Chronic myeloid leukemia (CML) is a cancer of the hematopoietic system and has been treated with the drug Imatinib relatively successfully. Drug resistance, acquired by mutations, is an obstacle to success. Two additional drugs are now considered and could be combined with Imatinib to prevent resistance, Dasatinib and Nilotinib. While most mutations conferring resistance to one drug do not confer resistance to the other drugs, there is one mutation (T315I) that induces resistance against all three drugs. Using computational methods, the combination of two drugs is found to increase the probability of treatment success despite this cross-resistance. Combining more than two drugs, however, does not provide further advantages. We also explore possible combination therapies using drugs currently under development. We conclude that among the targeted drugs currently available for the treamtent of CML, only the two most effective ones should be used in combination for the prevention of drug resistance.
Citation: Komarova NL, Katouli AA, Wodarz D (2009) Combination of Two but Not Three Current Targeted Drugs Can Improve Therapy of Chronic Myeloid Leukemia. PLoS ONE 4(2): e4423. https://doi.org/10.1371/journal.pone.0004423
Editor: Cathal Seoighe, University of Cape Town, South Africa
Received: July 25, 2008; Accepted: December 4, 2008; Published: February 10, 2009
Copyright: © 2009 Komarova et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work has been funded in part by NIH grant R01 CA129286. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Chronic myelogenous leukemia (CML) is a cancer of the hematopoietic system and is initiated and driven by the product of the BCR-ABL fusion gene –. It proceeds in three stages: the chronic phase in which the number of cells is relatively low and the degree of cellular differentiation is relatively high; the accelerated phase during which the number of cells starts to rise to higher levels and the degree of differentiation declines; and blast crisis, which is characterized by explosive cell growth and a low degree of differentiation. Small molecules that specifically target the BCR-ABL gene product provide a successful treatment approach which can lead to a reduction of BCR-ABL+ cells below detectable levels, at least during the early stages of the disease. The drug Imatinib has been mostly used in this respect –. As the disease advances, however, the chances of treatment failure rise due to the presence of drug resistant mutants that are generated mostly through point mutations, but also through gene duplications , , –. Drug resistance can potentially be overcome by the combination of multiple drugs, where a mutation that confers resistance against one drug does not confer resistance against any of the other drugs in use. This is applied successfully in HIV infection . Computational work has shown that a combination of three or four different drugs against CML could similarly overcome the resistance problem, even during advanced stages . In addition to Imatinib, the drugs Dasatinib and Nilotinib are alternative inhibitors of the BCR-ABL gene product –. The reality, however, is that one mutation (T315I) can confer resistance against all those drugs –. We refer to this as cross resistance. In addition to this one mutation, more than 50 mutations have been identified that confer resistance against one of the drugs (in particular imatinib), but not against the others , . Given this situation, the question arises whether a combination of those three drugs can improve treatment outcome. Here we examine this question with a mathematical model.
Results and Discussion
We construct and analyze a mathematical model that is based on a previously published computational framework . This is a stochastic, continuous time, birth-death process with mutations, by which we describe the dynamics of a heterogeneous population of cancer cells. The cells can divide, die, and mutate with given probabilities. In particular, upon each division, a cell has a probability to acquire resistance against a given drug by mutation. Before the start of treatment, the colony undergoes a clonal expansion until it reaches size N. At this time, therapy is started; the drug(s) are assumed to increase the death rate of susceptible cells. This could potentially lead to a significant reduction of BCR-ABL+ cells in the absence of resistance. We calculate the probability that drug resistant mutants will lead to treatment failure, and the complementary probability of treatment success.
The model includes several biological parameters. The colony size at start of treatment is assumed to vary within 108–1013 cells. The death rate of cells in the absence of treatment is between 0 and 70% of their birth rate. In the presence of drug therapy, susceptible cells have an additional, drug-induced death rate which can be one to ten times greater than the cells' division rate. In the presence of multiple drugs, the susceptible cells are killed at the rate given by the maximum drug-induced rate in the combination. We assume that there are of the order 10–100 different point mutations that can confer resistance against only one of the drugs. There is also one particular mutation which can confer resistance against all drugs in use. Our general modeling approach continues the tradition of Goldie and Coldman – in that we view a cancerous colony as a stochastic birth-death process; the main differences include the possibility to use drugs in combination rather than sequentially; the existence of combinatorial mutation networks and a non-zero cellular death rate. More mathematical details of the model are provided in the supporting online Material S1.
First, consider the combination of two drugs. For reference, Figure 1 shows the probability of treatment success for one and two drugs assuming the absence of cross resistance. This is compared to the cross-resistance scenario. While the probability of treatment success is lower in the presence than in the absence of cross resistance, combining two drugs with cross-resistance clearly improves the probability of treatment success relative to the use of only one drug. The reason is that it is much more likely to acquire a mutation that confers resistance against only one drug than to acquire the cross-resistance mutation. This is because only one specific mutation can lead to cross-resistance, while many mutations can confer resistance against only one drug. Hence, for most mutations, combination therapy will not be challenged by cross-resistance. On a qualitative level, this result does not depend on the kinetic parameters of the model, such as the division and death rates. The advantage of combining two drugs becomes insignificant if the number of mutations that confer resistance to only one drug is very low (Figure 2b), or if the rate at which the cross-resistance mutation is acquired is relatively high. If the number of mutations that confer resistance against only one drug is on the order of 50–100, and if the rate at which resistance mutations are generated is on the order of 10−6–10−9, however, the model suggests that combining two drugs is advantageous to the patient, even if cross-resistance is possible.
The cross-resistance networks are presented by using connected and disconnected nodes. The number of nodes corresponds to the number of drugs used. Connected nodes correspond to the existence of a cross-resistant mutation. Identical connecting lines indicate that the same mutation confers cross-resistance to all connected drugs. Different (single, double, dashed) lines correspond to different mutations. Simulation parameters are as follows: μ = 10−9, k = 100, M0 = 100, d/l = 0.5, h/l = 3. The symbols “I”, “D”, “N”, and “K” stand for “Imatinib”, “Dasatinib”, “Nilotinib” and a future drug which can bind to T315 mutants.
(a) The number of non-cross-resistant mutations is low (k = 10) and the mutation rate for cross-resistance is 10 times higher than in figure 1. Conclusion: combining more than one cross-resistant drugs does not improve the chances of treatment success. (b) The number of non-cross-resistant mutations is high, k = 104. Conclusion: combining three cross-resistant drugs improves the chances of treatment success compared with two cross-resistant or non-cross-resistant drugs (which in turn is better than using only one drug). The rest of parameters are as in figure 1.
Next, consider the combination of three drugs, i.e. Imatinib, Dasatinib and Nilotinib. The model shows that combining three drugs will not lead to any further advantage compared to the combination of two drugs (Figure 1). For triple combination therapy to be advantageous, most resistant cells must harbor mutations that render them resistant against two of the drugs (but not the third one). Accumulating two separate resistance mutations, however, is a relatively rare event. It is much more likely that a cell acquires the single cross-resistance mutation. Hence, triple combination therapy does not improve the probability of treatment success compared to double combination therapy. Triple combination therapy can only provide an additional advantage if the number of mutations that confer resistance against only one drug is unrealistically high (Figure 2b). Assuming reasonable values for the cellular division, death, and mutation rates, there must be at least k = 1000 mutations that confer resistance against only one drug for triple combination therapy to somewhat improve the chances of treatment success. The improvement becomes significant for k = 10,000 or more mutations (Figure 2b). For such high values of k we observe that the effect of cross-resistance is insignificant and treatment failure occurs as a result of mutations that confer resistance to one drug at a time. In this case combining three (cross-resistant) drugs gives a better outcome than two non-cross-resistant drugs (see figure 2b). Again, these results do not qualitatively depend on the kinetic parameters of the model. Figure 3 demonstrates how the probability of treatment success for one, two and three cross-resistant drug therapies changes as a function of the natural cancer cell death rate (a) and drug-induced death rate (b).
(a) The role of the natural death-rate of the cancer cells: d/l = 0 (solid lines) and d/l = 0.8 (dashed lines); h/l = 10. (b) The role of the drug-induced death rate: h/l = 3 (solid lines) and h/l = 400 (dotted lines); d/l = 0. Other parameters are the same as in figure 1.
In summary, the analysis shows that in the context of the currently available drugs for CML therapy (Imatinib, Dasatinib, and Nilotinib), combining two of them can provide an advantage over using one drug alone. Recent experimental data support this notion . However, combining all of these drugs does not improve the chances of treatment success beyond double combination therapy. Hence, the two most effective drugs should be given simultaneously to treat CML.
While these results are robust and do not break down in the context of biologically reasonable values of the cell death rates and drug-induced death rates (figure 3), the largest degree of uncertainty exists regarding the relationship between the number of tumor cells and the probability of treatment success. As evident in Figure 1, the probability of treatment success is only high if the number of tumor cells upon start of therapy is no larger than 1010. However, many more tumor cells than this are found in patients when the disease becomes detectable, and especially during more advanced stages. On the other hand, the tumor cell population is heterogeneous, and it is not clear to what extent different cell populations contribute to the disease dynamics , . For example, if only the more primitive tumor stem cells and progenitor cells are important determinants of tumor growth, then the effective population size that our model refers to is significantly lower than the overall number of tumor cells detected in a patient.
In general, we examined all possible resistance networks for three-drug therapies. These networks are represented by circles in Figures 1 & 2. They consist of nodes corresponding to different drugs; each pair of circles is connected if there exists a mutation event that confers cross-resistance against the two drugs. Different lines (single, double, dashed) corespond to different mutations. There are five possible three-drug cross-resistance networks, including the absence of resistance and triple cross-resistance, exemplified by the case of Imatinib, Dasatinib and Nilotinib. Here we examine all intermediate possibilities. Let us suppose that there is some cross-resistance between drug K and, say, I. In other words, even though the T315I mutation does not confer resistance to K, there may be a different mutation which makes the cell resistant to both K and I, but not to D. In this situation, treating with I and K is obviously better than just treating with I (this follows from our previous results), but what is interesting, adding D does give an advantage in this case. Adding D will even give an advantage if there is a third mutation which confers resistance to both K and D (but not I). In other words, even though treating with three drugs characterized by triple cross-resistance does not improve the chances of treatment success compared to two such drugs, treating with three pairwise-cross-resistant drugs is advantageous compared to treating with two such drugs.
General cross-resistance networks have relevance for future generation treatment options. If drugs are used that show no cross resistance with Imatinib, Dasatinib, and Nilotinib, such as VX-680 , –, would it be advantageous to combine such a drug with two or more drugs that do not inhibit the T315I mutation? Our calculations show that it is advantageous to combine such a future generation drug (call it drug “K”) with two drugs that cannot act on the T315I mutation , say, Imatinib (I) and Dasatinib (D). This is seen for example in figure 1 when we compare the probability of treatment success for two cross-resistant drugs with that for “pairwise cross-resistant” drugs. We observe that any of the pairwise cross-resistant combinations of three drugs gives a significant improvement. Note that adding a third cross-resistant drug does not improve the chances of successful therapy. In future work, we will explore the possibility of treatment combinations of currently used drugs, as well drugs under development, based on in vitro data such as published in , .
To conclude, we performed quantitative studies of drug cross-resistance in CML treatment. The mathematical framework created here is general enough to be applied to other cancers. However, the model analyzed is specific to CML and the existing drugs Imatinib, Nilotinib and Dasatinib. We predict that (i) combining two cross-resistant drugs improves the chances of treatment success, (ii) combining more than two drugs that do not inhibit the T315I mutation does not increase the probability of treatment success compared to combinations of two drugs, and (iii) once a drug effective against T315I mutants becomes available, the most effective treatment strategy is to combine that drug with two of the three presently existing drugs.
Materials and Methods
Our work is based on mathematical analysis of stochastic processes, describing the dynamics of drug sensitive and drug resistant cell populations, both in the pre-treatment phase and during therapy. Mathematical details are given in the supporting online Material S1.
Wrote the paper: NK DW. Conceived and designed the mathematical models: DW NK. Performed the analysis of the mathematical models: DW NK AK.
- 1. Melo JV, Hughes TP, Apperley JF (2003) Chronic myeloid leukemia. Hematology (Am Soc Hematol Educ Program) 132–152.JV MeloTP HughesJF Apperley2003Chronic myeloid leukemia.Hematology (Am Soc Hematol Educ Program)132152
- 2. Calabretta B, Perrotti D (2004) The biology of CML blast crisis. Blood 103: 4010–4022.B. CalabrettaD. Perrotti2004The biology of CML blast crisis.Blood10340104022
- 3. Shet AS, Jahagirdar BN, Verfaillie CM (2002) Chronic myelogenous leukemia: mechanisms underlying disease progression. Leukemia 16: 1402–1411.AS ShetBN JahagirdarCM Verfaillie2002Chronic myelogenous leukemia: mechanisms underlying disease progression.Leukemia1614021411
- 4. Yoshida C, Melo JV (2004) Biology of chronic myeloid leukemia and possible therapeutic approaches to imatinib-resistant disease. Int J Hematol 79: 420–433.C. YoshidaJV Melo2004Biology of chronic myeloid leukemia and possible therapeutic approaches to imatinib-resistant disease.Int J Hematol79420433
- 5. Gorre ME, Mohammed M, Ellwood K, Hsu N, Paquette R, et al. (2001) Clinical resistance to STI-571 cancer therapy caused by BCR-ABL gene mutation or amplification. Science 293: 876–880.ME GorreM. MohammedK. EllwoodN. HsuR. Paquette2001Clinical resistance to STI-571 cancer therapy caused by BCR-ABL gene mutation or amplification.Science293876880
- 6. Shah NP, Tran C, Lee FY, Chen P, Norris D, et al. (2004) Overriding imatinib resistance with a novel ABL kinase inhibitor. Science 305: 399–401.NP ShahC. TranFY LeeP. ChenD. Norris2004Overriding imatinib resistance with a novel ABL kinase inhibitor.Science305399401
- 7. Daley GQ (2003) Towards combination target-directed chemotherapy for chronic myeloid leukemia: role of farnesyl transferase inhibitors. Semin Hematol 40: 11–14.GQ Daley2003Towards combination target-directed chemotherapy for chronic myeloid leukemia: role of farnesyl transferase inhibitors.Semin Hematol401114
- 8. Deininger MW, Druker BJ (2003) Specific targeted therapy of chronic myelogenous leukemia with imatinib. Pharmacol Rev 55: 401–423.MW DeiningerBJ Druker2003Specific targeted therapy of chronic myelogenous leukemia with imatinib.Pharmacol Rev55401423
- 9. O'Dwyer ME, Mauro MJ, Druker BJ (2002) Recent advancements in the treatment of chronic myelogenous leukemia. Annu Rev Med 53: 369–381.ME O'DwyerMJ MauroBJ Druker2002Recent advancements in the treatment of chronic myelogenous leukemia.Annu Rev Med53369381
- 10. Daub H, Specht K, Ullrich A (2004) Strategies to overcome resistance to targeted protein kinase inhibitors. Nat Rev Drug Discov 3: 1001–1010.H. DaubK. SpechtA. Ullrich2004Strategies to overcome resistance to targeted protein kinase inhibitors.Nat Rev Drug Discov310011010
- 11. Druker BJ (2003) Overcoming resistance to imatinib by combining targeted agents. Mol Cancer Ther 2: 225–226.BJ Druker2003Overcoming resistance to imatinib by combining targeted agents.Mol Cancer Ther2225226
- 12. Druker BJ (2004) Imatinib as a paradigm of targeted therapies. Adv Cancer Res 91: 1–30.BJ Druker2004Imatinib as a paradigm of targeted therapies.Adv Cancer Res91130
- 13. Gambacorti-Passerini CB, Gunby RH, Piazza R, Galietta A, Rostagno R, et al. (2003) Molecular mechanisms of resistance to imatinib in Philadelphia-chromosome-positive leukaemias. Lancet Oncol 4: 75–85.CB Gambacorti-PasseriniRH GunbyR. PiazzaA. GaliettaR. Rostagno2003Molecular mechanisms of resistance to imatinib in Philadelphia-chromosome-positive leukaemias.Lancet Oncol47585
- 14. Nardi V, Azam M, Daley GQ (2004) Mechanisms and implications of imatinib resistance mutations in BCR-ABL. Curr Opin Hematol 11: 35–43.V. NardiM. AzamGQ Daley2004Mechanisms and implications of imatinib resistance mutations in BCR-ABL.Curr Opin Hematol113543
- 15. Shannon KM (2002) Resistance in the land of molecular cancer therapeutics. Cancer Cell 2: 99–102.KM Shannon2002Resistance in the land of molecular cancer therapeutics.Cancer Cell299102
- 16. Ribeiro RM, Bonhoeffer S (2000) Production of resistant HIV mutants during antiretroviral therapy. Proc Natl Acad Sci U S A 97: 7681–7686.RM RibeiroS. Bonhoeffer2000Production of resistant HIV mutants during antiretroviral therapy.Proc Natl Acad Sci U S A9776817686
- 17. Komarova NL, Wodarz D (2005) Drug resistance in cancer: principles of emergence and prevention. Proc Natl Acad Sci U S A 102: 9714–9719.NL KomarovaD. Wodarz2005Drug resistance in cancer: principles of emergence and prevention.Proc Natl Acad Sci U S A10297149719
- 18. Talpaz M, Shah NP, Kantarjian H, Donato N, Nicoll J, et al. (2006) Dasatinib in imatinib-resistant Philadelphia chromosome-positive leukemias. N Engl J Med 354: 2531–2541.M. TalpazNP ShahH. KantarjianN. DonatoJ. Nicoll2006Dasatinib in imatinib-resistant Philadelphia chromosome-positive leukemias.N Engl J Med35425312541
- 19. Bradeen HA, Eide CA, O'Hare T, Johnson KJ, Willis SG, et al. (2006) Comparison of imatinib mesylate, dasatinib (BMS-354825), and nilotinib (AMN107) in an N-ethyl-N-nitrosourea (ENU)-based mutagenesis screen: high efficacy of drug combinations. Blood 108: 2332–2338.HA BradeenCA EideT. O'HareKJ JohnsonSG Willis2006Comparison of imatinib mesylate, dasatinib (BMS-354825), and nilotinib (AMN107) in an N-ethyl-N-nitrosourea (ENU)-based mutagenesis screen: high efficacy of drug combinations.Blood10823322338
- 20. Deininger MW (2007) Optimizing therapy of chronic myeloid leukemia. Exp Hematol 35: 144–154.MW Deininger2007Optimizing therapy of chronic myeloid leukemia.Exp Hematol35144154
- 21. Weisberg E, Manley PW, Cowan-Jacob SW, Hochhaus A, Griffin JD (2007) Second generation inhibitors of BCR-ABL for the treatment of imatinib-resistant chronic myeloid leukaemia. Nat Rev Cancer 7: 345–356.E. WeisbergPW ManleySW Cowan-JacobA. HochhausJD Griffin2007Second generation inhibitors of BCR-ABL for the treatment of imatinib-resistant chronic myeloid leukaemia.Nat Rev Cancer7345356
- 22. Quintas-Cardama A, Kantarjian H, Cortes J (2007) Flying under the radar: the new wave of BCR-ABL inhibitors. Nat Rev Drug Discov 6: 834–848.A. Quintas-CardamaH. KantarjianJ. Cortes2007Flying under the radar: the new wave of BCR-ABL inhibitors.Nat Rev Drug Discov6834848
- 23. Goldie JH, Coldman AJ (1983) A model for resistance of tumor cells to cancer chemotherapeutic agents. Mathematical Biosciences 65: 291–307.JH GoldieAJ Coldman1983A model for resistance of tumor cells to cancer chemotherapeutic agents.Mathematical Biosciences65291307
- 24. Goldie JH, Coldman AJ (1998) Drug resistance in cancer: mechanisms and models: Cambridge University Press.JH GoldieAJ Coldman1998Drug resistance in cancer: mechanisms and models: Cambridge University Press.
- 25. Coldman AJ, Goldie JH (1986) A stochastic model for the origin and treatment of tumors containing drug-resistant cells. Bull Math Biol 48: 279–292.AJ ColdmanJH Goldie1986A stochastic model for the origin and treatment of tumors containing drug-resistant cells.Bull Math Biol48279292
- 26. Wicha MS, Liu S, Dontu G (2006) Cancer stem cells: an old idea–a paradigm shift. Cancer Res 66: 1883–1890.MS WichaS. LiuG. Dontu2006Cancer stem cells: an old idea–a paradigm shift.Cancer Res6618831890discussion 1895–1886. discussion 1895–1886.
- 27. Reya T, Morrison SJ, Clarke MF, Weissman IL (2001) Stem cells, cancer, and cancer stem cells. Nature 414: 105–111.T. ReyaSJ MorrisonMF ClarkeIL Weissman2001Stem cells, cancer, and cancer stem cells.Nature414105111
- 28. Burley SK (2005) Application of FAST™ fragment-based lead discovery and structure-guided design to discovery of small molecule inhibitors of BCR-ABL tyrosine kinase active against the T315I imatinib-resistant mutant. Blood 106: 206a.SK Burley2005Application of FAST™ fragment-based lead discovery and structure-guided design to discovery of small molecule inhibitors of BCR-ABL tyrosine kinase active against the T315I imatinib-resistant mutant.Blood106206a
- 29. Carter TA, Wodicka LM, Shah NP, Velasco AM, Fabian MA, et al. (2005) Inhibition of drug-resistant mutants of ABL, KIT, and EGF receptor kinases. Proc Natl Acad Sci U S A 102: 11011–11016.TA CarterLM WodickaNP ShahAM VelascoMA Fabian2005Inhibition of drug-resistant mutants of ABL, KIT, and EGF receptor kinases.Proc Natl Acad Sci U S A1021101111016
- 30. Melo JV, Chuah C (2007) Resistance to imatinib mesylate in chronic myeloid leukaemia. Cancer Lett 249: 121–132.JV MeloC. Chuah2007Resistance to imatinib mesylate in chronic myeloid leukaemia.Cancer Lett249121132
- 31. Harrington EA, Bebbington D, Moore J, Rasmussen RK, Ajose-Adeogun AO, et al. (2004) VX-680, a potent and selective small-molecule inhibitor of the Aurora kinases, suppresses tumor growth in vivo. Nat Med 10: 262–267.EA HarringtonD. BebbingtonJ. MooreRK RasmussenAO Ajose-Adeogun2004VX-680, a potent and selective small-molecule inhibitor of the Aurora kinases, suppresses tumor growth in vivo.Nat Med10262267
- 32. Young MA, Shah NP, Chao LH, Seeliger M, Milanov ZV, et al. (2006) Structure of the kinase domain of an imatinib-resistant Abl mutant in complex with the Aurora kinase inhibitor VX-680. Cancer Res 66: 1007–1014.MA YoungNP ShahLH ChaoM. SeeligerZV Milanov2006Structure of the kinase domain of an imatinib-resistant Abl mutant in complex with the Aurora kinase inhibitor VX-680.Cancer Res6610071014
- 33. Duncan EA, Goetz CA, Stein SJ, Mayo KJ, Skaggs BJ, et al. (2008) IkappaB kinase beta inhibition induces cell death in Imatinib-resistant and T315I Dasatinib-resistant BCR-ABL+ cells. Mol Cancer Ther 7: 391–397.EA DuncanCA GoetzSJ SteinKJ MayoBJ Skaggs2008IkappaB kinase beta inhibition induces cell death in Imatinib-resistant and T315I Dasatinib-resistant BCR-ABL+ cells.Mol Cancer Ther7391397
- 34. O'Hare T, Eide CA, Tyner JW, Corbin AS, Wong MJ, et al. (2008) SGX393 inhibits the CML mutant Bcr-AblT315I and preempts in vitro resistance when combined with nilotinib or dasatinib. Proc Natl Acad Sci U S A 105: 5507–5512.T. O'HareCA EideJW TynerAS CorbinMJ Wong2008SGX393 inhibits the CML mutant Bcr-AblT315I and preempts in vitro resistance when combined with nilotinib or dasatinib.Proc Natl Acad Sci U S A10555075512