Bruce A. Chabner
Cancer treatment requires the cooperative efforts of multiple medical specialties. Although surgeons traditionally have been the first specialists to treat the cancer patient, newer modalities have created important roles for the radiotherapist and medical oncologist in the initial management of cancer patients, and responsibility for care of the majority of patients with metastatic cancer is in the hands of these specialists. The array of alternatives for the treatment of cancer is constantly expanding. With the demonstration of effectiveness of new drugs and new biologics, and with the evolution of more effective strategies for integrating chemotherapy, surgery, and radiation, the development of a treatment plan becomes increasingly complex. The plan must be based on a thorough understanding of the potential for beneficial response and an awareness of the acute and later toxicities of each component of the treatment regimen.
As a general rule, the medical oncologist is urged to use standard regimens as described in the Physician Data Query (PDQ) system of the National Cancer Institute (NCI).a PDQ contains information on state-of-the-art treatments for each pathologic type of cancer, as well as a listing of experimental protocols for each disease. An important alternative to “standard” therapy is the clinical trial, which should be considered for every eligible patient. Such trials offer alternative treatment that is thought by a panel of experts to be at least as effective as the recognized standard of care. In phase III (randomized) trials, a standard regimen is compared with a new one that may represent an improvement. With either choice, standard therapy or a clinical trial, the medical oncologist must understand the potential benefits and risks of using specific drugs or combinations of drugs, or combinations of drugs and biologics, often integrated with surgery and irradiation. All these considerations enter into the choice of a treatment plan. Steps in the treatment decision-making process are discussed in this chapter to provide the reader with an understanding of the overall role of drugs in cancer treatment.
DETERMINANTS OF TREATMENT PLANNING
The first and primary determinant of treatment is the histologic diagnosis. Malignant neoplasms occur in over 100 different pathologic forms, each with a characteristic natural history, pattern of progression, and responsiveness to treatment. Thus, the histologic diagnosis, usually made by surgical biopsy or excision of a primary tumor, is of critical importance as a first step in treatment planning. The clinical oncologist must be alert to the possibility of atypical presentations of treatable and even curable tumors, such as germ cell tumors of the testis and breast cancer, and must ask for special immunohistologic or molecular tests to rule in or rule out a potentially curable tumor type. For example, germ cell tumors may arise occasionally in the thoracic or abdominal cavity in the absence of a primary testicular tumor; still, these unusual presentations retain an excellent response to appropriate chemotherapy. Treatment in cases lacking a precise histologic diagnosis is usually directed against the most responsive tumor type within the realm of possible diagnoses, such as testicular carcinoma in a patient with poorly differentiated carcinoma of uncertain origin, and occasionally produces durable responses.1
In certain cases—for example, lung carcinoma or the non-Hodgkin's lymphomas—accurate subtyping of tumors is important because the subtypes of these diseases have different patterns of clinical response to treatment. Subtyping may require the characterization of cell surface immunologic markers (e.g., to distinguish T-cell and B-cell lymphomas), the identification of specific intracellular secretory granules or enzymatic markers, such as dopa decarboxylase in small cell carcinoma of the lung, or the S100 antigen to rule out malignant melanoma. Molecular or genetic analysis may reveal important prognostic information for subtyping leukemias, lymphomas, and lung cancers.2 Epidermal growth factor receptor mutations identify a unique subgroup of patients with non-small cell lung cancer highly responsive to the epidermal growth factor receptor (EGFR) inhibitor (Iressa).3 The essential point is that the precise identity of a tumor is the single most important determinant of treatment choice and patient management. In premenopausal women with stage I breast cancer (less than 2 cm primary, node negative), adverse tumor features, such as a high S-phase (DNA synthetic phase) fraction, absence of estrogen or progesterone receptors, or high expression of the c-erbB-2/HER-2-neu oncogene, may guide the selection of drugs for adjuvant therapy, such as dose-intensive adjuvant chemotherapy and herceptin. Her-2-neu positive metastatic disease is also best treated with herceptin and chemotherapy.4
While gene array studies have provided significant insight into subgroups of cancer with favorable or unfavorable prognosis, or with a high propensity for metastasis,4 they have not yet demonstrated the ability to guide specific choice of therapies. Gene profiles that interrogate drug resistance markers for cytotoxic or hormonal therapy5, 6, 7 may provide this guidance in the future. Molecular analysis of target molecules such as receptor tyrosine kinases and signaling pathways will undoubtedly be useful in developing a functional classification of human tumors that will in guide the selection of therapy for most kinds of cancer.8
The next step in treatment planning is to determine the extent of disease, and specifically whether the tumor is curable by local treatment measures such as surgery or radiation therapy. The process of determining the extent of disease is termed staging and plays an important role in making therapeutic choices for diseases that are responsive to multiple types of treatment. For example, non-Hodgkin's lymphomas with “indolent” histology are curable with radiotherapy in a majority of cases when the tumor is confined to a single lymph node region (stage I), but are rarely curable, even with aggressive early chemotherapy, when more extensive lymph node involvement or dissemination to extranodal sites is present. Immediate treatment of stage I disease with radiation is indicated, whereas, paradoxically, no immediate therapy may be indicated for patients with advanced disease.
Individualizing Treatment Choice
The choice of specific therapies depends on histology, molecular and immunologic subtype, stage, and an additional factor, the patient's probable tolerance for the side effects of the various possible treatments. Although chemotherapy cures a substantial fraction of patients with diffuse large cell lymphoma, not all patients with this diagnosis are suitable candidates for intensive treatment. Severely debilitated patients and those with underlying medical problems—for example, heart disease, renal failure, diabetes, or chronic obstructive pulmonary disease—might well suffer severely disabling or fatal complications from potentially curative regimens, as indicated in Table 1.1. In such cases, the physician and patient may choose a less toxic, palliative regimen. The ultimate decision must be based on a thorough understanding of the disease process under consideration, the clinical pharmacology of the drugs in question, and the potential benefits and risks of alternative forms of treatment, such as chemotherapy, radiotherapy, or surgery.
Finally, possessing the information about histology, stage, and other tumor-related variables and about the patient's age and baseline health, the oncologist must decide whether a realistic opportunity exists for curative treatment. A decision to treat with curative intent demands a high degree of adherence to drug dosage and scheduling requirements, as specified in the standard or experimental regimen, and an acceptance of treatment-related toxicity. When cure is not a realistic expectation, a decision to treat must be based on an expectation for prolongation of the patient's life or an improvement in the quality of life. In these cases, treatment-related side effects may be minimized by dosage adjustments or treatment delays, when necessary, but at the cost of antitumor efficacy. When the probability for benefit is low, chemotherapy should be offered only after frank and thorough discussion of the likely outcome. In such cases, experimental phase I or phase II drugs may be a more attractive alternative in the setting of a clinical trial.
TABLE 1.1 TOXICITY OF CHOP REGIMEN FOR TREATING DIFFUSE LARGE CELL LYMPHOMA
DRUGS IN CANCER TREATMENT
Drugs are now used at some time during the course of the illness of most cancer patients. Cytotoxic drugs can cure some disseminated cancers (Table 1.2) and can be effective in decreasing tumor volume, alleviating symptoms, and even prolonging life in many other types of metastatic cancer. Adjuvant chemotherapy regimens are used in patients who have had primary tumors resected and who, although possibly cured by surgery, are at significant risk of recurrence. Adjuvant therapy has been shown in randomized trials to delay tumor recurrence and prolong survival in patients with breast cancer, colorectal cancer, non-small cell lung cancer, osteosarcoma, and other tumors. Neoadjuvant chemotherapy is used to reduce the bulk of primary tumors before surgical resection or irradiation of locally extensive head and neck carcinomas, esophageal cancer, non-small cell lung cancer, osteosarcoma and soft tissue sarcomas, bladder cancer, and locally advanced breast cancer. This approach can improve the probability of total surgical resection, decrease local recurrence, and allow organ preservation. Furthermore, the initial clinical response of the tumor mass can serve as an indication to continue therapy after surgery.
The design of drug treatment regimens is based on a number of considerations. These include (a) prior knowledge of the responsiveness of the pathologic category of tumor to specific drugs, (b) an understanding of the biochemical mechanisms of the drugs' cytotoxic activity as well as the mechanisms of resistance to the drugs, and (c) knowledge of the drugs' pharmacokinetic behavior and of patterns of normal organ toxicity. Some chemotherapy regimens have been designed to minimize emergence of drug resistance, based on the predictions of theoretical models of drug resistance. The biologic and pharmacokinetic features of individual drugs are considered in detail in succeeding chapters, but the impact of these and other factors, such as cell kinetics and drug mechanism of action and mechanism of resistance, on trial design is reviewed briefly at this juncture to provide a framework for understanding the use of individual agents.
Kinetic Basis of Drug Therapy
The objective of cancer treatment is to reduce the tumor cell population to zero cells. Chemotherapy experiments using rapidly growing transplanted tumors in mice have established the validity of the fractional cell kill hypothesis, which states that a given drug concentration applied for a defined time period will kill a constant fraction of the cell population, independent of the absolute number of cells. Regrowth of tumor occurs during the drug-free interval between cycles. Thus, each treatment cycle kills a specific fraction of the remaining cells. The results of treatment are a direct function of (a) the dose of drug administered and (b) the number and frequency of repetitions of treatment.
Most current chemotherapy regimens are based on cytokinetic considerations and use cycles of intensive therapy repeated as frequently as allowed by the tolerance of dose-limiting tissues, such as bone marrow or gastrointestinal tract. The object of these cycles is to reduce the absolute number of remaining tumor cells to zero (or less than one) through the multiplicative effect of successive fractional cell kills. (For example, given 99% cell kill per cycle, a tumor burden of 1011 cells will be reduced to less than one cell with six cycles of treatment: [1011 cells] × [0.01]6 <1.)
The fractional cell kill hypothesis was defined initially in animal models of leukemia and was applied most successfully in human leukemia and lymphoma.9 The fundamental assumption of constant fractional cell kill per cycle with constant dosing is unlikely to be valid for the more heterogeneous, slowly growing solid tumors in humans. Most clinical neoplasms are recognized at a stage of decelerating growth, which may be due to poor tumor vascularity with resulting hypoxia or poor nutrient supply, or to other unidentified factors. These tumors contain a high fraction of slowly dividing or noncycling cells (termed G0 cells). Antineoplastic agents, particularly the antimetabolites and antitumor antibiotics, are most effective against rapidly dividing cells and some are phase-specific (i.e., most effective in killing cells in a specific phase in the cell cycle). The initial kinetic features of a large, poorly vascularized tumor are unfavorable for treatment with most antimetabolites, which kill most effectively in the S phase. Drugs that attack DNA integrity, such as alkylators and adduct forming platinum derivatives retain activity against nondividing or slowly dividing cells, and are often used to reduce tumor bulk. An initial reduction in cell numbers produced by surgery, radiotherapy, or non–cell cycle-specific drugs improves blood flow, pushes the slowly dividing cells into more rapid cell division, and may recruit nondividing cells into the cell cycle, where they become increasingly susceptible to therapy with cell cycle-specific agents. Thus, an initially slowly responding tumor may become more responsive to therapy after surgical debulking or with continued treatment, and fractional cell kill may actually increase with sequential courses of treatment.
TABLE 1.2 CURABILITY OF DISSEMINATED CANCER WITH DRUGS
Biochemical heterogeneity of human tumors introduces additional complexity to the simple hypothesis that multiple cycles of fractional cell kill translate into tumor cure. Isoenzyme typing and karyotypic analysis of tumors demonstrate that most human tumors studied thus far have evolved clonally from a single malignant cell.10 Techniques for in vitro cloning of solid tumors have shown, however, that this original homogeneity does not persist during later stages of tumor growth; in fact, both experimental and human tumors are composed of cell types with differing biochemical, morphologic, and drug-response characteristics.11 This heterogeneity results from the inherent genetic instability of malignant cells. Indeed, mutations in cell-cycle checkpoint control genes, such as p53, and in DNA repair genes, such as the MSH genes in familial colon cancer, may be the initial event in malignant transformation of many tumor types, establishing a fundamentally mutable clone from which diverse subclones evolve. Thus, gene amplifications, deletions, or other alterations of genes coding for target proteins that control drug response and cell cycle lead to heterogeneity of the tumor cell population and probably account for outgrowth of resistant tumor cells during relapse of formerly sensitive tumors. This has been clearly demonstrated in the isolation of Gleevec-resistant cells from selected patients with chronic myelogenous leukemia prior to treatment.12 When cells are subjected to the selective pressure of drug treatment, sensitive tumor cells are destroyed, but subpopulations of resistant cells survive and proliferate. With the possible exceptions of treatment of chronic myelogenous leukemia with Gleevec, gestational choriocarcinoma treated with methotrexate, cyclophosphamide treatment for African Burkitt lymphoma, and cladribine treatment for hairy cell leukemia, single-agent chemotherapy has rarely produced long-term survival or cure of advanced malignancies. The most successful drug treatment regimens have combined multiple agents with different mechanisms of action.
Prediction of Drug Response to Individual Agents
The selection of drugs for treating specific types of cancer is largely based on the results of previous clinical trials and is often empirical. To avoid the needless toxicity of ineffective agents, especially in diseases with only modest rates of response, predicting sensitivity for the specific tumor and patient at hand would be desirable. Various experimental systems for testing tumor cells or tumor fragments for response to panels of drugs have been studied intensively in the hope that they would accurately predict response in patients. Although some of these tests have been accurate in predicting resistance to various drugs in populations of highly resistant patients, with the possible exception of cell-based assays for resistance in acute lymphoblastic leukemia7 and ovarian cancer,13 limited data exist to justify their routine use.
Biochemical tests for the presence of a specific enzyme critical to the response of an antimetabolite (such as deoxycytidine kinase for cytosine arabinoside, or thymidylate synthase for the fluoropyrimidines) or the presence of a specific cytoplasmic receptor, such as the estrogen receptor for hormonal therapy of breast cancer, can discriminate responders from nonresponders. A few of these tests—most notably the test for estrogen receptor or HER-2-neu expression in breast cancer—have become cornerstones for therapeutic decision-making; other tests offer considerable promise. A study of non-small cell lung cancer showed that expression of components of the DNA repair pathway correlated with patient survival after platinum-based treatment.14 Similarly, high concentrations of dihydrofolate reductase have been associated with resistance to methotrexate, as is a failure to transport or polyglutamate the drug.15, 16High levels of the DNA repair enzyme O6-alkyl-guanine alkyl transferase predict resistance to nitrosoureas, dacarbazine, and temozolomide, which damage DNA by alkylating the O6 position of guanine.17 The latter test may prove useful in predicting which patients will benefit from pretreatment with O6-benzyl guanine, an irreversible inhibitor of the alkyl transferase enzyme that is in clinical testing. Mutations in mismatch DNA repair are associated experimentally with cisplatin resistance.18 None of these molecular biochemical tests have been studied prospectively in a sizable patient population to prove their value in selecting routine treatment with cytotoxic drugs. In each case, other potentially important changes are known to occur in experimental examples of resistance to the various drugs.
Molecularly targeted drug discovery offers the hope of identifying new drugs tailored specifically to the mutations critical for malignant transformation and tumor progression, with limited toxicity to normal tissues lacking the molecular target. One such target, the tyrosine kinase, results from the bcr-abl translocation in chronic myelocytic leukemia.18 Gleevec, an inhibitor of the BCR-abl kinase, has striking activity in chronic and blastic phases of chronic myelocytic leukemia.19 Because it also inhibits the c-kit tyrosine kinase, it was used against gastrointestinal stromal tumors, which exhibit high levels of expression of this receptor; it proved to be particularly active against gastrointestinal stromal tumors that have mutated and constitutively activated c-kit.20
The epidermal growth factor receptor has also become the object of targeted drug discovery. Antibodies or small molecules that bind to and inhibit this receptor show potent tumor inhibitory activity against epithelial tumors that overexpress the EGFR. In addition Erbitux, a monoclonal antibody against EGFR, shows synergy with irinotecan against relapsed colon cancer.21 The synergy is believed to result from blockade of a critical growth factor signal, an event that lowers the tumor cell threshold for apoptosis. As mentioned previously, mutations and deletions that activate EGFR predict for response to Iressa.3 Other interactions between cytotoxic drugs and signal pathway inhibitors, leading to synergistic cell kill, have been described for antiangiogenic drugs against colon cancer and the HER-2-neu receptor inhibitor, herceptin, with taxanes or anthracyclines against breast cancer. This general principle of drug-signal inhibitor interaction will undoubtedly be further exploited in treating cancers that present biologic targets related to overexpressed growth factor pathways. The reader is referred to the chapter on molecularly targeted therapies for a more detailed discussion of this subject.
PHARMACOKINETIC DETERMINANTS OF RESPONSE
Although the outcome of cancer chemotherapy depends in large part on the inherent sensitivity of the specific tumor being treated, the chances for success, even in patients with sensitive tumors, can be compromised by failure to consider important pharmacokinetic factors such as drug absorption, metabolism, and elimination in designing protocols that determine the dose, schedule, and route of drug administration.
Not only may protocol design affect pharmacokinetics and response, but even among patients with apparently normal hepatic and renal function, considerable variability is seen in peak drug concentration, area under the concentration × time curve (AUC).22 The origin of this variability is uncertain. Clearly, pharmacogenetics (polymorphisms in expression of drug-metabolizing enzymes) plays an important role in determining the rate of elimination and thus the toxicity of some drugs, including irinotecan hydrochloride (by glucuronyl transferases), 6-mercaptopurine (by thiopurine methyl transferase), and 5-fluorouracil (by dihydropyrimidine dehydrogenase).23 Genetic polymorphisms may also influence the susceptibility of tumors to cytotoxic and targeted drugs. In addition, differences in hepatic P-450 isoenzyme activity, protein binding of drug, and age-related changes in renal tubular function all contribute to this variability. The fact remains that most pharmacokinetic studies show at least a long range of drug concentration and AUC for a given dose of drug.
Pharmacokinetic factors are important not only in designing general protocol but also in determining specific modifications of dosage in individual patients. Dosage may be increased or decreased based on observed patterns of toxicity or lack of same and, in some cases, may be based on direct drug concentration measurements. Renal or hepatic dysfunction leads to delayed drug elimination, which sometimes results in overwhelming toxicity. To avoid such toxicity, dosages of certain agents must be modified based on estimates of renal or hepatic function.
Interindividual variations are not predictable solely on the basis of renal or hepatic function, however, and direct measurement of plasma drug concentrations can provide a better guide for dosage adjustment to ensure adequate and safe drug exposure, as shown by studies of maintenance therapy in children with acute lymphocytic leukemia and who receive methotrexate and 6-mercaptopurine.24 One important source of the interindividual variability in pharmacokinetics is the variable oral absorption of a number of agents, including hexamethylmelamine, etoposide, methotrexate (in doses of more than 15 mg/m2), 6-mercaptopurine, 5-fluorouracil (5-FU), busulfan in high-dose therapy, and phenylalanine mustard, and even the targeted drug, Gleevec. This problem has been documented by pharmacokinetic studies, and may lead to a poor therapeutic result. Drug concentration monitoring may provide a valuable guide to delayed elimination of agents such as methotrexate.24 Monitoring allows dosage adjustment of the cytotoxic drug in later cycles and early institution or prolongation of rescue procedures. Reliable assays are available for many antineoplastic agents; most assays use high-pressure liquid chromatography, a technique available in most cancer centers. A few, such as the assays for methotrexate, have established importance as guides to the prediction of drug toxicity in high-dose therapy (Table 1.3). The utility of various assays is indicated in the discussion of individual agents in subsequent chapters.
Rationale for Combination Chemotherapy
Although the first effective drugs for treating cancer were brought to clinical trial in the 1940s, initial therapeutic results were disappointing. Impressive regressions of acute lymphocytic leukemia and adult lymphomas were obtained with single agents such as nitrogen mustard, antifolates, corticosteroids, and the vinca alkaloids, but responses were only partial and of short duration. When complete remissions were obtained, as in acute lymphocytic leukemia, they lasted less than 9 months, and relapse was associated with resistance to the original drug. The introduction of cyclic combination chemotherapy for acute lymphocytic leukemia of childhood in the late 1950s marked a turning point in the effective treatment of neoplastic disease. Such combinations are now a standard component of most treatment strategies for advanced cancer. The superior results of combination chemotherapy compared with single-agent treatment derive from the following considerations. First, initial resistance to any given single agent is frequent, even in the most responsive tumors; for example, in patients with Hodgkin's disease, the complete response rates to alkylating agents or procarbazine do not exceed 20%, and virtually all patients relapse. Second, initially responsive tumors rapidly acquire resistance after drug exposure, probably owing to selection of preexisting resistant tumor cells from a heterogeneous tumor cell population. Some anticancer drugs themselves increase the rate of mutation to resistance in experimental studies, as does hypoxia.25 The use of multiple agents, each with cytotoxic activity in the disease under consideration but with different mechanisms of action, allows independent cell killing by each agent. Cells resistant to one agent might still be sensitive to the other drugs in the regimen.
TABLE 1.3 DRUG MONITORING IN CANCER THERAPY
Patterns of cross-resistance must be taken into consideration in formulating drug combinations. Resistance to many agents may result from unique and specific mutations, for example as may occur in the target enzymes of antimetabolites or targeted agents such as Gleevec.12 Mutations that alter binding of inhibitors of topoisomerase II, an enzyme that promotes DNA strand breaks in the presence of anthracyclines and epipodophyllotoxins, may mediate resistance to each of these agents.26 In other cases, a single mutational change may lead to multidrug resistance. Table 1.4 describes cross-resistance patterns for some of the well-defined mechanisms of multidrug resistance. The most thoroughly studied and undoubtedly one of the more important mechanisms of multidrug resistance is increased expression of the MDR-1 gene. This gene codes for the P-170 membrane glycoprotein, which promotes the efflux of vinca alkaloids, anthracyclines, taxanes, actinomycin D, epipodophyllotoxins, and other natural products. This protein occurs constitutively in many normal tissues, including epithelial cells of the kidney, large bowel, and adrenal gland,27 and has been identified in tumors derived from these tissues, as well as in posttreatment lymphomas, leukemias, non-small cell lung cancer, multiple myeloma, and other cancers.28 P-170–mediated resistance results from decreased intracellular drug levels and can be reversed experimentally by administration of calcium-channel blockers, amiodarone, quinidine, and derivatives of cyclosporine, as well as by a variety of aprotic polar solvents. At this time, evidence suggests that P-170 contributes to clinical drug resistance in multiple myeloma, non-Hodgkin's lymphomas, pediatric sarcomas, and acute nonlymphocytic leukemia.27 Many clinical trials investigating the use of agents reversing multidrug resistance have been initiated, but the results of these trials are inconclusive. Many MDR inhibitors also inhibit hepatic clearance of doxorubicin hydrochloride, which significantly complicates the design and interpretation of these studies.28
TABLE 1.4 MECHANISMS OF RESISTANCE
A second mechanism for multidrug resistance involves the family of multidrug resistance proteins (MRPs), which, in experimental tumors, promotes drug efflux and confers resistance to anthracyclines, etoposide, and vinca alkaloids. Members of the MRP family may also mediate efflux of methotrexate, 6-mercaptopurine, camptothecin derivatives, and others.29 Multiple members of the MRP gene family have been identified, and, again, their role in clinical drug resistance remains uncertain. The MRP family of genes is widely expressed in epithelial tumors,27 and their potential for mediating multiagent resistance deserves further study.
Finally, classic alkylating agents (cyclophosphamide, melphalan hydrochloride, nitrogen mustard) may share cross-resistance related to enhanced DNA repair mediated by nucleotide excision repair enzymes and by increased levels of intracellular nucleophilic thiols, such as glutathione. There is clinical evidence that increased expression of nucleotide excision repair components correlates with a poor outcome in ovarian cancer treated with platinum-based regimens.30 Not all alkylating agents share cross-resistance. As mentioned earlier, resistance to the nitrosourea, procarbazine, and dacarbazine classes of alkylators is mediated by increased levels of a different enzyme, O6-alkyl- guanine alkyl transferase. Thus, there is a clear rationale for combining different alkylators in a single regimen. Undoubtedly, most resistant tumors have acquired a variety of mechanisms for avoiding the toxic effects of chemotherapy.
The acquisition of drug resistance is widely believed to be the product of random mutations in a tumor cell population. A corollary to this hypothesis is the concept that the probability of de novo drug resistance in any tumor population increases with increasing number of cells and number of cell divisions. More than 20 years ago, Goldie et al.31 proposed a mathematical model based on the random-mutation hypothesis. It suggested several important considerations in protocol design to minimize treatment failure due to acquired drug resistance: (a) Treatment should begin as early as possible when the malignant cell population is at its smallest. (b) To avoid selection of doubly resistant mutants by sequential chemotherapy, multiple mutually non–cross-resistant drugs should be used together. (c) To achieve maximal kill of both sensitive and moderately resistant cells, cytotoxic drugs should be administered as frequently as possible and in doses well above the minimally cytotoxic doses. The Goldie-Coldman model serves to explain the success of combination therapies against hematologic malignancies such as non-Hodgkin's lymphoma and Hodgkin's disease.32 However, randomized trials comparing new and more elaborate regimens, based on the Goldie-Coldman hypothesis, with older, empirically designed four-drug regimens have failed to demonstrate any improvement in the cure rates of either non-Hodgkin's lymphoma33 or Hodgkin's disease.34 Although these studies do not negate the basic tenants of Goldie-Coldman, they do suggest that the assumptions made to allow the formal mathematical modeling were overly simple. For instance, the model assumes that resistance develops to individual agents, one at a time, and thus does not account for multidrug resistance patterns. Multidrug resistance and broad resistance to apoptosis as conferred by inactivation of p53 or overexpression of bcl-2 are not considered. It is my opinion, unconfirmed, that a “model” can not substitute for the design of drug combinations based on a precise understanding of the molecular basis for drug resistance for each agent in the regimen.12
A third consideration supports combination chemotherapy. If drugs have nonoverlapping patterns of normal organ toxicity, each can be used in full dosage, and the effectiveness of each agent will be fully maintained in the combination. Drugs such as vincristine sulfate, prednisone, bleomycin sulfate, hexamethylmelamine, L-asparaginase, high-dose methotrexate/leucovorin calcium, and biologics, all lacking bone marrow toxicity, are particularly valuable in combination with traditional myelosuppressive agents. Based on these principles, curative combinations have been devised for diseases that are not curable with single-agent treatment, including acute lymphocytic leukemia (vincristine, prednisone, doxorubicin, and L-asparaginase), Hodgkin's disease (mechlorethamine, Oncovin [vincristine], procarbazine, and prednisone [MOPP] and Adriamycin (doxorubicin), bleomycin, vinblastine, and dacarbazine [ABVD]), diffuse large cell lymphoma (the combination of cyclophosphamide, doxorubicin, vincristine, and prednisone) and testicular carcinoma (bleomycin, cisplatin, and vinblastine or etoposide).
Schedule Development in Combination Therapy: Kinetic and Biochemical Considerations
The detailed scheduling of drugs in multidrug regimens was based initially on both practical and theoretical considerations. Intermittent cycles of treatment were used to allow periods of recovery of host bone marrow, gastrointestinal tract, and immune function, with the expectation that recovery of the tumor cell population would be slower than that of the injured normal tissues. This strategy allowed retreatment with full therapeutic doses as frequently as possible in keeping with the fractional cell kill hypothesis. A commonly used strategy is to incorporate myelotoxic agents on day 1 of each cycle, while delivering nonmyelosuppressive agents, such as bleomycin, vincristine, prednisone, or high-dose methotrexate with leucovorin rescue, in the period of bone marrow suppression (e.g., on day 8 of a 21-day cycle) to provide continuous suppression of tumor growth while allowing maximum time for marrow recovery. High-dose methotrexate with leucovorin rescue has proved to be particularly useful in this capacity during the “off period” because of its minimal effect on white blood cell and platelet counts.
Cytokinetic considerations also influence the specific sequencing of drugs in combination regimens. S-phase–specific drugs, such as cytosine arabinoside and methotrexate, are capable of killing cells only when they are present during the period of DNA synthesis. Experimentally, these agents are most effective if administered during the period of rapid recovery of DNA synthesis that follows a period of suppression of DNA synthesis. Thus, an initial phase of cytoreduction with drugs that are not cell cycle-phase–specific, such as the bifunctional alkylating agents, reduces tumor bulk and recruits slowly dividing cells into active DNA synthesis. These drugs can then be followed within the same cycle of treatment by cell cycle-phase–specific agents such as methotrexate or the fluoropyrimidines, which kill cells during periods of DNA synthesis.
Although most of the common anticancer regimens use intermittent bolus delivery of drugs, in recent years advantages of constant-infusion chemotherapy have been suggested. Early clinical trials have suggested improved therapeutic ratios for several drugs, including doxorubicin, 5- FU, etoposide, ifosfamide, and cytosine arabinoside.35, 36 Constant exposure of cells to cell cycle-phase–specific agents such as antimetabolites or cytosine arabinoside allows a greater fraction of the tumor cell population to cycle through the sensitive phase than is likely to occur with intermittent bolus therapy. The orally administered fluoropyrimidine, capecitabine, provides prolonged, low concentration exposure of tumors to the active metabolite of 5-fluorouracil, FdUMP. An additional consideration is the experimental evidence suggesting that constant exposure of agents such as natural products, resistance to which is mediated by p-170, may overwhelm the pump, which allows the killing of otherwise resistant cells.37 Finally, infusional regimens may change the toxicity of cancer drugs. For example, the cardiotoxicity associated with anthracyclines is more closely correlated with the peak concentration than with AUC. Liposomal preparations of doxorubicin and daunorubicin provide the same advantage of prolonged exposure to drug, low peak concentrations, and decreased cardiotoxicity as does continuous intravenous infusion of the same agents.38
Additional Considerations in Combination Chemotherapy: Drug Resistance and Drug Interactions with Targeted, Anti-apoptotic Agents
Drug resistance, either apparent with initial treatment or emerging at the time of relapse after an initial response, inevitably occurs in all but the few cancer types that are curable with chemotherapy (Table 1.2). A panoply of potential mechanisms conferring resistance to cancer cells has been described (Table 1.4). As has become increasingly obvious, resistance is, in most cases, a complex process involving multiple mechanisms that may emerge in parallel or in series. With the appreciation of this complexity has come increasing skepticism that strategies aimed at reversal of discrete pathways conferring resistance to specific drugs or drug classes will have a major impact on the treatment of the common solid tumors. On the other hand, recent research suggests that a very common process conferring resistance to many, if not virtually all, chemotherapeutic agents is the suppression or inactivation of apoptosis, or programmed cell death. Inactivation of apoptotic mechanisms can be mediated by inactivation of the p53 tumor suppressor gene (also termed a death pathway gene) or by inappropriate overexpression of genes that suppress apoptosis. The p53 gene is mutated in almost 50% of human cancers at diagnosis, and in an even higher percentage of tumors at the time of emergence of drug resistance.39 Thus, genes intimately involved with the fundamental processes of malignant transformation are now known to contribute directly to drug resistance.
Apoptosis is an active, energy-requiring, and protein synthesis-dependent process whereby cells, in response to specific signals, undergo an orderly, programmed series of intracellular events that lead to death. This process is a necessary component of normal development in all multicellular organisms and is required for the maintenance of normal function of many proliferating or renewable tissues such as the lymphatic system. Suppression of apoptosis is a common feature of neoplastic transformation. It may be the result of either overexpression of antiapoptotic genes such as bcl-2, or activation of growth factor pathways such as EGF epithelial cancers or HER-2-neu in breast cancer. Overexpression of bcl-2 is linked to the pathogenesis of B- cell lymphomas.40Activation of other protective factors such as NF-KBand the PI-3 kinase pathway in response to DNA damage suppresses cytotoxicity of chemotherapy drugs and radiation.41 Lowe et al42 elegantly demonstrated that, in the presence of normal p53, transformation of normal mouse embryo fibroblasts (MEF) with the adenovirus E1A transforming gene (functionally equivalent to loss of c-myc regulation) created a cell line with supersensitivity to doxorubicin, 5-FU, and etoposide, as well as x-irradiation, and that the cells died through the process of apoptosis. MEF cells lacking the p53 gene were resistant to doxorubicin, 5-FU, and etoposide, as well as x-irradiation. This experiment may explain the selectivity of chemotherapeutic agents for malignant cells over nonmalignant cells with similar proliferative rates and, reinforcing the results of other studies linking loss of cell-cycle control to resistance to chemotherapeutic agents,43 offer an explanation for the high rate of inherent resistance of many p53-mutated solid tumors to chemotherapeutic agents (Fig. 1.1). Furthermore, these results suggest potential targets for effectively bypassing the elaborate defense machinery available to the cancer cell. Peptidomimetic drugs that activate proapoptotic molecules have shown interesting ability to promote death of tumor cells in mice.44
Figure 1.1 Effect of c-myc regulation, p53, and bcl-2 on sensitivity of normal and malignant cells proliferating at comparable rates. (The dose of drug or radiation causing apoptosis of normal stem cells with regulated c-myc is much higher than the dose causing apoptosis of malignant cells with normal p53, but unregulated c-myc or analogous oncogene.)
Dose intensification has received increasing emphasis in recent years as a strategy for overcoming resistance to chemotherapy. Citron and colleagues45 have shown that the intensity of conventional treatment, that is, the dose per time unit, is an important consideration in adjuvant therapy of breast cancer. By decreasing the interval between treatments, using a “dose-dense” regimen, they found improvement in relapse-free survival. A steep dose-response effect for drug-responsive tumors has long been known, and the importance of delivering maximum tolerated doses in potentially curable diseases has been emphasized repeatedly. The concept of dose intensity, defined as the milligram per square meter of delivered drug per week of therapy, has been used by Levin and Hryniuk46 in retrospective comparisons of published response rates obtained with different chemotherapeutic regimens (using the published protocol doses rather than actual delivered doses) in breast cancer, colon cancer, and ovarian cancer trials. They concluded that a dose-response correlation exists for 5-FU in colon cancer, doxorubicin in breast cancer, and cisplatin in ovarian cancer. Similarly, retrospective analyses of MOPP or related regimens in Hodgkin's disease have concluded that delivered doses of vincristine, as well as of alkylators and procarbazine, correlate with response rates and disease-free intervals.47 These studies have been criticized because they are retrospective (i.e., an alternative and very plausible interpretation is that tumor-related or patient-related factors that are associated with inability to tolerate full chemotherapy doses also predict for poor response) and, in addition, in the case of the studies by Levin and Hryniuk,46 because of the rather general assumptions the authors made regarding drug equivalency, which allowed numerical dose-intensity assignments to regimens containing different drugs.
The dose-response relationships discerned from acute leukemia trials48 are probably applicable to other diseases, despite the lack of randomized studies. That is, in potentially curable cancers, readily tolerable (“standard”) doses of effective combination chemotherapy drugs are sufficient for a subset of patients with sensitive tumors, whereas higher and, in some cases, very high doses may be necessary for the subset of patients with relative drug resistance. The challenge is to develop reliable de novo predictive markers (such as, potentially, bcl-2 overexpression or mutations in p53 or K-ras genes) for each tumor to determine which patients will benefit from the higher doses. In the absence of such markers, treating every potentially curable patient with maximally tolerated doses, as established by the published or experimental protocol, is important. The following dosing principles have been used for the treatment of Hodgkin's disease,49but we believe they are broadly applicable to other potentially curable cancers: (a) Do not modify planned doses or schedules of chemotherapy in anticipation of toxicity that has not yet happened, nor for short-term, non–life-threatening toxicity, such as emesis or mild neuropathy. (b) Because significant individual variation may exist in the pharmacokinetics of drugs or in the sensitivity of the bone marrow (and other normal organs) to drug-related toxicity, the granulocyte count should be used as an in vivo biologic assay of the individual dosage limits of those agents with predominant myelotoxicity. If 100% of the planned doses do not produce a nadir granulocyte count of less than 1,000/mm3, the doses of the myelotoxic drugs are probably too low for that patient and should be increased in subsequent cycles to achieve a significant but tolerable level of myelotoxicity. (As a guideline, we aim for granulocyte nadirs in each cycle of between 500 and 1,000 per mm3.) (c) Tumor response should be assessed at regular intervals throughout therapy. If evidence is seen of lack of response or of tumor regrowth, an alternative, non–cross-resistant regimen should be started.
The use of recombinant hematopoietic growth factors can mitigate the bone marrow toxicity of chemotherapy. Two recombinant agents, granulocyte colony-stimulating factor and granulocyte-macrophage colony-stimulating factor, are effective in decreasing the duration of granulocyte nadir after myelotoxic chemotherapy, although neither affects thrombocytopenia. Other factors that may have roles in ameliorating both the platelet and granulocyte toxicities, such as thrombopoietin, are under investigation.
High Dose Chemotherapy
Marrow-ablative dosages of chemotherapy are used to increase tumor cell kill. It is then possible rescue the host with either autologous bone marrow or peripheral blood stem cells, or stem cells, or marrow from a histocompatible donor. During the past 20 years, this approach has been investigated in many centers as salvage therapy for patients with relapsed leukemias, Hodgkin's and non-Hodgkin's lymphomas, as well as some solid tumors. Rescue with marrow from a human leukocyte antigen-compatible donor has the advantage of being free of malignant cells. Marrow donated by a second person contains T lymphocytes, however, which may cause graft-versus-host disease, a potentially lethal complication. On the other hand, evidence exists that a “graft-versus-tumor” effect may be more beneficial in prolonging remissions, in comparison with syngeneic or autologous marrow rescue. The drugs used in these programs have myelosuppression as the primary dose-limiting toxicity and are used in doses above the lethal dose to bone marrow in the absence of marrow reinfusion but below the limits of nonhematologic toxicity. Alkylators such as busulfan, ifosfamide, and cyclophosphamide are prominent in most ablative regimens because characteristically their extra myeloid toxicity occurs at twofold to sevenfold higher dosage than the myeloablative dosage. Some high-dose toxicities, such as cystitis, can be prevented through the use of mesna, a disulfide that inactivates alkylating mioeties in the acid environment of urine. Total-body, total-lymphoid, or limited-field radiation has been used frequently as an adjunct to chemotherapy. Hematopoietic growth factors and peripheral blood stem cells have been used in conjunction with high-dose chemotherapy and marrow reinfusion to shorten the duration of marrow aplasia and reduce infection complications.
Randomized trials comparing high-dose regimens with best conventional therapy generally have not proven the value of dose escalation in patients with metastatic breast cancer.50 High-dose regimens with allogeneic bone marrow transplant appear to be very effective in some younger patients with acute myeloid leukemia and in chronic myelogenous leukemia, whereas autologous bone marrow transplant or peripheral blood stem cell transplant regimens appear to be effective in drug-responsive Hodgkin's disease in first or second relapse and in intermediate-grade and high-grade non-Hodgkin's lymphoma in first relapse. Reported trials generally have consisted of relatively small numbers of highly selected patients, however, and follow-up in most cases is still brief. One should remember that both early and late toxicities of high-dose chemotherapy bone marrow transplant regimens, both allogeneic and autologous, may be serious.51 Acute pulmonary toxicity and vascular occlusive disease with liver failure contribute to acute mortality due to a high-dose regimen. Later, acute lung toxicity, and secondary acute leukemia and myelodysplasia are seen with increasing frequency.52 The risk of treatment-related death from allogeneic programs may be as high as 15 to 40%, depending on the age and underlying health of the patient.
Drug Interactions in Combination Chemotherapy
Specific drug interactions, both favorable and unfavorable, must be considered in developing combination regimens. These interactions may take the form of pharmacokinetic, cytokinetic, or biochemical effects of one drug that influences the effectiveness of a second component of a combination. Patterns of overlapping toxicity are a primary concern. Drugs that cause renal toxicity, such as cisplatin, must be used cautiously in combination with other agents (such as methotrexate or bleomycin) that depend on renal elimination as their primary mechanism of excretion. Regimens that use cisplatin before methotrexate, as in the treatment of head and neck cancer, must incorporate careful monitoring of renal function, pretreatment plasma volume expansion, and dose adjustment for methotrexate to ensure that altered methotrexate excretion does not lead to severe drug toxicity. The sequence of drug administration may be critical; in many experimental systems, administration of paclitaxel before cisplatin gives additive or synergistic results, whereas the opposite sequence yields antagonism and increased toxicity.53 Taxol delays the clearance of doxorubicin and increases the risk of cardiotoxicity.54 Extensive interactions between P-450 inducers such as phenytoin or phenobarbital and P-450 substrates such as irinotecan, paclitaxel, or vincristine lead to marked increases in drug clearance and the need for upward dosage adjustment (see Chapter 21). The potential for important interactions between cancer drugs and other medications must be kept in mind during the routine care of cancer patients.
Biochemical interactions also may be important considerations in determining the choice of agents and their sequence of administration. Both synergistic and antagonistic interactions have been described. A chemotherapeutic drug may be modulated by a second agent that has no antitumor activity in its own right but that enhances the intracellular activation or target binding of the primary agent or inhibits the repair of lesions produced by the primary drug. The best example of this synergy is the use of leucovorin (5-formyltetrahydrofolate), which itself has no cytotoxic effect but which enhances the affinity of the binding of 5-FU to its target enzyme, thymidylate synthase, by forming a ternary complex among the enzyme, 5-FU, and folate.55 This combination is more effective clinically than 5-FU alone in colorectal cancer. A number of such combinations have reached the clinic and are described in greater detail in subsequent chapters.
Combination of Chemotherapy with Radiotherapy or Biologic Agents
A further innovation in the use of antineoplastic drugs is to combine drugs with irradiation or biologic agents. Many clinical protocols have been designed to take advantage of the well-documented synergy between irradiation and drugs such as cisplatin, paclitaxel and 5-FU.
The design of integrated chemotherapy-radiotherapy trials presents special problems because of the synergistic effects of the two therapies on both normal and malignant tissue. The normal tissue of greatest concern is the bone marrow, although the heart, lungs, and brain may also be affected by such interactions.56 Radiation given to the pelvic or midline abdominal areas produces a decline in blood counts, myelofibrosis, and a decrease in bone marrow reserve. This can severely compromise the ability to deliver myelotoxic chemotherapy, even months or years after the radiation. The use of conformal irradiation, administered through multiple portals, can preserve a greater portion of the marrow-bearing tissue. For some toxicities, the sequence of administration may be crucial. For example, mediastinal irradiation after combination chemotherapy for massive mediastinal Hodgkin's disease has proven to be practicable and effective. Because the initial chemotherapy results in significant shrinkage of the mediastinal tumor, smaller radiation portals can be used to encompass the residual tumor completely, with proportionately less resultant radiation pneumonitis. In small cell carcinoma of the lung confined to the thorax, simultaneous administration of radiotherapy and chemotherapy has produced better results than either therapy alone or in sequence.57 Similarly, simultaneous radiation and chemotherapy is superior to radiotherapy alone in adjuvant therapy for cervical cancer58 and rectal cancer.59 Thus, although considering the cumulative toxicities of chemotherapy and radiation on bone marrow and other vulnerable tissues in the radiation field is essential, the net benefits of simultaneous irradiation and chemotherapy often outweigh the disadvantages.
Many chemotherapeutic agents greatly potentiate the effects of irradiation and may lead to synergistic toxicity for organs usually resistant to radiation damage. Doxorubicin sensitizes both normal and malignant cells to radiation damage, possibly because both doxorubicin and x-rays produce free-radical damage to tissues. Doxorubicin adjuvant chemotherapy given in conjunction with irradiation to the left chest wall increases the risk of increased cardiac toxicity.60 Similarly, bleomycin and radiation cause synergistic pulmonary toxicity. When more than one effective chemotherapy regimen is available, the choice must be informed by consideration of these possible mutually potentiating toxicities.
A final consideration in the combined use of radiotherapy and chemotherapy is the carcinogenicity of both. The most important late side effect of cancer treatment among patients who are cured of their primary tumors is a secondary solid tumor induced by ionizing radiation. In studies of patients cured of Hodgkin's disease, the risk for secondary solid tumors begins to increase significantly after 10 years and continues to increase steadily into the second and third decades. The cumulative risk for all secondary radiation-induced (i.e., occurring within the radiation portal) solid tumors is approximately 15% at 15 years and may be as high as 20% at 25 years. On the other hand, little evidence exists that concomitant chemotherapy increases this risk of secondary solid tumors. The most important chemotherapy-related second malignancy is leukemia due to DNA alkylating or metalating agents. Among the most potently leukemogenic agents are the mustard-type alkylators, nitrosoureas, and procarbazine. The risk for leukemia increases with cumulative dose of alkylators, a fact that must be considered when long-term or high-dose alkylator use is contemplated. Leukemia has been reported after therapy for Hodgkin's disease, non-Hodgkin's lymphoma, breast cancer (adjuvant therapy), ovarian cancer, multiple myeloma, and other kinds of cancer (see Chapter 5). The most thoroughly studied group of patients consists of long- term survivors of Hodgkin's disease after MOPP chemotherapy. The cumulative risk for leukemia or myelodysplasia after MOPP is approximately 3% at 10 years. The risk for secondary myeloid malignancy decreases rapidly thereafter and approaches the age-related baseline 10 years after MOPP. Although earlier reports suggested a further increased risk for myeloid malignancy when radiation was added, either before or after MOPP, many analyses of large numbers of patients treated with both radiation and MOPP have not demonstrated any significant increased risk of acute myelogenous leukemia due to radiation.61 One should mention that the risk for myeloid leukemia is not increased after ABVD therapy for Hodgkin disease. A qualitatively different type of secondary nonlymphocytic leukemia is associated with topoisomerase II inhibitors, including etoposide, teniposide, and doxorubicin.62 Characteristically, acute myelogenous leukemia associated with topoisomerase II inhibitor therapy has a much shorter latency period than does alkylator-induced leukemia, is frequently of the myelomonocytic or monocytic FAB subtypes (M-4 or M-5, respectively), and is frequently associated with reciprocal chromosomal translocations involving band 11q23. The risk of this type of leukemia is associated with higher total cumulative dose of the topoisomerase II inhibitor and with a weekly or twice-weekly schedule. In addition, the risk may be increased when the topoisomerase II agent is combined with high-dose alkylators or with agents that inhibit DNA repair.
In summary, cancer chemotherapeutic agents have had a profound influence on the treatment and survival of patients with cancer. Because these agents have the potential for causing severe or disabling toxicity and yet must be used at maximal dosages to ensure full therapeutic benefit, the physician is literally walking a therapeutic tightrope and must constantly balance gain against likely toxicities. In this effort, every advantage afforded by knowledge of the patient, the disease, and the therapy must be used to achieve maximum benefit. The foregoing discussion should make it apparent that an intimate knowledge of drug action, drug disposition, and drug interactions, as well as late drug effects, is essential to the design and application of effective cancer chemotherapy. The essential information for this task is presented in the following chapters on individual drugs and is summarized in the initial tables that describe key features of each agent. This information can only enhance the chances of success in the difficult but rewarding task of treating cancer.
1. Greco FA, Vaughn WK, Hainsworth JD. Advanced poorly differentiated carcinoma of unknown primary site: recognition of a treatable syndrome. Ann Intern Med 1986;104:547–553.
2. Kwiatkowski DJ, Harpole DH, Godleski J, et al. Molecular-pathologic substaging in 244 stage I non-small cell lung cancer patients: clinical implications. J Clin Oncol 1998;16:2468–2477.
3. Lynch TJ, Bell DW, Sordella R, et al. Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small-cell lung cancer to Gefitinib. N Engl J Med 2004; 350:2129–2139.
4. Ramaswamy S, Ross KN, Lander ES, Golub TR. A molecular signature of metastasis in primary solid tumors. Nature Genetics 2003;33:49–54.
5. Szakacs G, Annereau JP, Lababidi S, et al. Predicting drug sensitivity and resistance: profiling ABC transporter genes in cancer cells. Cancer Cell 2004;6:129–137.
6. Ma XJ, Wang Z, Ryan PD, et al. A two-gene expression ratio predicts clinical outcome in breast cancer patients treated with tamoxifen. Cancer Cell 2004;5:607–617.
7. Holleman A, Cheok MH, Den Boer ML, et al. Gene-expression patterns in drug-resistant acute lymphoblastic leukemia cells and response to treatment. N Engl J Med 2004;351:533–542.
8. Roberts TG Jr, Chabner BA. Beyond fast track for drug approvals. N Engl J Med. 2004 Jul 29;351(5):501–505.
9. Skipper HE, Schabel FM Jr, Mellett LB, et al. Implications of biochemical, cytokinetic, pharmacologic, and toxicologic relationships in the design of optimal therapeutic schedules. Cancer Chemother Rep 1970;54:431–450.
10. Fialkow PJ. Clonal origin of human tumors. Biochim Biophys Acta 1976;458:283–321.
11. Shapiro JR, Shapiro WR. Clonal tumor cell heterogeneity. Prog Exp Tumor Res 1984;27:49–66.
12. Deininger MW, Druker BJ. SRCircumventing imatinib resistance. Cancer Cell 2004;6:108–110.
13. Holloway RW, Mehta RS, Finkler NJ, et al. Association between in vitro platinum resistance in the EDR assay and clinical outcomes for ovarian cancer patients. Gyn Oncol 2002;87:8–16.
14. Gurubhagavatula S, Liu G, Park S, et al. XPD and XRCC1 genetic polymorphisms are prognostic factors in advanced non-small-cell lung cancer patients treated with platinum chemotherapy. J Clin Oncol 2004; 22:2594–2601.
15. Curt GA, Carney DN, Cowan KH, et al. Unstable methotrexate resistance in human small-cell carcinoma associated with double-minute chromosomes. N Engl J Med 1983;308:199–202.
16. Gorlick R, Goker E, Trippett T, et al. Defective transport is a common mechanism of acquired methotrexate resistance in acute lymphocytic leukemia and is associated with decreased reduced folate carrier expression. Blood 1997;89:1013–1018.
17. Scudiero DA, Meyer SA, Clatterbuck BE, et al. Sensitivity of human cell strains having different abilities to repair O6- methyl guanine in DNA to inactivation by alkylating agents including chloroethyl nitrosoureas. Cancer Res 1984;44:2467–2474.
18. Fedier A, Fink D. Mutations in DNA mismatch repair genes: implications for DNA damage signaling and drug sensitivity. Int J Oncol 2004;24:1039–1047.
19. Druker B. Perspectives on the development of a molecularly targeted agent. Cancer Cell 2002;1:31–36.
20. Demetri GD, von Mehren M, Blanke CD, et al. Efficacy and safety of imatinib mesylate in advanced gastrointestinal stromal tumors. N Engl J Med 2002;347:472–480.
21. Cunningham D, Humblet Y, Siena S, et al. Cetuximab monotherapy and cetuximab plus irinotecan in irinotecan-refractory metastatic colorectal cancer. N Engl J Med 2004;351:337–345.
22. Hande K, Messenger M, Wagner J, et al. Inter- and intrapatient variability in etoposide kinetics with oral and intravenous drug administration. Clin Cancer Res 1999;5(10):2742–2747.
23. Watters JW, Kraja A, Meucci MA, et al. Genome-wide discovery of loci influencing chemotherapy cytotoxicity. PNAS 2004;101: 11809–11814.
24. Pui CH, Cheng C, Leung W, et al. Extended follow-up of long-term survivors of childhood acute lymphoblastic leukemia. N Engl J Med 2003;349:640–649.
25. Rice GC, Hoy C, Schimke RT. Transient hypoxia enhances the frequency of DHFR gene amplification in Chinese hamster ovary cells. Proc Natl Acad Sci U S A 1986;83:5978–5982.
26. Sugimoto Y, Tsukahara S, Oh-hara T, et al. Decreased expression of DNA topoisomerase I in camptothecin-resistant tumor cell lines as determined by monoclonal antibody. Cancer Res 1990; 50:6925–6930.
27. Ross DD, Doyle LA. Mining our ABCs: Pharmacogenomic approach for evaluating transporter function in cancer drug resistance. Cancer Cell 2004;6:105–107.
28. Marie JP, Zittoun R, Sikic BI. Multidrug resistance (mdr1) gene expression in adult acute leukemias: correlation with treatment outcomes and in vitro drug sensitivity. Blood 1991;78:586–592.
29. Lee K, Belinsky MG, Bell DW, et al. Isolation of MOAT-B, a widely expressed multidrug resistance-associated protein/ canalicular multispecific organic anion transporter-related transporter. Cancer Res 1998;58:2741–2747.
30. Taniguchi T, Tischkowitz M, Ameziane N, et al. Disruption of the Fanconi anemia–BRCA pathway in cisplatin-sensitive ovarian tumors. Nat Med 2003;9:568–574.
31. Goldie JH, Coldman AJ, Gudanskas GA. Rationale for the use of alternating non-cross-resistant chemotherapy. Cancer Treat Rep 1982;66:439–449.
32. DeVita VT, Hubbard SM, Longo DL. The chemotherapy of lymphomas: looking back, moving forward—The Richard and Hinda Rosenthal Foundation Award Lecture. Cancer Res 1987;47: 5810–5824.
33. Gordon LI, Harrington D, Andersen J, et al. Comparison of a second-generation combination chemotherapeutic regimen (m- BACOD) with a standard regimen (CHOP) for advanced diffuse non-Hodgkin's lymphoma. N Engl J Med 1992;327:1342–1349.
34. Connors JM, Klimo P, Adams G, et al. Treatment of advanced Hodgkin's disease with chemotherapy—comparison of MOPP/ ABV hybrid regimen with alternating courses of MOPP and ABVD: a report from the National Cancer Institute of Canada clinical trials group. J Clin Oncol 1997;15:2762.
35. Anderson H, Hopwood P, Prendiville J, et al. A randomized study of bolus versus continuous pump infusion of ifosfamide and doxorubicin with oral etoposide for small cell lung cancer. Br J Cancer 1993;67:1385–1390.
36. De Gramont A, Bosset JF, Milan C, et al. Randomized trial comparing monthly low-dose leucovorin and fluorouracil bolus with bimonthly high-dose leucovorin and fluorouracil bolus plus continuous infusion for advanced colorectal cancer: a French intergroup study. J Clin Oncol 1997;15:808–815.
37. Cowens JW, Creaven PJ, Greco WR, et al. Initial clinical (phase I) trial of TLC-D-99 (doxorubicin encapsulated in lyposomes). Cancer Res 1993;53:2796–2802.
38. Bottini A, Bersiga A, Brizzi M. p53 but not bcl-2 immunostaining is predictive for poor clinical complete response to primary chemotherapy in breast cancer patients. Clin Cancer Res 2000; 6:2751–2758.
39. Lai GM, Chen YN, Mickley LA, et al. P-glycoprotein expression and schedule dependence of Adriamycin cytotoxicity in human colon carcinoma cell lines. Int J Cancer 1991;49:696–703.
40. Korsmeyer SJ. Bcl-2 initiates a new category of oncogenes: regulators of cell death. Blood 1992;80:879–886.
41. Green DR, Bissonnette RP, Cotter TG. Apoptosis and cancer. PPO Updates 1994;8:37–52.
42. Lowe SW, Ruley HE, Jacks T, et al. p53-dependent apoptosis modulates the cytotoxicity of anticancer agents. Cell 1993;74:957–967.
43. Kohn KW, Jackman J, O'Connor PM. Cell cycle control and cancer chemotherapy. J Cell Biochem 1994;54:440–452.
44. Denicourt C, Dowdy SF. Medicine: Targeting Apoptotic Pathways in Cancer. Science 2004;305:1411–1413.
45. Citron ML, Berry DA, Cirrincione C, et al. Randomized trial of dose-dense versus conventionally scheduled and sequential versus concurrent combination chemotherapy as postoperative adjuvant treatment of node-positive primary breast cancer: first report of Intergroup Trial C9741/Cancer and Leukemia Group B Trial 9741. J Clin Oncol 2003;21:2226.
46. Levin L, Hryniuk WM. Dose intensity analysis of chemotherapy regimens in ovarian carcinoma. J Clin Oncol 1987;5:756–767.
47. Longo DL, Young RC, Wesley M, et al. Twenty years of MOPP therapy for Hodgkin's disease. J Clin Oncol 1986;4:1295–1306.
48. Greco FA, Vaughn WK, Hainsworth JD. Advanced poorly differentiated carcinoma of unknown primary site: recognition of a treatable syndrome. Ann Intern Med 1986;104:547–553.
49. Kaufman D, Longo DL. Hodgkin's disease. Crit Rev Oncol Hematol 1992;13:135–187.
50. Stadtmauer EA, O'Neill A, Goldstein LJ, et al., and the Philadelphia Bone Marrow Transplant Group. Conventional-dose chemotherapy compared with high-dose chemotherapy plus autologous hematopoietic stem-cell transplantation for metastatic breast cancer. N Engl J Med 2000;342:1069–1076.
51. Socié G, Stone JV, Wingard JR, et al. Long term survival and late deaths after allogeneic bone marrow transplantation. N Engl J Med 1999;341:14–21.
52. Howe R, Micallef INM, Inwards DJ, et al. Secondary myelodysplastic syndrome and acute myelogenous leukemia are significant complications following autologous stem cell transplantation for lymphoma. Bone Marrow Transpl 2003;32:317–324.
53. Liebmann J, Fisher J, Teague D, et al. Sequence dependence of paclitaxel (Taxol®) combined with cisplatin or alkylators in human cancer cells. Oncol Res 1994;6:25–31.
54. Holmes FA and Rowinsky EK. Pharmacokinetic profiles of doxorubicin in combination with taxanes. Semin Oncol 2001; 28(Suppl 12):8–14.
55. Sotos GA, Grogan L, Allegra CJ. Preclinical and clinical aspects of biomodulation of 5-fluorouracil. Cancer Treat Rev 1994;20:11–49.
56. Pihkala J, Saarinen UM, Lundstrom U, et al. Myocardial function in children and adolescents after therapy with anthracyclines and chest irradiation. Eur J Cancer 1996;32A:97–103.
57. Bunn PA, Lichter AS, Makuch RW, et al. Chemotherapy alone or chemotherapy plus chest radiation therapy in limited- stage small-cell lung cancer. Ann Intern Med 1987;106:655–662.
58. Rose PG, Bundy BN, Watkins EB, et al. Concurrent cisplatin-based chemoradiation improves progression-free and overall survival in advanced cervical cancer: results of a randomized Gynecologic Oncology Group study. N Engl J Med 1999;340:1144–1153.
59. Fisher B, Wolmark N, Rockette H, et al. Postoperative adjuvant chemotherapy or radiation therapy for rectal cancer: results from NSABP protocol R-01. J Natl Cancer Inst 1988;80:21–29.
60. Shapiro CL, Harrigan Hardenbergh P, Gelman R, et al. Cardiac effects in adjuvant doxorubicin and radiation therapy in breast cancer patients. J Clin Oncol 1998;16:3493–3501.
61. Lavey RS, Eby NL, Prosnitz LR. Impact on second malignancy risk of the combined use of radiation and chemotherapy for lymphomas. Cancer 1990;66:80–88.
62. Pui CH, Ribeiro RC, Hancock ML, et al. Acute myeloid leukemia in children treated with epipodophylotoxins for acute lymphoblastic leukemia. N Engl J Med 1991;321:1682–1687.
aPDQ is available to physicians at most medical libraries, at many hospitals, through the Internet, or through private computer software vendors. Access to PDQ is also available directly from the NCI through the NCI Information Associates Program. For information about accessing PDQ, call the Information Associates Program Customer Service Desk at 1-800-NCI-7890 (or call 301-816-2083 outside the United States).