Features of HLA‐A*02 in Identifying Eligible Patients for Tecelra TCR‐T Therapy

This study investigates the molecular mechanisms behind HLA-A*02 subtypes' differential responses to Tecelra, a TCR-engineered T-cell therapy targeting MAGE-A4. Using computational tools, the study identifies specific HLA-A*02 alleles, such as HLA-A*02:07, HLA-A*02:05, and others, with low affinity for the GV10 peptide or a risk of inducing alloreactivity. These findings provide insights into patient selection for Tecelra therapy, suggesting exclusion criteria for certain HLA-A*02 subtypes to optimize treatment efficacy and minimize adverse reactions. Tecelra (afamitresgene autoleucel) is an autologous T-cell therapy engineered with a lentiviral vector to express an affinity-enhanced TCR (ADP-A2M4) targeting the MAGE-A4230–239 peptide, GVYDGREHTV (GV10), presented by HLA-A*02 [1, 2]. Preclinical studies show that HLA-A*02:01/02/03/06 effectively presents GV10 and are recognized by ADP-A2M4, while HLA-A*02:05 leads to alloreactivity and ADP-A2M4 does not respond to HLA-A*02:07+MAGE-A4+ cell lines [3]. Recent studies have indicated that while HLA-A*02:01 is the most common HLA-A*02 allele in the Whites, HLA-A*02:02/03/06 increase HLA eligibility in other populations [4]. August 2024, the U.S. FDA approved Tecelra for treating synovial sarcoma in HLA-A*02:01/02/03/06+ patients with MAGE-A4+ [5]. This marks the first approval of a TCR-engineered T-cell therapy specifically for solid tumors. As the HLA-A*02 subtype encompasses the highest number of identified alleles, understanding the mechanisms behind HLA-A*02:05 inducing alloreactivity and HLA-A*02:07 failing to trigger a T-cell response is vital for informing patient selection in future clinical trials and therapies. Using NetMHCpan 4.1 and other prediction tools, GV10 binding to HLA-A*02:01 through HLA-A*02:50 was predicted (Material and Methods in Doc S1). The results indicated extremely low affinity for HLA-A*02:07 (Figure 1A, Table S1), aligning with previous findings that ADP-A2M4 does not effectively recognize MAGE-A4+ cell lines with HLA-A*02:07+ [3]. Phylogenetic analysis revealed that HLA-A*02:18 and HLA-A*02:33 are evolutionarily close to HLA-A*02:07 and exhibit similarly low GV10 binding affinity (Figure 1A). Sequences of HLA-A*02:01 were compared with HLA-A*02:07/18/33, revealing mutations at residue 99, where Y99 in HLA-A*02:01 is replaced by C99 or S99 in HLA-A*02:07/18/33 (Figure 1C and Figure S1A). Structural analysis revealed that the residue at position 99 primarily affects pocket D, which is critical for binding the position 3 residue Y (p3Y) of GV10 (Figure 1E and Figure S2) [6]. Additionally, several alleles, including HLA-A*02:04, also exhibit weak binding affinity for GV10 (Figure 1A). We validated our predictions with a refolding assay and differential scanning fluorimetry. HLA-A*02:07, predicted to have very weak affinity, failed to refold efficiently, while HLA-A*02:04, with weak affinity, showed a low Tm value of 49°C (Figure 1B). For other alleles, predicted affinity also closely correlated with their Tm values (Figure 1B). Our results suggest that HLA-A*02:04/07/18/33 alleles might be considered for exclusion in future treatments. Previous studies have shown that HLA-A*02:05 can cause alloreactivity to MAGE-A4−cell line [3], leading to the exclusion of HLA-A*02:05+ patients from clinical trials [1, 2]. Alleles with similar peptide-binding grooves (PBG) may have the same effect. Phylogenetic analysis identified HLA-A*02:14 as evolutionarily closest to HLA-A*02:05, followed by HLA-A*02:08 and HLA-A*02:02 (Figure 1A). Sequence analysis revealed that the key difference is Y9 in HLA-A*02:05/08/14, compared to F9 in HLA-A*02:02 (Figure 1D and Figure S1B). Further structural analysis suggests this difference at position 9 might lead to different peptide binding (Figure 1F and Figure S3A–D), which may explain why HLA-A*02:05 causes alloreactivity, whereas HLA-A*02:02 does not. Although HLA-A*02:06 also carries Y9, it does not induce alloreactivity. Comparing HLA-A*02:05 and HLA-A*02:06 sequences (HLA-A*02:05: R43, L95, W156; HLA-A*02:06: Q43, V95, L156), we found that the L95V substitution causes a rotational shift in Y116, and combined with W156L, alters the PBG, which may explain the lack of alloreactivity in HLA-A*02:06 (Figure S3B). We further analyzed the structure of the ADP-A2M2 recognizing GV10 presented by HLA-A*02:02/05/06. Compared to the non-alloreactive HLA-A*02:02/06, the HLA-A*02:05-GV10 complex exhibits structural differences upon recognition of ADP-A2M4 (Figure S3E,F). Taking together, these analyses suggest that HLA-A*02:14 and HLA-A*02:08 are highly similar to HLA-A*02:05, indicating a strong potential for alloreactivity, which may justify their exclusion from treatments and clinical trials. Although further experimental validation is needed, our analysis may help quickly identify patients who meet treatment criteria while excluding those at risk for adverse reactions, potentially improving the efficiency of patient selection for Tecelra treatment. Min Yang: data curation, formal analysis, investigation, writing – original draft. Peiluan Zhong: formal analysis, investigation, writing – review and editing. Huifang Jiao: software. Pengcheng Wei: conceptualization, funding acquisition, methodology, project administration, resources, validation, visualization, writing – review and editing. The authors have nothing to report. Approval of the research protocol by an Institutional Reviewer Board: N/A. Informed Consent: N/A. Registry and the Registration No. of the study/trial: N/A. Animal Studies: N/A. The authors declare no conflicts of interest. Figure S1. Sequence alignment of HLA-A*02. Figure S2. The electrostatic potential map of HLA-A*02. Figure S3. Structural feature analysis of HLA-A*02-GV10. Appendix S1. Materials and methods. Table S1. Prediction of GV10 binding affinity to HLA-A*02. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.