Hematologic Toxicity of Drug Therapy

Mechanisms of Hematologic Toxicities

AE from drugs may be expected (pharmacologic) or unpredictable (idiosyncratic). In addition, AE may be secondary to direct toxicity to cells or to immunologic reactions. In general, toxic reactions are insidious in onset, developing over weeks to months, although some toxic hematologic reactions may be seen earlier. With drug rechallenge, there is a latent period before onset of symptoms. On the other hand, immunologic reactions appear relatively early in the course of therapy. These reactions take days to weeks to develop, and, once established, often follow a more explosive course. Immunologic reactions recur promptly after re-exposure to even small doses of the causative agent.

Cytotoxic agents are examples of drugs that commonly and predictably produce cytopenias. In cancer therapy, these can be used as a marker of response and/or survival in patients undergoing treatment. Most cause cell death through apoptosis, although the pathways to that end vary by class of drug. In general, cytopenias from cytotoxic agents represent toxic reactions that are dose dependent and fairly predictable. A unique situation exists with azathioprine. Q63.1 The hematologic toxicities from this drug are determined not only by the dose of the drug, but also by the level of an enzyme, thiopurine methyltransferase (TPMT), that is important in the metabolism of the drug (see ‘Azathioprine’ section in this chapter).

Idiosyncratic reactions from drugs also may be either toxic or immunologic. They often are related to genetic variability in metabolism of drugs, particularly from common enzyme polymorphisms. Q63.2 For example, sulfonamides undergo N -acetylation in the liver, as a major part of their metabolism. Over 90% of patients demonstrating hypersensitivity to sulfonamides are slow acetylators. In these patients, metabolism of sulfonamides shifts to greater formation of nitroso and aryl-hydroxylamine metabolites, which can be directly toxic to cells of various organs, including bone marrow cells. Ascorbic acid deficiency has been demonstrated to potentiate this effect. A surrogate marker for sulfonamide antibiotic hypersensitivity has been proposed with a lymphocyte toxicity assay, however results were not found to be significantly different between drug-tolerant controls and patients who exhibited hypersensitivity reactions. A major limitation of this study was the small number of patients recruited, but it was suggested that the test was more sensitive in patients with more serious systemic organ involvement. This could potentially be used to predict which patients may be more prone to hypersensitivity reactions.

Q63.3 Idiosyncratic immunologic reactions account for certain adverse hematologic reactions. They are common mechanisms for drug-induced thrombocytopenia from noncytotoxic agents, such as quinidine, quinine, and heparin. In addition to the previously mentioned toxic mechanisms, immune mechanisms may play a role in hypersensitivity because of sulfonamides. Q63.2 The aniline structure (4-amino) of sulfonamide metabolites may alter antigen presentation by immune cells, and thus render a patient more susceptible to immunologic reactions from this class of drugs. Q63.3 In some cases of drug-induced agranulocytosis, including those associated with β-lactam or antithyroid medications, drug-dependent antineutrophil autoantibodies have been demonstrated. For most noncytotoxic drugs though, the mechanism causing hematologic toxicities is not well defined.

Timing of Hematologic Toxicity

The type and onset of various cytopenias from cytotoxic drugs depends on the category of drug. The cytotoxic drugs used in dermatology, especially the antimetabolites and alkylating agents, have leukopenia as their most common hematologic effect. The half-life of neutrophils is only 6 to 8 hours, compared with half-lives of 5 to 7 days for platelets and 60 days for red blood cells (RBC). A fall in the white blood cell (WBC) count often begins 5 to 14 days after initial administration of a cytotoxic agent, and recovery usually starts 7 to 10 days, after cessation of drug use. Drugs that are cell cycle-specific (such as antimetabolites) generally show earlier onset, and shorter duration, of myelosuppression. In contrast, cytopenias from busulfan or from the nitrosoureas (such as carmustine [BCNU]) occur later, typically starting 4 to 6 weeks after initiation of therapy. This applies to topical BCNU, as well as to systemic BCNU for mycosis fungoides.

The onset of idiosyncratic reactions is quite variable and depends on both the causative agent and the type of hematologic toxicity. These hypersensitivity reactions may be either metabolic or immunologic idiosyncratic reactions. The minimum time required for development of drug-induced immunocytopenias is about 6 days, if the drug is administered for the first time, but occurs within minutes to hours if the drug is readministered.

Agranulocytosis, hemolytic anemia, and thrombocytopenia are often acute reactions occurring in the first month of therapy. However, there are notable exceptions. Sulfasalazine- and dapsone-induced agranulocytosis generally occurs between weeks 3 and 12. Levamisole may produce autoantibodies against any of the blood cell lines after several months of therapy. Cytopenias from these drugs may persist for months, rather than the expected several weeks, after the offending drug is stopped.

Neutropenia and thrombocytopenia are unlikely to be related to a drug if they develop more than 1 month after cessation of the drug. However, aplastic anemia can show a different time course. It sometimes occurs after months to a year of drug use. It may also appear 4 months or occasionally longer after cessation of the drug, and often does not resolve despite stopping the suspected drug.

Prediction of Risk for Hematologic Toxicities

Determination of the exact risk ratio for hematologic toxicity from drugs used in dermatology is difficult; estimates can be inferred from knowledge of pharmacogenetics and from epidemiologic studies. Drug toxicities are typically either dose dependent or idiosyncratic. In general, cytotoxic agents produce toxicities that are dose dependent and follow a sigmoid-shaped curve. At low drug levels, myelosuppression may not occur. With increasing drug doses, cell death is proportional to drug concentration. With very high drug levels, the myelosuppressive effect plateaus.

The incidence of idiosyncratic reactions, whether caused by genetics or by random events, can be estimated from population studies. There are several large population studies of drug-related blood dyscrasias. Q63.4 The ability to extrapolate the relative risk of a drug’s toxicity from these reports depends on the following:

• 1.

  Completeness of the reported data;

• 2.

  Frequency of the various drugs’ use;

• 3.

  The population studied; and

• 4.

  The ability to choose one culprit drug in patients on multiple medications.

Studies that report the number of AE from a drug per population per year, ideally use case controls and account for the number of prescriptions written for a particular drug during the study period. If not, the study will likely underestimate the incidence of problems, from less frequently prescribed medications, and overestimate the incidence, from drugs that are frequently prescribed.

Agranulocytosis

Agranulocytosis is defined as a neutrophil count of less than 500/mm ³ (although significant caution is advised at levels 2 to 3 times this definition in dermatologic drug therapy) in the absence of significant anemia and thrombocytopenia. It is often accompanied by fever, pharyngitis, dysphagia, and oral ulcerations. Mortality, which is often secondary to septicemia, has decreased to 5% (0%–23%), likely from earlier recognition and the availability of new antibiotics and granulocyte colony-stimulating factors (G-CSF). Advanced age, severe infections, renal failure, and other serious comorbid conditions are associated with a poorer prognosis.

Q63.5 In up to 97% of cases, drugs are implicated as ‘probable’ or ‘possible’ causes of agranulocytosis, using the World Health Organization (WHO) definitions. The incidence is one to nine cases per million individuals per year, although an unexpectedly low incidence of 0.38 cases per 1 million per year was found in Latin American countries. In Hong Kong, the incidence ranged up to 15.4 cases per 1 million per year. In a recent pediatric study, incidence of agranulocytosis secondary to medications was 3.92 cases per 10,000 pediatric patients.

In a review of population-based studies on the incidence of drug-induced agranulocytosis, covering years 1960 to 2001, Garbe noted that the drugs presenting the greatest risk were antithyroid drugs, dipyrone, macrolides, trimethoprim-sulfamethoxazole (TMP-SMX), and anticonvulsant drugs ( Table 63.1 ). Very similar data were obtained from Saskatchewan, where antithyroid drugs, sulfasalazine, chlorpromazine, carbamazepine, TMP-SMX, and β-lactam antibiotics showed the strongest association with agranulocytosis. Andersohn reviewed case reports of drugs definitely or probably related to agranulocytosis. Drugs for which more than 10 reports were available included dapsone, penicillin, rituximab, sulfasalazine, and anticonvulsants, antithyroid and antiplatelet agents. A study in Spain found the annual incidence of drug-induced agranulocytosis to be 3.46 cases/1 million patient-years. Ticlopidine, calcium dobesilate, antithyroid drugs, dipyrone, and spironolactone were most strongly associated with agranulocytosis. Other drugs with a significant risk included sulfonamides, phenytoin, β-lactam antibiotics, erythromycin, and diclofenac. In Hong Kong, a total of 59 drugs were identified to have an association with agranulocytosis. Antithyroid drugs caused the majority, with carbimazole being the most commonly implicated drug, followed by antimicrobials and anticonvulsants. Human leucocyte antigen (HLA) alleles have been shown to be associated with drug-induced agranulocytosis. For example two alleles, HLA-B∗ 38:02 and HLA-DRB1∗ 08:03, seemed to be associated with antithyroid drugs. Huerta and colleagues estimated the risk of blood dyscrasia in patients on the full spectrum of antibiotics to be 3.3/100,000 person-years. The relative risk increased if the patient was over 60 years of age, was taking phenothiazines, or receiving more than one antibiotic simultaneously.

Aplastic Anemia (Pancytopenia)

Aplastic anemia results from a severe decrease in bone marrow production of blood cells, resulting in pancytopenia. In general, the neutrophil count is less than 1500/mm ³ , the platelet count less than 50,000/mm ³ , and the hemoglobin concentration less than 10 g/dL. The mortality rate is 46%. The International Agranulocytosis and Aplastic Anemia Study reported that 27% of their cases of aplastic anemia were likely drug-related. Penicillamine, gold, and carbamazepine were most commonly implicated ( Table 63.2 ). A French group compared drug use in patients with aplastic anemia to hospitalized patients and neighborhood control groups. The use of gold, d -penicillamine, or colchicine was associated with higher risk of aplastic anemia. The Swedish Blood Dyscrasia study also found that 25% of their cases of aplastic anemia were probably drug associated. TMP-SMX had a reported risk of 13 cases/10 ⁶ patient-years. However, concomitant viral disease might have played a role in the development of aplasia in some patients. The risk of aplastic anemia from TMP-SMX was much lower in other studies (1.4 cases/10 ⁶ users per 5-month period (see ‘Sulfonamides’ section). Other drugs used in dermatology and associated with risk of aplastic anemia include nonsteroidal anti-inflammatory drugs (NSAID), dapsone, β-lactam antibiotics, and chloroquine.

The larger studies cited addressing drug causation of aplastic anemia are from decades ago. Interestingly, the most recent review of aplastic anemia mentions drugs, only as causation of chemotherapy related and usually reversible pancytopenia. In addition, they emphasized increasing evidence for mutations leading to genetic predisposition to aplastic anemia and the induction of probable immune mechanisms of cytopenia, triggered by chemical, viruses, drugs, or antigens.

Thrombocytopenia

Thrombocytopenia is defined as a platelet count of 100,000/mm ³ . Reviews of published case reports looked at medications implicated in five or more reports of drug-induced thrombocytopenia ( Table 63.3 ). Heparin, quinidine, gold, certain antibiotics (linezolid, rifampin, sulfonamides, vancomycin), anticonvulsants, cimetidine, certain analgesics (acetaminophen, diclofenac, naproxen), and thiazide diuretics were the most commonly implicated drugs. A Dutch case-control study looking at nonheparin-related drug-induced thrombocytopenia, between 1990 and 2002, found an odds ratio of 7.8 for β-lactam antibiotic users and a odds ratio of 5.7 for those taking TMP-SMX. In the Swedish registry study, TMP-SMX had a risk of 96 cases of thrombocytopenia per 10 ⁶ patient-years. Concomitant viral illness might have been a confounding factor in some cases. Severe thrombocytopenia (<20,000/mm ³ ) warrants aggressive treatment with platelet transfusions to prevent intracranial bleeding. Recovery usually starts 1 to 2 days after cessation of the causative drug.

Neoplasia

The potential for induction of neoplasia is of special concern. For most cytotoxic drugs, hematologic malignancies and cutaneous carcinomas are more likely than are other solid tumors. The incidence of malignancy also depends on the agent used. Q63.6 Therapy-related myeloid neoplasms include acute myeloid leukemia and myelodysplastic syndrome. Hematologic malignancies are seen less commonly with antimetabolites (methotrexate, azathioprine, 6-mercaptopurine [6-MP]) than with alkylating agents (cyclophosphamide, chlorambucil), because the former do not directly affect the structure of deoxyribonucleic acid (DNA). However, a recent multicenter, case-control study showed a seven-fold risk of proven therapy-related myeloid neoplasms with azathioprine. Rare cases of myelodysplastic syndrome after treatment with cyclophosphamide have also been reported. In some instances, it is difficult to calculate the relative risk for development of malignancy because data on the potential association between a drug and neoplasm consist predominantly of individual case reports. Also, different patient populations may have inherently higher risks of developing malignancies, despite using the same drug at similar doses. For instance, a study of psoriasis patients found that they have a 2.95 relative risk (95% confidence interval [CI], 1.83–4.76) for developing lymphoma. This risk changed little when patients who used methotrexate were excluded.

Q63.7 Patients who develop a lymphoproliferative disorder while on immunosuppressive therapy may show regression of disease on withdrawal of the immunosuppressive drug. This is especially true for lymphomas demonstrating the Epstein–Barr viral genome. Therefore a period of watchful waiting after cessation of the immunosuppressive drug(s) is appropriate in this setting, before deciding on the need for chemotherapy. However, acute leukemia, developing during or after immunosuppressive or cytotoxic drug use, virtually never shows spontaneous regression. These leukemias are often associated with cytogenetic abnormalities of chromosomes 5 or 7 and have a very poor prognosis. Development of solid tumors in association with drug use is discussed elsewhere in this book. (See Chapter 64 .)

Methotrexate

Azathioprine

Cyclophosphamide

Chlorambucil