The 293T cell line and the human gastric cancer cell line Hs 746T were purchased from the American Type Culture Collection (ATCC), while the cell line SNU-638 was obtained from the Korean Cell Line Bank (Seoul National University, Seoul, Korea) (23). 293T and Hs 746T were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM, Corning) supplemented with 10% heat-inactivated fetal bovine serum (FBS, GeminiBio). SNU-638 cells were cultured in RPMI-1640 (Corning) supplemented with 10% FBS. To enable bioluminescence imaging of tumor growth in animal models, Hs 746T and SNU-638 tumor cell lines were transduced with a firefly luciferase-F2A-GFP (FLuc-GFP) lentiviral vector (Biosettia, cat. no. GlowCell-16-1). Cells were maintained in a humidified incubator at 37°C with 5% CO₂ and were routinely tested for mycoplasma using a MycoAlert™ detection kit (Lonza).

Lentiviral vectors encoding CARs specific to intercellular adhesion molecule 1 (ICAM-1) and epithelial cellular adhesion molecule (EpCAM) have been previously described (24, 25). The ICAM-1-specific CAR was composed of a c-Myc tag, LFA-1 I domain (F292A), the CD8α hinge and transmembrane domain, 4-1BB co-stimulatory domain and CD3ζ intracellular domain. The EpCAM-specific CAR construct contained from the 5′-LTR end: a c-Myc tag, a single-chain variable fragment (scFv) derived from the anti-EpCAM monoclonal antibody UBS54 (26), the CD8α hinge, the transmembrane and intracellular domains of CD28, and the intracellular domain of CD3ζ. SSTR2 was incorporated at the C-terminus of both CARs using a porcine teschovirus-1 2A (P2A) ribosome-skipping sequence to obtain comparable expression levels of both CAR and SSTR2.

Lentivirus was produced via transient transfection of 293T cells with transfer plasmid and LV-MAX lentiviral packaging mix (Gibco, cat. no. A43237) using Lipofectamine 3000 transfection reagent (Invitrogen, cat. no. L3000015), following the manufacturer’s instructions. Lentiviral supernatant was collected after 72 hours, filtered through a 0.45 μm filter, and concentrated using PEG-it virus precipitation solution (System Biosciences, cat. no. LV825A-1). The lentivirus was then aliquoted and stored at −80°C.

To manufacture CAR T cells, leukopaks were obtained either from healthy donors (AllCells) commercially or from advanced thyroid cancer (ATC) patients who provided informed consent under the protocol approved by the Weill Cornell Medicine Institutional Review Board (Protocol number, 19-12021154). Primary T cells were isolated through MACS separation using CD4 (Miltenyi Biotec, cat. no. 130-045-101) and CD8 (Miltenyi Biotec, cat. no. 130-045-201) microbeads, and then cultured in TexMACS medium (Miltenyi Biotec, cat. no. 170-076-307) supplemented with 5% human AB serum (Sigma, cat. no. H4522), 12.5 ng/ml of IL-7 (Miltenyi Biotec, cat. no. 170-076-111), and 12.5 ng/ml of IL-15 (Miltenyi Biotec, 170-076-114). After activation with Human T-Expander CD3/CD28 Dynabeads (Gibco, cat. no. 11141D) at a bead:cell ratio of 1:1 (day 0), T cells were transduced twice with lentivirus at a multiplicity of infection (MOI) of 5 on days 1 and 2. Subsequently, T cells were expanded in G-Rex 6M well plate (Wilson Wolf Manufacturing, cat. no. 80660M). On day 10, CAR T cell products were harvested and cryopreserved in a 1:2 mixture of T cell complete growth medium and CryoStor CS10 (STEMCELL Technologies, cat. no. 07930).

DOTATATE was purchased from Macrocyclics (Texas, USA) and dissolved in water at 1 mg/ml, and then stored at –20°C in aliquots. ¹⁷⁷LuCl₃ was obtained from ITM Pharma Solutions GmbH (Munich, Germany). To prepare ¹⁷⁷Lu-DOTATATE, ¹⁷⁷LuCl₃ was diluted in a solution containing 0.2 M sodium acetate and 0.02 M ascorbic acid to achieve a concentration of 50 mCi/ml. DOTATATE was added into the reaction vial at a ratio of 2 mCi/nmol. The reaction mixture was then incubated in a dry-block heater at 90°C for 30 minutes, followed by cooling to room temperature. High-performance liquid chromatography (HPLC) using a 4.6 mm × 100 mm Symmetry C18 column was employed to determine the radiochemical yield and purity, which consistently exceeded 95%. The final product was diluted with 0.9% saline and passed through a 0.22 μm filter before injection.

NOTA-Octreotide precursor (1,4,7-Triazaclononane-1,4,7-triacetic acid-octreotide) was obtained from ABX advanced biochemical compounds GmbH (Radeberg, Germany) and radiolabeled with ¹⁸F as previously described (20).

Six to eight-week-old male NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG) mice were purchased from The Jackson Laboratory (Stock # 005557). All experimental mice were co-housed in the Animal Core Facility at Weill Cornell Medicine (New York, NY) under specific pathogen-free conditions and provided with sterile food and water. They were maintained at an ambient temperature of 21–27°C and humidity of 40–60%, with a 12 h light/dark cycle. All procedures involving animals were approved by the Institutional Animal Care and Use Committee at Weill Cornell Medicine. Hs 746T and SNU-638 tumor cells were inoculated subcutaneously into the flank of mice in 100 μl of culture media:Matrigel (1:1) solution. Relapsed SNU-638 tumor model was established by subcutaneous implantation of the relapsed SNU-638 tumor fragments into the flank of mice. For the systemic ATC tumor model, ATC001 tumor cells were injected intravenously via tail vein. CAR T cells manufactured with healthy donors were used for the treatment of gastric tumors (Hs 746T and SNU-638), whereas autologous ICAM-1 CAR T cells manufactured from ATC patient T cells were employed in the ATC001 tumor model. All CAR T cells were cryopreserved and adoptively administered via tail vein freshly after thawing. ¹⁷⁷Lu-DOTATATE was injected intravenously. The treatment time and dose are indicated in the schematics and/or figure legends. Tumor progression was monitored by bioluminescent imaging using the IVIS imaging system (PerkinElmer). Bioluminescence images were acquired 15 min after intraperitoneal injection of 200 μl of 15 mg/ml D-luciferin (Gold Biotechnology) and analyzed using the Living Image v.4.7.2. (PerkinElmer). Whole-body bioluminescence flux was used to estimate tumor burdens. Tumor volume (V) was measured with a caliper on a weekly basis and calculated using the formula V = [length × (width)²]/2 for subcutaneous tumor models. Randomization of mice was performed on the basis of bioluminescence imaging or tumor size measurements to ensure equal tumor burden prior to CAR T cell treatment. PET-CT imaging was performed to monitor CAR T cell expansion within the tumor using a micro-PET-CT scanner (Inveon, Siemens) and the Inveon acquisition workplace (Siemens) 2 hours after intravenous injection of ¹⁸F-NOTA-Octreotide tracer. PET-CT images were analyzed using the AMIDE v.1.0.5. Peripheral blood from subcutaneous Hs 746T model was harvested via retro-orbital blood collection under isoflurane anesthesia at indicated time points. Serum was collected post-centrifugation at 2,000 × g for 15 min at 4°C and stored at −80°C for cytokine analysis. The health condition of mice was monitored on a daily basis by the veterinary staff, independent of the investigators and studies. Mice were euthanized using a CO₂ chamber, followed by cervical dislocation, when they reached either a maximum tumor size of 3,000 mm³ or a humane endpoint, such as losing >25% of body weight or experiencing symptoms of overt disease (e.g., ruffled fur, difficulty with diet, hunched back posture, or abnormal grooming behavior).

Serial noninvasive SPECT-CT imaging was performed on a SPECT-CT dedicated animal scanner (Inveon, Siemens) to verify tumor targeting and calculate tumor dosimetry following ¹⁷⁷Lu-DOTATATE treatment. Regions of interest (ROIs) were drawn on the co-registered CT images, and the activities within the spheres were quantified in the reconstructed SPECT images. Non-decay corrected time–activity concentration data were fit to a single exponential model. The cumulated activities were estimated through a combination of mathematical model fitting and area under the curve calculation for the measured data points. Absorbed doses were calculated using the IDAC-dose 2.1 software (27), assuming complete local absorption of the ¹⁷⁷Lu β-emissions.

For the ex vivo analysis of tumor-infiltrating lymphocytes (TILs), tumor tissues from the lungs of the ATC001 tumor model were collected 29 days following T cell treatment. Tumors were dissected into small fragments and digested in RPMI-1640 medium containing collagenase type IV (200 U/mL, Gibco) for 1 hour at 37°C. The samples were then dissociated using a serological pipette and filtered through a 70 μm cell strainer to generate single cell suspensions. TILs were isolated through MACS separation using human CD4 (Miltenyi Biotec, cat. no. 130-045-101) and human CD8 (Miltenyi Biotec, cat. no. 130-045-201) microbeads. Subsequently, the samples were stained with Pacific Blue-conjugated human CD3 antibody (Biolegend, clone HIT3a), FITC-conjugated c-Myc antibody (Miltenyi Biotec, clone SH1-26E7.1.3), and APC-conjugated human SSTR2 antibody (R&D Systems, clone 402038). A fraction of the TIL sample was cultured in complete T cell growth media for 24 hours and then analyzed by flow cytometry with the same settings as above. The pre-infusion CAR T cell product was also assessed and served as the baseline for the expression levels of CAR and SSTR2 on the cell surface. Cell staining was performed at room temperature in the dark for 15 minutes, followed by two washes and a staining with propidium iodide before analysis on a Gallios flow cytometer (Beckman Coulter). Flow cytometry data were analyzed using FlowJo v.10.8.1 (Tree Star, Inc.), with CD3⁺ cells gated as T cells.

Cytokine levels in clarified mouse plasma were quantified using a custom LEGENDplex™ panel (BioLegend), following the manufacturers’ instructions. Plasma from untreated mice was used as the control. Cytokine concentrations were calculated using a standard curve (standards provided within the kit).

Statistical analysis was performed using Prism 10 (GraphPad, Inc.). An unpaired, two-tailed student’s t-test was performed to evaluate differences between groups. Mouse survival curves were generated using the method of Kaplan–Meier, and the significance was analyzed with the log-rank (Mantel–Cox) test. P < 0.05 was considered statistically significant.

To leverage the capabilities of CAR T cells in delivering targeted radiation via SSTR2-specific ¹⁷⁷Lu-DOTATATE, it is essential to monitor the expansion of CAR T cells within the tumor. This tracking is crucial for identifying the optimal timing of ¹⁷⁷Lu-DOTATATE treatment. To this end, we performed longitudinal PET-CT imaging of ICAM-1 CAR T cells using ¹⁸F-NOTA-Octreotide in a Hs 746T gastric cancer model, where tumor cells lack overexpression of SSTR2 on their surfaces ( Figures 1A–C ). Maximum intensity projection PET images revealed the specific uptake and retention of ¹⁸F-NOTA-Octreotide in tumors treated with ICAM-1 CAR T cells. The total uptake of ¹⁸F-NOTA-Octreotide (% ID) displayed a rapid and continued increase as tumors progressed. Meanwhile, the standardized uptake (% ID/cm³) increased rapidly in the first 3 weeks, maintaining a range between 3–7% ID/cm³ thereafter. In contrast, no detectable signals were observed in Hs 746T tumors from control cohort of No T mice that did not receive CAR T cell treatment ( Figure 1B ). These results suggest that ¹⁷⁷Lu-DOTATATE treatment could be guided by ¹⁸F-NOTA-Octreotide PET imaging, allowing treatment initiation at a point with substantial CAR T cell infiltration within tumors (>3% ID/cm³).

We further validated the specific uptake of ¹⁸F-NOTA-Octreotide by autologous tumor-infiltrating CAR T cells in a systemic tumor model of advanced thyroid cancer, where tumor burden was observed in the lungs following intravenous injection of tumor cells ( Figures 1D, E ). Additionally, we isolated TILs from the lungs and observed a loss of surface CAR expression within TILs, as indicated by reduction of cMyc tag staining ( Figure 1E ). The downregulation of CAR expression following tumor antigen stimulation has been recognized as one of the major limiting factors of CAR T cell functionality and antitumor efficacy within the tumor microenvironment (28, 29). Consistent with the findings from others, we observed recovery of CAR expression in TILs following a 24-hour resting period in vitro without tumor stimulation (30). In contrast, we did not detect any significant downregulation of SSTR2 expression, indicating that the loss of surface CAR was primarily driven by antigen stimulation ( Figure 1E ). This sustained expression of SSTR2 underscores its value in tracking CAR T cells and the image-guided delivery of ¹⁷⁷Lu-DOTATATE for TRT.

In the study using Hs 746T gastric cancer model (as illustrated in Figure 1A ), we administrated 7.4 MBq of ¹⁷⁷Lu-DOTATATE to mice on day 28 following CAR T cell treatment. Subsequently, we conducted serial SPECT-CT imaging at 24, 144, 360 and 432 hours post intravenous injection of ¹⁷⁷Lu-DOTATATE. The SPECT-CT images demonstrated a highly selective localization of ¹⁷⁷Lu-DOTATATE within the tumor tissue, with minimum background levels recorded in the rest of the body ( Figure 2A ). At 24 hours post-injection, the effective ¹⁷⁷Lu activity in the tumor was measured to be 0.29 ± 0.16 MBq/g, corresponding to an uptake of 4.29 ± 2.37% of injected dose of ¹⁷⁷Lu-DOTATATE per gram of tumor ( Figure 2B ). Furthermore, the non-decay corrected activity in tumor was 0.12 ± 0.05 MBq/g at 144 hours post-injection, representing a 71% retention of ¹⁷⁷Lu activity after accounting for the physical decay of ¹⁷⁷Lu (0.17 MBq/g). This sustained retention of ¹⁷⁷Lu activity in the tumor suggests minimal biological or metabolic clearance of ¹⁷⁷Lu from the tumor and alludes to survival of CAR T cells. The cumulated absorbed doses for the tumor, liver, and kidneys were calculated to be 2.55 ± 1.17 Gy, 0.58 ± 0.19 Gy, and 0.75 ± 0.28 Gy, respectively ( Figure 2C ). Importantly, the absorbed dose in the kidneys (0.75 Gy) was well below the threshold nephrotoxicity dose (24 Gy) in mice (31).

We then set out to investigate the potential of enhancing the anti-tumor immune response through the delivery of low-dose radiation. In the aggressive gastric cancer model of Hs 746T, the activity of ICAM-1 CAR T cells was mostly modest, despite a substantial expansion of CAR T cells within the tumor ( Figures 3A–C ; 1B, C ). Four weeks following CAR T cell injection, a time when significant CAR T expansion was observed in tumor (¹⁸F-NOTA-Octreotide uptake >3% ID/cm³), a subset of mice received a single dose of 7.4 MBq of ¹⁷⁷Lu-DOTATATE. Impressively, all radiation-treated mice displayed rapid tumor shrinkage and survived significantly longer ( Figures 3B–D ). Within 3–4 weeks of ¹⁷⁷Lu-DOTATATE treatment, 5 out of 6 mice showed complete tumor regression, while 1 mouse maintained a stable tumor burden after initial tumor reduction ( Figure 3C ). The uptake of ¹⁸F-NOTA-Octreotide by tumor-infiltrating CAR T cells remained consistently high in the weeks after ¹⁷⁷Lu-DOTATATE injection ( Figure 3E ), suggesting minimal radiation-mediated CAR T cell death at the dosage of 2.5 Gy. Of note, the elevated levels of IFN-γ and perforin detected in mouse serum following ¹⁷⁷Lu-DOTATATE treatment underscored a radiation-induced effect on T cell immunity ( Figure 3F ).

We have previously developed EpCAM-targeting CAR T cells for the treatment of gastric cancer, which demonstrated potent efficacy in achieving complete responses in subcutaneous gastric cancer models (24). However, tumor recurrence occurred frequently after short duration of complete response. When we transplanted these relapsed tumors into NSG mice and administered fresh EpCAM CAR T cell products, we observed a modest partial response to CAR T cell therapy, despite a substantial expansion of CAR T cells within the tumor. This indicated that the relapsed tumors had acquired resistance, limiting the potential benefits of further CAR T cell treatment. Consequently, we pursued a strategy involving the delivery of higher dose of ¹⁷⁷Lu-DOTATATE to achieve direct killing of these resistant gastric tumors ( Figure 4A ). The administration of 37 MBq of ¹⁷⁷Lu-DOTATATE following CAR T cell therapy resulted in absorbed doses of 6.02 ± 1.35 Gy within the tumor, leading to tumor growth control ( Figures 4B–D ). Additionally, SPECT images demonstrated the selective uptake of ¹⁷⁷Lu-DOTATATE by tumor-infiltrating CAR T cells. In contrast, mice treated solely with ¹⁷⁷Lu-DOTATATE, without CAR T treatment, did not exhibit detectable tumor uptake of ¹⁷⁷Lu-DOTATATE ( Figures 4B, C ). Mice that received either CAR T or ¹⁷⁷Lu-DOTATATE alone experienced tumor progression, although at a slower rate compared to mice receiving no treatment ( Figure 4D ). Longitudinal monitoring of CAR T cells by ¹⁸F-NOTA-Octreotide PET-CT revealed the simultaneous killing of CAR T cells along with tumor elimination ( Figure 4E ). Similar to the observation in Hs 746T tumor model and ICAM-1 CAR T cells ( Figures 1A–C ), PET-CT imaging showed continuous expansion of CAR T cells as the tumors progressed in mice receiving CAR T cells only ( Figure 4E ). Overall, these findings suggest that TRT holds the potential to effectively eliminate resistant tumors that are unresponsive to CAR T cell treatment.