Longitudinal Studies of Cardiac Troponin-I Concentrations in Serum from Male Sprague Dawley Rats: Baseline Reference Ranges and Effects of Handling and Placebo Dosing on Biological Variability

Abstract

Serum cardiac troponin-I has been validated as a biomarker for cardiotoxicity in numerous animal models; however, baseline reference ranges for cTnI concentration in a healthy population of laboratory rats, as well as an investigation of biological cTnI variability in rats with respect to time, handling, and placebo dosing methods, have not been reported. In this study, we used an ultrasensitive cTnI immunoassay to quantify hourly concentrations of cTnI in live rats handled under standard laboratory conditions using 15 μL of serum per determination. The baseline reference range (mean 4.94 pg/mL, range 1–15 pg/mL, 99% confidence interval [CI]) of cTnI concentration in rats was consistent with previously reported reference ranges for cTnI in humans (1–12 pg/mL) and with preliminary studies in dogs (1–4 pg/mL) and monkeys (4–5 pg/mL) using the same cTnI assay method. In addition, cTnI concentrations in individual rat serum samples show minimal biological variability over a twenty-four-hour interval when compared to a meaningful reference change value of 193% to 206%. Furthermore, measurements of cTnI concentration were consistent within the reference limits in individual rats over long periods and under three different standard laboratory handling conditions. Thus, using this new method, rats can be followed longitudinally at hourly intervals, and a doubling of cTnI concentration would be significant above biological variability. This is a new paradigm for preclinical testing, which allows transient changes in cTnI concentration to be accurately quantified. This understanding of baseline and biological variability in rats will be fundamental for designing and analyzing future studies that assess potential cardiotoxicity in drug development.

Introduction

One of the largest challenges to drug discovery is the translation of safety and efficacy studies across preclinical and clinical development programs. Meaningful translation of results across this divide depends on the applicability of surrogate human biomarkers to animal models (Berridge et al. 2009).

For many years, measurement of cardiac troponin concentrations has been considered the gold standard for the detection of acute coronary syndrome and myocardial infarction in human beings (Antman et al. 2000) and use of serum cardiac troponin I (cTnI) concentration as a biomarker for detection of these cardiac lesions in human clinical trials is becoming an industry standard and is supported by the European Society/American College of Cardiology (Thygesen et al. 2007). In addition, measurement of cardiac troponin concentrations in human beings and animals has proved valuable in detection of cardiac injury from various causes (other than acute coronary syndromes and heart failure) such as toxic myocardial injury (NACP Writing Group et al. 2007; O’Brien 2006).

Cross-species reactivity of commercially available clinical cTnI immunoassays has been demonstrated in several animal models, including rats, dogs, and monkeys (Adamcová et al. 2007 ;Apple et al. 2008; O’Brien 2008; O’Brien et al. 2006; Schultze et al. 2008). However, in preclinical models, the challenge remains to quantify minute changes in the concentrations of cTnI to very low doses of cardiotoxins.

A specific issue for rat model systems is that many cross-species cTnI clinical assays do not provide sufficient sensitivity to quantify cTnI concentration in animals at baseline. In these cases, cTnI values are reported as less than the lower limit of detection for that particular assay. In addition, a second issue of major consequence to scientists involved in the assessment of cardiac injury is the relatively large sample volume required by some commercially available instruments and immunoassays used to measure cTnI concentrations. Such limitations of commercially available cTnI immunoassays have direct consequences on preclinical investigations of cardiotoxicity in small animals, for which sample volume is often a limiting factor. For example, pre-and post-dose longitudinal studies in rats that have acute changes in cTnI concentration in individual animals over time have been difficult. From a preclinical study design perspective, these limitations have several indirect consequences. Studies to date have relied on intra-cohort comparisons of changes in cTnI concentrations from large sample volumes. Such studies require large numbers of rats that are euthanized so as to collect the requisite sample volumes for accurate cTnI quantification.

This issue also becomes problematic when introducing new animal models for investigation of cardiotoxicity that may require changes in traditional toxicology study designs and animal handling techniques. In rat studies in particular, novel automated sampling methods may require multiple blood draws from a single individual. Without pre- and post-handling comparisons within a healthy individual, it is difficult to assess whether or not the handling method introduces variability in the measurement of cTnI concentrations resulting from stress. Most importantly, it has been difficult to rigorously document biological variability of cTnI in individual rats or to validate a baseline reference range for cTnI concentrations in rats without baseline longitudinal data.

In this study, we investigated the use of the Erenna cTnI ultra-sensitive assay (Singulex, Alameda, CA, USA) for longitudinal studies that quantify baseline concentrations of cTnI in rats. The Erenna system has been shown to accurately quantify cTnI in small sample volumes (2.5 to 50 μL) of serum or plasma in several species, including rats, dogs, and monkeys (Schultze et al. 2008). In addition, it has been used to quantify cTnI distribution (Wu et al. 2006) and biological variability (Wu et al. 2009) in healthy human subjects and in subjects with myocardial infarction. Here, we show the preliminary validation of baseline reference ranges for cTnI in a population of laboratory rats, as well as investigate biological variability of rat cTnI with respect to time, handling, and placebo dosing methods.

Materials and Methods

Institutional Compliance Statement

Rats were housed in an Association for Assessment and Accreditation of Laboratory Animal Care–accredited facility at Lilly Research Laboratories (Greenfield, IN, USA). Study protocols were approved by the Eli Lilly Institutional Animal Care and Use Committee.

Automated Blood Sampling

Automated blood sampling (ABS) is a fully automated, mechanical process used for collection of serial blood samples from the femoral artery and/or vein of cannulated rats that are affixed with a harness and tether system. This method of phlebotomy allows for minimal animal stress owing to reduced handling, provides improved sample consistency, and reduces manpower for high-throughput bioavailability and other pharmacokinetic studies (Bundgaard et al. 2007; Nolan et al. 2004).

Rat Model

Male Sprague Dawley rats (approximately 300 g) were purchased from Charles River Laboratories (Portage, MI, USA). Rats were surgically modified by the vendor. Polyurethane cannulas were placed in the femoral artery and vein. Intravenous access lines were tunneled subcutaneously from the femoral region to the back of the neck and were exteriorized in the dorsal cervical region of each rat and connected to the sampling harnesses. Rats were housed individually in shoebox cages with hardwood chip bedding (Harlan Teklad, Madison, WI, USA) and provided with two micron–filtered municipal tap water and fed Harlan Teklad 2014 diet (Harlan Teklad Global Diets, Madison, WI, USA) ad libitum. Intravenous lines for each rat were connected aseptically via a vascular access harness and tether system to the DiLab Accusampler (DiLab Inc., Littleton MA, USA). Temperature (22.2°C ± 4.4°C), humidity (20%–80%), and light cycle (twelve-hour light:dark) were controlled for these studies. All blood samples described in these studies were collected from the femoral artery.

Blood Collection and Processing

Each blood sample drawn from the femoral artery was placed in a vial without anticoagulant. Blood collection vials were housed in a refrigerated collection plate on each DiLab Accusampler. Clotted samples were spun in a centrifuge, and serum was divided into aliquots. To avoid evaporation, serum samples were stored in ninety-six–well plates with individual sample lids.

Sampling Methods

Resting (Longitudinal Studies)

Six rats were housed in a “resting” state; no handling occurred after connection of the rat to the DiLab Accusamplers. Blood samples of 80 μL volume were drawn from each rat every hour for twenty-four hours.

Oral Dosing (Gavage)

Six rats were housed in a study room with minimal interruption. Blood samples of 80 μL volume were drawn from each rat every hour for twenty-four hours. Approximately five hours after study start, each rat was given an oral gavage (16-gauge stainless steel animal feeding needle) with 1 mL of filtered tap water.

Second Oral Dosing

Nine rats were housed in a study room as described in Oral Dosing above. Samples of 80 μL were drawn from each animal every hour for twelve hours. Approximately three hours after study start, each rat was given an oral gavage as described above.

Simulated Transport

Six rats were housed in a study room with minimal interruption. Samples of 80 μL volume were drawn from each rat every hour for twenty-four hours. Approximately five hours following the start of sample collection, each rat was disconnected from the DiLab Accusampler and placed along with its shoebox cage on a stainless steel transport cart. The cart was pushed from the study room down an adjoining hallway to simulate transportation to the necropsy room following study completion. Rats were then returned to the study room and reconnected to the DiLab Accusamplers. Blood sample collection continued with minimal interruption in the study room.

Sample Analysis

All serum samples were provided by Lilly Research Laboratories (Greenfield, IN, USA) and tested in a manner blinded to reference test results at Singulex (Alameda, CA, USA). Samples were analyzed with the Erenna cTnI Immunoassay System (Singulex, Alameda, CA, USA; LoD = 0.2 pg/mL, LLoQ = 0.8 pg/mL with ≤10%CV), which has been described previously (Schultze et al. 2008; Todd et al. 2007). In brief, frozen samples were thawed and tested in 15-μL sample volumes for analysis in duplicate. Samples were each brought up to 50 μL volume with calibrator diluent and were added to 150 μL assay buffer containing microparticles (MPs) coated with biotinylated cTnI capture antibody. The resulting mixture was incubated for one hour in a ninety-six–well plate. The MPs were then magnetically separated, washed, and incubated with 20 μL fluorescent dye–labeled detection antibody for thirty minutes. After five washes via magnetic separation, 20 μL of elution buffer was added, and the eluted detection antibody was separated from MPs using a 384-well filter plate. The eluate was passed through the Erenna Immunoassay System, and detection antibodies were quantified using single-molecule counting instrumentation. All sera and calibrators were tested in duplicate and the results are presented as the mean of the two measurements. The human National Institute for Standards and Technology standard was used as the standard reference for calibration, which has been shown previously to provide accurate determinations of cTnI concentration in rat serum with this assay (Schultze et al. 2008).

Statistical Analysis

Data were calculated from duplicate values, and are presented as the mean ± SD. Because our data were not both normally distributed (Kolmogorov-Smirnov test) and with equal variance (Levene’s test), we used the nonparametric Kruskal-Wallis test to determine whether the data in each treatment group came from the same population. In addition, we evaluated our data for the presence of outliers using box-and-whisker plots and compared the results of the Kruksal-Wallis test on our data with and without outliers excluded. Multiple treatment groups were compared against a single set of control values with Dunnett’s test. Multiple time points within treatment groups were also compared to test for time-related trends. All statistical calculations were performed using StatsDirect software version 2.7.0, [7th July 2008], (Altrincham, Cheshire, UK), and GraphPad Prism version 5.01 for Windows (GraphPad Software, [1st Oct, 2008], San Diego, CA, USA). We determined the reference change values (RCVs) for a statistically significant (p ≤ .05) change (increase or decrease) in cardiac troponin I (cTnI) concentrations using data from resting rats and the lognormal approach (Fokkema et al. 2006). Using this approach, RCV values are always nonsymmetrical (Fokkema et al. 2006).

Results

Intra-cohort Comparison of cTn Concentrations

The distribution of cTnI concentrations in resting rats (RR) and rats after saline gavage oral dosing (OD) or simulated transport (ST) is shown in Figure 1A through 1C. Compared to resting rats (mean 4.11 ± 2.05 pg/mL), a statistically significant increase in mean concentrations of cTnI were observed in orally dosed rats (mean 6.89 ± 3.13 pg/mL, Table 1). The difference in cTnI values between resting rats and those that underwent saline gavage remained statistically significant (p <.001), even after exclusion of outliers (n = 5) identified in the box-and-whisker plot (Figure 1, Table 1). However, the cTnI concentration in simulated transported rats (mean 5.34 ± 2.54 pg/mL) was not significantly different when compared to that of resting rats.

Inter-cohort Time-Course Comparison of cTnI Concentrations

To investigate whether apparent changes in mean cTnI concentrations between treatment groups was a result of handling, or represented preexisting biological variation between treatment groups, we compared pre- and post-handling concentrations of cTnI for each treatment cohort. An hourly time course of mean cTnI concentrations from six rats per each of three handling groups (RR, resting rats; OD, oral dose by saline gavage; ST, simulated transport) over a twenty-four-hour period is shown (Figure 2A). There were no apparent upward or downward trends over time in any of the three treatment groups before or after time of handling, as indicated by the vertical bar at time = 0. The distribution of cTnI concentrations before and after handling for resting rats (RR) and rats after saline gavage oral dosing (OD) or simulated transport (ST) is shown in Figure 2B. When post-handling results were tested for significance against pre-handling results within the same cohort, no statistically significant change in cTnI concentration in resting rats or in saline gavage rats was observed. However, there was a slight significant decrease (p <.05) in cTnI concentrations in rats following simulated transport (Figure 2B).

Effect of Saline Gavage on cTnI Concentrations

To evaluate further the effect of saline gavage on cTnI concentrations, we performed saline gavage on a second group of rats (n = 9) and monitored cTnI concentrations for twelve hours at one-hour time intervals post-gavage. The distribution of cTnI concentrations in this treatment group of rats at each time point post-saline gavage is shown in Figure 3. The data in each time interval came from the same population (p = .59), indicating that there was no time-dependent effect of saline gavage on cTnI concentrations. In addition, a comparison of cTnI concentrations within each time group against the control group (T0) indicated no statistically significant difference between these groups (p value range:.29 to >.99). Using cTnI data (N = 72) from six resting-state rats, lognormal RCV values for a statistically significant increase or decrease in cTnI values were 206.7% and 67.4%, respectively. Therefore, cTnI concentrations would have to double (or decrease by 67.4%) in rats undergoing any procedure before a statistically significant change could be considered to have occurred as a result of the procedure/drug administered that was greater than the change expected due to the combined effects of analytical and biological variation alone on cTnI concentrations in resting rats. Using cTnI data (N = 117) from the nine rats that underwent saline gavage and were sampled at hourly intervals for twelve hours, lognormal RCV values statistically significant increase or decrease in cTnI values were 193.6% and 65.9%, respectively.

Distribution and Biological Variation of cTnI in Laboratory Rats

To resolve the apparent discrepancy between the statistically significant difference between mean cTnI concentration in resting rats compared to orally dosed rat cohorts versus the nonsignificant change in cTnI concentrations between pre- and post-dose rats, we investisgated the biological variability of cTnI concentration within the laboratory rat population as a whole compared to cTnI concentration variability in individuals. A presumptive baseline reference range for cTnI concentrations representing all laboratory rats was determined. All results for all time points for all groups (N = 87 time points) were pooled and used to calculate frequency distributions of cTnI in the representative laboratory rat population (Figure 4A). The mean (± SD) concentration of cTnI inrats was 4.94 ± 2.62 pg/mL with an upper 99th% reference limit of 15.38 pg/mL. Since there was slight fluctuation in concentrations of cTnI over time in the time-course study, we investigated the biological variation of cTnI concentration for individual rats over a twenty-four-hour time period (Figure 4B). Individual rats from each handling cohort were selected for which a minimum of twenty-four hours of sampling occurred and are shown as the mean and absolute range of cTnI concentration. Two rats in the oral dose saline gavage group exhibited more variability and higher cTnI concentrations compared to rats in the other two groups (RR and ST), which exhibited lower degrees of biological variability.

Discussion

In this study, we demonstrated use of a highly sensitive cTnI immunoassay using as low as 15 μL of serum per determination to detect hourly concentrations of cTnI in live rats handled under standard laboratory conditions, as might occur in routine toxicity testing of new drugs in development. Our results suggest that a baseline reference range (99% CI) of cTnI concentration in male rat serum should be 1–15 pg/mL when measured with the Erenna cTnI assay system. These results are consistent with previously reported reference ranges for cTnI in humans (1–12 pg/mL) and with preliminary studies in dogs (1–4 pg/mL) and monkeys (4–5 pg/mL) using the same cTnI immunoassay system (Table 2). In addition, we have demonstrated that rat serum cTnI concentrations show minimal biological variability over a twenty-four-hour interval when compared to a meaningful reference change value of 193% to 206%. Furthermore, measurements of cTnI concentration are consistently within the reference limits in individual rats over long periods and under three different standard laboratory handling conditions. Thus, the natural variation of cTnI concentration observed within individual male rats, although quantifiable with this ultrasensitive method, is likely to represent low concentration variation of cTnI in laboratory rats.

These results indicate that for an experimentally observed increase of cTnI in an individual rat to be considered biologically relevant, the observed value would need to at least double in concentration from the previous measurement (RCV of 193%–206%) with a resulting increased cTnI concentration greater than 15.38 pg/mL. Thus, using this new method, rats can be followed longitudinally at hourly intervals, and a doubling of cTnI concentration above the 99th% cutoff value would be significant above biological variability. This is a new paradigm for preclinical testing, which allows transient changes in cTnI concentration to be accurately quantified. These results are consistent with previous studies, which correlated increased cTnI concentration in rats with microscopic lesions in myocardial tissues (Schultze et al. 2008), and further studies on these effects of low-dose cardiotoxic compounds are warranted. Similarly, studies on biological variation in other commonly used laboratory rat strains, especially with regard to potential differences in baseline cTnI concentration with respect to variables such as age or sex, would be of significant benefit.

To our knowledge, this is the first such study to follow rats longitudinally and to successfully quantify cTnI concentration at hourly intervals, as opposed to cohort-based studies. It is interesting to note that in this study of baseline animals, the results did differ in level of significance depending on whether the data were analyzed within or between cohorts. This result attests to the value of time-course studies within individuals and points to inappropriate conclusions that can be made from some intra-cohort comparisons. Our results reaffirm that the best study design includes pre- and posttreatment analysis of inter-individual data monitoring over time, which has been difficult in rats.

Compared to the preliminary baseline reference range (mean 4.94 pg/mL, range 1–15 pg/mL), there are a few important implications for the required sensitivity and reading range of cTnI immunoassays used for longitudinal, preclinical cardiotoxicity studies in rats. First, to quantify a doubling in the lower quartile (<25th%, cTnI < 3 pg/mL) and lower 50% (median cTnI < 4.5 pg/mL) of an initial population would require a corresponding assay sensitivity (LLoQ) of at least < 3 pg/mL and < 9 pg/mL, respectively. Second, the ULoQ should be adequate to quantify large relative increases of cTnI concentration above baseline in studies of cardiotoxic compounds in rats. For example, cTnI concentrations in rats have been reported to be well over 1000 pg/mL following administration of isoproterenol at doses ≥ 0.5 mg/kg (Mitra et al. 2008; Schultze et al. 2008; York et al. 2007), which is significantly greater than a fivefold change from our observed reference range. Thus, to provide maximum coverage of all possible and expected cTnI concentrations, such longitudinal studies require a sensitive immunoassay (LLoQ of at least 3 pg/mL) coupled with a wide dynamic range of over four logs.

We propose that the method used in this study to quantify cTnI concentration shows significant improvement over previous studies in rats and will further improve on preclinical study design and implementation in several ways. First, the increased sensitivity of this cTnI assay allows quantification in serum from rats from a small volume of serum. Second, the small sample volume allows for longitudinal time-course studies to be made, allowing pre- and post-dose concentrations of cTnI to be quantified within the same treated individual. In contrast, previous studies have relied on cohort comparisons, which are not always in agreement with pre- and post-intervention studies. Third, this method may decrease the need for euthanasia of rats at every time point in some studies, which can be costly. Lastly, this method has been used successfully to establish baseline reference ranges and to establish the biological variability of cTnI concentrations in rats and in rats undergoing gavage dosing and simulated transport.

This understanding of baseline and biological variability in rats will be fundamental for designing and analyzing future studies that assess potential cardiotoxicity in drug development. Similar studies in other commonly used animal models for cardiotoxicity will be important for developing a validated set of species-specific baseline reference ranges for cTnI with which to evaluate cross-species comparisons of cTnI between commonly used animal model systems. This information will be indispensable for showing bioequivalence of cardiotoxicity across species, and to ensure meaningful translation of results from preclinical into clinical research.