EMIM 2018 ControlCenter

Online Program Overview Session: PS-08

To search for a specific ID please enter the hash sign followed by the ID number (e.g. #123).

Imaging Therapies

Session chair: Emmanuelle Canes-Soulas - Lyon, France; Claus Gluer - Kiel, Germany
Shortcut: PS-08
Date: Wednesday, 21 March, 2018, 4:00 PM
Room: Lecture Room 4/5 | level -1
Session type: Parallel Session


Click on an contribution preview the abstract content.

4:00 PM PS-08-1

Introductory Lecture by Marleen Keyaerts - Brussels, Belgium

This talk provides an overview of state-of-the-art research and refers to the following presentations selected from abstract submissions.

4:20 PM PS-08-2

Discriminating Radiation Necrosis from Recurrent Tumor with [18F]PARPi and Amino Acid PET in Mouse Models (#516)

P. L. Donabedian1, S. Kossatz1, B. Carney1, 2, J. A. Engelbach3, 4, S. A. Jannetti2, 1, W. Weber1, J. Garbow3, 4, T. Reiner1

1 Memorial Sloan-Kettering Cancer Center, Radiology, New York, New York, United States of America
2 Graduate Center of the City University of New York, Ph.D Program in Chemistry, New York, New York, United States of America
3 Washington University, Radiology, St. Louis, Missouri, United States of America
4 Washington University, Alvin J. Siteman Cancer Center, St. Louis, Missouri, United States of America


Radiation necrosis can appear similar to recurrent tumor on standard imaging. Current algorithms for this differential diagnosis require one or more follow-up imaging studies, long dynamic acquisitions, or complex image postprocessing. The best PET imaging protocol available for this setting is amino acid PET, which has not seen widespread adoption in the clinical environment. Using mouse models of both glioblastoma and radiation necrosis, we tested the potential of poly(ADP-ribose) polymerase (PARP)-targeted PET imaging with [18F]PARPi to better discriminate radiation necrosis from tumor.


The mouse model of radiation necrosis was generated by 60-Gy stereotactic irradiation of the left cerebral hemisphere. The glioblastoma mouse model was generated by intracranial injection of U251 MG cells.  PARP1 expression of both models was determined using immunohistochemistry. We chose [18F]Fluoroethyltyrosine ([18F]FET) as an amino acid PET tracer.  Both tracers were synthesized using cyclotron-produced fluorine-18, and DCE-MR and PET/CT imaging carried out on paired cohorts of mice. Analysis of PET data was carried out to quantify differences in tracer uptake in lesioned and contralateral brain regions between [18F]FET-PET and [18F]PARPi-PET in both glioblastoma and radiation necrosis mice.



In mice with experimental radiation necrosis, lesion uptake on [18F]PARPi-PET was barely higher than contralateral (lesion: 0.23±0.29 %ID/ccmean; contralateral: 0.15±0.20 %ID/ccmean; p = 0.73), while [18F]FET-PET clearly delineated the contrast-enhancing region on MR with a region of uptake averaging 1.7 times background (lesion: 9.7±3.2 %ID/ccmean; contralateral: 5.9±1.7 %ID/ccmean; p = 0.14). In mice bearing focal intracranial U251 xenografts, Visual delineation of the tumor from background was much easier on PARPi-PET than FET-PET, and both lesion-to-contralateral activity ratios (max/max, p = 0.034) and tumor-to-background ratios (max/mean, p = 0.009) were higher on PARPi-PET than FET-PET.


We present preclinical data showing that experimental murine radiation necrosis is not significantly [18F]PARPi-avid, and that [18F]PARPi-PET outperforms [18F]FET-PET in distinguishing radiation necrosis from focal intracranial xenografts. Efficient discrimination between recurrent tumor and radiation necrosis represents an added value for [18F]PARPi-PET used for postsurgical treatment planning and patient selection for treatment with pharmacological or radiotherapeutic PARP inhibitors.


1. Walker, A. J. et al. Postradiation imaging changes in the CNS: how can we differentiate between treatment effect and disease progression? Futur. Oncol. 10, 1277–1297 (2014).

2. Carney, B. et al. Non-invasive PET Imaging of PARP1 Expression in Glioblastoma Models. Mol. Imaging Biol. 18, 386–392 (2016).

3. Jiang, X. et al. A gamma-knife-enabled mouse model of cerebral single-hemisphere delayed radiation necrosis. PLoS One 10, 1–13 (2015).


This work was supported by the National Institutes of Health grants R01CA204441, R21CA191679 and P30CA008748. We also thank the MSKCC Center for Molecular Imaging and Nanotechnology, Small Animal Imaging Core Facility, Radiochemistry & Molecular Imaging Probes Core Facility and Molecular Cytology Core Facility.

DCE-MR and PARPi- and FET-PET/CT of radiation necrosis and tumor.
Left: Axial slices of DCE-MR (left columns) and fused PET/CT images (right columns) of animals with radiation necrosis (A) or U251 tumors (C) injected with [18F]PARPi (top rows) or [18F]FET (bottom rows). Right: Lesioned-to-contralateral (L/CL) whole-hemisphere %ID/ccmax ratios in radiation necrosis (B), U251 tumor (D), and treatment naive mice [18F]FET-PET and [18F]PARPi-PET images.
Keywords: radiation necrosis, PET imaging, PARP, amino acids
4:30 PM PS-08-3

Omics landscape of FDG-PET based heterogeneity in solid tumors. (#514)

M. A. Jarboui1, J. A. Disselhorst1, M. A. Krueger1, C. Trautwein1, P. Katiyar1, B. J. Pichler1

1 Werner Siemens Imaging Center, Eberhard Karls University Tuebingen, Tuebingen, Baden-Württemberg, Germany


The heterogeneity observed within solid tumors and the myriad of genetic alterations shape tumors evolution and their molecular environment. Despite the development of quantitative analytical methods in molecular system biology, several studies struggle to capture the biological signature of tumor heterogeneity. Accordingly, we developed an image-guided milling machine (IGMM) that allows for accurate delineation of tumor tissues based on tracers uptake with subsequent multiplex omics analysis.


FDG-PET imaging was performed on an MMTV-PyMT breast cancer mouse model with highly heterogeneous tumors. Imaging data were used to delineate 131 regions of interest (ROIs) with differential FDG uptake in 24 mice. ROIs were isolated with our designed IGMM. Using a multiplex extraction method, metabolites and total proteins were isolated. Metabolites were analyzed using either, the Biocrates targeted metabolomic platform where metabolites were detected by LC-MS analysis (ABSCIEX QTRAP 6500) or by untargeted 1H NMR spectroscopy (600 MHz Bruker Avance III). Equal amounts of total protein from each ROI were acquired using the high-resolution LC-MS LTQ Orbitrap Fusion (Thermo Scientific). Acquired MS spectra were further analyzed using Maxquant label-free quantification algorithm (LFQ).


Unbiased PCA analysis of high versus low FDG uptake regions showed a clear separation based on protein and metabolites abundance. We detected alterations of specific tumor driver proteins, transcription factors, and interferon-induced proteins. The expression of the transcription factor Sp1, a hallmark of oncogenesis and zinc finger proteins, the largest transcription factor family, correlate with increased FDG uptake. Targeted metabolomic analysis showed an increase in glycogenic amino-acids and polyamines. While spermine, spermidine, kynurenine and putrescine levels positively correlate with FDG uptake, histamine and serotonin were negatively correlated. Free carnitine and short chains acylcarnitines were abundant in high uptake regions, unlike the long chains. Untargeted NMR data are currently under investigation, partial analysis of 10 samples subset of 10 metabolites show a strong positive correlation of taurine,  alanine, creatine/creatinephosphate and lactate with FDG uptake.


Our analysis shows a specific proteomic and metabolomic profile defined by the FDG-PET uptake of tumor regions within oncogenic tissue. Regions of high FDG uptake were characterized by high transcriptional activation, increased beta-oxidation, active amino-acid metabolism and polyamines accumulation. As we are finalizing data analysis and integration, our investigation provides a detailed molecular landscape of solid tumor heterogeneity and is the first attempt to use imaging-guided isolation of tumor regions in tandem with proteomics/metabolomics investigation.

Keywords: Proteomics, Metabolomics, cancer, imaging
4:40 PM PS-08-4

Effects of spacer region on the development of an αCD20 CAR construct (#519)

W. Al Rawashdeh1, J. Brauner1, S. Rüberg1, N. Mockel-Tinbrink1, T. Toepfer1, C. Barth1, D. Lock1, D. Schneider2, G. Rauser1, M. Jurk1, A. Kaiser1, O. Hardt1

1 Miltenyi Biotec, Bergisch Gladbach, Germany
2 Lentigen Corporation, Gaithersburg, United States of America


Chimeric antigen receptor (CAR) modified T cells have emerged as the hottest topic in cellular immunotherapy as of late. T cells transduced with second generation CARs have demonstrated remarkable complete remission rates in patients with refractory B cell malignancies and the potency and persistence of CAR T cells have been shown to depend, in part, on the activating and co-stimulatory domains of the CAR construct. However, the choice of the spacer is also critical. Here, two αCD20 CAR T cell constructs, differing from the spacer region, were investigated as preclinical candidates.


Autologous T cells were transduced using lentiviral vectors to express one of two αCD20 CARs with different spacer domains using a 12-day TCT process on the CliniMACS Prodigy®, which performs all manufacturing steps in a single automated and closed system1. Cytotoxicity and cytokine release tests were assessed in vitro1. In vivo efficacy and persistence studies were performed using NSG mice and i.v. injection of hCD20+ lymphoma CDX (Raji), while sensitivity and specificity studies were performed using s.c. co-injection of wildtype and CD20-transduced melanoma CDX (Mel-526) in different percentages along with the CAR T cells. Therapy dose was normalized to 1x106 CD20 CAR+ cells in all in vivo studies. In vivo imaging was performed using IVIS Lumina III and XenoLight RediJect D-Luc Ultra.


TCT runs yielded ~6x109 CD3+ cells (viability ≥ 97%) and transduction efficiencies ˃10% (release specification) were always achieved. In vitro, cytokine release and cytotoxicity upon co-culture with CD20+ cells were comparable for both constructs (Fig. 1). However, CART-1 failed to produce any therapeutic effect in 2 independent in vivo studies (n=5, n=5) revealing a tumor burden similar to Mock-treated (n=5, n=7) and untreated controls (n=6, n=7) (Fig. 2). In contrast, CART-2 , efficiently eradicated hCD20+ cells (n=6, n=7). CART-2 cells persisted in vivo and were detected in blood, spleen and bone marrow many weeks post therapy, suggesting a durable clinical response. CART-2 did not cause any weight loss and no pathological alterations were found in 8 major organs by ex vivo histopathological analysis, indicating lack of toxicity. CART-2 displayed high specificity and sensitivity, where 2% hCD20+ cells were sufficient to delay tumor growth while hCD20- cells were not killed (n=30).


The impact of the different spacers on efficacy was not accurately reflected by in vitro studies and was only observed in vivo. This outcome motivates the efforts to find improved in vitro predictive assays. 


1. Lock. D, Mockel-Tenbrink. N, Drechsel. K, Barth. C, Mauer. D, Schaser. T, Kolbe. C, Al Rawashdeh. W, Brauner. J, Hardt. O, Pflug. N, Holtick. U, Borchmann. P, Assenmacher. M, Kaiser. A. “Automated manufacturing of potent CD20 directed CAR T cells for clinical use”. Hum. Gen. Ther. (2017) Oct;28(10):914-925

Fig. 1: Comparison of in vitro cytotoxicity of T cells transduced with the different αCD20 CAR const
CART-1 and CART-2 T cells demonstrated strong in vitro killing of CD20+ Jeko-1 cells, ~100% killing at 5:1 E:T ratio, in comparable to Mock-GFP T cells. In vitro killing of CART-1 and CART-2 was very similar.  
Fig.2: In vivo efficacy second study.
Total flux (p/s) shows that CART-2 achieved a highly significant therapeutic effect (p<0.01). Mice treated with CART-2 cells were the only group to achieve complete remission where the BLI signal was comparable to background (< 1x106 p/s) starting day 15, while CART-1 and Mock-GFP T cells had no therapeutic effect as the increase in tumor burden was similar to the untreated group (Tumor only).
Keywords: CAR T cells, preclinical trials, clinical translation, TCT automated manufacturing, immunotherapy
4:50 PM PS-08-5

Characterizing, imaging and targeting triple negative breast cancer tumors and metastases (#542)

F. De Lorenzi1, L. Rizzo1, S. Von Stillfried2, F. Gremse1, C. Rijcken3, C. - C. Glüer4, F. Kiessling1, T. Lammers1

1 Institute for Experimental Molecular Imaging, Aachen, Germany
2 RWTH Aachen University, Pathology Institute, Aachen, Germany
3 Cristal Therapeutics, Maastricht, Netherlands
4 Kiel University, Department of Radiology and Neuroradiology, Kiel, Germany


The lack of clinically established targeted therapies for triple negative breast cancer (TNBC) implies the need to systematically characterize primary tumors and secondary lesions in order to identify new therapeutic targets1. We here developed an optically imageable pre-clinical TNBC model which spontaneously metastasizes to distant organs, allowing non-invasive and longitudinal monitoring of cancer progression. Multiple imaging modalities were employed to characterize human and murine TNBC lesions as well as to address the biodistribution and targeting efficacy of nanomedicines.


4T1/iRFP breast cancer cells (680 nm) were orthotopically implanted in female nude mice. Solid tumor growth and metastases were monitored non-invasively through hybrid computer tomography and fluorescence molecular tomography (CT-FMT). Upon CT-based identification of metastases, mice were injected with fluorescently labeled micelles (750 nm) and monitored for nanoparticle (NP) biodistribution. Before sacrifice, mice were injected with lectin and excised organs were scanned by fluorescence reflectance imaging (FRI) to assess the colocalization of metastases (680 nm) and nanocarriers (750 nm). Human and murine primary and secondary lesions were stained for the characterization of vascular and stromal parameters, which taken together determine the targeting efficiency of NP on EPR-basis.


In vivo (passive) target site accumulation of micelles was seen over time in primary tumors (yellow arrows) and metastases (red arrows) (Fig 2A-B) over healthy organs (Fig 2C). The micelles signal (750 nm) overlaps with that of malignant cells (680 nm), at metastases which were macroscopically recognized at bright field images (Fig 2C). Histological characterization of the vasculature and stromal components (Fig 2D-I) gave evidence that metastases are more extensively vascularized than solid tumors, and comprise a denser ECM matrix (collagen fibers crosslinking through LOX). These factors might determine the targetability of malignant lesions. Additionally, further investigation of integrin targets on both murine and human primary tumors and metastases is currently ongoing through histopathology as well as proteomics and mass spectrometry-“based” methods.



This study evidences that metastases at different organs can be efficiently targeted with micellar nanomedicines, with a preferential accumulation at primary tumors in comparison to secondary lesions. Such systematic characterizations of microenvironmental factors influencing EPR and target site accumulation are of high relevance for the clinical translatability of nanomedicines and shall bring valuable insights into the discovery of new biomarkers for TNBC.


1"Triple-negative breast cancer: challenges and opportunities of a heterogeneous disease", G. Bianchini et al., Nature Reviews Clinical Oncology volume13, pages674–690(2016)

Multimodal and non-invasive metastasis imaging
Keywords: triple negative breast cancer, non-invasive imaging, tumor biomarkers, core-crosslinked polymeric micelles, metastases targeting
5:00 PM PS-08-6

Intratumoral activity distribution of 177Lu-PSMA-617 in a mouse model of prostate cancer (#543)

A. Örbom1, S. - E. Strand2, O. Vilhelmsson Timmermand1

1 Lund University, Oncology and Pathology, Lund, Sweden
2 Lund University, Medical Radiation Physics, Lund, Sweden


Despite the promising results of PSMA-targeted peptide radionuclide therapy of disseminated prostate cancer, and its rapid translation into the clinic, very little is published on the small-scale activity, and consequently absorbed dose, distribution. A handful of studies have examined patient biopsies [1-2], but the limited literature using preclinical models seem to omit this aspect [3]. This study aims to study the intratumoral activity distribution of a therapeutic PSMA-targeted peptide, and compare it to histology, antigen expression and some diagnostic PET-tracers.


BALB/c nude male mice were given subcutaneous tumor xenografts of human PSMA-expressing LNCaP cells. PSMA-617 (ABX, Radeberg, Germany) was labeled with 177Lu and the mice (n=8, n=7 with tumors) were given 20 MBq by i.v. injection. The animals were sacrificed at 20 min (n=3, n=2 with tumors), 70 min (n=2), 90 min (n=1), and 120 min (n=2) after which blood, tumor, kidneys and salivary glands were collected, weighed and their activity measured by gamma counter. One kidney and part of the tumor was frozen on dry ice and cryosectioned at 10 µm thickness. Autoradiography was performed using a double-sided silicon strip detector (Biomolex Imager 700, Biomolex, Oslo, Norway) with a 50 µm intrinsic spatial resolution [4] after which the same tissue sections were stained with hematoxylin and eosin.


The percent injected activity per gram was low at all time-points in salivary glands (<1.1% IA/g), which do not express PSMA in this animal model, and blood (<1.7% IA/g). Median kidney uptake was 72.4% IA/g at 20 min p.i., 108.2% IA/g at 70 min p.i., and 24.8% IA/g at 120 min p.i. For the tumor, median uptake was 16.7% IA/g at 20 min p.i., 10.1% IA/g at 70 min p.i. and 21.0% IA/g at 120 min p.i. Activity distribution in the kidney was almost entirely in the cortex, where it exhibited a spotty uptake pattern at all time-points, possibly indicating uptake in the glomeruli (Figure 1). The tumor activity distribution appears to become more homogeneous over time, however there are still areas with viable tumor cells with comparably low uptake at 120 min p.i. (Figure 2).


The small-scale activity distribution influences the absorbed dose delivered to both tumor and risk organs and is relevant to the choice of fractionation schedule and radioprotective measures. This study has begun to investigare the small-scale activity distribution of 177Lu-PSMA-617 in a mouse model of prostate cancer and currently ongoing work will increase the sample sizes as well as include comparisons with uptake of some clinical tracers using multi-radionuclide imaging and with antigen distribution etc using immunohistochemistry.


1. Pyka, Thomas, et al. "68Ga-PSMA-HBED-CC PET for differential diagnosis of suggestive lung lesions in patients with prostate cancer." Journal of Nuclear Medicine 57.3 (2016): 367-371.

2. Schottelius, Margret, et al. "[111 In] PSMA-I&T: expanding the spectrum of PSMA-I&T applications towards SPECT and radioguided surgery." EJNMMI research 5.1 (2015): 68.

3.Benešová, Martina, et al. "Preclinical evaluation of a tailor-made DOTA-conjugated PSMA inhibitor with optimized linker moiety for imaging and endoradiotherapy of prostate cancer." Journal of Nuclear Medicine 56.6 (2015): 914-920.

4. Örbom, Anders, et al. "Characterization of a double‐sided silicon strip detector autoradiography system." Medical physics 42.2 (2015): 575-584.

Figure 1
Activity distribution within kidneys of mice injected with 177Lu-PSMA-617.
Figure 2
Bottom: Activity distribution within tumors of mice injected with 177Lu-PSMA-617. Top: The same tissue sections stained with hematoxylin and eosin.
Keywords: PSMA, Lu-177, Autoradiography, Small-scale, Prostate Cancer
5:10 PM PS-08-7

Generation of Trimodal Imaging Stem Cells for Quantification and Optimization of Stem Cell Cancer Therapy (#526)

M. Zaw-Thin1, R. Bofinger2, J. Connell1, P. S. Patrick1, D. Stuckey1, H. C. Hailes2, A. B. Tabor2, M. F. Lythgoe1, T. L. Kalber1

1 University College London, Centre for Advanced Biomedical Imaging, London, United Kingdom
2 University College London, Department of Chemistry, London, United Kingdom


Stem cells have been used as selective anticancer agents1. In vivo imaging of stem cell distribution informs on methods needed to enhance cell migration to tumours. However, imaging cells distribution throughout the body is challenging and no single imaging modality can provide a complete answer. The aim of this study was to develop stem cells labelled with novel bimodal nanoparticles (SPECT/MRI) in combination with luciferase (Bioluminescence - BLI) to assess transplanted cell distribution from different injection routes and the ability of cells to home to tumours.


111In-SPION nanoparticle: DOTA was functionalised with an amine to form peptide bonds with carboxyls on superparamagnetic iron oxide nanoparticles (SPION)2. SPION were radiolabeled with 111In (HEPES-pH 5.5) with magnetic purification.

Cell labelling: luciferase positive human adipose derived stem cells (ADSC) were incubated overnight with 111In-SPION at 37 °C (Figure 1a). 1.5x105 ADSCs were injected intravenously (i.v) or intracardially (i.c) into NOD/SCID mice and imaged with SPECT/CT (Nanoscan-Mediso), MRI (ICON-Bruker) and BLI (IVIS-PerkinElmer) at day 0, 1, 3 and 7 after injection.

Tumour migration: 1.5x105 ADSCs were stimulated with IL-6 and SDF-1α (50 ng/ml)3 and injected i.v or i.c into NOD/SCID mice bearing a murine 4T1 orthotopic breast tumour and imaged with BLI as above.


SPECT/CT images from cell distribution study suggested that 1h after i.v injection, ADSCs were in lungs (50%) and liver (10%). This correlated with BLI and MRI (Figure 1b, c), which showed a reduction in T2 in the liver. 1h after i.c injection, SPECT/CT images showed ADSC distribution was predominantly in liver (42%), lungs (6%), kidney (2%) and brain (1%). This also correlated with BLI and MRI (Figure 1e, f), which showed focal hypointensities in the brain and kidney. SPECT and MRI co-registered images of liver and brain suggested the cells were dual labelled (Figure 1d, g).

BLI images from the tumour migration study suggested that there were ADSCs presence within the tumour as early as 1h after i.c injection, (Figure 2b). However, ADSCs were only detected within the tumour at day 3 after i.v injection (Figure 2a).


These results demonstrate the advantages of combining a genetic BLI reporter with a novel SPECT/MRI nanoparticle to quantitatively assess whole body distribution patterns of ADSC after different injection routes. The results from the tumour migration study suggest that the i.c injection route provides much more efficient ADSC delivery to tumour tissue. Future experiments will utilise this trimodal imaging strategy to quantitatively assess ADSC homing after cytokine stimulation and via different injection routes.


1) Sage et al.Thorax 2014;69:638. 2) Mitchell et al. Biomaterials 2013;34:1179. 3) Shi et al. Haematologica 2007;92:897.

Figure 1
(a) Diagram of ADSC labelled 111In-SPION & luciferase, (b, e) SPECT/CT images of ADSC distribution 1h after intravenous (i.v) (lung & liver) and intracardiac (i.c) injection (lungs, liver & kidneys), (c, f) BLI images of viable cells after i.v (lung) and i.c injection (brain, liver & kidneys), (d, g) Co-registered SPECT/MRI images of dual labelled cells after i.v (liver) and i.c injection (brain).
Figure 2

BLI images of 4T1 breast tumour bearing NOD/SCID mice showing the presence of ADSCs in the tumour (a) day 3 after intravenous injection and (b) 1h after intracardiac injection.


Keywords: Multimodal, Nanoparticle, SPECT/CT, MRI, Bioluminescence, Cell therapy, Cancer
5:20 PM PS-08-8

In vivo monitoring of intracellular pO2 in response to CAR T cell immunotherapy against glioma (#540)

F. Chapelin1, W. Zhu2, H. Okada3, E. T. Ahrens2

1 University of California San Diego, Bioengineering, La Jolla, California, United States of America
2 University of California San Diego, Radiology, La Jolla, California, United States of America
3 University of California San Francisco, Neurological Surgery, San Francisco, California, United States of America


Hypoxia is associated with tumor recurrence, and malignant progression1. Monitoring tumor pO2 levels can provide a preclinical biomarker for the effectiveness of emerging immunotherapies2. Perfluoro-crown-ether (PCE) nanoemulsion (NE) dissolves O2, causing a linear increase in the 19F spin-lattice relaxation rate (R1) with increasing pO2. PCE NE can intracellularly label tumor cells ex vivo pre-implantation. Using 19F MRI/MRS, we tested the hypothesis that an increase in pO2 is commensurate with CD8+ T cell apoptotic processes in a mouse model of glioma treated with human CAR T cells.


U87 glioma cells expressing EGFRvIII and luciferase were labeled ex vivo with PCE NE3 in media. Human T cells were transduced with a CAR vector4 to express a surface antibody against EGFRvIII. Female SCID mice (N=15) received unilateral subcutaneous injections of 5×106 PCE-labeled glioma cells. All mice were subjected to MRI and BLI four days post-tumor implantation, then received intravenous cell therapy (day 0). Groups 1-3 (N=5 per group) received 20×106 CAR T cells, 20×106 untransduced T cells, and no T cells, respectively. MRI and BLI scans were acquired on days 1, 3, 7, and 10. The 19F R1 was measured over entire tumor using PRESS to yield pO2 values, calculated with a calibration curve5. Tumors and spleens were harvested and fixed for histological correlation.


Prior to implantation, U87-EGFRvIII-Luc cells were labeled ex vivo with PCE NE to level ~7×1012 F-atoms/cell measured via 19F NMR. Following sub-cutaneous injection, labeled glioma cells appear as an MRI 19F hotspot with SNR~10 (Fig. 1A). Longitudinal in vivo measurements show a transient spike in tumor pO2 approximately three days after CAR T cell infusion (R1=0.99±0.12 s-1, pO2=134±25 mmHg) compared to untransduced T cells (pO2=61±20 mmHg) and control (pO2=40±9 mmHg, p = 0.026, Fig. 1B). These data suggest specific CAR T cell homing to the tumor tissue, presumably initiating a target killing cascade, and altering intracellular pO2. Longitudinal bioluminescence measurements show significant tumor regression 7 days post CAR treatment compared to  both control groups (p=0.012, Fig. 1C).  Histopathological staining confirmed the presence of CAR T cells in greater amounts than untransduced T cells in the tumors at day 3 post-infusion (data not shown6), consistent with the MRS results.


In this study, we show that 19F NE enables temporal measurements of tumor cell oxygen tension in response to CAR T cell therapy. Peak pO2 three days post-infusion suggests significant CAR T cell infiltration and tumor cell killing. Overall, these data support the view that 19F pO2 MRI and MRS can serve as a biomarker for cell-mediated apoptosis and provide insight into the modes of action of engineered T cell immunotherapy against cancer.


1 Tatum, J. L. et al. Hypoxia: importance in tumor biology, noninvasive measurement by imaging, and value of its measurement in the management of cancer therapy. Int J Radiat Biol 82, 699-757, (2006).

2 Matsuo, M. et al. Magnetic resonance imaging of the tumor microenvironment in radiotherapy: perfusion, hypoxia, and metabolism. Semin Radiat Oncol 24, 210-217, (2014).

3 Kadayakkara, D.K.K., et al. In vivo observation of intracellular oximetry in perfluorocarbon-labeled glioma cells and chemotherapeutic response in the CNS using fluorine-19 MRI. Magn Reson Med 64(5): 1252–1259, (2010).

4 Ohno, M. et al. Expression of miR-17-92 enhances anti-tumor activity of T-cells transduced with the anti-EGFRvIII chimeric antigen receptor in mice bearing human GBM xenografts. J Immunother Cancer 1, 21, (2013).

5 Zhong, J. et al. In vivo intracellular oxygen dynamics in murine brain glioma and immunotherapeutic response of cytotoxic T cells observed by fluorine-19 magnetic resonance imaging. PLoS One 8, e59479, (2013).

6 Chapelin, F. et al. Fluorine-19 nuclear magnetic resonance of chimeric antigen receptor T cell biodistribution in murine cancel model. Scientific Reports, 18;7(1):17748 (2017).

Figure 1: In vivo longitudinal pO2 and luminescence changes in U87 tumors.
(A) In vivo 1H/19F MRI image showing labeled U87 cancer cells in mouse flank. (B) pO2 measurements following delivery of CAR T cells or controls. A significant increase in tumor pO2 in CAR T cell-treated animals is observed at day 3 (*, p = 0.026). (C) BLI shows twice lower radiance in CAR-treated animals compared to controls at day 7 (*, p = 0.01), representing significant tumor growth reduction.
Keywords: fluorine MRI, cancer, immunotherapy, oximetry, perfluorocarbon