Population quantitative systems pharmacology model of distinct CAR-T cell phenotypes and CD19-specific metabolic tumour volume reveals sources of high pharmacokinetic variability and overall survival
Anna M. Mc Laughlin (1,2), Nahum Puebla-Osorio (3), Robin Michelet (1), Michael Green (3), Annette Künkele (4,5), Wilhelm Huisinga (6), Paolo Strati (3), Beth Chasen (7), Sattva S. Neelapu (3), Cassian Yee (8,9), Charlotte Kloft (1)
(1) Dept. of Clinical Pharmacy and Biochemistry, Institute of Pharmacy, Freie Universitaet Berlin, Germany, (2) Graduate Research Training Program PharMetrX, Germany, (3) Department of Lymphoma and Myeloma, Division of Cancer Medicine, The University of Texas MD Anderson Cancer Center, Houston, USA, (4) Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin, Humboldt – Universität zu Berlin, and Berlin Institute of Health, Department of Pediatric Oncology and Hematology, Berlin, Germany, (5) Berlin Institute of Health (BIH), Berlin, Germany, (6) Institute of Mathematics, University of Potsdam, Potsdam, Germany, (7) Department of Nuclear Medicine, Division of Diagnostic Imaging, The University of Texas MD Anderson Cancer Center, Houston, USA, (8) Department of Melanoma Medical Oncology, The University of Texas MD Anderson Cancer Center, Houston, USA, (9) Department of Immunology, The University of Texas MD Anderson Cancer Center, Houston, USA
Objectives: Chimeric antigen receptor (CAR)-T cell therapy has increased the prognosis of patients with relapsed/refractory Non-Hodgkin lymphoma (NHL) significantly [1–3]. Yet, only 40%-64% of patients respond long-term [1,4,5]. Thus, early response prediction would be valuable. CAR-T cell expansion and persistence are the two main determinants of treatment success [6], but exhibit significant interindividual variability (IIV) [6,7]. Reported potential covariates include the CAR design [8], the manufacturing method [9], the ‘fitness’ [10], phenotype [11,12], and CD4+/CD8+ subset composition of the T cells [12,13] in manufacturing and infusion products, tumour burden [10,14], tumour microenvironment [15,16], and the lymphodepleting chemotherapy regimen [17,18]. While correlative analyses identified factors associated with favourable outcome [6,19], no analysis has quantified their impact on distinct CAR-T cell kinetic phases. Previous CAR-T cell models characterised the (pre-)clinical kinetics and dynamics of CAR-T cells [7,20,21]; however, they did not consider distinct CAR-T cell phenotypes, a desired feature due to their different expansion, killing potential, and survival [22]. Thus, this work aimed to characterise the kinetics and dynamics of four CAR-T cell phenotypes and the CD19+ metabolic tumour volume in a clinical ‘real-world’ population using a population quantitative systems pharmacology (QSP) model.
Methods: Data from a ‘real-world’ study involving 19 NHL patients with follow-up≤24 months containing 133 CAR-T cell phenotype and 48 metabolic tumour volume measurements was acquired. CAR-T cell concentrations in peripheral blood were measured on days 7, 14 and 28 after CAR-T cell infusion using flow cytometry. CD19+ metabolic tumour volume was measured using positron emission tomography – computed tomography (PET-CT) at baseline, at one month and 3 months after CAR-T cell infusion. Amalgamating literature knowledge on immunology, physiology and adoptive cell therapy with insights from previously published CAR-T cell models [7,20,21] and the clinical data, a nonlinear mixed-effects population QSP model for CD19+ CAR-T cell therapy was developed. Patient characteristics as well as progression-free survival (PFS) and overall survival (OS), were analysed in different, model-defined, subpopulations using exploratory data analysis and Kaplan-Meier plots, respectively.
To propose an early survival prediction framework for use in clinical practice, which would be based on but not require using the model, correlations between clinical composite scores (CCS) of the ratios between Cmax and baseline metabolic tumour volume, and a selected model parameter were tested for the sum of all CAR-T cell phenotypes and each phenotype separately. For the CCS with the highest correlation, a receiver operating characteristic (ROC) analysis was performed to determine a CCS cut-off value for early survival prediction. Finally, a Cox-proportional hazard model was used to estimate the reduction in risk for death associated with a CCS above the cut-off value.
Results: Following the progressive differentiation T cell lineage model [23,24], our population QSP model postulated the differentiation of naïve CAR-T cells (TN) via central memory CAR-T cells (TCM), and effector memory CAR-T cells (TEM) to effector CAR-T cells (TEFF). For all phenotypes, except the terminally differentiated TEFF, two expansion terms were implemented: (i) linear homeostatic proliferation and (ii) nonlinear expansion upon tumour contact with maximum expansion rate per mL tumour volume (Vmax1) and concentration at half maximum expansion (KM1). For TN, TCM and TEM, linear differentiation processes to TCM, TEM and TEFF, respectively, were implemented. For all CAR-T cell phenotypes, natural death was characterised using first order reactions. The homeostasis of metabolic tumour volume was characterised using a logistic growth model; CAR-T cell tumour killing was implemented using a Michaelis-Menten model. A mixture model for parameter Vmax1 distinguished between the reference population (n=15) with adequate expansion and a low expansion subpopulation (n=4) with a 92% reduced capacity. IIV was implemented on Vmax1 and the maximum tumour killing capacity. A log-transformed both sides approach was used to model residual variability. Two covariates were identified on Vmax1: a previous autologous stem cell transplantation (ASCT), implemented as a dichotomous covariate using a fractional change model, was estimated to increase the expansion capacity by 3.53-fold. The ratio of CD4+/CD8+ CAR-T cells at day seven was centred to the median and implemented as a power function. An increase in the CD4+/CD8+ fraction moderately decreased Vmax1. Implementation of the mixture model and inclusion of both covariates considerably reduced the IIV on Vmax1 from 446% CV to 150% CV. All parameters were estimated with acceptable precision (RSE: ≤ 45%) and model predictions captured the observations well as indicated by goodness-of-fit plots.
Patients in the low expansion subpopulation had significantly higher metabolic tumour volumes at baseline and had received fewer previous ASCTs. Patients in this subpopulation and without previous ASCT had shorter PFS and OS compared to patients in the reference population and with previous ASCT: 100% of patients with previous ASCT (n=7) were alive at the end of follow-up. In contrast, 44.4% of patients without ASCT in the reference population (n=9), and none of the 3 patients without ASCT in the low expansion subpopulation were still alive.
Among the tested correlations for different CCS and selected model parameter Vmax1, the correlation was highest for the CCS for naïve T cells (CCSTN) (r=0.98). The ROC analysis determined an optimal CCSTN cut-off of 0.00136 (cells/µL)/mL tumour to identify patients in the low expansion subpopulation (AUC: 91.7%). Patients with CCSTN above this cut-off had longer PFS and OS compared to patients with CCSTN below the cut-off value. In the Cox-proportional hazards model, a CCSTN value above the cut-off was associated with an 88% reduced risk of death.
Conclusions: Despite the small sample size, a population QSP model framework describing four CAR-T cell phenotypes and CD19+ metabolic tumour volume was successfully developed. We identified factors explaining 2/3 of the high IIV in CAR-T cell expansion and showed how the model could be used to identify factors allowing early survival prediction. Our clinical composite score, which is highly correlated with one of the model parameters but can be calculated without the model, provides non-modellers access to our early survival predictions. Nevertheless, due to the low sample size, our findings remain exploratory. Thus, future investigations with a much larger sample size are needed to test our hypotheses, potentially refine our model, and determine a robust CCSTN value.
References:
[1] F.L. Locke, A. Ghobadi, C.A. Jacobson, D.B. Miklos, L.J. Lekakis, O.O. Oluwole, Y. Lin, I. Braunschweig, B.T. Hill, J.M. Timmerman, A. Deol, P.M. Reagan, P. Stiff, I.W. Flinn, U. Farooq, A. Goy, P.A. McSweeney, J. Munoz, T. Siddiqi, J.C. Chavez, A.F. Herrera, N.L. Bartlett, J.S. Wiezorek, L. Navale, A. Xue, Y. Jiang, A. Bot, J.M. Rossi, J.J. Kim, W.Y. Go, S.S. Neelapu. Long-term safety and activity of axicabtagene ciloleucel in refractory large B-cell lymphoma (ZUMA-1): a single-arm, multicentre, phase 1–2 trial. Lancet Oncol. 20: 31–42 (2019).
[2] S.J. Schuster, M.R. Bishop, C.S. Tam, E.K. Waller, P. Borchmann, J.P. McGuirk, U. Jäger, S. Jaglowski, C. Andreadis, J.R. Westin, I. Fleury, V. Bachanova, S.R. Foley, P.J. Ho, S. Mielke, J.M. Magenau, H. Holte, S. Pantano, L.B. Pacaud, R. Awasthi, J. Chu, Ö. Anak, G. Salles, R.T. Maziarz. Tisagenlecleucel in adult relapsed or refractory diffuse large B-cell lymphoma. N. Engl. J. Med. 380: 45–56 (2019).
[3] S.S. Neelapu, F.L. Locke, N.L. Bartlett, L.J. Lekakis, D.B. Miklos, C.A. Jacobson, I. Braunschweig, O.O. Oluwole, T. Siddiqi, Y. Lin, J.M. Timmerman, P.J. Stiff, J.W. Friedberg, I.W. Flinn, A. Goy, B.T. Hill, M.R. Smith, A. Deol, U. Farooq, P. McSweeney, J. Munoz, I. Avivi, J.E. Castro, J.R. Westin, J.C. Chavez, A. Ghobadi, K. V. Komanduri, R. Levy, E.D. Jacobsen, T.E. Witzig, P. Reagan, A. Bot, J. Rossi, L. Navale, Y. Jiang, J. Aycock, M. Elias, D. Chang, J. Wiezorek, W.Y. Go. Axicabtagene Ciloleucel CAR T-Cell Therapy in Refractory Large B-Cell Lymphoma. N. Engl. J. Med. 377: 2531–2544 (2017).
[4] L.J. Nastoupil, M.D. Jain, L. Feng, J.Y. Spiegel, A. Ghobadi, Y. Lin, S. Dahiya, M. Lunning, L. Lekakis, P. Reagan, O. Oluwole, J. McGuirk, A. Deol, A.R. Sehgal, A. Goy, B.T. Hill, K. Vu, C. Andreadis, J. Munoz, J. Westin, J.C. Chavez, A. Cashen, N.N. Bennani, A.P. Rapoport, J.M. Vose, D.B. Miklos, S.S. Neelapu, F.L. Locke. Standard-of-care axicabtagene ciloleucel for relapsed or refractory large B-cell lymphoma: Results from the US lymphoma CAR T consortium. J. Clin. Oncol. 38: 3119–3128 (2020).
[5] S.J. Schuster, M.R. Bishop, C.S. Tam, P. Borchmann, U. Jaeger, E.K. Waller, H. Holte, J. McGuirk, S. Jaglowski, K. Tobinai, C. Andreadis, I. Fleury, S. Mielke, T. Teshima, J.R. Westin, V. Bachanova, S.R. Foley, J. Ho, J.M. Magenau, N. Wagner-Johnston, K. Kato, M.J. Kersten, K. van Besien, J. Chu, A. Jary, O. Anak, G.A. Salles, R.T. Maziarz. Long-Term Follow-up of Tisagenlecleucel in Adult Patients with Relapsed or Refractory Diffuse Large B-Cell Lymphoma: Updated Analysis of Juliet Study. Biol. Blood Marrow Transplant. 25: S20–S21 (2019).
[6] K.T. Mueller, S.L. Maude, D.L. Porter, N. Frey, P. Wood, X. Han, E. Waldron, A. Chakraborty, R. Awasthi, B.L. Levine, J.J. Melenhorst, S.A. Grupp, C.H. June, S.F. Lacey. Cellular kinetics of CTL019 in relapsed / refractory B-cell acute lymphoblastic leukemia and chronic lymphocytic leukemia. Blood 130: 2317–2326 (2017).
[7] A.M. Stein, S.A. Grupp, J.E. Levine, T.W. Laetsch, M.A. Pulsipher, M.W. Boyer, K.J. August, B.L. Levine, L. Tomassian, S. Shah, M. Leung, P.H. Huang, R. Awasthi, K.T. Mueller, P.A. Wood, C.H. June. Tisagenlecleucel Model-Based Cellular Kinetic Analysis of Chimeric Antigen Receptor–T Cells. CPT Pharmacometrics Syst. Pharmacol. 8: 285–295 (2019).
[8] S. Guedan, H. Calderon, A.D. Posey, M. V. Maus. Engineering and Design of Chimeric Antigen Receptors. Mol. Ther. - Methods Clin. Dev. 12: 145–156 (2019).
[9] T. Kaartinen, A. Luostarinen, P. Maliniemi, J. Keto, M. Arvas, H. Belt, J. Koponen, P.I. Mäkinen, A. Loskog, S. Mustjoki, K. Porkka, S. Ylä-Herttuala, M. Korhonen. Low interleukin-2 concentration favors generation of early memory T cells over effector phenotypes during chimeric antigen receptor T-cell expansion. Cytotherapy 19: 689–702 (2017).
[10] O.C. Finney, H. Brakke, S. Rawlings-Rhea, R. Hicks, D. Doolittle, M. Lopez, B. Futrell, R.J. Orentas, D. Li, R. Gardner, M.C. Jensen. CD19 CAR T cell product and disease attributes predict leukemia remission durability. J. Clin. Invest. 129: 2123–2132 (2019).
[11] C. Berger, M.C. Jensen, P.M. Lansdorp, M. Gough, C. Elliott, S.R. Riddell, others. Adoptive transfer of effector CD8+ T cells derived from central memory cells establishes persistent T cell memory in primates. J. Clin. Invest. 118: 294–305 (2008).
[12] X. Wang, L.L. Popplewell, J.R. Wagner, A. Naranjo, M.S. Blanchard, M.R. Mott, A.P. Norris, C.W. Wong, R.Z. Urak, W. Chang, S.K. Khaled, T. Siddiqi, L.E. Budde, J. Xu, B. Chang, N. Gidwaney, S.H. Thomas, L.J.N. Cooper, S.R. Riddell, C.E. Brown, M.C. Jensen, S.J. Forman. Phase 1 studies of central memory – derived CD19 CAR T – cell therapy following autologous HSCT in patients with B-cell NHL. Blood 127: 2980–2991 (2016).
[13] C.J. Turtle, S.R. Riddell, D.G. Maloney, C.J. Turtle, L. Hanafi, C. Berger, T.A. Gooley, S. Cherian, M. Hudecek, D. Sommermeyer, K. Melville, B. Pender, T.M. Budiarto, E. Robinson, N.N. Steevens, C. Chaney, L. Soma, X. Chen, C. Yeung, B. Wood, D. Li, J. Cao, S. Heimfeld, M.C. Jensen, S.R. Riddell, D.G. Maloney. CD19 CAR – T cells of defined CD4 + : CD8 + composition in adult B cell ALL patients. J. Clin. Invest. 126: 2123–2138 (2016).
[14] E.A. Dean, R.S. Mhaskar, H. Lu, M.S. Mousa, G.S. Krivenko, A. Lazaryan, C.A. Bachmeier, J.C. Chavez, T. Nishihori, M.L. Davila, F. Khimani, H.D. Liu, J. Pinilla-Ibarz, B.D. Shah, M.D. Jain, Y. Balagurunathan, F.L. Locke. High metabolic tumor volume is associated with decreased efficacy of axicabtagene ciloleucel in large B-cell lymphoma. Blood Adv. 4: 3268–3276 (2020).
[15] K.N. Kodumudi, A. Weber, A.A. Sarnaik, S. Pilon-Thomas. Blockade of Myeloid-Derived Suppressor Cells after Induction of Lymphopenia Improves Adoptive T Cell Therapy in a Murine Model of Melanoma. J. Immunol. 189: 5147–5154 (2012).
[16] J. Baba, S. Watanabe, Y. Saida, T. Tanaka, T. Miyabayashi, J. Koshio, K. Ichikawa, K. Nozaki, T. Koya, K. Deguchi, C. Tan, S. Miura, H. Tanaka, J. Tanaka, H. Kagamu, H. Yoshizawa, K. Nakata, I. Narita. Depletion of radio-resistant regulatory T cells enhances antitumor immunity during recovery from lymphopenia. Blood 120: 2417–2427 (2012).
[17] A. V. Hirayama, J. Gauthier, K.A. Hay, J.M. Voutsinas, Q. Wu, T. Gooley, D. Li, S. Cherian, X. Chen, B.S. Pender, R.M. Hawkins, A. Vakil, R.N. Steinmetz, U.H. Acharya, R.D. Cassaday, A.G. Chapuis, T.M. Dhawale, P.C. Hendrie, H.P. Kiem, R.C. Lynch, J. Ramos, M. Shadman, B.G. Till, S.R. Riddell, D.G. Maloney, C.J. Turtle. The response to lymphodepletion impacts PFS in patients with aggressive non-Hodgkin lymphoma treated with CD19 CAR T cells. Blood 133: 1876–1887 (2019).
[18] C.J. Turtle, L.A. Hanafi, C. Berger, M. Hudecek, B. Pender, E. Robinson, R. Hawkins, C. Chaney, S. Cherian, X. Chen, L. Soma, B. Wood, D. Li, S. Heimfeld, S.R. Riddell, D.G. Maloney. Immunotherapy of non-Hodgkin’s lymphoma with a defined ratio of CD8+ and CD4+ CD19-specific chimeric antigen receptor-modified T cells. Sci. Transl. Med. 8: 335RA116 (2016).
[19] F.L. Locke, J.M. Rossi, S.S. Neelapu, C.A. Jacobson, D.B. Miklos, A. Ghobadi, O.O. Oluwole, P.M. Reagan, L.J. Lekakis, Y. Lin, M. Sherman, M. Better, W.Y. Go, J.S. Wiezorek, A. Xue, A. Bot. Tumor burden, inflammation, and product attributes determine outcomes of axicabtagene ciloleucel in large B-cell lymphoma. Blood Adv. 4: 4898–4911 (2020).
[20] C. Liu, V.S. Ayyar, X. Zheng, W. Chen, S. Zheng, H. Mody, W. Wang, D. Heald, A.P. Singh, Y. Cao. Model-Based Cellular Kinetic Analysis of Chimeric Antigen Receptor-T Cells in Humans. Clin. Pharmacol. Ther. 109: 716–727 (2020).
[21] A.P. Singh, X. Zheng, X. Lin-Schmidt, W. Chen, T.J. Carpenter, A. Zong, W. Wang, D.L. Heald. Development of a quantitative relationship between CAR-affinity, antigen abundance, tumor cell depletion and CAR-T cell expansion using a multiscale systems PK-PD model. MAbs 12: 1688616 (2020).
[22] A. Chaudhury, X. Zhu, L. Chu, A. Goliaei, C.H. June, J.D. Kearns, A.M. Stein. Chimeric Antigen Receptor T Cell Therapies: A Review of Cellular Kinetic-Pharmacodynamic Modeling Approaches. J. Clin. Pharmacol. 60: S147–S159 (2020).
[23] A. V. Gett, F. Sallusto, A. Lanzavecchia, J. Geginat. T cell fitness determined by signal strength. Nat. Immunol. 4: 355–360 (2003).
[24] A. Lanzavecchia, F. Sallusto. Progressive differentiation and selection of the fittest in the immune response. Nat. Rev. Immunol. 2: 982–987 (2002).