Single Institution Analysis of Lymphoma Treatment Related Post-CAR Myeloid Neoplasms

Background: Treatment related myeloid neoplasm (tMN) is a growing concern following chimeric antigen receptor (CAR) T-cell therapy. tMN represents the second most common cause of post-CAR non-relapse mortality after infection. Though rates of post-CAR tMN are similar to comparator therapies such as autologous hematopoietic stem cell transplant (HCT), the clinical characteristics and molecular origin of these malignancies are currently poorly described. Results: We evaluated 533 patients treated for non-Hodgkin lymphoma (NHL) at Stanford. At a median follow up of 552 days we identified 16 patients with tMN including 14 with myelodysplastic syndrome (MDS) and 2 with acute myeloid leukemia (AML). In total 5/16 (31.3%) had prior HCT and 5/16 (31.3%) were treated with multiple CAR T-cell therapies. In total 10/16 (62.5%) of tMN patients had multiple cellular therapies (HCT or CAR) prior to transformation, however, in this analysis, multiple cellular therapies did not significantly increase the risk for tMN (p = 0.17). To compare the clinical features of post-HCT tMN we queried 1007 NHL patients treated with HCT (median follow up 2198 days) finding n = 29 post-HCT patients who did not undergo CAR T-cell therapy and developed tMN (24 MDS and 5 AML). The post-HCT tMN population had a greater number of patients treated for T-cell lymphoma (7 versus 0) and fewer pre-tMN lines of therapy (3.0 versus 4.5). There was no difference in the Revised International Prognostic Scoring System (IPSSR) score of patients with MDS, the rate of complex karyotype, or the rate of chromosomes 7 or 5 deletion between groups. Additionally, there was no significant difference in the rates of event-free survival (EFS) or overall survival (OS) after tMN diagnosis (EFS 37.4% vs 49.4% and OS 53.1% vs 52.4% in the HCT vs CAR groups respectively). To define the molecular origin of post-CAR tMN we analyzed genomic DNA from n = 99 pre-infusion CD19-CAR samples and found that pre-existing clonal hematopoiesis (CH) mutations are common occurring in 43.4% of patients at AF ≥ 2% and 75.7% of patients at any VAF. TP53 mutations existed in 11.1% and 23.2% of patients in pre-infusion samples at the same thresholds. TP53 gene mutation was present in 8/12 evaluable patients with tMN and was significantly associated with development of tMN among thirty patients assessed for tMN due to post-CAR cytopenia (odds ratio 1.85, 95% CI 1.38 - 2.48). For patients developing tMN, median pre-infusion TP53 mutant VAF was 0.36% relative to 32% post-infusion (n=8, 88-fold median increase). Single cell DNA sequencing (scDNA-seq) performed on 12 post-CAR samples with CH, including two tMN cases, showed that 22/29 (74.8%) unique CH containing clones contained a single mutation and most CH mutations existed within the myeloid lineage. Of all 29 clones 7/15 DNMT3A containing clones demonstrated a plurality of mutated cells in the lymphoid lineage. CAR integration into DNMT3A+ CH clones was observed in three patients, none of whom developed tMN. TP53 mutant burden was primarily in the myeloid lineage. In one case, we captured the transition from early MDS to aggressive disease, marked by loss of heterozygosity at the alternate TP53 allele and formation of a novel CD34+ population. Conclusions: Post-CAR tMN is rare occurring in 16/533 (3%) patients in our analysis. The majority of post-CAR tMN is TP53 mutated and occurred after tandem cellular therapies. TP53-derived post-CAR tMN often forms from low VAF TP53 mutations present in a large fraction of patients prior to CAR infusion. Using scDNA sequencing data we identified myeloid predominance for TP53 clonal hematopoiesis post-infusion and found rare integration of CAR19 vector into DNMT3A+ clonal hematopoiesis. As compared to post-HCT outcomes, we did not find differences in molecular risk, EFS, or OS between post-CAR tMN and post-HCT tMN indicating similar disease phenotype and clinical course.