Model-Based Inference and Classification of Immunologic Control Mechanisms from TKI Cessation and Dose Reduction in Patients with CML
Total Page:16
File Type:pdf, Size:1020Kb
Published OnlineFirst February 10, 2020; DOI: 10.1158/0008-5472.CAN-19-2175 CANCER RESEARCH | CONVERGENCE AND TECHNOLOGIES Model-Based Inference and Classification of Immunologic Control Mechanisms from TKI Cessation and Dose Reduction in Patients with CML Tom Hahnel€ 1, Christoph Baldow1,Joelle€ Guilhot2, Francois¸ Guilhot2, Susanne Saussele3, Satu Mustjoki4,5, Stefanie Jilg6, Philipp J. Jost6, Stephanie Dulucq7, Francois-Xavier¸ Mahon8, Ingo Roeder1,9, Artur C. Fassoni10, and Ingmar Glauche1 ABSTRACT ◥ Recent clinical findings in patients with chronic myeloid leuke- cessation. Among them were a third class of patients that main- mia (CML) suggest that the risk of molecular recurrence after tained TFR only if an optimal balance between leukemia abundance stopping tyrosine kinase inhibitor (TKI) treatment substantially and immunologic activation was achieved before treatment cessa- depends on an individual's leukemia-specific immune response. tion. Model simulations further suggested that changes in the However, it is still not possible to prospectively identify patients that BCR-ABL1 dynamics resulting from TKI dose reduction convey will remain in treatment-free remission (TFR). Here, we used an information about the patient-specific immune system and allow ordinary differential equation model for CML, which explicitly prediction of outcome after treatment cessation. This inference of includes an antileukemic immunologic effect, and applied it to individual immunologic configurations based on treatment altera- 21 patients with CML for whom BCR-ABL1/ABL1 time courses had tions can also be applied to other cancer types in which the been quantified before and after TKI cessation. Immunologic endogenous immune system supports maintenance therapy, control was conceptually necessary to explain TFR as observed in long-term disease control, or even cure. about half of the patients. Fitting the model simulations to data, we identified patient-specific parameters and classified patients into Significance: This mathematical modeling approach provides three different groups according to their predicted immune system strong evidence that different immunologic configurations in configuration (“immunologic landscapes”). While one class of patients with CML determine their response to therapy cessation patients required complete CML eradication to achieve TFR, other and that dose reductions can help to prospectively infer different patients were able to control residual leukemia levels after treatment risk groups. Introduction chromosomal translocation between chromosomes 9 and 22. The resulting BCR-ABL1 fusion protein acts as constitutively activated Chronic myeloid leukemia (CML) is a myeloproliferative disor- tyrosine kinase triggering a cascade of protein phosphorylation, der, which is characterized by the unregulated proliferation of which deregulate cell cycle, apoptosis regulation, cell adhesion, and immature myeloid cells in the bone marrow. CML is caused by a genetic stability. Because of their unregulated growth and their distorted differentiation, immature leukemic cells accumulate and impair normal hematopoiesis in the bone marrow, leading to the 1Institute for Medical Informatics and Biometry, Carl Gustav Carus Faculty of 2 patient's death if left untreated. Medicine, Technische Universitat€ Dresden, Dresden, Germany. INSERM CIC fi 3 € Tyrosine kinase inhibitors (TKI) speci cally target the kinase 1402 - CHU Poitiers, Poitiers, France. III. Medizinische Klinik, Universitatsmedizin fi Mannheim, Heidelberg University, Heidelberg, Germany. 4Hematology Research activity of the BCR-ABL1 protein with high ef ciency and have Unit Helsinki, Helsinki University Hospital Comprehensive Cancer Center, been established as the first line treatment for patients with CML (1). Helsinki, Finland. 5Translational Immunology Research Program and Depart- Individual treatment responses are monitored by measuring the ment of Clinical Chemistry and Hematology, University of Helsinki, Helsinki, proportion of BCR-ABL1 transcripts relative to a reference gene, for 6 Finland. III. Medizinische Klinik und Poliklinik, Klinikum rechts der Isar, Tech- example, ABL1 or GUS, in blood cell samples by using reverse nische Universitat€ Munchen,€ Munchen,€ Germany. 7Laboratory of Hematology, transcription and quantitative real-time PCR (qRT-PCR; refs. 2–4). University Hospital of Bordeaux, Bordeaux, France. 8Bergonie Institute, INSERM U1218, University of Bordeaux, Bordeaux, France. 9National Center for Tumor Most patients show a typical biexponential treatment response with Diseases (NCT), Partner Site Dresden, Dresden, Germany. 10Instituto de a rapid, initial decline (a slope), followed by a moderate, second Matematica e Computac¸ao,~ Universidade Federal de Itajuba, Itajuba, Brazil. decline (b slope; refs. 5–7). Whereas the initial decline can be Note: Supplementary data for this article are available at Cancer Research attributed to the eradication of proliferating leukemic cells, the Online (http://cancerres.aacrjournals.org/). second decline has been suggested to result from a slower eradi- fi A.C. Fassoni and I. Glauche contributed equally to this article. cation of quiescent leukemic stem cells (3, 4, 8, 9). Within ve years of treatment, about two thirds of the patients achieve a major € Corresponding Author: Ingmar Glauche, Technische Universitat Dresden, molecular remission (MMR), that is, a BCR-ABL1 reduction of three Fetscherstrasse 74, Dresden D-01307, Germany. Phone: 49-351-458-6051; Fax: 49-351-458-7222; E-mail: [email protected] logs from the baseline (MR3), while at least one third of these additionally achieve a deep molecular remission (DMR, i.e., MR4 or Cancer Res 2020;XX:XX–XX lower; refs. 4, 7, 10). doi: 10.1158/0008-5472.CAN-19-2175 TKI discontinuation has been established as an experimental Ó2020 American Association for Cancer Research. option for well responding patients with DMR for at least one AACRJournals.org | OF1 Downloaded from cancerres.aacrjournals.org on September 23, 2021. © 2020 American Association for Cancer Research. Published OnlineFirst February 10, 2020; DOI: 10.1158/0008-5472.CAN-19-2175 Hahnel€ et al. year (11, 12). Different studies independently confirmed that Here, we used BCR-ABL1 time courses of TKI-treated pati- about half of the patients show a molecular recurrence, while the ents with CML that were enrolled in previously published TKI others stay in sustained treatment-free remission (TFR) after TKI discontinuation studies from different centers in Europe. In par- stop. Consistently, most patients present with a recurrence within ticular, we focused on patients for which complete time courses 6 months, while only a few cases are observed thereafter (11–14). during the initial TKI therapy and after treatment cessation are The overall good response of those patients after restarting treat- available. Therefore, potential correlations between response ment with the previously administered TKI indicates that clonal dynamics, remission occurrences, and timings after cessation transformation and resistance occurrence is not a primary problem become accessible. Motivated by the observation that the initial in CML. As it appears unlikely that even a sustained remission treatment response before TKI cessation does not show obvious truly indicates a complete eradication of the leukemic cells, other correlations with remission occurrences, we aim to explain the factors have to account for a continuing control of a minimal, resulting dynamics in terms of an ordinary differential equation potentially undetectable residual leukemic load. Although treat- (ODE) model of TKI-treated CML. Explicitly including a patient- ment discontinuation is highly desirable to reduce treatment- specific, CML-dependent immune component, we are able to related side-effects and lower financial expenditures (15, 16), it is demonstrate that three different immunologic configurations can still not possible to prospectively identify those patients that are at determine the overall outcome after treatment cessation. We risk for a molecular recurrence. Investigations of clinical markers further investigate how this patient-specificconfiguration can and scores to predict the recurrence behavior of patients after the be estimated from system perturbations, such as TKI dose reduc- treatment cessation revealed that both TKI treatment duration and tion scenarios prior to treatment cessation. Our predictions the duration of a DMR were also associated with a higher prob- closely resemble recent clinical findings substantiating our con- ability of TFR (11, 13, 17, 18). However, it is still unclear whether clusion that treatment response during TKI dose reduction is the dynamics of the initial TKI treatment response (e.g., the initial indeed informative to predict a patient's future outcome after slope of decline) correlate with the remission occurrence after stopping therapy (40). treatment discontinuation. The underlying mechanisms of the recurrence behavior after TKI stop are still controversial. While fewer recurrences for Patients and Methods patients with longer treatment suggest that a leukemic stem cell Patient selection fi exhaustion is an important determinant, it is not a suf cient We analyzed time courses of 60 TKI-treated patients with CML, criteria to prospectively identify nonrecurring patients (13, 17). for whom TKI therapy had been stopped as a clinical intervention. Favorable outcomes of treatment discontinuation for patients that Informed written