Identification of a CAR-Derived Clone by NGS-Based MRD After Fully Human BCMA CAR T-Cell Therapy in Multiple Myeloma

B cell maturation antigen (BCMA) chimeric antigen receptor (CAR)-T cell therapy has significantly improved survival outcomes in patients with relapsed or refractory multiple myeloma (RRMM), achieving unprecedented depth of response compared to conventional salvage regimens [1-3]. Minimal residual disease (MRD) negativity has emerged as a strong predictor of favorable prognosis in this setting [4] and is increasingly recognized as a surrogate endpoint for accelerated drug approvals of novel agents.

Next-generation flow cytometry (NGF) and next-generation sequencing (NGS)-based MRD assessments are the most widely adopted approaches, offering sensitivities of approximately 10⁻⁵ and 10⁻⁶, respectively. In addition to their high sensitivity, NGS-MRD enables longitudinal tracking of clonal dynamics and the identification of potential subclonal evolution. However, the interpretation of NGS-MRD in the context of genetically engineered T cell products remains limited and poorly characterized.

In our center, most patients attained NGS-MRD negativity following BCMA CAR-T infusion in real-world clinical settings. Unexpectedly, in a subset of these patients, we identified a persistent novel clone sequence that was not related to the original tumor clone post-infusion. This novel clone gradually declined over time but remained detectable in serial MRD assessments. Whether this uncommon finding reflects clonal evolution, tumor relapse, secondary neoplasm, or a previously unrecognized signal derived from the CAR-T product itself remains to be determined. To investigate this phenomenon, we conducted a retrospective study utilizing a multi-modal approach to characterize the origin and clinical relevance of this novel clone.

We retrospectively analyzed 55 myeloma patients who received BCMA CAR-T cell therapy at our center and had at least one NGS-MRD assessment post-infusion, as part of real-world observational studies or investigator-initiated trials (IIT). Written informed consent was obtained from all patients, and the study was approved by the Ethics Committee of the Blood Diseases Hospital.

Fluorescence in situ hybridization (FISH) and NGF-based MRD were performed as previously described [5]. NGS-MRD analysis was conducted using the Neo-MRD assay (Neoimmune, Shenzhen, China), targeting V(D)J rearrangements in IGH, IGK, and IGL genes. A two-step PCR was applied: a 28-cycle multiplex PCR to amplify V and J segments, followed by a 12-cycle universal PCR to add Illumina adaptors. Libraries were sequenced on the NovaSeq 6000 platform (paired-end, 150 bp). CDR3 sequences with fewer than three reads per million were excluded. Dominant clones were defined based on > 3% clonal frequency and > 0.2% nucleated cell content and were tracked across timepoints.

Lentiviral vector copy number (VCN) was quantified using digital droplet PCR (ddPCR) in 27 bone marrow (BM) DNA samples collected at MRD timepoints. Single-cell RNA, BCR, and TCR sequencing were performed on bone marrow mononuclear cells from P22 at T4 (D28 post-infusion) to explore the clonal origin of the novel clone. All statistical analyses and graph generation were conducted using R (version 4.1.2).

A total of 55 patients were included in this study, with a median age of 63 years. Among them, 56.4% had RRMM, 63.6% were enrolled in IITs, and 49.1% received equecabtagene autoleucel (eque-cel), a fully human-derived BCMA CAR-T product. The remaining patients received academic BCMA CAR-T constructs. Detailed clinical data are summarized in Table S1.

Of these, 25 patients (45.5%) exhibited the emergence of a persistent novel clone following CAR-T infusion. Comparable to patients without a novel clone (n = 30), the novel clone cohort had similar disease features and treatment-related adverse events, including cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS). Notably, the only distinguishing demographic factor was a significantly younger median age in the novel clone group (60 vs. 66, p = 0.009). But the distribution of CAR-T products between the two cohorts differed entirely. All 25 patients with novel clone emergence received eque-cel, whereas only two eque-cel recipients did not exhibit this phenomenon. This pattern strongly suggests that the emergence of the novel clone is specifically associated with the eque-cel product.

Clonal dynamics were assessed in 20 patients who had at least two serial NGS-MRD assessments (Figure 1). All patients showed complete clearance of baseline tumor clone following CAR-T infusion (Table S2). In most cases, the novel clone appeared concurrently in the first post-infusion bone marrow sample. Its abundance ranged from 10⁻⁵ to 10⁻³, except for P22, who showed 42.76% at Day 28 post-infusion. The longest persistence observed was 479 days until last follow-up.

FIGURE 1

Open in figure viewerPowerPoint

Clonal dynamics over time as assessed by NGS-MRD: Clearance of the baseline myeloma clone and emergence of the novel clone are shown across serial bone marrow timepoints. Eque-cel, equecabtagene autoleucel, a fully human-derived BCMA CAR-T product; MRD, minimal residual disease; NGS, next-generation sequencing; T, timepoint.

To investigate the molecular origin of the novel clone and its relationship to the CAR-T product, we reanalyzed NGS-MRD raw data. Remarkably, all patients exhibiting the novel clone shared an identical nucleotide sequence: GGTTCAGTGGAAGTGGATCTGGGACAGATTTCACTCTCACCATCAGCAGTCTGCAACCTGAAGATTTTGCAACTTACTACTGTCAGCAAAAATACGACCTCCTCACTTTTGGCGGAGGGACCAAGGTGGAGATCAAACGTAAG. Sequence alignment confirmed overlap with the single-chain variable fragment (scFv) of the eque-cel CAR construct, specifically mapping to the FR3–CDR3–FR4 region within the variable light (VL) domain (Figure S1). This alignment strongly supports the hypothesis that the novel clone represents a fragment of the human-derived CAR transgene.

Interestingly, in P10, both the original tumor clone and the novel clone were detected simultaneously at 1 year post-infusion, followed by BCMA-negative relapse (Figure 1). A similar observation was made in P23, who exhibited sustained novel clone detection concurrent with BCMA-negative extramedullary relapse (Figure 1). These findings suggest that the presence of the novel clone does not indicate active disease. Instead, antigen escape, rather than the loss of CAR-T persistence, may underlie relapse in these cases, further reinforcing the non-tumor nature of the novel clone signal [6].

CAR-T cell persistence was evaluated by qualifying eque-cel VCN using ddPCR in 27 available BM samples at MRD timepoints. One sample lacked no detectable CAR signal; the remaining 26 showed a median VCN of 0.91% (range, 0.01%–54.52%). Importantly, VCN levels strongly correlated with the abundance of the novel clone detected by NGS-MRD (R = 0.88, p < 0.001; Figure 2A), supporting a direct link between this novel clone and the CAR transgene.

FIGURE 2

Open in figure viewerPowerPoint

Molecular and single-cell characterization of the novel clone: (A) Correlation between novel clone abundance by NGS-MRD and vector copy number by droplet digital PCR (ddPCR); (B) workflow illustrating sample processing and sequencing strategy for single-cell RNA (scRNA), T-cell receptor (TCR), and B-cell receptor (BCR) analysis; (C) uniform Manifold Approximation and Projection (UMAP) plot of 10 179 high-quality cells showing 11 major cell clusters; (D) heatmap of marker gene expression across the 11 clusters; (E) UMAP plot showing distribution of CAR-positive cells and comparison of T cell subsets between CAR+ and non-CAR+ T cells; (F) UMAP plot for mapping the novel clone sequence against the single-cell TCR/BCR dataset; (G) bar plot of TCR clonotype group proportions in CAR+ versus non-CAR+ T cells. BMMCs, bone marrow mononuclear cells; CAR, chimeric antigen receptor; MRD, minimal residual disease; NGS, next-generation sequencing.

We further explore the cellular basis of the novel clone through single-cell RNA, BCR, and TCR sequencing on bone marrow mononuclear cells from P22 at the T4 timepoint (Figure 2B). After quality control, 10 179 high-quality cells were classified into 11 major cell types according to respective marker genes (Figure 2C,D). No plasma cells and only 354 B cells were identified. Among these, 989 cells were identified as CAR-positive based on alignment with the eque-cel transgene. These CAR+ T cells were broadly distributed among T cell subsets, with a modest enrichment of CD4+ T cells compared to non-CAR+ T populations (Figure 2E). Notably, the novel clone sequence did not match any BCR or TCR sequence in the single-cell dataset, excluding derivation from endogenous T or B cells (Figure 2F). Moreover, scTCR-seq analysis further revealed a polyclonal CAR+ T cell repertoire with no evidence of abnormal clonal dominance or skewing (Figure 2G). These results strongly support that the novel clone represents a persistent CAR transgene, not a malignant or reactive clonal expansion.

In this study, we report for the first time the unexpected and recurrent detection of a persistent novel clone by NGS-based MRD assays following infusion of fully human BCMA CAR-T product (eque-cel). Comprehensive sequence alignment and vector tracking analyses revealed that the novel clone originates from the scFv region of the eque-cel construct, rather than from residual malignant or endogenous lymphoid clones.

This observation carries two important implications. First, it reveals a novel analytical artifact in the interpretation of NGS-MRD in the context of humanized CAR-T products. Since NGS-MRD relies on identifying immunoglobulin gene rearrangements, particularly within the CDR3 region of IGH/IGK/IGL, the presence of a human-sequence-derived scFv segment within the CAR construct may yield false signals resembling clonal B-cell populations. This phenomenon was not observed in patients treated with nonhuman-derived BCMA CAR-T products such as ide-cel [1] or cilta-cel [2], likely due to the absence of homology between their CAR constructs and human immunoglobulin gene segments. No similar findings have been reported with CD19 CAR-T in B-cell malignancies. Our data underscore the need for heightened awareness of this phenomenon when interpreting MRD results after CAR T-cell therapy—particularly as humanized CAR designs become increasingly adopted. Second, the strong correlation between novel clone abundance and VCN suggests that NGS-MRD may incidentally serve as a highly sensitive surrogate for in vivo tracking of CAR-T cell persistence.

Importantly, single-cell multiomic profiling of representative patients confirmed that the novel clone did not correspond to endogenous T or B cell clonotypes, nor did it represent monoclonal expansion. CAR+ T cells demonstrated a polyclonal and phenotypically diverse profile, with no evidence of transformation or expansion bias. Together with the absence of BCR/TCR sequence overlap, these data argue strongly against a neoplastic or immune-dysregulated origin of the novel clone. While prolonged CAR+ cell persistence warrants long-term clinical monitoring, our findings do not suggest an association with secondary neoplasms.

Collectively, our findings demonstrate that the persistent novel clone observed post-eque-cel therapy represents an analytical artifact derived from the CAR transgene, not residual disease or secondary lymphoproliferative processes. Awareness of this signal is essential to prevent misinterpretation of disease status and to ensure accurate clinical decision-making in the post-CAR-T setting.