ALBERT

Description: Abstract 2044 Current established protocols for the culture and differentiation of embryonic stem cells (ESCs) utilise two-dimensional (2D) tissue culture flasks/dishes. These culture methods are cumbersome and inefficient involving three stages: a) maintenance/expansion of undifferentiated ESCs, b) spontaneous differentiation through formation of embryoid bodies (EBs), and c) dissociation of EBs and replating leading to the terminal differentiation to the desired lineages. One of the major challenges in the use of ESCs for the production of red blood cells is controlling their differentiation pathway(s). Optimal culture conditions and requirements as well as precise differentiation mechanisms and cellular interactions within EBs are still not well characterised, resulting in sub-optimal control of homogenous differentiation especially due to the formation of all three germ layers. Furthermore, cavitation within EBs results in loss of available cell numbers, which reduces the yield and quality of the cellular product outcomes. To date, the most efficient protocols for the generation of oxygen-transporting, enucleated red blood cells from ESCs require co-culture with feeder cells and a multi-step process that lasts for approximately one month rendering such protocols difficult to scale-up. We have developed an integrated, single-step bioprocess that: a) uses conditioned medium (CM) derived from HepG2, a human hepatocarcinoma cell line, that stimulates mesoderm formation, b) facilitates 3D culture through encapsulation of undifferentiated mESCs in hydrogels, c) bypasses EB formation, and d) involves culture in a rotating wall vessel bioreactor that does not require passaging of the cells and is scalable and automatable. Previously, we have shown that in traditional 2D culture systems use of HepG2-CM facilitated early differentiation of mESCs into hematopoietic cells, as shown by expression of C-Myb, C-kit, Gata-2, SCL, and beta-globin genes, in comparison with that of control cultures. A significantly higher number (p≤0.001) of hematopoietic colonies was also achieved in conditioned medium-treated murine embryonic stem cells (CM)-mESC, at day 7 and 14, with a two-fold enhancement of all myeloid-erythroid progenitor colonies. Nucleated eythrocytes and macrophages were identified in the CM-mESC group as early as day 7 of culture. However, attempting to bypass the EB formation step in the 2D culture system did not produce any hematopoietic cells even by the conditioned medium-treated embryonic stem cells. Here, we now demonstrate, that single-step 3D cultures of encapsulated mESCs can produce hematopoietic cells bypassing EB formation. Specifically, undifferentiated mESCs were encapsulated (20,000 cells per hydrogel bead) and placed inside the rotating wall vessel bioreactor. The experimental group was exposed to conditioned medium supplemented with LIF for 3 days to stimulate mesoderm formation bypassing EB formation and terminal hematopoietic differentiation was accomplished by simply changing the culture medium and replacing it with 3U/ml hEPO and 40ng/ml mSCF. A significant increase in the number (p≤ 0.05) of hematopoietic colonies was observed from CM-mESCs at day 14, with a total five-fold expansion as well as enhancement of erythroid progenitors, BFU-E and CFU-E formation. A higher expression of hematopoietic genes, C-Myb, C-kit, Gata-2, SCL, as well as erythroid genes, EKLF and beta-major globin was also reported at 3 weeks of culture in low concentrations of cytokines in the bioreactor. Immunophenotypic analysis of the highly viable cells collected from the CM-mESCs group confirmed the positive expression of the proerythroblast markers (TER-119 and CD71). In conclusion, we have devised a scalable, automatable, single-step process for the derivation of mature erythrocytes from mESCs which may be used for further study of erythroid development and applications in the human system. Disclosures: No relevant conflicts of interest to declare.

Description: Abstract 276 Cord blood mononuclear cells (CBMNCs) are an ideal cell source for therapeutic applications although maintenance of fetal globin in expanded erythroid cells is problematic. We have previously developed a 3D bone marrow (BM) biomimicry through the use of a synthetic scaffold made of polyurethane (PU) coated with collagen type I which expanded CBMNCs in a cytokine-free environment for at least 28 days, with or without addition of serum. Addition of near physiological concentrations (0.2U/mL) recombinant human erythropoietin (rhuEPO) to the 3D CBMNCs serum- and cytokine-free cultures at day 7 enhanced normal erythropoiesis to enucleation (from CD45+/CD71+/CD235− to CD45−/CD71+/CD235+ stages) and promoted erythroid clonogenic capacity (CFU-E and BFU-E). However, mechanisms of erythroid expansion and globin-maturity of the cells are ill-defined. We have now extended our investigation to evaluate in situ hemoglobin and cytokine gene expression and detect cellular hemoglobin protein and cytokine production in culture supernatants. CBMNCs were separated by Ficoll-Paque density gradient centrifugation and seeded onto collagen-coated PU 3D scaffolds at 2.5×106cells/scaffold (5×5×5mm3). Cultures were established with full-medium exchange every other day in 4 conditions: 2 controls (serum-free and serum-containing cultures without rhuEPO) and serum-free cultures with addition of rhuEPO at 1.845U/mL or 0.2U/mL from day 7 onwards. Culture output was evaluated by in situ analysis and by physically extracting cells from the scaffolds. β-globin (HBB) and γ-globin (HBG) genes were detected by in situ PCR from day 0 and were consistently expressed throughout the culture period in all conditions (HBB expression was consistently higher). A decline in expression of both globins was noted at days 21 and 28 only in serum-free cultures without addition of rhuEPO. Western blot analysis of extracted cells confirmed both HBB and HBG expression in the serum-containing rhuEPO-free culture. In serum-free conditions, both HBB and HBG proteins were noted only at day 7; once rhuEPO was added to these cultures, a hemoglobin switch occurred with persistence of only HBB at days 21 and 28. Even in serum-free and rhuEPO-free cultures, HBB was present, although in lower amounts than in the rhuEPO-exposed cells, which suggested that fetal globin switching and maturation towards adult erythropoiesis occurred primarily in the serum-free cultures exposed to rhuEPO. Increased expression of GM-CSF, IL-1β and TNF was observed to day 7 and declined thereafter while IL-6 expression was observed only at day 7. There was constant high expression of TGF-β throughout the culture in all conditions, whereas TPO was not detected in any condition. The erythropoietin-receptor (EPO-R) gene was detected in all conditions, yet EPO-R was shed into the supernatant mostly in the first 7 days (maximum 42pg/ml) of serum-free cultures, declining by day 14, which corresponded to the timing of rhuEPO addition. Flt-3 was consistently detected in supernatant of the serum-containing cultures, yet declined after day 14 in serum-free cultures with and without rhuEPO. Stem cell factor (SCF), critical in early stages of erythropoiesis, was not present in culture supernatants at any time-point. Interestingly, endogenous production of EPO (maximum 0.8 mU/ml) was detected by ELISA in the first 2 weeks of both serum-containing and serum-free cultures, prior to addition of rhuEPO. High EPO concentration (maximum 2600mU/ml) was still observed in the serum-free, rhuEPO-containing cultures over the entire 28 days, suggesting that EPO was not completely utilized by the maturing cells during culture and that even lower concentrations of EPO could be beneficial. EPO detection was maximal at day 21 for serum-free cultures exposed to rhuEPO; this day 21 peak, in conjunction with the known erythroid maturation kinetics within the 3D culture, the detection of endogenous EPO production at day 14, the shedding profile of EPO-R and the hemoglobin switch, suggested that the 14–21 day time-points of culture will be important for the future study of erythroid physiology within this system. In conclusion, the 3D BM biomimicry is a good model to study erythropoiesis ex vivo, using physiological concentrations of rhuEPO and serum-free conditions rendering it suitable for future clinical applications. Disclosures: No relevant conflicts of interest to declare.

Description: Chemotherapy for AML is currently dosed using body surface area (BSA) formulas. For overweight patients (defined as actual body weight 〉20% that of ideal), calculations based on ideal body weight (IBW) are used in order to limit potential toxicity [1]. However, an analysis of the consequences of this adjustment on leukemic and normal bone marrow hematopoietic (HC) cells hasn't been performed. We have developed a mathematical model to simulate personalized chemotherapy for AML [2,3]. The model is used to predict outcomes with standard 3+10 DA treatment [3 doses of 60mg/m2 daunorubicin (DNR) 1h on days 1-3-5 and 2 pulses a day of cytarabine (Ara-C) 100mg/m2 days 1-10]. In this study, we focus on the impact of using ideal vs actual body weight (ABW) for the calculation of BSA using Mosteller's equation on both leukemic and normal HC based on simulating treatment response with our mathematical model. Data were obtained from patients with AML undergoing standard chemotherapy with DA. Cell cycle kinetics of 12 patients was obtained based on matching the % blasts measured on BM biopsies before chemotherapy start and after recovery, which correlates with total cell number in the cell cycle model. Treatments were then simulated based on drug dosage for IBW and ABW. The outcomes (number of leukemic cells remaining) in both cases are recorded after the resting period (Fig. 1). According to these results, patients with ABW closer to IBW will have the same outcome since the BSA is similar; patients with higher ABW will have a better outcome when the ABW is used for the BSA dose-based chemotherapy calculation compared with the outcome obtained by using the IBW; patients with lower ABW than IBW will have a better outcome compared with that obtained with the IBW drug calculation. Seven additional hypothetical patients were simulated in order to validate the results obtained with actual patients; they all fell within the 95% confidence region of the logarithmic fit observed. In order to determine toxicity, as defined by time to normal cell recovery after chemotherapy, HCs were simulated for all 12 patients on both IBW and ABW dosage schemes (Fig. 2). For most patients, the final HC outcome on the ABW was within 10% of the outcome on the IBW dose. Only for one patient was the outcome using ABW significantly lower (25%) than for that using IBW. Put together with the information from Figure 1, these results suggest that DA chemotherapy based on ABW instead of IBW calculations for patients whose weight is over 20% of IBW have a positive impact on reducing leukemic burden, while not significantly affecting HC recovery. Only for very extreme cases (e.g. P12), is the HC recovery impaired, in which case DA dosage based on IBW instead of ABW is preferable. There is an unmet need to standardise dosing of chemotherapy to achieve the best anti-leukemic effect and to quantify (and limit) potential toxicities. The use of IBW for dosing of chemotherapy in AML patients may result in under-treatment and poorer outcomes. Our model suggests that the use of ABW for dose-determination could improve treatment outcomes in AML in terms of leukaemia cell kill but not at the expense of normal HC recovery, except in extreme cases. Ultimately, the use of mathematical models predicting disease progression and targeted treatment outcomes will be critical to realise the potential of precision medicine for the treatment of AML and other cancers. References 1. Berger, N.A., A time to stop, a time to start: high-dose chemotherapy in overweight and obese patients. Bone Marrow Transplant, 2015. 50 (5): p. 617-8. 2. Panoskaltsis, N., et al., Optimized Patient- and Leukemia-Specific Chemotherapy Protocols For The Treatment Of Acute Myeloid Leukemia. Blood, 2013. 122 (21). 3. Fuentes-Gari, M., et al., A mathematical model of subpopulation kinetics for the deconvolution of leukaemia heterogeneity. Journal of the Royal Society Interface, 2015. 12 (108): p. 20150276. Figure 1. Ratio of predicted leukemic cell outcomes (#leukemic cells remaining) for ABW and IBW, versus ratio of BSA to ideal BSA, after the last resting period for 12 patients, with ABW within 20% or over 20% IBW. Logarithmic fit: y-0.274 ∙ ln(x)+0.9844 (95% confidence regions) Figure 1. Ratio of predicted leukemic cell outcomes (#leukemic cells remaining) for ABW and IBW, versus ratio of BSA to ideal BSA, after the last resting period for 12 patients, with ABW within 20% or over 20% IBW. Logarithmic fit: y-0.274 ∙ ln(x)+0.9844 (95% confidence regions) Figure 2. Ratio of predicted normal HC cell outcomes (#normal cells remaining) for ABW and IBW, versus ratio of BSA to ideal BSA, after the last resting period for 12 patients, with ABW within 20% or over 20% IBW. y=0.7488 ∙ ln(x)+1.0035; (95% confidence regions) Figure 2. Ratio of predicted normal HC cell outcomes (#normal cells remaining) for ABW and IBW, versus ratio of BSA to ideal BSA, after the last resting period for 12 patients, with ABW within 20% or over 20% IBW. y=0.7488 ∙ ln(x)+1.0035; (95% confidence regions) Disclosures No relevant conflicts of interest to declare.

Description: Treatment for Acute Myeloid Leukaemia (AML) with chemotherapy may result in acute and long-term life-threatening complications due to drug toxicity. Only relatively few patient-and leukemia-specific factors are taken into consideration in current protocols and choice of treatment often depends on the treating physician’s experience. With the advent of novel treatments and large amounts of patient- and leukemia-specific genomic data, there is a clear need for a systematic approach to the design and execution of chemotherapy regimens. Mathematical modeling is a tool that can be used for the automation of chemotherapy treatment due to its advantages in systematically exploring extensive datasets in order to capture a system’s dynamics and subsequently provide better insight for process enhancement. An AML-specific model that combines the leukemia-specific actions of the cell cycle (i.e. drug target) with patient-specific pharmacology of the drugs (pharmacokinetics) was developed. Specifically, it simulates the response of patients with AML undergoing treatment with two standard chemotherapy protocols, one intensive and the other non-intensive: (a) Daunarubicin (DNR) and Cytosine Arabinoside (Ara-C) used in standard intravenous (iv) doses (DA – 3+7 or 3+10) and (b) low dose Ara-C (LDAC) administered subcutaneously (sc). Sensitivity analysis of the model identifies cell cycle times as the critical parameters that control treatment outcome. For model analysis, clinical data from 6 patients who underwent chemotherapy are used for the estimation of cell cycle time distribution. The patient data are comprised of disease characteristics (tumor burden, cell cycle times, normal cell population) as well as patient-specific characteristics (gender, age, weight and height). The estimated mean S-phase duration (Ts) is 15 hrs (range: 9-21 hrs) and mean whole cell cycle duration (Tc) is 47.5 hrs (range: 33-68 hrs). The estimated data reveal a clear relationship of cell cycle times to treatment outcomes. Specifically, low Ts duration combined with high Tc duration indicates worse treatment outcomes, whereas, the reverse combination is indicative of a good response to treatment. In order to improve effectiveness of AML therapy and reduction of toxicity, chemotherapy treatment is presented as an optimal control problem with the main aim of obtaining a treatment schedule which could maximize leukemic cell kill yet minimize death of the normal cell population in the bone marrow. By the end of treatment, the leukemic population should be reduced to a level of approximately 109 cells at which point BM hypoplasia is achieved. Out of the 6 patients studied, 2 patients had a successful treatment with leukemic hypoplasia achieved, 2 had a reduction of leukemic cells without achieving the hypoplasia target and 2 had disease progression on chemotherapy. The optimization algorithm is formulated and solved for all patients for both intensive and non-intensive treatment protocols with maximal and minimal thresholds set for efficacy and toxicity, respectively. For iv Ara-C, total drug administration is set between 50mg – 4000mg with infusion duration between 1 min to 24 hours. The window for DNR dose optimization is stricter due to potential toxic effects and the only independent variable is dose with 30mg – 90mg per infusion. For sc Ara-C, the maximum dose per day is 40mg and doses are permitted up to four times daily for a maximum period of 20 days. Optimization results obtained for the 6 patients indicate that continuous infusions are more effective for leukaemia cell kill than rapid infusions. For non-intensive chemotherapy, 40mg of Ara-C in continuous infusion over 10 days is better than daily divided doses with leukemic hypoplasia achieved for all patient case studies. For the intensive protocol, dose increase of DNR to 90 mg/m2 combined with Ara-C daily infusion is the optimal chemotherapy regimen. Ara-C doses differ between patients and the optimal dose range is between 200 to 250 mg/m2. In summary, this work presents the potential for improved treatment design in AML therapy, dependent on disease and patient characteristics, defined on a case-by-case basis. This design would provide the opportunity to personalize treatment protocols for gold standard intensive and non-intensive therapies as well as for novel drugs through the use of optimization methods. Disclosures: No relevant conflicts of interest to declare.

Description: Acute Myeloid Leukemia (AML) is a heterogeneous and complex malignancy of the blood and bone marrow (BM). Current treatment options only cure 40% of patients under the age of 65 years and, over the age of 65 years only 10% will be cured. Hence, there is great urgency to further understand pathogenesis, optimize treatment and identify markers for diagnosis, prognosis and targets for novel therapies. It is increasingly evident that metabolism and BM microenvironment are key factors in AML pathogenesis and progression - entities that cannot be accurately investigated within current in vitro methods of two-dimensional (2D) culture. The metabolome is the final downstream product of gene transcription that also responds dynamically to its environment. Metabolomics is an established technique to identify biomarkers with prognostic relevance, such as 2-hydroxygluterate, an onco-metabolite derived from IDH mutations in AML. Targeting metabolic pathways such as oxidative phosphorylation, glycolysis and specific mutations that affect metabolism such as isocitrate dehydrogenase are currently being assessed in clinical trials. An in vitro system that could more accurately represent the in vivo BM may help optimise these metabolically-targeted regimens. Here we show that a previously established three-dimensional (3D) culture within our laboratory recapitulates elements of BM structure enabling long-term culture of AML cell lines without disrupting metabolism, as is the case with serial passage in 2D culture. Metabolism was assessed using gas chromatography-mass spectrometry (GC-MS) metabolomics analysis which reflects the phenotype of the leukemic cell lines. Porous polyurethane scaffolds were fabricated using dioxin by thermally induced phase separation as previously described; scaffolds have a pore size of 100-250µm and a porosity of 90-95% cut to 0.5cm3, coated with type I collagen and seeded with 0.5x106cells of either K562 (erythroleukemia) or HL-60 (promyelocytic leukemia) cells, after expansion in 2D culture. In parallel, 2D cultures of the same cell lines were maintained using passage every 2 days. K562 and HL-60 cells were cultured in IMDM, 1% penicillin and streptomycin and 20% or 10% foetal bovine serum, respectively. Serial sampling occurred at 4 time points over 21 days from 10 scaffolds of each condition. Cells were extracted from the 3D scaffolds using two techniques - needle extraction or TrypLE express (Thermo Fisher Scientific) to assess whether these physical or chemical extraction methods disrupt the metabolome. Once cells were extracted, they were suspended in methanol at 1x106cells/mL cold methanol for metabolite quenching; metabolites were subsequently extracted with methanol/water and derivatised with Methoxamine and N-Methyl-N-(trimethylsilyl)trifluoroacetamide. Metabolomics analysis was then performed using a Shimadzu QP2010 Ultra GC-MS machine detecting 138 metabolites including those of pertinent pathways: glycolysis, tricarboxylic acid, pentose phosphate, urea, glutaminolysis and amino acids. Bioinformatics analysis was performed with MeV TM4(http://mev.tm4.org). Using unsupervised clustering techniques, including hierarchical clustering and principal component analysis, we identified that the metabolome of TrypLE express and needle-extracted cells from 3D scaffolds have a similar metabolic signature and group closely with each other as well as with the metabolome of the seeded cells (day 0); they do not vary significantly over the 21 days of culture. Conversely, with each passage, the metabolomes of the 2D-cultured cells differ to those of day 0 and vary to each other over the same 21-day period. These results highlight limitations in the use of 2D cultures to address AML biology as metabolic changes with passage reflect change in phenotype. Based on these findings, we conclude that 3D cultures provide a more stable environment for leukemic cell culture and assessment of leukemia biology, irrespective of method of cell extraction. The 3D culture platform is more suited than standard 2D in vitro cultures to investigate metabolism, microenvironment and drug targets in AML. Disclosures No relevant conflicts of interest to declare.

Description: Reproduction of dynamic physiologic erythropoiesis in vitro requires a three-dimensional (3D) architecture, erythroblast-macrophage interactions and cytokines such as erythropoietin (EPO). The role of oxygen concentration gradients in this process is unclear. We have created a 3D bone marrow (BM) biomimicry using collagen-coated polyurethane scaffolds (5mm3) to expand cord blood mononuclear cells (CBMNCs) in a cytokine-free environment for 28 days (D). Addition of EPO to this system induces mature erythropoiesis. We hypothesised that physiologic concentrations of cytokines - stem cell factor (SCF) / EPO - and a hypoxia (H)/normoxia (N) schedule to mimic BM oxygen gradients would enhance erythropoiesis. CBMNCs were seeded (4x106 cells/scaffold) in 3D serum-free cultures supplemented with 10ng/mL SCF (D0-D28), and 100mU/mL EPO (D7-D28), with medium exchange every 3D. Three conditions were compared: N (20%), H (5%) and 2-step oxygenation HN (H D0-D7 and N thereafter). Erythroid maturation was monitored weekly by flow cytometry (CD45/CD71/CD235a) both in situ (i.e., in scaffolds) and in supernatant (S/N) cells. D0-7 H was more efficient in early induction of CD235a in the absence of exogenous EPO (H 13% vs N 8% CD45loCD71+CD235alo cells, p

Description: Erythropoiesis is regulated by microenvironmental factors including bone marrow (BM) architecture, cell-cell and cell-matrix interactions, growth factors, oxygen, etc. Traditional 2D hematopoietic cell cultures require high concentrations of exogenous cytokines and neglect BM architecture which favours formation of niches, including erythroblastic islands, wherein paracrine and autocrine communication occurs. Consequently, these 2D systems do not facilitate the study of erythropoiesis within a biomimetic hematopoietic microenvironment, are costly due to high concentrations of cytokines required, and not scalable. We have previously created a 3D BM biomimicry using porous collagen-coated polyurethane scaffolds for culture of human cord blood mononuclear cells (MNCs) in cytokine-free and serum-free conditions for 28 days. In addition, low concentration (200 mU/mL) of EPO was shown to enhance erythroid differentiation. We now extend our studies to evaluate the role of oxygen and the microenvironmental niches in supporting erythropoiesis. MNCs were isolated from cord blood and seeded into the 3D system (4x106 cells/scaffold). Serum-free cultures were supplemented with near physiologic concentrations of cytokines: 10 ng/mL SCF (D0-D28), and 100 mU/mL EPO (D7-D28), with medium exchange every 3 days. Three conditions were investigated: normoxia (20%), hypoxia (5%) and 2-step oxygenation (hypoxia from D0-D7 and normoxia thereafter). Proliferation was evaluated in the scaffolds in situ, of the cells released into the supernatant, and of cells extracted from scaffolds at different time-points. Maximum cell expansion was detected in situ at D14 under normoxic (1.6±0.3-fold) and 2-step oxygenation (1.8±0.3-fold) conditions (p

Description: Abstract 341 Ex vivo expansion of cord blood mononuclear cells (CBMNCs) could provide a safe, flexible and ample supply of blood components for cellular therapies. Traditionally, hematopoietic cell expansion has been performed in 2D tissue culture flask or well-plate static cultures using abnormally high concentrations of cytokines which is expensive, reduces the self-renewal capacity, and skews normal differentiation. We have previously developed a 3D bone marrow biomimicry through the use of a synthetic scaffold made of polyurethane (PU) coated with collagen type I which could expand CBMNCs in a cytokine-free environment for at least 28 days ex vivo, with or without the addition of serum to the media. We hypothesised that the addition of near physiological concentrations (0.2U/mL and 1.845U/mL) of exogenous erythropoietin (EPO) to these established 3D CBMNC ex vivo cultures at day 14 in a serum-free and cytokine-free environment would be sufficient to enhance erythropoiesis. CBMNCs were separated by Ficoll-Paque density gradient and seeded onto collagen-coated PU 3D scaffolds at a cell density of 2.5×106cells per scaffold (5×5×5mm3). Cultures were established in serum-free conditions and only EPO was added at days 14–28, with full-medium exchange every 2 days. Culture output was evaluated at days 14, 21 and 28 both by physically extracting cells from the scaffolds and by in situ analysis. Over 28 days, most stages of maturation, from erythroid progenitors to enucleated erythrocytes were observed by light microscopy of cytospins and by immunophenotypic analysis of extracted cells (CD45−/CD71+/CD235+), with more maturation occurring by day 28 of culture, after the addition of EPO. Although both concentrations of EPO produced comparable erythroid differentiation of cells, even by CFU assay, the viability (75% vs. 61%, p

Description: Patients with AML have heterogeneous features, including those specific to the patient as well as those specific to the disease, such as leukemic burden and dynamic sub-clonal populations. Outside of clinical trials, few of these components are used to determine treatment. In order to move towards precision medicine, we have developed πiChemo, a computational application based on a dynamic mathematical modelling framework, using patient-, leukemia- and treatment-specific data to predict outcomes and optimize chemotherapy regimens for patients with AML. The model consists of a pharmacokinetic and pharmacodynamic (PK/PD) module that calculates the concentration and effect of Cytarabine Arabinoside (Ara-C) and Daunorubicin (DNR) in bone marrow (BM); and a population balance models (PBMs) module that describes normal populations (stem cells, progenitors, precursors) and abnormal populations (leukemic sensitive blasts (LSB) and leukemic resistant blasts (LRB)) in BM. The PBMs module also determines mature cell numbers in three lineages found in BM and peripheral blood (PB): (1) red blood cells (RBC), (2) white blood cells (WBC) and (3) lymphocytes (L). Model structure was analysed by global sensitivity analysis, which identified the most significant parameters on outcome predictions, re-estimated for each patient. The final integrated PK/PD & PBMs model has 1,295 differential equations, 8,044 algebraic equations, 9,335 variables, 25 fixed parameters and 4 degrees of freedom or variables to be optimized (Ara-C dose, Ara-C injection duration, DNR dose and DNR injection duration). Model validation, predictions and optimizations were performed using anonymised retrospective data from 28 patients with AML. The model required: (i) patient features: height, weight, age and gender, (ii) patient status: initial BM differential and PB cell counts, (iii) leukemia data: cellularity, presence of dysplasia and initial blast percentage and, (iv) treatment data: type (low-dose (LD) or intensive (DA)), dose, administration route (SC vs IV), administration mode (bolus injection vs infusion), time between injections and between cycles. The model predicted the absolute numbers of stem cells, progenitors, precursors, WBC, RBC, L, LSB and LRB in BM, and WBC, RBC, L and neutrophil count in PB during treatment for all patients. Model simulations predicted outcomes for 18 patients who achieved complete remission (7 LD & 11 DA), 4 patients who entered partial remission (2 LD & 2 DA) and 6 patients who relapsed (2 LD & 4 DA). The most remarkable results are those of prediction for BM blast percentage after each chemotherapy cycle and the PB neutrophil count for all patients. The notable fit between model predictions and daily patient data demonstrate model robustness and accuracy in the capacity to track patient-specific restaging BM and daily PB count evolution before, during and after treatment. The same patient datasets were used to apply an optimization algorithm that could maximize normal cell number and reduce leukemia burden, to personalize chemotherapy dose and administration for best outcomes. The results show that doses and administration methods vary between patients and between chemotherapy cycles for the same patient, depending on the evolution of normal and abnormal populations in BM. Low-dose continuous Ara-C infusions were more effective than rapid bolus injections, due to reduced chemotherapy effects on normal cells and subsequent quicker recovery in the normal BM compartments. RBC progenitors and precursors recovered faster than WBC and L lineages, and the recovery of normal BM cells was faster than that of normal mature cells in PB. The πiChemo tool requires only patient- and leukemia-specific initial conditions at diagnosis, easily obtained in standard clinical practice, for outcome predictions and treatment optimizations. Real-time model-fit testing and comparison of model results against daily PB cell counts would enable the re-estimation of significant parameters, increasing model accuracy and treatment effectiveness whilst therapy is ongoing. The πiChemo precision therapy tool has the potential to personalize optimal standard and novel treatments for AML in real-time. Disclosures No relevant conflicts of interest to declare.

Description: Current in vitro human erythroid culture platforms require abnormally high cytokine supplementation and use lower cell density (107/mL) when perfused with cytokine-free media in long-term culture. In order to study the role of this manufactured HFBR microenvironment on spatiotemporal physiologic erythropoiesis, we now extend our previous reports by implementing 5-fold less cytokine concentrations than those used in typical ex vivo erythropoietic cultures. Herein, we show a 〉107 red cell harvest from the HFBR culture over 28-days with spontaneous expansion of stromal cells, maintenance of erythroid progenitor pools, and formation of stromal and erythroid cell niches in defined areas within the HFBR structure with differential in situ production of 23 growth factors varying over time. The 5.25 mL HFBR scaffold was inoculated with 108 CBMNCs and HFs were rapidly perfused (20 mL/h) with serum-free StemSpan medium gradually supplemented with a cytokine gradient of decreasing SCF (50 - 0 ng/mL) and increasing EPO (0 - 0.3 U/mL) over 28 days in order to maintain progenitor cells whilst inducing erythropoiesis. Quantitative confocal microscopy analyses of HFBR sections demonstrated that DAPI+ CBMNCs maintained high cell density (〉107/mL), and high viability (〉80%), while more than 107 enucleated cells were filtered through HFs over the 28-day culture. Inside the HFBR, hematopoietic progenitor cells were maintained (total of 3.1∙106 CD34+ and 5.5∙106 CKIT+ MNCs) while erythroid cells were expanded across various stages of maturation (28-day total increase of 1.2∙107 EPOR+, 1.8∙107 CD71+, and 2.3∙107 CD235a+ MNCs); CD235a+mature red cell phenotypes were enriched 10-fold in the HF filtrate over 28 days. Stromal cells expanded and differentiated during the 28-day HFBR culture with a total increase of mesenchymal stem cell marker Stro-1 (2.2∙107 cells), pre-osteoblast marker osterix (OSx; 1.6∙107 cells), and mature osteoblast marker osteopontin (OPN; 0.5∙107 cells). Expression of human collagen-1, fibronectin, and laminin-2 was detected by microscopy, while enzyme-linked immunoassays on HFBR filtrate detected 23 multilineal, unsupplemented cytokine profiles including interleukins produced primarily from day 0-12 (IL-6, IL-10, IL-21) as well as colony stimulating factors and stromal growth factors which increased in production from day 20-28 (G-CSF, GM-CSF, EGF, VEGF, Ang-2, PDGF, FGF-β). Using a novel confocal microscopy computational analyses that we have developed, DAPI+MNCs were found to self-associate into expanding 50-500µm clusters throughout the 28-day culture which increased local cell density 10-20 fold, representing niche-like areas. At day 14 and 28, MNCs formed clustered niches far from HFs which expressed hypoxic (HIF1a, PIMO), stromal, and erythroid markers (Stro-1, OSx, collagen-1, laminin-2, VCAM-1, CD45, EPOR: 〉1400µm from HFs). At day 28, 3-fold more MNC clusters formed near HFs and were comprised of hematopoietic progenitor and erythroid phenotypes (CD45, CD34, CKIT, CD235a, CD71: