Chemical-Induced Atrial Thrombosis in NTP Rodent Studies

Abstract

Cardiac thrombosis, one of the causes of sudden death throughout the world, plays a principal role in several cardiovascular diseases, such as myocardial infarction and stroke in humans. Data from studies of induction of chemical thrombosis in rodents help to identify substances in our environment that may contribute to cardiac thrombosis. Results for more than 500 chemicals tested in rodents in 2-year bioassays have been published as Technical Reports of the National Toxicology Program (NTP) 〈http://ntp-server.niehs.nih.gov/index〉. We evaluated atrial thrombosis induced by these chemical exposures and compared it to similarly induced lesions reported in the literature. Spontaneous rates of cardiac thrombosis were determined for control Fischer 344 rats and B6C3F1 mice: 0% in rats and mice in 90-day studies and, in 2-year studies, 0.7% in both genders of mice, 4% in male rats, and 1% in female rats. Incidences of atrial thrombosis were increased in high-dosed groups involving 13 compounds (incidence rate: 20–100%): 2-butoxyethanol, C.I. Direct Blue 15, bis(2-chloroethoxy)methane, diazoaminobenzene, diethanolamine, 3,3′-dimethoxybenzidine dihydrochloride, hexachloroethane, isobutene, methyleugenol, oxazepam, C.I. Pigment Red 23, C.I. Acid Red 114, and 4,4′-thiobis(6-t-butyl-m-cresol). The main localization of spontaneously occurring and chemically induced thromboses occurred in the left atrium. The literature survey suggested that chemical-induced atrial thrombosis might be closely related to myocardial injury, endothelial injury, circulatory stasis, hypercoagulability, and impaired atrial mechanical activity, such as atrial fibrillation, which could cause stasis of blood within the left atrial appendage, contributing to left atrial thrombosis.

Supplementary data referenced in this paper are not printed in this issue of Toxicologic Pathology. They are available as downloadable files at http:taylorandfrancis.metapress.com/openurl.asp?genre=journal&issn=0192-6233. To access them, click on the issue link for 33(5), then select this article. A download option appears at the bottom of this abstract. In order to access the full article online, you must either have an individual subscription or a member subscription accessed through www.toxpath.org.

Introduction

Thrombosis constitutes a serious disorder that can prove fatal in humans with classical risk factors associated with myocardial infarction, one of the most important of which is change in the vascular wall (Lowe, 2004). Venous thrombosis becomes significant, for example, in aged women undergoing oral estrogen therapies; blood stasis and coagulability are involved in the induction of thrombosis (Rosendaal et al., 2002; Lowe, 2004). Mitral valve disease, atrial fibrillation, dilatation of the left atrium, bradycardia, a low cardiac output, and hypercoagulability generally summarize the causes of formation of left atrial thrombus in human cases; systemic diseases, such as malignant tumors, amyloidosis, and nephritic syndrome, have also contributed to development of intracardiac thrombus (Aoyagi et al., 2002). Atrial thrombosis sometimes occurs as a side effect after therapy involving central venous catheterization, heart valve replacement, or maintenance hemodialysis (Platt et al., 1980; Rotellar et al., 1996; Kingdon et al., 2001; Shapiro et al., 2002; Douketis, 2003; Nishimura et al., 2003). Recently, a cyclo-oxygenase-2(COX-2)-inhibitor, one of a category of nonsteroidal anti-inflammatory drugs, was withdrawn from the market because of a significant risk of cardiovascular thrombotic complications, such as development of cardiac thrombus and myocardial infarction (Konstam and Weir, 1999; Bing and Lomnicka, 2002; Schmidt et al., 2004). Inhibition of COX-2 may lead to increased prothrombotic activity by tipping the balance of prostacyclin/thromboxane in favor of thromboxane, a prethrombotic eicosaid (Mukherjee et al., 2001; Bing and Lomnicka, 2002). Thus, much attention has been focused upon cardiac thrombosis as one of several side effects induced by certain medical treatments in humans.

Of approximately 85,000 chemicals registered for use in the United States, with an additional 2,000 introduced annually, complete toxicological-screening data are obtainable for only 7%; for 40%, no data are available (Bennett and Davis, 2002). Although the NTP has compiled a large database of incidences of lesions seen in chemical-treated animals 〈http://ntp-server.niehs.nih.gov/index〉, atrial thrombosis in rodents was not reported to be increased (Elwell and Mahler, 1999) until recently, when atrial thrombosis was induced by exposure to bis(2-chloroethoxy)methane (CEM) and 2-butoxyethanol (2-BE) via different pathogeneses (NTP, 1998a, 2003a, 2003b; Ezov et al., 2002). We, therefore examined all of the NTP data for more than 500 chemicals and identified 13 that appeared to induce atrial thrombosis. Supplementary data presenting the names and structures of these compounds, as well as their usages and effects in humans, are provided online (Please see the end of the abstract for the URL). In addition, the rates of spontaneous atrial thrombosis were determined for control F344 rats and B6C3F1 mice from these 90-day and 2-year toxicity studies. This article, one of a series of works highlighting specific findings from the NTP studies, focuses on the incidences and morphologic aspects of atrial thrombosis induced by chemicals. We discuss reports from the literature of other chemically induced atrial thromboses and potential mechanisms of induction. Such results contribute to the identification of environmental substances that may affect the induction of cardiac thrombosis.

Methods Used in the NTP Studies

The NTP has customarily used 2-year and short-term, preliminary rodent bioassays to assess the overall toxicity and carcinogenicity of the tested chemical and to identify potential human carcinogens. We searched all of the NTP data for more than 500 chemicals and identified 13 that appeared to induce atrial thrombosis. Table 1 lists aspects of the study design for each of these chemicals identified as possible inducers of atrial thrombosis, such as the animal species and the route, duration, and dose of the exposure. Using data from NTP rodent studies, we described the incidences of thrombosis in overall and early sacrificed animals exposed to these chemicals 〈http://ntp-server.niehs.nih.gov/index〉, in addition to control animals. The standard bioassay included male and female B6C3F1 mice and F344 rats, although other strains were occasionally included, such as Swiss Webster mice that were used in the study of oxazepam. Rodents were typically exposed to a chemical at 6–8 weeks of age by a route of exposure chosen for its relevance to the known or suspected route(s) of human exposure. All procedures, care, and treatment of animals were in accordance with the principles of humane treatment outlined by the National Institutes of Health (Grossblatt, 1996).

Results from the testing of more than 500 chemicals in the NTP studies led to the identification of 8 compounds, some reported in the original investigations, that induced atrial thrombosis in rodent models: bis(2-chloroethoxy)methane, 2-butoxyethanol, C.I. Acid Red 114, C.I. Direct Blue 15, diazoaminobenzene, 3,3′-dimethoxybenzidine dihydrochloride, isobutene, and oxazepam. Moreover, our examination of the data revealed 5 additional chemicals that may also have induced this abnormality, although no description of the increased incidence of thrombosis was noted in the reports of the NTP studies: C.I. Pigment Red 23, diethanolamine, hexachloroethane, methyleugenol, and 4,4′-thiobis(6-t-butyl-m-cresol).

For pathological analysis, complete necropsies were performed on all animals using standardized methodology. At necropsy, all tissues, including masses and other macroscopic abnormalities, were removed and fixed in 10% neutral buffered formalin for microscopical evaluation. After fixation, the tissues were trimmed, dehydrated, cleared, and paraffin-embedded. Five-μm-thick sections were mounted onto glass slides, stained with hematoxylin and eosin (H&E), and examined microscopically. According to the “Guides for Toxicologic Pathology” (Ruben et al., 2000), we used the following criterion for the diagnosis of atrial thrombosis: the composition of fibrin, platelets, and mixed inflammatory cells in the atrial lumen. All data, including those from pathological examination, were obtained according to GLP standards, underwent extensive pathology peer review by an external expert panel advisory board, and are available online 〈http://ntp.niehs.nih.gov〉.

The function of endothelium-derived nitric oxide synthase (eNOS) in early atrial thrombotic change was investigated. Paraffin-embedded heart sections from F344 rats, each of 10 controls and 10 or 13 animals exposed dermally to 600 mg/kg CEM for 2, 3, 5, and 16 days were analyzed immunohistochemically to determine the effects of treatment on the expression of eNOS 〈http://dir.niehs.gov/dirlep/immuno/retrievals.htm〉. These tissues were derived from the mechanistic study of CEM-induced cardiac toxicities (NTP, 2003a). Immunohistochemistry was performed using the avidinbiotin-peroxidase technique. Tissue slides were deparaffinized in xylene and alcohols and placed in a 3% hydrogen peroxide solution for 15 minutes. Antigen retrieval was achieved by placing the slides in 1X citrate buffer, pH 6.0 (Biocare Medical, Walnut Creek, CA), boiling them in the decloaker, and cooling for 10 minutes. Rabbit anti-NOS3 (eNOS) antibody (1:600, Santa Cruz Biotechnology, Santa Cruz, CA) was applied for 1 hour. The secondary antibody (1:800, Vector Labs Burlingame, CA) was applied for 30 minutes at room temperature. The immunohistochemical localization was visualized by using liquid DAB (3,3′-diaminobenzidine tetrahydrochloride) (DakoCytomation, Carpinteria, CA), applied for 30 minutes at room temperature according to instructions given in the Vectastain Elite Kit (Vector Labs, Burlingame, CA), and counterstaining with a modified Harris hematoxylin (Richard Allen, Kalamazoo, MI). All incubations were performed in a humidifying chamber. Slides were dehydrated and coverslipped with Permount (Surgipath, Richmond, IL).

To designate the grading of eNOS expression, the intensity of the immunopositivity and the relative area of the sections exhibiting staining in the endocardial cells in the left atrium were graded by 2 pathologists using a scale ranging from 0 (−) to 2 (++) as follows: (0) = no specific immunohistological reaction visible in endocardial cells, (1) = up to 50% of total area showing a weakly positive reaction, and (2) = up to 100% of total area showing a strongly positive reaction.

Spontaneous Occurrence of Atrial Thrombosis in F344 Rats and B6C3F1 Mice

Historical control data from the NTP reports of cardiac thrombosis in both genders of B6C3F1 mice and F344 rats are shown in Table 2. In 90-day studies of rats and mice, no cardiac thrombosis occurred spontaneously. The rate of occurrence of cardiac thrombosis in 2-year mouse studies, however, was 20 of 2,798 males and 21 of 3,110 females (incidence rates: 0.71% and 0.68%, respectively). In the 2-year rat studies, the incidence rates were higher than those of mice: 4.11% in males (134/3,257) and 1.01% in females (44/4,352). In the 2-year studies, the incidence rates of dead or sacrificed moribund animals exhibiting thrombosis were 85% (17/20) for male mice, 81% (17/21) for female mice, 88% (118/134) for male rats, and 91% (40/44) for female rats. All of the cardiac thrombosis in these data was atrial, of which the main localization was left atrial. Although these historical control data include results from all kinds of dosing routes, such as feed, gavage, inhalation, dermal, drinking water, and vaginal application, some difference(s) might, nonetheless, exist among animals dosed by different routes (data not shown).

The spontaneous occurrence of atrial thrombosis, also called auricular thrombosis, has been reported in several animals: rats (MacKenzie and Alison, 1990; Lewis, 1992; Elwell and Mahler, 1999; Ruben et al., 2000; Elangbam et al., 2002), mice (Maita et al., 1988; Hagiwara et al., 1996; Elangbam et al., 2002), cotton rats (Sorden and Watts, 1996), hamsters (Liu and Tilley, 1980; Allen et al., 1985), monkeys (Wood et al., 1978; Kessler and London, 1982; Allen et al., 1985), dogs (Jubb and Kennedy, 1993; Ayers and Jones, 1978), and cats (Jubb and Kennedy, 1993; Ayers and Jones, 1978; Liu and Tilley, 1980). The highest incidence of spontaneous atrial thrombosis reported was 65% in retired breeding BALB/c female mice, probably related to abnormalities of blood coagulation (Hagiwara et al., 1996). Atrial thrombosis has been described in aged animals, including rats and mice, often associated with myocardial lesions, such as degeneration, focal inflammation, mineralization, amyloid deposition, or degenerative myxoid lesions in heart valves (Sorden and Watts, 1996; Ruben et al., 2000; Elangbam et al., 2002). Our historical control data of spontaneously occurring atrial thrombosis might be useful in the evaluation of possible chemical- and drug-related changes in the incidence of this condition, since, in F344 rats and B6C3F1 mice, few explicit reports exist (Ayers and Jones, 1978; Lewis, 1992). Our data reveal a lower incidence of spontaneous atrial thrombosis in F344 rats and B6C3F1 mice, compared to that in other murine strains. The relationship between the occurrence of atrial thrombosis and other cardiac lesions, especially spontaneous cardiomyopathy, is not clear, because animals with atrial thrombosis do not always exhibit a severe degree of cardiomyopathy, and animals with a severe degree of cardiomyopathy do not always develop atrial thrombosis (data not shown). Thus, our data indicate that spontaneous atrial thrombosis might not always be related to cardiomyopathy in F344 rats and B6C3F1 mice.

Genetic modifiers of the coagulation response have been examined as risk factors in cardiac thrombosis using several genetically modified mice: factor IX-overexpression mice with an abnormal blood-coagulation system (Ameri et al., 2003), tissue-type and/or urokinase plasminogen-activator knockout mice with disturbances in the fibrinolytic/coagulation cascade (Christie et al., 1999), and At^m/m mice showing a mutation in the antithrombin gene (Dewerchin et al., 2003). Additional reports have presented relationships between the occurrence of cardiac thrombosis and gene-modified murine strains, including ja/ja (jaundiced) mice with a deficiency in erythroid β-sectrin (Kaysser et al., 1997); β-tropomyosin-overexpression mice exhibiting abnormal cardiomyocytic contraction and relaxation, which culminates in altered blood flow and thrombus formation (Muthuchamy et al., 1998); tumor necrosis factor-α-overexpression mice with severely reduced cardiac output and enhanced thrombogenic potential of the endothelium as a direct result of local expression of cytokines and up-regulation of molecules, such as tissue factor (Bryant et al., 1998); β₂-adrenergic receptor-overexpression mice exhibiting severe myocardial hypertrophy and fibrosis (Du et al., 2000); and the spontaneously hypertensive (SHR) rat, with a hydrodynamic or rheologic problem in the lumen and structural changes in the vascular and cardiac walls (Nagaoka et al., 1970). Dietary fat also contributes to the development of heart thrombosis in some murine strains: C3H/OUJ mice with a secondary change in dystrophic cardiac calcinosis (Everitt et al., 1988); TS (Taconic Swiss), RF (Rockefeller Institute strain), pregnant RF, and Swiss mice; and epinephrine- and endotoxin-treated Holtzman rats (Ball et al., 1965; Clower, 1968; Clower and Douglas, 1968; Lockwood et al., 1969; Renaud and Godu, 1969; Wicks et al., 1969; Clower and Lockwood, 1972; Davenport and Ball, 1981; Hagiwara et al., 1996) that develop fat-induced cardiovascular lesions, endocardial damage, or increased susceptibility of platelets to thrombin-induced aggregation. Mural thrombosis might be related to severe anemia induced by excessive fat in the diet, since atrial thrombosis was prevented after therapeutic injection of red blood cells into Swiss mice (Ashburn et al., 1972). Dietary copper deficiency also induced atrial thrombosis in Swiss Webster mice, acccompanied by cardiac enlargement due to decreased coronary resistance and/or hemodynamic overload related to severe anemia (Lynch and Klevay, 1994; Klevay, 1985). Spontaneous induction of atrial thrombosis in B6C3F1 mice and F344 rats has not been clearly related to genetic or dietary factors.

Chemical-Induced Atrial Thrombosis

Table 1 presents the results of our examination of the data of these 13 compounds to determine the incidences and main localization of heart thrombosis. In the chemical-exposed animals exhibiting cardiac thrombosis, the incidence was 20–100% among the groups (Table 1). The incidence rate of dead or sacrificed-moribund animals exhibiting thrombosis was 78% (274/351 animals), similar to that of the spontaneous occurrence of thrombosis in control animals (Table 2). Thrombosis constitutes a serious disorder that can prove fatal in humans (Virmani et al., 2001; Lowe, 2004) and animals (Jubb and Kennedy, 1993). The data that we uncovered suggest that heart thrombosis could be the main cause of death and moribundity in rats and mice exposed to these chemicals. We formed 3 categories indicating possible relationships to other lesions: cardiac damage, hematological change, and the occurrence of multiple tumors.

Relationship Between Chemical-Induced Atrial Thrombosis and Cardiac Damage

Chronic heart failure in humans, known to confer upon patients a greater risk of thromboembolism, is likely related to numerous, diverse factors, such as vascular abnormalities, increased coagulability, impaired blood flow, and low cardiac output that promotes formation of fibrin-rich clots. In addition, defective endothelial function and significant levels of circulating aggregates of platelets have been demonstrated in heart-failure patients (De Lorenzo et al., 2003). Atrial thrombosis in humans can occur in association with cardiac dysfunction and abnormal blood flow (whirlpool) in the atrium (Suetsugu et al., 1988; Bilge et al., 1999; De Lorenzo et al., 2003). Autopsies and echocardiographic studies have indicated near-50% incidences of thromboembolic events in patients with acute and chronic heart failures (Asinger et al., 1981; De Lorenzo et al., 2003).

We were able to identify 3 chemicals from the NTP database, which might induce atrial thrombosis secondary to myocardial injury. We made this identification based upon the progression and localization of this injury.

Bis(2-chloroethoxy)methane (CEM)

In the 13-week rat dermal study (NTP, 2003b), atrial thrombosis with myocardial degenerative change was noted in 3 of 10 male rats treated with 600 mg/kg CEM (Table 1). In the time-course study of CEM-induced cardiac toxicity, atrial thrombosis was interpreted to be secondary to increased severity of heart damage (NTP, 2003a). This heart lesion was characterized by myocardial vacuolization, myocytic necrosis, mononuclear-cell infiltration, and fibrosis in all myocardial areas, including the left atrium (Figure 1a, b) (Dunnick et al., 2004a, 2004b). Damage in myocardial mitochondria, constituting one of the earliest changes of myocardial damage, led to vacuolar formation (Dunnick et al., 2004a). The mechanism of damage appears related to one of the metabolites of CEM, thiodiglycoic acid, known to cause mitochondrial damage in the myocardium.

Diethanolamine (DEL)

In the mouse 2-year dermal study (NTP, 1999), higher incidences of atrial thrombosis were noted in male and female mice treated with 160 mg/kg DEL (Table 1). The lesions were detected in 6 of 50 males and 4 of 50 females (incidence rates: 12% and 8%, respectively) without other heart lesions; however, atrial thrombosis was not attributed to chemical-induced toxicity in the original report. In the previous 13-week mouse studies, myocardial degeneration could be seen in high-dose groups (NTP, 1992c; Melnick et al., 1994). The possibility exists that DEL induced cardiac effects in the 2-year mouse study, although myocardial damage was not detected histopathologically. Atrial thrombosis may, therefore, have been induced by DEL via cardiac functional changes.

Oxazepam (OZP)

In the 57-week study using Swiss Webster mice, the lower survival of animals receiving OZP was attributable to an increase in the extent and severity of amyloid deposits in many organs. Atrial thrombosis and pulmonary lesions consistent with chronic heart failure occurred at higher incidences and with greater severity (Figure 1c, 1d) (NTP, 1993; Bucher et al., 1994). The incidence of atrial thrombosis was 34/60 males (57%) and 23/59 females (40%) in the 2,500-ppm groups and 35/60 males (58%) and 31/59 females (53%) in the 5,000-ppm OZP-treated groups (Table 1). In CD mice, systemic amyloidosis was seen, complicated with auricular thrombosis (Maita et al., 1988). The Swiss Webster mouse strain has been reported to exhibit a high incidence of cardiac amyloidosis, which is related to chronic dermatitis (Gruys et al., 1996; Elwell and Mahler, 1999), compared to naturally occurring amyloidosis in the hearts of B6C3F1 mice that were used in the 2-year NTP study (Elwell and Mahler, 1999). The reason that atrial thrombosis occurred only in the study using Swiss Webster mice might be related to the functional damage induced by amyloidosis in the heart. Systemic amyloidosis in humans induced abnormal prothrombin time, fibrinogenopenia, increased fibrinolysis, abnormal coagulation factors, and/or hemolytic anemia, accompanied by chronic intravascular coagulation, probably due to vascular damage produced by perivascular amyloid deposition (Bowie et al., 1969; Santarone et al., 1999; Bick, 2001). A high prevalence of atrial thrombosis was found in human patients with cardiac amyloidosis, arising from atrial dysfunction ascribed to the combination of amyloid infiltration of atrial walls and increased atrial afterload caused by restrictive hemodynamics (Santarone et al., 1999). Atrial amyloidosis has also induced atrial fibrillation, followed by atrial stretch and endocardial remodeling (Goette and Lendeckel, 2004).

Other Chemicals That Induced Atrial Thrombosis Related to Myocardial Damage

Several chemicals have been reported to induce severe myocardial damage, followed by atrial thrombosis. This pathology has occurred at an incidence of up to 75% (left atrium: 9/12 mice, right atrium: 2/12 mice) in male ICR mice treated intravenously with 4 mg/kg of doxorubicin (Fujihira et al., 1993). An intraperitoneal injection of 100 mg/kg of 3-nitropropionic acid induced atrial thrombosis in several strains of mice, but not B6C3F1 (Gabrielson et al., 2001). Feeding of 0.1% hydrochlorothiazide (Lijinsky and Reuber, 1987) or more than 250 ppm quinacrine to F344 rats (Reuber and Lijinsky, 1984) induced thrombosis. Of these 4 chemicals mentioned, the NTP conducted the toxicology and carcinogenesis feeding studies of 2, using F344 rats and B6C3F1 mice: 3-nitropropionic acid (NTP, 1978) and hydrochlorothiazide (NTP, 1989b). In the NTP report of 3-nitropropionic acid, an increased incidence of atrial thrombosis was not noted, although this chemical induced myocardial damage similar to that reported in the study of Gabrielson et al. (2001). The reason for the differing results may be dependent on the different mouse strains and/or dosing routes used. In the NTP report on hydrochlorothiazide, the induction of atrial thrombosis was not detected in the same strain of rats treated with a dosing volume higher than that used in the Lijinsky study. No reasons were offered for the differences in the results of the induction of atrial thrombosis between the Lijinsky study (Lijinsky and Reuber, 1987) and the NTP study (NTP, 1989b). In the Netherlands, the TNO conducted, in Wistar rats, a 29-month toxicity study of methyl bromide, which induced atrial thrombosis accompanied by myocardial degeneration (Reuzel et al., 1991). Causes of this atrial thrombosis were suggested to be possibly related to mild sustained endocardial damage and changes in the blood flow secondary to cardiac muscular damage. Catecholamines induced myocardial damage and intravascular aggregation of platelets (thrombus formation) in many species of animals, including rats (Haft, 1974). In rats treated with troglitazine for 2 years, atrial thrombosis occurred accompanied by myocardial damage, such as atrial fibrosis and karyomegaly (Herman et al., 2002).

Relationship Between Chemical-Induced Atrial Thrombosis and Hematological Changes

In humans, chemically induced hemolysis is a well-known phenomenon occurring with some association with thrombosis. Antineoplastic drugs, such as deoxycoformycin, pentostatin, cisplatin, and mitomycin, have been associated in humans with thrombotic thrombocytopenic purpura (TTP) characterized by hemolysis and formation of microthrombi in many organs, including heart (Bonner and Erslev, 1994; Leach et al., 1999; Ezov et al., 2002; Dlott et al., 2004). The most tenable hypothesis holds that TPP results from the introduction into the circulation of one or more platelet-aggregating substances due to immune-mediated or drug-induced direct toxicity (Bonner and Erslev, 1994; Ezov et al., 2002; Dlott et al., 2004). The direct toxicity of mitomycin on endothelial function might play an important role in the pathogenesis of TTP (Dlott et al., 2004). In human myeloma patients, thalidomide induced atrial or deep-vein thrombosis, probably due to sinus rhythm in the heart (Urbauer et al., 2002; Jego et al., 2003). Retinoic acid induced intraventricular thrombosis in human leukemia patients (Barbui et al., 1998; Falanga et al., 2003). The mechanism has been in part mediated by an increased expression of adhesion molecules that facilitate adhesion of cells to vascular endothelium, thereby promoting localized coagulation (Torromeo et al., 2001).

Several chemicals and drugs have induced methemoglobinemia, such as the antimalarials, chloroquine and primaquine; local anesthetics (lignocaine, benzocaine, and prilocaine); glyceryl trinitrate; sulphonamides; and phenacetin (Coleman and Coleman, 1996; Hall et al., 1986). The formation of methemoglobin induced by these chemicals, resulting in the production of blood clots, apparently occurs predictably within a thrombus (Moody, 2003). Drug-induced platelet antibodies have been demonstrated to downregulate or enhance platelet function (Kekomaki, 2003). Sulfonamides induced immune thrombocytopenia, decreased platelet production, or increased destruction of platelets (Van den Bemt et al., 2004). Heparin-induced thrombocytopenia complicated by thrombosis was associated with high levels of drug-dependent antibodies (Kekomaki, 2003).

From the NTP database, we were able to select 5 chemicals that might induce atrial thrombosis secondary to or probably related to hematological changes, especially hemolytic anemia.

2-Butoxyethanol (2-BE)

In the 14-week rat inhalation study (NTP, 1998a), atrial thrombosis without myocardial degenerative change was noted in one of 5 females treated with 500 ppm 2-BE (Table 1, Figure 1e). In the mechanistic study of BE-induced toxicity, considerable evidence was reported of systemic disseminated thrombosis in tissues including heart (atrium), nasal cavity, incisor, liver, lung, femur, brain, eye, and/or tail (coccygeal vertebrae), followed by infarction in some organs (NTP, 1998a; Ghanayem et al., 2001; Ezov et al., 2002; Nyska et al., 2003, 2005; Redlich et al., 2004; Shabat et al., 2004; Lewis et al., 2005). Hematological analysis indicated severe hemolytic anemia and mild thrombocytosis. The thrombosis in rats may represent a cycle of events in which an initial low level of endothelial activation and/or dysfunction triggered by hemolysis and hypoxia result in additional vascular problems, including enhanced erythrocytic adherence to endothelial cells, overexpressions of vascular cell adhesion molecule-1 (VCAM-1) and endothelial intercellular adhesion molecule-1 (ICAM-1), poor blood flow, vascular occlusion, thrombosis, infarction, and additional hypoxia (Koshkaryev et al., 2003; Nyska et al., 2003; Potti et al., 2004; Redlich et al., 2004; Shabat et al., 2004).

These histopathological changes in rats have provided models of sickle cell anemia in humans (Ghanayem et al., 2001; Ezov et al., 2002; Pathare et al., 2003). Metabolic activation of BE to form butoxyacetic acid (BAA) is a prerequisite for the development of hematotoxicity (Ghanayem et al., 1987, 2001; Ghanayem and Sullivan, 1993). Human erythrocytes exposed to BAA have manifested changes similar to those seen in rat erythrocytes, such as deformability, osmotic fragility, and changes in sodium content, though human erythrocytes appear to be relatively resistant to the hemolytic effects of BAA (Ghanayem, 1989; Ghanayem and Sullivan, 1993; Udden, 2002).

C.I. Pigment Red 23 (Red 23)

In the NTP 2-year rat study (NTP, 1992b), atrial thrombosis was not attributable to chemical-induced toxicity; however, the incidence was significantly higher in the 50,000-ppm Red 23-treated male group than in the control group (Table 1; incidence: 20%, 10/50 males). In the 2-week preliminary study and the interim evaluations of the 2-year study (3 and 15 months), hemolytic anemia was noted at treatment levels greater than 50,000 ppm. The presence of anemia in rats and its absence in mice may be related to the difference in life span of erythrocytes—50 to 65 days for rats versus 20 to 30 days for mice (NTP, 1992b). The short life span of erythrocytes in mice enables them to replace damaged cells faster than rats, thus maintaining these hematologic parameters within normal values. The mechanism of atrial thrombosis induced by Red 23 has not been clarified; however, chemical-induced anemia may be attributable to the induction of atrial thrombosis.

Diazoaminobenzene (DAB)

In the NTP 2-week mouse study (NTP, 2002), atrial thrombosis was noted in chemical-induced lesions without myocardial changes in groups of males treated with 100 mg/kg or more (Table 1; incidence: 5/5 males, 100 mg/kg; 5/5 males, 200 mg/kg). Clinical-pathology data indicated chemical-induced methemoglobinemia, Heinz-body formation, and hemolytic anemia. Methemoglobin is hemoglobin with iron oxidized to the ferric (Fe⁺⁺⁺) state from the normal, or reduced, ferrous (Fe⁺⁺) state and rendered incapable of transporting oxygen to all of the organs. The production of methemoglobin causes the production of blood clots (Moody, 2003). The higher incidence of atrial thrombosis might be related to methemoglobinemia induced by DAB.

Methyleugenol (MEG)

Our search for atrial thrombosis in the 2-year mouse study revealed the higher incidence of this lesion in groups of females treated with 75 mg/kg or more of MEG. The incidence showed 6/48 and 9/44 rats in the 75 and 150 mg/kg groups, respectively (Table 1). In the NTP and other previous reports in mice (Johnson et al., 2000; NTP, 2000; Abdo et al., 2001; Smith et al., 2002), atrial thrombosis was not noted among the chemical-induced lesions, although the incidence was higher in the 2-year study than that seen in control and background data. Although hematological analysis was not conducted in any mouse studies, evidence indicated the occurrence of thrombocytosis, demonstrated by increased platelet counts in the groups that received 100 mg/kg or greater in the 14-week rat study (NTP, 2000; Abdo et al., 2001). Anemia was noted in groups that received more than 300 mg/kg, and bone marrow changes were noted in both rat and mouse studies; therefore, anemia might have occurred in the 2-year mouse studies. Disturbed blood flow may result from alterations in the properties of erythrocytes, such as increased adherence to the endothelium of the blood vessel wall (Koshkaryev et al., 2003). Recent research showed that methyleugenol elicited hypotension and bradycardia, effects that appeared related to an active vascular relaxation (Lahlou et al., 2004). The mechanism of the induction of atrial thrombosis has not been elucidated; however, chemical-induced anemia, probably hemolytic, and cardiovascular effects, such as decreased blood flow in the atrium, may contribute to the induction of atrial thrombosis.

4,4′-Thiobis(6-t-butyl-m-cresol) (TBBC)

In the NTP 2-year rat study (NTP, 1994), atrial thrombosis was not considered a chemical-induced effect; however, the incidence was increased in groups of males that received 500 ppm or more of TBBC, without dose-dependency (Table 1; 12%, 8%, 6% incidence rates in 500, 1000, 2500 ppm-treated groups, respectively). Anemia and an increased number of platelets induced by TBBC were noted in the 2500 ppm-treated rat group and 1000 ppm-treated mouse group. In the original report, alteration in platelets was observed in association with inflammation, trauma, surgery, hyposplenic or asplenic states, malignancies, acute blood loss, and hyperadreno corticism. These associations appear to offer an incomplete explanation of the basic processes involved in the induction of atrial thrombosis. While the mechanism of induction has not been clarified, TBBC-induced anemia (probably hemolytic) should be considered a possible cause of atrial thrombosis.

Relationship Between Chemical-Induced Atrial Thrombosis and Occurrence of Multiple Tumors

In cancer of humans and animals, the development of thrombosis involves a complex interaction between the tumor cell, the patient, and the hemostatic system (Khato et al., 1977; Tanabe et al., 1999; Schafer et al., 2003). Hyperfibrinogenemia contributes to the hypercoagulable state due to a compensatory overproduction of clotting factors in cancer patients (Khato et al., 1977; Tanabe et al., 1999). Tumor-bearing mice given TNF exhibited intravascular clot formation with fibrin deposition in vivo. Activation of coagulation of the matrix from TNF-stimulated human endothelial cells was dependent on the presence of platelets, indicating their important role in propagating reactions leading to formation of fibrin in vitro (Tijburg et al., 1991; Jaimes et al., 2001). Neoplastic and/or endothelial cells in tumors, through expression of tissue factors or cytokines, such as TNFα, can activate coagulation (Ray, 2000; Philipp et al., 2003; Schafer et al., 2003). Some kinds of malignant tumors, such as lymphoma, stimulate megakaryocytopoiesis and platelet production during growth. The mechanisms may be related to the production by malignant tumors of an array of cytokines, such as IL-6, a potent stimulator of platelet production (Ray, 2000).

In attempting to select chemicals from the NTP studies that might be inducers of atrial thrombosis and probably involved also in the occurrence of multiple tumors, we found and describe 3 of them next.

C.I. Acid Red 114 (Red 114)

In the NTP 2-year rat study (NTP, 1991), atrial thrombosis was noted among the chemical-induced lesions. The incidence was higher, compared to controls, in groups of males that received 150 ppm or more (Table 1; 28%, 36% in 150 and 300 ppm Red 114-treated groups, respectively). The NTP report described the following potential mechanism: “Thrombosis may have formed as a consequence of debilitation in tumor-bearing animals. Debilitation may have led to impaired circulation which allowed pooling of blood within the atrium, resulting in thrombus formation.” In the 3-month preliminary study and the interim evaluations (9 and 15 months) of the 2-year rat studies, anemia—the poorly regenerative type–and hypocellarity of bone marrow were noted, suggesting that this chemical exerts a direct effect on hematopoietic cells at high-dose levels. Chemical-induced multiple tumors were noted in skin; zymbal, clitoral, mammary, and adrenal glands; liver; oral cavity; intestine; and lung. Some animals also developed mononuclear-cell leukemia. Although the mechanism of the induction of atrial thrombosis has not been elucidated, chemical-induced anemia and tumorigenesis may be involved.

C.I. Direct Blue 15 (Blue 15)

In the NTP report (NTP, 1992a), atrial thrombosis was noted among the chemical-induced lesions in rats, without any description of the mechanism (Figure 1f). The incidence was higher in groups of males that received 1250 ppm or more Blue 15 (Table 1; 26% and 24% in 1250 and 2500 ppm-treated groups, respectively). Blue 15-induced multiple tumors were noted in skin; zymbal, clitoral, preputial, and adrenal glands; liver; oral cavity; intestine; uterus; and brain. Some animals developed mononuclear-cell leukemia. In the 3-, 9- and 15-month rat studies, no induction of any cardiac lesions was documented (Morgan et al., 1989; NTP, 1992a). The mechanism of the induction of atrial thrombosis by Blue 15 has not been clarified, but we speculate that it might be related to tumorigenesis.

3,3′-Dimethoxybenzidine Dihydrochloride (DMOB)

In the NTP 2-year rat study (NTP, 1990), increased incidences of atrial thrombosis in exposed males led to impaired circulation and sludging of blood in the atrial chambers, as well as increased morbidity. The incidence rates were 15/44 (34%), 27/75 (36%), and 23/60 rats (38%) in 80, 170, and 330 ppm DMOB-treated male groups, respectively (Table 1). The compound induced many kinds of tumors in different tissues (liver; intestine; zymbal, preputial, clitoral, and mammary glands; oral cavity; skin; brain; uterus; mesothelium); no vascular tumors were seen in any organs. Since this chemical is one of the metabolites of Blue 15 (NTP, 1990, 1992a), we speculate that the induction of atrial thrombosis by DMOB might be related to multiple tumorigenesis, as suggested for Blue 15.

Other Chemicals That Induced Atrial Thrombosis Related to Vascular Tumors

Vascular tumors may sometimes be related to the pathogenesis of thrombosis in other sites due to endothelial damage, although the mechanisms are not clear (Miyamoto et al., 1992). Multiple hemorrhagic and thrombotic episodes involving vascular tumors can lead to thrombocytopenic and hemorrhagic crises (Herman et al., 2002). The compound, 4-aminobiphenyl, induced left atrial thrombosis and hemangiosarcoma in numerous organs in a mouse carcinogenesis study (Schieferstein et al., 1985). Although the cause of the induction may have involved a marked toxic effect on the hematopoietic system, the relationship between the occurrence of atrial thrombosis and hemangiosarcoma should be analyzed. In our investigation of 13 chemicals found in the NTP database, no induction of vascular tumors in applicable rodent studies involved chemical exposure.

Unknown Mechanisms of Chemical-Induced Atrial Thrombosis

We could not speculate concerning the mechanism(s) of atrial thrombosis induced by 2 chemicals obtained from the NTP database and literature search, because no hematological changes or related lesions were induced.

Hexachloroethane (HCE)

An increased incidence of atrial thrombosis (incidence rate, 10%) occurred in the male group that received 20 mg/kg in the NTP 2-year rat study (Table 1; NTP, 1989a); however, induction of toxicity by HCE was not determined (NTP, 1989a). No descriptions of hematological changes and cardiovascular effects have been reported in rat studies (Weeks et al., 1979; Gorzinski et al., 1985), similar to reports from human cases (ATSDR, 1997).

Isobutene (IBT)

In the NTP 2-year rat study (NTP, 1998b), the higher incidence of atrial thrombosis in the 8000 ppm IBT-treated male group was noted (12%, 6/50). The NTP decided, however, that this marginal increase was not related to chemical exposure, since this lesion occurs commonly and spontaneously at a low incidence in chronic rat studies. Hematological analysis was not conducted in this study.

Both of the aforementioned chemicals induced atrial thrombosis at increased incidences, compared to controls and NTP historical background data. Both HCE and IBT should be investigated further to distinguish between true and false positivity and clarify the mechanism(s) of the induction of atrial thrombosis.

Relationship Between Chemically Induced Atrial Thrombosis and Damage in Other Organs

Severe renal disease sometimes results in secondary myocardial fibrosis with the appearance of left atrial thrombosis in more advanced cases (Glaister, 1986; Citak et al., 2000). Researchers have speculated that lipoprotein in the human nephritic syndrome may promote thrombosis (Stenvinkel et al., 1993). Thromboembolism in patients with renal disease may be related to low levels of plasma antithrombin III and albumin and high levels of fibrinogen and cholesterol (Citak et al., 2000; Aoyagi et al., 2002). In our studies of 13 chemicals used to treat mice and rats, some compounds induced renal lesions in the same groups in which the increased incidences of atrial thrombosis occurred. The occurrence of atrial thrombosis, however, was not by itself related to chemical-induced renal lesions (data not shown).

Spontaneous hippocampal neuronal necrosis that developed in aged male F344 rats used in NTP studies was reported to result from an impairment of cerebral perfusion, secondary to vascular obstruction caused by atrial thrombosis or the occurrence of leukemic cells and haemolytic anemia concomitant with mononuclear-cell leukemia, which commonly occurs in F344 rats (Barbolt and Everette, 1990). In our cases, no evidence could be discovered that atrial thrombosis was related to the occurrence of any neural lesions or leukemia induced by chemical exposure alone (data not shown).

Repeated inhalation of 3 ppm methyl isocyanate for 4 days induced left atrial thrombosis in F344 rats; dead animals exhibited severe lung toxicity (Mitsumori et al., 1987). Although this study was conducted by the NTP, we were unable to include this chemical in our investigation because of the inability to locate the study findings at the given web site 〈http://ntp-server.niehs.nih.gov/index〉. The mechanism of the induction of atrial thrombosis was not clarified but, in the study report, was considered secondary to alveolar damage or tissue hypoxia, because thrombus formation was always restricted to the left atrium that drains the pulmonary vein (Mitsumori et al., 1987). Large atrial thrombosis, however, causes secondary passive pulmonary lesions, such as congestion, sometimes with alveolar fibroplasias (Lewis, 1992). Thrombosis may not, therefore, occur secondary to lung damage.

Certain infectious pathogens have long been suspected of playing a role in the process leading to cardiac thromboembolic complications in humans and animals, such as acute myocarditis induced by encephalomyocarditis in mice (Tomioka et al., 1985), bacterial endocarditis or myocarditis in monkeys infected by Staphylococcus or other bacteria (Kessler and London, 1982; Wood et al., 1978), and chronic infection with Helicobactor pylori in mice resulting in increased platelet embolization after damage to mesenteric arterioles (Aguejouf et al., 2003). In our cases, no evidence was found that atrial thrombosis was related to the occurrence of systemic pathogenic infection (data not shown).

Differences in Responses by Gender and Animal to Chemically Induced Atrial Thrombosis

Our research suggests differing responses to chemical-induced atrial thrombosis of sex and strain of animals. The reason for the gender differences in 2-BE-induced toxicity may be different rates of production of sufficiently high levels of the hematotoxic metabolite, butoxyacetic acid (Koshkaryev et al., 2003). With respect to the strain difference of the occurrence of myocardial damage induced by CEM, the greater severity of CEM-induced heart toxicity in rats than mice may have been due to their higher rates of production of thiodiglycolic acid, a metabolite of CEM (Dunnick et al., 2004b). No data were available, however, from toxicokinetic studies of other chemicals, and differing sensitivities by sexes and strains were not described in these NTP studies. Future detailed investigations of the toxicokinetic characteristics of each chemical and its metabolite(s) are needed to clarify underlying reasons for gender and strain differences in chemically-induced atrial thrombosis.

Function of Endothelium-Derived Nitric Oxide Synthase (eNOS) in Early Change of Atrial Thrombosis

Endothelial dysfunction is considered the major risk factor and a very early indicator of cardiovascular disease, including murine thrombosis (Carter and Gavin, 1989; Triggle et al., 2003). Impaired endothelium-dependent functions result primarily from decreased synthesis of endothelium-derived NO and/or an increase in the production of reactive oxygen species, such as superoxide (Triggle et al., 2003; Davis et al., 2004). Porcine atrial fibrillation causes a downregulation of the production of atrial eNOS and NO and a comparative increase in the expression of plasminogen activator inhibitor-1 in the left atrium during alterations comprising endocardial remodeling; these collective changes were considered one of the potential mechanisms for induction of left atrial thrombosis (Cai et al., 2002; Goette and Lendeckel, 2004). Nitric oxide also exerts powerful antithrombotic effects in atrial endothelium and causes inhibition of platelet activation (Goette and Lendeckel, 2004).

We analyzed the expression of eNOS in the left atrial endocardial cells in rats exposed to 600 mg/kg CEM for 2, 3, 5, and 16 days (Figure 2a,b,c). In the endocardial cells of control rats, the mean scores were 1.7 to 2.2 (Figure 2a,b). In contrast, the scores of animals treated with CEM for 2 and 16 days decreased with statistical significance and were 1.4 and 0.7, respectively (Figure 2a,c). Histological evaluations indicating the significance of damage at these time points have been described (Dunnick et al., 2004a). In addition, atrial myocardial cells exhibited weak positivity for eNOS in some control animals (Figure 2b), similar to a localization reported previously (Balligand and Cannon, 1997). Treatment with CEM for 13 weeks induced left atrial thrombosis in rats (Table 1). Our data and review of the literature suggest that decreased expression of eNOS in atrial endocardial cells may be an important and critical factor involved in early changes leading to thrombus formation. Additional investigations are necessary to clarify the relationship(s) between eNOS expression and the induction of atrial thrombosis by the other chemicals that we have listed in this paper.

Why Spontaneously Occurring and Chemical-Induced Atrial Thrombosis Develops Mainly on the Left Side

Data from the NTP rodent studies show that both spontaneously occurring and chemically inducible atrial thrombosis occurred mainly on the left side (Tables 1 and 2). Several researchers have addressed the reasons for such localization in humans and animals (Ayers and Jones, 1978; Lewis, 1992; Al-Saady et al., 1999; Bilge et al., 1999; Elwell and Mahler, 1999; Ruben et al., 2000; Cai et al., 2002). In humans, left atrial thrombosis has usually been considered the source of embolic events in acute infarction (Bilge et al., 1999), although it has been infrequently detected in the presence of sinus rhythm in the heart (Agmon et al., 2002). In canine cases, atrial thrombosis occurred as a terminal event in atrial fibrillation; the majority of cases may be attributable to eddying or stasis of blood in the atrium or its auricular appendage (Jubb and Kennedy, 1970). The incidence of thrombosis in atrial fibrillation implies a role for an atrial hemodynamic factor; atrial thrombosis has been associated with atrial fibrillation-induced structural changes in the atrium, such as decreased contraction and dilatation of the atrial appendage (Bankl et al., 1995; Al-Saady et al., 1999; Goette and Lendeckel, 2004). Approximately 90% of atrial thrombi in nonrheumatic atrial fibrillation were seen within the left atrial appendage in human patients (Al-Saady et al., 1999).

Atrial fibrillation has been manifested in baseline electrocardiographs as cardiac rhythm showing irregular undulations of varying amplitude, contour, and spacing (Aronow, 2002). Also a common complication of cardiac operations leading to increased risk for thromboembolism, it is attributable to age-related structural changes in the human atrium, such as dilatation, muscular atrophy, decreased conduction throughout tissue, and fibrosis (Hogue and Hyder, 2000). The left atrial appendage, a long, tubular, hooked structure, is the remnant of the original embryonic left atrium; usually crenellated, with a narrow junction with the venous component of the atrium (Al-Saady et al., 1999); and closely related in its superior aspect to the pulmonary artery and inferomedially to the free wall of the left ventricle (Al-Saady et al., 1999; Nishimura et al., 2003). A possible reason for the onset of fibrillation may be stagnation of blood in the left atrial appendage (Suetsugu et al., 1988), which is a muscular chamber acting as a contractile pump with a characteristic pattern of contraction (Bilge et al., 1999). In rodents, this appendage appears to have a location similar to that in humans; cardiac hypertrophy occurs therein at an incidence of 77% in F344 rats aged 9 to 27 months (Boluyt et al., 1999). Local radiation administered to the rat heart induced fibrosis and thrombosis in the left atrium; considerable evidence has indicated that the left atrial appendage significantly contributes to left ventricular filling and plays a pivotal role in maintaining normal cardiac status, especially in states of cardiac disease (Kruse et al., 2001). The reason for the predilection for the appendage in atrial thrombosis may involve not only its distinctive anatomy, with the inner surface marked by muscular ridges, but also abnormalities in blood-flow patterns (Aronow, 1991; Bilge et al., 1999). Collectively, all of this information indicates that spontaneously occurring and chemical-induced murine atrial thromboses occur chiefly in the left side because of distinctive anatomical and hemodynamic characteristics of that region.

Mechanisms of Chemical Induction of Atrial Thrombosis and Possible Human Risk

Mechanisms of putative pathogenesis of thrombosis have been indicated by several specific effects caused by thrombosis-producing compounds: endothelial damage (homocysteine, endotoxin, sodium acetriozate); alterations in pathophysiologic circulatory dynamics (ergotamine, pitressin, oral contraceptives, acetylcholine, autonomic blockers); changes in platelets (serotonin, progesterone, testosterone, somatotropic hormone, vincristine, congo red, ristocein, thrombin, epinephrine, adenosin diphosphate, Evans blue); and transformations in clotting factors (epinephrine, guanethidine, debrisoquin, thyramine, lactic acid, long-chain fatty acids, catecholamines, ACTH, thymoleptics, nictotine, oral contraceptives, mercuric chloride, corticosteroids, aminocaproic acid, aprotine) (Ramos et al., 2001). Our retrospective investigation and review of the literature suggest that endothelial injury, circulatory stasis, and/or hypercoagulability—even though indirect, or secondary, effects induced by chemicals—might predispose the individual to cardiac thrombi. The resulting impaired atrial mechanical activity, occurring as atrial fibrillation and congestive heart failure, might cause stasis of blood within the left atrium, contributing to left atrial thrombosis (Figure 3).

There are reports in the literature of human left atrial thrombosis following atrial fibrillation (Al-Saady et al., 1999; Goette and Lendeckel, 2004). No reports are available concerning the risk of heart thrombosis in humans exposed to any one of the 13 compounds that we have reported to be associated with atrial thrombosis in the NTP studies. Some COX-2-inhibitors, such as Vioxx, pose a significant risk of cardiovascular thrombotic complications (Konstam and Weir, 1999; Bing and Lomnicka, 2002; Schmidt et al., 2004). Preclinical studies of these compounds did not reveal any potential risk for development of cardiac thrombosis (U.S. FDA, 1999). Investigations are needed to elucidate the reasons for the present inability to detect potential risk of thrombotic development in laboratory animals exposed to COX-2 inhibitors. Potential factors may include suitability of the animal model, doses selected for the testing, and different mechanisms leading to development of thrombosis.

Future detailed investigations of hematological and electrocardiological functions following exposure must be conducted to determine precisely which of these chemicals act as real inducers of cardiac toxicities. To concentrate on molecular functioning in such investigations could enhance understanding of the pathogenesis of chemical-induced atrial thrombosis, since the progression and risks to humans of this toxicity remain to be completely elucidated. Additional research must be completed to analyze the precise mechanism(s) of induction and provide understanding of potential extrapolations from rodents to humans of chemical-induced cardiovascular alterations.