Introduction

Immunotherapies have revolutionized the field of oncology1 with immune checkpoint inhibitors being approved for the treatment of over 85 indications in the USA.2 Chimeric antigen receptor (CAR)-T cells are transforming the management of hematological malignancies with several approved products to date.3 Developing CAR-T cells for solid tumors has been more challenging due to paucity of suitable target antigens, increased risk for on-target/off-tumor toxicity, immunosuppressive intratumoral microenvironment and impaired T cell infiltration into tumor tissue.4 5 Genetically unmodified tumor infiltrating lymphocytes (TILs), expanded ex vivo from resected tumor lesions, have emerged as a potentially efficacious treatment for solid malignancies.6–8 Classical TIL therapy does not target an a priori defined tumor-specific antigen but relies on the amplification of intrinsic antitumor T cell immune activity.

In contrast, autologous T cells genetically modified to express specific T cell receptors (TCRs), often derived from healthy donors or cancer patients, allow for directing T cell activity against tumor antigens of choice. Compared with CAR-T cells, TCR-T cell therapy offers more flexibility, as it allows targeting of intracellular as well as cell surface antigens.9 Even though TCR-engineered T cells can be directed against tumor-specific mutations,10 11 most currently developed product candidates focus on tumor associated or more specifically cancer testis antigens.4 5 These proteins are expressed in tumor cells, but their expression is very limited in normal cells, except for germ cells, making them ideal targets for adoptive cell therapy. Melanoma-associated antigen A1 (MAGEA1) protein is a cancer testis antigen prevalent in melanoma, liver, lung, esophageal, head and neck and gastric cancers.12–15 To date, there are no clinical data available on TCR-engineered T cells directed against this antigen. Hence, our objective was to study the safety and efficacy of IMA202 in a first-in-human, dose escalation, multisite, basket trial in human leucocyte antigen (HLA)-A*02:01 positive patients with MAGEA1-positive advanced solid tumors. IMA202 consists of autologous genetically modified cytotoxic CD8+ T cells expressing a TCR, which is specific for a nine amino acid peptide derived from MAGEA1.16

Materials and methodsPatients

Eligible patients were ≥18 years of age and presented pathologically confirmed advanced and/or metastatic solid tumor with recurrent/progressing and/or refractory disease, HLA-A*02:01 expression, MAGEA1-positive tumor as assessed by quantitative PCR (qPCR) from a fresh biopsy, Eastern Cooperative Oncology Group (ECOG) performance status of 0–1. In addition, eligible patients presented adequate organ and marrow function, measurable disease according to Response Evaluation Criteria in Solid Tumors (RECIST) V.1.1, adequate hepatic, renal and pulmonary function, life expectancy >3 months, adequate serum creatinine level, acceptable coagulation status and received available standard-of-care treatments. Patients with a history of other malignancies within the last 3 years, prior stem-cell or organ transplantation, active viral infections, autoimmune diseases or active brain metastasis, as well as pregnant or nursing women were excluded. A full description of the eligibility criteria can be found in online supplemental section.

Trial design and statistical analysis

IMA202-101 was a multicenter, 2+2 dose escalation, open-label, phase 1 basket trial evaluating the safety and tolerability of treatment with IMA202 in MAGEA1-positive recurrent and/or refractory solid tumor patients. The primary objective was the evaluation of safety and tolerability of IMA202 and primary endpoints were the incidence and nature of treatment-emergent adverse events (TEAEs), adverse events (AEs) of special interest, treatment-emergent serious AEs, dose-limiting toxicities (DLTs) as well as the maximum tolerated dose or the recommended phase 2 dose.

Secondary endpoints included the evaluation of TCR-engineered T cell persistence in vivo and antitumor activity including tumor response measured according to RECIST V.1.1, immune-related RECIST (irRECIST) and duration of response.

From August 29, 2018 to November 16, 2021, patients were recruited and the trial was conducted at different locations in the US (University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania; University of Texas MD Anderson Cancer Center, Houston, Texas) and Germany (University Hospital Würzburg, Würzburg, Bavaria; University Hospital Bonn, Bonn, North Rhine-Westphalia; University Hospital C.–G.-Carus Dresden, Dresden, Saxony; University Medical Center Hamburg-Eppendorf, Hamburg).

The following TEAEs occurring from day 0 (IMA202 infusion) to day 21 were defined as DLTs: Any National Cancer Institute–Common Terminology Criteria for Adverse Events (NCI-CTCAE) grade 4 or 5 TEAEs and any NCI-CTCAE grade 3 TEAEs not having resolved to ≤grade 2 within 7 days and having been assessed as at least possibly related to IMA202 (excluding hematological laboratory values); any treatment-emergent autoimmune toxicity ≥grade 3 regardless of attribution. AEs were coded using the Medical Dictionary for Regulatory Activities and severity was graded according to NCI-CTCAE V.5.0 except for cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS) for which we applied the criteria published by Neelapu et al and Lee et al.17 18

We investigated a total of three dose levels (DL) starting at 50×106 transduced CD8+ T cells (CD3+CD8+ dextramer positive T cells) per m2 body surface area (BSA)±20% (DL1) and then escalating to 300×106 cells/m2 BSA±20% (DL2) and 1000×106 cells/m2 BSA±20% (DL3). Additionally, patients were allowed to be enrolled at DLs already cleared for safety or at any intermediate DLs to better understand the safety and tolerability of IMA202 and to provide a T cell product to patients in need.

The 2+2 trial design was an algorithmic-driven dose escalation design based on a maximally acceptable DLT rate of 25%. Based on the number of patients with DLTs, the dose was escalated, enriched to four patients, or de-escalated. The sample size was driven by the algorithmic design with a maximum of 16 patients (4 cohorts with up to four patients).

Progression-free survival (PFS) and overall survival (OS) were summarized using the Kaplan-Meier method to estimate the median survival time (GraphPad Prism V.9). All patients in the analyzed safety analysis set (SAS population) received IMA202 infusion. Correlation analysis of IMA202 peak frequency vs dose and T cell infiltration vs dose were performed using Spearman test.

The trial was completed on March 17, 2023. Strengthening the Reporting of Observational Studies in Epidemiology cohort reporting guidelines were used.19

Trial procedures and treatment

After confirmation of eligibility, patients underwent leukapheresis and IMA202 product was manufactured under current Good Manufacturing Practice-compliant conditions following 7–10 days manufacturing process. Briefly, peripheral blood mononuclear cells (PBMCs) were isolated from fresh leukapheresis and cryopreserved using a controlled rate freezer in a medium containing 10% DMSO until the start of manufacturing. On day 0, PBMCs were thawed, rested for 4–6 hours, and activated overnight using anti-CD3 and anti-CD28 antibodies. On day 1, activated T cells were transduced using a third-generation lentiviral vector encoding the MAGEA1-specific TCR. This was followed by expansion in complete media supplemented with cytokines until harvest. In-process testing including cell counts and TCR expression were performed between day 6 and day 8 to ensure adequate expansion to meet cell dose and determine the day of harvest based on the available number of transduced cells. Eight out of 16 infused products were harvested on day 10, 3 each on day 8 and day 9 and 2 on day 7. Downstream processing included washing, concentration, formulation into drug products using a commercially available cryoprotectant, and cryopreservation. A comprehensive release testing was performed on each drug product for critical quality attributes, that is, safety, purity, identity, and quantity and only passing products were released for infusion. MAGEA1-specific T cells in the products were assessed by pHLA multimer staining (dextramer staining) for dose determination.

Daily lymphodepletion with fludarabine (influenza; 20 mg/m² BSA for hepatocellular carcinoma (HCC) patients and 40 mg/m² BSA for all other solid tumor types) and cyclophosphamide (CY; 250 mg/m² BSA for HCC patients and 500 mg/m² BSA for all other solid tumor types infused according to institutional standards) was applied intravenously on four consecutive days (day −6 to day −3) before IMA202 single infusion at day 0. Protocol prespecified dose adaptions were allowed in case of impaired renal or bone marrow function and advanced age. Online supplemental table 1 summarizes total doses of influenza and CY administered to each patient.

Starting approximately 6 hours after IMA202 infusion, low-dose interleukin (IL)-2 (flat dose of 1×106 IU) was administered subcutaneously every 12 hours for 14 days (online supplemental table 1). IL-2 administration was interrupted at the discretion of the investigator in case of toxicities. After IMA202 infusion patients were closely observed during the treatment and observation period until progressive disease (PD) or death. Thereafter, the follow-up started which ranged from 0 to 9.1 months (median 2.6 months; time from end of treatment and observation period until death (n=14) or withdrawal of consent (n=2)). No patient was lost to follow-up. During follow-up, patients were evaluated for changes in health status, vital signs, physical examination, tumor assessment and OS. In addition, blood samples were collected to test for replication competent lentivirus.

Treatment of CRS and T cell-associated neurotoxicity followed established guidelines17 with more aggressive treatment being possible for patients with high fever (temperature ≥39.5°C). Interruption of IL-2 application was recommended in case of CRS ≥grade 2.

Tumor response was assessed according to RECIST V.1.120 and irRECIST.21

Determination of MAGEA1 expression

Patient tumors had to express MAGEA1, as assessed by an assay based on a reverse transcriptase qPCR analysis of a fresh tumor biopsy specimen stored in RNAlater stabilization solution (ThermoFisher Scientific). For MAGEA1, a correlation between messenger ribonucleic acid (mRNA) and immunopeptidome levels (both in-house data) was established as demonstrated before.22 From this correlation, a reads per kilobase per million mapped reads threshold was generated and translated into a quantitative real-time PCR assay threshold in which MAGEA1 was considered positive if expression levels were above a target-specific delta-cycle threshold (DCt). For DCt generation, Ct average of three reference genes (RPLP0, OAZ1, RPL37A) was calculated and subtracted from observed Ct value for MAGEA1. A threshold (DCt of 7.83 for MAGEA1) was chosen to maximize the sensitivity and specificity of prediction of peptide presentation as described previously.22

Identification and characterization of the MAGEA1-specific TCRIn vitro T cell priming

For priming of CD8+ T cells from healthy donors, streptavidin-coated microspheres served as artificial antigen-presenting cells and were loaded with anti-CD28 antibody (clone 9.3, purified from mouse hybridoma supernatant, University of Tübingen, Germany) and MAGEA1-derived target peptide (KVLEYVIKV)-HLA (pHLA) monomers.23 After 3 weeks of culture with repeated stimulation and medium exchange, cells were analyzed for primed populations using specific MAGEA1-target HLA-A*02:01 tetramers (MAGEA1-Brilliant Violet 650 and PE-Cy7), viability and anti-CD8 staining. Single cells of 2D target tetramer-binding populations were sorted on a BD ARIAIII FACS device into lysis buffer (64.9 mmol/L Tris, 810.8 mmol/L LiCl, 6.5 mmol/L EDTA, pH 7.5) for single cell rapid amplification of cDNA 5’ ends (5’RACE).

5’RACE and TCR assembly

After cell lysis, mRNA was captured by paramagnetic oligo-deoxythymine-beads and cDNA was synthesized using TCR gene-specific primers. TCR transcripts were amplified via nested multiplex PCR.24 The resulting PCR products were analyzed by Sanger sequencing. The sequencing data were used to assemble full length coding DNA sequences in silico using BLAST, CDR3 determination and final chain assembly. The TCRs were resynthesized via gene synthesis at GenScript (Rijswijk, Netherlands).

TCR expression

For transient re-expression, TCR mRNA was in vitro transcribed with the help of mMESSAGE mMACHINE T7 Transcription Kit according to the manufacturer’s instructions. As a template for in vitro transcription, individual TCR chains were PCR amplified with T7 and Kozak sequences at the 5’ end and a 64-adenine 3’ tail. Primary CD8+ T cells were isolated from leukaphereses from HLA-A*02-positive donors via CD8+ magnetic-activated cell sorting (Miltenyi Biotec, Bergisch Gladbach). After 1 day resting, the CD8+ cells were prestimulated for 3–5 days with plate-bound anti-CD3 (10 µg/mL coating concentration) and soluble anti-CD28 (0.1 µg/mL) antibodies. Cells were harvested, electroporated with TCR mRNA in ECM830 electroporator (BTX, Holliston, USA) at 500 V for 3 ms and rested for 20 hours. TCR re-expression was evaluated via tetramer staining along with anti-CD3 and viability staining.

For stable TCR expression, PBMC-derived T cells were activated overnight using immobilized anti-CD3 and anti-CD28 antibodies (0.5 µg/mL each), followed by transduction with a lentiviral vector encoding the MAGEA1-specific TCR. Cells were cultured in the presence of cytokines for additional 6–9 days and harvested for monitoring transduction efficiency, transgene expression, and functional assessment.

T cell functionality

TCR mRNA electroporated T cells were used for activation assay by interferon (IFN)-γ release ELISA after co-culture with target cells loaded with peptides or target-expressing tumor cell lines as well as primary cells from healthy tissues. Released IFN-γ levels were determined after 20 hours of co-culture with the help of BD OptEIA Human IFN-γ ELISA or Biolegend Human IFN-γ-ELISA MAX Deluxe Kits. Primary cells from healthy tissues were obtained from PromoCell (Heidelberg, Germany) or induced pluripotent stem cell-derived cell types were obtained from FUJIFILM Cellular Dynamics (Madison, USA). Tumor cell lines were obtained from ATCC (Virginia, USA) or DSMZ (Braunschweig, Germany). All cells were cultured according to the manufacturer’s instructions and genotyped for HLA-A*02. Culture periods were kept short to maintain cellular characteristics. The T cell activation assays were performed in T cell medium to enable optimal activity of the effector cells. T cell medium consists of RPMI 1640 GlutaMAX supplemented with 10% heat-inactivated human serum, 1% penicillin/streptomycin, 0.2% gentamycin and 1% sodium pyruvate. The EC50 of the MAGEA1-specific TCR was determined from two donors using GraphPad Prism V.6 via nonlinear fit (sigmoidal, 4PL) of log-dose (loaded peptide concentration) versus response (IFN-γ release).

For TCR specificity testing, 10 similar peptides were selected based on their sequence similarity to the target peptide. For this purpose, an in-house database was searched for peptides that share at least five identical amino acids with the target and have been detected at least once on an HLA-A*02-positive healthy tissue by LC-MS/MS. For the selection of a representative set of similar peptides, parameters such as prediction of binding to HLA-A*02:01 (NetMHCpan≤0.5), the number of identical amino acids (with or without anchoring positions), similarity to the target peptide (based on PMBEC-score), display of a unique similarity motif and number of detections on healthy tissues were considered.

Cytotoxicity assay

Cytotoxic response of MAGEA1 TCR-positive transduced and non-transduced (NT) T cells was measured against red fluorescent protein (RFP)-labeled U2OS and UACC-257 MAGEA1-positive (HLA-A*02-positive) tumor cells using IncuCyte live imaging. The assay was performed at various effector-to-target (E:T) ratios and fold-tumor growth monitored for at least 72 hours based on RFP fluorescence signal. Results are presented as mean±SD of three replicates at each imaging time point.

TCR affinity determination

The MAGEA1-specific TCR was expressed as soluble protein and refolded according to a published protocol in Escherichia coli.25 Refolded TCRs were purified via anion exchange and size exclusion chromatography. The protein concentration was determined using Bradford assays and refolding was determined via native and denaturing polyacrylamide gel electrophoresis. Biolayer interferometry technology was used to determine the affinity of the refolded TCR toward target pHLA compared with unrelated pHLA.

Quantification of IMA202 T cells

Genomic DNA was extracted from PBMC and/or tissue biopsies using QIAamp DNA Mini Kit and AllPrep DNA/RNA Mini Kit (both Qiagen), respectively, according to manufacturer’s instruction. The quantity of IMA202 T cells was assessed in DNA-samples using qPCR specific for Psi sequence of the lentiviral construct. The limit of detection for the assay is four copies/μg genomic DNA. The number of analyzed samples varied according to sample availability.

Serum cytokine analysis

Concentration of serum cytokines was measured using ProcartaPlex 34 Plex immunoassays (Invitrogen) at baseline; the day of T cell infusion (day 0) and days 1, 3, 7, 14, and 28 postinfusion. Cytokine signals were detected using the Luminex xMap technology in a microplate format on a Luminex 200 system. Data were acquired using the xPonent software. Raw data were then imported to the ProcartaPlex Analysis App (ThermoFisher Scientific) for data analysis. Cytokine analysis was based on standard curves generated by regression model of four or five parameter logistic (4PL/5PL standard curve fitting in the ProcartaPlex Analysis App). Coefficient of variation (%) of replicates was set at ≤20% for both standards and unknown analytes. Geometric mean of the regulated cytokine levels was plotted in log-scale against time. Graphs were generated using GraphPad Prism software. The number of analyzed samples varied according to sample availability postinfusion and meeting the sensitivity criteria of the assay.

Phenotypical T cell analysis

For flow cytometry-based ex vivo immunomonitoring, isolated and cryopreserved PBMC collected at different time points before and after infusion were subjected to pHLA multimer (tetramer) and cell surface staining. PBMCs were rested overnight in RPMI 1640+HEPES+10% human serum+1 ng/mL IL-15 and 20 U/mL IL-2. Potential aggregates were removed by centrifugation. Between 5×105 and 5×106 cells were treated with fixable Viability Stain BV510 (Becton Dickinson), followed by multimer staining in PE and/or PE-Cy7 for 20 min at room temperature (each multimer at a concentration of 0.8 µg/mL) and surface staining in two separate panels using antibodies listed in online supplemental table 2. All washing steps were carried out in phosphate-buffered saline (PBS), 2% fetal calf serum (FCS), 2 mM EDTA, and 0.01% azide. Stained cells, fixed using PBS with 1% FCS and 1% formaldehyde, were acquired on an LSRII SORP flow cytometer and analyzed using FlowJo software, V.10.4 (Tree Star).

The frequency of MAGEA1-specific T cells (tetramer-positive) was assessed in the product and postinfusion samples. The frequencies of naïve, central memory (CM), effector memory (TEM) and terminally differentiated effector memory (TEMRA) cells as well as frequencies of CD27−, CD28−, CD62L−, programmed cell death protein (PD)-1−, CD45RO−, CD57−, T cell immunoglobulin and mucin domain (TIM)-3− and lymphocyte activation gene (LAG)-3-positive cells were analyzed on MAGEA1-specific T cells (tetramer-positive) and non-specific cell (tetramer-negative). Memory T cell subsets were classified using the markers CD197 (C-C chemokine receptor (CCR)7) and CD45RA, with naïve being CCR7+CD45RA+, CM being CCR7+CD45RA−, TEM being CCR7−CD45RA− and TEMRA being CCR7−CD45RA+. The number of analyzed samples per patient varied according to sample availability.

Multiplex immunofluorescence staining and image analysis

Multiplex immunofluorescence staining was performed using similar methods that have been previously described and optimized.26 Briefly, 4 µm thick formalin-fixed, paraffin-embedded sample sections were stained with H&E and an anti-CD8 antibody (clone C8/144B, catalog#MS-457-S from Thermo Fisher Scientific). The slides were scanned using the Vectra/Polaris V.3.0.3 (Akoya Biosciences) at ×10 magnification (1.0 µm/pixel) through the full emission spectrum and using positive tonsil controls from the run staining to calibrate the spectral image scanner protocol.27 A pathologist selected five regions of interest (ROIs) for scanning in high magnification using the Phenochart Software image viewer 1.0.12 (931×698 µm size at resolution ×20) in order to capture various elements of tissue heterogeneity. Each ROI was analyzed by a pathologist using InForm V.2.4.8 image analysis software (Akoya Biosciences). Densities of CD8+ cytotoxic T cells were quantified and the final data was expressed as number of cells/mm2.27 The data were consolidated using the R studio V.3.5.3 (Phenopter V.0.2.2 packet, Akoya Biosciences). The number of analyzed samples varied according to sample availability and meeting the sensitivity criteria of the assay.

HLA peptide isolation and relative quantitation of pHLA

Primary human tissue samples were extracted surgically or postmortem from HLA-A*02-positive normal tissue donors. The resulting sample set covered 42 different organs. Tissue samples were snap-frozen in liquid nitrogen after excision and stored until isolation at −80°C for subsequent pHLA analyses.

After tissue homogenization and lysis, pHLA complexes were isolated by immunoprecipitation using BB7.2 (Department of Immunology, University of Tübingen, Germany) coupled to cyanogen bromide-activated sepharose resin (GE Healthcare Europe). Peptides were eluted from antibody resin by acid treatment and purified by ultrafiltration. HLA peptidomics was performed using an in-house analysis pipeline as previously described.22 Briefly, peptidome samples were separated by reversed-phase ultraperformance LC (UPLC) (nanoAcquity Waters) using ACQUITY UPLC BEH C18 columns (75 µm×250 mm, Waters) and a gradient ranging from 1% to 34.5% acetonitrile over the course of 70 or 190 min. MS was performed on online coupled Orbitrap mass spectrometers Fusion, Velos, and Linear trap quadrupole (Thermo Fisher Scientific) in data-dependent acquisition mode. Samples were analyzed in at least three replicate runs, acquiring MS/MS data in collision-induced dissociation and higher collisional energy dissociation mode. Data processing was performed using a proprietary pipeline, which combines database search, spectral clustering, feature detection, retention time alignment, and global normalization for the generation of population-scale, peptide presentation profiles.

ResultsTarget and T cell receptor characteristics

Between September 2019 and September 2022, we prescreened 242 HLA-A*02:01 positive patients with advanced solid tumors for MAGEA1 expression at four clinical sites in Germany and five sites in the USA using a qPCR assay on fresh tumor biopsies. Overall, the target antigen prevalence in HLA-A*02:01 positive patients was 28%. Tumors with the highest MAGEA1 prevalence were HCC (59%) and melanoma (36%), which is in line with expression prevalence in The Cancer Genome Atlas datasets (online supplemental figure 1).

Patient-individual MAGEA1-targeting T cells (IMA202) were generated on the basis of a highly specific TCR recognizing a nine amino acid peptide derived from MAGEA1.16 The TCR was selected as clinical candidate among >130 MAGEA1 TCRs derived from a TCR discovery campaign with different healthy human donors. Among the tested TCRs, the IMA202 TCR showed the highest specificity, functionality, and no signs of cross-reactivity. Re-expression of the MAGEA1 TCR via mRNA transfection rendered human CD8+ T cells strongly responsive to the MAGEA1 peptide loaded onto HLA-A*02-positive T2 cells, resulting in half-maximal IFN-γ release at a low MAGEA1 peptide concentration of 11 nM (figure 1A). Comparable results were obtained when the TCR was stably expressed in donor cells via lentiviral transduction (data are not shown). The TCR was CD8 coreceptor dependent and thus did not show binding to pHLA complex and downstream functionality in CD8− T cells (online supplemental figure 2). Specificity of MAGEA1 recognition was confirmed by testing the TCR against 10 peptides sharing high sequence similarity with the target peptide, that is, the similar peptides had five or six amino acids identical to the MAGEA1 target peptide. The similar peptides were chosen to cover the entire target peptide sequence, and the presence of those similar peptides on human normal tissue samples was verified by mass spectrometry (online supplemental figure 3), making them relevant for detection of cross-reactivity. As shown in figure 1B, the TCR only recognized the MAGEA1 target peptide, even though T2 cells were loaded with high similar peptide concentrations of 10 µM. During the course of TCR characterization, the IMA202 TCR was expressed in multiple HLA-A*02:01-positive healthy donor T cells and tested against MAGEA1 negative HLA-A*02:01 positive tumor cell lines without any signs of cross-reactivity or alloreactivity toward the second HLA allele of the respective donors (online supplemental table 3). Furthermore, lentiviral expression of the TCR in CD8+ T cells resulted in strong recognition of the MAGEA1-positive and HLA-A*02:01-positive tumor cell lines UACC-257 and U266B1 (online supplemental table 4) while no T cell activation was detected in coculture with HLA-A*02:01-positive human primary normal tissue cells, supporting the highly tumor-specific nature of the TCR (figure 1C). When analyzed in an Incucyte killing assay, lentiviral transduced CD8+ T cells completely eliminated HLA-A*02:01-positive U2OS and UACC-257 tumor cells expressing MAGEA1 at different levels while NT T cells failed to control tumor cell growth (figure 1D,E). To further investigate the therapeutic suitability of the MAGEA1 TCR, we generated a soluble TCR version for binding affinity analysis. The TCR exhibited a high binding affinity toward immobilized MAGEA1:HLA-A*02:01 complexes with a KD value of 8.7 µM (figure 1F).

Figure 1

Characterization of the MAGEA1-specific TCR. (A) Functional avidity measurement using IFN-γ release of IMA202 TCR-expressing CD8+ T cells on co-culture with peptide-loaded T2 cells. One representative of two donors using mRNA electroporation for TCR expression is shown. The mean EC50 value was 11 nM. MAGEA1 peptide concentrations for peptide loading ranged from 10 µM to 10 pM, and 2×104 T cells were used at an E:T ratio of 1:1. (B) Specificity characterization. The IMA202 TCR recognizes the target peptide MAGEA1 loaded on T2 cells, but no similar peptides or controls. CD8+ T cell activation after 20 hours of co-culture with peptide-loaded T2 cells was measured by IFN-γ ELISA. One of two independent donors is shown and 2×104 T cells were used at an E:T ratio of 1:1. The sequence and the target peptide motif similarity of each similar peptide is depicted in the table next to the graph. (C) Reactivity toward HLA-A*02 positive (A2+) human primary cells. T cells were lentivirally transduced with the IMA202 TCR and co-cultured with different primary human normal MAGEA1-negative cells and the MAGEA1-positive tumor cell lines UACC-257 and U266B1 at an E:T ratio of 3:1, with 6×104 effector cells. The mean of IFN-γ release after 20 hours from replicates is shown. Error bars indicate SDs. One representative donor of two is depicted. (D) Functional assessment. T cells were lentivirally transduced with the IMA202 TCR and co-cultured with 5×103 MAGEA1+ fluorescently labeled U2OS cell lines at indicated E:T ratios normalized to TCR positivity for 86 hours. One representative donor of two transductions is shown. (E) Cytotoxic potential of transduced IMA202 T cells demonstrated against another MAGEA1-positive cell line UACC-257 across five healthy donors in an InCucyte assay at E:T of 3:1. NT represent non-transduced T cells from 4/5 donors (F) Biolayer interferometry analysis of binding affinity of the MAGEA1-specific TCR used in the IMA202 trial revealed a KD value of 8.7 µM. E:T, effector-to-target; HCASMC, Human Coronary Artery Smooth Muscle Cells; HCM, Human Cardiomyocytes; HCMEC, Human Cardiac Microvascular Endothelial Cells; HREpC, Human Renal Epithelial Cells; HTSMC, Human Tracheal Smooth Muscle Cells; iCell HA, induced pluripotent stem cell-derived human astrocytes; iCell HH, induced pluripotent stem cell-derived human hepatocytes; NHDF, Normal Human Dermal Fibroblasts.

Patient characteristics

We limited eligibility to tumors with a MAGEA1 mRNA expression level above a defined threshold (online supplemental figure 4) ensuring a reasonable likelihood for HLA-presentation of the target peptide.22 A total of 46 (28%) patients tested for MAGEA1 expression fulfilled this criterion, of whom 25 underwent leukapheresis. Of these, 16 (64%) were infused with IMA202 in the dose escalation part of this trial. Due to death or withdrawn consent, none of the treated patients finished the protocol-specified follow-up of 2 years. Detailed information is depicted in figure 2.

Figure 2

CONSORT diagram and number of infused T cells. Study consort diagram (left). Figure depicting dose level based on TCR-T cell dose (cells/m²) for all 16 treated patients (right). DL1 is defined as 50×106 transduced cells/m² BSA±20% (range 40 to 60 x 106 cells/m²), DL2 as 300×106 transduced cells/m² BSA±20% (range 240–360×106 cells/m²), and DL3 as 1000×106 transduced cells/m² BSA±20% (range 800–1200×106 cells/m²). Enrichment cohorts represent intermediate dose levels between DL1 and DL2 as well as DL2 and DL3, respectively. BSA, body surface area; CONSORT, Consolidated Standards of Reporting Trials; I/E, inclusion/exclusion.

The characteristics of the 16 patients treated with IMA202 are outlined in table 1.

Table 1

Baseline characteristics and DLs

Seven patients had melanoma, two squamous cell carcinoma (SCC) of the anus, two HCC, two non-small cell lung cancer (NSCLC) adenocarcinoma and the remaining three patients had osteosarcoma, rhabdomyosarcoma, and oropharyngeal SCC, respectively. The median age of patients was 57 years (range 20–72 years) and 12 out of 16 (75%) patients had an ECOG performance status of 1. Patients had received a median of 5 prior lines of systemic therapies (range 1–7), including chemotherapy (n=11), radiotherapy (n=10), immunotherapy (n=14) and targeted therapy (n=10). All patients had tumors relapsed or refractory to all standard treatments. The median serum lactate dehydrogenase (LDH) level was 1.15× upper limit of normal (ULN) (range 0.6–2.6×ULN), with 6 (37.5%) patients having an LDH level≤1×ULN and 10 (62.5%) patients >1×ULN. The median serum albumin levels were 3.6 g/dL (range 2.7–4.8 g/dL) and the median sum of longest diameters of target lesions was 90.55 mm (range 26.4–272.0 mm) in pretreatment radiology evaluations. One patient had pre-existing, stable brain metastases.

Overall, we included heavily pretreated patients with various solid tumors and poor prognostic characteristics.

Product characteristics

Manufacturing of IMA202 was successful in 24/25 (96%) of patients that underwent leukapheresis and took a median of 8 days (range 7–10 days), which corresponds to a median of 4.6 population doublings. The median time from manufacturing to product release was 29.5 days. Patients received a median of 1.415×109 viable MAGEA1-specific T cells (range 0.086×109–2.57×109) (figure 2).

The median CD8/CD4 ratio of the final product for infused patients was 0.71 (range 0.27–1.73) (online supplemental figure 5A and table 5). The frozen product contained a median of 42% (range 12%–73%) MAGEA1-specific CD3+CD8+ T cells (online supplemental figure 5B and table 5) as the active ingredient. T cell memory characterization demonstrated that the MAGEA1-specific CD8+fraction was predominantly of TEM phenotype (frequency among MAGEA1-specific CD8+T cells: median, 82%; range 19%–95%) but also contained naïve (frequency among MAGEA1-specific CD8+T cells: median, 8.4%; range 0%–54%), CM (frequency among MAGEA1-specific CD8+T cells: median, 7.5%; range 3%–22%) and TEMRA cells (frequency among MAGEA1-specific CD8+T cells: median, 1.9%; range 0%–19%) (online supplemental figures 6A and 7A). Consistent with the memory phenotype distribution, MAGEA1-specific CD8+T cells widely expressed CD45RO (frequency among MAGEA1-specific CD8+T cells: median, 89%; range 28%–99%) while CD57 expression was observed in a minor subset (frequency among MAGEA1-specific CD8+T cells: median, 12%; range 1%–33%). Frozen T cells expressed molecules required for homing and costimulation such as CD62L (frequency among MAGEA1-specific CD8+ T cells: median, 57%; range 15%–92%), CD28 (frequency among MAGEA1-specific CD8+T cells: median, 66%; range 54%–92%), and CD27 (frequency among MAGEA1-specific CD8+T cells: median, 37%; range 11%–92%) (online supplemental figures 6B and 7B). MAGEA1-specific CD8+T cells also predominantly expressed TIM-3 on activation during manufacturing (frequency among MAGEA1-specific CD8+T cells: median, 84%; range, 50%–93%), however, median PD-1 and LAG-3 expression ratio remained less than 15% in the final product (online supplemental figure 7B).

Safety

Patients received doses between 0.086×109 and 2.57×109 transduced CD8+ T cells. No DLT was observed in patients who received IMA202 up to DL3 (ie, 1.60×109 to 2.57×109 transduced cells). TEAEs were as expected for adoptive T cell therapy (table 2).

Table 2

Treatment-emergent AEs

All 16 treated patients experienced at least one AE. 15 (93.8%) patients had AEs of ≥grade 3, and 1 patient had a grade 5 AE (dyspnea, unrelated to treatment). The most common grade 3–4 AEs reported were cytopenias (neutropenia (13/16), lymphopenia (12/16), anemia (8/16), thrombocytopenia (8/16) and leukopenia (4/16) related to lymphodepletion). Other common TEAEs (≥20% of subjects) were nausea (43.8%), diarrhea (31.3%), fatigue (31.3%), fever (31.3%), rash (31.3%), and chills (25%). Among those common TEAEs, only one patient experienced grade 3 fatigue and one grade 3 rash, respectively. 13 patients experienced CRS, including 1 patient who experienced grade 3 CRS (fever, hypotension, hypoxia and increased alanine aminotransferase/aspartate transaminase). Onset and duration of CRS by grade are shown in figure 3. The median time from IMA202 infusion to onset of CRS was 1 day (range 0–6 days) and the median duration of CRS was 11 days (range 2–21 days). ICANS was observed in two patients (both grade 1). CRS was managed with tocilizumab in eight patients and one patient required systemic corticosteroids. IL-2 administration was interrupted or permanently discontinued in nine patients. No signs of potential “on-target, off-tumor” toxicity were noted.

Figure 3

Cytokine release syndrome. The figure shows CRS grading over time post-IMA202 Infusion. The x-axis indicates time from IMA202 infusion while patient numbers and DLs are given on the y-axis. Grades for CRS were determined according to CARTOX criteria17 and are color-coded according to actual severity. Enrichment cohorts were opened to obtain further safety data and provide products for patients in need. CRS, cytokine release syndrome; DL, dose level; EC, enrichment cohort.

In summary, IMA202 showed a manageable tolerability profile.

Clinical activity

The date of data cut-off for the primary analysis was January 26, 2024, which represents the date of the lock of the clinical trial database. All patients who received IMA202 treatment were evaluable for response assessment by RECIST V.1.120 and 15 patients were evaluable by irRECIST.21

Of the 16 IMA202-treated patients, 11 (68.8%) patients had stable disease (SD), and 5 (31.3%) had PD as their best overall response according to RECIST V.1.1 (figure 4A). Five patients had initial tumor shrinkage in the sum of their target lesions (figure 4A,B). One patient (#03) with SD at the day 42 assessment experienced further shrinkage in the sum of target lesions diameter by 35.4% in total at 3 months. At that point, the patient had, however, progressive non-target lesion in the lung and therefore was classified as PD (figure 4C). The median duration of disease stabilization in the 11 patients with SD was 11 weeks (range 6.1–25.7 weeks). Overall, the median PFS was 9.1 weeks (range 3.1–25.7 weeks) and the median OS was 25.3 weeks (range 7.0–47.0 weeks). Of the 15 (93.8%) patients who were evaluable for response assessment according to irRECIST, 12 (80.0%) had SD and 3 (20.0%) patients had PD as best overall response.

Figure 4

Clinical outcomes of treated patients. (A) Shown are best changes in sum of diameter of target lesions and best overall response based on RECIST 1.1 for all 16 patients. *Maximum change of target lesions and RECIST 1.1 best overall response at different time points. #Patient #09 had clinical progression at week 6. (B) The graph shows tumor responses postinfusion over time. #Patient #09 had clinical progression at week 6. (C) An exemplary CT scan of patient #03 at baseline as well as 6 weeks and twelve weeks after IMA202 infusion is depicted. BOR, best overall response; DL, dose level; EC, enrichment cohort; NSCLC, non-small cell lung cancer; PD, progressive disease; RECIST, Response Evaluation Criteria in Solid Tumors; SCC, squamous cell carcinoma; SD, stable disease.

In conclusion, IMA202 demonstrated signs of tumor shrinkage in this heavily pretreated, high risk cohort of patients.

Engraftment and persistence of IMA202 in peripheral blood

Rapid IMA202 T cell engraftment occurred in all 16 patients (median peak, day 3; range days 1–7). IMA202 levels in peripheral blood tended to slowly decline over time, but no patient showed loss of IMA202 T cells during the period of assessment. The median IMA202 persistence was 66 days and the longest was 300 days postinfusion (figure 5A). A median peak frequency of 20.56% MAGEA1-specific CD8+T cells within all CD8+T cells and 6.63% within all CD3+ T cells was observed postinfusion (online supplemental figure 8). There was a significant correlation between peak expansion of IMA202 cells in peripheral blood and the number of TCR-engineered T cells infused (r=0.5882, p=0.018, figure 5B; time point of peak expansion for each patient is listed in online supplemental table 6). No correlation was found though between peak IMA202 vector copies or infused dose with clinical responses (online supplemental figure 9).

Figure 5

Biological characterization of the IMA202 TCR-T product postinfusion. (A) IMA202 product persistence up to 300 days postinfusion was determined using qPCR assay. (B) Correlative data of peak expansion and total infused cells. (C) Bar graph showing IMA202 TCR-T cell infiltration into tumors in individual patients at day 42 postinfusion. (D) Correlative data of tumor infiltration and total infused cells. *Sample was collected on day 60 postinfusion. (E) Immunofluorescent staining images of CD8+ T cells in tumor preinfusion and postinfusion of IMA202 cells. In addition, the respective H&E staining is shown for each sample; blue stain.

Longitudinal phenotyping of peripheral blood T cells indicated that MAGEA1-specific CD8+T cells remained mostly of TEM phenotype (frequency among MAGEA1-specific CD8+T cells at week 1: median, 92%; range 63%–97%) with an increase in TEMRA cells at later time points (frequency among MAGEA1-specific CD8+T cells at week 8: median, 17%; range 3%–35%) (online supplemental figure 7A). Similar to TEM phenotype, MAGEA1-specific CD8+T cells expressed CD45RO widely which peaked at week 1 while CD57 showed a similar kinetics with TEMRA cells with an increase trend over time (