Efficacy and safety of novel multiple-chain DAP-CAR-T cells targeting mesothelin in ovarian cancer and mesothelioma: a single-arm, open-label and first-in-human study

Abstract

Background

Despite remarkable achievements in applying chimeric antigen receptor (CAR)-T cells to treat hematological malignancies, they remain much less effective against solid tumors, facing several challenges affecting their clinical use. We previously showed that multichain DNAX-activating protein (DAP) CAR structures could enhance the safety and efficacy of CAR-T cells when used against solid tumors. In particular, mesothelin (MSLN)-targeted CAR-T cell therapy has therapeutic potential in MSLN-positive solid tumors, including ovarian cancer and mesothelioma.

Methods

In vitro cell killing assays and xenograft model were utilized to determine the anti-tumor efficacy of MSLN targeting DAP-CAR-T cells and other CAR-T cells. ELISA and flow cytometry analysis were used to assess the cytokine secretion capacity and proliferation ability. Eight patients with MSLN expression were enrolled to evaluate the safety and efficacy of MSLN-DAP CAR-T cell therapy. Single-cell sequencing was performed to explore the dynamics of immune cells in patients during treatment and to identify the transcriptomic signatures associated with efficacy and toxicity.

Results

We found that multichain DAP-CAR formed by combining a natural killer cell immunoglobulin-like receptor truncator and DAP12 exhibited better cytotoxicity and tumor-killing capacity than other natural killer cell-activated receptors associated with DAP12, DAP10, or CD3Z. The safety and efficacy of MSLN-DAP CAR-T cell therapy in patients with ovarian cancer and mesothelioma were evaluated in a single-arm, open-label clinical trial (ChiCTR2100046544); two patients achieved partial response, while four patients had a stable disease status. Furthermore, single-cell sequencing analysis indicated that KT032 CAR-T cell infusion could recruit more immune cells and temporarily remodel the TME.

Conclusions

Our study highlights the safety and therapeutic efficacy of multiple-chain DAP-CAR-T cell therapy targeting MSLN to treat patients with ovarian cancer and mesothelioma.

Trial registration

ChiCTR.org.cn, ChiCTR2100046544. May 21, 2021.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13073-024-01405-5.

Background

Chimeric antigen receptor (CAR)-T cell therapy has been revolutionary, achieving remarkably effective and durable clinical outcomes against hematological malignancies [1, 2]. In a recent clinical trial, patients with relapsed or refractory multiple myeloma treated with ciltacabtagene autoleucel (cilta-cel; a humanized anti-B-cell maturation antigen CAR-T cell therapy) achieved overall remission rates and stringent complete response rates of up to 84.6% and 73.1%, respectively, while their progression-free survival at 12 months was 75.9% [3, 4]. However, CAR-T cell therapy shows limited efficacy against solid tumors due to their complexity. The most substantial obstacle to their use is the limited availability of suitable tumor-associated antigens [5]. Theoretically, the ideal antigen would be uniformly overexpressed on malignant cells with little or no expression on healthy cells, thus avoiding off-target toxicity [6]. Moreover, an immunosuppressive tumor microenvironment and inaccessible tumor sites limit the efficacy of CAR-T cells [7–9]. Considering these barriers, optimizing CAR structure may play a critical role in improving the ability of CAR-T cells to eradicate solid tumors.

CARs comprise an extracellular binding (generally a single-stranded variable fragment [scFv]), a transmembrane, a spacer, and one or more cytoplasmic domains [10]. ScFv is used to develop CAR-T therapies because of its small size, high affinity, and antigen specificity [11]. It comprises the antibody V_H and V_L regions connected by a linker. However, the variable structural domain of scFv (V_H-V_L) may induce CAR aggregation and trigger a cytotoxic signaling cascade that leads to T cell exhaustion [12]. Nanoantibodies (also known as VHH) can avoid the above shortcomings [13]. Thus, VHH antibodies are currently used as alternatives to scFv as the target structural domain for CAR [14]. Additionally, some studies have genetically engineered CAR-T cells by recruiting chemokines, such as chemokine (C–C motif) ligand (CCL19) and interleukin (IL)-7, and synthesizing T cell receptor/synthetic T cell receptor and antigen receptor double-chain chimeric receptors to enhance the antitumor capacity of conventional CAR-T cells [15, 16]. Our previous studies have demonstrated that multichain CAR structures, such as killer cell immunoglobulin-like receptors (KIR)-CAR/DNAX-activating protein (DAP12), could enhance the safety and efficacy of CAR-T cells for solid tumors; in particular, multichain CAR-T targeting mesothelin (MSLN) significantly improved T cell depletion [17, 18]. MSLN is a glycosylated phosphatidylinositol-anchored membrane glycoprotein that is usually confined to tumor differentiation antigens on the mesothelium [19]. Specifically, the level of MSLN in the serum, which is considered a tumor marker, is elevated in ovarian cancer and mesothelioma [20]. MSLN is significantly overexpressed in various solid tumors but is either unexpressed or expressed at very low levels in normal tissues [21]. Thus, MSLN has emerged as an attractive target for developing novel immunotherapies. Previous studies have indicated that MSLN-targeted CAR-T cells can treat ovarian cancer, while anti-MSLN CAR-T cells show promise for improving the prognosis of patients with malignant mesothelioma [22, 23]. Based on these findings, we designed a human anti-MSLN CAR and conducted a clinical trial in patients with MSLN-expressing ovarian cancer or mesothelioma to further explore the clinical safety and feasibility of MSLN CAR-T cell therapy (ClinicalTrials.gov: ChiCTR2100046544) at the First Affiliated Hospital of Nanjing Medical University, China, between August 2021 and February 2024.

Our current lack of a comprehensive description of T cell behavior in vivo has complicated research and hindered the widespread application of CAR-T cell therapy, which has various adverse reactions and low remission rates. Therefore, a more thorough exploration of the mechanism of CAR-induced T cell activation will help us understand the reasons for the success or failure of CAR-T cell immune responses. Single-cell RNA sequencing (scRNA-seq) technology has progressed rapidly and has become a major tool for deciphering immune cell phenotypes in immunotherapy [24]. In addition, scRNA-seq can help identify the behavioral and molecular profiles of CAR-T cells before and after treatment. Therefore, we performed scRNA-seq analyses on CAR-T cells, tissue, and blood (pre- and post-treatment) samples from patients receiving MSLN-targeted CAR-T cell (KT032) therapy. We attempted to determine the transcriptional profile of cells associated with CAR-T cell infusion.

Methods

Study design

This was a researcher-initiated, single-arm, open-label, single-center, first-in-human clinical trial (ChiCTR2100046544) designed to evaluate the safety and efficacy of KT032 in patients with mesothelin-expressing advanced solid tumors between August 2021 and February 2024, at the First Affiliated Hospital of Nanjing Medical University. Tumor biopsies were evaluated for mesothelin (MSLN) expression at enrolment, followed by leukapheresis and KT032 manufacture. Positron emission tomography (PET) or computerized tomography (CT) scans were performed at baseline and at 1, 2, 3, and 6 months after KT032 infusion. The primary objective of this phase I study was to evaluate safety and efficacy. The secondary objective was to assess the efficacy of this treatment by determining the objective response rate, disease control rate, duration of response, PFS, and OS. Responding patients included those who achieved partial or complete response, whereas non-responders included those with PD. The exploratory objective was correlating responses with KT032 expansion, persistence, and functionality. For the phase I dose escalation part of the study, patients received a single intravenous infusion dose with or without LD. The protocol-specific DLs were 0.5 × 10⁶/kg, 1 × 10⁶/kg, and 1.5 × 10⁶/kg. Informed consent was obtained from all participants in the clinical trial, as stated in the clinical trial protocol. Participants were not compensated for their participation in the clinical trial. The trial was approved by the institutional review boards of the participating center, and all patients provided written informed consent. The participating site was Jiangsu Province Hospital, Jiang Su, China. Eight subjects with ovarian cancer and mesothelioma were enrolled from August 31, 2021, to September 16, 2022.

Patient eligibility

Eligibility for the study required patients to meet all the following criteria: Ages ranging from 18 to 70 years, regardless of sex. Histologically or pathologically diagnosed advanced ovarian cancer or malignant mesothelioma. Recurrence after receiving second-line or higher treatment, such as chemotherapy or targeted drugs. The tumor was pathologically reviewed with confirmed positive mesothelin expression on ≥ 15% of tumor cells using immunohistochemistry. The patients had at least one lesion that met the evaluable and measurable criteria defined by RECIST version 1.1. ECOG performance status of 0 or 1. The patient must have adequate organ function according to the following laboratory values: left ventricular ejection fraction ≥ 40% as measured using resting echocardiogram; creatinine ≤ 200 μmol/L; absolute neutrophil count ≥ 1.5 × 10⁹/L; platelets ≥ 80 × 10⁹/L; hemoglobin ≥ 80 g/L; oxygen saturation of blood > 91%; total bilirubin ≤ 2 × ULN; alanine aminotransferase, and aspartate aminotransferase ≤ 2.5 × ULN. Venous access with venous or leukapheresis. The patients voluntarily participated in the study and signed an informed consent form.

Patients who met any of the following criteria were excluded from the study. Pregnant and lactating women. Patients who underwent chemotherapy or radiotherapy within 3 days before leukapheresis. Patients were administered systemic steroid drugs within the first 5 days of the blood collection period (excluding those who had recently or are currently using inhaled steroids). Patients were administered drugs that stimulated bone marrow hematopoietic cell generation within 5 days before blood collection. Patients who had previously used gene or cell therapy products. Patients who had a history of epilepsy or other central nervous system diseases. Active infection with HIV, hepatitis B, hepatitis C, or human T lymphotropic virus. Patients who suffered from other tumors in the past 5 years. Patients with severe pleural and ascitic fluid. Patients who had active or uncontrollable infections requiring systemic treatment within 14 days before enrollment. Patients who received other antitumor treatments (excluding pretreatment chemotherapy) within 2 weeks before the start of the study. Patients who were unable or unwilling to comply with the study protocol requirements, as judged by the researchers.

IHC analysis

Mesothelin expression was assessed in all patients by pathology laboratory of Jiangsu Provincial Hospital using a validated IHC assay performed on 3 ~ 5 mm-thick sections of formalin-fixed paraffin-embedded (FFPE) tumor tissue. Immunostaining was performed which is a monoclonal primary antibody that binds to mesothelin in paraffin-embedded tissue sections. Mesothelin protein expression was determined using both the percentage of stained viable tumor cells and the staining intensity. Study eligibility required biopsy tissue to express membranous mesothelin in at least 15% of the viable tumor cells at a staining intensity of 2 + .

KT032 manufacturing, preconditioning, and bridging therapy

The autologous MSLN-targeted CAR-T cells (KT032) were manufactured in the GMP facility at Nanjing CART Medical Technology Co., Ltd. Subjects completed single apheresis at the clinical center and transported to the GMP facility. Then, PBMCs were isolated and purified by Ficoll density gradient centrifugation. CD3 + T cells were separated and seeded in culture medium supplemented with human IL-2. CD3/CD28 dynabeads were used for T cell activation. After 24 h of activation, T cells were transduced with a lentivirus encoding MSLN-targeted CAR. T cells were expanded in the presence of Optimizer medium (Gibco) and IL-2. KT032 was harvested and formulated after the cell quantity reached the required dose. The in-process and release tests were performed for each lot of KT032 cells. After release, the patients received an infusion of KT032 (day 0, week 0) within 7 days of preconditioning therapy. Nonsteroidal anti-inflammatory drugs and antihistamines were administered within half an hour of cell infusion. Preconditioning therapy was administered as follows: fludarabine on days –4 and –3; cyclophosphamide on days –4, –3, and –2. Bridging chemotherapy was allowed for patients whose tumor burden was heavy or who had the potential to rapidly progress, according to the investigators’ discretion during the KT032 manufacturing period. Investigators could also individualize the regimen according to the patient’s previous antitumor therapy and clinical condition.

KT032 detection using qPCR in peripheral blood

The persistence of KT032 cells in peripheral blood was determined by quantifying the junction region between the 4-1BB and T2A regions of the lentivirus transgene using qPCR. Patient samples (2 mL of peripheral blood) were collected in K2EDTA tubes at baseline and after infusion. gDNA was extracted using the QIAamp DNA Midi Kit (Qiagen, Hilden, Germany). The standard curve for transcript copy number was established by amplifying a tenfold serial dilution of a plasmid encoding KT032 CAR from 4 × 10⁶ to 40 copies. The number of transgene copies per microgram of gDNA was determined using a 7500 Fast (Thermo Fisher Scientific) instrument, with samples in triplicate.

KT032 detection using flow cytometry in peripheral blood

Peripheral whole blood samples were used to lyse red blood cells stained with Mouse Anti-Human CD45-PE (BD Biosciences, 555483), mouse anti-human CD3-FITC (BD Biosciences, 555332), Biotin-SP (long spacer) AffiniPure Goat Anti-Mouse IgG, F(ab') fragment specific (Jackson, 115–065-072) antibodies, and streptavidin-APC (SouthernBiotech, 7100-11 M) for CAR detection. Flow cytometry was performed using a NovoCyte 3110 instrument (Agilent).

Cytokine measurement

Baseline and post-infusion peripheral blood samples were collected from patients and processed for plasma. Plasma was analyzed for the following cytokines: IL-2, IL-4, IL-6, IL-10, TNF-α, and IFN-γ using a BD cytokine cytometric bead array kit and conducted according to the manufacturer’s guidelines.

Single-cell sequencing

The single-cell suspension was counted using a Countess® II Automated Cell Counter based on trypan blue staining. The optimum concentration of cells should be at least 700 cells/µL and the proportion of living cells was more than 80% by quality control. The cell suspension, barcoded gel beads, and partitioning oil were loaded onto a 10 × Genomics Chromium Chip that generates single-cell Gel Bead-In-EMlusion (GEMs). Captured cells were lysed, and the transcripts were barcoded using reverse transcription inside individual GEMs. Then, cDNA and cell barcodes were PCR-amplified. The scRNA-seq libraries were constructed using Chromium Next GEM Single Cell 5’ Reagent Kits v2.0 (PN-1000395 and PN-1000352) and sequenced on an Illumina NovaSeq6000 platform to generate 2 × 150-bp paired-end reads. The single cell by gene matrices for each sample were calculated by using Cell Ranger (version 5.0.0) with the human reference genome version GRCh38. Before cell clustering by using the Seurat R package (version 3.1.1), high-quality cells were preserved by the following parameters: first, each sample was multicell-filtered using DoubletFinder software, followed by further filtering based on the number of genes identified, the number of unique molecular identifiers (UMIs), and the proportion of mitochondrial genes expressed in individual cells. In ovarian cancer, cells that were expressing 200 to 7800 genes, less than 56,000 UMIs, and mitochondrial gene percent of < 25% for further analysis. In mesothelioma, the number of genes expressed per cell ranged from 200 to 10,000 and less than 30,000 UMIs per cell, and mitochondrial gene expression of < 20% was retained. In manufactured CAR-T cells, the number of genes expressed per cell ranged from 210 to 7000, the number of UMIs per cell < 54,000, and maximum mitochondrial content of 15% as a high cutoff. Gene expression of filtered matrices was normalized through “LogNormalize” method and used Harmony for data merging and batch effect correction. For visualization of the high-dimension data, principal component analysis (PCA) was performed for dimensional reduction with top2000 highly variable genes selected by FindVariableFeature. Cell clustering was used FindClusters with 10 PCs and 0.5 resolution parameter. Cell’s cluster information was generated to visualize with the same PCs using Uniform Manifold Approximation and Projection (UMAP).

Gene differential expression and enrichment analysis

In the Seurat package, FindMarkers was used to analyze the differences between the two groups of cells. The threshold is | log₂FC |≥ 1, and the ratio of cells expressing the target gene in any group is ≥ 0.1. The MAST obstacle model was used to test the significance of differences. Significant p-values of the differentially expressed genes were corrected using Seurat’s Benjamini–Hochberg method. Genes with corrected p-values ≤ 0.05 were selected as significant DEGs and were used for subsequent enrichment analysis. Gene Ontology (GO) is an internationally standardized gene functional classification system with three ontologies: molecular function, cellular component, and biological process. To reveal the biological function of DEPs, all peak-related genes were mapped to GO terms in the GO database [GO.db 3.8.2 (April 26, 2019)]. Gene numbers were calculated for every term and significantly enriched GO terms in DEG compared with the genome background were defined by a hypergeometric test. Genes interact with each other to perform specific biological functions. Pathway-based analysis helps further understand the biological functions of genes. The Kyoto Encyclopedia of Genes and Genomes is a major public pathway-related database (Release 94). Pathway enrichment analysis identified significantly enriched metabolic pathways or signal transduction pathways in the DEGs compared with the whole genome background.

Xenograft model

The NOD/ShiLtJGpt-Prkdcem26Cd52Il2rgem26Cd22/Gpt (NSG) mice were purchased from Gempharmatech (Nanjing, China) and raised under SPF condition. Subcutaneous xenografts were established by injecting 1 × 107 SKOV3 cells suspended in 200 µl PBS subcutaneously into nude mice (Gempharmatech, Nanjing, China). Tumor volumes were examined twice a week using the following formula: volume (mm3) = length × width2/2. Mice were humanely euthanized with carbon dioxide (CO₂) inhalation method according to ethical guidelines at the designated time point of 5 weeks after inoculation. The study was approved by the Institutional Ethics Committee of Nanjing Medical University (IACUC2103036).

Statistical analysis

Experimental data were analyzed using GraphPad Prism version 8.0 and the SPSS software package. Differences between groups were assessed using unpaired Student’s t-tests or two-way ANOVA, with statistical significance defined as a P < 0.05. The results are presented as mean with standard deviation, and error bars were labeled in the graphs to illustrate the differences in measurements between replicates.

Results

MSLN-KIR CAR-T cells show robust cytotoxicity and antitumor efficacy in vitro and in vivo

Besides KIR-activation receptors, other activation receptors exist on the membranes of natural killer (NK) cells, including NKP44, NKP46, NKP30, and NKG2D. The functionality of these receptors in the CAR form of the NKR complex, instead of in the KIR, remains unknown. Here, we engineered new multiple-chain MSLN targeting CARs, each containing only one of the NKP44, NKP46, NKP30, or NKG2D intracellular, transmembrane, and short extracellular domains (Fig. 1A). As shown in Fig. 1B, flow cytometry revealed that all CARs were successfully expressed and presented to the T cell membrane, with the highest positivity for the KIR CAR. No significant differences were observed in the proliferative capacity of T cells transduced with different CAR structures (Fig. 1C). Furthermore, all CAR-T cells showed robust tumor cell-killing ability against MSLN high-expressing OVCAR3 cells; however, KIR CAR-T cells were significantly better at killing MSLN low-expressing SKOV3 cells (Fig. 1D). Additionally, KIR CAR-T cells secreted more interferon (IFN)-γ than other CAR-T cells under SKOV3 cell stimulation. In contrast, the secretion of IFN-γ by 30 CAR-T and two-dimensional (2D) CAR-T cells was relatively lower than that in other CAR-T cells when co-cultured with OVCAR3 cells (Fig. 1E). Moreover, KIR CAR-T cells secreted a higher level of IL-2 following stimulation with OVCAR3 cells than 30 CAR-T and 2D CAR-T cells (Fig. 1F). Moreover, the Tcm ratio of KIR CAR-T cells was higher than that of the other CAR-T cells (Fig. 1G). Since the cytotoxicity and cytokine secretion capacity of 2D CAR-T cells were poor, our subsequent in vivo efficacy assessments did not include 2D CAR-T cells. In the SKOV3 subcutaneous xenograft model, mice were infused with different CAR-T cells or control non-transduced (NTD) cells. As shown in Fig. 1H, only KIR CAR-T cells substantially inhibited or eradicated tumors, whereas the other CAR-T cells merely slowed tumor growth. In conclusion, KIR CAR confers T cells with stronger cytotoxic and antitumor effects than other activated NKR CARs, which could be a potential therapeutic strategy for inducing MSLN expression in patients with cancer.

Fig. 1.

Killer cell immunoglobulin-like receptor (KIR) Chimeric antigen receptor (CAR) is the optimal CAR structure based on natural killer (NK) cell activation receptors. A CAR design based on various NK cell activation receptor combinations. B Flow cytometric analysis of CAR-positive T cell percentage 7 days after lentiviral transduction. C Expansion of different CAR-T and control non-transduced (NTD) cells. D xCELLigence RTCA system evaluation of in vitro cytotoxicity of different MSLN CAR-T and NTD cells against SKOV37 and OVCAR3 cells. E ELISA analysis of IFN-γ secretion by different MSLN CAR-T cells following co-culture with SKOV37 and OVCAR3 cells (E:T, 2:1). F ELISA analysis of IL-2 secretion by different MSLN CAR-T cells following co-culture with OVCAR3 cells (E:T, 2:1). G Flow cytometry-based gating strategy for identifying Tn, Tcm, Tem, and Tef subsets in NTD and different MSLN CAR-T cells. H Tumor growth curves demonstrating the antitumor efficacy of NTD or different MSLN CAR-T cells in a SKOV3-derived xenograft model. Statistical significance: **P < 0.01

Efficacy of MSLN-KIR CAR-T cells as a treatment for ovarian cancer and mesothelioma

To evaluate the safety and efficacy of MSLN-KIR CAR-T cells (KT032) in ovarian cancer and mesothelioma therapy, we conducted a single-arm, open-label clinical trial at the First Affiliated Hospital of Nanjing Medical University, China, between August 2021 and February 2024. Nine MSLN-positive patients with advanced recurrent refractory ovarian cancer (eight patients) or mesothelioma (one patient) were enrolled in the trial after MSLN screening, and signed informed consent forms were obtained from all participants from August 2021 to September 2022. All nine patients underwent leukapheresis, which failed in one patient with ovarian cancer (Fig. 2A). All eight patients with successful leukapheresis had previously received at least two prior standard lines of therapy, which are shown in Table 1. MSLN KIR CAR-T cells were manufactured via lentiviral transduction in the GMP facility at Nanjing CART Medical Technology Co., Ltd., then cryopreserved and transferred to the First Affiliated Hospital of Nanjing Medical University. Among the eight patients, three received a low dose (0.5 × 10⁶ cells/kg), two received a medium dose (1.0 × 10⁶ cells/kg), and three received a high dose (1.5 × 10⁶ cells/kg) of CAR-T cells, which were infusion following fludarabine- and cyclophosphamide-mediated lymphodepletion. We adjusted the doses of fludarabine (50–93 mg∙m⁻² day⁻¹, 1 day) and cyclophosphamide (500–930 mg∙m⁻² day⁻¹, 1 day) according to the physical condition of each patient (Table 1). As of the evaluation cutoff date (28 days after infusion), two of eight patients discontinued this study after their CAR-T cell infusion due to rapid disease progression (Fig. 2A).

Fig. 2.

A–I Evaluation of the efficacy and safety of KT032 in patients with ovarian cancer and mesothelioma. A The study schema is shown. Subjects were screened for enrolment, underwent leukapheresis for CAR-T cell manufacturing, and infused with varying doses of CAR-T cells. B A swimmer plot was drawn to show patients’ response and time to disease progression in months. C A waterfall plot was drawn to show the determined percent change in tumor burden. D CT images showing the longest dimension of an initially responding tumor lesion at baseline, and at different time points after receiving KT032 cell infusion

Table 1.

Patient characteristics

Patient	Age (years)	Gender	Weight (kg)		Dose (cells/kg)		CAR-T positive (%)	Response			CRS		Indication
1	55	Female	62		1.5 × 106/kg		48.4%	PR			3		Ovarian cancer
2	68	Female	57		1.0 × 106/kg		55.8%	PR			1		Ovarian cancer
3	58	Female	57		0.5 × 106/kg		45.6%	SD			1		Ovarian cancer
4	39	Female	60		0.5 × 106/kg		44.8%	SD			1		Mesothelioma
5	54	Female	35.5		1.5 × 106/kg		71.4%	SD			1		Ovarian cancer
6	50	Female	64		0.5 × 106/kg		63.2%	SD			0		Ovarian cancer

Of the six evaluable patients, two achieved partial response (PR), and four achieved stable disease (SD) status. However, all the six patients progressed within 9 months, and two patients were still alive until the latest follow-up. The median progression-free survival (PFS) was 5.5 months, and the median overall survival (OS) was 10.5 months (Fig. 2B and C). An enhanced computed tomography (CT) scan indicated that liver metastases were reduced by 31% and peritoneal effusion disappeared in Patient 1. Patient 2 exhibited a 42% reduction in the total maximum diameter of the measurable lesion. Meanwhile, the changes in tumor lesions or metastases were not significant in other patients (− 18.7% to + 4.56%) (Fig. 2D–I). Notably, the level of cancer antigen 125 (CA125) in Patient 1 was significantly decreased from 1687 U/mL to 150 U/mL 32 d after CAR-T therapy, while Patients 2 and 3 also showed a decrease in CA125 after CAR-T treatment, albeit with a rebound increase observed at day 20. For Patients 4 and 5, CAR-T therapy failed to reduce the CA125 level (Additional file2: Fig S1A-B).

Pharmacokinetics and safety of MSLN-KIR CAR-T cells

Flow cytometry and qPCR analysis showed that MSLN-KIR CAR-T cells in the peripheral blood peaked at approximately 7 days (6–10 days) after infusion (Fig. 3A). However, the peak of CAR-T cell expansion represented by flow cytometry was not very consistent with data from the qPCR analysis, which may be due to the discrepancy in total cell count in peripheral blood. Activation and expansion of MSLN-KIR CAR-T cells induced the secretion of high levels of cytokines IL6 and IFNγ in five of six patients (besides Patient 6, who showed slight changes in IL6 and IFNγ levels), while levels of TNFα, IL2, IL4, and IL10, etc. were slightly affected by CAR-T cell treatment. As a result, all five patients besides Patient 6, developed cytokine release syndrome (CRS) (one patient with grade 3, four patients with grade 1) accompanied by high levels of IL-6 and IFNγ (Fig. 3B–G); accordingly, corticosteroids, indomethacin, or tocilizumab (8 mg/kg) were used to treat CRS. Among the five patients with CRS, Patient 1 experienced immune pneumonia and received mechanical ventilation and high-dose corticosteroids prednisone (1 g/day) combined with tocilizumab (8 mg/kg). The patient’s condition improved significantly after treatment. Moreover, CT evaluation indicated that immune-related pneumonia was relieved (Additional file2: Fig S1C). Notably, the proliferation of CAR-T cells was limited due to the high-dose corticosteroids in this patient, while the number of CAR-T cells increased again (constituting 6.65% of the total peripheral blood T cells) on the 25th day after infusion, indicating that CAR-T cell activity may be restored by the withdrawal of corticosteroids (Fig. 3A and B). The other most common adverse effects during the MSLN-KIR CAR-T cell therapy were nausea, which was observed in 83% (5/6 patients), and no immune effector cell-associated neurotoxicity syndrome was observed in any patients. Other adverse events related to the MSLN-KIR CAR-T cell treatments were summarized in Table 2.

Fig. 3.

KT032 cell pharmacokinetics and cytokine levels in peripheral blood of patients. A The pharmacokinetics of KT032 cell copy number in patients were evaluated using qPCR. The pharmacokinetics of KT032 cells in patients’ peripheral blood were assessed using qPCR and flow cytometry. B ELISA analysis of IFN-γ levels in patients’ peripheral blood. C ELISA analysis of TNFα levels in patients’ peripheral blood. D ELISA analysis of IL-2 levels in patients’ peripheral blood. E ELISA analysis of IL-4 levels in patients’ peripheral blood. F ELISA analysis of IL-6 levels in patients’ peripheral blood. G ELISA analysis of IL-10 levels in patients’ peripheral blood

Table 2.

Summary of reported adverse events related to KT032 cells by grade reported in more than one subject (unless ≥ grade 3)

All Subjects (n = 6)				Grade 1, n		Grade 2, n			Grade 3, n			Grade 4, n		Total, n
Clinical events
Fatigue				1		1			1			0		3
Nausea				4		1			0			0		5
Ascites				0		0			1			1		2
Vomiting				1		1			0			0		2
Confusion				0		0			1			0		1
Diarrhea				1		1			0			0		2
Dysgeusia				2		0			0			0		2
Fever				2		2			1			0		5
Abdominal pain				0		1			1			0		2
Anorexia				1		1			0			0		2
Anxiety				2		0			0			0		2
Chills				1		2			2			0		5
Constipation				1		1			0			0		2
Dizziness				2		0			0			0		2
Myalgia				1		0			0			0		1
Paroxysmal atrial tachycardia				1		1			0			0		2
Pleural effusion				0		1			0			0		1
Sore throat				1		0			0			0		1
Abdominal distension				0		2			0			0		2
Dyspnea				1		0			0			1		2
Hematologic Events
Anemia				1		1			0			0		2
DIC				0		0			1			0		1
leukopenia				0		0			5			0		5
Nonhematologic Events
Alkaline phosphatase increased				0		1			0			0		1
ALT increased				0		0			1			0		1
AST increased				0		0			1			0		1
Blood bilirubin increased				0		0			1			0		1
Total				23		17			16			2		58

 ALT alanine aminotransferase, AST aspartate aminotransferase, DIC disseminated intravascular coagulation

Associations between MSLN CAR-T cell subtypes and clinical response

The proportion of CAR-positive T cells in the manufactured CAR-T products and the CD4/CD8 ratio of CAR-T cells were not related to patient responses. We hypothesized that subpopulations and functional gene signatures of individual CAR-T products may influence patient outcomes; hence, we conducted scRNA-seq on MSLN CAR-T cells produced from four patients. Cryopreserved CAR-T cells from four patients with ovarian cancer undergoing treatment were thawed and prepared for scRNA-seq. The number of CAR-T cells that passed the initial quality control ranged from 8093 to 9498 per patient. Further dimensionality reduction and unsupervised clustering analysis showed that 12 clusters (0–11) were identified, and CAR-T cells belonging to each cluster were detectable in all four patients. Next, we evaluated the differences in the clusters and transcriptomic signatures of the KT032 products from patients who had a good response (R) compared with those who did not. Patients 1 (HUME) and 4 (YUQI) were defined as having a good response, as Patient 1 achieved PR, while Patient 4 demonstrated signs of tumor lysis, as evaluated using CT during the two weeks after KT032 treatment. Meanwhile, Patients 5 (YLFE) and 6 (SXME) were defined as not responding, since both were determined to have progressed using fluorodeoxyglucose positron emission tomography. No significant differences in CD4 + and CD8 + T cell subsets were observed between the two patient groups (Fig. 4A–C). Differential gene expression analysis showed significant upregulation of GSTM1 and NOTCH2NLB gene expression in CAR-T cells from the responding patients (Fig. 4D). GSTM11 is essential for controlling the intracellular redox state of immune cells, while NOTCH2NLB is functionally similar to the intracellular structural domain (NICD) of Notch1 and activates the Notch signaling pathway. These findings indicate that the control of the intracellular redox state and the Notch signaling pathway may be important for the long-term antitumor activity of MSLN-KIR CAR-T cells.

Fig. 4.

Single-cell sequencing analysis of the properties of KT032 from different patients. A The bar chart shows the proportional distribution (left) and the absolute distribution (right) of various T cell types in four cases with KT032 cell transfusion. B The bubble plot displays the expression levels of key marker genes in various types of T cells with KT032 cell transfusion. The size of the bubbles represents the percentage of gene expression in the cells, and the color of the bubbles represents the average level of gene expression. C The distribution of various types of T cells in the t-distributed stochastic neighbor embedding (tSNE) plot for four cases with KT032 cell transfusion. D The volcano plot of the differential analysis shows the upregulated and downregulated genes in various T cell clusters with KT032 cell transfusion products compared with PR samples. Upregulated genes are shown in red, downregulated genes in blue, and the symbols of the top five genes with the largest differences in upregulation and downregulation are labeled. The differential threshold is set as fold change > 2 and P < 0.05

Gene expression signatures of immune and cancer cells in the ovarian cancer tumor microenvironment (TME) after MSLN-CAR-T cells infusion determined using scRNA-seq

To further explore whether immune cell components and their gene signatures in the immune microenvironment of patients with ovarian cancer correlated with clinical responses, we first compared the representation of immune cell types and functional states between one patient with PR and one with SD. As shown in Fig. 5A, T cells, fibroblasts, and pericytes were significantly enriched within the TME of the patient who achieved PR, while epithelial cells were enriched within the TME of the patient with SD. Moreover, dimensionality reduction and unsupervised clustering analysis identified 11 CD4 + and CD8 + T cell clusters in both patients, but patients with PR had relatively higher proportions of CD4 naive, CD4 Teff, CD4 Tcm, and CD8 naive cells, while patients with SD had much more CD8 Teff cells, including CD8 Teff, CD8 Teff CD74, and CD8 Teff mitotic cells (Fig. 5A–D). This suggests that more T cells, especially CD4 clusters, within the TME may be associated with a good response to CAR-T cell infusion.

Fig. 5.

Single-cell sequencing analysis of the properties of the immune and tumor cells after infusion of KT032 in patients with ovarian cancer. A The bar chart shows the proportional distribution of various cell types in tumor samples from patients with ovarian cancer before and after KT032 treatment. B The bar chart shows the proportional distribution of various T cells in tumor samples from patients with ovarian cancer before and after KT032 treatment. C The tSNE plot illustrates the distribution of different types of T cells in tumor samples from patients with ovarian cancer before and after KT032 treatment. D The bubble plot displays the expression levels of key marker genes for different types of T cells in tumors from patients with ovarian cancer. The size of the bubbles represents the percentage of gene expression in the cells, while the color represents the average level of gene expression. E The volcano plot of the differential analysis shows the upregulated and downregulated genes in various T cell clusters of tumors from patients with ovarian cancer after treatment compared with before treatment. Upregulated genes are shown in red, downregulated genes in blue, and the symbols of the top five genes with the largest differential expression are labeled. The differential threshold is fold change (FC) > 2 and P <0.05. F The bar chart illustrates the proportional distribution of different myeloid cells in tumor samples from patients with ovarian cancer before and after CAR-T treatment. G The bar chart displays the proportional distribution of different malignant cells in tumor samples from patients with ovarian cancer before and after CAR-T treatment. H The tSNE plot shows the distribution of different malignant cells in tumor samples of patients with ovarian cancer before and after CAR-T treatment. I The bubble plot shows the expression levels of the top five upregulated genes in various malignant cells in ovarian cancer tumor samples. The size of the bubbles represents the percentage of gene expression in the cells, while the color represents the average level of gene expression. J The volcano plot of differential analysis shows the upregulated and downregulated genes in tumor samples from patients with ovarian cancer after treatment compared with before treatment. Upregulated genes are shown in red, downregulated genes in blue, and the symbols of the top five genes with the largest differential expression are labeled. The differential threshold is FC > 2 and P < 0.05

The proportions of myeloid cells, T cells, and NK_NKT cells increased while that of epithelial cells decreased (Fig. 5A). For T cell clusters, only the proportions of CD4 + Tregs and CD8 + Tact cells increased slightly, while other clusters were not significantly influenced by CAR-T cell infusion (Fig. 5B). Differentially expressed gene (DEG) enrichment analysis showed that these DEGs in CD4 + Tregs post-CAR-T cell infusion were enriched in response to cytokine, cytokine-mediated signaling pathways, immune response, etc., indicating that CAR-T cells and their secreted cytokines may influence Treg cell phenotypes (Additional file2: Fig S2A). Moreover, T cells from patients with PR showed a higher expression of IL12RB2, GZMB, and IL10, indicating greater cytotoxicity and immunomodulatory properties (Fig. 5E). Furthermore, patients with PR had much more MDSC-like macrophages but less CD24-TAM abundance compared with patients with SD. In contrast, the proportions of FN1-TAM and MDSC-like macrophages increased after CAR-T cell therapy (Fig. 5F and Additional file2: Fig S2B-E). Further analysis demonstrated that dysregulated genes in these CD24-TAM cells were enriched in antigen processing and presentation of peptide antigen via MHC class I, cellular response to interferon-gamma, myeloid leukocyte mediated immunity, etc., which indicated that CAR-T cell therapy might induce the transformation of the TAM (Additional file2: Fig S2F). Moreover, the DEGs in MDSC-like macrophages post CAR-T cell treatment were enriched in immune response, cellular response to cytokine stimulus, response to cytokines, etc. as well as CD4 + Tregs (Additional file2: Fig S2G), which indicated that CAR-T cells would alter the pre-existing suppressive tumor microenvironment. These findings indicate that CAR-T cell infusion may recruit more immune cells and temporarily remodel the TME.

In addition, to investigate which genes in tumor cells may influence ovarian cancer cell resistance to MSLN-DAP CAR-T cell-mediated killing, we compared the changes in tumor cell subpopulations and gene expression profiles after CAR-T cell infusion. Seven clusters were identified by dimensionality reduction and unsupervised clustering analysis (Fig. 5G–I). As shown in Fig. 5J, patients with PR had fewer M0 tumor cells that highly expressed FLOR3, PAGE2B, PAGE2, and C19orf33, and more M1 cells that had low expression of FLOR3, PAGE2B, PAGE2, and C19orf33 compared with patients with SD. The number of M2 cluster cells with high expression of NFKB1, MIR222HG, and LINC00472 was significantly decreased in both patients, while there was an increase in the number of M3 cluster cells that expressed CCL4, NKG7, and CD53. These findings suggest that tumor cells expressing NFKB1, MIR222HG, and LINC00472 may be more sensitive to MSLN-DAP CAR-T cell killing, whereas tumor cells expressing CCL4, NKG7, and CD53 may be resistant to MSLN-DAP CAR-T cell-mediated eradication.

Gene expression signatures of immune and cancer cells in mesothelioma after MSLN-CAR-T cell infusion

We further explored the representation of immune cell types and functional states in patients with mesothelioma who had the longest survival times of 15 and 45 days after CAR-T cell infusion. As shown in Fig. 6A–B, dimensionality reduction and unsupervised clustering analysis showed significant enrichment of NK cells, endothelial cells, and pericytes within the TME 45 days after CAR-T cell therapy compared with 15 days. Conversely, the ratio of mesothelial to myeloid cells decreased, whereas the proportion of T cells did not. Compared with day 15, the increased endothelial cells highly expressed KLF4, BMP4, and FCN3 and had low expression of CHI3L1, HLA-DRA, and PRG4, while the NK cells had upregulated ZNF420 expression, and the pericytes showed increased levels of SNAI1, EFNA1, and BRD9 (Fig. 6C). Moreover, a total of seven clusters of CD4 + and CD8 + T cells were identified, and relatively higher proportions of CD8 + Tact and a lower ratio of CD8 Mitotic characterized in the TME 45 days after CAR-T cell therapy compared with 15 days (Fig. 6D). These altered genes were enriched for protein binding, NADH dehydrogenase (ubiquinone) activity, and catalytic activity (Fig. 6E). Additionally, seven clusters of mesothelioma cells were identified using dimensionality reduction and unsupervised clustering analyses. As shown in Fig. 6F and G, the ratio of M1 clusters with highly expressed SPATC1L, DDX52, BMPR2, and TMEM126B increased 45 days after CAR-T cell infusion, while the ratio of M3 clusters with highly expressed EPGN, NRK, WFDC2, and PRR9 decreased. Moreover, the DEGs in the specimen 45 days after CAR-T cell therapy were enriched in C–C motif chemokine receptor 1 chemokine receptor binding, chemokine receptor binding, and T cell receptor binding (Fig. 6H), indicating that CAR-T cell treatment may recruit and reactivate immune cells in the TME, which in turn induces an antitumor immune response.

Fig. 6.

Single-cell sequencing analysis of the properties of the immune and tumor cells after infusion of KT032 in a patient with mesothelioma. A The bar chart shows the proportional distribution of different cell types in tumor samples from patients with mesothelioma after KT032 therapy in two stages. B The tSNE plot illustrates the distribution of different cell types in tumor samples from patients with mesothelioma after KT032 therapy in two stages. C The volcano plot of differential analysis displays the upregulated and downregulated genes in tumor samples from patients with mesothelioma after KT032 therapy in stage two compared with stage one. Upregulated genes are shown in red, downregulated genes in blue, and the top five genes with the largest differences in upregulation or downregulation are labeled with their symbols. The differential threshold is FC > 2 and P < 0.05. D The bar chart presents the proportional distribution of different T cell types in tumor samples from patients with mesothelioma after KT032 therapy in two stages. E The bubble plot shows the top 20 enriched pathways (q-value) for differentially expressed genes between CD8+ mitotic T cells (left) and CD8+ Tact cells (right) in tumor samples from patients with mesothelioma after KT032 therapy in two stages. F The bar chart displays the proportional distribution of different malignant cell types in tumor samples from patients with mesothelioma after KT032 therapy in two stages. G The heatmap illustrates the expression levels of the top five upregulated genes in different malignant cell types in tumor samples from patients with mesothelioma. The color represents the average gene expression level. H The bubble plot shows the top 20 enriched pathways (q-value) for upregulated genes in the M3 subgroup (left) and M4 subgroup (right) of tumor samples from patients with mesothelioma after KT032 therapy

Discussion

Immunotherapy has become a new hope in humanity’s fight against cancer, and CAR-T therapy has been one of the most promising tumor immunotherapies to emerge in recent years. Since the FDA’s approval of the first CAR-T cell product in 2017, six CAR-T products have since been approved for treating B-cell lymphoma and multiple myeloma, two targeting the B-cell maturation antigen and four targeting CD19. Despite the success of CAR-T cells in treating hematological tumors and the significant prolongation of patient survival, they have not achieved comparable therapeutic effects when used to treat solid tumors. In addition to the lack of specific antigens and tumor heterogeneity, the complex microenvironment of solid tumors is the most significant limitation of CAR-T cell therapy. To improve the efficacy of CAR-T cell therapy in solid tumors, Zheng et al. utilized DAP10/DAP12 to replace 41BB or CD28 and CD3ζ as an alternative costimulation/signaling combination. This led to a significant decrease in IFNγ and IL-2 secretion and showed significantly more cytotoxicity in vitro and in vivo than standard BBZ CAR T cell constructs [25]. In our previous studies, we developed a multichain DAP-CAR mimicking properties of natural immune receptors by using a full-length immune-activated receptor-associated protein (DAP12) to replace CD3ζ as the activation domain, retaining its transmembrane structure and the immunoreceptor tyrosine-based activation motifs. This was combined with the NK cell-activated receptor (a truncated form of KIR2DS2) and the co-stimulatory molecule 4-1BB, which does not depend on CD3ζ signaling and thus resists the inhibition of CAR expression by the TME. The safety of the DAP12 CAR structure was evaluated by clinical trials with patients with B-cell acute lymphoblastic leukemia. In this study, we compared the ability of other NCR structures to activate T cells and found that DAP-CAR was the optimal structure. To this end, we conducted a clinical trial in patients with ovarian cancer and mesothelioma to evaluate the safety and efficacy of MSLN-targeted DAP-CAR-T cells. MSLN-DAP CAR-T cells achieved a 33.3% (2/6) objective response rate and a 100% disease control rate in six patients over 3 months, demonstrating its potential for clinical application.

Adverse events mainly included fever, leukopenia, and erythropenia; additionally, one patient had a pulmonary infection. Neurotoxicity and CRS are the hallmark clinical toxicity events associated with CAR-T cell therapy [26]. CRS is usually observed in patients within one week following treatment and is characterized by fever, hypotension, and neurological deficits, accompanied by drastic increases in cytokine levels, such as IL-6 and TNF-γ [27]. In this study, we found that IL-6 levels were elevated in patients on the day of CRS symptoms. Moreover, we observed a negative correlation between the efficacy of CAR-T cell therapy and adverse immune events. This evidence supports our findings, as some studies have suggested that a certain degree of toxicity is expected to produce an effective response to therapy [28]. However, owing to the small sample size, these findings should be interpreted cautiously. The efficacy of CAR-T cell therapy is positively correlated with the proliferative ability of T cells in vivo [29]. Several factors, such as the number of residual lymphocytes after lymphoblastic conditioning and the use of corticosteroids, may affect the proliferation and persistence of CAR-T cells. Our study found more residual lymphocytes in Patients 5 (0.21 × 10⁹/L) and 6 (0.35 × 10⁹/L), which may be related to the poor proliferation of T cells after CAR-T cell infusion. Moreover, in Patient 1, high doses of corticosteroids were observed to inhibit CAR-T cell proliferation in vivo, whereas cell proliferation was restored after hormone discontinuation. Thus, strategies that enhance cell proliferation may improve patient outcomes. Another critical issue that needs to be addressed is the amount of CAR-T cell infusion. We found that an infusion of 1.0 × 10⁶ cells/kg was relatively safe for patients. In addition, the myeloablative conditioning regimen before KT032 therapy needed to be adjusted according to the patient’s performance score. The maximal depletion of autologous lymphocytes is a prerequisite for effective KT032 therapy.

Evidence indicates that the proportion of different cell subtypes in CAR-T cells correlates with patient outcomes. In particular, a higher percentage of memory T cells usually results in better patient outcomes. However, our scRNA-seq results showed no significant differences in the proportions of each subpopulation of KT032 CAR-T cell products between patients with good or poor responses. In addition to CAR-T cell trials in solid tumors, we observed a relatively short persistence even when KT032 exhibited improved CAR-T cell expansion in the blood. Unlike CART19 cells, which can persist for months and even years after infusion in hematological malignancies, KT032 was undetectable in the peripheral blood of most patients 40 days after infusion and was undetectable in patients even 70 days after infusion. Nevertheless, while KT032 did not meet our expectations, it still exhibited prolonged in vivo persistence compared with BBZ CAR-T cells. Moreover, few KT032 cells were detected within the punctured tumors using scRNA-seq, suggesting infiltration is challenging for KT032 cells in solid tumor therapy. Next, we need to enhance the infiltration ability of KT032 cells by expressing additional chemokines, such as CCL7. In addition to the properties of KT032, we explored whether the heterogeneity of immune cells in the TME was associated with the outcome of KT032 therapy using scRNA-seq. T cells from patients with PR highly expressed IL12RB2, GZMB, IL10, and many more MDSC-like macrophages, but fewer CD24-TAM were observed in patients with PR. Our findings showed that KT032 CAR-T cell infusion could recruit more immune cells and temporarily remodel the TME. However, because single-cell sequencing was conducted in three patients, these findings must be verified in further clinical trials with a larger cohort.

Beatty et al. conducted a phase I study to evaluate the safety and activity of second-generation CAR-T cells against mesothelin, using the same scFv fused to 4-1BB and CD3ζ (BBZ CAR-T) in patients with ovarian cancer, malignant pleural mesothelioma, and pancreatic ductal adenocarcinoma. Although BBZ CAR-T-meso cells were well tolerated and expanded in all patients, only limited clinical activity was achieved [30]. Similarly, Guo et al. reported three patients with ovarian cancer who received second-generation anti-MSLN CAR-T cell therapy, and no effective treatment activity was observed [31]. This, coupled with the therapeutic effects we achieved, suggests that the second-generation BBZ CAR structure may have difficulty breaking through the therapeutic barrier for solid tumors, whereas DAP-CAR might be expected to crack the door to solid tumor cell therapy.

Conclusions

Although the application of CAR-T cells to treat solid tumors has been attempted for several years, many clinical trials have shown that CAR-T cell therapies have not yet achieved breakthroughs in solid tumor treatment. This study demonstrated the safety, efficacy, and feasibility of administering MSLN targeting novel multiple-chain CAR-T cells in patients with ovarian cancer and mesothelioma. Moreover, we characterized the correlation between patient outcomes and the properties of CAR-T cells or heterogeneity of the TME. Our findings from the first human clinical trial of KT032 provide an optimized direction for subsequent studies to test next-generation KT032-based modified CAR-T cells targeting mesothelin solid tumors. This study has some limitations. First, additional multifaceted optimization strategies that prolong KT032 cell persistence and promote trafficking to tumors are needed to improve outcomes with KT032 cells in MSLN-expressing solid tumors. In addition, only eight patients were enrolled in this clinical trial, and more patients are needed to further verify the safety and efficacy of KT032. Furthermore, single-cell sequencing was only conducted on samples from three patients, and these results need to be verified in a larger cohort of patients.

Authors’ contributions

Concept and design, T.X., E.W., M.S. and Y.S.; provision of study materials or patients, H.Z., X.Z., X.C., C.Z., S.F., C.S., T.W.; collection and assembly of data, all authors; data analysis and interpretation, T.X., C.W., J.Z., M.S. and Y.S.; manuscript writing and critical evaluation. All authors read and approved the final manuscript.

Funding

This work was supported by grants from the National Natural Science Foundation of China (82372625 to M Sun, 82060559 to E.X. Wang, 82172889 to Y.Q. Shu, 82372593 to T.P. Xu); Natural Science Foundation of Jiangsu Province (BK20220259 to M. Sun, BK20211381 to T.P. Xu); The Affiliated Suzhou Hospital of Nanjing Medical University, Suzhou Municipal Hospital, Gusu School (GSKY20210112 to M. Sun); and Suzhou GuSu Health Talent Research Project (Grant number GSWS2023001 to M.S.).

Data availability

The dataset(s) supporting the findings of this study are included within the article.

The raw sequence data generated within this study have been deposited to the Genome Sequence Archive in BIG Data Center. Raw sequencing reads are available from Beijing Institute of Genomics (BIG) HRA006416 (https://ngdc.cncb.ac.cn/search/all?&q=HRA006416) [32]

The GSVA package and related code are available from https://github.com/rcastelo/GSVA [33]. The CellphoneDB package and related code are available from https://github.com/Teichlab/cellphonedb [34].

Declarations

Ethics approval and consent to participate

This study was approved by the Institution Review Board of The First Affiliated Hospital of Nanjing Medical University (2021-SR-084). This study conformed to the principles of the Helsinki Declaration. Healthy donors or patients consented to participate in the study. All participants provided written informed consent prior to participation. The animal experiment in this study was approved by the Institutional Ethics Committee of Nanjing Medical University (IACUC2103036).

Consent for publication

All participants provided written informed consent for publication.

Competing interests

Authors EX-W, C-W, and JZ-Z were employed by Nanjing CART Medical Technology Co., Ltd. The other authors declare that they have no competing interests.