Serum Chemical Biomarkers of Cardiac Injury for Nonclinical Safety Testing

Abstract

There has been a substantial increase in the number and efficacy of laboratory biomarkers for the evaluation of human cardiac injury over the last decade. The advantages of these over traditional laboratory tests have encouraged adoption of comparable markers in nonclinical studies for cardiac injury assessment. Of particular interest are markers that are not only more sensitive and/or specific than traditional parameters for cardiac injury, but also those that may directly bridge human and laboratory animal safety testing. However, a majority of recently developed markers are quantified through antibody-based assays, and cross-reactivity with the comparable analyte in nonhuman samples is difficult to predict and often species-variable. The utility of these novel biomarkers thus, depends upon adequate assay validation with each laboratory species of interest. In contrast, traditional laboratory parameters of cardiac injury lack the properties of an ideal biomarker, but are well established and have an extensive database in nonclinical safety assessment. The current status and utility of both recently developed and traditional biomarkers of cardiac injury in nonclinical testing, and considerations for validation of novel biomarkers of cardiac injury are reviewed.

Introduction

The need for noninvasive, specific and sensitive biomarkers for the evaluation of cardiac injury has been long-standing in both human clinical, and laboratory animal safety testing. For laboratory animal safety testing, such markers also must be applicable to multiple species and directly relevant to the clinical setting to gain broad utility. Traditional serum biomarkers of cardiac injury for laboratory animals have deficiencies in virtually all of these areas, with poor cardiac specificity and sensitivity, inconsistent behavior across species and limited relevance to clinical testing methods. Because of these deficiencies, nonclinical safety evaluation of cardiac injury has historically relied nearly exclusively on functional, and postmortem findings to predict human risk rather than laboratory testing. This practice may now be shifting, however, due to recent evidence on the limitations of these nonlaboratory methods alone for predicting human risk, and the expanding number of clinical laboratory markers of cardiac injury that have shown promise as comparable markers in animal models.

As with all laboratory parameters, novel markers to assess cardiac injury in laboratory animals must undergo rigorous analytical and biological validation before being accepted as standard markers of myocardial conditions. Analytic validation assesses the competence of a specific assay with exemplary test specimens, while biological validation requires appropriate animal models and reference endpoint(s) to define the in vivo utility of the parameter. Both of these validation steps require considerable resources and thus, new biomarkers selected for evaluation of cardiac injury in laboratory animal safety testing should be those likely to most improve upon traditional nonclinical evaluation of cardiac injury.

While traditional nonclinical serum markers of cardiac injury have significant limitations when assessed in isolation, they also have the advantage of a large historical database on their behavior in laboratory species and can still provide useful ancillary information pertaining to cardiac injury in safety assessment. In particular, the isoenzymes of lactacte dehydrogenase and creatinine kinase in combination with more specific markers of cardiac injury, such as the cardiac troponins, can provide information on the relative severity, extent or duration of myocardial injury. A multimarker approach with inclusion of both traditional and validated novel parameters of cardiac injury may thus, have benefit with second tier nonclinical investigations of cardiac effects in laboratory animals.

The following discussion is a review of the utility and deficiencies of both traditional, and more recently recognized laboratory biomarkers of myocardial injury for evaluation of laboratory species. The considerations for validation of novel markers of cardiac injury for laboratory animal testing, using the cardiac troponins as specific examples, are also outlined.

Adapting Clinical Biomarkers for Preclinical Testing of Cardiac Injury

While the concept of a bridging biomarker seems straight forward, there are some expectations of a model biomarker for cardiac injury that directly contrast between clinical and nonclinical testing. Clinical biomarkers of cardiac injury are chosen, in part, to address the priorities of high diagnostic sensitivity and relatively rapid clearance to enable early diagnosis and close monitoring of treatment success. In contrast, with nonclinical evaluation, biomarker specificity is of high priority because of the limited number of pertinent endpoints that can be feasibly evaluated, and biomarker circulation time is of sufficient duration to allow for limited collection intervals. In addition, the “gold standard” reference for validation of a biomarker of cardiac injury with clinical medicine is most often a clinical outcome or functional endpoint; while light microscopic findings are the reference with nonclinical testing. These differences in reference standards need to be considered when establishing criteria for biologic validation of laboratory parameters as bridging biomarkers of cardiac injury.

Validation of a clinical parameter for nonclinical testing also requires establishing comparable analytical behavior of an assay with the homologous marker in animal specimens. Clinical assays for biomarkers of cardiac injury are being increasingly based upon immunologic methodology and currently, a majority of these assays utilize proprietary reagent antibodies that recognize assay-specific epitopes and report in nonstandardized units. Cross-reactivity of these reagents with the nonhuman antigen may be incomplete or absent. Specific information on cross-reactivity of the reagent antibodies with homologous marker in animals is often not available and, if available, commonly not helpful in determining actual assay reactivity with nonhuman samples. Instead each immunoassay that incorporates different antibody reagents and methodology requires full analytical validation for each species to be tested before the assay results can be reliably interpreted.

Nonspecific Variables in Study Design and Results Interpretation

Several nonspecific variables can significantly and/or commonly impact the results of cardiac-associated biomarker evaluation in laboratory animals and need to be considered with study design and data interpretation. In particular, due to the limited circulating half-life of many cardiac injury biomarkers, timing of sample collection can be critical. Because, frequent sampling may be a limited option particularly with rodents studies, collection intervals should be established based on available information on the type or onset of expected cardiac injury, as well as the behavior of the analyte in the test species. The difficulty in predicting optimal timing of sample collection without this information is one of the major objections to including cardiac injury markers into routine nonclinical testing.

Blood collection method is also an important variable, and should be consistent within a study, and as atraumatic as possible. The extent and means of animal handling for blood collection as well as dosing, and the type of anesthesia used can affect values of some biomarkers of cardiac injury (Friedel et al., 1975; Walter, 1999; Yerroum et al., 1999; Aartsen et al., 2000; Kim et al., 2005). Cardiac puncture for specimen collection readily results in artifactually high muscle enzyme activity and cardiac troponin concentrations (O’Brien et al., 1997a). Additionally, concentrations of cardiac injury biomarkers can be altered in samples collected at necropsy or from deeply sedated animals as a result of generalized hypoxia rather than an in-life effect.

General toxicity or exaggerated pharmacology that impairs cardiovascular homeostasis can also alter cardiac biomarker results. Conditions such as electrolyte imbalance, blood volume depletion, endotoxemia, uremia, seizures or direct sympathetic effects can secondarily alter cardiac function and cardiac biomarker results (Fredericks, 2002; Wallace et al., 2004). Age has been found to be a critical variable in rat studies in that older animals can show greater incidence of background cardiac pathology and greater sensitivity to cardiotoxic effects of a test compound. Fasting has also been reported to affect nonspecific markers of muscle injury (Baetz and Mengeling, 1971). Thus, whether animals are fasted or not prior to specimen collection should ideally be kept consistent between collection intervals. Rodent strain may be a consideration in the design of long-term studies that include evaluation of cardiac injury biomarkers due to strain differences in incidence of background myocardial lesions and sensitivity to cardiotoxicants (Ruben et al., 1984; Kemi et al., 2000; Faulx et al., 2005).

Enzymes of Muscle Injury

Lactate dehydrogenase, creatinine kinase and aspartate aminotransferase (AST) catalyze anaerobic or aerobic metabolic reactions in cells. Their enzyme activity is typically analyzed by automated kinetic assays with standardized reference material. Therefore, results are expected to be comparable between laboratories, despite differences in assay methodology.

While these analytes have this advantage, their measured serum activity has overall poor specificity and sensitivity for cardiac tissue in animals, as in humans. Serum activity demonstrates a large dynamic range in all laboratory species and most commonly reflects release from noncardiac tissue. Reference ranges are generally without benefit, in part, as there is no cutoff point for a clearly abnormal serum value. Serum activity of these enzymes is also considered to reflect the mass of the tissue source, which is greater with skeletal muscle compared to cardiac muscle. Thus, routine handling and restraint, as well as intramuscular injections and vigorous activity are commonly associated with transient high serum levels presumably due principally to skeletal muscle release. However, assessment of total serum lactate dehydrogenase and creatinine kinase actvity is required for evaluation of their isoenzyme distribution that can have more specific utility in the evaluation of cardiac injury.

Lactate Dehydrogenase (LD)

Lactate dehydrogenase is found within the cytoplasm of all cells, including non-nucleated erythrocytes and cytoplasmic fragments, e.g., platelets (Friedel and Mattenheim, 1970, Chow et al., 1977). Total LD activity varies widely in different tissues and for each species, likely in association with the metabolic activity of the tissue (Baba and Sharma, 1971; Yasuda et al., 1990). For example, skeletal muscle has more LD activity on a gram wet weight basis than cardiac muscle in the rats, whereas the opposite ratio has been reported for dog (Clampitt and Hart, 1978; Keller, 1981; Boyd, 1983). Total LD activity in these different tissues has also been shown to vary within a species with age, gender, and skeletal muscle group or location within the heart muscle (Baba and Sharma, 1971; Bogatskaya and Litoshenko, 1975; Keller, 1981). These complex variables in tissue distribution are among several factors that likely influence total serum LD activity.

Circulating half-life of the enzyme also influences measured serum activity. However, no general pattern characterizes the serum half-life or kinetics of total LD activity with cardiac injury in nonclinical studies. Broad species differences in half-life of each LD isoenzyme, and isoenzyme tissue distribution likely profoundly affect LD kinetics with any tissue injury and are discussed in more detail in that section (Boyd, 1983). However, these species differences in tissue distribution and circulating half-life indicate that LD lacks key properties of a biomarker for bridging clinical and nonclinical testing.

With both clinical and non-clinical testing, specimen handling can significantly influence LD activity. Because of inhibition by chelating agents, serum rather than plasma is generally recommended for analysis (Otto and Birkenmeier, 1993; Moss and Henderson, 1999). Because erythrocytes, platelets and leukocytes can also contribute to sample enzyme activity, hemolysis and delays in separating serum from the clot can affect values obtained (Friedel and Mattenheim, 1970). These handling concerns may be particularly important with nonclinical testing due to higher total LD activity relative to humans in, and diverse LD isoenzyme content between, common laboratory species in peripheral blood cellular elements (Friedel and Mattenheimer, 1970; Chow et al., 1977). While some human LD isoenzymes have been found to be labile when exposed to refrigeration temperatures, the extent to which this applies to laboratory species samples is unclear. Stable total LD and isoenzyme values in frozen serum samples has been reported or observed for some nonhuman species (Preus et al., 1989). However, some laboratories may prefer fresh, nonrefrigerated samples for isoenzyme analysis due to the potential for differential isoenzyme liability (Tanishima et al., 1995).

LD Isoenzymes

Evaluation of serum LD isozymes is usually performed to increase analyte tissue-source specificity, and most commonly, cardiac-specificity. Lactate dehydrogenase is a tetrameric protein of 2 different subunits, termed the M subunit for skeletal muscle and the H subunit for heart. The DNA or cDNA encoding these subunits are termed LD-A and LD-B, respectively (Liao et al., 2001; Rossignol et al., 2003). Five major LD isoenzymes are recognized in the tissues of essentially all laboratory mammals. The predominant isoenzymes in the heart are LD₁ (with 4 H subunits) and LD₂ (3 H and 1 M subunit). The isoenzyme with the greatest proportion of M subunits, LD₅, predominates in skeletal muscle and liver of most, but not all laboratory species (Yasuda et al., 1990). Most other tissues contain a broader distribution of the different LD isoenzymes.

Although the predominant isoenzymes in the heart are similar among laboratory species, the proportional activity of LD₁ and LD₂ in cardiac muscle and other tissues (on a gram wet weight basis) varies considerably between species (Yasuda et al., 1990). Similarly, the proportion of LD₁ and LD₂ activity relative to the other isozymes in serum in health is widely disparate between species (Preus et al., 1989). For example, in the rat and dog, serum LD₁ activity is overshadowed by the activity of the other measured isoenzymes; while in the Cynomolgus monkey, rabbit and pig, the serum LD isoenzyme profile is more closely aligned with that of humans with predominant activity attributed to LD₁. These differences in serum and tissue activity of the isoenzymes in health influence the utility and/or sensitivity of LD isoenzyme analysis among different laboratory species in the assessment of cardiac injury.

Isoenzyme analysis of laboratory animal specimens is currently most appropriately done by electrophoresis or ion exchange, with the former being more common. There are various other methods available for clinical use, including an automated kinetic assay using hydroxybutyrate, a favored substrate of human LD₁ and LD₂ isoenzymes. However, these alternative kinetic, or immunologic assays are highly unreliable with laboratory animal specimens (Negroni et al., 1982; Barrett et al., 1988). Yet, electrophoretic methods are non-standardized with variable use of agar, agarose or cellulose acetate gels, and different pH, temperature and voltage conditions. These differences in methodology, as well as physiologic and pathologic differences between animals and most profoundly, species differences influence the serum LD isoenzyme profiles obtained with electrophoresis. Therefore, comparison of profiles between laboratories and studies can only be very general.

The units used for comparisons between groups in isoenzyme profiles, particularly for the cardiac-predominant isoenzymes also vary. Comparative units reported with nonclinical studies have included: absolute LD₁ activity, percent LD₁, LD₁ relative to LD₂ activity (LD₁/LD₂), or even LD₁ relative to hydroxybutyrate dehydrogenase activity. Because total serum LD activity demonstrates wide interindividual variability in health and absolute values are magnified by small differences in percentages, the most common and probably most reliable comparative unit is percent activity of each isoenzyme relative the total. Absolute LD activity for each specimen must also be known in order to interpret and understand the significance of these percentages. However, other comparative units might be appropriate for some animal models.

For all laboratory species, increased relative serum activity in one or both of the cardiac-predominant isoenzymes, LD₁ and LD₂, is considered generally supportive of myocardial injury. However, these isoenzymes are not specific for cardiac injury; hemolysis, as well as renal, pancreatic, liver or gastrointestinal sources may contribute to their serum activity in some species. There is also no common pattern that characterizes the kinetics of LD isoenzyme activity associated with cardiac injury in laboratory animals. The occurrence, peak, and persistence of increases in LD₁ and/or LD₂ have been shown to vary with the type, location and severity of myocardial injury (Wolf et al., 1986; Ladi et al., 1990).

Broad variation in circulating half-life of these two isoenzymes between species has also been found, with a 12 to 15-fold difference in serum half-life for each isoenzyme between common laboratory species (Boyd, 1983). Thus, although LD₁ has been found to have the longest circulating half-life of the LD isoenzymes for all species evaluated, reported half-lives range from a few hours to days. Within a species, although the half-life of LD₁ in serum has generally been considered to be longer than that of the CK isoen-zymes, this is far from consistent in studies of cardiac injury in laboratory animals (Wolf et al., 1986; Bertinchant et al., 2000). Serum LD isoenzyme analysis may also not clearly identify the extent of increases in LD1 and LD2 activity due to relative changes in other isoenzymes. For example, LD₅ has relatively high activity in liver and with cardiovascular system compromise, liver-derived increase in this isoenzyme may mask potential effects on LD₁ or LD₂ activity (Ljubuncic et al., 1992).

For the many reasons given above, the analysis of serum LD isoenzyme distribution cannot confirm or negate the presence of cardiac injury in preclinical studies. However, as part of a panel of cardiac injury markers, comparison of LD isoenzyme profiles within a study has shown utility in evaluation of animal models of cardiac ischemia, cardiotoxicity and cardioprotection. Relative shifts in the serum LD₁/LD₂ ratio has been shown to correlate with severity and duration of cardiac injury in some of these models (Wolf et al, 1986; Preus et al., 1988; Ljubuncic et al., 1992). Evaluation of LD isoenzyme shifts within homogenized myocardial tissues has also shown utility in defining chronic cardiac injury, including chronic cardiotoxicity and pharmacologic cardioprotection (Jelinkova et al., 1970; Awaji et al., 1990; Muders et al., 2001). Thus, LD isoenzyme profile analysis may aid in better characterizing a cardiac injury that is confirmed by more specific preclinical testing methods.

Creatinine Kinase (CK)

As with LD, CK is predominantly found in the cytoplasm; within myocytes, the enzyme occurs in close association with the sarcoplasmic reticulum, mitochondria and myofibrils. In contrast with LD, CK is relatively tissue specific with activity predominantly in skeletal and cardiac muscle in both humans and laboratory animals (Boyd, 1983; Moss and Henderson, 1999). Variable, but much lower CK activity in the brain, intestinal smooth muscle, kidney and platelets of evaluated laboratory species has been reported (Boyd, 1983; Aktas et al., 1995, 1994). CK activity is also greater in skeletal muscle than cardiac muscle, and reported to be greater by approximately 2-fold in dogs, 3- to 4-fold in humans and Cynomologus monkeys, and 3 to 7-fold in rats relative to cardiac muscle on a wet weight or protein basis (Boyd, 1983; Aktas et al., 1993; Loeb, 1999; Moss and Henderson, 1999). These species differences in tissue activity presumably influences species differences in serum CK activity following skeletal or cardiac muscle injury.

The circulating half-life of CK is also more limited than that of LD in absolute range between laboratory species and between isoenzymes. The serum half-life of total CK is approximately 0.5 to 1 hour in rats, 2 to 3 hours in dogs and up to 9 hours in rabbits (Friedel et al., 1976; Boyd, 1983; Aktas et al., 1993, 1995; Lefebvre et al., 1993). The half-life is slightly longer in humans at approximately 12 hours (Janssen et al., 1989; Moss and Henderson, 1999). The circulating half-life of the skeletal muscle-predominant isoenzyme is only slightly longer than that of the cardiac-associated isoenzyme in evaluated species (Aktas et al., 1995). Thus, with acute skeletal or cardiac muscle injury in all laboratory animals, an increase in serum CK activity is expected to be relatively short in duration, especially in rodents. With all laboratory species, sporadic and transient high serum CK activity among individual animals in association with animal handling, intramuscular injections, or surgical procedures are well documented (Wolf et al., 1986; Bertinchant et al., 2000). In general, because of the unavoidable variable of animal handling, combined with greater skeletal muscle concentration and mass relative to that of myocardial tissues, serum total CK activity is more reliably a biomarker of skeletal, than cardiac muscle injury in non-clinical studies.

Specimens for CK analysis should be processed quickly and stored well-sealed and away from light. Enzyme activity in samples from some species can degrade significantly over a few hours if samples are stored at room temperature, in bright light or in loosely capped tubes (Buttery et al., 1992). Creatine kinase total and isoenzyme activity is stable is samples subjected to short-term refrigeration or freezing, although some laboratories evaluating isozymes may prefer fresh samples to avoid any possible differential isoenzyme degradation with storage (Preus et al., 1989; Ercan and Grossman, 2003).

CK Isoenzymes

Serum CK isoenzyme analysis is most commonly performed to amplify analyte specificity for cardiac muscle, although other indications have been reported (Delahunty, 1984; Gupta et al., 1991). The enzyme is a dimer of the same, or two genetically distinct subunits termed M-type for skeletal muscle and B-type for brain. At least 3 isozymes have been identified in all evaluated mammalian species: CK-MM (or CK-3), CK-MB (CK-2), and CK-BB (CK-1). Mitochondrial CK (CK mt) has also been detected in serum electrophoretic patterns of some species (Ishikawa et al., 1997; Hironaka et al., 2003), but is currently of limited diagnostic utility in nonclinical studies. Additional circulating degradation products of the three major isoenzymes have been distinguished with high-voltage electrophoresis of human and animal specimens (George et al., 1984; Billadello et al., 1989; Moss and Henderson, 1999). Terminology for these catabolic products can be confusing as they have been termed CK-3₁, CK-3₂ and CK-2₁ according to their electrophoretic mobility, or alternatively, as CK-3b, CK-3c and CK-2b or CK-MM₂, CK-MM₃ and CK-MB₂ in the literature. While these additional isoforms emphasize the lability of the enzyme, they currently lack diagnostic utility in nonclinical studies. Macro-CK, has been identified in some human samples and represents immunoglobulin-bound or oligomeric CK; this isoform is of primary significance in that it can interfere with accuracy of select clinical CK-MB assay results. However, macro-CK has not been identified in nonhuman samples. Thus, CK-MB and CK-MM, and to a lesser extent, CK-BB are the primary isoenzymes with diagnostic utility in nonclinical studies.

In humans and most laboratory animal species evaluated, the tissue with the greatest proportion of CK activity attributed to CK-MB is the myocardium (Yasuda et al., 1990). However, the majority of CK activity in both cardiac and skeletal muscle is attributed to CK-MM, which has been reported to represent 90 to 100% and 50 to 99%, respectively of the total activity in these tissues (Clampitt and Hart, 1978; Graeber et al., 1985; Sharkey et al., 1991; Loeb, 1999; Fredericks et al, 2001; Hironaka et al., 2003). The variability in specific proportion of tissue CK isoenzyme activity detected in different studies is partly dependent upon sampling location and method of analysis. However, species differences are also a major contributing factor, particularly in the proportion of myocardial CK-MB activity detected. For example, myocardial CK-MB relative to total CK activity is commonly reported to be between 10% to 25% in rats and humans, while that for the dog is considerably less, generally 1% to 3% (Fontanet et al., 1991; Sharkey et al., 1991; Moss and Henderson, 1999; Hironaka et al., 2003). Within a species, relative myocardial, as well as skeletal CK-MB activity has also been shown to vary with physiologic adaptation or chronic pathologic conditions (Sharkey et al., 1991; Hironaka et al., 2003). The extent to which these proportional differences in myocardial CK-MB activity influence serum CK-MB activity with cardiac or skeletal injury is unclear.

As with LD, serum CK-MB determination of laboratory animal specimens is currently most appropriately done by electrophoresis or ion exchange. In contrast, with human samples, serum CK-MB activity is more commonly determined by rapid, automated isoenzyme-selective immunoinhibition or immunoprecipitation assays. Determination of CK-MB protein (“CK-MB mass”) by automated immunoassay is also increasingly replacing analysis for CK-MB activity with human samples. Unfortunately, these more practicable immunological methods lack recognized validation with specimens from laboratory species to support their reliability with non-clinical toxicity testing (Roberts and Sobel, 1977; Ishikawa et al., 1997). Results with these clinical antibody-based assays with non-human samples can be highly inaccurate despite the well-conserved amino acid sequence of the isoenzyme subunits between mammals (Roman et al., 1985; Billadello et al., 1986). Notably, clinical immunoinhibition assays for CK-MB activity typically yield values of the isoenzyme in rat or dog serum that are up to 3-fold greater than the total serum CK activity due to incomplete antibody neutralization of the targeted M subunit and relatively high CK-BB activity in these species serum. Most current commercial immunoassays for CK-MB mass utilize the same monoclonal antibody (“Conan-MB”) that has shown comparable immunoaffinity with human, dog and rabbit tissue CK-MB (Vaidya et al., 1986; Landt et al., 1989; Voss et al., 1995; Pinelli et al., 2002). However, other antibody reagents of these CK-MB mass assays vary and significantly influence results obtained with nonhuman samples. Thus, each of these assays for CK-MB mass must be independently validated for the tested species. However, while electrophoresis or ion exchange is a reliable method for CK isoenzyme analysis with all species samples, comparison of results between laboratories and studies can only be general due to differences in methodology and reagents.

Circulating CK-MB in serum of laboratory species, as in humans, is a minor to non-detectable component of total serum CK activity in health. Proportionally greater serum CK-MB, relative to the total CK activity can reflect acute myocardial injury. This has been supported by studies with animal models in which serum CK-MB activity was more sensitive and specific than total CK when both analytes were determined at sequential intervals following cardiac injury (Wolf et al., 1986; Przybysewski et al., 1996). For these analyses, CK-MB activity was analyzed as a percentage or an absolute value, and evaluation of both can be advantageous in the interpretation of study results (Wolf et al., 1986). This is, in part, because skeletal muscle injury with surgical manipulation and other procedures may result in a large rise in serum CK-MM that can mask a change in the percent myocardial CK-MB activity. However, since CK-MB occurs with variable activity in skeletal muscle, increased absolute CK-MB activity may also occur with substantial skeletal muscle injury alone (Graeber et al., 1985; Fredericks et al., 2001).

The potential for CK-MB results such as these examples that are nondiscriminating as to tissue source may be more common in nonclinical, than clinical studies due to the frequency of concurrent skeletal muscle injury, and in some laboratory species (e.g., dog), the overlapping proportion of CK-MB in cardiac and select skeletal muscles. Thus, serum CK-MB analysis is not appropriate as sole biomarker to detect cardiac injury in preclinical studies. However, as part of a panel of cardiac injury markers, comparison of CK isoen-zyme profiles within nonclinical studies has shown utility for comparing the the extent or severity of a myocardial injury in animal models. As with LD isoenzyme analysis, evaluation of shifts in CK isoenzyme profiles within heart muscle tissue has also shown value in investigating changes in myocardial metabolism associated with chronic cardiac injury in animals (Sharkey et al., 1991; Hironaka et al., 2003).

Myoglobin

Myoglobin, as with CK, is primarily found in the cytoplasm of both skeletal and cardiac myocytes. In clinical medicine, the primary utility for serum myoglobin analysis is in early detection of acute myocardial infarction and response to thrombolytic treatment. Although the analyte lacks specificity for cardiac muscle and occurs in higher (though species- and age- variable) proportion in skeletal muscle, it displays a clinically advantageous short circulating half-life for early postinjury monitoring. A circulating half-life of about 20 minutes to 20 hours in humans, and less than 10 minutes in dogs has been reported (Sylven, 1978; Klocke et al., 1982). Accordingly, serum myoglobin in humans and laboratory species has generally been found to increase early following myocardial injury (with adequate perfusion) and return rapidly to baseline with resolution of tissue damage (Ellis et al., 1985; Spangenthal and Ellis, 1995). Peak arterial plasma myoglobin in dogs, as determined with an in-house developed enzyme-linked immunoassay (ELISA), occurred within 20 to 40 minutes after release of a 2-hour coronary artery occlusion (Ellis et al., 1985). Similarly, an acute transient increase in myoglobin, as determined by a noncommercial radioimmunoassay, was observed in the serum of rats following a cardiotoxic dose of isoproterenol (McMurtry and Wexler, 1979).

Currently, commercial ELISA kits that are marketed for evaluation of myoglobin in several different laboratory species are available. These assays utilize species-specific reference material. Automated clinical assays for human myoglobin incorporate a variety of proprietary reagent antibodies with nonstandardized reference material and manufacturers do not claim cross-reactivity with the nonhuman specimens (Panteghini et al., 2004). Notably, despite extensive conservation of myoglobin amino acid sequences across mammalian species, cross-reactivity of antibodies developed against the protein in one species with myoglobin from another has not been in accordance with regional sequence similarity. This is partly due to amino acid substitutions that influence tertiary structure and other antigenic features of the protein (East et al., 1982; Twining et al., 1990).

Past studies to develop antibodies to myoglobin have shown at least some antibody preparations are capable of comparative recognition of human and nonhuman primate myoglobin (Kagen and Gurevich, 1967; Adams et al., 1978). While other polyclonal preparations against human myoglobin have not shown significant cross-reactivity with the protein in rodents and dogs (Vanderbroucke et al., 1980). Thus, the currently available commercial assays for serum myoglobin require independent validation for each species to be tested to ensure reliable and interpretable results. However, because of the lack of cardiac-specificity and very short circulating half-life, serum myoglobin has only limited overall utility as a biomarker for nonclinical evaluation of cardiac injury.

Heart Fatty Acid Binding Protein (H-FABP)

Similar to myoglobin, H-FABP is an early clinical laboratory marker of myocardial injury and response to thrombolytic treatment (Tanaka et al., 1991; de Groot et al., 2001). Heart-FABP is also of comparable size with myoglobin, found in high concentration in myocyte cytoplasm, and rapidly released into blood with cell injury. Heart-FABP is one of several structurally diverse long-chain fatty acid binding proteins that have differential tissue expression. At least 9 immunologically and genetically distinct fatty-acid binding proteins have been reported in humans, and at least three distinct FABPs have been found in rodent tissues (Heuckeroth et al., 1987; Glatz and van der Vusse, 1990; Veerkamp et al., 1991). A physiological role commonly assigned to H-FABP in all species is transport of hydrophobic long-chain fatty acids from the cell membrane to intracellular sites of metabolism in the mitochondria. Other functions of the protein have also been suggested.

In contrast with myoglobin, H-FABP is considered relatively more cardiospecific because the concentration in heart muscle is several-fold greater than that in skeletal muscle. In humans, myogobin has approximately a 2-fold greater concentration in skeletal muscle than heart (when expressed per gram tissue wet weight), while H-FABP is approximately 10-fold greater in heart than in most skeletal muscles (Yoshimoto et al., 1995). Similarly in rodents, H-FABP has been found to occur in all striated muscle, but is greatest in proportion in cardiac muscle (Crisman et al., 1987; Heuckeroth et al., 1987; Paulussen et al., 1989). The protein or gene expression of H-FABP in rodents has also been found in a few other tissues, including kidney and brain, although generally in much lower levels. The actual concentration of H-FABP in cardiac muscle relative to other tissues in nonhuman species however, is variable between species, as well as between muscle region and during development (Crisman et al., 1987; Paulussen et al., 1989; Maurice et al., 1999).

In humans with myocardial infarction, circulating H-FABP generally peaks within 6 hours and returns to reference limits by 24 hours after the onset of the injury (Tanaka et al., 1991; Kleine et al., 1992; Okamoto et al., 2000). H-FABP also appears rapidly in urine, and both plasma and urinary H-FABP concentrations have been correlated with severity of myocardial injury and infarct size in humans (Tanaka et al., 1991; Glatz et al., 1994; VanNieuwenhoven et al., 1995).

Similarly in vitro and in vivo studies of the rat and dog have indicated that H-FABP is rapidly released into circulation from injured myocytes and quickly cleared intact through the kidneys (Knowlton et al., 1989; Sohmiya et al., 1993; Vork et al., 1993). In a rat model of myocardial ischemia, peak plasma H-FABP occurred within 15 minutes and concentrations were generally proportional to the size of affected area (Knowlton et al., 1989). Elimination kinetics in dogs administered exogenous H-FABP indicated a mean plasma half-life of approximately 30 minutes, and peak level in urine of 7 minutes. While in a dog model of coronary artery occlusion, plasma and urinary H-FABP levels showed rapid increase after reperfusion (Sohmiya et al., 1993). To strengthen the distinction between skeletal and cardiac muscle injury with H-FABP analysis in these and clinical studies, the ratio of myoglobin to H-FABP has been proposed (Van Nieuwenhoven et al., 1995). Use of the ratio has been particularly advocated for nonclinical studies due to common concomitant skeletal muscle injury that occur in association with animal handling. However, studies on the benefit of this ratio compared to simple analysis of H-FABP absolute values in nonclinical studies have not been published. Notably, because H-FABP renal clearance is different than that of myoglobin, impaired renal excretion can also alter this ratio.

Commercial automated assays for analysis of human H-FABP are available. However, considerable differences in amino acid sequences between the human and rodent protein have been documented, and at least some antibodies to human H-FABP tested with rodent and nonrodent animal samples have shown only 26% to 60% cross-reactivity with the nonhuman protein (Heuckeroth et al., 1987; Schreiber et al., 1998; Maurice et al., 1999). Commercial species-specific ELISA kits for H-FABP are available for several laboratory species.

In conclusion, H-FABP levels in circulation and urine has potential value as a bridging biomarker that is more cardiac-specific than myoglobin for early detection of myocardial injury. H-FABP may also be useful in estimation of severity or extent of acute myocardial injury in nonhuman species. However, as with myoglobin, H-FABP has only narrow overall utility in nonclinical testing for cardiac injury because of the lack of complete tissue specificity, and very short circulating half-life.

Cardiac Troponins

Literature on the application of cardiac troponins (cTn) in nonclinical studies for assessment of cardiac injury is abundant. Most of this material has been generated following availability of an improved second generation cardiac troponin T (cTnT), and a variety of cardiac troponin I (cTnI) commercial assays. In clinical medicine, cTn T and I are considered the most reliable of serum biomarkers for diagnosis of acute myocardial infarction, and have shown utility with multiple other, particularly acute, myocardial conditions (Adams et al., 1994, 1996; see review in Jeremias and Gibson, 2005). The utility of cTn is attributed to its high specificity for cardiac myocyte injury. Cardiac troponins are protein isofoms associated with myocyte thin filament contractile elements, and have approximately 10 to 30 additional amino acids relative to the skeletal isoforms (Townsend et al., 1994; see review in Wallace et al., 2004). The specificity of cTn assays is attributed to reagent specificity for these nonhomologous regions. The cTn sequences appear to be well-conserved across species, and cTn T and I assays have shown comparable cardiac specificity in laboratory species as in humans (O’Brien et al., 1997b; Fredericks et al., 2001, 2002).

The growing database on the use of cTn in nonclinical studies also suggests that cTnT and cTnI behave similarly across species as serum biomarkers of comparable types of induced myocardial injury, and show good concordance with findings in humans. Serum cTn T and I have been found to be similar in onset of detection and magnitude of change in animal models of acute ischemia and cardiotoxicity as occurs clinically with these conditions (Bertinchant et al., 2000; O’Brien et al., 1997a). Time of peak concentration, and clearance rate of cTn T and I in these models has also appeared to approximate that in clinical patients with comparable conditions (Bertinchant et al., 2000).

The results of multiple studies supporting the utility of cTn T and I as bridging biomarkers, as well as the biology of the proteins and commercial assays are reviewed in Wallace et al. (2004). This referenced review was contributed by an Expert Working Group on Biomarkers of Cardiac Drug-Induced Toxicity of the Nonclinical Studies Subcommittee reporting to the Center for Drug Evaluation and Research (CDER). This group identified cTn T and I as the most suitable candidates for application in nonclinical studies and validation as a bridging biomarker of cardiotoxicity. Based on this report, the International Life Sciences Institute, Health and Environmental Sciences Institute (ILSI-HESI) Biomarker Technical Committee, in collaboration with academic institutions and government agencies, recently adopted cardiac troponins as candidate markers to validate for nonclinical testing.

The ILSI-HESI Biomarker Technical Committee-Troponin Expert Working Group is approaching completion of the analytical validation, and has planned biological validation studies with selected commercial assays. Analytical validation of these assays is expected to establish the relative degree of cross-reactivity and imprecision of each assay with serum of selected species (Cynomolgus and Rhesus monkey, rat and Beagle dog). Biological validation will allow evaluation of the utility of selected assays with animal models of acute and chronic cardiotoxicity. The biological validation studies are expected to provide information in the tested species on the diagnostic window and kinetics of circulating cTn, and the correlation of assay results with cardiac histopathology. Since collected samples will be evaluated for both cTnT and cTnI, the advantages of testing each alone and together will also be assessed for each model. The assays being validated under this initiative were selected based on their platform type, availability and common usage in non-clinical and clinical laboratory testing.

Results of the analytical validation study have demonstrated expected inherent differences in assay reactivity with positive cTn serum from the different species. These assay differences are attributed to collective differences in reagent antibodies, reference material and assay methodology. Analytical validation has also shown that all tested assays demonstrate adequate, though variable cross-reactivity with linearly diluted serum pools of high, medium and low cTn concentrations from dog, and Cynomolgous and Rhesus monkeys. Several automated assays, however, showed limited to no reactivity with cTn-positive serum from Han Wistar or Sprague–Dawley rats. These findings support the utility of some, but not all of the cTn assays with samples of common laboratory species. The results also demonstrate the unpredictable reactivity and noncomparability of commercially available clinical cTnI immunoassays with analysis of non-human specimens.

Several nonclinical studies in which results of cTn testing have been compared with those of other laboratory biomarkers of cardiac injury have been published. In these studies, cTn results were comparable to, or superior to the other markers in correlating with reference endpoints (Cummins and Cummins, 1987; Bachmaier et al., 1995; Bleuel et al., 1995; O’Brien et al., 1997a; Bertinchant et al., 2000). Specifically, cTn results were superior to that of other markers in dynamic range of change, width of diagnostic window and correlation with histologic or functional severity of cardiac injury. These diagnostic advantages of cTn analysis are similar to those reported with clinical studies.

Analysis of cTn has also shown utility as a biomarker of chronic cardiac injury in nonclinical studies (O’Brien et al., 1997; Herman et al., 1998; Bertinchant et al., 2003). However, effects on serum cTn with chronic, compared with acute cardiac injury in animal models are generally of lower magnitude and less consistent among individual animals and with histologic findings. In several studies of chronic cardiotoxicity in rats, including published reports, background spontaneous cardiac lesions have also been noted histologically. The extent of correlation between these spontaneous lesions and serum cTn values remains unclear. The interpretation of positive values for cTn, or other cardiac injury biomarkers in the presence of background lesions such as spontaneous cardiomyopathy in rats, or in the absence of routine histologic findings is expected to be further evaluated during the biological validation of these assays.

Atrial and Brain Naturietic Peptides (ANP, BNP)

The utility the biomarkers discussed above has been demonstrated primarily in the evaluation of acute cardiac myocyte injury. In contrast, laboratory markers that can bridge nonclinical and clinical testing for chronic and/or functional cardiac injury have yet to be established. The naturietic peptides, ANP and BNP, when evaluated in combination with serum cTn, have shown promise as bridging biomarkers for this purpose.

Atrial naturietic peptide and BNP are structurally related hormones secreted from the heart (as prohormones) in response to increased atrial or ventricular myocardial wall tension (see reviews in Stein and Levin, 1998; Doust et al., 2005). Atrial naturietic peptide is primarily synthesized in the atria, whereas BNP is mainly released from the ventricles. BNP is also found in the brain, while ANP is additionally expressed in the pituitary, kidney and lung (Stein and Levin, 1998). Inducers of ANP or BNP release other than myocardial wall stretch, may include neuroendocrine vasoconstrictors (epinephrine, arginine vasopressin and angiotensin II) and/or their resulting hemodynamic effects. Accordingly, the naturietic peptides act as counterregulatory hormones to the effects of these neurohormones.

At rest and in health in humans, ANP concentrations are higher than BNP, and responsive to a wide variety of global hemodynamic alterations (Stein and Levin, 1998; Ohba et al., 2001). In contrast, significant BNP responses have been found to be more limited, and can occur in the absence of global hemodynamic changes, such as with regionalized ventricular wall stress at the border of infarcted areas (Stein and Levin, 1998; Doust et al., 2005). In clinical studies of outcome after myocardial infarction, plasma BNP has been correlated with the severity of left ventricular dysfunction and remodeling, as well as patient survival (Omland et al., 1996; Morrow and Braun Wald, 2003). BNP and/or ANP have also shown value as a clinical biomarker for the diagnosis of congestive heart failure, asymptomatic left ventricular dysfunction and anthracycline-induced cardiotoxicity (Omland et al., 1996; Okumura et al., 2000).

These naturietic hormones appear to have conserved functional roles across species; in a study of rats chronically dosed with doxorubicin, results of plasma ANP and BNP testing were found to be comparable with those in humans (Stein and Leven, 1998; Koh et al., 2004). In this study, significantly greater circulating BNP and cTnT, but not ANP, relative to controls and baseline values were observed after 10 weeks of dosing. The level of significance relative to the control group for BNP was also higher, and more closely correlated with left ventricular function (ejection fraction) in comparison with cTnT. Although, more investigation is needed, these and other early findings suggest ANP and BNP testing may have an important role in nonclinical evaluation of cardiac injury, particularly in evaluation of chronic, or subtle functional cardiac effects.

Both radioimmunoassays and ELISAs for plasma ANP and BNP quantitation in several laboratory species are commercially marketed (Peninsula Laboratories. Inc., San Carlos, CA). Automated clinical assays for the naturietic proteins have not been shown to cross-react with non-human plasma specimens.

Conclusion

Laboratory evaluation for cardiac injury in nonclinical studies is shifting in strategy. This change is partly motivated by recognition that traditional laboratory biochemical markers of muscle injury, including CK, LD and their isoenzyme fractions, alone, are inadequate for evaluation of cardiac injury in nonhuman species. These biomarkers lack tissue-specificity, and ideal properties to bridge clinical and non-clinical testing. However, because they also have well documented behavior in association with cardiac injury in laboratory animals, these parameters when assessed in conjunction with more specific bridging markers of cardiac injury, may continue to have utility in nonclinical testing, particularly for further characterizing the injury.

The changing strategy for laboratory assessment of cardiac injury in nonclinical studies is predominantly motivated, however, by the increasing availability of more specific laboratory markers of cardiac injury. Among these markers, the cardiac troponins have demonstrated not only comparable cardiac-specificity in animals as in humans, but also properties of ideal bridging biomarkers. Results of serum cTn evaluation with a wide variety of animal models of cardiac injury have been highly concordant with that observed clinically in humans with comparable injury. Still to be established, however, are the most reliable cTn assay platforms, and the appropriate diagnostic window for cTn evaluation in non-human species. Assessment of the utility of the cTn with animal models of different types of cardiac injury also needs further investigation. Currently, the difficulty in predicting optimal timing for specimen collection, and the potentially confounding effects of general toxicity suggests evaluation of cTn, as with other markers of cardiac injury, is still apt to be unreliable when applied to routine early toxicity studies. The utility of these laboratory markers is likely to be greater when the potential onset, mechanism and/or functional or microscopic changes that provoked concern of cardiac injury are defined.

Additional biomarkers of cardiac injury that have recently been evaluated in animal models and show promise as bridging biomarkers include the naturietic peptides, ANP and particularly, BNP. Further evaluation of these markers in other models and a variety of species is needed to assess their utility in nonclinical studies.

Validation of novel biomarkers of cardiac injury, such as the cTns, is necessary for interpretation of assay results. Validation studies should be designed to first assess the analytic reliability of an assay, and subsequently, offer key information on behavior of the biomarker in each tested species. Validation studies of cardiac injury biomarkers for nonclinical testing particularly need to support selected collection intervals, and differences between species and types of cardiac injury. Because validation of each novel biomarker for non-clinical testing requires considerable resource investment, efforts need to be focused on biomarkers that offer the most potential to advance current nonclinical testing for cardiac injury.