Treatment of evolving cancers will require dynamic decision support

Abstract

•Cancer is a heterogeneous and evolving disease, yet most drugs are administered to simply give the maximum dose.•Treatment scheduling currently ignores the complex and dynamic response of the tumor and its environment.•Patients should be carefully mapped to schedules that match their current ‘tumorscape’ and treatment objectives.•We propose a five-step ‘ADAPT’ paradigm that leverages mathematical modeling to tackle this complex, multifactorial problem. Cancer research has traditionally focused on developing new agents, but an underexplored question is that of the dose and frequency of existing drugs. Based on the modus operandi established in the early days of chemotherapies, most drugs are administered according to predetermined schedules that seek to deliver the maximum tolerated dose and are only adjusted for toxicity. However, we believe that the complex, evolving nature of cancer requires a more dynamic and personalized approach. Chronicling the milestones of the field, we show that the impact of schedule choice crucially depends on processes driving treatment response and failure. As such, cancer heterogeneity and evolution dictate that a one-size-fits-all solution is unlikely—instead, each patient should be mapped to the strategy that best matches their current disease characteristics and treatment objectives (i.e. their ‘tumorscape’). To achieve this level of personalization, we need mathematical modeling. In this perspective, we propose a five-step ‘Adaptive Dosing Adjusted for Personalized Tumorscapes (ADAPT)’ paradigm to integrate data and understanding across scales and derive dynamic and personalized schedules. We conclude with promising examples of model-guided schedule personalization and a call to action to address key outstanding challenges surrounding data collection, model development, and integration. Cancer research has traditionally focused on developing new agents, but an underexplored question is that of the dose and frequency of existing drugs. Based on the modus operandi established in the early days of chemotherapies, most drugs are administered according to predetermined schedules that seek to deliver the maximum tolerated dose and are only adjusted for toxicity. However, we believe that the complex, evolving nature of cancer requires a more dynamic and personalized approach. Chronicling the milestones of the field, we show that the impact of schedule choice crucially depends on processes driving treatment response and failure. As such, cancer heterogeneity and evolution dictate that a one-size-fits-all solution is unlikely—instead, each patient should be mapped to the strategy that best matches their current disease characteristics and treatment objectives (i.e. their ‘tumorscape’). To achieve this level of personalization, we need mathematical modeling. In this perspective, we propose a five-step ‘Adaptive Dosing Adjusted for Personalized Tumorscapes (ADAPT)’ paradigm to integrate data and understanding across scales and derive dynamic and personalized schedules. We conclude with promising examples of model-guided schedule personalization and a call to action to address key outstanding challenges surrounding data collection, model development, and integration. ‘The dose makes the poison’—this principle by Paracelsus, a 16th century physician and pioneer of modern Western medicine, epitomizes the inherent trade-offs of systemic cancer therapy. If given at the right dose, then anticancer agents can overcome disease incurable by surgery or radiotherapy alone; however, given at too high or too low a dose they can incur life-threatening toxicity or be ineffectual. But what is this ‘magic bullet dose’1Ehrlich P. Experimental researches on specific therapy: on immunity with special reference to the relationship between distribution and action of antigens: first Harben lecture.in: Himmelweit F. The Collected Papers of Paul Ehrlich. Pergamon/Elsevier, Oxford, UK1960: 106-117Google Scholar? Current clinical practice is largely guided by the maximum tolerated dose (MTD) paradigm, which emerged from early pioneering work on chemotherapies for the treatment of leukemias.2Skipper H.E. Schabel F.M. Wilcox W.S. Experimental evaluation of potential anticancer agents.Cancer Chemother Reports. 1964; 35: 1-111Google Scholar, 3De Vita V.T. Chu E.A. History of cancer chemotherapy.Cancer Res. 2008; 68: 8643-8654Google Scholar, 4Korn E.L. Freidlin B. Clinical trial designs in oncology.Abeloff’s Clin Oncol. 2020; : 296-307Google Scholar It seeks to maximize the tumor cell kill, and thus the chance of cure, by administering the highest dose which can be given before unacceptable side-effects are observed. This dose is established in phase I clinical trials, and in practice is adjusted only in the case of toxicity, interference with other treatments, insurance questions, or tumor progression.4Korn E.L. Freidlin B. Clinical trial designs in oncology.Abeloff’s Clin Oncol. 2020; : 296-307Google Scholar, 5Perry M.C. The Chemotherapy Source Book. Lippincott Williams & Wilkins, 2008Google Scholar, 6Stampfer H.G. Gabb G.M. Dimmitt S.B. Why maximum tolerated dose?.Br J Clin Pharmacol. 2019; 85: 2213-2217Google Scholar Rapid and potentially significant burden reduction is a characteristic feature of a patient with cancer successfully treated at MTD (Figure 1A), and has been pivotal in establishing systemic therapies as a pillar of modern oncology.3De Vita V.T. Chu E.A. History of cancer chemotherapy.Cancer Res. 2008; 68: 8643-8654Google Scholar,5Perry M.C. The Chemotherapy Source Book. Lippincott Williams & Wilkins, 2008Google Scholar Yet, two equally defining features are severe toxicity, which results in treatment interruptions and dose modifications, and the fact that all too often improvements are only temporary and/or are only achieved in a subset of the patient’s lesions. For example, >75% of adults treated for acute lymphoblastic leukemia will have no evidence of disease after 4-6 weeks of induction chemotherapy, but only ∼30% will be cured.7Bassan R. Hoelzer D. Modern therapy of acute lymphoblastic leukemia.J Clin Oncol. 2011; 29: 532-543Google Scholar Upon recurrence this pattern repeats so that most patients will experience a sequence of different agents administered according to MTD principles, but with often progressively shorter lasting success due to the emergence of acquired or intrinsic drug resistance and accumulating toxicity. Eventually treatment may only be given with palliative rather than curative intent, at which point doses may be adjusted to maximize quality of life (Figure 1A). This illustrates that the MTD paradigm is a crucial tool in oncology, but that on its own it is not enough to address two central challenges in cancer treatment: heterogeneity and evolution. There is mounting evidence that the ‘more is better’ paradigm that motivated MTD for chemotherapies may not apply to targeted and immunotherapies, which can reach their maximum effect before dose-limiting toxicities occur and where adverse events can be harder to quantify.4Korn E.L. Freidlin B. Clinical trial designs in oncology.Abeloff’s Clin Oncol. 2020; : 296-307Google Scholar,8Sachs J.R. Mayawala K. Gadamsetty S. Kang S.P. de Alwis D.P. Optimal dosing for targeted therapies in oncology: drug development cases leading by example.Clin Cancer Res. 2016; 22: 1318-1324Google Scholar Moreover, cancer is now widely regarded as an evolutionary and ecological disease in which populations of tumor and nontumor cells compete with each other for limited space and resources, are predated upon by the immune system, and can enter symbiotic or parasitic relationships with each other.9Merlo L.M. Pepper J.W. Reid B.J. Maley C.C. Cancer as an evolutionary and ecological process.Nat Rev Scholar, Maley C.C. a of cancer and resistance 2011; Scholar, The of cancer as Scholar, D. ecological to cancer and Scholar, D. and an of Scholar The MTD paradigm is predetermined dosing and dose and/or drug only upon a toxicity or evidence of in nature and often ignores significant to disease as with the tumor for drug or (Figure results in a complex, of which the relationship between the treatment and toxicity in a The of this is to that the evolutionary nature of cancer a dynamic and to treatment by mathematical modeling. the MTD paradigm is for tumor burden more is not dosing which cell with other as the or the of resistance we should seek to match each patient with the strategy that will achieve the for their disease, treatment and treatment should be the of therapy. we to key and results from different with a and across in the to tackle the evolutionary nature of In we seek to personalized K. Optimal cancer in the Scholar, de the cancer to deliver more cancer Clin Book. 2016; Scholar, to personalized cancer in a case Oncol. 2020; by to not what drug is but and at what dose it is clinical treatment are on in the of a disease For example, not the MTD paradigm is on the that the the cell kill, the the of However, this may to the and/or patient at treatment requires a to across and of data across and can a and for and to an in of Scholar we to which modeling or seeks not to the data but the processes which have given to the a for the Scholar For example, a model an that the of a to the treatment that the patient this no on and the treatment tumor a mathematical model of the data seek to the treatment tumor and treatment response to give to the pattern of modeling has a in clinical with each in treatment scheduling by or with mathematical of the of the of mathematical Scholar, S. of from MTD Scholar, F. The of Rev Scholar, the of Cancer Scholar, D. the mathematical treatment schedules for Scholar of cancer are by a of that the of should cancer be 2020; Scholar of the of the of mathematical F. The of Rev by drug and therapy: and their Rev Scholar, to Scholar, D. B. F. mathematical of drug for Rev Clin Oncol. 2016; Scholar, personalized oncology: from of to to The Scholar, D. for personalized in oncology: and Clin 2020; Scholar, D. F. paradigm for in Scholar, The of mathematical Scholar for of key and In this is limited by current all Oncol. 2020; Scholar but is a for mathematical Rev 2008; Scholar and B. R. of and in a new with a mathematical Res. 2008; 68: Scholar, mathematical a and tumor model to clinical Scholar, K. K. K. for cancer mathematical Scholar, D. dosing mathematical and of cancer 2016; Scholar, and treatment in acute 2019; Scholar, R. of to and not 2019; Scholar, for cancer a mathematical model by 2020; 22: In this we a for treatment schedule in which we integrate data with mathematical to tackle the complex and evolving nature of a patient’s In the we key scheduling and show mathematical modeling has to dosing We that their and for and at what dose to give treatment can be by the fact that each is on a tumor model with of which understanding of This will the for the in which we the of and evolution to a of a patient’s We propose that different tumor can be as specific of this of or the patient at different in This a more of scheduling to address the complex and nature of the In we a five-step ‘Adaptive Dosing Adjusted for Personalized Tumorscapes (ADAPT)’ with which to patients to the model and strategy that best matches their current disease characteristics and treatment To we examples of model-guided schedule personalization and key research in data collection, model development, and to dynamic in The have not only an of anticancer agents, but a of as to agents should be and In the we will work by the of Cancer Scholar and F. The of Rev D. B. F. mathematical of drug for Rev Clin Oncol. 2016; D. B. F. mathematical of drug for Rev Clin Oncol. 2016; D. F. paradigm for in Scholar and key results from across different In we will show that what at first may a P. The model in of tumor Res. Scholar in a by a the to address the complex and evolving nature of cancer (Figure and We propose that as understanding has new scheduling have been by with in as as a for more personalized dosing to for in of a of a patient’s disease that we should seek to integrate a evolving by to clinical dosing the first agents in the and their limited and In a and H.E. Schabel F.M. Wilcox W.S. Experimental evaluation of potential anticancer agents.Cancer Chemother Reports. 1964; 35: 1-111Google Scholar that in to achieve more it to carefully the treatment the effect of with a low dose of or a high that the dose in a reduction and a of cure, the schedule administered a mathematical H.E. Schabel F.M. Wilcox W.S. Experimental evaluation of potential anticancer agents.Cancer Chemother Reports. 1964; 35: 1-111Google H.E. The of on the of cell Res. Scholar and R. to and of treatment Scholar that this most leukemia cells in the the and so that treatment at a dose the of This as the has a of mathematical H.E. Schabel F.M. Wilcox W.S. Experimental evaluation of potential anticancer agents.Cancer Chemother Reports. 1964; 35: 1-111Google H.E. The of on the of cell Res. Scholar that to maximize the chance of cure, treatment should be given at the highest dose and frequency which toxicity (i.e. the motivated therapy for acute leukemia of Cancer Res. 1964; Scholar to a for leukemia in which chemotherapies at maximum dose and at maximum frequency and for the first and in cases in this Vita V.T. Chu E.A. History of cancer chemotherapy.Cancer Res. 2008; 68: 8643-8654Google therapy for acute leukemia of Cancer Res. 1964; S. The of of and Scholar This the treatment of leukemia and to MTD and its mathematical as of the key in cancer M.C. The Chemotherapy Source Book. Lippincott Williams & Wilkins, 2008Google Scholar the success of the in to has been Vita V.T. Chu E.A. History of cancer chemotherapy.Cancer Res. 2008; 68: 8643-8654Google Scholar to address in and R. to and of treatment Scholar an model to for the more complex treatment and toxicity in (Figure of that that the of cells in with tumor K. of the cell of of J Scholar, from Scholar, The relationship between and of J Cancer Scholar This that in work in H.E. Schabel F.M. Wilcox W.S. Experimental evaluation of potential anticancer agents.Cancer Chemother Reports. 1964; 35: 1-111Google H.E. The of on the of cell Res. Scholar the of cells by treatment in the of therapy as the a of the and R. to and of treatment Scholar that to maximize the chance of it crucial that disease treated at the maximum dose and For chemotherapies, this is due to the fact that cell are it that treatment with of the In they that the MTD paradigm results in severe toxicity. In this that treatment has to be by doses and/or the M.C. The Chemotherapy Source Book. Lippincott Williams & Wilkins, 2008Google Scholar they that in rather than at the maximum from the treatment should be dose and/or Moreover, drugs are should be given to of each that in this it be to disease at the chance of R. to and of treatment Scholar and has an in the development of R. The more and Clin Oncol. and cancer a and a to Oncol. 2019; Scholar is at disease after surgery in to in in patients in the cancer has but has not yet at the of in the that this principle in patients with yet to results or to a in and cancer a and a to Oncol. 2019; by and more frequency and/or to as and more complex, M.C. The Chemotherapy Source Book. Lippincott Williams & Wilkins, 2008Google R. The more and Clin Oncol. and cancer a and a to Oncol. 2019; Scholar For example, trial of and as treatment of first of and Clin Oncol. Scholar administered or weeks and that the more treatment the of disease recurrence by and the of by as have established therapy as a key tool of modern and mathematical modeling can to the development of new the of therapy between treatment the of in cancer is to be at and cancer a and a to Oncol. 2019; that research is on to and agents to cancer the model for drug to with tumor it that no heterogeneity between tumor cells and that a question of and To research that cells intrinsic drug in the and a more complex treatment model to the impact of a on mathematical they that the of a resistance depends on the tumor and the the tumor is to after model for the drug of to their Scholar, for the of chemotherapy.Cancer Scholar, model for the and treatment of drug Scholar that treatment should be as early and as to the chance of resistance (Figure Moreover, toxicity not of they that the should be to maximize the chance of a model for the resistance of tumor cells to cancer The of drug resistance in for systemic Res. Scholar In this the of different be as as the chance of model for the resistance of tumor cells to cancer The of drug resistance in for systemic Res. Scholar As such, and they that model but by on resistance rather than the cell and by that cells from they different the best of drug dosing the of However, evidence for the clinical of the paradigm is limited M.C. The Chemotherapy Source Book. Lippincott Williams & Wilkins, 2008Google Scholar for an For example, in a of early that a between surgery and therapy is with a shorter to R. with early treatment in an cancer Clin Oncol. P. D. of on of for J Clin Oncol. with to treatment as early as However, as have been by evidence for the effect of treatment has Cancer for J S. S. of the after early of the Cancer J that resistance evolution is more complex than by the tumor cells are the of and scheduling of Res. Scholar, S. therapy with and tumor Clin Scholar, D. is dosing of drugs can tumor in Clin Scholar in the that tumor and treatment response are not only by the characteristics of the tumor cells but by their in on the of tumor they that of drugs at their MTD often requires significant interruptions tumor to which to To and to the effect of chemotherapies, they to treatment at a low dose, but at high to the of a a at its and Scholar (Figure therapy has promising results in a of and clinical the of Cancer personalized Rev Clin Oncol. Scholar for in developing schedules is of a dose level that is tolerated given at a high frequency the D. for Chemother Pharmacol. Scholar modeling is to address this as by a in which mathematical to derive a schedule that promising results in a phase I in cancer and F. P. modeling for phase I cancer a of for cell cancer and F. with mathematical a phase I trial in Chemother Pharmacol. Scholar Moreover, research that therapy has as doses to immune cells and a more of cell a at its and Scholar have been in this to the schedule that with tumor cell and of mathematical to what can be from Scholar, of personalized the immune response Res. 2019; Scholar, for evolution and can more Res. 2020; Scholar MTD the in the the of therapy have it as a paradigm that the of a more of cancer treatment for The nature of chemotherapies that agents can be administered for only a As a the so have been with response and/or the chance of the of the of and targeted which be given for the to acquired resistance which to research in different disease as To D. of on the cell of the Res. K. of on and Scholar or F. resistance in a strategy to drug to in 2019; Scholar has that resistance can be and to in 2019; Scholar treatment is This has schedules in which treatment is at with the of the tumor and quality of life by toxicity (Figure However, with promising To D. of on the cell of the Res. Scholar, K. of on and Scholar, F. resistance in a strategy to drug Scholar, to in 2019; Scholar clinical has so results of the phase trial of for in recurrence after radiotherapy for clinical Scholar, for level after J Scholar, in J Scholar, therapy phase clinical the is in the Clin Oncol. 2016; Scholar, and in patients with a phase 2020; Scholar, Dosing of and in no point in a 2020; Scholar, E.A. and treatment for drug Rev Clin Oncol. Scholar This is for which are a of therapy in which treatment is given at doses the MTD to cells on the that the doses to be tolerated with is an treatment Res. Scholar, D. and in the Scholar, of dosing for cell cancer with evolutionary cancer 2011; Scholar, for from cell Oncol. 2011; for the in results practice in this be that different have been in this the cancer that it may be by a of on the characteristics of cancer cells in To D. of on the cell of the Res. K. of on and Scholar drug F. resistance in a strategy to drug Scholar or cell to in 2019; Scholar in As a it is that a given is only for a specific subset of and may a different schedule or shorter or will not at To address this mathematical modeling has been to that therapy is and to potentially treatment on the response of a patient’s to in 2019; K. and cancer J R. to 2020; In this targeted therapies have treatment personalization not only in of which agents are administered but have a of treatment schedules. for mathematical schedule and potential personalization has been in the of drug toxicity. toxicity only as a on drug but have been mathematical of their own for example, the work by of with across Clin Oncol. In the complex processes that drug and a relationship between the dosing and the achieved drug level in tumor and nontumor The of and mathematical modeling to and and toxicity, and have key of modern drug development, doses and dosing of and for & Scholar, S. and modeling for clinical in 2020; Scholar, R. D. de Alwis D.P. and to clinical cancer tumor dynamic Cancer Res. 2020; Scholar For example, a by dosing for an of drug development in the Chemother Pharmacol. 2019; Scholar that and of the to to the which in their by the and the need for However, a key is that due to in drug and or can be significant and between patients so that a dose can in a of and To for more drug and dosing have drug is no personalized cancer Scholar, K. has dosing not yet clinical from the and a for the Scholar, F. R. cancer and 2020; Scholar drug with of toxicity or can the treatment dose to and (Figure In oncology, or have been in with where and can have potentially as treatment in with for acute lymphoblastic J Scholar, The of limited for cancer drug Cancer Res. Scholar, K. S. of modeling and 2008; Scholar, D. model to therapy dosing for acute lymphoblastic Scholar, D. D. Optimal for a strategy for Scholar or D. dosing mathematical and of cancer 2016; D. dosing the phase trial of in cancer Cancer 2016; As such, their has been more on the processes that drug and its impact on nontumor have mathematical can to a more and personalized of the that treatment response schedule the evidence for the and of is which illustrates of the and challenges in more dynamic and personalized scheduling clinical K. has dosing not yet clinical from the and a for the R. R. S. dosing at the challenges and Pharmacol. Scholar The most of in scheduling has been by the of from and to and has the a more complex of the tumor and the need for dynamic scheduling (Figure In it has that the choice of schedule depends not only on what we but on we The most is that of therapy which to tumor rather than cure, in patients where may be Res. of strategy in the on Scholar The central therapy is that cells are in with cells for space and resources, and that potentially they are at a in the of treatment due to a of resistance the of drug therapy to treatment in to cells to cells (Figure The of therapy is on the to the between of tumor and the of tumor cell therapy to 2020; Scholar, therapy for from patient mathematical Scholar, the and of a cancer evolutionary dosing Scholar modeling has been to this and to new data from R. to 2020; R. response to therapy in cancer Scholar Res. J.W. of evolutionary to cancer Res. Scholar and clinical to in 2019; evolutionary treatment of Scholar of therapy have on mathematical modeling to the of heterogeneity and evolutionary to recurrence in and cancer Res. Scholar, K. R. resistance to targeted cancer Scholar, The impact of the heterogeneity of cells and on treatment Scholar, therapy by Scholar of the need for a of resistance in Res. 2020; Scholar Res. 2020; Scholar, cancer therapy in an Cancer Res. 2019; Scholar, modeling of drug heterogeneity to Scholar dose K. Optimal to for Scholar and R. of Scholar In a therapy trial on cancer treatment personalized a treatment on a of tumor the a of of the and is upon or the the only patients and its of a reduction in drug and a in to to a are evolutionary treatment of R. mathematical response to in cancer and to Scholar clinical for example, cancer or cancer are In to is a of treatment which tumor and treatment as a complex, evolutionary personalized dosing (Figure and can be evolutionary and ecological therapy from events from the cancer and the of Res. 2020; Scholar in a first therapy to the and heterogeneity of a tumor by a or therapy to the populations leading to and evolution Rev Scholar, D. Maley C.C. from Rev Clin Oncol. Scholar, in from the evolutionary of Res. 2019; Scholar have been in D. trial for the treatment of widely by evolutionary Scholar and evolutionary for 2020; Scholar and are in cancer a is an evolutionary strategy where resistance to drug to to a D. evolution to drug therapy and the Scholar has emerged as an to R. as a strategy cancer resistance resistance and Scholar and has promising results in of dosing D. is on the of 2019; Scholar may be by D. evolutionary to drug in 2020; K. for drug resistance in J Scholar from the of mathematical modeling to the of Res. 2020; modeling of drug heterogeneity to S. the and of evolution with Scholar, R. Optimal therapy scheduling on a of Scholar, tumor evolution to Scholar The illustrates different to treatment scheduling have been the But in and clinical data to new no ‘magic treatment strategy has In the by different not and can be We propose that the for this is the complex, evolving nature of where on different and scales a patient will to a given treatment we that is a of to the characteristics of cancer as a can be as to the and tumor of the and nature of drug resistance of drug

Topics & Concepts