Chapter 11: Toxic Responses of the Blood

Hematotoxicology is the study of adverse effects of drugs, nontherapeutic chemicals, and other chemicals in our environment on blood and blood-forming tissues. This subspecialty draws on the discipline of hematology and the principles of toxicology. Scientific understanding of the former began with the contributions of Leeuwenhoek and others in the 17th century with the microscopic examination of blood. Hematology was later recognized as an applied laboratory science but limited to quantitation of formed elements of the blood and the study of their morphology, along with that of bone marrow, spleen, and lymphoid tissues. It is now a diverse medical specialty, which—perhaps more than any other discipline—has made tremendous contributions to molecular medicine.

The vital functions that blood cells perform, together with the susceptibility of this highly proliferative tissue to intoxication, make the hematopoietic system unique as a target organ. Accordingly, it ranks with liver and kidney as among the most important considerations in the risk assessment of individual patient populations exposed to potential toxicants in the environment, workplace, and medicine cabinet.

The delivery of oxygen to tissues throughout the body, maintenance of vascular integrity, and provision of the many affector and effector immune functions necessary for host defense require a prodigious proliferative and regenerative capacity. Each of the various blood cells (erythrocytes, granulocytes, and platelets) is produced at a rate of approximately one to three million/s in a healthy adult and several times that rate in conditions where demand for these cells is high, as in hemolytic anemia or suppurative inflammation (Kaushansky, 2006). As with intestinal mucosa and gonads, this characteristic makes hematopoietic tissue a particularly sensitive target for cytoreductive or antimitotic agents, such as those used to treat cancer, infection, and immune-mediated disorders. This tissue is also susceptible to secondary effects of toxic agents that affect the supply of nutrients, such as iron; the clearance of toxins and metabolites, such as urea; or the production of vital growth factors, such as erythropoietin and granulocyte colony-stimulating factor (G-CSF).

The consequences of direct or indirect damage to blood cells and their precursors are predictable and potentially life-threatening. They include hypoxia, hemorrhage, and infection. These effects may be subclinical and slowly progressive or acute and fulminant, with dramatic clinical presentations. Hematotoxicity is usually assessed in the context of risk versus benefit. It may be used to define dosage in treatment modalities in which these effects are limiting, such as those employing certain anticancer, antiviral, and antithrombotic agents.

Hematotoxicity is generally regarded as unacceptable, however, in treatments for less serious illnesses, such as mild hypertension or arthritis, or following exposure to contaminated foods or environmental contaminants. Risk-versus-benefit decisions involving hematotoxicity may be controversial, especially when the incidence of these effects is very low. Irrespective of whether the effect is linked to the pharmacologic action of the drug, as with cytoreductive or thrombolytic agents, or unrelated to its intended action, the right balance between risk and benefit is not always clear.

Hematotoxicity may be regarded as primary, where one or more blood components are directly affected, or secondary, where the toxic effect is a consequence of other tissue injury or systemic disturbances. Primary toxicity is regarded as among the more common serious effects of xenobiotics, particularly drugs (Vandendries and Drews, 2006). Secondary toxicity is exceedingly common, due to the propensity of blood cells to reflect a wide range of local and systemic effects of toxicants on other tissues. These secondary effects on hematopoietic tissue are often more reactive or compensatory than toxic and provide the toxicologist and clinician with an important and accessible tool for monitoring and characterizing toxic responses.

The production of blood cells, or hematopoiesis, is a highly regulated sequence of events by which blood cell precursors proliferate and differentiate to meet the relentless needs of oxygen transport, host defense and repair, hemostasis, and other vital functions described previously. The bone marrow is the principal site of hematopoiesis in humans and most laboratory and domestic animals. The spleen has little function in blood cell production in the healthy human but plays a critical role in the clearance of defective or senescent cells, as well as in host defense. In the human fetus, hematopoiesis can be found in the liver, spleen, bone marrow, thymus, and lymph nodes. The bone marrow is the dominant hematopoietic organ in the latter half of gestation and the only blood-cell-producing organ at birth (Moore, 1975). All marrow is active, or “red marrow,” at birth (Hudson, 2006). During early childhood, hematopoiesis recedes in long bones and, in adults, is confined to the axial skeleton and proximal humerus and femur (Custer and Ahlfeldt, 1932). The marrow in the distal long bones becomes “yellow” or fatty. When demand for blood cell production is great, as with certain disease states, fatty marrow can be reactivated as sites of hematopoiesis (Fig. 11-1). This can be useful in toxicology studies as a marker of sustained hematopoietic stress, as exemplified in studies on the hematopathology of cephalosporin toxicity in the dog (Bloom et al., 1987). Under extreme conditions, embryonic patterns of hematopoiesis may reappear as extramedullary hematopoiesis (Young and Weiss, 1997).

While the central function of bone marrow is hematopoiesis and lymphopoiesis, bone marrow is also one of the sites of the mononuclear phagocyte system (MPS), contributing monocytes that differentiate into a variety of MPS cells located in liver (Kupffer cells), spleen (littoral cells), lymph nodes, and other tissues. Conventional histologic and cytologic sampling of bone marrow reveals a very limited picture of an exceedingly complex tissue containing erythroid, granulocytic, megakaryocytic, MPS, and lymphoid precursors in varied stages of maturation; stromal cells; and vasculature all encased by bone (Fig. 11-1). Routine examinations of such specimens in our pathology and toxicology laboratories cannot possibly reveal the sophisticated interactions that mediate lineage commitment, proliferation, differentiation, acquisition of functional characteristics, and trafficking that result in the delivery of mature cells to the circulation, as required in sickness and in health. Exactly how the process of hematopoietic progenitor cell differentiation and maturation and subsequent release into the peripheral circulation is so tightly regulated is not fully known. Early and elegant morphologic studies revealed a complex interplay of developing cells with stromal cells, extracellular matrix components, and cytokines that make up the hematopoietic inductive microenvironment (HIM) (Young and Weiss, 1997). More recent studies have shown that each lineage, and even stage of maturation, is supported within a specific niche that is maintained by the surrounding stromal cells (Heissig et al., 2005). An array of cytokines and chemokines directs a particular progenitor cell to the appropriate niche (Lataillade et al., 2004). Today’s powerful research tools, which include the use of knockout mice, have begun to define how hematopoietic growth factors, cytokines, and chemokines interact with the HIM and other tissues to control the production and trafficking of blood cells (Kaushansky, 2006; Laurence, 2006). These tools also include the evolving applications of systems biology, which uses a combination of mathematical, engineering, and computational tools for constructing and validating models for studying complex biologic phenomena such as hematopoieses (Whichard et al., 2010). These understandings are providing opportunities to develop promising therapies that are now presenting new pharmacologic and toxicologic challenges.

The Erythrocyte

Erythrocytes (red blood cells [RBCs]) make up 40% to 45% of the circulating blood volume and serve as the principal vehicle of transportation of oxygen from the lungs to the peripheral tissues. In addition, erythrocytes are involved in the transport of carbon dioxide from tissues to the lung, maintenance of a constant pH in blood, and regulation of blood flow to tissues (Hsia, 1998; Kim-Shapiro et al., 2005). Erythrocytes help modulate the inflammatory response through clearance of immune complexes containing complement components and through interaction with nitric oxide, a potent vasodilator (Kim-Shapiro et al., 2005; Lindorfer et al., 2001). An area of developing interest is the role of erythrocytes as a carrier and/or reservoir for drugs and toxins (Schrijvers et al., 1999). The effect of xenobiotics on erythrocytes has been extensively evaluated, because of both the ready access to the tissue and the frequency with which xenobiotics cause changes in this critical tissue.

Xenobiotics may affect the production, function, and/or survival of erythrocytes. These effects are most frequently manifest as a change in the circulating red cell mass, usually resulting in a decrease (anemia). Occasionally, agents that increase oxygen affinity lead to an increase in red cell mass (erythrocytosis), but this is distinctly less common. Shifts in plasma volume can alter the relative concentration of erythrocytes/hemoglobin and can be easily confused with true anemia or erythrocytosis.

There are two general mechanisms that lead to true anemia—either decreased production or increased erythrocyte destruction. Both mechanisms may be operative in some disorders, or a combination may arise due to the imposition of a second disorder on a compensated underlying problem. For example, patients with compensated congenital hemolytic anemias are very susceptible to additional insults that may precipitate an acute drop in a previously stable red cell mass, such as parvovirus infection-associated suppression of erythropoiesis.

Evaluation of a peripheral blood sample can provide evidence for the underlying mechanism of anemia. The usual parameters of a complete blood count (CBC)—including the RBC count, hemoglobin concentration (Hbg), and hematocrit (also referred to as packed cell volume [PCV])—can establish the presence of anemia. Two additional parameters helpful in classifying an anemia are the mean corpuscular volume (MCV) and the reticulocyte count. Increased destruction is usually accompanied by an increase in reticulocytes (young erythrocytes containing residual RNA), which are easily enumerated using appropriate stains. The introduction of automated methods has improved the precision of reticulocyte counting and introduced new parameters that aid in characterization of red cell production. With these new methods, reticulocyte counting may also be useful in conditions associated with decreased production, particularly when assessing response to therapy. Other readily performed parameters helpful in the evaluation of the human erythron include erythrocyte morphology (eg, megaloblastic changes, erythrocyte fragmentation, sickled RBCs); serum concentration of haptoglobin, lactic dehydrogenase (LD), free hemoglobin, vitamin B₁₂, folate, iron, and ferritin; direct and indirect red cell antiglobulin tests; and bone marrow morphology (Prchal, 2006; Ryan, 2006).

Alterations in Red Cell Production

Erythrocyte production is a continuous process that is dependent on frequent cell division and a high rate of hemoglobin synthesis. Human adult hemoglobin (hemoglobin A), the major constituent of the erythrocyte cytoplasm, is a tetramer composed of two α-globin and two β-globin chains, each with a heme residue located in a stereospecific pocket of the globin chain. Synthesis of hemoglobin is dependent on coordinated production of globin chains and heme moieties. Abnormalities that lead to decreased hemoglobin synthesis are relatively common (eg, iron deficiency) and are often associated with a decrease in the MCV and hypochromasia (increased central pallor of RBCs on stained blood films due to the low hemoglobin concentration).

An imbalance between α- and β-chain production is the basis of congenital thalassemia syndromes and results in decreased hemoglobin production and microcytosis (Weatherall, 2006). Xenobiotics can affect globin chain synthesis and alter the composition of hemoglobin within erythrocytes. This is perhaps best demonstrated by hydroxyurea, which has been found to increase the synthesis of γ-globin chains. The γ-globin chains are a normal constituent of hemoglobin during fetal development, replacing the β chains in the hemoglobin tetramer (hemoglobin F, α₂γ₂). Hemoglobin F has a higher affinity for oxygen than hemoglobin A and can protect against crystallization (sickling) of deoxyhemoglobin S in sickle cell disease (Steinberg, 2006).

Synthesis of heme requires incorporation of iron into a porphyrin ring (Fig. 11-2) (Napier et al., 2005; Ponka, 1997). Iron deficiency is usually the result of dietary deficiency or increased blood loss. Drugs that contribute to blood loss, such as nonsteroidal anti-inflammatory agents, with their increased risk of gastrointestinal ulceration and bleeding, may potentiate the risk of developing iron deficiency anemia. Defects in the synthesis of porphyrin ring of heme can lead to sideroblastic anemia, with its characteristic accumulation of iron in bone marrow erythroblasts. The accumulated iron precipitates within mitochondria in a complex with mitochondrial ferritin, causing the characteristic staining pattern of ringed sideroblasts evident on iron stains such as Prussian blue (Cazzola et al., 2003). A number of xenobiotics (Table 11-1) can interfere with one or more of the steps in erythroblast heme synthesis and result in sideroblastic anemia (Alcindor and Bridges, 2002; Beutler, 2006a,b,c; Fiske et al., 1994).

Hematopoiesis requires active DNA synthesis and frequent mitoses. Folate and vitamin B₁₂ are necessary to maintain synthesis of thymidine for incorporation into DNA (Fig. 11-3). Deficiency of folate and/or vitamin B₁₂ results in megaloblastic anemia, with its characteristic morphologic and biochemical changes (Table 11-2), which commonly affect erythroid, myeloid, and megakaryocytic lineages. A number of xenobiotics may contribute to a deficiency of vitamin B₁₂ and/or folate (Table 11-3), leading to megaloblastic anemia (Babior, 2006).

Many of the antiproliferative drugs used in the treatment of malignancy predictably inhibit hematopoiesis, including erythropoiesis. The resulting bone marrow toxicity may be dose-limiting, as previously discussed. Drugs, such as amifostine, have been developed that may help protect against the marrow toxicity of these agents (Phillips, 2002). The development of recombinant forms of some of the growth factors that regulate hematopoiesis has helped shorten the duration of bone marrow suppression. As with other therapeutic proteins, there is a risk of antibody formation in response to administration of these proteins; if the antibody reacts with the endogenous growth factor, it may cause profound cytopenia (Bennett et al., 2004; Li et al., 2001).

Erythropoietin is commonly used to support red cell production in patients undergoing chemotherapy and with renal failure. Following a change in formulation, a series of cases of red cell aplasia associated with erythropoietin use was reported (Bennett et al., 2004). The etiology was antibodies to the synthetic protein that cross-reacted with endogenous erythropoietin. A change in formulation in combination with the nature of the storage container and route of administration is thought to have promoted the formation of protein aggregates, a phenomenon known to be associated with an increased risk of antibody formation (Koren et al., 2002). The incidence of red cell aplasia appears to have diminished following a change in packaging and administration of erythropoietin by intravenous injection (Bennett et al., 2004).

Drug-induced aplastic anemia may represent either a predictable or idiosyncratic reaction to a xenobiotic. This life-threatening disorder is characterized by peripheral blood pancytopenia, reticulocytopenia, and bone marrow hypoplasia (Vandendries and Drews, 2006; Young, 1999, 2000). Agents such as benzene and radiation have a predictable effect on hematopoietic progenitors, and the resulting aplastic anemia corresponds to the magnitude of the exposure to these agents. In contrast, idiosyncratic aplastic anemia does not appear to be related to the dose of the agent initiating the process. A long list of chemicals has been associated with the development of aplastic anemia (Table 11-4), many of which have been reported in only a few patients. The principal mechanisms of aplastic anemia, as with the more common agranulocytosis, are still debated. Immune mechanisms have long been thought to contribute to the development of both idiosyncratic disorders, as discussed in greater detail in the next section. However, it has been, up until recently, difficult to obtain definitive evidence for humoral and/or cellular mechanisms of marrow suppression (Vandendries and Drews, 2006; Young, 2000). Recent observations have strongly suggested that both toxicities are immune-mediated, often involving altered proteins through reactive metabolite–mediated damage (Zhang et al., 2011).

Pure red cell aplasia is a syndrome in which the decrease in marrow production is limited to the erythroid lineage (Djaldetti et al., 2003; Fisch et al., 2000). It is an uncommon disorder that may be due to genetic defects, infection (parvovirus B19), immune-mediated injury, myelodysplasia, drugs, or other toxins. As pure red cell aplasia occurs sporadically and infrequently, the linkage between drug exposure and pathogenesis of the aplasia remains speculative for some agents. The drugs most clearly implicated and for which there are multiple case reports include isoniazid, phenytoin, and azathioprine. The mechanism of drug-induced pure red cell aplasia is unknown, but some evidence suggests that it may also be immune-mediated. Patients with drug-induced red cell aplasia should not be reexposed to the purported offending chemical. As noted above, pure red aplasia may also occur as a consequence of an immune response to therapeutic erythropoietin.

Alterations in the Respiratory Function of Hemoglobin

Hemoglobin is necessary for effective transport of oxygen and carbon dioxide between the lungs and tissues. The respiratory function of hemoglobin has been studied in detail, revealing an intricately balanced system for the transport of oxygen from the lungs to the tissues (Hsia, 1998). Electrostatic charges hold the globin chains of deoxyhemoglobin in a “tense” (T) conformation characterized by a relatively low affinity for oxygen. Binding of oxygen alters this conformation to a “relaxed” (R) conformation that is associated with a 500-fold increase in oxygen affinity. Thus, the individual globin units show cooperativity in the binding of oxygen, resulting in the familiar sigmoid shape to the oxygen dissociation curve (Fig. 11-4). The ability of hemoglobin to safely and efficiently transport oxygen is dependent on both intrinsic (homotropic) and extrinsic (heterotropic) factors that affect the performance of this system.

Homotropic Effects

One of the most important homotropic properties of oxyhemoglobin is the slow but consistent oxidation of heme iron to the ferric state to form methemoglobin (Percy et al., 2005). Methemoglobin is not capable of binding and transporting oxygen. In addition, the presence of methemoglobin in a hemoglobin tetramer has allosteric effects that increase the affinity of oxyhemoglobin for oxygen, resulting in a leftward shift of the oxygen dissociation curve (Fig. 11-4). The combination of decreased oxygen content and increased affinity may significantly impair delivery of oxygen to tissues when the concentration of methemoglobin rises beyond critical levels (Hsia, 1998; Percy et al., 2005).

Not surprisingly, the normal erythrocyte has metabolic mechanisms for reducing heme iron back to the ferrous state; these mechanisms are normally capable of maintaining the concentration of methemoglobin at less than 1% of total hemoglobin (Percy et al., 2005). The predominant pathway is cytochrome b₅ methemoglobin reductase, which is dependent on reduced nicotine adenine dinucleotide (NADH) and is also known as NADH diaphorase. An alternate pathway involves a reduced nicotine adenine dinucleotide phosphate (NADPH) diaphorase that reduces a flavin that in turn reduces methemoglobin. This pathway usually accounts for less than 5% of the reduction of methemoglobin, but its activity can be greatly enhanced by methylene blue, which is reduced to leukomethylene blue by NADPH diaphorase. Leukomethylene blue then reduces methemoglobin to deoxyhemoglobin.

A failure of these control mechanisms leads to increased levels of methemoglobin, or methemoglobinemia. The most common cause of methemoglobinemia is exposure to an oxidizing xenobiotic that overwhelms the NADH diaphorase system. A large number of chemicals and therapeutic agents may cause methemoglobinemia (Table 11-5) (Bradberry et al., 2001; Bradberry, 2003; Coleman and Coleman, 1996). These chemicals may be divided into direct oxidizers, which are capable of inducing methemoglobin formation when added to erythrocytes in vitro or in vivo, and indirect oxidizers, which do not induce methemoglobin formation when exposed to erythrocytes in vitro but do so after metabolic modification in vivo. Nitrites appear to be able to interact directly with heme to facilitate oxidation of heme iron, but the precise mechanism that leads to methemoglobin formation is unknown for many of the other substances listed in Table 11-5.

The development of methemoglobinemia may be slow and insidious or abrupt in onset, as with the use of some topical anesthetics (Bradberry et al., 2001; Bradberry, 2003; Khan and Kruse, 1999; Nguyen et al., 2000). Most patients tolerate low levels (<10%) of methemoglobin without clinical symptoms. Cyanosis is often evident when the methemoglobin concentration exceeds 5% to 10%. Levels above 20% are generally clinically significant and some patients may begin to manifest symptoms related to tissue hypoxemia at methemoglobin levels between 10% and 20%. The severity of clinical manifestations increases as the concentration rises above 20% to 30%, with methemoglobin levels above 70% being life-threatening. Intravenous administration of one to two mg/kg methylene blue is effective in rapidly reversing methemoglobinemia through activation of the NADPH diaphorase pathway (Clifton and Leikin, 2003). The effect of methylene blue is dependent on an adequate supply of NADPH. Consequently, methylene blue is not effective in patients with glucose-6-phosphate dehydrogenase (G-6-PD) deficiency because of the decreased capacity to form NADPH (Bradberry, 2003; Coleman and Coleman, 1996).

Heterotropic Effects

There are three major heterotropic effectors of hemoglobin function: pH, erythrocyte 2,3-bisphosphoglycerate (2,3-BPG, formerly designated 2,3-diphosphoglycerate) concentration, and temperature (Hsia, 1998). A decrease in pH (eg, lactic acid, carbon dioxide) lowers the affinity of hemoglobin for oxygen, that is, it causes a right shift in the oxygen dissociation curve, facilitating the delivery of oxygen to tissues (Fig. 11-4). As bicarbonate and carbon dioxide equilibrate in the lung, the hydrogen ion concentration decreases, increasing the affinity of hemoglobin for oxygen and facilitating oxygen uptake. Thus, the buffering capacity of hemoglobin also serves to improve oxygen uptake and delivery.

The binding site for 2,3-BPG is located in a pocket formed by the 2 β chains of a hemoglobin tetramer. Binding of 2,3-BPG to deoxyhemoglobin results in stabilization of the “T” conformation, with reduced oxygen affinity (a shift to the right of the oxygen dissociation curve). The conformational change induced by binding of oxygen alters the binding site for 2,3-BPG and results in release of 2,3-BPG from hemoglobin. This facilitates uptake of more oxygen for delivery to tissues. The concentration of 2,3-BPG increases whenever there is tissue hypoxemia but may decrease in the presence of acidosis or hypophosphatemia. Thus, hypophosphatemia may result in a left shift of the oxygen dissociation curve.

Clofibric acid and bezafibrate are capable of lowering the oxygen affinity of hemoglobin, analogous to 2,3-BPG (Poyart et al., 1994). However, the association constant of bezafibrate for hemoglobin is too low for there to be a meaningful effect in vivo. Work continues on bezafibrate derivatives that may lower oxygen affinity and enhance tissue oxygenation. In contrast, some aromatic benzaldehydes have been shown to increase oxygen affinity and shift the dissociation curve to the left. It was thought that these compounds may be useful in preventing the sickling of deoxyhemoglobin S in patients with sickle cell disease. However, these and other chemicals evaluated for their effect on hemoglobin oxygen affinity have not progressed into clinical usage (Papassotiriou et al., 1998; Poyart et al., 1994).

The oxygen affinity of hemoglobin decreases as the body temperature increases (Hsia, 1998). This facilitates delivery of oxygen to tissues during periods of extreme exercise and febrile illnesses associated with increased temperature. Correspondingly, oxygen affinity increases during hypothermia, which may lead to decreased oxygen delivery under these conditions. This must be taken into consideration during surgical procedures during which there is induction of deep hypothermia.

The respiratory function of hemoglobin may also be impaired by blockade of the ligand binding site following interaction with other substances, most notably carbon monoxide (Hsia, 1998). Carbon monoxide has a relatively low rate of association with deoxyhemoglobin but has high affinity once bound. The affinity is about 200 times that of oxygen, and thus persistent exposure to a low level of carbon monoxide (eg, 0.1%) may lead to 50% saturation of hemoglobin. Binding of carbon monoxide also results in stabilization of the hemoglobin molecule in the high-affinity “R” conformation. Consequently, the oxygen dissociation curve is shifted to the left, further compromising oxygen delivery to the tissues. Carbon monoxide is produced at low levels by the body through the metabolism of heme and equilibrates across the pulmonary capillary/alveolar bed. Low concentrations of carboxyhemoglobin can be cytoprotective during inflammatory stress or ischemia/reperfusion injury and the therapeutic use of low concentrations of carbon monoxide is being explored (Kao and Nanagas, 2005; Ryter and Otterbein, 2004).

The major sources of significant exogenous exposure to carbon monoxide are smoking and burning of fossil fuels (including automobiles), particularly in enclosed spaces. Heavy smoking during pregnancy may result in significant levels of carboxyhemoglobin in fetal blood and diminished oxygenation of fetal tissues. Symptoms of carbon monoxide toxicity, such as dizziness, shortness of breath, and headache, begin to appear when carboxyhemoglobin levels reach 20%. Levels of 50% to 80% carboxyhemoglobin may be lethal. In addition to effects due to interaction with hemoglobin, CO may cause toxicity through effects on myoglobin, cytochromes, guanylate cyclase, or nitric oxide. The key to therapy is removal from the source of carbon monoxide and provision of an adequate supply of oxygen; hyperbaric hyperoxia may be used in serious cases (Kao and Nanagas, 2005; Ryter and Otterbein, 2004).

Methemoglobin can combine reversibly with a variety of chemical substances, including cyanide, sulfides, peroxides, fluorides, and azides. The affinity of methemoglobin for cyanide is utilized in two settings. First, nitrites are administered in cyanide poisoning to form methemoglobin, which then binds free cyanide, sparing other critical cellular respiratory enzymes (Cummings, 2004). Second, formation of cyanmethemoglobin by reaction of hemoglobin with potassium ferricyanide is a standard method for measurement of hemoglobin concentration.

Nitric oxide, an important vasodilator that modulates vascular tone, binds avidly to heme iron. An additional function of erythrocytes is related to this interaction, which can influence the availability of nitric oxide in parts of the circulation (Hsia, 1998; Lundberg and Weitzberg, 2005). Solutions of hemoglobin have been evaluated as a potential replacement for RBC transfusions. However, these trials have been halted due to the toxicity associated with administration of hemoglobin solutions. Vascular instability is one of the complications associated with infusion of hemoglobin solutions and is thought to be related to the scavenging of essential nitric oxide by the administered hemoglobin (Moore et al., 2005; Rother et al., 2005).

Alterations in Erythrocyte Survival

The normal survival of erythrocytes in the circulation is about 120 days (Dessypris, 1999). During this period the erythrocytes are exposed to a variety of oxidative injuries and must negotiate the tortuous passages of the microcirculation and the spleen. This requires a deformable cell membrane and energy to maintain the sodium–potassium gradients and repair mechanisms (Van Wijk and van Solinge, 2005). Very little protein synthesis occurs during this time, as erythrocytes are anucleate when they enter the circulation and residual mRNA is rapidly lost over the first one to two days in the circulation. Consequently, senescence occurs over time until the aged erythrocytes are removed by the spleen, where the iron is recovered for reutilization in heme synthesis. Any insult that increases oxidative injury, decreases metabolism, or alters the membrane may cause a decrease in erythrocyte concentration and a corresponding anemia.

Nonimmune Hemolytic Anemia

Microangiopathic Anemias

Intravascular fragmentation of erythrocytes gives rise to the microangiopathic hemolytic anemias (Baker, 2006). The hallmark of this process is the presence of schistocytes (fragmented RBCs) in the peripheral blood. These abnormal cellular fragments are usually promptly cleared from the circulation by the spleen. Thus, their presence in peripheral blood samples indicates either an increased rate of formation or abnormal clearance function of the spleen. The formation of fibrin strands in the microcirculation is a common mechanism for RBC fragmentation. This may occur in the setting of disseminated intravascular coagulation, sepsis, the hemolytic-uremic syndrome (HUS), and thrombotic thrombocytopenic purpura (TTP). The erythrocytes are essentially sliced into fragments by the fibrin strands that extend across the vascular lumen and impede the flow of erythrocytes through the vasculature. Excessive fragmentation can also be seen in the presence of abnormal vasculature, as occurs with damaged cardiac valves, arteriovenous malformations, vasculitis, and widely metastatic carcinoma (Baker, 2006). The high shear associated with malignant hypertension may also lead to RBC fragmentation.

Other Mechanical Injuries

March hemoglobinuria is an episodic disorder characterized by destruction of RBCs during vigorous exercise or marching (Abarbanel et al., 1990; Sagov, 1970). The erythrocytes appear to be destroyed by mechanical trauma in the feet. Sufficient hemoglobin may be released to cause hemoglobinuria. The disorder should be distinguished from other causes of intermittent hemoglobinuria such as paroxysmal nocturnal hemoglobinuria. The introduction of improved footgear for athletes and soldiers has significantly decreased the incidence of this problem.

Major thermal burns are also associated with a hemolytic process. The erythrocyte membrane becomes unstable as the temperature increases. With major burns there can be significant heat-dependent lysis of erythrocytes. Small RBC fragments break off, with resealing of the cell membrane. These cell fragments usually assume a spherical shape and are not as deformable as normal erythrocytes. Consequently, these abnormal cell fragments are removed in the spleen, leading to an anemia. The burden of RBC fragments may impair the phagocytic function of the spleen, contributing to the increased susceptibility to endotoxic shock following major burns (Hatherill et al., 1986; Schneidkraut and Loegering, 1984).

Infectious Diseases

A variety of infectious diseases may be associated with significant hemolysis, by either direct effect on the erythrocyte or development of an immune-mediated hemolytic process (Berkowitz, 1991; Beutler, 2006a,b,c). The most common agents that directly cause hemolysis include malaria, babesiosis, clostridial infections, and bartonellosis. Erythrocytes are parasitized in malaria and babesiosis, leading to their destruction. Clostridial infections are associated with release of hemolytic toxins that enter the circulation and lyse erythrocytes. The hemolysis can be severe with significant hemoglobinuria, even with apparently localized infections. Bartonella bacilliformis is thought to adhere to the erythrocyte, leading to rapid removal from the circulation. The hemolysis can be severe and the mortality rate in this disorder (Oroya fever) is high.

Oxidative Hemolysis

Molecular oxygen is a reactive and potentially toxic chemical species; consequently, the normal respiratory function of erythrocytes generates oxidative stress on a continuous basis. The major mechanisms that protect against oxidative injury in erythrocytes include NADH diaphorase, superoxide dismutase, catalase, and the glutathione pathway (Coleman and Coleman, 1996; Njalsson and Norgren, 2005). As indicated previously, a small amount of methemoglobin is continuously formed during the process of loading and unloading of oxygen from hemoglobin. Formation of methemoglobin is associated with formation of superoxide free radicals, which must be detoxified to prevent oxidative injury to hemoglobin and other critical erythrocyte components. Under physiologic conditions, superoxide dismutase converts superoxide into hydrogen peroxide, which is then metabolized by catalase and glutathione peroxidase (Fig. 11-5).

A number of xenobiotics, particularly compounds containing aromatic amines, are capable of inducing oxidative injury in erythrocytes (Table 11-6) (Bradberry, 2003; Percy et al., 2005). These chemicals appear to potentiate the normal redox reactions and are capable of overwhelming the usual protective mechanisms. The interaction between these xenobiotics and hemoglobin leads to the formation of free radicals that denature critical proteins, including hemoglobin, thiol-dependent enzymes, and components of the erythrocyte membrane. In the presence of hydrogen peroxide and xenobiotics such as hydroxylamine, hydroxamic acid, and phenolic compounds, a reactive ferryl (Fe⁴⁺) hemoglobin intermediate may be formed according to the following reaction:

H₂O₂ + Hgb − Fe³⁺ → H₂O + Hgb(•+) − Fe⁴⁺ − O⁻.

In this intermediate, referred to as compound 1, tyrosine may donate the extra electron, turning it into a reactive free radical. Compound 1 may undergo further reaction with organic compounds (AH₂ in the following equations) to yield additional free radicals according to the following reactions:

Hgb(•+) − Fe⁴⁺ − O + AH₂ → Hgb − Fe⁴⁺ − O⁻ + AH• + H⁺;

Hgb − Fe⁴⁺ − O⁻ + AH₂ + H⁺ → Hgb − Fe³⁺ + AH• + H₂O.

Hemoglobin contains exposed free cysteines (β93) that are critical for the structural integrity of the molecule. Oxidation of these groups can denature hemoglobin and decrease its solubility. The oxidized, denatured hemoglobin species comprise what has been designated sulfhemoglobin. The denatured hemoglobin can form aggregates that bind to the cell membrane to form inclusions called Heinz bodies, a hallmark of oxidative injury to erythrocytes (Jandl, 1987). Heinz bodies can be visualized by use of phase contrast microscopy or supravital stains such as crystal violet. These membrane-associated inclusions impair the deformability of the erythrocyte membrane and thus impede movement of erythrocytes through the microcirculation and the spleen. Heinz bodies are effectively removed from the erythrocyte by the spleen, so they are not often observed in peripheral blood samples from patients despite ongoing oxidative injury. However, the culling of Heinz bodies can alter the morphology of the affected cells, giving rise to what are called “bite” cells and “blister” cells, which may provide an important clue as to the ongoing process (Yoo and Lessin, 1992). These cells look as though a portion of the cytoplasm had been cut away. Heinz body formation can be induced by in vitro exposure to oxidizing agents and patients with oxidative hemolysis often show increased in vitro formation of Heinz bodies.

Oxidative denaturation of the globin chain decreases its affinity for the heme group, which may dissociate from the globin chain during oxidative injury (Kumar and Bandyopadhyay, 2005). Free heme itself is toxic to cells and can induce tissue injury through formation of reactive oxygen species. The ferric iron in the heme ring may react with chloride to form a complex called hemin. Hemin is hydrophobic and intercalates into the erythrocyte membrane from which it is removed by interaction with albumin. However, if the rate of hemin formation exceeds the rate of removal by albumin, hemin accumulates in the membrane, where it can cause rapid lysis of the erythrocyte.

The generation of free radicals may also lead to peroxidation of membrane lipids (Jandl, 1987; Kumar and Bandyopadhyay, 2005). This may affect the deformability of the erythrocyte and the permeability of the membrane to potassium. The alteration of the Na⁺/K⁺ gradient is independent of injury to the Na⁺/K⁺ pump and is potentially lethal to the affected erythrocyte. Oxidative injury also impairs the metabolic machinery of the erythrocyte, resulting in a decrease in the concentration of ATP (Tavazzi et al., 2000). Damage to the membrane can also permit leakage of denatured hemoglobin from the cell. Such free denatured hemoglobin can be toxic on its own. Free hemoglobin may irreversibly bind nitric oxide, resulting in vasoconstriction. Released hemoglobin may form nephrotoxic hemoglobin dimers, leading to kidney damage.

Oxidative injury thus results in a number of changes that decrease the viability of erythrocytes. Protection against many of the free radical–induced modifications is mediated by reduced glutathione (Njalsson and Norgren, 2005). Formation of reduced glutathione is dependent on NADPH and the hexose monophosphate shunt (Fig. 11-6). Significant oxidative injury usually occurs when the concentration of the xenobiotic is high enough (due to either high exposure or decreased metabolism of the xenobiotic) to overcome the normal protective mechanisms, or, more commonly, when there is an underlying defect in the protective mechanisms.

The most common enzyme defect associated with oxidative hemolysis is G-6-PD deficiency, a relatively common sex-linked disorder characterized by alterations in the primary structure of G-6-PD that diminish its functional activity (Beutler, 1996). It is often clinically asymptomatic until the erythrocytes are exposed to oxidative stress. The stress may come from the host response to infection or exposure to xenobiotics. The level of G-6-PD normally decreases as the erythrocytes age. In the African type of G-6-PD deficiency, the enzyme is less stable than normal; thus, the loss of activity is accelerated compared with normals. In the Mediterranean type of G-6-PD deficiency, the rate of loss of enzyme activity is even higher. Consequently, the older erythrocytes with the lowest levels of G-6-PD are most susceptible to hemolysis, with the degree of hemolysis affected by the residual amount of enzyme activity as well as the magnitude of the oxidative injury.

Erythrocyte-reduced glutathione is rapidly depleted on exposure to an oxidizing agent in patients with G-6-PD deficiency. This leads to the series of oxidative injuries described above with the development of intravascular and extravascular hemolysis. Oxidative hemolysis is usually reversible if the process is promptly recognized and the offending agent is removed. Occasionally the hemolysis may be sufficiently severe to result in death or serious morbidity (eg, renal failure). Hemolytic anemia may occur in patients with deficiency of glutathione synthetase due to the reduced intracellular concentration of glutathione (Njalsson and Norgren, 2005).

Nonoxidative Chemical-Induced Hemolysis

Exposure to some xenobiotics is associated with hemolysis without significant oxidative injury (Beutler, 2006a,b,c). Arsenic hydride is a gas that is formed during several industrial processes. Inhalation of the gas can result in severe hemolysis, with anemia, jaundice, and hemoglobinuria. The mechanism of hemolysis in arsine toxicity is not understood. Lead poisoning is associated with defects in heme synthesis and a shortening of erythrocyte survival. The cause of the hemolysis is uncertain, but lead can cause membrane damage and interfere with the Na⁺/K⁺ pump. These effects may cause premature removal of erythrocytes from the circulation. Excess copper has been associated with hemolytic anemia. The pathogenesis may relate to inhibitory effects on the hexose monophosphate shunt and the Embden–Meyerhof pathway. Ingestion of excess chromium may result in hemolytic anemia and thrombocytopenia, although the mechanism is not known (Cerulli et al., 1998). Significant hemolysis may also occur with biologic toxins found in insect and snake venoms (Beutler, 2006a,b,c).

Immune Hemolytic Anemia

Immunologic destruction of erythrocytes is mediated by the interaction of IgG or IgM antibodies with antigens expressed on the surface of the erythrocyte. In the case of autoimmune hemolytic anemia the antigens are intrinsic components of the patient’s own erythrocytes. A large number of drugs have been associated with enhanced binding of immunoglobulin to the erythrocyte surface and shortened RBC survival (Arndt and Garratty, 2005).

A number of mechanisms have been implicated in xenobiotic-mediated antibody binding to erythrocytes (Arndt and Garratty, 2005). Some drugs, of which penicillin is a prototype, appear to bind to the surface of the cell, with the “foreign” drug acting as a hapten and eliciting an immune response. The antibodies that arise in this type of response only bind to drug-coated erythrocytes. Other drugs, of which quinidine is a prototype, bind to components of the erythrocyte surface and induce a conformational change in one or more components of the membrane. This type of interaction can give rise to a confusing array of antibody specificities. Some of the antibodies recognize only the drug–membrane component complex; others are specific for the membrane component, but only when drug is present; while still others may recognize the membrane component in the presence or absence of the drug. A third mechanism, for which α-methyldopa is a prototype, results in production of a drug-induced autoantibody that cannot be distinguished from the antibodies arising in idiopathic autoimmune hemolytic anemia. The mechanism for induction of this group of antibodies is not understood, but may be related to development of an autoimmune response. A variant of this type of response is the augmentation of autoimmune hemolytic anemia that may occur during therapy of some lymphoproliferative disorders. Autoimmune phenomena, including autoimmune hemolytic anemia, are known to occur in lymphoproliferative disorders such as chronic lymphocytic leukemia (CLL). Treatment of these disorders with some drugs has been associated with worsening of the hemolytic anemia (Gonzalez et al., 1998). It has been hypothesized that therapy further disrupts regulation of the autoimmune phenomenon, allowing increased antibody production.

Some xenobiotics are associated with nonspecific deposition of proteins on erythrocytes. This was first associated with cephalosporins but has also been seen with other drugs, including cisplatin and the β-lactamase inhibitors sulbactam and clavulanate (Arndt and Garratty, 2005). Immunoglobulin and complement proteins may be among the proteins deposited on the erythrocyte surface. These proteins may cause a positive direct antiglobulin test, suggesting a drug-induced antibody response. However, there is no evidence of a drug-dependent antibody in the patient’s serum, and drug-treated erythrocytes may bind antibody from normal nondrug exposed serum. This form of antibody deposition is generally not associated with hemolysis, although the possibility of hemolysis related to this type of reaction has been raised.

Drug-induced intravascular hemolysis is often a dramatic clinical event and may be associated with fever, chills, back pain, hypotension, a rapid fall in hemoglobin concentration, a decrease in serum haptoglobin, a marked increase in serum LD, and hemoglobinuria (Arndt and Garratty, 2005). The clinical picture of extravascular hemolysis depends on the rate of hemolysis but is usually less dramatic. Often there is evidence of reticulocytosis, polychromasia, spherocytosis, a moderate increase in serum LD, and an increase in serum bilirubin. Serologic studies usually show evidence of IgG and/or complement on the surface of erythrocytes, although it may be difficult to document that antibody binding is drug-dependent. The mainstay of therapy in patients with drug-induced hemolytic anemia is removal of the offending agent and avoidance of reexposure.

Components of Blood Leukocytes

The leukon consists of leukocytes, or white blood cells. They include granulocytes, which may be subdivided into neutrophils, eosinophils, and basophils; monocytes; and lymphocytes. Granulocytes and monocytes are nucleated ameboid cells that are phagocytic. They play a central role in the inflammatory response and host defense. Unlike the RBC, which resides exclusively within blood, granulocytes and monocytes generally pass through the blood on their way to the extravascular tissues, where they reside in large numbers, although it is now understood that senescent neutrophils that remain in the circulation return to the bone marrow through the stromal-derived factor-1 (SDF-1)α/CXCR4 chemokine axis (Martin et al., 2003).

Granulocytes are defined by the characteristics of their cytoplasmic granules as they appear on a blood smear stained with a polychromatic (Romanovsky) stain. Neutrophils, the largest component of blood leukocytes, are highly specialized in the mediation of inflammation and the ingestion and destruction of pathogenic microorganisms. The turnover of neutrophils is enormous and increases dramatically in times of inflammation and infection, elevating the number of these cells released from the bone marrow. Eosinophils and basophils modulate inflammation through the release of various mediators and play an important role in other homeostatic functions. All these are influenced by cellular and humoral immune responses, as discussed in greater detail in Chap. 12.

In the world of clinical and experimental toxicology, the neutrophil is the focus of concern when evaluating granulocytes as possible targets for drug and nontherapeutic chemical effects. Eosinophils and basophils are far more difficult to study, with changes in these populations most frequently associated with reactions to other target organ or systemic toxicity. Examples include the eosinophilia observed with the toxic oil syndrome that resulted from exposure to rapeseed oil denatured in aniline utilized in northwestern Spain (Kilbourne et al., 1991), and the eosinophilia–myalgia syndrome associated with λ-tryptophan preparations contaminated with 1,1-ethylidene-bis[tryptophan] (Varga et al., 1992). Peripheral eosinophilia is often but not reliably observed with hypersensitivity reactions to drugs (Roujeau, 2005), while tissue eosinophilia can be diagnostic, in the context of a suggestive clinical course, in conditions such as drug-induced cutaneous vasculitis (Bahrami et al., 2006) and eosinophilic pneumonia (Flieder and Travis, 2004). This variability in systemic response can be genetically predisposed, as demonstrated in studies using transgenic mice on genetic restrictions in people afflicted by the aforementioned toxic oil syndrome (Gallardo et al., 2005). The time course of the reaction can also influence whether eosinophilia can be demonstrated in hypersensitivity disease (Roujeau, 2005).

Evaluation of Granulocytes

The most informative test to assess the neutrophil compartment is the blood neutrophil count. Accurate interpretation requires an understanding of neutrophil kinetics and the response of this tissue to physiologic and pathologic changes. In the blood, neutrophils are distributed between circulating and marginated pools, which are of equal size in humans and in constant equilibrium (Athens et al., 1961). A blood neutrophil count assesses only the circulating pool, which remains between 1800 and 7500/μL in a healthy adult human (Dale, 2006).

This constancy is remarkable, considering that as many as 10¹¹ neutrophils are released from the marrow daily, that this circulating pool represents only 1% of the total body neutrophils (Semerad et al., 2002), and that the circulating half-life of these cells is only approximately six hours (Cartwright et al., 1964). How this extraordinary regulation is achieved is only partially understood. Studies using knockout mice suggest that G-CSF is an essential regulator of both granulopoiesis and neutrophil release from the bone marrow (Semerad et al., 2002). It is induced by the T-cell-derived cytokine IL-17 that, in turn, is controlled by IL-23 that is provided by dendritic cells and macrophages (Stark et al., 2005). The latter is downregulated by the phagocytosis of apoptotic neutrophils in the tissues, which provides an important negative feedback loop. The upregulation and downregulation of chemokine receptors further controls the release of neutrophils from the bone marrow (as discussed below) and their return following senescence (Martin et al., 2003). Pharmaceutical companies are currently developing recombinant proteins that function as agonists and inhibitors of these mediators, which have great potential as exciting new therapies. Many will also be shown to cause unacceptable immunotoxicity and hematotoxicity, which portends exciting times for the academic and industrial hematopathologist and toxicologist.

Neutrophil kinetics and response to disease will vary substantially among animal species (Feldman et al., 2006). Thus, a thorough understanding of these features in any animal model used in investigative toxicology is required before informed interpretations can be made. In humans, clinically significant neutropenia occurs when the blood neutrophil count is less than 1000/μL, but serious recurrent infections do not usually occur until counts fall below 500/μL (Dale, 2006). Morphologic assessment of peripheral blood granulocytes can be helpful in characterizing neutropenia. In humans and most healthy animal species, mature (segmented) and a few immature (band) neutrophils can be identified on blood films stained with Wright or Giemsa stain. During inflammation, a “shift to the left” may occur, which refers to an increased number of immature (nonsegmented) granulocytes in the peripheral blood, which may include bands, metamyelocytes, and occasionally myelocytes. During such times, neutrophils may also show “toxic” granulation, Döhle bodies, and cytoplasmic vacuoles. These morphologic changes may be prominent in sepsis or as a result of drug or chemical intoxication. In order to fully characterize such changes or understand the pathogenesis of the abnormality, bone marrow must be examined using marrow aspirates and biopsies. These provide information on rates of production, bone marrow reserves, and abnormalities in cell distribution, and occasionally specific clues as to etiology. In vitro stem cell assays may be used to assess the granulocyte progenitor cell compartment, which may include colony-forming-units granulocyte/monocyte (CFU-GM) performed in a semisolid medium, such as agar or methylcellulose, which contains appropriate growth factors, as discussed later in this chapter. Normal human marrow specimens contain approximately 50 to 1000 CFU-GM per 10⁶ nucleated cells cultured (Liesveld and Lichtman, 1997). Marrow stem cell reserves can be assessed in vitro after administration of G-CSF (Demirer and Bensinger, 1995), which stimulates increased production and release of neutrophil precursors. Glucocorticoids (Peters et al., 1972) and epinephrine (Babior and Golde, 1995) may also be used for this purpose but are rarely used in a clinical setting.

The recent understanding that the CXC chemokine ligand SDF-1/CXCR4 mediates the retention of granulopoietic stem cells within their bone marrow niche (Laurence, 2006; Sharma et al., 2011), as well as mature neutrophils within the bone marrow pool (Semerad et al., 2002), has led to the development of an inhibitor of this ligand that, when administered with G-CSF, can cause a transient release of neutrophils and CD34+ (stem) cells into the circulation (Liles et al., 2005). The latter can be collected and re-engrafted to form new functioning bone marrow (Broxmeyer et al., 2005). The homing and retention of such hematopoietic stem/progenitor cells is influenced by complex and highly coordinated interactions among adhesion molecules, cytokines, growth factors, and regulatory cofactors that are only partially understood (Sharma et al., 2011). The ability to manipulate this system, as in the above experiments and through the many recombinant proteins under development, will continue to provide important research, diagnostic, and therapeutic tools for the hematologist, oncologist, and toxicologist.

The degree of proliferation in the granulocyte compartment can also be assessed using older techniques that employ ³H-thymidine suicide assays or DNA-binding dyes with fluorescence-activated cell sorting analyses (Keng, 1986).

Toxic Effects on Granulocytes

The toxicologist is concerned with the effect of xenobiotics on granulocytes as relates to proliferation (granulopoiesis) and kinetics, the extent to which a drug or chemical contaminant can impair the vital functions these cells perform, and how neutrophils mediate or exacerbate inflammatory disease or other target organ toxicity. The last two areas are discussed in Chap. 12, as they relate to their role as important effector cells of the immune system. However, it is difficult to separate effects on granulopoiesis and neutrophil kinetics from that of function. Both are complex and highly regulated, as discussed above, through an array of growth factors, chemokines, cytokines, and interactions with monocytes, dendritic cells, and lymphocytes in a bidirectional, multicompartmental manner (Nathan, 2006). Such complexity is not surprising, given the daunting task these cells perform, which is elegantly described in a review by Nathan (2006): “The [neutrophil] must remain non-sticky as it hurtles through the arterial and arteriolar circulation; then it must squeeze through capillaries smaller in diameter than itself, without allowing collision, friction, or distortion to activate it. A fraction of the population must adhere tightly enough to the normal endothelium of post-capillary venules to resist being washed away in the circulation, but loosely enough to roll while scouting for evidence of tissue damage and microbial infection. If such evidence is received, the cell must crawl to a boundary between endothelial cells, penetrate the junctions and the underlying basement membrane without damaging these structures, move up the chemotactic gradient, and decide whether its original information remains valid. If the answer is negative, the cell must execute itself by apoptosis. If the answer is positive, the cell must attempt to engulf and destroy microbes. If it cannot locate microbes quickly, it must attempt to destroy them at a distance by releasing every weapon at its disposal.” Many of the mediators and interactions that enable this feat have now become targets for the therapeutic dysregulation of these processes, which has led to the development of candidate drugs that may prove to be uniquely efficacious and/or toxic, as discussed below.

Effects on Proliferation and Kinetics

As with other hematopoietic tissue, the high rate of proliferation of neutrophils makes their progenitor and precursor granulocyte pool particularly susceptible to inhibitors of mitosis. Such effects by cytotoxic drugs are generally nonspecific, as they similarly affect cells of the dermis, gastrointestinal tract, and other rapidly dividing tissues. Agents that affect both neutrophils and monocytes pose a greater risk for toxic sequelae, such as infection (Dale, 2006). Such effects tend to be dose-related, with mononuclear phagocyte recovery preceding neutrophil recovery (Arneborn and Palmblad, 1982).

Myelotoxicity in clinical medicine and preclinical safety studies today is most commonly seen with cytoreductive cancer chemotherapy agents. However, this is changing, as more cancer cell–targeted, normal-tissue-sparing anticancer agents are being developed. In fact, most published reviews today on drug-induced neutropenia and agranulocytosis do not even address that associated with such cytotoxic agents. Accordingly, the term commonly refers to idiosyncratic reactions, most often secondary to accelerated immune-mediated destruction of neutrophils and their progenitors, as discussed below. The toxicity associated with cytotoxic drugs, however, remains important in that it is often dose-limiting (even with some of the newer drugs) with serious manifestations that include febrile neutropenia associated with life-threatening infections (Kuderer et al., 2002). These drugs vary in terms of their mechanism, the kinetics of the cytopenias they induce, and how individual patients or animals respond. Most act to inhibit DNA synthesis or directly attack its integrity through the formation of DNA adducts or enzyme-mediated breaks (Chabner et al., 2006). While cytoreductive drugs such as alkylating agents, cisplatin, and nitrosureas can be toxic to both resting and actively dividing cells, nonproliferating cells such as metamyelocytes, bands, and mature neutrophils are relatively resistant (Friberg and Karlsson, 2003). Generally, stem cells cycle slowly and are therefore minimally affected by a single administration of a cytotoxic drug such as 5-fluorouricil; however, such exposure can stimulate cycling activity, making these cells more vulnerable to doses administered three to five days later (Harrison and Lerner, 1991). Cytokines have long been thought to enhance these effects by driving cells into the S phase (Smith et al., 1994). Sustained exposure to drugs affecting slowly cycling stem cells is believed to cause more prolonged myelosuppression, similar to that observed with idiosyncratic toxic neutropenia (Tannock, 1986). Recent studies suggest that these features are shared by cancer stem cells, which is the basis for the more sophisticated cancer interventions and regimens under development today (Eramo et al., 2008; Vermeulen et al., 2010; Ricci-Vitiani et al., 2010).

Finally, there is considerable variation among individuals as regards susceptibility to bone marrow toxicity; this can relate to how the drug is metabolized as with 5-fluorouracil (Sundman-Engberg et al., 1998) and 6-mercaptopurine (Chabner et al., 2006), gender (Huang et al., 2007; Gamazon et al., 2011), and other factors. Based on these and other data, including plasma drug concentrations, “semi-mechanistic” pharmacokinetic/pharmacodynamic models of myelosuppression have been developed to tailor doses and treatment regimens to individual patients (Friberg and Karlsson, 2003). Pharmacokinetic monitoring is now routinely performed with some anticancer treatment regimens, particularly high-dose methotrexate (Chabner et al., 2006).

Two innovations have had a dramatic impact on cancer chemotherapy and the dose-limiting myelotoxicity associated with these drugs: (1) the aforementioned development of drugs with cancer cell–specific molecular targets that are relatively bone marrow sparing, such as those that target aberrant growth factor receptor signaling, apoptosis, angiogenesis, and other metabolic, immune, inflammatory, and mutation-promoting pathways that selectively advantage tumor cells (Hanahan and Weinberg, 2011); and (2) the use of hematopoietic growth factors, the cotreatment with which mitigates or successfully rescues patients from the effects of myelosuppression (Andres et al., 2010). Most notable of the latter has been the development and application of the G-CSFs filgrastim and the longer acting pegfilgrastim, the action of which was discussed previously. Treatment with these recombinant proteins can substantially reduce the incidence, severity, and duration of neutropenia and its complications (Rader, 2006). Antagonists to the aforementioned chemokine ligand CXCR4 are also under development for treatment alone or in combination with G-CSF, which have been shown to be effective in mobilizing both stem cells and mature neutrophils (De Clercq, 2005; Larochelle et al., 2006). Cytokine-induced differentiation therapy of leukemias is another exciting treatment modality (Leung et al., 2005). The prospect of exaggerated pharmacology and off-target effects of these sophisticated interventions should provide the preclinical toxicologist and oncologist with interesting hematotoxicologic challenges.

Lindane, an insecticide used to treat seeds and soil, has been associated with leukopenia (Parent-Massin et al., 1994). It is cytotoxic for human CFU-GMs at concentrations observed in blood and adipose tissue from exposed human subjects. An example of chemicals affecting mature cells is methyl methacrylate monomer, which has been used in orthopedic surgical procedures and is cytotoxic to both neutrophils and monocytes at clinically relevant concentrations (Dahl et al., 1994).

Chemicals that affect granulocyte kinetics can cause neutropenia or neutrophilia that has variable toxicologic significance. The effects of epinephrine and glucocorticoids on granulocyte kinetics were discussed previously. Dexamethasone has long been known to cause neutrophilia through enhanced release of mature neutrophils from the bone marrow and demargination, with the latter being the largest contributor to the expanded circulating pool (Nakagawa et al., 1998). It is now clear that the reduced margination of neutrophils is mediated by multiple effects, including altered chemotaxis, expression of adhesion molecules, and the release of mediators from other cells (Barnes, 2006; Caramori and Adcock, 2005). It is widely assumed that inhibition of margination and homing are among the important mechanisms of the anti-inflammatory and immunosuppressive effects of these widely used drugs. The development of drugs with more selective effects on neutrophils is among the most active areas of investigative pharmacology and toxicology today.

Inhibitors of the CXC chemokine ligands such as CXCR4 are under development as anticancer drugs and often associated with transient neutropenia (unpublished data). The commonly observed (yet remarkable) late-onset neutropenia (LAN) following treatment with rituximab, an anti-CD20 monoclonal antibody used to treat CD20-positive lymphoma and rheumatoid arthritis, is a good example of a common drug-induced neutropenia—in this case, typically self-limiting and thought to be of limited clinical significance. Informed hypotheses on the mechanism of LAN include perturbations of granulocyte homeostasis, mediated by the complex interaction between B-cell recovery and SDF-1 (Grant et al., 2011) and polymorphisms in immunoglobulin Fc receptor genotypes, which enhance ADCC by rituximab, resulting in increased killing of malignant and normal B cells. The latter could lead to the release of lysosomal enzymes and the destruction of neutrophils as “bystanders” (Hincks et al., 2011). Consistent with these hypotheses is the recent observation of LAN being a good prognostic indicator in lymphoproliferative disorders (Hincks et al., 2011).

Effects on Function

While there are a variety of disorders associated with defects in the parameters of neutrophil function discussed above, demonstrable in vivo effects associated with drugs and nontherapeutic chemicals are surprisingly few (Borregaard and Boxer, 2006). Examples include ethanol and glucocorticoids, which impair phagocytosis and microbe ingestion in vitro and in vivo (Brayton et al., 1970). Iohexol and ioxaglate, components of radiographic contrast media, have also been reported to inhibit phagocytosis (Lillevang et al., 1994). Superoxide production, required for microbial killing and chemotaxis, has been reported to be reduced in patients using parenteral heroin as well as in former opiate abusers on long-term methadone maintenance (Mazzone et al., 1994).

In addition to glucocorticoids, several drugs and nontherapeutic chemicals have been shown to inhibit neutrophil chemotaxis. Examples include macrolide antibiotics, which suppress the expression of the adhesion molecule ICAM (Tamaoki, 2004); zinc salts, which are found in antiacne preparations (Dreno et al., 1992); the insecticide chlordane (Miyagi et al., 1998); and mercuric chloride/methylmercuric chloride (Contrino et al., 1988). More common is the activation of neutrophils with the potential for proinflammatory consequences, specifically through increased phagocytosis, O₂^- production, or both. Examples include the environmental contaminants sodium sulfite, mercuric chloride, chlordane, and toxaphene (Girard, 2003). This toxicologic potential of xenobiotics will be discussed in more detail in Chap. 12.

Idiosyncratic Toxic Neutropenia

Of greater concern are chemicals that unexpectedly damage neutrophils and granulocyte precursors—particularly to the extent of inducing agranulocytosis, which is characterized by a profound depletion in blood neutrophils to less than 500/μL (Pisciotta, 1973). The term was first used by Schultz in 1922 in patients with severe sore throat associated with a marked reduction of granulocytes, followed by sepsis and death (Schultz, 1922). Drug association was first demonstrated through rechallenge experiments performed on two patients with aminopyrine-induced agranulocytosis, who developed leucopenia within two hours of reexposure (Madison and Squier, 1934). A later study demonstrated that blood transferred from an agranulocytosis patient to normal controls resulted in a rapid drop in neutrophil count, suggesting a role for a preformed blood factor(s) such as antibodies (Moeschlin and Wagner, 1952). Such toxicity occurs in specifically conditioned individuals, and is therefore termed “idiosyncratic.” While rare, the seriousness of this disorder, as with aplastic anemia, makes it among the most important hematotoxicities, as regards regulatory scrutiny and risk:benefit considerations, which justifies the detailed discussion below. Acute agranulocytosis is attributed to drugs in >70% of cases (Garb, 2007).

Mechanisms of idiosyncratic damage often do not relate to pharmacologic properties of the parent drug, which makes managing this risk a particular challenge to hematologists and toxicologists, as discussed later in this chapter. Preclinical toxicology is rarely predictive of these effects, which are generally detected and characterized following exposure of a large population to the drug (Dieckhaus et al., 2002; Szarfman et al., 2002). Idiosyncratic drug-induced neutropenia may be dose-related and involve a nonselective disruption of protein synthesis or cell replication resulting in agranulocytosis, as discussed below. Alternatively, and more commonly, it may not be dose-dependent (as qualified below), in which case it is usually thought to be allergic or immunologic in origin. The latter has been observed with many drugs, and is more frequently observed in women, older patients, and patients with a history of allergies (Dale, 2006).

Idiosyncratic xenobiotic-induced agranulocytosis may involve a sudden depletion of circulating neutrophils concomitant with exposure, which may persist as long as the chemical or its metabolites persist in the circulation. Hematopoietic function is usually restored when the agent is detoxified or excreted. Suppression of granulopoiesis, however, is more prevalent than peripheral lysis of neutrophils and is asymptomatic unless sepsis supervenes (Pisciotta, 1973). The onset of leukopenia in the former is more gradual, but may be precipitous if lysis of circulating neutrophils also occurs. The pattern of the disease varies with the stage of granulopoiesis affected, which has been well defined for several drugs that cause bone marrow toxicity (Table 11-7). Toxicants affecting uncommitted stem cells induce total marrow failure, as seen in aplastic anemia, which generally carries a worse prognosis than chemicals affecting more differentiated precursors, such as CFU-G. It is thought that, in the latter case, surviving uncommitted stem cells eventually produce recovery, provided that the risk of infection is successfully managed during the leukopenic episodes (Pisciotta, 1973).

The incidence of drug-induced idiosyncratic agranulocytosis ranges from 2 to 15 cases per million patients exposed to drugs per year (Andres et al., 2002). While all drugs may be causative, the most commonly incriminated drugs include antithyroid agents and antibiotics, particularly sulfonamides (Andres et al., 2002; Berliner et al., 2004). An extensive case–control study on drug-induced agranulocytosis in Barcelona, Spain, followed 177 community cases (representing 78.73 million person-years), which were compared with 586 sex-, age-, and hospital-matched control subjects previously treated with the same putative drugs (Ibanez et al., 2005). The annual incidence was 3.46:1 million, which increased with age. The fatality and mortality rates were 7.0% and 0.24:1 million, respectively. The drugs most frequently implicated (in decreasing order of odds ratio) were ticlopidine hydrochloride, calcium dobesilate, antithyroid drugs, dipyrone, and spironolactone. Other drugs associated with significant risk were pyrithyldione, cinepazide, aprindine hydrochloride, carbamazepine, sulfonamides, phenytoin and phenytoin sodium, β-lactam antibiotics, erythromycin stearate and erythromycin ethylsuccinate, and diclofenac sodium. These drugs accounted for 68.6% of all cases. Some drugs commonly implicated in the past, such as phenylbutazone, chloramphenicol, and ticlopidine, are used less commonly today due to this and other toxicity. Curiously, the incidence of this idiosyncratic reaction has not changed in the western hemisphere over the past 30 years, despite this evolution of putative drugs, suggesting that host factors are critical in the pathogenesis of the toxicity (Tesfa et al., 2009).

The severity of the neutropenia often causes severe sepsis or localized infections, such as sore throat, pneumonia, or various cutaneous infections. Prior to the use of hematopoietic growth factors, the mortality was 10% to 20% (Julia et al., 1991); with appropriate management it is now less than 5% (Andres et al., 2010).

Clozapine-induced agranulocytosis is unique, as a genetic predisposition has been established (Turbay et al., 1997; Yunis et al., 1995). Prior to an aggressive risk management program that included careful screening of prospective patients and early detection through hematologic monitoring, the incidence of agranulocytosis with this highly efficacious atypical antipsychotic was as high as 1% to 2%. This measure reduced the incidence to 0.38% and the mortality by 90% (Honigfeld et al., 1998).

Mechanisms of Toxic Neutropenia

Because cases of drug-induced neutropenia are relatively rare, sporadic, or transient, studies on the pathogenesis of this hematotoxicity have been limited. Toxic neutropenia has historically been classified according to mechanism as immune-mediated or nonimmune-mediated. There has long been a debate in the literature as to whether the principal mechanism of idiosyncratic drug-induced neutropenia (including agranulocytosis) is immune-mediated, or subsequent to the generation of toxic metabolites—both involving “preconditioning” through genetically determined immune responses to, or metabolism of, the putative drug, respectively. Both are consistent with the aforementioned early observations of Madison and Squier (1934) where transfusion of blood from affected patients to normal subjects induced neutropenia, presumably involving a humoral factor—presumably an antibody. There is now consensus that the mechanism of most drug-induced idiosyncratic diseases, including hepatotoxicity, Stevens–Johnson syndrome, agranulocytosis, and aplastic anemia, usually is immune-mediated, often involving altered proteins through reactive metabolite–mediated damage (Zhang et al., 2011). The Th-17 T lymphocyte is emerging as an important mediator of this through a cellular immune mechanism demonstrated in aplastic anemia (de Latour et al., 2010; Platanias, 2010). Two hypotheses as to the mechanism (or alternative pathogeneses) for these idiosyncratic reactions that have emerged based on observations over the past 10 years include the hapten hypothesis and the danger hypothesis (Zhang et al., 2011). The former involves a reactive metabolite binding to a protein making it “foreign,” which in turn induces an immune response that leads to the toxicity. The latter has a reactive metabolite damaging a cell, which elicits an immune response against the drug or an autoimmune response. The “perfect storm” in the rare individual in which these reactions occur is thought to be caused by preconditioned (or individual-specific) circumstances that drive both the metabolism of the drug and the immune reactions to the altered proteins. Consistent with these hypotheses is the fact that aplastic anemia and the more common agranulocytosis can be induced by many of the same drugs, most of which can be oxidized to reactive metabolites by the myeloperoxidase system of neutrophils, macrophages, and/or their precursors (Uetrecht, 1990).

The incidence of xenobiotic-induced immune neutropenia is considerably less than that of immune hemolytic anemias (Vandendries and Drews, 2006). In immune-mediated neutropenia, antigen–antibody reactions lead to destruction of peripheral neutrophils, granulocyte precursors, or both. As with RBCs, an immunogenic xenobiotic can act as a hapten, where the chemical must be physically present to cause cell damage, or may induce immunogenic cells to produce antineutrophil antibodies that do not require the drug to be present (Salama et al., 1989). Also like immune hemolytic anemia, drug-induced autoimmune neutropenia has been observed (Capsoni et al., 2005). Examples of drugs that have been implicated include fludarabine (Stern et al., 1999), propylthiouracil (Sato et al., 1985), and rituximab (Voog et al., 2003). Xenobiotic-induced immune-mediated damage may also be cell-mediated (Pisciotta, 1973).

Detection of xenobiotic-induced neutrophil antibodies is also considerably more difficult than that of RBCs or platelets. This is because the neutrophil is relatively fragile and short-lived, and becomes easily activated (Palmblad et al., 2001). There is no good test for direct antigranulocyte antibodies comparable to the Coombs (antiglobulin) test. Several assays have been used, which can be grouped into four categories: those measuring end points of leukoagglutination, cytotoxic inhibition of neutrophil function, immunoglobulin binding, and those using cell-mediated mechanisms. Among the specific challenges these assays pose are the tendency of neutrophils to stick to each other in vitro, attract immunoglobulin nonspecifically to their surface, and reflect membrane damage through indirect and semiquantitative changes (Pisciotta, 1973). Most neutrophil-associated antibody assays applied clinically today employ flow cytometry, with the indirect granulocyte immunofluorescence test being the most promising (Sella et al., 2010). The reader is referred elsewhere for a more detailed discussion of assays for immune-mediated neutrophil damage (Hagen et al., 1993; Nifli et al., 2006; Palmblad et al., 2001; Sella et al., 2010).

Some nonimmune-mediated toxic neutropenias have long been known to have a genetic predisposition (Pisciotta, 1973). Direct damage may cause inhibition of granulopoiesis or neutrophil function. It may entail failure to detoxify or excrete a xenobiotic or its metabolites, which subsequently build up to toxic proportions (Gerson and Meltzer, 1992; Gerson et al., 1983; Uetrecht, 1990). Some studies suggest that a buildup of toxic oxidants generated by leukocytes can result in neutrophil damage, as with the reactive intermediates derived from the interaction between clozapine and neutrophils. The resulting superoxide and hypochlorous acid production by the myeloperoxidase system are thought to contribute to clozapine-induced neutropenia (Uetrecht, 1990). Accumulation of nitrenium ion, a metabolite of clozapine that causes a depletion of ATP and reduced glutathione, rendering the neutrophil highly susceptible to oxidant-induced apoptosis, is now thought to be the principal mechanism of this disorder (Williams et al., 2000).

Examples of agents associated with immune and nonimmune neutropenia/agranulocytosis are listed in Table 11-8.

Human Leukemias

Leukemias are proliferative disorders of hematopoietic tissue that are monoclonal in origin and thus originate from individual bone marrow cells. Historically they have been classified as myeloid or lymphoid, referring to the major lineages for erythrocytes/granulocytes/thrombocytes or lymphocytes, respectively. Because the degree of leukemic cell differentiation has also loosely correlated with the rate of disease progression, poorly differentiated phenotypes have been designated as “acute,” whereas well-differentiated ones are referred to as “chronic” leukemias. The classification of human leukemias proposed by the French–American–British (FAB) Cooperative Group, based on the above and other morphologic features (Bennett and Sanger, 1982; Bennett et al., 1985; Levine and Bloomfield, 1992), has largely been replaced by the WHO classification. The WHO classification of acute myelogenous leukemia (AML) incorporates some of the traditional features associated with AML subtyping while placing increasing significance on molecular features, particularly recurrent cytogenetic abnormalities such as t(8;21), inv(16), t(15;17), and 11q23. The revised WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues provides the most up-to-date diagnostic framework for classifying CLL, chronic myelogenous leukemia (CML), acute lymphoblastic leukemia (ALL), AML, myeloproliferative neoplasms (MPN), and the myelodysplastic syndromes (MDS), along with various subtypes of these disorders (Swerdlow et al., 2008). The result of the WHO classification is that hematologic diseases that are biologically homogenous are subtyped, leading to greater prognostic and therapeutic relevance (Altucci et al., 2005). These early correlations imply that the biology and clinical features of these proliferative disorders relate to the stage of differentiation of the target cell, which is now being linked to individual gene alterations, as well as epigenetic factors such as cytokine stimulation (Altucci et al., 2005; Look, 2005).

There is considerable evidence supporting the notion that leukemogenesis is a multievent progression (Look, 2005; Pedersen-Bjergaard et al., 1995; Varmus and Weinberg, 1993; Vogelstein et al., 1988; Williams and Whitaker, 1990). These studies suggest that factors involved in the regulation of hematopoiesis also influence neoplastic transformation. Such factors include cellular growth factors (cytokines), proto-oncogenes, and other growth-promoting genes, as well as additional genetic and epigenetic factors that govern survival, proliferation, and differentiation.

Secondary leukemia is a term used to describe patients with AML or MDS who have a history of environmental, occupational, or therapeutic exposure to hematotoxins or radiation. It also includes patients with AML evolving from antecedent myelodysplastic or other myeloid stem cell disorders. Therapy-related AML and MDS is a term applied to the former group; both are used to distinguish from features of AML that arise de novo. Various cytogenetic findings have been associated with prognosis and response to therapy. It has been suggested that secondary leukemias be redefined as any leukemia with a specific cytogenic or molecular poor prognostic feature due to a presumed predisposing factor (Rowe, 2002). While it is generally accepted that the incidence of secondary AML is increasing, the true incidence remains unknown (Preiss et al., 2010).

Mechanisms of Toxic Leukemogenesis

The understanding that certain chemicals and radiation can dysregulate hematopoiesis, resulting in leukemogenesis, is a relatively recent one. While suggested by Hunter as early as 1939, following his observations on benzene exposure and AML (Hunter, 1939), it was not until the introduction of radiation and chemotherapy as treatments for neoplasia that these agents became associated with blood dyscrasias that included (or led to) AML (Andersen et al., 1981; Casciato and Scott, 1979; Foucar et al., 1979). The notion emerged that myelotoxic agents, under certain circumstances, can be leukemogenic.

Curiously, AML is the dominant leukemia associated with drug or chemical exposure, followed by MDS (Andersen et al., 1981; Casciato and Scott, 1979; Irons, 1997). The evidence that this represents a continuum of one toxic response is compelling (Irons, 1997; Look, 2005). This has also been linked to cytogenetic abnormalities, particularly nonrandom deletions that include 7q^-, 5q^-, 20q^-, 6q^-, 11q^-, and 13q^-, which activate tumor suppressor genes required for normal myeloid cell development (Bench et al., 1998), as well as translocations implicating the mixed lineage leukemia (MLL) gene located in the 11q23 region. Monosomal karyotype, including that for 5 and 7, together with the latter translocations, is associated with a highly adverse prognosis in secondary AML like those following treatment of acute promyelocytic leukemia (Sanz and Montesinos, 2011). The frequency of at least one of the aforementioned deletions in patients who develop MDS and/or AML after treatment with alkylating or other antineoplastic chemicals has historically run from 67% to 95%, depending on the study (Bitter et al., 1987; Johansson et al., 1991; Le Beau et al., 1986; Pedersen-Bjergaard et al., 1984; Rowley et al., 1981). The cytogenetic classification known as complex karyotypes, or complex chromosomal aberrations (CCAs, defined as more than three aberrations), has been shown to occur in up to 50% of therapy-related AML and MDS cases (Mauritzson et al., 2000; Rossi et al., 2000). CAAs are also associated with the most unfavorable prognosis among the subtypes of MDS and AML (Look, 2005). These observations, together with the understandings on the pathogenesis of leukemia previously discussed, have further corroborated the model proposed by Irons 15 years ago for the evolution of toxic leukemogenesis, which is illustrated in Fig. 11-7 (Irons, 1997).

The relatively low frequency of deletions in chromosomes 5 and 7 thought to occur in de novo, as compared with secondary AML, has led to using these cytogenetic markers to discriminate between toxic exposures and other etiologies of this leukemia. However, a recent population-based cytogenetic study of adult secondary AML and age- and sex-matched de novo AML patients showed that, with the exception of monosomy 7, the distributions of chromosomal abnormalities among the groups were comparable (Preiss et al., 2010). The same study found a higher degree of genetic instability in the therapy-related AML patients, compared with secondary AML that evolved from antecedent stem cell disorders such as MDS, which was characterized by a more frequent combination of numerical and structural aberrations, complex karyotypes, abnormal mitoses only, and hypodiploidy.

Larson and LeBeau reported the results of cytogenetic analyses on 306 consecutive patients with therapy-related AML, which were largely consistent with the above understandings (Larson and Le Beau, 2005; Look, 2005). They went on, however, to perform gene expression profiling of CD34+ hematopoietic progenitor cells from these patients, through which they were able to identify distinct subtypes of AML with characteristic gene expressions. Not surprisingly, these early progenitor cells showed expression patterns typical of arrested differentiation. Through this work they positioned therapy-related AML as an important model that provides a unique opportunity to examine the effects of mutagens on carcinogenesis in humans, as well as the role of genetic susceptibility. The fact that therapy-related AML behaves differently than the de novo disease with the same chromosomal abnormalities suggests that there is indeed a fundamental difference on the genetic level. Tools such as next-generation transcriptome sequencing are now being used to identify perturbations in genetic and molecular pathways that are unique to therapy-related leukemias, which will have implications for interventions with the new sophisticated targeted agents under development today (Stoddart et al., 2011).

Some of the same chromosomal deletion changes described with therapy-related AML have been observed in AML patients occupationally exposed to benzene (Bitter et al., 1987; Cuneo et al., 1992; Fagioli et al., 1992; Golomb et al., 1982; Mitelman et al., 1978, 1981), who also show aneuploidy with a high frequency of involvement of chromosome 7 (Irons, 1997). However, more recent evidence suggests that chromosomes 5 and 7 are not preferential targets in benzene-related cases, which is consistent with the fact that benzene metabolites are actually very weak alkylating agents (Pyatt and Thirman, 2011). Molecular mechanisms for benzene-induced hematotoxicity, which include both leukemogenesis and aplastic anemia (Hoffmann et al., 2001), are thought to include the bioactivation of reactive metabolites (Powley and Carlson, 2001), the formation of reactive oxygen species (Kolachana et al., 1993), and activation of the aryl hydrocarbon receptor (AhR) and the oncogene C-Myb (Wan et al., 2005). Studies have also suggested that polymorphisms in genes involved in DNA double strand break repair might influence susceptibility to these effects (Shen et al., 2006).

Other forms of leukemia—including CML, CLL, ALL, and multiple myeloma—have shown weak correlations with occupational exposure or treatment with alkylating agents (Irons, 1997). The latter has been repeatedly associated with exposure to benzene, although a causal relationship has yet to be demonstrated (Bergsagel et al., 1999).

Leukemogenic Agents

The overwhelming majority of toxic or secondary leukemias seen today are therapy-related (Godley and Larson, 2001), due, in part, to stricter workplace and environmental regulations that have limited exposure to potential carcinogens. Putative agents can be classified by mechanism, as discussed previously, and include alkylating agents, topoisomerase II inhibitors, antimetabolite, and antitubulin drugs. The clinical syndrome is a late complication of these cytotoxic therapies, with a latency period between primary diagnosis and treatment-related disease ranging from months to years. While the morphologic abnormalities may vary (Godley and Larson, 2001; Jaffe, 2001; Thirman and Larson, 1996), there is a continuum in the percentage of marrow blasts from a MDS to overt AML (Larson and Le Beau, 2005).

Most alkylating agents used in cancer chemotherapy can cause MDS and/or AML, including cyclophosphamide, melphalan, busulfan, chlorambucil, and nitrosurea compounds such as carmustine, or BCNU (Casciato and Scott, 1979; Greene et al., 1986). Other oncolytic agents implicated include azathioprine, procarbazine, doxorubicin, and bleomycin (Carver et al., 1979; Valagussa et al., 1979; Vismans et al., 1980). The risk these drugs pose varies considerably with the therapeutic regimen. The incidence of MDS/AML in patients treated with alkylating agents has been reported to be 0.6% to 17%, with an average of 100-fold relative risk. Moreover, treatment-related MDS is associated with a substantially higher rate of transformation to AML than is primary or spontaneous MDS (Bitter et al., 1987; Kantarjian et al., 1986).

Treatment with topoisomerase II inhibitors, particularly the epipodophyllotoxins etoposide and teniposide, can induce AML, the clinical course of which has the following distinguishing characteristics: (1) the absence of a preleukemic phase, (2) a short latency period, (3) frequent involvement of an M4/5 subtype, and (4) balanced chromosome aberrations involving chromosomes 11q23 and 21q22, among others (Larson and Le Beau, 2005; Murphy, 1993). Similar cytogenetic features have been observed following treatment with doxorubicin or dactinomycin (intercalating topoisomerase II inhibitors) in conjunction with alkylating agents and irradiation (Sandoval et al., 1993).

Of the aromatic hydrocarbons, only benzene has been proven to be leukemogenic. While substituted aromatic hydrocarbons have long been suspected to be causative, due to the fact that preparations of xylene and toluene in the past contained as much as 20% benzene (Browning, 1965), clinical or experimental evidence for that is lacking (Irons, 1997).

Exposure to high-dose γ- or x-ray radiation has long been associated with ALL, AML, and CML, as demonstrated in survivors of the atom bombings of Nagasaki and Hiroshima (Cartwright et al., 1964; Shimizu et al., 1989). Less clear is the association of these diseases with low-dose radiation secondary to fallout or diagnostic radiographs (Cartwright et al., 1964).

Other controversial agents include 1,3-butadiene, nonionizing radiation (electromagnetic, microwave, infrared, visible, and the high end of the ultraviolet spectrum), and cigarette smoking, for which published studies on the relationship to leukemia incidence is confusing, contradictory, or difficult to interpret based on dose response (Irons, 1997). Data suggesting that formaldehyde may be leukemogenic in humans have also been recently reviewed (Golden et al., 2006). The results were similarly inconclusive.

Hemostasis is a multicomponent system responsible for preventing the loss of blood from sites of vascular injury and maintaining circulating blood in a fluid state. Loss of blood is prevented by formation of stable hemostatic plugs mediated by the procoagulant arm of hemostasis. This procoagulant response is normally limited to sites of vascular injury by the multicomponent regulatory arm of hemostasis. The dynamically modulated balance between procoagulant and regulatory pathways permits a rapid, localized response to injury. The major constituents of the hemostatic system include circulating platelets, a variety of plasma proteins, and vascular endothelial cells. The role of other cells in hemostasis, especially leukocytes, has been documented (Lane et al., 2005; Monroe and Hoffman, 2005). Alterations in these components or systemic activation of this system can lead to the clinical manifestations of deranged hemostasis, including excessive bleeding and thrombosis. The hemostatic system is a frequent target of therapeutic intervention as well as inadvertent expression of the toxic effect of a variety of xenobiotics. This section briefly reviews the inadvertent effects of xenobiotics on hemostasis and the toxic effects of chemicals used to manipulate the hemostatic system.

Toxic Effects on Platelets

The Thrombocyte

Platelets are essential for formation of a stable hemostatic plug in response to vascular injury. They initially adhere to the damaged wall through binding of von Willebrand factor (vWF) with the platelet glycoprotein Ib/IX/V (GP Ib/IX/V) receptor complex (Jurk and Kehrel, 2005). Ligand binding to GP Ib/IX/V or interaction of other platelet agonists (eg, thrombin, collagen, ADP, thromboxane A₂) with their specific receptors initiates biochemical response pathways that lead to shape change, platelet contraction, platelet secretion of granule contents, activation of the GP IIb/IIIa receptor, and externalization of phosphatidylserine (Jurk and Kehrel, 2005). Activation of the GP IIb/IIIa receptor permits fibrinogen and other multivalent adhesive molecules to form cross-links between nearby platelets, resulting in platelet aggregation. Xenobiotics may interfere with the platelet response by causing thrombocytopenia or interfering with platelet function; some agents are capable of affecting both platelet number and function.

Thrombocytopenia

Like anemia, thrombocytopenia may be due to decreased production or increased destruction. Thrombocytopenia is a common side effect of intensive chemotherapy, due to the predictable effect of antiproliferative agents on hematopoietic precursors, including those of the megakaryocytic lineage. It is a clinically significant component of idiosyncratic xenobiotic-induced aplastic anemia, as discussed previously. Indeed, the initial manifestation of aplastic anemia may be mucocutaneous bleeding secondary to thrombocytopenia.

Exposure to xenobiotics may cause increased immune-mediated platelet destruction through any one of several mechanisms (Table 11-9) (Aster, 2005; Aster and Bougie, 2007; Aster et al., 2009; van den Bemt et al., 2004). Some drugs function as haptens, binding to platelet membrane components and eliciting an immune response that is specific for the hapten. The responding antibody then binds to the hapten on the platelet surface, leading to removal of the antibody-coated platelet from the circulation. This type of antibody interaction can often be blocked in vitro by excess soluble drug that binds to the antibody and prevents its interaction with the platelet surface (van den Bemt et al., 2004).

A second mechanism of immune thrombocytopenia is initiated by xenobiotic-induced exposure of a neoepitope on a platelet membrane glycoprotein. This elicits an antibody response, with the responding antibody binding to this altered platelet antigen in the presence of drug, resulting in removal of the platelet from the circulation by the mononuclear phagocytic system. The epitope specificity can be quite selective, as there is often little or no cross-reactivity between drugs having a very similar structure (eg, quinine and quinidine). This type of interaction is not inhibited in vitro by excess soluble drug, as the antibody target is a platelet-dependent epitope. Quinidine is a prototype of this type of mechanism and can induce antibodies directed at GP Ib/IX/V, GP IIb/IIIa, and/or platelet endothelial cell adhesion molecule-1 (PECAM-1) (van den Bemt et al., 2004). Epitope heterogeneity has been described, regarding the binding site of these antibodies (Peterson et al., 2008), which has been hypothesized to be due to the drug binding to and stabilizing denatured or nonnative conformations in these platelet receptors, which are then recognized as foreign and initiate a specific immune response (Li, 2010).

The diagnosis of drug-dependent antiplatelet antibodies can be quite difficult. Despite availability limited to specialized centers, flow cytometry is the most sensitive method for detecting drug-dependent antiplatelet antibodies. However, even this methodology has limitations, and must be interpreted with the clinical information. Therefore, these assays are not used in routine clinical practice. Consequently, the diagnosis is usually established by observing the resolution of thrombocytopenia following discontinuation of the offending drug. In most cases, the platelet count returns to normal within 5 to 10 days of drug discontinuation. Although a large number of chemicals have been implicated in the development of immune thrombocytopenia, the supporting evidence in many cases is weak (van den Bemt et al., 2004).

A recent survey was conducted using three methods for identifying drugs that are associated with immune thrombocytopenia: published case reports; analysis of serum samples from affected patients for drug-dependent, platelet-reactive antibodies; and the Food and Drug Administration’s Adverse Event Reporting System (Reese et al., 2010). An astonishing 1444 drugs were found to have at least one report associated with thrombocytopenia, of which 573 demonstrated a statistically distinctive reporting association. Of 1468 drugs suspected as causal, 102 were evaluated by all three methods, of which 23 were found to have an association in all three.

Thrombocytopenia is an uncommon but serious complication of inhibitors of GP IIb/IIIa such as abciximab (Huxtable et al., 2006). The mechanism appears to be related to exposure of epitopes on GP IIb/IIIa that react with naturally occurring antibodies. Because the reaction is dependent on antibodies formed prior to exposure to drug, it may occur shortly after the first exposure to the drug. Ligand binding is known to alter the conformation of GP IIb/IIIa. The GP IIb/IIIa inhibitors bind at the ligand binding site and also cause a conformational change in GP IIb/IIIa, permitting naturally occurring antibodies to bind to and initiate clearance of platelets by the mononuclear phagocytic system.

Heparin-induced thrombocytopenia (HIT) represents another mechanism of immune-mediated platelet destruction. This disorder is due to the development of antibodies that react with a multimolecular complex formed by the interaction between heparin and a protein, usually platelet factor 4 (PF 4) (Davoren and Aster, 2006; Warkentin and Greinacher, 2004). When the relative concentration of heparin to PF 4 is appropriate, formation of this complex is associated with exposure of a neoepitope on PF 4 (or another target protein) and development of an IgG response to the neoepitope. The IgG then binds to the PF 4–heparin complex to form an immune complex that binds to the platelet Fc receptor, FcγRIIa. Clustering of platelet FcγRIIa by the immune complex activates biochemical signaling pathways mediated by the cytoplasmic domain of FcγRIIa. This results in platelet activation and aggregation. During the process of platelet activation, platelet microparticles that promote thrombin generation are released. Consequently, HIT is associated with both thrombocytopenia and an increased risk of arterial and venous thrombosis. Other drug–antibody complexes (eg, streptokinase–IgG) may trigger platelet activation and thrombocytopenia through a similar mechanism (Deckmyn et al., 1998; McRedmond et al., 2000).

TTP is a syndrome characterized by the sudden onset of thrombocytopenia, a microangiopathic hemolytic anemia, and multisystem organ failure, which often includes neurologic dysfunction. The syndrome tends to occur following an infectious disease but may also occur following administration of some drugs. The pathogenesis of TTP appears to be related to the ability of unusually large vWF multimers to activate platelets, even in the absence of significant vascular damage (Lian, 2005). Although these large multimers are normally secreted into blood by endothelial cells, they are rapidly processed into smaller multimers by a protease (ADAMTS13) present in plasma. Acquired TTP is associated with the development of an antibody that inhibits this protease, permitting the very large vWF multimers to persist in the circulation (Lian, 2005; Veyradier and Meyer, 2005). Consequently, these multimers bind to platelet GP Ib/IX/V and induce platelet activation and aggregation. The organ failure and hemolysis in TTP is due to the formation of platelet-rich microthrombi throughout the circulation. The development of TTP or TTP-like syndromes has been associated with drugs such as ticlopidine, clopidogrel, cocaine, mitomycin, and cyclosporine (Veyradier and Meyer, 2005; Zakarija and Bennett, 2005). In some cases, drug-induced TTP is related to development of antibodies to ADAMTS13, whereas in other cases it may be due to direct endothelial cell toxicity from the drug.

The HUS is a disorder characterized by clinical features similar to those of TTP, with microangiopathic hemolytic anemia, thrombocytopenia, and renal failure (Zakarija and Bennett, 2005). Neurologic complications tend to be less severe, while renal failure often dominates the clinical picture. Sporadic cases of HUS have been linked to infection with verocytotoxin-producing Escherichia coli, but they may also occur during therapy with some drugs, including mitomycin. In contrast to TTP, the vWF-cleaving protease is normal in patients with HUS (Furlan et al., 1998). The pathogenesis of the thrombocytopenia and microangiopathic changes in HUS is still uncertain, but there is experimental evidence suggesting that it is related to endothelial cell injury, with subsequent platelet activation and thrombus formation (Delvaeye et al., 2009).

Desmopressin, a vasopressin analog, is an example of nonimmune-mediated increased platelet destruction. It induces a 2- to 5-fold increase in the plasma concentration of vWF and factor VIII. It is commonly used in the treatment of patients with von Willebrand disease and other mild bleeding syndromes. Desmopressin has been associated with the development or accentuation of thrombocytopenia in some patients with type 2B von Willebrand disease. The thrombocytopenia in such cases is related to the release of an abnormal vWF from endothelial cells. The abnormal vWF has enhanced affinity for GP Ib/IX/V and the interaction of the vWF with its receptor leads to platelet clearance from the circulation (Mannucci, 1998).

Toxic Effects on Platelet Function

Platelet function is dependent on the coordinated interaction of a number of biochemical response pathways. A variety of drugs and foods have been found to inhibit platelet function, either in vivo or in vitro (Abrams, 2006). Major drug groups that affect platelet function include nonsteroidal anti-inflammatory drugs, β-lactam-containing antibiotics, cardiovascular drugs, particularly β-blockers, psychotropic drugs, anesthetics, antihistamines, and some chemotherapeutic agents. The effect of these drugs can vary between individuals, perhaps due to subclinical variations in underlying platelet function.

Xenobiotics may interfere with platelet function through a variety of mechanisms. Some drugs inhibit the phospholipase A₂/cyclooxygenase pathway and synthesis of thromboxane A₂ (eg, nonsteroidal anti-inflammatory agents). Other drugs appear to interfere with the interaction between platelet agonists and their receptors (eg, antibiotics, ticlopidine, clopidogrel). As the platelet response is dependent on a rapid increase in cytoplasmic calcium, any chemical that interferes with translocation of calcium may inhibit platelet function (eg, calcium channel blockers). Occasionally, drug-induced antibodies will bind to a critical platelet receptor and inhibit its function. The functional defect induced by such antibodies may potentiate the bleeding risk associated with the xenobiotic-induced thrombocytopenia. In some cases, the mechanism of inhibition is not known.

The effect of xenobiotics on platelet function can often be studied following in vitro exposure of platelets to the chemical of interest. However, evaluation following exposure is preferred, as metabolites of the parent compound may contain the platelet inhibitory activity. The most common method of assessing platelet function is turbidimetric platelet aggregation using platelet-rich plasma, but alternate techniques are available, including the PFA 100 analyzer, flow cytometry, and whole-blood impedance aggregometry (Cesar et al., 2005; Matzdorff, 2005).

Toxic Effects on Fibrin Clot Formation

Coagulation

Fibrin clot formation is the result of sequential activation of a series of serine proteases that culminates in the formation of thrombin (Lane et al., 2005; Monroe and Hoffman, 2005). Thrombin is a multifunctional enzyme that converts fibrinogen to fibrin; activates factors V, VIII, XI, XIII, protein C, and platelets; and interacts with a variety of cells (eg, leukocytes and endothelial cells), activating cellular signaling pathways (Lane et al., 2005).

The most common toxic effect of xenobiotics on fibrin clot formation are related to a decreased level of one or more of the critical proteins necessary for this process. The decrease in clotting factor activity may be due to decreased synthesis of the protein(s) or increased clearance from the circulation. Decreased synthesis is most often a reflection of hepatocellular damage or interference with vitamin K metabolism, as discussed below, whereas increased clearance is usually associated with the development of an antibody to a specific coagulation factor.

Decreased Synthesis of Coagulation Proteins

The majority of proteins involved in the coagulation cascade are synthesized in the liver. Therefore, any chemical that impairs liver function may cause a decrease in production of coagulation factors. The common tests of the coagulation cascade, the prothrombin time (PT) and activated partial thromboplastin time (aPTT), may be used to screen for liver dysfunction and a decrease in clotting factors. These assays are often performed as part of the safety evaluation of a new chemical entity. The half-life of clotting factors in the circulation varies significantly, with factor VII having the shortest half-life (three–four hours). Therefore, with acute toxicity (eg, acetaminophen overdose), the effect on blood coagulation may be first seen as a decrease in the level of factor VII. Such a decrease would lead to prolongation of the PT with a normal aPTT. With a more chronic process, the PT, the aPTT, or both may be affected.

Factors II, VII, IX, and X are dependent on vitamin K for their complete synthesis (Stafford, 2005). Anything that interferes with vitamin K metabolism may lead to a deficiency of these factors and a bleeding tendency. This may occur with xenobiotics that interfere with absorption of vitamin K from the intestine or with chemicals that interfere with the reduction of vitamin K epoxide (Table 11-10). The combination of antibiotic therapy and limited oral intake is a common cause of acquired deficiency of vitamin K–dependent proteins among hospitalized patients (Chakraverty et al., 1996). The “super rodenticides” are another cause of acquired vitamin K deficiency (Berry et al., 2000; Chua and Friedenberg, 1998). These agents have a very prolonged half-life in vivo; thus, the coagulation defect may persist for weeks or months following exposure. Rodenticide exposure may occur accidentally, as part of a Munchausen syndrome, in association with a suicide attempt, or as part of a homicide attempt. At times it may be important to distinguish between a true vitamin K deficiency and interference with the reduction of vitamin K epoxide. This is most readily accomplished by measuring the level of vitamin K and vitamin K epoxide in serum or plasma. In the case of vitamin K deficiency, both vitamin K and vitamin K epoxide are decreased, whereas in the case of inhibition of vitamin K reduction, vitamin K epoxide is significantly increased. Specific rodenticides may be measured using HPLC techniques, but it is important to specify which active agent (eg, brodifacoum) should be measured, as the assays may not show cross-reactivity between agents.

Increased Clearance of Coagulation Factors

Idiosyncratic reactions to xenobiotics include the formation of antibodies that react with coagulation proteins. These antibodies bind to the coagulation factor, forming an immune complex that is rapidly cleared from the circulation and resulting in deficiency of the factor. The antibody is often reversible over time if the initiating agent is withdrawn. However, during the acute phase, these patients may have life-threatening bleeding. The factors that are most often affected include factors VIII, V, and XIII, vWF, prothrombin, and thrombin (Table 11-11) (Lollar, 2004, 2005). Many of these antibodies inhibit the functional activity of the coagulation factor in addition to increasing the rate of clearance. Other antibodies have been demonstrated to have catalytic activity, resulting in proteolysis of the target coagulation factor (Ananyeva et al., 2004).

Lupus anticoagulants are antibodies that interfere with in vitro phospholipid-dependent coagulation reactions (Bertolaccini et al., 2004). Although it was once hypothesized that these antibodies were directed against phospholipid, it is now evident that lupus anticoagulants are directed against phospholipid-binding proteins, including prothrombin and β₂-glycoprotein 1. These antibodies usually do not cause a deficiency of any specific coagulation factor. However, in vivo, these antibodies can potentiate procoagulant mechanisms and interfere with the protein C system. Consequently, these antibodies have been associated with an increased risk of thrombosis (Bertolaccini et al., 2004). The development of lupus anticoagulants has been seen in association with a variety of medications, including procainamide, chlorpromazine, and hydralazine.

Toxicology of Chemicals Used to Modulate Hemostasis

Patients with bleeding or thrombotic problems are commonly encountered in clinical practice. A variety of agents are available to treat such patients, ranging from recombinant hemostatic proteins to chemical entities that modulate the activity of the coagulation system. The major toxicologic reactions to plasma-derived products are infectious diseases (eg, hepatitis C) and allergic reactions, which can be severe. The use of some products, such as activated concentrates of vitamin K–dependent proteins (eg, Autoplex and FEIBA), has been associated with the development of disseminated intravascular coagulation and/or thrombosis in some patients (Mannucci, 1998).

Oral Anticoagulants

Oral anticoagulants (warfarin) interfere with vitamin K metabolism by preventing the reduction of vitamin K epoxide, resulting in a functional deficiency of reduced vitamin K (Ansell et al., 2004). These drugs are widely used for prophylaxis and therapy of venous and arterial thrombosis. The therapeutic window for oral anticoagulants is relatively narrow, and there is considerable interindividual variation in the response to a given dose. A number of factors, including concurrent medications and genetics, affect the individual response to oral anticoagulants (Ansell et al., 2004; D’Andrea et al., 2005; Rojas et al., 2005). For these reasons, therapy with these drugs must be routinely monitored to maximize both safety and efficacy. This is routinely performed with the PT, with results expressed in terms of the international normalized ratio (INR).

A number of xenobiotics, including foods, have been found to affect the response to oral anticoagulants (Ansell et al., 2004). Perhaps the most common mechanism for interference with oral anticoagulants is mediated by inhibition of CYP2C9 (Rojas et al., 2005). Other mechanisms of interference include induction of CYP2C9, which tends to diminish the effect of warfarin by shortening its half-life; interference with absorption of warfarin from the gastrointestinal tract; displacement of warfarin from albumin in plasma, which temporarily increases the bioavailability of warfarin until equilibrium is reestablished; diminished vitamin K availability, due to either dietary deficiency or interference with the absorption of this lipid-soluble vitamin; and inhibition of the reduction of vitamin K epoxide, which potentiates the effect of oral anticoagulants.

Just as other drugs interfere with the action of oral anticoagulants, administration of oral anticoagulants may affect the activity of other medications, particularly those that are metabolized by CYP2C9. Dicumarol administration prolongs the half-life of chlorpropamide and phenytoin, resulting in hypoglycemia in the case of chlorpropamide and an increased plasma drug concentration in the case of phenytoin. Bis-hydroxycoumarin, but not warfarin, potentiates the activity of tolbutamide, resulting in enhanced hypoglycemia (Harder and Thurmann, 1996).

Oral anticoagulants have been associated with the development of warfarin-induced skin necrosis (Ansell et al., 2004). This disorder is due to the development of extensive microvascular thrombosis in the affected skin. This uncommon toxic effect is thought to be related to a rapid drop in protein C following administration of the drug, resulting in impaired protein C function. The risk of developing warfarin-induced skin necrosis increases with the dose of warfarin used to initiate therapy, particularly when the initial dose exceeds 10 mg per day; this is one of the reasons that loading doses of warfarin are no longer recommended.

Vitamin K is necessary for the synthesis of proteins other than the coagulation-related factors, including osteocalcin, a major component of bone. Perhaps because of this, long-term administration of warfarin has been associated with bone demineralization (Ansell et al., 2004; Stafford, 2005). This effect can be important in patients with borderline bone density. Administration of warfarin during pregnancy, particularly the first 12 weeks of pregnancy, is associated with congenital anomalies in 25% to 30% of exposed infants (Bates et al., 2004). Many of the anomalies are related to abnormal bone formation. It is thought that warfarin may interfere with synthesis of proteins critical for normal structural development.

Heparin

Heparin is a widely used anticoagulant for both prophylaxis and therapy of acute venous thromboembolism (Hirsh and Raschke, 2004). In many hospitals, the majority of patients are exposed to this potent anticoagulant at some point during their hospitalization. The major complication associated with heparin therapy is bleeding, a direct manifestation of its anticoagulant activity. The risk of bleeding is related to the intensity of therapy, the patient’s body mass and underlying condition, and the presence of other hemostatic defects (eg, thrombocytopenia).

As discussed in the section “Toxic Effects on Platelets,” heparin administration is also associated with the development of HIT. For unknown reasons, this complication occurs more frequently with heparin derived from bovine sources than with that derived from porcine sources. The incidence of HIT is also significantly higher in patients receiving unfractionated heparin than it is in patients receiving low-molecular-weight heparin (Davoren and Aster, 2006; Warkentin and Greinacher, 2004).

Long-term administration of heparin is associated with an increased risk of clinically significant osteoporosis (Dinwoodey and Ansell, 2006; Hirsh and Raschke, 2004). The mechanism underlying the development of osteoporosis in these patients is not known. Patients may suffer from spontaneous vertebral fractures and demineralization of long bones of the arms and legs. The risk of osteoporosis may be less with low-molecular-weight heparin as compared with that with unfractionated heparin.

Heparin administration may also cause a transient rise in serum transaminases, suggesting significant liver dysfunction (Guevara et al., 1993). However, the rise is rapidly reversible on discontinuation of heparin and may reverse even before heparin is discontinued. The elevation of serum transaminases has not been associated with chronic liver dysfunction. The mechanism of heparin-induced increase in transaminases is not known.

In 2008, serious injuries and deaths were associated with the use of heparin originating in China. It was determined that certain sources of heparin were contaminated with oversulfated chondroitin sulfate (OSCS). The presence of OSCS is believed to be responsible for the observed severe allergic reactions, gastrointestinal disturbances, and hypotension (Blossom et al., 2008).

Fibrinolytic Drugs

Fibrinolytic drugs are used in the treatment of acute thromboembolic disease with the goal of dissolving the pathogenic thrombus (Collen and Lijnen, 2005). Each of these drugs works by converting plasminogen, an inactive zymogen, to plasmin, an active proteolytic enzyme. Plasmin is normally tightly regulated and is not freely present in the circulation. However, administration of fibrinolytic drugs regularly results in the generation of free plasmin leading to systemic fibrin(ogen)olysis. The toxicology of the fibrinolytic drugs can be divided into toxic effects of systemic plasmin activation and toxic effects of the activators themselves.

Systemic fibrinolysis is associated with the development of a complex coagulopathy characterized by a decrease in fibrinogen, factors V and VIII, and α₂-antiplasmin; an increase in circulating fibrin split products; degradation of platelet GP Ib/IX/V and IIb/IIIa; degradation of endothelial cell glycoproteins; degradation of fibronectin and thrombospondin; and prolongation of the PT, aPTT, and thrombin time (Hajjar, 2006). All of these effects potentiate the risk of bleeding. Anatomic locations that are frequently involved in bleeding complications include the cerebral circulation and sites of recent vascular access. As systemic plasmin can lyse physiologic as well as pathologic thrombi, reactivation of bleeding from sites of vascular access is not uncommon. Platelet inhibitors and heparin are commonly used in conjunction with fibrinolytic therapy to prevent recurrent thrombosis. As one might expect, the concurrent use of anticoagulants with systemic fibrinolysis may contribute to the risk of bleeding (Menon et al., 2004).

Another complication associated with fibrinolysis is recurrent thrombosis at the site of pathologic thrombosis. While rethrombosis may be related to underlying damage to the vascular wall, there is some evidence that fibrinolytic therapy may contribute to this process. For example, plasmin, in appropriate concentrations, can actually induce platelet activation (McRedmond et al., 2000). This process may be mediated by plasmin or streptokinase/plasminogen cleavage of the platelet thrombin receptor (protease activated receptor-1). Cleavage of the receptor is associated with activation of the platelet biochemical signaling pathways. There is sufficient “cross-talk” between the fibrinolytic system and the contact system of coagulation that one could also anticipate increased thrombin generation occurring as a result of fibrinolytic therapy (Schmaier et al., 1999).

Streptokinase is a protein derived from group C β-hemolytic streptococci and is antigenic in humans. Antibody formation to streptokinase occurs commonly in association with streptococcal infections as well as exposure to streptokinase. Acute allergic reactions may occur in 1% to 5% of patients exposed to streptokinase, and these allergic reactions may consist of minor symptoms such as hives and fever as well as major, life-threatening anaphylactic reactions. In addition, delayed hypersensitivity reactions associated with severe morbidity may occur (Curzen et al., 1998; Siebert et al., 1992). Allergic reactions also occur with other fibrinolytic agents containing streptokinase (eg, anisoylated plasminogen–streptokinase complex) or streptokinase-derived peptides. The immune complex formed by IgG and streptokinase is capable of binding to and clustering platelet FcγRIIa, initiating platelet activation and aggregation (McRedmond et al., 2000).

Urokinase and recombinant tissue plasminogen activator (t-PA) are generally not associated with allergic reactions. However, work is progressing on a number of genetically engineered forms of t-PA. Whether or not such mutant forms of t-PA are immunogenic has not been firmly established (Collen and Lijnen, 2005).

Inhibitors of Fibrinolysis

Inhibitors of fibrinolysis are commonly used to control bleeding in patients with congenital abnormalities of hemostasis, such as von Willebrand disease. Tranexamic acid and ε-aminocaproic acid are small molecules that block the binding of plasminogen and plasmin to fibrin and other substrate proteins through interaction with lysine binding sites on plasmin(ogen). Although relatively well tolerated, there is some evidence that administration of these chemicals may increase the risk of thrombosis due to the inhibition of the fibrinolytic system (Mannucci, 1998). In a single case, intravenous infusion of ε-aminocaproic acid in a patient with chronic renal failure was associated with acute hyperkalemia (Perazella and Biswas, 1999).

Aprotinin is a naturally occurring polypeptide inhibitor of serine proteases. It is usually derived from bovine material and consequently is immunogenic when administered to humans. Aprotinin is given by intravenous infusion, as it is inactive when given orally. Allergic reactions in response to aprotinin have been reported, ranging from minor cutaneous manifestations to anaphylactic reactions (Peters and Noble, 1999). A more recent observational study compared clinical outcomes after cardiac surgery performed with aprotinin, tranexamic acid, or e-aminocaproic acid with outcomes after cardiac surgery without an inhibitor of fibrinolysis (Mangano et al., 2006). In this study, use of antifibrinolytics was associated with decreased blood loss. However, use of aprotinin was associated with a significant increase in end-organ damage, including renal, cardiac, and cerebral events. The design and results of this study have been questioned, with calls for prospective randomized comparator studies (Sedrakyan et al., 2006). The results also point out the intricate balance within the hemostatic system and the potential for problems when modulating the activity of one portion of this system.

Assessing the risk that exposure to new drugs, chemical products, and other agents poses to humans—in terms of significant toxic effects on hematopoiesis and the functional integrity of blood cells and hemostatic mechanisms—can be logistically and intellectually challenging. This is due in part to the complexity of hematopoiesis and the range of important tasks that these components perform, as previously discussed. A central issue in drug and nontherapeutic chemical development is the predictive value of preclinical toxicology data and the expansive but inevitably limited preregistration clinical database for the occurrence of significant hematotoxicity on broad exposure to human populations. Appropriately, this area of well-resourced applied toxicology is highly regulated yet provides unique and exciting opportunities for sophisticated, well-controlled research (Bloom, 1993).

Preclinical Risk Assessment

Most preclinical studies that assess the potential for candidate drugs or nontherapeutic chemicals to induce hematotoxicity in humans are performed in industry as part of the routine safety evaluation of these molecules. These studies are largely prescribed by government regulatory bodies of the various countries and regions, including the United States, the European Union, and Japan (ICH, 2000; Hall, 1992, 1997). The unique accessibility of blood as a potential target organ, which allows direct ex vivo assessment of the tissue (compared with the indirect parameters used to detect effects on liver, kidney, etc), together with the use of hematologic monitoring to reflect systemic effects of other organ toxicity, enables more sensitive signal detection, and may account for the relatively high predictive value of these studies. In fact, earlier studies have shown that hematotoxicity that is preclinically identified predicts toxicity in human clinical trials with a 91% concordance (Olson et al., 2000). It is doubtful, however, that preclinical animal studies are as predictive for today’s candidate drug portfolios of highly targeted agents that include more human recombinant proteins, monoclonal antibodies, and targets that are specific to human tissue and disorders such as cancer and autoimmune disease. Another consequence is that more sophisticated biomarkers are increasingly required to demonstrate both pharmacodynamic and off-target effects in human and animal models, some of which provide opportunities for companion diagnostics and personalized medicine. Immunogenicity of heterologous proteins further complicates pharmacokinetic, pharmacodynamic, and safety assessment in animals. As a result, there is a greater sense of urgency to study these novel agents in humans. Accordingly, phase I clinical studies have become more important not only for safety assessment but also as a venue for early proof of concept experiments. Finally, as with other organ toxicity, the predictive value of preclinical studies for idiosyncratic iatrogenic disease in humans, such as aplastic anemia and agranulocytosis, is limited.

The issues relating to the assessment of blood as a target organ that confront the industrial toxicologist are largely similar to those of other target organs and include the selection of the appropriate animal model, how to best monitor for hematotoxicity, and the appreciation of species differences in responding to hematotoxic insults.

Animal Models and Hematologic Monitoring

Selection of a species that is practical to study and predictive for hematotoxicity in humans is always a challenge. While this is driven in part by the aforementioned regulatory requirements, the selection is also influenced by other considerations, including having a pharmacokinetic profile comparable to that of humans, prior information on sensitivity of a particular species to a class of compounds, the ability to fully characterize effects on peripheral blood and bone marrow, and practical considerations, such as logistics and economics (Bloom, 1993). These become of particular importance in choosing a model to fully characterize the toxicity of a chemical known to have a hematotoxic potential.

Of the commonly used animal species, rats and mice offer the advantage of their small size, which favorably impacts test compound requirements and number of subjects that can be economically housed and tested. Both have been well characterized hematologically (Moore, 2000a,b; Valli and McGrath, 1997). Blood volume limitations, however, often prohibit the frequent, or serial, evaluation of blood and bone marrow required to characterize the progression of a hematotoxic effect. While this can be addressed in part through serial sacrifices, the inability to fully characterize individual animals poses a significant disadvantage. Test results will also vary in accordance with the phlebotomy site and method, particularly in rodents (Suber and Kodell, 1985), and with the physical and chemical restraint employed (Loomis et al., 1980).

Serial blood and bone marrow sampling is practical in larger species, such as the dog and monkey. These models offer the additional advantage of being hematologically more similar to humans, as regards hematopoiesis and blood cell kinetics, which in the monkey extends to immunohematologic features (Ladiges et al., 1990; Shifrine and Wilson, 1980). The latter species, however, presents more interanimal hematologic variability, particularly in wild-caught primates, due to temperament, vascular access, and other influences, which include nutritional status and infection.

Tests used to assess blood and bone marrow in preclinical toxicology studies will vary with the phase or objective of the evaluation (acute, subacute, chronic), the intended use of the chemical, and what is understood or suspected regarding the toxicologic profile of the xenobiotic. Ideally, the studies in aggregate should provide information on the effects of single- and multiple-dose exposure on erythrocyte parameters (RBC, Hbg, PCV, MCV, MCHC), leukocyte parameters (WBC and absolute differential counts), thrombocyte counts, screening coagulation tests (PT, aPTT), peripheral blood cell morphology, and bone marrow cytologic and histologic features (Bloom, 1993; Lund, 2000; Weingand et al., 1996). Notable are the applications of flow cytometry, which have been refined in recent years and recently reviewed regarding the advantages and disadvantages of these techniques in preclinical studies (Reagan et al., 2011). Forward-angle scatter has been used to effectively type canine bone marrow hematopoietic cells (Weiss et al., 2000), where that combined with side-angle scatter and the use of fluorochromes and lymphocyte-specific antibodies has been accurate and reproducible in rats (Chriswell et al., 1998a). The latter correlates well with manual counts and has been used to characterize and quantitate the effects of phenylhydrazine, phlebotomy, cyclophosphamide, and various hematopoietic growth factors (Chriswell et al., 1998b, 2000). Other techniques using cell surface markers have been used to enumerate subpopulations, which must be validated for each species due to variation in surface antigen expression and other preanalytical variables (Reagan et al., 2011).

Additional tests should be employed in a problem-driven fashion, as required to better characterize findings from the aforementioned screening efforts or to more fully explore a class-specific effect or other hematotoxicologic potential of concern (Bloom, 1993). Examples of these tests are listed in Table 11-12. While much progress has been made in validating many of the more specialized assays in our principal animal models, additional validation that addresses laboratory- and species-specific preanalytical and analytical variables is often required.

Despite the importance of bone marrow assessment in detecting and characterizing hematotoxicity, there is remarkably little regulatory guidance on indications, frequency, or preferred techniques. The Federal Drug Administration (FDA) Centers for Drug (CDER) and Biologic (CBER) Evaluation and Research, and the European Medicines Agency (EMA) have issued guidelines largely in the context of immunotoxicity risk assessment (Food and Drug Administration et al., 2006; European Medicines Agency and Committee for Proprietary Medicinal Products, 2006). Limited guidance is also provided by Japan’s Ministry of Health, Labor and Welfare, the FDA’s Center for Veterinary Medicine, and the Environmental Protection Agency (Reagan et al., 2011). To address this gap, a joint effort of the American Society for Veterinary Clinical Pathology and the Society of Toxicologic Pathology (Bone Marrow Working Group) has resulted in a recent publication on best practices for evaluation of bone marrow in nonclinical toxicity studies (Reagan et al., 2011).

Because hematologic features and response to disease can vary substantially among animal species, it is essential that the toxicologist fully understands the hematology of the animal model used for preclinical risk assessment. While complete and accurate reference data are helpful, they do not provide information on pathophysiology that may be species-specific and required to accurately interpret the preclinical data. Examples of these features include the relative influence of preanalytical variables (blood collection technique, nutritional status, sample stability), response to blood loss or hemolysis, stress effects on the leukogram, susceptibility to secondary effects associated with other target organ toxicity, etc. It is beyond the scope of this chapter to fully discuss the comparative hematology of laboratory animals, which is provided in several excellent reviews (Feldman et al., 2006; Valli and McGrath, 1997).

In Vitro Bone Marrow Assays

As with other target organ risk assessment, in vitro methods for assessing potential hematotoxicity are attractive in that they are faster and less expensive than in vivo studies, while providing data that often suggest or clarify the mechanism of a toxic effect. Drug- or chemical-induced bone marrow suppression can result from effects on specific hematopoietic stem cells or on the hematopoietic microenvironment. These effects can be distinguished and confirmed using short-term clonogenic assays and long-term functional assays, respectively (Deldar, 1994; Naughton et al., 1992; Williams et al., 1988). The former include burst-forming-unit erythroid (BFU-E), colony-forming-unit erythroid (CFU-E), colony-forming-unit granulocyte/monocyte (CFU-GM), colony-forming-unit megakaryocyte (CFU-MK), and colony-forming-unit granulocyte, erythroid, megakaryocyte, monocyte (CFU-GEMM), which have been developed for several laboratory animal species (Deldar and Parchment, 1997; Pessina et al., 2005). It is therefore possible to examine effects on the myeloid, erythroid, and megakaryocytic lineages in a fashion where concentrations of the chemical are tightly controlled, as is duration of exposure to it.

While the most commonly applied clonogenic techniques used to assess the above still employ viscous or semisolid matrices and culture supplements that promote differentiation of the hematopoietic cells (Erickson-Miller et al., 1997; McMullin et al., 1998), some newer techniques measure intracellular ATP, which is influenced by artifacts such as apoptosis, cytotoxicity, necrosis, and fibroblast contamination (Fan and Wood, 2007; Slater, 2001). Several commercial services are available that provide stem cells of human, canine, primate, and rodent origin to which are added cytokines and growth factors to support cell development (Reagan et al., 2011).

In vitro clonogenic assays are best used in a preclinical setting in combination with in vivo testing. Used in this way, the predictive value of these assays is enhanced. This has been particularly true for anticancer and antiviral drugs, where the in vitro component of risk assessment has been used for therapeutic index-based screening to identify less myelosuppressive analogs, structure–toxicity relationships, and new-drug lead candidates (Deldar and Stevens, 1993; Parchment et al., 1993; Pessina et al., 2003). Other advantages of the in vitro hematopoietic stem cell assays include the opportunities they provide to test combinations of chemicals as well as their metabolites and effects of serum and other cell components, such as lymphocytes (Deldar and Parchment, 1997). Most important is the ability to test human hematopoietic cells directly in a preclinical setting, thus obviating extrapolation considerations. Concern for possible metabolic activation can be addressed by culturing the target cells in question with metabolizing systems in a cell-free extract (s9), with isolated hepatocytes, or with other CYP450-expressing cell types (Frazier, 1992; Negro et al., 2001).

Perhaps the most interesting use of these in vitro clonogenic assays in risk assessment has been their role in making practical interspecies comparisons regarding sensitivity to a particular agent or group of drugs or chemicals. Comparisons to the sensitivity of human cells can be made that have implications for the relative predictive value of various animal models for hematotoxicity in humans. Examples include the resistance of murine CFU-GM to the anticancer drug topotecan relative to that of the canine and human cells (Deldar and Stevens, 1993). This is consistent with the early observations of Marsh—that the dog is a particularly predictive model for the myelosuppression associated with anticancer drugs in humans (Marsh, 1985). Thus, while some chemicals show comparable suppressive activity across species lines (doxorubicin, pyrazoloacridine, hepsulfan, cyclopentenyl cytosine), others, such as camptothecins, carboxyamidotriazole, and fostriecin, show differences of as much as three log concentrations (Du et al., 1991; Horikoshi and Murphy, 1982; Reagan et al., 1993).

Because myelotoxicity, and particularly suppression of granulopoiesis, is a major limitation in administering anticancer drugs, and is often used to determine the optimal dose, the clonogenic assay for CFU-GM has received the most attention as a preclinical tool for predicting this response. It is the only assay that has been validated by an international study supported by the European Centre for Validation of Alternative Methods (ECVAM) (Pessina et al., 2003). Analytical validation was completed for the human and mouse assays, and a predictive model was developed to calculate the maximum tolerated dose (MTD) in humans using data from the mouse assay that is adjusted for the established interspecies variation (Pessina et al., 2001). The model was applied in an international blind trial to 20 drugs that included 14 antineoplastic drugs, the antiviral drugs zidovudine and acyclovir, and the pesticide lindane (Pessina et al., 2003). It predicted the MTD for all 20 chemicals, although extrapolation on the regression curve out of the range of the actual drug doses tested was required to derive the IC90 for 10 of these drugs. Since then, the Scientific Advisory Committee of the ECVAM endorsed the CFU-GM assay for predicting acute neutropenia in humans as a substitute to using a second species, such as the dog, for this purpose (24th meeting at ECVAM of the ECVAM Scientific Advisory Committee, European Commission, March 2006). Subsequent studies have confirmed the predictive value of this assay in both anticancer drugs (Masubuchi, 2006) and nononcology candidate drugs (Horn et al., 2008). The latter study illustrates the species specificity encountered and the importance of integrating the in vivo and in vitro approaches required to fully characterize the hematotoxicity observed in preclinical safety assessment today.

While this discussion has focused on in vitro hematopoietic clonogenic assays in the context of risk assessment, these assays have also proven to be extraordinarily useful tools for investigating mechanisms of toxic cytopenia in humans (Deldar, 1994). Parchment and Murphy (1997) review the application of these to four categories of hematologic toxicity observed clinically: (1) the reversible cytopenia following acute exposure to a cytotoxic or cytostatic chemical; (2) the permanent loss in the production of a mature blood cell type(s); (3) cytosis, or the dramatic increase in blood cell counts following single or repeated toxicant exposure; and (4) the progressive loss of one or more blood cell lineages during chronic exposure to a toxicant. In all these circumstances, in vitro and ex vivo hematopoietic clonogenic assays have proven useful in understanding the mechanism(s) of these toxic effects and formulating strategies for risk management and treatment.

Emerging Technologies

As discussed previously, the well-controlled and resource-intensive preclinical toxicology and safety studies prescribed for candidate drugs and nontherapeutic chemicals provide unique opportunities for studies on mechanisms of xenobiotic-induced hematotoxicity (Bloom, 1993; Deldar and Parchment, 1997). Together with the traditional biomarkers for toxicity, mechanism-based technologies such as toxicogenomics, proteomics, and metabonomics are now available to the toxicologist and are increasingly helpful in both predicting toxicity and determining whether a particular finding is relevant to humans (Todd and Ulrich, 1999). Challenges that the application of these tools presents relate to informatics, standardization, and validation (Reynolds, 2005).

Clinical Trials and Risk Assessment

As with preclinical risk assessment, most of the clinical research on hematotoxicity is driven by regulatory requirements and supported by the drug, cosmetic, and chemical industries. The challenges and opportunities this presents are similar to those in preclinical development with the following differences. Most clinical studies involve actual patients with the targeted disease, in contrast to the inbred, healthy, well-defined animals employed in preclinical studies. This presents additional variables and challenges to manage. Second, the scale of clinical trials, the volume of data produced, and the resources required exceed by orders of magnitude those of preclinical studies. Third, many clinical trials involve research cooperative groups that represent a network of clinical scientists from academic medical centers, such as the Eastern Cooperative Oncology Group (ECOG), the AIDS Clinical Trial Group (ACTG), Thrombolysis in Myocardial Infarction (TIMI), and others. Most of the information on drug- or chemical-induced hematotoxicity in humans is collected through this industry-sponsored and highly regulated clinical research.

It is well understood that the ways in which drugs and nontherapeutic chemicals affect the hematopoietic system are influenced by both the nature of the chemical and the response of the subject or target population. As discussed previously, many chemicals are known to induce dose-dependent hematotoxicity in a fashion that is highly predictable. Others cause toxicity in a small number of susceptible individuals, and these often include chemicals not otherwise hematotoxic in most individuals (Dieckhaus et al., 2002; Patton and Duffull, 1994). These idiosyncratic reactions present the biggest challenge as regards detection and characterization before human patients or populations are broadly exposed. They include aplastic anemia, thrombocytopenia, hemolysis, and leukopenia, which may or may not be immune-mediated (Salama et al., 1989).

Prevailing theories today, as regards mechanisms for these idiosyncratic reactions, have been discussed previously. The chemical structure can be a risk factor if it is similar to that of other known toxicants. Patient- or population-related risk factors include pharmacogenetic variations in drug metabolism and detoxification that lead to reduced clearance of the chemical or production of novel intermediate metabolites (Cunningham et al., 1974; Gerson et al., 1983; Mason and Fischer, 1992), histocompatibility antigens (Frickhofen et al., 1990), interaction with drugs or other chemicals (West et al., 1988), increased sensitivity of hematopoietic precursors to damage (Vincent, 1986), preexisting disease of the bone marrow, and metabolic defects that predispose to oxidative or other stresses associated with the chemical (Stern, 1989).

In drug development, the clinical evaluation of candidate molecules is usually performed in 3 phases: phase I examines the effect of single and multiple increasing doses in small numbers of normal and/or patient volunteers. Pharmacokinetic properties are usually addressed, as well as the routes of excretion and metabolism; and the assessment of active and inactive metabolites. The emphasis is usually on safety assessment. Phase II includes controlled studies in the target patient population that examine both safety and efficacy. They explore dose response and usually provide the first indication of benefit versus risk. Phase III entails larger studies designed to confirm efficacy in an expanded patient population and evaluate less frequent adverse effects, such as the aforementioned idiosyncratic blood dyscrasias.

Development of a demonstrably hematotoxic drug is usually stopped in phase I or II, unless the indication includes life-threatening conditions where toxicity is acceptable (eg, anticancer drugs). Thus, drugs tested in phase III generally show an acceptable safety profile in most subjects at the doses used. Even phase III studies, however, are not usually powered to detect the low incidence of idiosyncratic hematotoxicity previously discussed (Levine and Szarfman, 1996). In order to detect one adverse event affecting 1% of an exposed patient population at a 95% confidence level, a trial must include approximately 300 subjects (O’Neill, 1988). Clinical databases supporting new drug applications generally cannot be used to detect adverse events that occur below 1 per 1000 exposures (ICH, 2001), and most will not rule out events with a frequency of less than 1 per 500 (Szarfman et al., 1997). Thus, rare, delayed, or cumulative toxicity is often missed in preregistration clinical trials.

Detection of low-incidence hematotoxicity is usually achieved through postmarketing surveillance, such as the MedWatch program introduced by the FDA in 1993 (Szarfman et al., 1997). Other countries that practice comprehensive postmarketing surveillance include Canada, the United Kingdom, Sweden, Germany, France, Australia, and New Zealand. Adverse event data, including serious hematotoxicity, are provided to the WHO; this information is compiled by a computer-based recording system employing WHO terminology and system and organ classifications for adverse reactions (Edwards et al., 1990). Examples of iatrogenic blood dyscrasias detected through postmarketing surveillance include the hemolysis and thrombocytopenia associated with the antibiotic temafloxacin, the aplastic anemia linked to the antiepileptic felbamate, the hemolysis caused by the antidepressant nomifensine, and the agranulocytosis associated with the antiarrhythmic aprindine.

The WHO has also established criteria for grading hematotoxicity (WHO, 1979), which are summarized in Table 11-13. These have been particularly useful in establishing and communicating treatment strategies and guidelines for chemicals known to suppress hematopoiesis (cytoreductive oncolytic, immunosuppressive, and antiviral agents, etc) and for which this limiting toxicity is used to establish MTDs for individual patients.

Greater risk is acceptable with these drugs due to the life-threatening conditions they are used to treat. Similar risk–benefit decisions are also made regarding the use of drugs that cause blood dyscrasias in an idiosyncratic fashion, as previously discussed. Some are used to treat nonmalignant or life-threatening conditions, the risk of which is managed through rigorous laboratory monitoring. Examples include felbamate, ticlopidine, and clozapine, as discussed previously. Postmarketing surveillance plays a critical role in measuring the effectiveness of such monitoring.

Critical to the effectiveness of such surveillance by manufacturers and government regulatory agencies is the ability to detect a “signal,” such as that related to life-threatening idiosyncratic hematotoxicity. Over the past five years, regulatory agencies and drug monitoring centers have been developing computerized data mining methods to better identify reporting relationships in spontaneous reporting databases that have enabled and optimized such signal detection (Almenoff et al., 2005). This includes the use of screening algorithms and computer systems that efficiently signal higher-than-expected combinations of drugs and events in the FDA’s spontaneous reports database, such as aplastic anemia, agranulocytosis, and idiopathic thrombocytopenic purpura. Examples include the Multi-Item Gamma Poisson Shrinker (MGPS) program, which computes signal scores for pairs, and for higher-order (eg, triplet, quadruplet) combinations of xenobiotics and events that are significantly more frequent than their pairwise associations would predict (Szarfman et al., 2002). Such tools provide an objective and unprecedented systematic and simultaneous view of these large databases and alert government and manufacturers to critically important new safety signals that inform the toxicologist.