CD19 Intron Retention Is Mechanism of CAR-T Treatment Resistance in Non-Hodgkin Lymphoma

Background: Chimeric antigen receptor T-cell (CAR-T) therapy has become a viable treatment option for patients (pts) with relapsed or refractory diffuse large B-cell lymphoma (R/R DLBCL). Transcriptomic markers for CD19 directed CAR-T failure have been characterized in pre-treatment B-cell acute lymphoblastic leukemia (B-ALL) but not in DLBCL. Here, we report the identification of RNA alternative splicing (AS) events associated with CD19 directed CAR-T response in pre-treatment DLBCL pts and of RNA binding proteins (RNABP) regulating these AS events.

Methods: To identify putative AS markers for CAR-T treatment outcome, we performed differential AS on bulk RNA-seq samples of pre-treatment tumor biopsies in pts that had a durable response (DR), defined via remission up to 9 months post-treatment, and pts that had a non-durable response (NDR) with rMATS-turbo (Shen et al. PNAS. 2014). Intron retention (IR) events within CD19 were quantified with a secondary, more robust algorithm to ensure reproducibility and accuracy. Survival analyses were done using the survival and survminer libraries in R. Validation assays for the downstream effects CD19 IR were conducted in EJ1, Jeko-1, OCI-Ly3, SU-DHL-6, and Toledo cell lines. The surface protein expression of CD19 was assayed via flow cytometry (FC) followed by Quantibrite PE (BD Biosciences, San Jose, CA) quantification. Total CD19 protein expression was measured via western blot (WB) in the aforementioned cell lines. Analyses of CD19 IR was done via RT-PCR followed by electrophoresis, and bands were quantified with iBright Analysis Software (Thermo Fisher Scientific, Carlsbad, CA). The effect of RNABP was assessed via the creation of CRISPR/Cas9 knockout (KO) cell lines followed by WB of the target gene and quantification of CD19 surface protein expression by FC.

Results: Introns 2 and 6 of CD19 are more highly expressed in NDR pts. A combined metric estimating the proportion of normal CD19 isoforms that do not express introns 2 or 6 (normal CD19) was created and DR pts expressed normal CD19 isoforms significantly higher than NDR pts (Figure 1). Normal CD19 isoforms classification performance was assessed via leave one out cross validation and achieved an overall accuracy of 67.6%. Sensitivity (classifying NDR) was 61.1% and specificity (classifying DR) was 81.3%. The expression of normal CD19 relative to the median level within the cohort was significantly associated with progress free survival (Figure 2). The expression of CD19 intron 6 was negatively correlated with CD19 surface protein expression in DLBCL cell lines. Cell lines were sorted by cells that exhibited high CD19 surface expression (bright) and low expression (dim). Cell lines sorted into the dim category expressed higher levels of intron 6 than bright cells. The correlation of RNABP expression with CD19 intron 6 expression was done in this patient cohort and the NCICCR-DLBCL cohort. Six RNABP were significantly correlated in both cohorts (r² > 0.2 and p-value <0.01). NOP2, one of the RNABP that was positively correlated with intron 6 expression, was knocked out in Jeko-1 cell lines to validate the putative correlation with CD19 protein expression. The resulting knock out cells exhibited lower levels of intron 6 retention and higher CD19 protein expression by FC and WB than non-transformed cells.

Conclusion: Pre-treatment tumors of R/R DLBCL patients who relapse before 9 months following CD19 directed CAR-T therapy have higher expression of CD19 isoforms that retain introns 2 and 6. The upregulated expression of the CD19 isoforms retaining these introns is correlated with CD19 surface protein expression and is a marker for sustained remission following CAR-T treatment. NOP2 is correlated with CD19 intron retention and surface protein expression and could be a target for modulating CD19 expression in conjunction with CAR-T treatment.

Figure 1: The expression of normal CD19 isoforms in pre-treatment tumor samples of DR (red) and NDR patients (blue). ΔPSI denotes the difference in medians between the two groups. P-value yielded by a Wilcoxon test between the two groups is reported.

Figure 2: Progress free survival (PFS) stratified by expression of normal CD19 isoforms. Red – above cohort median expression of normal CD19 isoforms. Blue – below cohort median expression of normal CD19 isoforms. P-value yielded via log-rank test.