Biomarkers in Peripheral Blood Associated with Vascular Injury in Sprague-Dawley Rats Treated with the Phosphodiesterase IV Inhibitors SCH 351591 or SCH 534385

Abstract

Drug-associated vascular injury can be caused by phosphodiesterase (PDE) IV inhibitors and drugs from several other classes. The pathogenesis is poorly understood, but it appears to include vascular and innate immunological components. This research was undertaken to identify changes in peripheral blood associated with vascular injury caused by PDE IV inhibitors. We evaluated twelve proteins, serum nitrite, and leukocyte populations in peripheral blood of rats treated with experimental PDE IV inhibitors. We found that these compounds produced histological microvascular injury in a dose- and time-dependent manner. Measurement of these serum proteins showed changes in eight of the twelve examined. Changes were seen in the levels of: tissue inhibitor of metalloproteinase-1, α1-acid glycoprotein, GRO/CINC-1, vascular endothelial growth factor, C-reactive protein, haptoglobin, thrombomodulin, and interleukin-6. No changes were seen in levels of tumor necrosis factor-α, hepatocyte growth factor, nerve growth factor, and granulocyte-monocyte colony stimulating factor. Serum levels of nitrite were also increased. Circulating granulocyte numbers were increased, and lymphocyte numbers were decreased. The changes in these parameters showed both a dose- and time-dependent association with histopathologic changes. These biomarkers could provide an additional tool for the nonclinical and clinical evaluation of investigational compounds.

Introduction

The drug-induced vascular injury (DIVI) associated with phosphodiesterase inhibitors has a long history, and there has been limited success in understanding the pathogenesis of the toxicity. There are a number of excellent reviews of the field, which provide a good overview of the characteristic lesions and the variety of drugs that can cause this toxicity (Balazs et al. 1986; Boor et al. 1995; Expert Working Group 2005; Hanig 1991; Kerns et al. 1989; Kerns and Joseph 1989; Louden and Morgan 2001) The nonspecific PDE inhibitor caffeine was reported to cause polyarteritis nodosa in rats (Johannsson 1981). Since then, vascular injury has been reported in animal toxicology studies as a result of many compounds, in at least six classes of compounds. These lesions are characteristically found in mesenteric microvessels in the rat and in the coronary arteries in the dog, pig, or monkey. It is not clear whether similar drug-induced lesions occur in humans, in part owing to the absence of biomarkers shown to be associated with lesion development and a paucity of relevant human autopsies.

The mechanism(s) leading to lesion development have not been clearly identified. There are two major theories, each of which may be correct for some drugs. These are the biomechanical stress theory and the inflammatory theory (Expert Working Group 2005). The biomechanical theory is that the drugs cause excess relaxation in the affected vessels, and the stress of distension and/or increased flow causes splits in the endothelium, initiating the lesion. This theory has experimental support, for example in studies where coadministration of a vasoconstrictor prevented lesions seen with a PDE III inhibitor (Joseph 2000). The inflammatory theory is that some component of the innate immune system—such as granulocytes, mast cells, and/or monocytes—is activated by the drug and that the damage is secondary to the release of mediators by these cells. This theory has its primary support in the observation that pretreatment with the immunosuppressive drug dexamethasone blocks nearly all development of lesions seen following treatment with a PDE IV inhibitor (Slim et al. 2003). There is insufficient evidence to determine whether any of these theories is in fact correct.

Despite the lack of agreement on the mechanism behind the toxicity, there is a clear need for biomarkers for vascular injury in an accessible tissue such as peripheral blood. In this study we have identified a series of candidate biomarkers in peripheral blood that show a reasonable association with vascular lesions identified by histopathology. These candidate biomarkers are correlated with lesion severity in both dose response and in time course. The full description of the histopathology studies is in the companion paper to this study (Zhang et al. 2008).

Materials and Methods

Animals

Sprague-Dawley rats were obtained from Charles River Laboratories (Wilmington, DE) and were singly housed. Rats were fed Certified Purina Rodent Chow #5002 (Ralston Purina Co., St. Louis, MO) and water ad libitum. The experimental protocol was approved by the Institutional Animal Care and Use Committee, Center for Drug Evaluation and Research, FDA, and conducted in an AAALAC-accredited facility. All procedures for animal care and housing were in compliance with the Guide for the Care and Use of Laboratory Animals, 1996 ILAR (Institute of Laboratory Animal Resources).

Reagents

The phosphodiesterase IV inhibitors SCH 351591 and SCH 534385 (Schering-Plough Research Institute, Lafayette, NJ) were administered as a suspension in 1 mL of phosphate buffered saline by oral gavage. Doses ranged from 3 to 80 mg/kg/day, and study durations ranged from two to nine days. See Zhang et al. 2008for details of dosing concentrations and timing.

Histopathology

Tissues were fixed in 4% buffered formalin, sectioned at 5 μm, and stained with hematoxylin and eosin. The mesentery was pinned out at full extension, fixed, and sectioned as previously described (Zhang et al. 2002). The severity of mesenteric lesions was rated using a semiquantitative scale, as previously described (Zhang et al. 2002; Zhang et al. 2008).

Clinical Chemistry and Hematology Analysis

At necropsy, blood was collected from the inferior vena into serum collection tubes. Sera were analyzed using a Hitachi 717 clinical chemical analyzer (AniLytics, Inc., Gaithersburg, MD, USA). Blood cellular parameters were measured by a hematology analyzer (AniLytics, Inc., Gaithersburg, MD, USA).

Serum Nitrite Determination

The measurement of serum nitrite was performed using the Total Nitric Oxide Assay (R&D Systems, Inc., Minneapolis, MN, USA) according to the manufacturer’s instructions. Since nitric oxide (NO) is metabolized to nitrite (NO_2-) and nitrate (NO_3-), the determination of total serum nitrate and nitrite concentration in this assay provides a quantitative measurement of nitric oxide generated by nitric oxide synthases. Serum samples were diluted two-fold and passed through a 10 kD molecular weight filter (Biomax-10, Millipore, MA, USA) before NO measurement. The optical density (OD) was measured at 540 nm.

Candidate Serum Biomarker Analysis

Sera was stored at -70°C and tested on the following sandwich ELISA kits: rat interleukin-6 (IL-6), rat tumor necrosis factor α (TNF-α), rat monocyte chemotactic protein-1 (MCP-1), and rat macrophage inflammatory protein-2 (MIP-2) from BioSource International, Camarillo, CA, USA; rat hepatocyte growth factor (HGF), rat tissue inhibitor of metalloproteinase-1 (TIMP-1), and rat granulocyte-macrophage colony-stimulating factor (GM-CSF) from R&D Systems, Minneapolis, MN, USA; rat GRO/CINC EIA (human IL-8 homolog) from IBL Co., Japan; rat C-reactive protein (CRP) from BD BioSciences, PharMingen, San Diego, CA, USA; haptoglobin from Tri-Delta Diagnostics, Inc., Moris Plains, NJ, USA; rat α-1 acid glycoprotein from Cardio Tech Services, Inc., Minneapolis, MN, USA; mouse vascular endothelial growth factor (VEGF). from Oncogene Research Products, San Diego, CA, USA; nerve growth factor from Chemicon International, Temecula, CA, USA; and human thrombomodulin from American Diagnostica Inc., Greenwich, CT, USA. The thrombomodulin kit became unavailable during the early phases of this study, resulting in limited data for this biomarker. No other kit reactive with rat thrombomodulin has been identified.

Flow Cytometry

Leukocytes were stained and analyzed in whole blood using anti-rat antibodies CD11b-FITC, CD62L-PE, and CD45-PE/Cy5 (BD/PharmMingen Inc., San Diego, CA, USA). Data were collected by triggering on CD45-PE/Cy5 on a Coulter Elite flow cytometer (Beckman Coulter Inc., Hialeah, FL, USA), as previously described (Weaver et al. 2002).

Results

Candidate Biomarkers

Dose Response of SCH 351591

Significant dose-related elevations in serum levels of TIMP-1, α1-acid glycoprotein, GRO/CINC-1, VEGF, CRP, haptoglobin, and IL-6 were noted in drug-treated rats compared to those from vehicle-treated animals (Figure 1). The plasma levels of thromobomodulin were decreased in drug-treated rats. In addition, serum levels of MCP-1 and MIP-2 showed a nonsignificant positive trend, whereas serum levels of TNF-α, hepatocyte growth factor, nerve growth factor, and GM-CSF were not significantly changed (data not shown).

Time Course of SCH 351591:

The time course of changes in the serum levels of the biomarkers following treatment with a dose of 20 mg/kg/day for up to three days are shown in Figure 2. Altered values were seen in the early time points for several biomarkers. A significant increase at the four-hour time point was observed with GRO/CINC-1. Early but not significant decreases were seen with haptoglobin and CRP. Increases later in the time course were significant for α1-acid glycoprotein, GRO/CINC-1, and haptoglobin at twenty-four hours and for all five of these biomarkers at forty-eight and seventy-two hours. Significant changes at this dose were not observed for CRP or IL-6. Thrombomodulin was not evaluated because the test kit became unavailable.

Comparison with Histopathology Lesion Scores of SCH 351591:

The details of the mesenteric vascular lesion scores are presented in the companion paper (Zhang et al. 2008). The alterations of the inflammatory biomarkers were plotted versus the severity of mesenteric vascular injury for the dose response studies (Figure 3). The data show a good association between the changes in biomarkers and the histopathological changes for TIMP-1, α1-acid glycoprotein, GRO/CINC-1, CRP, haptoglobin, and IL-6. Haptoglobin showed a decrease at the lower doses and longer time points, but no changes at the higher doses. VEGF showed elevation at lower lesion scores, suggesting that this might be a more sensitive biomarker. A similar plot for the biomarker vs. histopathology scores for the time course study is shown in Figure 4. The responses were good for all biomarkers except CRP. The early change in GRO/CINC-1 at four hours was not matched by a high lesion score.

Responses to SCH 534385:

This PDEIV inhibitor was studied at two doses, 20 and 40 mg/kg/day, which are pharmacologically equivalent to 40 and 80 mg/kg/day of SCH 351591. The responses of the biomarker panel were similar to that seen with the original compound (SCH 351591). Significant responses were observed with all biomarkers evaluated except for CRP (Figure 5).

Hematology and Flow Cytometry

Dose Response:

An increase in the number of neutrophils and a decrease in the number of lymphocytes varied directly as the dose increased, particularly at doses of 40, 80, or 160 mg/kg/day for one to three days. Neutrophilia and lymphopenia occurred in rats treated with 3 mg/kg/day for three days, and significantly advanced in rats treated with 9 mg/kg/day for seven days, and reached a statistically significant plateau in groups at doses of 40, 80, or 160 mg/kg/day of SCH 351591 for one to three days (Figure 6A). The equivalent data for SCH 534385 are shown in Figure 6B, where significant neutrophilia and lymphopenia are again observed.

Time Course:

The increases at early time points show a generally linear increase in circulating granulocytes starting at four hours (Figure 6C). The rise in granulocytes caused an increase in total WBC that became significant at twenty-four hours (Figure 6C). The absolute numbers of circulating monocytes are unchanged. The number of lymphocytes was slightly decreased in this experiment, but the change was not statistically significant.

Serum Nitrate

Changes in this parameter as a function of time are shown in Figure 7. A small but not significant rise is seen at one hour. Significant elevations are seen at forty-eight hours and beyond.

Clinical Chemistry

No reliable and reproducible changes among standard rat serum clinical chemistry parameters were seen. The parameters measured in the panel were: sodium, chloride, potassium, calcium, phosphorus, lactate dehydrogenase, glucose, urea nitrogen, creatinine, creatine kinase, cholesterol, triglicerides, total bilirubirubin, alkaline phosphatase, alanine aminotransferase, aspartate aminotransferase, γ glutamyl transferase, uric acid, total protein, and albumin.

Discussion

In this study, we have identified a series of biomarkers that shows both temporal and dose-related correlation with the histopathological observation of microvascular lesions. These biomarkers include a number of acute-phase proteins, changes in blood leukocyte populations, and serum nitrite.

There are a number of possible cellular or tissue sources of the acute-phase proteins observed to change in response to VI inducers. The known cellular sources of the biomarkers tested in this study are shown in Table 1. The cell types listed in this table cover the majority of cell types present in resting and/or inflamed mesenteric tissue. In addition, hepatocytes are included, as they are a source of a number of acute phase proteins. Examination of the literature shows many examples of cross-talk among many of these inflammatory mediators. For example, IL-6 induces both TIMP-1 (Kralisch et al. 2005), and VEGF (Shin et al. 2005), however VEGF downregulates TIMP-1 (Pufe et al. 2004). To add to the uncertainty of the interpretation, the upregulation of these proteins could also represent a pharmacological response to the test drug rather than being an indicator of toxicity.

The only compounds known to inhibit lesion formation are dexamethasone (Slim et al. 2003) and arginine vasopressin (Joseph 2000). Dexamethosone has a very broad spectrum of immunosuppressive activity and could easily inhibit both toxic and pharmacological changes in biomarkers. The vasoactive compound arginine vasopressin required a twelve-hour continuous infusion to achieve effective blockade of lesion formation (Joseph 2000). In addition, vasopressin has been shown to activate some and suppress other specific activities of both acquired and innate arms of the immune system (Chikanza et al. 2000; Lagumdzija et al. 2004). These nonvascular activities complicate the interpretation of the results.

The cellular changes in circulating leukocyte populations occurred in parallel with the changes in serum biomarkers. These changes could be the result of direct action of the drug, a direct response to toxicity, or secondary to changes in other acute phase proteins. For example, granulocytosis has been observed after infusion with recombinant IL-8 (van Zee et al. 1992) in primates or following genetic modification to constitutively express IL-6 (Hawley et al. 1991).

Increases in serum nitrite are generally considered to result from recent production of nitric oxide (NO) by eNOS and/or iNOS (Tsikas 2005). These enzymes can be found either constitutively or following external stimulation in all cell types commonly found in the mesenteric vasculature (Table 1).

The most useful biomarkers for vascular injury would be ones that would pass four tests. The first test is that the bio-marker would be specific for the injury, and the second would be that it is causally related to the pathogenesis of the lesion. The third test is that the biomarker should be released very early in the development of the lesion, and the fourth test should show a good correlation with the extent of the lesion. The inflammatory biomarkers studied here do not pass the test of specificity for the actual lesion. These can be reasonably expected to rise in response to many types of tissue injury or other inflammatory stimuli. The second test for direct linkage to the mechanism causing vascular injury cannot be evaluated, as there is insufficient published data to identify the pathogenesis of the microvascular injury. The biomarker GRO/CINC-1 might be a candidate to pass the third test of early release. Additional studies are being conducted to evaluate the very early responses of this and several other candidate biomarkers. The final test of candidate biomarkers is that of correlation with the extent of the histopathological lesions. The data clearly show that the biomarkers TIMP-1, α1-acid glycopro-tein, GRO/CINC-1, VEGF, haptoglobin, and IL-6 pass this test. This finding is valid for both of the PDE IV inhibitors evaluated in this study.

Therefore, as biomarkers of DIVI, the group of serum proteins evaluated in this study have clear strengths and limitations. However, they do provide a set of tools that could potentially be used in nonclinical studies to determine the potential of an investigational drug to induce the biomarkers that are associated with DIVI. Additionally, those biomarkers might be included in a panel of biomarkers to be used in clinical studies where DIVI was observed in nonclinical studies.