Although a variety of specific drugs are used in the chemotherapy of cancer, their effects can be quite variable from patient to patient and even within the same patient at different periods of the treatment regimen. Such variability involves different factors, some of which we have discussed above and others of which are noted in Figure 20.6. In this figure, the overall pharmacological- therapeutic process from a drug dose to its therapeutic effect is depicted. However, since neopla- sia is a somewhat specific situation involving cell growth as well as specific humoral effects of the neoplasm on the host (Chapters 17 and 18), other variable factors come into play, some of which are considered here.
It is already obvious that drugs used in the chemotherapy of cancer have a variety of toxic effects in the host that may or may not be directly related to its effect on the neoplasm. Thus, in any such situation, the toxicity to normal tissues and the organism as a whole becomes a limiting factor in the dose of drugs that can be given for effective therapy without inducing excessive toxicity. Some of the more commonly seen toxicities to the patient are seen in Table 20.6. A number of the toxicities noted in Table 20.6 occur predominantly at high doses or high doses given for extended periods of time. In addition to the toxicities noted there in the table, gonadal
Figure 20.6 Scheme outlining the overall pharmacologic-therapeutic process from drug dose to a thera- peutic effect, which in cancer chemotherapy is destruction and/or elimination of neoplastic cells. (Adapted from Bjornsson, 1996, with permission of the author and publishers.)
dysfunction, hypersensitivity reactions, and dermatological complications as well as alopecia (loss of hair) occur not uncommonly as toxic manifestations of a number of chemotherapeutic regimens.
The relationship between the probability of some biological or toxic effect of a drug and the dose administered to induce antineoplastic efficacy may be seen in Figure 20.7. As noted, if the drug is to be useful, the “antitumor effect” curve, ideally giving complete clinical remission, should be displaced toward lower doses compared with the curve describing the probability of significant toxicity to normal tissue. The therapeutic index (or therapeutic ratio) may be defined
aThe list of drugs inducing these effects is not entirely complete. The reader is referred to the paper by Lowenthal and Eaton (1996) for further details of some of these toxicities.
from such relationships as the ratio of the dose required to produce a given probability of toxicity and of an antitumor effect. The therapeutic index seen in Figure 20.7 is represented as a ratio of the 5% level of probability to severe toxicity (referred to as toxic dose 05 or TD-05) and the 50% probability of an antineoplastic effect, referred to as the effective dose 50 or ED-50. The appropriate end points of neoplastic response and toxicity will depend on the limiting toxic- ity of the drug, the intent of the treatment (i.e., cure versus palliation), and whether treatment is given to a patient or an experimental animal. It should be noted, however, that dose-response curves similar to those of Figure 20.7 have rarely been obtained for drug effects in humans (Tannock, 1992).
Obviously, improvement in the therapeutic index of a drug is the goal of chemotherapy, whether experimental or clinical. Any method whereby the antineoplastic effect of a drug can be shifted such that the ED-50 gives very little if any normal tissue toxicity is to be the ultimate goal. Improvement in the therapeutic index still requires that any treatment modification leading to increased killing of neoplastic cells in experimental systems must be assessed for its effects on critical normal tissues prior to therapeutic trials.
Remission Versus Cure in Cancer Chemotherapy
It is now well established that the “curability” of cancer by chemotherapy is related to the body burden of viable neoplastic cells present at the time chemotherapy is initiated. Although treat- ment of the neoplasm in the patient may result in the disappearance of all clinical and laboratory findings pointing toward the presence of neoplastic cells in the patient, almost half of the cases of treated neoplasia in the United States recur after some period of time. Thus, it is more appro- priate to use the term remission to indicate the clinical situation of a patient who has been treated, apparently successfully, for a neoplastic condition. While it is now possible by a variety of technologies (Pantel et al., 1999; Lambrechts et al., 1998) to detect extremely low levels of neoplastic cells remaining in the organism after apparently successful therapy, for most solid
Figure 20.7 Schematic relation between the dose of a drug and (a) the probability of a given measure of antineoplastic effect, and (b) the probability of a given measure of normal-tissue toxicity. Although the therapeutic index might be defined as the ratio of doses to give 50% probabilities of normal tissue damage and antineoplastic effects, when the endpoint toxicity is severe, a more appropriate definition of the thera- peutic index should be at a lower probability of toxicity as noted in the box. (Adapted from Tannock, 1992, with permission of the author and publisher.)
neoplasms the limit of clinical and/or radiological detection is of the order of 1 g of tissue or about 109 cells. If these 109 cells are scattered throughout the organism rather than in a single locus, obviously there will be no clinical detection of the neoplastic disease. The question of how many neoplastic cells could remain in the host before a cure was effected was answered several decades ago in experiments using rodents.
The L-1210 Leukemia Model and Chemotherapy
In 1965, Howard Skipper reported investigations on the therapy of experimental leukemia in mice. Skipper demonstrated that in this system it was necessary to kill every leukemic cell in the host (regardless of the total number, their anatomical distribution, or metabolic heterogeneity) in order to effect a cure, since one single, viable L-1210 cell could grow, proliferate, and kill the mouse. Obviously, the major problem in this investigation was associated with the killing of a relatively small but persistent fraction of leukemic cells that survived the maximum toler- ated therapy because of the relative efficacy of the drug, drug resistance, or anatomical compartmentalization.
A hypothetical illustration of the possible importance of drug level and schedule in at- tempts to achieve a total cure in experimental leukemia in animals is seen in Figure 20.8. As indicated in the figure, if one initially administers 105 leukemic cells to a mouse, one finds that the cells, after a 2-day lag, proliferate logarithmically until the mouse is killed when 109 cells are present in the body. Therefore, the time of survival is inversely related to the number of leukemic cells in the mouse at any one time. Line A in the figure represents the number of leukemic cells in untreated animals as a function of days after inoculation of the cells. Line B, representing the daily drug treatment, termed low-level, long-term (until death), is plotted to show a daily 50% “drug kill” of the leukemic cell population in the animal together with a daily quadrupling of the surviving leukemic cells. The percentage of cells killed by a given dose of a given active drug is constant or, in other words, a constant fraction of cells is killed with each dose. This phenome-non is termed the first-order kinetics of killing. Line C, representing the daily drug treatment that is termed moderate-level, long-term, is plotted to show a daily 75% drug kill of the animal’s leukemic cell population. Theoretically, this can result in a different host survival rate, assuming that cumulative drug toxicity and development of drug resistance or compartmentalization, such as in the central nervous system, do not occur. Line D, representing the daily drug treatment that is termed high-level, short-term, is plotted to show a daily 99% drug kill of the animal’s leuke-mic cell population and, barring other complications, a “cure” of a 105 cell inoculum.
With this model, it is possible to devise drug schedules, usually intermittent, in which all tumor cells are killed and the mice cured, by allowing time for normal dividing cells to recover from toxicity. Although this model has potentially interesting implications, many cancer chemo- therapeutic drugs kill only dividing cells (cf. Skipper and Schabel, 1982). Therefore, the L-1210 leukemia model has found its greatest applicability in the human in acute leukemias (cf. Frei,1984), where most of the cells are dividing. Other studies have indicated that logarithmic de- creases in tumor cell survival, resulting from drugs affecting dividing cells, require logarithmic rather than linear increases in dosages.
The application of Figure 20.8 to the situation in the human in the case of most solid neo- plasms is seen in Figure 20.9. In this figure, the decreases noted in the number of neoplastic cells occurs to the point where there is a clinical “complete remission.” However, at this point there are still more than 108 viable neoplastic cells remaining in the patient. At a later time these con-tinue to grow and even in the face of continued therapy may have developed some resistance in the patient by a variety of mechanisms, some of which we will discuss now. Takahashi and Nish- ioka (1995) have also argued that significant gain in survival may occur without actual reduction in the amount of neoplasm by cell death or other such mechanisms. They have argued that, espe- cially in solid neoplasms, the survival times of most patients depend more on an induced cyto- static phase rather than loss of tumor mass. This concept may be important in the prolonged survival sometimes noted with other means of therapy, such as immunotherapy.
Drug Resistance in Neoplastic Cells
In his early experiments, Skipper (1965) believed that the primary cause of death in leukemia in the face of continued daily treatment was the relatively rapid selection of a mutant, drug-resis- tant leukemia cell population. This result was especially notable when only single drugs were used at low or moderate levels. It is now evident that a variety of different mechanisms of drug resistance of neoplastic cells exist involving mutations, chromosomal abnormalities, gene ampli- fication, and alternate metabolic pathways. A summary of several of the molecular mechanisms involved in resistance to chemotherapeutic drugs in neoplastic cells is seen in Figure 20.10.
The Multidrug-Resistance Phenotype
Some three decades ago, a number of descriptions demonstrating that neoplastic cell lines display a resistance to multiple different chemotherapeutic drugs was reported. The drugs in- volved showed no obvious strong chemical similarity but later were found to have similar physi- cal properties in that they are relatively hydrophobic. Some of the drugs to which cells with the multidrug resistance (MDR) phenotype exhibit resistance include actinomycin D, etoposide, paclitaxel, and vinblastine (Bellamy, 1996). More recent investigations have demonstrated that a common feature of many of these drug-resistant lines is the existence of the P-glycoprotein, a specific protein in the plasma membrane of such cells. The function of this glycoprotein is an energy-dependent outward drug efflux from the cell, such that cells exhibiting the MDR pheno- type fail to accumulate drugs (Gottesman, 1993). In the membrane, the molecule has 12
Figure 20.8 Diagram of drug schedule and cell killing in mice inoculated with the L-1210 leukemia. At day 0 the animal is inoculated with 105 cells. The subsequent four conditions (A, B, C, and D) represent the growth curves of the cells under several different drug regimens as described in the text. (After Skipper,1965, with permission of the author and publisher.)
transmembrane regions as well as two large intracytoplasmic components. A diagram of its appearance and possible mechanism of action is seen in Figure 20.11. In the figure a cationic drug, such as doxorubicin, is shown entering the plasma membrane and subsequently a “pore” of the P-glycoprotein either within the membrane or from the cytoplasm. Such transport back to the outside of the cell requires energy, as noted in the figure (Gottesman, 1993). The P-glycoprotein occurs in normal tissues, especially those involved in major physiological transport mechanisms such as the liver, intestine, kidney, and brain (Schinkel, 1997). In these tissues, especially in the gut, the presumed function is to remove potentially deleterious exogenous materials from the cell rapidly. In neoplasms, expression of high levels of the P-glycoprotein usually is associated with a poor prognosis with a variety of different neoplasms (cf. Bellamy, 1996; Dicato et al.,1997).
The P-glycoprotein is one member of a large superfamily of similar transport protein com- plexes termed the ABC superfamily (Bellamy, 1996). Several transport proteins involving non-P- glycoprotein–mediated multidrug resistance have also been described (cf. Bellamy, 1996; Ya-
Figure 20.9 Relationship between remission of a neoplasm and a complete cure as a function of the remaining numbers of neoplastic cells in the host. In this hypothetical example, treatment is initiated when there are about 100 g (1011 cells) present in the host. Each treatment, given at monthly intervals, eliminates
90% of the cells present, leading to complete disappearance of any clinical evidence of the neoplasm in the host or a “complete remission.” However, more than 108 viable cells are present, many of which have now become resistant to the therapy utilized, with subsequent regrowth requiring alternate methods of therapy or ultimately leading to demise of the patient. Note that despite the attainment of a complete remission or clinical response, more than 108 viable neoplastic cells remain and that the reduction in cell numbers is small compared to that required for a cure. (Adapted from Tannock, 1992, with permission of the author and publisher.)
mada et al., 1997; Baggetto, 1997). However, a variety of agents that are capable of modulating the function of the P-glycoprotein have been described (Kavallaris, 1997), but as yet it does not appear feasible to employ such agents together with the effective chemotherapeutic drugs that are transported by this protein.