Introduction

Reviewing the past century’s history, we must pay heartfelt tribute to the three representative drugs of anti-cytotoxic T-lymphocyte antigen-4 (CTLA-4) antibody ipilimumab and anti-programmed cell death protein-1 (PD-1) antibodies nivolumab and pembrolizumab, which are like the great explorers Columbus/Da Gama/Magellan in the era of geographical discovery, transforming the concept of tumor immunity into clinical practice of immunotherapy, thus benefiting the vast number of patients with tumor.1–4 However, the urgent clinical need for more personalized checkpoint blockade therapy remains pressing, owing to the relatively low overall response rate, unpredictable immune responses and primary and acquired immune resistance.5 This has instigated research into the development of predictive biomarkers and rational combination therapy to improve their efficacy,6 7 and novel immunotargets have been explored and identified as such.8 9 Of these, lymphocyte-activated gene 3 (LAG-3) is considered the third immune checkpoint of clinical significance,10 right next to CTLA-4 and PD-1.

LAG-3, also known as CD223, is identified as an inhibitory receptor highly expressed on activated lymphocytes including CD4+T, CD8+T, regulatory T (Treg) cells, B lymphocytes, natural killer (NK) cells, dendritic cells (DCs) and macrophages.10 11 It transmits inhibitory immune signals through binding to the ligands such as major histocompatibility complex class II (MHC-II), fibrinogen-like protein 1, galectin-3, LSECtin, α-synuclein.10 12–14 Elevated LAG-3 expression was linked to T-cell exhaustion in a broad spectrum of cancers.15 16 Moreover, T-cell function was further disabled by the co-expression of LAG-3 and PD-1, which is also correlated with the resistance to anti-programmed death-ligand 1 (PD-L1)/PD-1 inhibitors.17 18 Thus, these observations make LAG-3 a promising immunotherapeutic target complementary to current immunotherapies.

To date, there are more than 20 anti-LAG-3 therapeutics in clinical trials, with the approval of relatlimab from the US Food and Drug Administration in combination with nivolumab to treat unresectable or metastatic melanoma.10 19 20 In phase 1–2 trial (NCT019680109), the objective response rate (ORR) was 11.5% in relatlimab-nivolumab group.21 The ORR of patients with LAG-3 expression ≥1% was 3.5 times that of LAG-3<1%. In the key RELATIVITY-047 trial (NCT0347092), it was observed that the median progression-free survival and progression-free survival rate at 12 months were significantly improved with relatlimab-nivolumab as compared with nivolumab (10.1 months vs 4.6 months, 47.7% vs 36.0%).22 The median progression-free survival was longer for patients with LAG-3 expression ≥1% (12.58 months vs 4.83 months). These results underscore that patient stratification based on LAG-3 expression could benefit for the anti-LAG-3-related therapy. In addition, growing attention has been paid to the prognostic significance of LAG-3 for checkpoint blockade therapy.23 24 Given this, the strategies for quantitative analysis of LAG-3 expression are urgently needed for clinical use.

Positron emission tomography (PET) is just one such technology, which has been exemplified with radiotracers for better assessment of PD-1/PD-L1, immunoglobulin and ITIM domain, interleukin (IL)-2, and CD69 expression than immunohistochemistry (IHC) in preclinical and clinical settings.25–30 Currently, the potential use of [89Zr]Zr-REGN3767 and [89Zr]Zr-BI754111 for PET imaging of LAG-3 expression is ongoing in clinical trials (NCT04566978, NCT04706715, and NCT03780725). Preliminary results of [89Zr]Zr-BI754111 in patients with non-small cell lung cancer (NSCLC) and head and neck squamous cell carcinoma revealed that tracer uptake in tumors was clearly visible and correlated to immune cell-derived RNA signatures, suggesting a potential predictive imaging biomarker for LAG-3-directed therapies.31 However, these two tracers suffered from relatively low target-to-background ratio, due to their long blood circulation and slow clearance from the body. Breckpot et al developed a nanobody-based tracer for non-invasively quantifying LAG-3 expression on tumor-infiltrating leukocytes and predicting the response of immune checkpoint blockade.32 33 Nevertheless, this nanobody is specific for murine LAG-3 protein, with no clinical translational potential. For the sake of clinical practicality, high-contrast PET imaging should be preferentially achieved within 2 hours using low molecule and human LAG-3-binding moieties.

In this study, a series of cyclic peptides with high affinity for both human and murine LAG-3 were employed as the lead compounds to design and develop 68Ga-labeled PET tracers, which were comparatively characterized in vitro and in vivo. With a candidate [68Ga]Ga-CC09-1 in hit, this tracer was further evaluated for PET imaging of LAG-3 expression on infiltrating lymphocytes in humanized mice models and in first-in-human studies.

Material and methodsRadiosynthesis

In brief, 25–30 µg of NOTA-peptides in 1 mL 0.25 M NaOAc buffer was added to 370–925 MBq 68GaCl3 solution eluted from an ITM 68Ge/68Ga generator with 4 mL 0.05 M HCl. The solution was heated to 90°C for 10 min, and the final product was obtained after C18 sep-Pak cartridge purification. The radiochemical yield, radiochemical purity, and molar activity were determined by radio-HPLC analysis, respectively.

Molecular docking

The protein crystal structures of LAG-3 and MHC-II were extracted from the protein data bank (PDB codes: 7TZG and 1KT2) and their structural interface was obtained using Protein Preparation module in Maestro V.12.0. The cyclic peptides ligands were constructed with two-dimensional structures using ChemDraw, followed by three-dimensional structures using Maestro workspace, and then docked using the Induced Fit Docking module. Finally, their interactions were analyzed in the Maestro workspace.

Cell culture and tumor models

Chinese hamster ovary (CHO) cell line was obtained from Procell Life Science & Technology (Wuhan, China). The construction of CHO-expressed human LAG-3 (CHO-hLAG-3) cells and separation of peripheral blood mononuclear cells (PBMCs) as well as their knock-down of LAG-3 expression via sgRNAs were described in Supporting Information. All animal experimental protocols and procedures were approved by Animal Care and Use Committee at Central South University (CSU-2022–0290). The LAG-3-KO mice (Cat. NO. NM-KO-18048) were purchased from the Shanghai Model Organisms Center and identified through sequencing. The C57BL/6 or Balb/c nude mice were purchased from Hunan Silaike Jingda Experimental Animal, China. For the subcutaneous xenograft models, 1×106 cells were resuspended in a total volume of 100 µL phosphate-buffered saline and injected into the right flank of mice. For the generation of PBL models, NSG mice (Shanghai Model Organisms Center, China) were reconstituted with 5×106 PBMCs via vein injection and were then used for imaging studies.

Cell uptake

CHO, CHO-hLAG-3, unstimulated PBMCs, stimulated PBMCs and LAG-3-knock-down PBMCs were seeded in 24-well plate, incubated for 24 hours and then added with 74 kBq of [68Ga]Ga-CC09-1, [68Ga]Ga-CC09-2 and [68Ga]Ga-CC09-3 per well, respectively. The cells were incubated at 37°C for 15, 30, 60, and 120 min. For in vitro blocking experiments, cells were co-incubated with 10 µg of cold cyclic peptides precursor. After incubation, these cells were washed, collected and measured with a gamma-counter (WIZARD 2480, Pekin Elmer).

Cell binding assay

CHO-hLAG-3 cells were added into a 24-well plate with a concentration of 2×105 cells per well. A series of cold cyclic peptides precursor were prepared, with different final concentrations ranging from 1×10–11 M to 1×10–4 M. Then, cells were co-incubated with these solutions and 68Ga-labeled cyclic peptides tracers (74 kBq/well) for 1 hour at 37°C. After incubation, cells were washed, collected and analyzed via the gamma counter. The IC50 values of 68Ga-labeled cyclic peptides tracers were determined using GraphPad Prism V.8.0.2 software, respectively.

Micro-PET imaging

The tumor-bearing mice, PBL and NSG model mice received the intravenous injection of 7.4–14.8 MBq radiotracers. Mice were anesthetized and imaged using a micro-PET/CT scanner (Mediso, Hungary). Static PET imaging was acquired for 10 min, while dynamic imaging was collected for 60 min. After image reconstruction, region-of-interest was selected to determine per cent injected dose per gram (% ID/g) in various tissues to obtain the time-activity curves and tumor-to-non-tumor ratios. For in vivo blocking experiments, 100 µg of cold cyclic peptide precursor was intravenously injected 1-hour before radiotracer administration.

Biodistribution in mice

The tumor-bearing mice were intravenously injected with 1.85 MBq of [68Ga]Ga-CC09-1. Tissues of interest were collected at 30, 60, and 120 min after the administration, weighed and then measured with a gamma-counter. In the blocking experiments, 100 µg of cold CC09-1 was injected 1-hour prior to the radiotracer. The values of % ID/g for each organ of interest were calculated. The pharmacokinetic study was carried out using C57BL/6 mice intravenously injected with 7.4 MBq of [68Ga]Ga-CC09-1. Blood samples at different time points were collected, weighed, and measured with a gamma-counter. The blood distribution (% ID/g) was plotted over time to obtain the time-activity curve. Pharmacokinetic parameters were calculated using DAS V.2.0.

Study participants

All participants were sequentially recruited for enrollment in this study from June 2022 to September 2022 and have been asked to sign an informed consent.

Imaging protocol and reconstruction procedure

Three healthy volunteers and five patients were intravenously injected with 3.7–5.2 MBq/kg of [68Ga]Ga-CC09-1 and underwent dynamic imaging for 60 min on a GE Discovery PET/CT 690 Elite scanner. The [18F]FDG PET/CT scan was performed in patients within 1 week. [18F]FDG PET/CT and [68Ga]Ga-CC09-1 examinations were performed according to the imaging protocols, which are detailed in Supporting Information.

Safety evaluation

Volunteers and patients were requested to report any side effects (such as dizziness, vomiting, or abdominal discomfort) experienced during injection and scanning. Vital signs (heart rate and blood pressure) were continuously monitored throughout the administration process. Physical examinations, routine blood tests (hematology, chemistry), urine tests, as well as assessments of liver and renal function, were conducted within 1-week following the scan. Any alterations and adverse events observed during the [68Ga]Ga-CC09-1 PET/CT scanning procedure and subsequent examination were systematically recorded and analyzed in accordance with the Common Terminology Criteria for Adverse Events.

Image interpretation

After reconstruction, images were interpreted by two experienced nuclear medicine physicians. The maximum standardized uptake value (SUVmax) and mean standardized uptake value (SUVmean) of tumors and other organs were measured by volume regions of interest method.

Statistical analysis

GraphPad Prism Software V.8.0.2 (San Diego, California, USA) was used for all data analysis. The significance of differences between groups were analyzed using the unpaired Student’s t-test. P value<0.05 was considered statistically significant. All the experimental data were presented as (mean±SD).

ResultsRadiolabeling cyclic peptides with gallium-68

NOTA was successfully conjugated to the cyclic peptides with a PEG-based linker to obtain three precursors (denoted as NOTA-CC09-1, NOTA-CC09-2, and NOTA-CC09-3), as confirmed by MALDI-MS (online supplemental figures S1–S3). These precursors were radiolabeled with 68GaCl3 and purified via C18 solid phase extraction cartridges, which were verified by radio-HPLC (figure 1A and online supplemental figure S4). The radiochemical yields of the radiotracers [68Ga]Ga-CC09-1, [68Ga]Ga-CC09-2 and [68Ga]Ga-CC09-3 were 79.5±2.8%, 77.9±3.3% and 81.3±3.6% with their radiochemical purity of >99% (n=5). As shown in online supplemental table S1, the molar activity was determined to be 53.21±9.07 GBq/μmol for [68Ga]Ga-CC09-1, 48.32±5.47 GBq/μmol for [68Ga]Ga-CC09-2 and 39.83±7.63 GBq/μmol for [68Ga]Ga-CC09-3 (n=4). The partition coefficient (log PpH=7.4) was measured to be −2.04±0.27, –1.34±0.18 and −1.36±0.52, respectively (n=4), suggesting their hydrophilicity. The in vitro saline and serum stability of three radiotracers were >95% out to 4 hours (online supplemental figure S5).

Figure 1

Design and development of LAG-3 targeting radiotracers. (A) Chemical structure of [68Ga]Ga-CC09-1, [68Ga]Ga-CC09-2 and [68Ga]Ga-CC09-3. (B) The binding models of [68Ga]Ga-CC09-1, [68Ga]Ga-CC09-2 and [68Ga]Ga-CC09-3 at the LAG-3/MHC-II interface (protein data bank codes: 7TZG and 1KT2). LAG-3, lymphocyte-activated gene 3; MHC-II, major histocompatibility complex class II.

Molecular docking

Blockade of LAG-3/MHC-II interaction has been considered as an established therapeutic strategy.34 To clarify the binding modes of cyclic peptides tracers, we defined the LAG-3/MHC-II interface (PDB code 7TZG and 1KT2) and conducted molecular docking via Maestro V.18.0. As shown in figure 1B, all these radiotracers could well bind to the LAG-3/MHC-II interface. In detail, the ligand-mediated recognition was via Gln49, Thr172, Arg121 and Thr208 for [69Ga]Ga-CC09-1 with an optimized binding energy of −10.113 kcal/mol, Ser-174 and Arg121 for [69Ga]Ga-CC09-2 with an optimized binding energy of −6.899 kcal/mol as well as Ser170 and Gln49 for [69Ga]Ga-CC09-3 with an optimized binding energy of −6.506 kcal/mol.

Cell uptake and binding assay

To perform the cell uptake and binding assay, the CHO cells were stably transfected with human LAG-3 (figure 2A, online supplemental figures S6 and S7). The cell uptake of 68Ga-labeled cyclic peptides tracers was initially performed in CHO-hLAG-3 and CHO cells (n=5). As shown in figure 2B, [68Ga]Ga-CC09-1 displayed the time-dependent cellular uptake within 2 hours in CHO-hLAG-3 cells, as compared with low non-specific uptake in LAG-3-negative CHO cells. The cell uptake of [68Ga]Ga-CC09-1 in CHO-hLAG-3 cells was about 3.5 higher than that in CHO cells at 60 min (p<0.001). Similar results were observed in the cell uptake experiment of [68Ga]Ga-CC09-2 and [68Ga]Ga-CC09-3 (online supplemental figure S8). In the blocking study, the cell uptake for three radiotracers could be significantly decreased with an overdose of unlabeled cyclic peptides (figure 2C, p<0.001), indicating the specific binding of tracers to hLAG-3. In addition, [68Ga]Ga-CC09-1 showed higher cell uptake than [68Ga]Ga-CC09-2 and [68Ga]Ga-CC09-3 in CHO-hLAG-3 cells (figure 2D, p<0.01, p<0.05). Competitive binding experiments were performed in CHO-hLAG-3 cells and demonstrated an IC50 of 0.940 µM for [68Ga]Ga-CC09-1, 1.512 µM for [68Ga]Ga-CC09-2 and 1.252 µM for [68Ga]Ga-CC09-3, respectively (figure 2E). The affinity of [68Ga]Ga-CC09-1 was further assessed using microscale thermophoresis and the KD value was determined to be 0.895 µM (figure 2F).

Figure 2

In vitro characterization of 68Ga-labeled cyclic peptides tracers. (A) Fluorescence confocal microscopy of hLAG-3 expression. CHO cells transfected with a plasmid expressing human LAG-3 were stained with anti-LAG-3 antibody and CoraLite594-conjugated second antibody (red) and DAPI (blue). (B) Time-dependent uptake of [68Ga]Ga-CC09-1 in CHO-hLAG-3 and CHO cells (n=5). (C) and (D) Cell uptake of [68Ga]Ga-CC09-1, [68Ga]Ga-CC09-2 and [68Ga]Ga-CC09-3 without and with cold cyclic peptides after 60 min of incubation, respectively (n=5). (E) Competitive binding assays of [68Ga]Ga-CC09-1, [68Ga]Ga-CC09-2 and [68Ga]Ga-CC09-3 (n=5). (F) The MST binding curves of [69Ga]Ga-CC09-1 (n=3). (G) Cell uptake of [68Ga]Ga-CC09-1 in 1×106 unstimulated and stimulated PBMC without and with cold cyclic peptides after 60 min of incubation, respectively (n=4). (H) Cell uptake of [68Ga]Ga-CC09-1, [68Ga]Ga-CC09-2 and [68Ga]Ga-CC09-3 in 1×106 stimulated PBMC after 60 min of incubation (n=4). (I) Cell uptake of [68Ga]Ga-CC09-1 in 1×105 unstimulated and stimulated PBMC as well as stimulated PBMC with pretreatment of sgRNA after 60 min of incubation, respectively (n=3). The data are shown as mean±SD, ***p<0.001, **p<0.01, *p<0.05. CHO, Chinese hamster ovary; CPM, counts per minute; DAPI, 4',6-diamidino-2-phenylindole; hLAG-3, human LAG-3; LAG-3, lymphocyte-activated gene 3; MST, microscale thermophoresis; PBMC, peripheral blood mononuclear cell.

The human PBMCs were activated and employed for cell uptake assay (n=4). The expression of LAG-3 was remarkably upregulated after stimulation (online supplemental figure S9, p<0.01). The cell uptake of [68Ga]Ga-CC09-1 at 60 min of incubation was 4.27±0.71% in 1×106 activated PBMCs, 1.24±0.38% in unstimulated PBMCs and 1.00±0.04% in the blocking group (figure 2G, p<0.01, p<0.01). [68Ga]Ga-CC09-1 had the higher cell uptake in the activated PBMCs than [68Ga]Ga-CC09-2 and [68Ga]Ga-CC09-3 (figure 2H, 4.60±0.53% vs 3.14±0.42% vs 3.34±0.38%, p<0.05, p<0.05). Furthermore, the knock-down of LAG-3 protein via sgRNAs in activated PBMCs was performed and demonstrated by western blotting and PCR analysis (online supplemental figure S10). The cell uptake of [68Ga]Ga-CC09-1 at 60 min of incubation was 0.44±0.06% (p<0.001), 0.50±0.04% (p<0.001) and 0.48±0.13% (p<0.01) in 1×105 activated PBMCs with pretreatment of three types of sgRNA, compared with 1.12±0.08% in activated PBMCs (n=3, figure 2I).

Micro-PET imaging of LAG-3 expression in mice

The potential of [68Ga]Ga-CC09-1, [68Ga]Ga-CC09-2 and [68Ga]Ga-CC09-3 for PET imaging of LAG-3 expression was performed and compared in CHO-hLAG-3 cell-derived xenograft models (n=4). The radioactivity uptake in brain, heart, liver, lung, muscle and tumor over time was quantified by using dynamic imaging to plot the time-activity curves (online supplemental figure S11). All three radiotracers could enable the visualization of LAG-3 expression in tumors within 2 hours (figure 3A). The tumor accumulation reached saturation at 60 min post-injection, and then gradually reduced over time (figure 3B). [68Ga]Ga-CC09-1 uptake in CHO-hLAG-3 tumors at 60 min was 1.35±0.33 % ID/g, compared with [68Ga]Ga-CC09-2 uptake of 0.37±0.02 % ID/g (p<0.01) and [68Ga]Ga-CC09-3 uptake of 0.28±0.02 % ID/g (p<0.01). These radiotracers were quickly distributed in the whole body and rapidly cleared from non-specific organs/tissues. As shown in figure 3C, the optimized imaging contrasts were obtained at 60 min after intravenous injection, with a tumor/muscle ratio of 17.18±3.20 for [68Ga]Ga-CC09-1, 4.14±2.52 for [68Ga]Ga-CC09-2 (p<0.01) and 4.00±0.480 for [68Ga]Ga-CC09-3 (p<0.01). The tumor-to-other tissues ratios of [68Ga]Ga-CC09-1 were observably higher than that of [68Ga]Ga-CC09-2 and [68Ga]Ga-CC09-3 (figure 3D). These results demonstrated the preferable capacity of [68Ga]Ga-CC09-1 for profiling of LAG-3 expression.

Figure 3

Micro-PET imaging of 68Ga-labeled cyclic peptides tracers in Chinese hamster ovary human lymphocyte-activated gene 3 cell-derived xenograft models (n=4). (A) Representative small-animal PET images of [68Ga]Ga-CC09-1, [68Ga]Ga-CC09-2 and [68Ga]Ga-CC09-3. (B) Time-activity curves of xenografts. (C) Tumor-to-muscle ratio at different time points and (D) tumor-to-non-tumor ratio at 60 min post-injection were derived from PET imaging of [68Ga]Ga-CC09-1, [68Ga]Ga-CC09-2 and [68Ga]Ga-CC09-3. The data are shown as mean±SD, **p<0.01, *p<0.05, n.s., not significant. PET, positron emission tomography; % ID/g, per cent injected dose per gram.

The specificity of [68Ga]Ga-CC09-1 for LAG-3 was further corroborated. We observed the uptake of [68Ga]Ga-CC09-1 in CHO-hLAG-3 tumors, which was markedly blocked with co-injection of CC09-1 (figure 4A and online supplemental figure S12). The CHO tumors showed comparatively low radioactivity uptake. The tumor uptake in CHO-hLAG-3, CHO and blocking group at 60 min after injection was 1.32±0.16 % ID/g, 0.11±0.01 % ID/g and 0.18±0.03 % ID/g (figure 4B, p<0.0001, p<0.0001). The corresponding tumor-to-muscle ratios were measured to be 20.49±5.01, 0.73±0.34 and 0.74±0.36 (online supplemental figure S12B, p<0.01, p<0.01). The ratios of tumor/other tissues were significantly decreased in CHO and blocking groups (online supplemental figure S12C). The ex vivo tumor tissues at 60 min post-injection in three groups were imaged by using autoradiography and their LAG-3 expression was confirmed by IHC staining, collaboratively supporting the PET results (figure 4C and online supplemental figure S12D). In addition, the murine LAG-3 expression were also detected by [68Ga]Ga-CC09-1 PET imaging in MC38, PANC02 and B16F10 tumor-bearing models (online supplemental figure S13). To further confirm its specificity, PET imaging of [68Ga]Ga-CC09-1 was performed in wild-type (WT) and LAG-3-KO mice bearing MC38 tumors (n=3, figure 4D and online supplemental figure S14). Tumor uptake of [68Ga]Ga-CC09-1 in WT mice was significantly higher than that in LAG-3-KO mice and WT mice with blocking at 30, 60, and 90 min post-injection. The tumor uptake at 60 min was determined to be 2.32±0.25 % ID/g, 0.18±0.02 % ID/g and 0.25±0.06 % ID/g, respectively (figure 4E, p<0.001, p<0.001), with the corresponding tumor-to-muscle ratios of 21.85±0.11, 2.37±0.30 and 2.48±0.15 (online supplemental figure S14B, p<0.0001, p<0.0001). The expression of murine LAG-3 in MC38 tumor microenvironment on WT and LAG-3-KO mice was verified by immunofluorescence staining (figure 4F and online supplemental figure S15). Furthermore, we quantified the LAG-3-positive tumor-infiltrating lymphocytes subsets in MC38 tumor microenvironment on WT and LAG-3-KO mice, respectively (online supplemental figure S16). As shown in figure 4G, LAG-3-positive CD19, NK, CD3+T, CD4+T, CD8+T, Treg and DC cells in WT mice were particularly higher than that in LAG-3-KO mice.

Figure 4

In vivo specificity and biodistribution of [68Ga]Ga-CC09-1. (A) PET imaging of [68Ga]Ga-CC09-1 in CHO and CHO-hLAG-3 cell-derived xenograft models without and with pretreatment with cold CC09-1 (n=4). (B) Tumor uptake at different time points were derived from their PET imaging data. (C) Immunohistochemical staining of hLAG-3 expression in CHO and CHO-hLAG-3 xenografts. (D) PET imaging of [68Ga]Ga-CC09-1 in wild-type (WT) and LAG-3-KO mice bearing MC38 tumors (n=3). (E) Tumor uptake at different time points were derived from PET imaging data. (F) Immunofluorescence staining of mLAG-3 expression in MC38 tumor microenvironment on WT and LAG-3-KO mice. (G) The quantitative expression of LAG-3 on the subsets of tumor-infiltrating lymphocytes in MC38 tumor microenvironment on WT and LAG-3-KO mice (n=10 and 3, respectively). (H) The biodistribution of [68Ga]Ga-CC09-1 in CHO and CHO-hLAG-3 cell-derived xenograft models (n=4). The data are shown as mean±SD, ****p<0.0001, ***p<0.001, **p<0.01, *p<0.05. CHO, Chinese hamster ovary; hLAG-3, human LAG-3; LAG-3, lymphocyte-activated gene 3; PET, positron emission tomography; % ID/g, per cent injected dose per gram.

Biodistribution, pharmacokinetics and radiation dosimetry

The biodistribution of [68Ga]Ga-CC09-1 was investigated in CHO-hLAG-3 and CHO cells-bearing mice (n=4, figure 4H). The findings in biodistribution study were in agreement with that of PET imaging. The radioactivity in kidneys was predominant with an uptake value of 19.27±3.94 % ID/g at 60 min post-injection, indicating its excretion through the urinary system. The tumors had a moderate uptake of tracers and the highest uptake value was 2.24±0.10 % ID/g at 60 min after injection. Correspondingly, the CHO-hLAG-3 tumor uptake was decreased to 0.66±0.10 % ID/g by pretreatment with CC09-1 (p<0.001). In contrast, CHO tumor showed relatively low radiotracer accumulation (0.55±0.17 % ID/g, p<0.0001). The low uptake and rapid elimination in other non-target organs generated a high imaging contrast. CHO-hLAG-3 tumor-to-muscle ratio was up to 11.39±0.76 at 60 min, as compared with CHO tumor of 3.80±1.36 and blocking group of 7.04±0.91 (online supplemental figure S17A, p<0.001, p<0.01). These observations together demonstrated the specificity of [68Ga]Ga-CC09-1. In addition, the a similar biodistribution of [68Ga]Ga-CC09-1 was also observed in MC38 tumor models (online supplemental figure S18).

Furthermore, the pharmacokinetic parameters of [68Ga]Ga-CC09-1 were computed and shown in online supplemental table S2. The blood half-life of [68Ga]Ga-CC09-1 was determined to be 19.25 min (online supplemental figure S17B). In the in vivo metabolic stability study, approximately 85% of intact radiotracer was analyzed in blood within 60 min after injection, while a large proportion of tracer was decomposed in urine (online supplemental figure S19).

The radiation dosimetry of [68Ga]Ga-CC09-1 was estimated for the safety of human use (online supplemental table S3). The kidneys and urinary bladder wall were doomed to receive the highest absorbed doses, owing to urinary excretion of [68Ga]Ga-CC09-1. The effective dose of [68Ga]Ga-CC09-1 was calculated as 9.85E-03 mSv/MBq for adult men and 1.21E-02 mSv/MBq for adult women, as comparable to previously reported 18F-FDG of 1.90E-02 mSv/MBq.35

ImmunoPET imaging of LAG-3+ lymphocytes infiltrates in PBMC-humanized mice

We next verified the capacity of [68Ga]Ga-CC09-1 for mapping of LAG-3+ lymphocytes infiltrates using longitudinal PET imaging in hu-PBL-SCID (PBL) mice and NOD-scid IL-2Rgammanull (NSG) mice (n=3). Until 30 days after injection of PBMC, the prominent [68Ga]Ga-CC09-1 uptake was observed in the spleen and lung of PBL mice, along with a slight increase in the salivary gland (figure 5A). The spleen, lung and salivary gland uptake at 30 days were 1.80±0.14 % ID/g, 0.62±0.06 % ID/g and 0.88±0.20 % ID/g, compared with that of 0.44±0.09 % ID/g (p<0.001), 0.21±0.04 % ID/g (p<0.001) and 0.27±0.05 % ID/g (p<0.01) at 17 days (figure 5B). However, there was no difference in the distribution of [68Ga]Ga-CC09-1 in NSG mice over time (online supplemental figure S20). As shown in figure 5C,D and online supplemental figure S21, in PBL mice relative to NSG mice without PBMCs injection, the radiotracer uptake in spleen, lung and salivary gland was substantially increased from 0.23±0.05 % ID/g to 1.80±0.14 % ID/g (p<0.0001), 0.18±0.03 % ID/g to 0.62±0.06 % ID/g (p<0.001) and 0.34±0.04 % ID/g to 0.88±0.20 % ID/g (p<0.05), respectively. The contrast ratios of spleen, lung and salivary gland to muscle were also improved from 1.54±0.18 to 8.88±2.15, 1.22±0.30 to 3.05±0.78 and 2.25±0.29 to 4.44±1.71 (figure 5E, p<0.01, p<0.05, not significant). IHC staining of spleen in PBL and NSG mice illustrated the LAG-3+ lymphocytes infiltrate, corresponding with the uptake of [68Ga]Ga-CC09-1 as delineated by PET (figure 5F).

Figure 5

ImmunoPET Imaging of LAG-3+ lymphocytes infiltrates using [68Ga]Ga-CC09-1 in PBMC-humanized mice (n=3). NSG mice were injected with human PBMCs and humanized for 10, 17 and 30 days. (A) Representative PET/CT images and (B) biodistribution of PBL mice at 60 min after injection of [68Ga]Ga-CC09-1. (C) Representative PET images of spleen in PBL and NSG mice. (D) [68Ga]Ga-CC09-1 uptake in spleen, lung and salivary glands. (E) The ratios of spleen, lung and salivary glands to muscle. (F) LAG-3 immunohistochemistry of spleen. The data are shown as mean±SD, ****p<0.0001, ***p<0.001, **p<0.01, *p<0.05, n.s., not significant. LAG-3, lymphocyte-activated gene 3; PBMC, peripheral blood mononuclear cell; PET, positron emission tomography; % ID/g, per cent injected dose per gram.

Safety evaluation

None of participants reported subjective symptoms either during or after tracer injection. There were no adverse events and no significant changes in vital signs or clinical laboratory tests observed in both healthy volunteers and patients (online supplemental table S4). The mean administered activity of [68Ga]Ga-CC09-1 was 196±15 MBq (range 183–211 MBq). According to the specific activities of [68Ga]Ga-CC09-1, the administered mass of precursor was estimated to be as low as 4.4–7.2 µg per participant.

Biodistribution in healthy volunteers

The biodistribution of [68Ga]Ga-CC09-1 in healthy volunteers was consistent with that observed in animal experiments. Representative PET/CT maximum intensity projection images and quantitative data of normal organs were shown in figure 6. The accumulation of [68Ga]Ga-CC09-1 was mainly observed in kidneys and bladder with the SUVmean of 3.33±0.39 and 41.31±1.77 at 60 min and with 54.75±5.07% of total injected radioactivity present in the urine at 30 min after injection. The cardiac blood pool showed moderate uptake, with an SUVmean of 1.27±0.03 at 60 min. Low radioactivities in the brain, lung, and muscle were found with the SUVmean of 0.03±0.01, 0.44±0.06, 0.28±0.03 at 60 min, respectively. The radiotracer displayed the rapid distribution-elimination balance with 1 hour and urinary system excretion, in favor of high-contrast imaging in clinical settings.

Figure 6

Positron emission tomography imaging of [68Ga]Ga-CC09-1 in healthy volunteers (n=3). (A) Maximum intensity projection images at different time points after injection. The SUVmax (B) and SUVmean (C) values of normal organ uptake of [68Ga]Ga-CC09-1. SUVmax, maximum standardized uptake value; SUVmean, mean standardized uptake value.

PET/CT results

Five patients underwent [68Ga]Ga-CC09-1 and [18F]FDG PET scan within 1-week, respectively. The patients’ clinical information and PET data were detailedly summarized in online supplemental table S5. All the primary tumors were visualized in [68Ga]Ga-CC09-1 and [18F]FDG PET/CT images (figure 7, online supplemental figures S22 and S23). The median SUVmax and SUVmean of [68Ga]Ga-CC09-1 PET/CT were 2.25 (range, 2.05–4.46) and 1.54 (range, 0.89–2.91). Correspondingly, the median SUVmax and SUVmean of 18F-FDG PET/CT were 7.7 (range, 2.85–22.96) and 4.82 (range, 1.95–13.92). The median tumor-to-muscle ratios of [68Ga]Ga-CC09-1 and [18F]FDG were measured to be 3.83 (range, 2.42–3.92) and 6.86 (range, 2.72–20.81). Interestingly, tumor uptake of [68Ga]Ga-CC09-1 was comparable to that of [18F]FDG in patient 2 with melanoma (SUVmax 4.13 vs 2.05, SUVmean2.61 vs 1.45 and T/M 2.89 vs 2.42), which was further confirmed by LAG-3 IHC (online supplemental figure S24). Compared with [18F]FDG, the uptake of [68Ga]Ga-CC09-1 was uneven in the large tumor of patient 1 with NSCLC (SUVmax=4.46, SUVmean=2.91 and T/M=3.83), implying the heterogeneity of tumor LAG-3 expression. Unfortunately, the tumor specimen from fine needle biopsy of this patient was negative for LAG-3 immunostaining (online supplemental figure S24).

Figure 7

[68Ga]Ga-CC09-1 and [18F]FDG PET/CT imaging in (A)patients with melanoma and (B) NSCLC (n=5). All the primary tumors were visualized in [68Ga]Ga-CC09-1 and [18F]FDG PET/CT images. NSCLC, non-small cell lung cancer; PET, positron emission tomography; SUV, standardized uptake value.

Discussion

LAG-3 has been considered as the next-generation of immune checkpoint and a promising prognostic biomarker of immunotherapy.10 23 As with PD-1/PD-L1 and CTLA-4 inhibitors, the PET imaging strategy for patient stratification and therapeutic response monitoring could guide the development of clinical decision-making of LAG-3-related therapy. In this study, we developed and validated 68Ga-labeled cyclic peptides tracers for PET imaging of LAG-3 expression on tumor-infiltrating lymphocytes in vitro and in vivo. [68Ga]Ga-CC09-1 was chosen in preference for further detecting the LAG-3+ lymphocytes infiltrates in humanized mice models and assessing the safety, biodistribution and potential of quantifying LAG-3 expression in first-in-human studies.

Currently, the majority of immunoPET agents have been developed with radiolabeling of therapeutic antibodies.36 Although characterized by high affinity and specificity, these tracers suffered from drawbacks such as immunogenicity, long blood circulation and slow clearance from body as well as inadequate tumor penetration. Imagine that it might encounter difficulties in real-time monitoring the changes of immunotarget expression, when the tracer took days to obtain optimized imaging contrast. In contrast to monoclonal antibodies, peptides offered the superiority of rapid imaging and clearance within hours, favorable pharmacokinetics and better tumor penetration as diagnostic agents.37 The first cyclic peptides capable of interfering with the interaction between LAG-3 and MHC-II were screened using phage display technology,38 which contributed to the development of 68Ga-labeled cyclic peptides tracers for translational immunoPET imaging of LAG-3 expression in tumor microenvironment.

The cyclic peptides were modified with PEG liker and NOTA for preserving their affinity and 68Ga-labeling. 68Ga-labeled cyclic peptides tracers ([68Ga]Ga-CC09-1, [68Ga]Ga-CC09-2 and [68Ga]Ga-CC09-3) were produced with high radiochemical yield, radiochemical purity, and molar activity. We first demonstrated the potential binding mechanisms of these tracers targeting LAG-3/MHC-II interface. Their specific cell uptake was modest for CHO-hLAG-3 cells and activated PBMCs, with IC50 and KD value in the range of 10–6 M. In CHO-hLAG-3 tumor-bearing mice, their capacity for PET imaging of LAG-3 expression was exhaustively compared in view of tumor uptake and imaging contrast. Among them, [68Ga]Ga-CC09-1 was preferred with the optimized tumor-to-non-tumor ratio at 60 min post-injection. The specificity of [68Ga]Ga-CC09-1 was confirmed by the markedly decreased tumor uptake in the CHO tumor-bearing mice and blocking groups pretreated with cold CC09-1. In addition, [68Ga]Ga-CC09-1 PET could also quantify the murine LAG-3 expression in tumor microenvironment, which non-invasively assesses the response of immunotherapy in preclinical studies. Its specificity was strictly verified in WT and LAG-3-KO mice bearing MC38 tumors. Owing to its hydrophilicity, the accumulation of [68Ga]Ga-CC09-1 was mainly observed in kidneys, indicating the urinary excretion. The rapid distribution and clearance from the body generated the optimized tumor-to-non-tumor ratios in a short time, which is inconsistent with that of PET data and conducive to clinical translation. Radiation dosimetry estimation supported that [68Ga]Ga-CC09-1 is safe for human use. Furthermore, [68Ga]Ga-CC09-1 could track the LAG-3+ lymphocytes infiltrates that are located in spleen, lung and salivary gland of PBL mice, which is similar to the findings of several other groups.25 39 All the observations of our preclinical studies proved that [68Ga]Ga-CC09-1 is a promising PET tracer for quantifying the LAG-3 expression in tumor microenvironment.