PET Reporter Gene Imaging and Ganciclovir-Mediated Ablation of Chimeric Antigen Receptor T Cells in Solid Tumors

Chimeric antigen receptor (CAR) T cells allow for specific tumor antigen recognition via antibody-derived single-chain variable fragments. Numerous clinical trials have utilized CARs to redirect T cells against targeted tumor antigens, with CD19-targeting CAR T cells achieving high complete response rates against therapy-resistant B-cell malignancies (1–5). Imaging strategies to monitor CAR T-cell biodistribution and proliferation would help to optimize their widespread clinical use and the monitoring of individual therapeutic responses, especially in solid tumors.

Noninvasive molecular imaging allows for real-time monitoring of immunotherapies, providing more information than changes in anatomic tumor volume alone. Such imaging techniques can provide quantitative spatiotemporal data on immune response and could potentially shed light on how the dynamics of T-cell response, such as proliferation and trafficking, correlate with therapeutic efficacy (6–9). Current methods to monitor adoptively transferred T cells include ex vivo cell labeling, radiolabeled antibodies, small-molecule metabolic probes, and reporter genes (10). Ex vivo labeling of immune cells with a radiotracer prior to patient administration does not allow for longitudinally imaging of the cell population due to tracer dilution upon cell division, limited intracellular radioactivity, cellular toxicity, relatively shorter radioisotope half-life, and high tracer efflux from the cell (6, 11–15). Antibody-based imaging of surface markers, while useful for monitoring of specific cell phenotypes, suffers from enhanced blood half-life and nonspecific tumor uptake, precluding clear delineation of immune cell populations from tumor cells (7, 16–18). Small-molecule probes, such as [¹⁸F]-2-fluoro-2-deoxyglucose ([¹⁸F]FDG), can be limited in monitoring immunotherapies due to the inability to distinguish between immune cell and tumor cell uptake (19).

Reporter gene PET imaging of CAR T cells leverages the engineered T cell for highly specific in vivo monitoring. Adoptively transferred immune cells can be engineered with PET reporter genes to encode for proteins that are not normally expressed by cells of interest, allowing for imaging of the immune cells using the corresponding PET tracer. The herpes simplex virus type 1 thymidine kinase (HSV1-tk) gene has previously been utilized for imaging of CAR T cells in animal models and patients with glioblastoma (20, 21). These patients showed a two-fold increase in 9-(4-[¹⁸F]-fluoro-3-[hydroxymethyl]butyl)guanine ([¹⁸F]FHBG) accumulation within the recurrent tumor or resection site upon infusion of over 1 × 10⁷ HSV1-tk–expressing CAR T cells, though a high background signal was detected prior to CAR T-cell infusion due to a disrupted blood–brain barrier (21).

The high response rates seen in clinical trials of CD19-directed CAR T cells in hematologic malignancies have led to a surge of interest in extending this therapeutic paradigm to other malignancies. Applying this therapy to target solid tumor antigens, which are often also expressed on essential healthy tissues, poses the risk for on-target, off-tumor toxicity (22). It is critical that the next generation of CAR T cells be engineered with control programs to allow clinicians to intervene in the case of adverse events. Suicide switches offer the ability to ablate transferred cells in the case of severe toxicities. In clinical trials, the inducible caspase-9 (iCasp9) system has been shown to be effective in depleting alloreactive T cells after transplant and resolving GVHD (23, 24). However, around 10% of the transferred cells were resistant to ablation. Although these residual cell populations did not induce a recurrence of GVHD, such incomplete ablation of CAR T cells might not be sufficient to mitigate CAR T cell–associated toxicities.

The HSV1-tk system has a dual benefit of acting not only as a PET reporter gene but also as a suicide gene. The HSV1-tk gene acts as a suicide switch when a prodrug substrate, such as ganciclovir, is administered, thereby inducing apoptosis of cells expressing the HSV1-tk reporter (25, 26). A mutated version of the thymidine kinase gene (sr39tk) was developed with increased catalytic activity compared with the original HSV1-tk gene and with less efficacy at phosphorylating endogenous substrates, leading to increased sensitivity and specificity to both [¹⁸F]FHBG and ganciclovir (27–30). Preclinical animal studies have shown intraperitoneal administration of ganciclovir can induce ablation of both engineered hematopoietic stem cells and tumor cells expressing the HSV1-sr39tk reporter without off-target toxicity (27, 29). With the clear potential benefit of a dual-function system, we characterized the utility of sr39tk as both a PET imaging reporter and suicide gene specifically in CAR T cells, which has yet to be demonstrated.

B7H3 (CD276) is a checkpoint molecule expressed at high levels on numerous pediatric solid tumors, including sarcomas and gliomas (31–33). Recent preclinical studies suggest that B7H3-targeting CAR T cells could be a promising therapy for various pediatric solid tumors (33). However, because B7H3 CAR T cells have not been tested clinically, it would be of clinical value to have the capability to monitor and ablate these cells. Toward this goal, we engineered B7H3 CAR T cells with the sr39tk gene and evaluated their efficacy, distribution, and ablation in a mouse model of osteosarcoma (Fig. 1). Inclusion of the sr39tk gene did not significantly hinder in vitro or in vivo B7H3 CAR efficacy. Visualization of B7H3-sr39tk CAR T cells revealed tumor homing and expansion within the spleen as expected. In addition, sr39tk-modified human CAR T cells intravenously administered to NSG mice were ablated within 3 days of ganciclovir treatment. Thus, this construct can both noninvasively provide insight into systemic CAR T-cell migration and proliferation through [¹⁸F]FHBG PET tracer administration while simultaneously allowing for rapid and specific CAR T-cell ablation through ganciclovir prodrug administration.

Human cell lines used in this study were provided by the following: NALM-6- GL by S. Grupp (University of Pennsylvania, Philadelphia, PA), 143B and MG63.3 human osteosarcoma by C. Khanna (NCI, NIH, Bethesda, MD). STR fingerprinting was conducted to verify the identity of all cell lines, and each cell line was validated to be Mycoplasma free by qPCR or MycoAlert (Lonza). Cells were cultured in RPMI1640 media supplemented with 2 mmol/L l-glutamine, 10 mmol/L HEPES, 100 U/mL penicillin, 100 mg/mL streptomycin (Invitrogen), and 10% heat-inactivated FBS.

Retroviral supernatant was produced by transient transfection of 293GP cells with the RD114 envelope plasmid and corresponding CAR plasmid as described previously (34). T-cell transduction was performed as described previously (34). Briefly, T cells, isolated by negative selection using the RosetteSep Human T Cell Enrichment Cocktail Kit (STEMCELL Technologies) from human buffy coats, were thawed and activated with anti-CD3/CD28 beads at a 3:1 bead to T-cell ratio and cultured in T-cell culture media (AIM V + 5% heat-inactivated FBS, 2 mmol/L l-glutamine, 10 mmol/L HEPES, 100 U/mL penicillin, 100 mg/mL streptomycin, and 100 U/mL rhIL-2). Virus-coated plates were prepared by spinning 1 mL of 1.9 × 10⁸ TU/mL of virus on Retronectin-coated (Takara/Clonetech) plates at 3,200 rpm for 2 hours. T cells were then cultured on these plates for 24 hours. This transduction process was performed twice on days 2 and 3 after activation. Beads were magnetically removed on day 4, and cells were expanded 10–20 fold in T-cell media until use. For both in vivo and in vitro assays, CAR T cells were used on day 10 postactivation.

T cells were stained using the following antibodies CD3 (BD Biosciences, clone UCHT1), CD4 (BD Biosciences, clone SK3), CD8 (BD Biosciences, clone SK1), PD-1 (Thermo Fisher Scientific, clone J105), CD45RA (BD Biosciences, clone HI100), CD62L (BD Biosciences, clone DREG-56), Ki-67 (BioLegend, clone 16A8). CAR staining was performed using recombinant human B7-H3 Fc (R&D Systems) conjugated using the DyLight 650 Microscale Antibody Labeling Kit (Thermo Fisher Scientific). Flow cytometry was performed on a FACS Fortessa instrument (BD Biosciences) and analyzed with FlowJo software.

Cytokine release was assayed by coincubating 5 × 10⁴ CAR T cells with 5 × 10⁴ tumor cells in complete RPMI1640. At 24 hours, culture media were collected and cytokines were measured by ELISA (Biolegend). Killing assays were performed by coculturing 5 × 10⁴ CAR T cells with 5 × 10⁴ fluorescence tumor cells in complete RPMI1640 in a 96-well plate and acquiring images every 2–3 hours using an Incucyte (Sartorius). Tumor cell proliferation killing was monitored by dividing the total fluorescence intensity at each time point by the fluorescence intensity at the first time point.

[¹⁸F]FHBG was produced according to previously described methods (20, 44) at the Stanford University Cyclotron and Radiochemistry Facility (Stanford, CA).

For uptake studies, prewarmed HBSS and approximately 5 μCi of [¹⁸F]FHBG was added to individual wells. Cells were incubated with the tracer at 37°C and 5% CO₂. At the specified time points, incubation was halted by removing tracer-containing media and placing the plates on ice to halt any residual uptake. Wells were then washed with ice-cold PBS (3 × 1 mL) and lysed in radioimmunoprecipitation assay buffer (Thermo Fisher Scientific Inc.; 200 μL). A portion of the cell lysates (150 μL) were used to determine the amount of decay-corrected radioactivity on a gamma counter (Hidex Gamma Counter). After sufficient time for radioactive decay, the remaining cell lysate was used to determine protein concentration using a bicinchoninic acid assay (Thermo Fisher Scientific Inc.). Uptake values were determined as percentage uptake by counting a control sample of the total radioactivity added to wells and dividing the uptake of the sample by this value. For retention studies, cells were incubated with appropriate radiotracer for 60 minutes and then washed with PBS (2 × 1 mL); a subsequent incubation was then completed at 37°C in fresh, prewarmed radiotracer-free media. At 30 minutes postmedia change, samples were processed as described above.

Immunodeficient NSG mice (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ) were purchased from The Jackson Laboratory or bred in house. Mice used for in vivo experiments were between 6 and 12 weeks old and were all female. All animal studies were carried out according to NCI and Stanford University Animal Care and Use Committee–approved protocols. MG63.3 osteosarcoma cell line was expanded under standard culture conditions (described above) and harvested with TrypLE (Gibco, Thermo Fisher Scientific). For MG63.3, 1 × 10⁶ cells were injected in the periosteum of the tibia. For in vivo antitumor efficacy experiments, 5 × 10⁶ CAR T cells (B7H3-sr39tk, B7H3, or blue fluorescent protein-sr39tk; matched for total T-cell dose) were injected intravenously into a tail vein 21 days after tumor inoculation. Mice were assigned to treatment groups based on their tumor sizes and groups were statistically identical. Tumor growth was monitored by bioluminescence (see below). Mice were euthanized when combined mean diameter for the sum of all masses reached 1.75 cm, as set by the institutional protocol. Overall timeline for survival studies were based on previously reported data with this model (33).

For bioluminescence imaging (BLI), mice were anesthetized, and imaging was performed using an IVIS Spectrum cooled charge-coupled device imaging system (Perkin Elmer). For tumor imaging, mice were injected with substrate d-luciferin (3 mg). For CAR T-cell imaging, mice were injected with Nano-Glo Luciferase Assay Substrate (Promega) diluted 40× in PBS. Images were analyzed with Living Image Software 4.1 (Perkin Elmer). The day 42 time point of Supplementary Fig. S5 was acquired on an SII Lago-X instrument and analyzed on Aura Imaging Software (Spectral instruments).

For PET/CT imaging, mice were anesthetized and injected with approximately 100–150 μCi of [¹⁸F]FHBG via the lateral tail-vein. Sixty minutes after injection, static PET images collected over a 10-minute acquisition period using a microPET/CT hybrid Inveon scanner (Siemens) for Fig. 2 or GNEXT PET/CT (Sofie) for Fig. 4. A CT image was acquired along with each PET scan for attenuation correction and anatomic reference frame for the respective PET data. All PET images were reconstructed with a three-dimensional ordered subsets expectation maximization algorithm and coregistered with CT images using the Inveon Research Workplace image analysis software (version 4.0; Siemens). To quantify tracer uptake within tumors in PET images, three-dimensional regions of interest were drawn over the tumor, contralateral muscle, and spleen. For tumor and contralateral muscle calculations, a 50% threshold was applied to each selected region, and uptake values were expressed as percentage injected dose per gram of tissue (%ID/g).

Ninety minutes after tracer administration, mice were euthanized, various tissues (tumor, muscle, spleen) were collected and weighed, and radioactivity was measured using an automated γ counter (Hidex Gamma Counter). Radioactivity was decay corrected to the time of radiotracer injection using diluted aliquots of the initial administered dose as standards. Data were expressed as percentage injected dose per gram of tissue (%ID/g) values.

Statistical analysis was performed using two-tailed unpaired t tests. Statistical differences in survival curves were determined using Mantel–Cox tests. IC₅₀ was calculated through a four-parameter nonlinear regression. All statistical analysis was performed in GraphPad Prism Version 8.0.2.

The B7H3 CAR was constructed using a single-chain variable fragment based on MGA271 antibody linked to a CD8 hinge and transmembrane domain along with a 4–1BB costimulatory domain, which was previously shown to have potent antitumor efficacy and clinical potential (33). We generated an MSGV1-based retroviral vector composed of the B7H3 CAR linked by a porcine teschovirus-1 2A ribosome skipping sequence to sr39tk (Fig. 2A) (33). Flow cytometry confirmed that B7H3-sr39tk CAR T cells exhibited similar B7H3 CAR expression as B7H3 CAR T cells (Fig. 2B). Similarly, there were no changes in exhaustion phenotype (PD-1 and LAG-3; Supplementary Fig. S1) or T-cell subsets (CD4 vs. CD8 and CD45Rα vs. CD62L; Supplementary Fig. S2A–S2C) upon addition of the sr39tk gene (Supplementary Fig. S1 and S2). In vitro coculture assays demonstrated similar antitumor efficacy between the B7H3-sr39tk and B7H3 CAR T cells against 143B osteosarcoma cells naturally expressing B7H3 or Nalm6 leukemia cells engineered to overexpress B7H3 (Fig. 2C). Antigen-specific activity was further confirmed by analysis of proinflammatory cytokines IFNγ and IL2 in coculture supernatants (Fig. 2D). An advanced orthotopic model of human osteosarcoma treated with a single tail-vein (intravenous) dose of 5 × 10⁶ B7H3-sr39tk CAR T cells exhibited statistically significant reductions in tumor volume (Fig. 2E) and improvements in overall survival (Fig. 2F) compared with sr39tk T cells alone.

B7H3-sr39tk CAR T cells were incubated ex vivo with [¹⁸F]FHBG, showing selective uptake and retention compared with B7H3 CAR T cells (Fig. 3A). Findings were validated with similar trends observed in the uptake and retention of [³H]Penciclovir, a tritiated-acyclic guanine analog of ganciclovir (Fig. 3B; refs. 35, 36). This study additionally confirmed improved [³H]Penciclovir uptake after 60 minutes of incubation with the B7H3-sr39tk CAR T cells compared with T cells transduced with a construct containing the sr39tk upstream of the B7H3 CAR, henceforth referred to as sr39tk-B7H3 CAR (Fig. 3B). As such, later studies utilized the B7H3-sr39tk CAR T cells for both imaging and ablation. While the B7H3–sr39tk construct had slightly less [¹⁸F]FHBG uptake than the sr39tk only construct, this is likely a result of reduced expression of transgenes in the larger B7H3–sr39tk construct.

NSG mice were orthotopically injected with 1 × 10⁶ human osteosarcoma MG63.3 cells in the hind leg. Twenty-one days later, animals received either 5 × 10⁶ B7H3-sr39tk CAR T cells, 5 × 10⁶ sr39tk T cells, or saline intravenously (Fig. 4A). PET/CT images of mice 60 minutes postadministration of the PET reporter probe [¹⁸F]FHBG (∼150 μCi) and 8 days following T-cell administration (intravenous) showed increased tracer uptake in the tumor and spleen of mice treated with B7H3-sr39tk CAR T cells compared with sr39tk T-cell treated or untreated mice (Fig. 4B). Quantitative image analysis corroborated our findings of significantly higher percent-injected dose per gram of tissue (% ID/g) uptake in the tumors (4.25 ± 0.37% ID/g) and spleens (0.25 ± 0.019% ID/g) of B7H3-sr39tk CAR T-cell treated mice relative to tumors (2.60 ± 0.11% ID/g) and spleens (0.05 ± 0.004% ID/g) of sr39tk T-cell treated mice, while showing similar low levels of [¹⁸F]FHBG uptake between cohorts in the contralateral muscle (2.02 ± 0.38 %ID/g for B7H3-sr39tk CAR T-cell treated mice versus 1.53 ± 0.24 %ID/g for B7H3-sr39tk CAR T-cell treated mice; Fig. 4C). A high background in the abdominal area was observed in all groups due to the radiotracer elimination through the biliary tree and the gastrointestinal tract in mice (27).

The suicide gene function in B7H3-sr39tk CAR T cells was first evaluated in vitro through incubation with ganciclovir. B7H3-sr39tk CAR T cells showed a dose-dependent response to ganciclovir with an IC₅₀ of 17.5 nmol/L in cell culture and 95% cell death at 1 μmol/L (Fig. 5A). Remaining viable cells following ganciclovir treatment lacked the B7H3 CAR, indicating that they are primarily nontransduced T cells (Supplementary Fig. S3). To assess the feasibility in an in vivo setting, T-cell ablation experiments were performed in NSG mice bearing MG63.3 tumors. T cells were engineered with the nanoluciferase reporter system to facilitate in vivo tracking by BLI. Mice were treated intravenously with 5 × 10⁶ of either B7H3-sr39tk CAR, B7H3 CAR, or sr39tk T cells 21 days posttumor implantation (Fig. 5B). Ganciclovir was administered intraperitoneally at a dose of 100 mg/kg on day 29 and 50 mg/kg on day 30 posttumor implantation. BLI confirmed B7H3-sr39tk CAR T-cell ablation within 2 days of ganciclovir administration and sustained ablation through 9 days following ganciclovir administration, with no significant effect on the control B7H3 CAR T cells (Fig. 5C and D). Long-term monitoring studies confirmed B7H3-sr39tk ablation up to 42 days following ganciclovir administration (Supplementary Fig. S4A and S4B). Flow cytometry of spleens 9 days following the final ganciclovir administration revealed that our ganciclovir dosing regimen resulted in a 99.4% ablation efficiency of B7H3-sr39tk CAR T cells compared with mice that only received B7H3-sr39tk CAR T cells (Fig. 5E). Remaining viable cells following ganciclovir treatment had lower levels of Ki67 and PD1 compared with B7H3-sr39tk CAR T cells in mice not treated with ganciclovir, indicative of lower CAR T-cell activation, division rate, and response to antigen (Supplementary Fig. S5).

In this study, we have demonstrated that the sr39tk system can provide non-invasive monitoring and ablation of transferred CAR T cells. Our in vitro data show that CAR T cells are sensitive to ganciclovir treatment with an IC₅₀ of 17.5 nmol/L. Importantly, the mean steady-state serum concentration of human patients treated with ganciclovir in common regimens has been found to be approximately 2.3 μmol/L, with peak plasma concentrations exceeding 30 μmol/L (37). Thus, clinically relevant concentrations of ganciclovir are 2–3 orders of magnitude higher than the IC₅₀ we determined in our in vitro model, demonstrating the clinical feasibility of our selective ablation CAR T-cell system. Our in vivo data quantified that only 0.58% of CAR T cells remain viable following ganciclovir treatment compared with untreated controls. In this study, we demonstrated ablation of B7H3-sr39tk CAR T cells using BLI for visualization and flow cytometry for quantification. In this setting, BLI was selected over [¹⁸F]FHBG PET/CT as our previous work demonstrated that higher sensitivity of optical imaging permits lower levels of reporter gene expression and/or lower numbers of expressing cells to be imaged relative to PET imaging (38). For quantification of ganciclovir ablation, flow cytometry was employed, as it is a standard technique for quantifying cell numbers from in vivo samples with a cellular level of sensitivity. Previous studies employing the iCasp9 suicide gene comparably determined ablation efficacies with flow cytometry (23, 39). The use of flow cytometry to determine ganciclovir ablation efficiency also permits direction comparison with ablation efficiencies reported in other studies. The sr39tk system reported here is the only system capable of serving the dual role as both an effective PET imaging reporter and suicide switch (40–42). Other classes of PET reporter genes can be utilized solely for T-cell monitoring, while other suicide switches do not allow noninvasive imaging in humans. While previous work has clinically monitored CAR T cells through the HSV1-tk gene, we are the first to image via the HSV1-sr39tk gene, which has been previously demonstrated to have increased sensitivity to [¹⁸F]FHBG (30).

The inclusion of the virally derived sr39tk gene in the B7H3 CAR introduces a concern of immunogenicity, which has been reported in human studies of HSV1-tk–modified T cells (43, 44). However, following lymphodepletion, immunogenicity via anti-TK effector CD8 T cells is not observed until several months after treatment, which could provide sufficient time for CAR-mediated tumor clearance (45). In addition, sr39tk T cells can be ablated through ganciclovir administration if toxicities are observed, emphasizing the benefit of the dual PET reporter and drug-induced suicide system. The other significant concern is the observed high gastrointestinal signal of the tracer that could occlude analysis of potential organs of interest. Although localization of a murine model is limited by the small body habitus, human trials have demonstrated that this PET tracer can provide sufficient separation to clearly visualize and differentiate T-cell localization in secondary lymph tissue from [¹⁸F]FHBG clearance in the hepatobiliary system (21, 46). Biodistribution of [¹⁸F]FHBG in humans indicates that imaging of the HSV1-sr39tk gene is feasible everywhere outside of the central nervous system, as the tracer is unable to cross the blood–brain barrier (47). In addition, voiding of the bladder and imaging at later time points allows for clearance and lower background signal from organs of high uptake (46).

While our model demonstrated the ability to selectively track B7H3-sr39tk CAR T cells, this reporter gene strategy has applicability across other adoptive cell therapeutic models. Our study investigated a single osteosarcoma model but has the potential for other solid tumor models outside of the central nervous system (20). The sr39tk reporter gene approach can be applied to other emerging immune cell therapies to track migration and proliferation systemically within patients. In particular, expressing CARs on NK cells (CAR-NK cells) have gained increased interest as a cancer immunotherapy. CAR-NK cells provide a unique opportunity to produce an off-the-shelf allogeneic product that could be readily available for immediate clinical use. Significant challenges and uncertainties exist within the application of adoptive NK-cell therapies, in particular, their limited in vivo persistence and suboptimal homing to tumors compared with T-cell therapies (47, 48). As such, strategies to monitor CAR-NK cell biodistribution and expansion are desperately needed. In addition, monitoring of tumor-homing macrophages through the sr39tk gene can provide insight into its potential as an early cancer detection tool (49). Incorporation of sr39tk within these macrophages would allow for PET visualization of regions of macrophage homing and sensing of the tumor microenvironment.

In summary, this work is the first to demonstrate the feasibility of incorporating the sr39tk gene for monitoring and control of CAR T-cell therapies and highlights the potential utility of this approach for management of adoptive cell therapies in the clinical setting.

L. Labanieh reports other compensation from National Science Foundation (graduate research fellowship) during the conduct of the study; other compensation from Lyell Immunopharma (consultant for Lyell and received stock options) outside the submitted work; in addition, L. Labanieh has a patent for DEGRON FUSION PROTEINS AND METHODS OF USING SAME pending, licensed, and with royalties paid from Lyell Immunopharma and a patent for REGULATABLE CELL SURFACE RECEPTORS AND RELATED COMPOSITIONS AND METHODS pending, licensed, and with royalties paid from Lyell Immunopharma. J.R. Cochran reports grants from NIH NCI, Goldman Sachs Foundation, St. Baldrick's/Stand Up to Cancer Pediatric Dream Team; personal fees and other compensation from xCella Biosciences, Inc. (cofounder and shareholder), Trapeze Therapeutics, Inc. (cofounder, director, and shareholder); other compensation from Aravive, Inc. (shareholder) outside the submitted work. R.G. Majzner reports personal fees from Lyell Immunopharma, Illumina Radiopharmaceuticals, Xyphos Inc., GammaDelta Therapeutics, and Zai Lab outside the submitted work; in addition, R.G. Majzner has a patent for B7-H3 CARs with enhanced activity pending. C.L. Mackall reports personal fees from Lyell Immunopharma, Neoimmune Tech, pricity; and other compensation from Allogene (equity) outside the submitted work. No potential conflicts of interest were disclosed by the other authors.

S. Murty: Conceptualization, data curation, formal analysis, methodology, writing-original draft, writing-review and editing. L. Labanieh: Data curation, formal analysis, investigation, writing-review and editing. T. Murty: Data curation, methodology. G. Gowrishankar: Conceptualization, data curation, supervision, investigation, writing-review and editing. T. Haywood: Data curation, methodology, writing-review and editing. I.S. Alam: Data curation, investigation, writing-review and editing. C. Beinat: Data curation, investigation, methodology, writing-original draft, writing-review and editing. E. Robinson: Data curation, investigation, writing-review and editing. A. Aalipour: Data curation, writing-review and editing. D.D. Klysz: Data curation, investigation, writing-review and editing. J.R. Cochran: Supervision, writing-review and editing. R.G. Majzner: Data curation, supervision, writing-review and editing. C.L. Mackall: Conceptualization, supervision, funding acquisition, writing-review and editing. S.S. Gambhir: Conceptualization, supervision, project administration, writing-review and editing.

The authors would like to acknowledge the Stanford Center for Innovation in In-Vivo Imaging (SCI3) and Timothy Doyle for their maintenance of the preclinical animal imaging facility. The authors would also like to thank the Stanford Radiochemistry Facility and Dr. Di Fan for their assistance in radiotracer production. Funding for this work was provided by the Canary Foundation (to S.S. Gambhir), the Ben and Catherine Ivy Foundation (to S.S. Gambhir), and the Virginia and D.K. Ludwig Fund for Cancer Research (to C.L. Mackall). This work was also supported by a St. Baldrick's–Stand Up to Cancer (SU2C) Dream Team translational research grant (SU2C-AACR-DT-27-17 to C.L. Mackall). SU2C is a division of the Entertainment Industry Foundation, and research grants are administered by the American Association for Cancer Research, the scientific partner of SU2C. S. Murty has received support from the National Science Foundation Graduate Student Fellowship. L. Labanieh has received support from the National Science Foundation Graduate Research Fellowship, Stanford Graduate Fellowship, and Stanford EDGE Fellowship. We dedicate this manuscript to the loving memory of our mentor, Sanjiv Sam Gambhir.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.