching principles and unique considerations for assessing adversity in CP data interpretation. The intent is not to provide a comprehensive review of all potentially adverse CP changes. Rather, it is to describe the basic principles, overall approach, and unique considerations for assessing adversity in CP data interpretation using specific examples.
While the objective of preclinical or general toxicity studies is to define a safe starting dose for first in human studies and/or to inform human safety risk assessment, the adversity determination for CP changes in preclinical studies is based on the context of the study under consideration, and thus, for the species in which these changes occur. There are many definitions of adversity in the toxicologic pathology literature (Integrated Risk Information System [IRIS] 2007; Renwick et al. 2003; Sergeant 2002; Eaton and Gilbert 2008; Kerlin et al. 2016; Hall et al. 2012; International Programme on Chemical Safety 2004; Keller et al. 2012; Lewis et al. 2002; Palazzi et al. 2016). For the purpose of this article, the current ESTP definition is used: In the context of a preclinical toxicity study, an adverse effect is a test item-related change in the morphology, physiology, growth, development, reproduction or life span of the animal model that likely results in an impairment of functional capacity to maintain homeostasis and/or an impairment of the capacity to respond to an additional challenge. (Palazzi et al. 2016, 3)
Adversity of CP Changes Is Based on Weight of Evidence and Requires Consideration of Pathogenesis
In most cases, CP changes are not harmful in and of themselves and are not the sole determinant of the no adverse effect level (NOAEL) for a study. When CP changes are not inherently adverse, assessment of adversity may be based on a pattern of changes and their pathophysiologic implications. Some CP changes are indicators of altered function, and as such might be considered adverse. For example, as stand-alone findings, high serum urea (conventionally known as blood urea nitrogen [BUN]) or creatinine concentrations have no harmful effects on the animal. However, BUN or creatinine increases may be considered to be markers for an adverse effect when associated with evidence of renal injury or dysfunction. Increases in BUN and creatinine concentrations are common in toxicity studies and indicate decreased glomerular filtration rate (Stockham and Scott 2008a, 429–33). These increases are usually due to prerenal causes (decreased renal perfusion associated with hemoconcentration or dehydration) or renal causes (loss of urine concentrating capacity). In the absence of histopathologic evidence of renal injury, increases in BUN and creatinine concentrations associated with low urine volume and high urine specific gravity and coupled with clinical signs indicative of dehydration would likely be attributable to prerenal azotemia. Collectively, this pattern would be considered to indicate the severity of dehydration but would be nonadverse. In contrast, increased BUN and creatinine associated with changes in urine volume such as oliguria, anuria, or polyuria, with low urine specific gravity, and morphologic findings indicative of renal injury would be related to renal disease and the totality of these findings would in most cases be considered adverse.
Adversity designations may be based on the context or cause of the change. As the most abundant plasma protein, albumin is largely responsible for producing the oncotic pressure that keeps fluid within the vascular system. Severe hypoalbuminemia results in loss of oncotic pressure causing edema and ascites due to accumulation of fluid in interstitial spaces (Mosier 2007). Marked or severe decreases in albumin associated with clinical edema would be adverse, irrespective of mechanism. However, the designation of adversity may be less clear and contingent upon the cause when edema is not clinically evident. For example, even less pronounced decreases in albumin may contribute to an adversity determination when associated with adverse mechanism such as reduced synthesis due to liver failure.
Although CP results are rarely the sole determinants of the NOAEL for a study, they are often useful in quantifying and describing the impact or identifying the underlying pathogenesis of an adverse finding. For example, increased serum alanine aminotransferase (ALT) activity is a relatively tissue-specific indicator of hepatocellular injury and is currently the most universally accepted liver injury marker (Amacher 1998; Kim et al. 2008). As a stand-alone finding, a high serum ALT activity has no harmful effects on the health of the animal. However, an ALT increase may be considered a marker of an adverse effect when associated with light microscopic evidence of hepatocellular injury. While an increase in ALT is generally considered indicative of hepatocellular injury and is often associated with histopathologic findings such as hepatocellular necrosis, increases in ALT can occur in the absence of hepatocellular injury due to a number of conditions unrelated to liver injury (Valentine et al. 1990; Travlos et al. 1996; Boone et al. 2005; Yang and Gong 2009; Radi et al. 2011). Thus, other CP indicators of hepatocellular and hepatobiliary injury or diminished liver function, microscopic observations in liver and other tissues, dose response, and study duration should be considered in the collective determination of adversity.
Rapidity of Onset and Persistence with Repeated Dosing Can Impact Adversity
In some cases, the rapidity of onset of a CP change or persistence of that change with repeated dosing can be considered adverse, reflecting insufficient time, or the inability to sufficiently compensate. The rapidity of change may not be determined when CP is only evaluated at the end of dosing, but it can be evaluated in studies where multiple CP time points are collected. Acute and severe changes in critical CP biomarkers such as electrolytes (e.g., potassium) may be considered as a potential cause of sudden death; however, these are not likely to be measured or documented prior to moribundity or death. In contrast, slowly progressing and persistent changes of similar magnitude could result in fewer clinical signs. A severe decrease in red blood cell mass (hemoglobin, hematocrit, and red blood cells) could be adverse due to functional consequences on tissue oxygenation (Ness and Kruskall 2005; Ettinger and Barrett 1995). Adverse clinical signs due to acute tissue hypoxia such as weakness, fainting, or death could occur after an acute decrease in red blood cell mass (Kerl and Hohenhaus 1993; Redondo et al. 1995). In contrast, clinical signs associated with a gradual decrease in red blood cell mass to a similar degree could be limited to mild lethargy due to adaptive responses (Péchereau, Martel, and Braun 1997). As another example, hypoglycemia may occur in response to drug treatment, either as an exaggerated pharmacologic response or off-target effect (Pettersen et al. 2014). The spectrum of clinical signs is dependent on the rapidity of onset and severity of hypoglycemia. Severe, rapid onset hypoglycemia can lead to seizures, coma, and death (Guyton and Hall 2006; Rana et al. 2011; Madey, Hannah, and Lazaridis 2013; Leow and Wyckoff 2005). In contrast, prolonged hypoglycemia, even if not severe, may result in neuronal degeneration (Mohseni 2001). While clinical signs differ between these manifestations of hypoglycemia, both effects would be considered adverse.
Few CP Biomarkers Have the Potential to Be Adverse in Isolation
Under some circumstances, an isolated CP change may be adverse. This most commonly applies to tightly regulated CP biomarkers that must be maintained within a specific range for homeostasis or those for which an excursion from the normal physiologic range would have detrimental effects on critical cellular, organ, or systemic functions such as hemostasis, immunity, oxygen tension, acid/base status, oncotic or osmotic pressure, and neuromuscular signal transduction. A substantial change affecting any of these systems has the potential to be adverse and results in clinically observable effects including mortality. Examples include marked/severe decreases in platelet count, neutrophil count, hemoglobin, glucose or albumin concentration, and electrolyte alterations, some of which are discussed throughout this article. Assessment of adversity of these CP changes is facilitated by general guidance in the literature for some of these CP biomarkers, which associate the potential for adverse outcome with the magnitude of change (Ness and Kruskall 2005; Russell 2010; Johnson, Thompson, and Calia 1985; Ettinger and Barrett 1995). This is illustrated in the following examples.
Profoundly decreased platelet counts can be considered adverse due to the increased likelihood of spontaneous bleeding or uncontrollable bleeding after injury (including venipuncture). Markedly to severely low platelet counts with evidence of hemorrhage and/or decreases or maturational changes in bone marrow megakaryocytes support the determination of adversity. However, clinical signs of bleeding or changes in bone marrow are not required for low platelet counts to be considered adverse, since thrombocytopenia of a sufficient magnitude increases the risk of bleeding. Platelet counts of ≤50,000/µl in dogs have been associated with a greater propensity for bleeding (Russell 2010). Animals with platelet counts of <10,000/µl (dogs) or <25,000/µl (mice) are at an even greater risk for bleeding (Morowski et al. 2013). Similar values in rats and nonhuman primates (NHP) are likely applicable, although not well-documented in the literature. Therefore, even in the absence of clinical hemorrhage, low absolute platelet counts can be considered adverse in light of consistency and/or time course and based on the likelihood that the animal will be unable to maintain hemostasis in the face of vascular injury.
A marked to severe decrease in neutrophils is another instance in which adversity can be determined based on severity of the change without the presence of concurrent adverse sequelae. The lower the absolute neutrophil count in dogs, the greater the likelihood of infection (Giguere 2013). In dogs with neutrophil counts of 500 to 1,000/µl, there is only a moderate risk of infection, but the risk of infection gets higher as the neutrophil count gets below 500/µl, and the risk of infection is very high at <200/µl (Dinauer and Coates 2005). Marked to severely decreased neutrophils in rats (<50/µl) and monkeys (<500/µl) may lead to increased risk of infection. Although the association in these species is not well-documented, one study showed that rats with neutrophil counts <50/µl were susceptible to bacterial challenge (Johnson, Thompson, and Calia 1985). In summary, when the risk of increased infection is considered high based on absolute values, consistency, and/or time course, severely decreased neutrophils may be considered adverse without associated findings. Concurrent decreased bone marrow myeloid cellularity, fever, and/or histopathologic evidence of infection, although not essential to make a call of adversity, would clearly add additional weight of evidence.
In some cases, increases in CP biomarkers such as cardiac troponin I (cTnI) can reflect injury to tissues associated with heightened concern. For example, markedly increased cTnI levels may be considered adverse due to the specificity for myocardial injury and short circulating half-life of cTnI (Clements et al. 2010), even in the absence of an identifiable cause such as microscopic evidence of myocardial necrosis or supraventricular tachycardia (Kanjwal et al. 2008).
In conclusion, for the instances when a CP change could be considered adverse in relative isolation, the determination is based on a combination of severity (absolute value and magnitude of change), rapidity of onset, persistence, and likelihood of life-threatening consequences. Although the change is potentially adverse in isolation, the determination of adversity will typically be further supported by other study findings.
A Designation of Adverse or Nondverse May Not Be Necessary
Analytic Constructs, Calculations and Artifacts
Analytic constructs, ratios, and other calculations, such as certain red blood cell indices and albumin/globulin ratios, are descriptive of a process and inform on the context but are not useful for assessing adversity. For example, a change in albumin to globulin (A:G) ratio would not be considered inherently adverse because it is an outcome of changes in albumin and globulin that are considered independently for adversity determination.
Large unstained cells (LUC) are another example of an analytic construct. The classification LUC includes nucleated cells in an area of an automated hematology analyzer scatterplot based on large size and low peroxidase activity (Moritz and Becker 2010). LUCs can be monocytes, basophils, large or reactive lymphocytes, plasma cells, or immature cells including blast cells of various lineages (Lilliehook and Tvedten 2011; Meintker et al. 2013; Gibbs 2014). The finding of blast cells resulting in increased LUCs in animals with test article-related leukemia is adverse, but the adversity determination is made on the constellation of changes in hematology and histopathology, rather than on LUC counts.
Since only test article–related changes should be considered when assessing adversity, a potentially adverse CP change would not be considered adverse if it is determined to be due to artifact and thus not test article related. For example, preanalytic or analytic effects can influence the magnitude of test article–related changes and produce results that would be consistent with adverse values. Preanalytic considerations include factors involved with blood sampling (time of sampling, anesthesia, stress due to restraint, fasting status, or feeding schedule), transport, and storage of blood samples (duration, refrigeration, freezing; Braun et al. 2015; Schultze and Irizarry 2016; Everds 2016). Knowledge of the biologic variability and preanalytic and analytic variations relative to the inherent total error of a particular test method is typically considered in the final assessment of the magnitude of observed changes. For example, when there is a test article-related decrease in red cell mass within an NHP study, concurrent menstruation may magnify the effect because blood loss due to menstruation can reach significant levels in individual females (to −33% baseline; Perigard et al. 2016). In this case, determination of adversity may require integration of specific in-life observations and/or documentation of menstrual cycling in female monkeys.
Sodium citrate tubes are commonly used for coagulation testing. A proper anticoagulant to blood ratio (1:9) must be maintained for accurate results, and sample hematocrit may affect the citrate concentration (O’Brien, Sellers, and Meyer 1995; Stockham and Scott 2008b, 276–77). Prolongations in coagulation times (prothrombin and activated partial thromboplastin times) may indicate a systemic hypocoagulability, and depending on the severity of the prolongation and associated clinical signs, may or may not be considered adverse. However, prolongations in coagulation times may also be observed in conjunction with significant dehydration or when there is a decrease in plasma relative to red cell mass (i.e., increased hematocrit) due to overcitration of a smaller volume of plasma in both cases. In these situations, prolongation in coagulation times is due to an altered blood to citrate ratio ex vivo (i.e., is not indicative of systemic hypocoagulability), and thus no adversity call is necessary.
CP Changes That Are Part of a Spectrum of Morbidity Should Not Be Designated as Adverse or Nonadverse
Some of the most pronounced CP changes occur in animals that are euthanized in poor clinical condition. CP changes can be confounded by nonspecific effects related to morbidity/moribundity such as poor sample quality and difficult phlebotomy due to dehydration, low blood pressure, stress, immune system dysregulation, multiple organ failure or homeostatic, hemostatic and acid–base disturbances (Hall 2013; Everds et al. 2013; Everds 2015). In general, CP changes associated with moribundity should be listed in the report in the appropriate pathophysiologic context and should be described as secondary to the clinical condition of the animal, which will inform the adversity decision.
Practical Implications for Reporting Adversity for CP
The principal goal of adversity determinations in preclinical studies is to identify health risks or hazards and establish the NOAEL. Ultimately, this is used to communicate the relevant findings in animal toxicity data for predicting potential human outcomes to physicians and regulators (Kerlin et al. 2016). Test article–related effects and the study NOAEL are communicated in the summary and conclusion sections of integrated toxicology reports, which are usually authored by study directors with input from the relevant contributing scientists including pathologists. Therefore, it is critical that pathologists’ interpretations regarding adversity be clearly and accurately communicated to the study director. Excessively detailed justifications, particularly for obviously adverse or nonadverse findings, make reports too long and complex and risk obscuring the more important points. Adversity of CP changes should only be addressed in the CP report if it makes sense to do so, that is, if the adverse CP change contributes to determination of overall NOAEL for the integrated toxicology study report.
Stand-alone CP subreports should clearly identify all test article–related changes. Once CP effects are determined to be test article related, their potential for adversity may be assessed. However, without associated clinical/antemortem and anatomic pathology data, adversity should only be addressed in the subreport when it is clearly linked to a clinical outcome in the study animals or a high risk for an adverse clinical consequence. Clearly adverse CP changes should be addressed in the CP report and should include a discussion of magnitude, incidence, and correlation with adverse clinical course or anatomic pathology findings. Changes that raise concern for adversity or potentially adverse CP changes may also be described and discussed in the stand-alone CP report to help the study director with building a weight of evidence for establishing the NOAEL.
Rationale for an adversity call or the lack thereof may be provided, as appropriate, to support the designation of adversity for CP biomarkers that have a potential to be adverse. This is most appropriate for changes that are not clearly adverse or those that could potentially be considered adverse. Explanatory material may consist of generic statements (e.g., nonadverse based on magnitude/severity or direction of change, lack of change in correlative end points), or specific statements (e.g., adverse because of the magnitude of platelet decrease and the associated potential for hemorrhage), and may include supportive references from the literature. However, use of references that are not specifically applicable for the study in question may encourage the reliance on inappropriate literature to justify decisions on adversity. References may provide support for remarks about pathogenesis (U.S. Food and Drug Administration 2000). However, adversity designations should primarily be based on the scientific judgment of those with expertise in the relevant disciplines. Statements of rationale may be helpful in the support of the designation of adversity but are not required for changes that are clearly adverse or nonadverse or where adversity is simply not relevant.
Although study scientists identify NOAELs based on their interpretation of findings on a single study, determination of the safe starting dose of a new chemical or biological entity for first in human studies is based on a consensus between sponsors, clinicians, and regulatory agencies, regarding the findings from multiple studies. In preclinical toxicology studies, determination of the NOAEL is specific for the context of each study; however, species differences can indirectly influence human risk assessment. For example, there are profound differences in lipoprotein metabolism across species (Yin et al. 2012; Bergen and Mersmann 2005); a marked increase in total cholesterol in rats would not be considered adverse, as the finding generally reflects high-density lipoprotein (HDL) as the predominant cholesterol carrying lipoprotein in the rat, and rats are resistant to atherogenesis (Bauer 1996). Similarly, a cholesterol increase in a dog study could also be considered nonadverse for similar reasons, as dogs rarely develop arteriosclerosis in the absence of hypothyroidism when massive elevations of serum cholesterol are seen (La Perle and Capen 2007; Van Vleet and Ferrans 2007). Although nonadverse in either rat or dog, test article-related total cholesterol changes in preclinical studies may identify a potential hazard for human safety. Thus, even nonadverse preclinical study findings may help inform human safety risk assessment.
Conclusion
In preclinical general toxicology studies, the most common end points for identifying adverse effects include mortality, microscopic findings, clinical signs, and body weight changes. Microscopic findings and clinical observations often define the NOAEL and histopathology is usually key for identification of target organ toxicity. End points such as CP, food consumption, organ weights, and macroscopic findings add to the weight of evidence and may provide a mechanistic link for test article-related effects without being adverse by themselves.
To provide direction on the determination of adversity for CP biomarkers in preclinical safety studies, toxicologic clinical pathologists representing the ASVCP RAC and the STP CPIG identified principles, overall approaches, and unique considerations for assessing adversity in CP data interpretation. A comprehensive review of all potentially adverse CP changes was not endeavored. As discussed, there are no universally applicable rules or specific decision limits for designating adversity of CP biomarkers. CP changes should be addressed on a case-by-case basis using the fundamental concepts and principles described. A designation of adversity should be made in the CP subreport if it assists in data interpretation. This is most relevant for CP biomarkers that have the potential to be adverse, that is, those that are maintained within critical physiologic ranges from which excursions could be life threatening. In a few instances, changes may be clearly adverse based on the mechanism involved, the rapidity or absolute magnitude of the change, or the resultant immediate critical effects in the organism. In these instances, CP may provide the primary basis for NOAEL and adversity determination, and providing a justification for this adversity designation may be necessary in CP subreports. Factors to consider include incidence and frequency, magnitude of change relative to control or baseline values, mechanism, severity and rapidity of onset of the change, influence of preanalytical (biological and procedure related) effects, and likelihood of an adverse consequence.
Most CP biomarkers do not have the potential to be adverse in isolation. Rather, they are primarily used to provide mechanistic insight or support adversity decisions in the context of other study findings. As such, they should be addressed as part of a constellation of findings that are collectively deemed adverse. The weight-of-evidence approach for CP data interpretation described herein is consistent with most definitions of adversity, which are based on potential detrimental effects to the entire organism, rather than individual changes.
This article represents a consensus opinion among industry clinical pathologists and has been endorsed by members of the ASVCP RAC and STP CPIG.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.