Abstract Many cancer therapies induce high response rates, with some resulting in “undetectable” disease as assessed using standard pathology techniques, yet many patients ultimately relapse due to persistent minimal residual disease (MRD) and die of their disease. This is particularly true in patients with leukemia and lymphoma, including those receiving CD19-targeted chimeric antigen receptor (CAR) T-cell therapy for mantle cell lymphoma (MCL). Long-term follow-up of brexucabtagene autoleucel (KTE-X19) CAR T-cell therapy in R/R MCL reported 68% initial complete remissions (CRs), but only 37% progression-free survival at 3 years. Furthermore, only 60% of initially MRD-negative patients, assessed by next-generation sequencing (reported sensitivity: 1 in 105 cells), remained in extended remission. This highlights a critical need to better understand the MRD that persists after therapy. The limited ability of current techniques for characterizing ultra-rare cells has impaired the development of informed strategies to eradicate MRD. Although circulating tumor DNA detection enables MRD detection, capturing ultra-rare MRD cells to study phenotypes and vulnerabilities is challenging. Fluorescence-activated cell sorting and sequencing (Sort-Seq) can reliably isolate MRD cells at a sensitivity of only 10-4. To interrogate ultra-rare MRD, we developed Live-cell Pick-Seq (LiP-Seq), an advanced platform leveraging multiplexed live-cell imaging to identify and retrieve single tumor cells, enabling the isolation of viable lymphoma cells at extremely low frequency for single-cell RNA-sequencing (scRNA-seq). First, we validated LiP-Seq's capability to isolate cells at frequencies below 10−6 and compared its scRNA-seq quality against Sort-Seq in MCL patient samples. Among transcripts detected by both methods, relative expression correlated well between picked and sorted cells. Moreover, 98% of picked cells contained reads covering = 5,000 genes (median 11,636 genes/cell), a higher fraction than obtained through sorting (88% covering = 5000 genes, median 8,726 genes/cell). 5,262 genes were detected only in LiP-Seq but not Sort-Seq; these had lower expression than genes detected by both platforms. These data indicate that LiP-Seq generates high-quality scRNA-seq data and enables detection of lower-expressed genes. We then applied LiP-Seq to a clinical MRD setting. We collected 44 paired peripheral blood samples at multiple timepoints from 11 MCL patients before and in CR after CD19 CAR T-cell therapy (brexucabtagene autoleucel). Cells from these patients were picked based on an MCL-enriching panel (CD45+CD79b+CD5+CD3-) for scRNA-seq. To accurately identify tumor cells, we reconstructed BCR repertoires and developed a method to classify cells based on tumor-specific single nucleotide variants (SNVs). ScRNA-seq analysis of post-treatment MRD cells, particularly those identified as tumor-derived by SNV analysis, revealed upregulation of Interferon-Induced Transmembrane Protein 2 (IFITM2) compared to pre-treatment tumor cells. This upregulation was observed across multiple patients and timepoints. The IFITM family mediates cellular responses to interferons, and we have previously shown that IFITM3 acts as a scaffold to amplify PI3 kinase growth signaling in malignant and non-malignant B cells. More broadly, expression of IFITM1, IFITM2, or IFITM3 are predictors of poor prognosis in several cancers. We therefore assessed IFITM2 levels using in vitro co-culture assays of MCL with CD19-targeted CAR T cells. Strikingly, rare Jeko-1 cells that remained alive following 4 days of exposure to CD19-targeted CAR T cells expressed IFITM2 at 3-fold higher levels than cells cocultured with untransduced T cells. Moreover, exogenous overexpression of IFITM2 in Jeko-1 cells was sufficient to confer relative protection from CD19 CAR T-cell-mediated killing. These data, in combination with LiP-Seq of primary MRD specimens, highlight the cell surface protein IFITM2 as a potential mediator of MRD survival and resistance to CD19 CAR T-cell therapy in MCL. In summary, LiP-Seq offers a novel platform capable of isolating and characterizing viable ultra-rare target cells, providing unprecedented insights into MRD biology and facilitating downstream biological assays. In the setting of cancer MRD, this approach may identify important features with clinical relevance that are not obvious from studying more abundant tumor cells at other timepoints or disease states.
Rice et al. (Mon,) studied this question.