Detection of intratumoral microorganisms in DLBCL impacts on CART-cell outcomes

Abstract Introduction: In lymphoma, a non-disrupted gut microbiome has been associated with clinical response to CD19 CAR T (Smith et al., Nat Med 2022). In diffuse large B-cell lymphoma (DLBCL), the tumor microenvironment (TME) is a critical component that contributes to the biology and potentially to patient outcomes. Intratumoral microorganisms (IMS) constitute a component of the TME in solid tumors and tend to localize in specific niches, interact with the TME, and correlate with response to therapy and outcomes. The role of IMS in clinical response to CAR T cell therapy for DLBCL patients is unknown. In this study, we hypothesized that IMS might predict response to CAR T cells. Methods: Next-generation sequencing (NGS) targeting exons frequently somatically mutated in human cancers was performed on pre-CART samples from 83 patients (pts) with DLBCL at our institution. The control group comprised 71 lymph node samples from pts with suspected hematologic disorders where the same NGS revealed no mutations, and 266 untreated DLBCL samples. Microbial reads from regions that do not align with the human reference genome were analyzed (Elkrief JCO 2024). Several databases were referenced to exclude contamination, and IMS were deemed positive if a minimal number of 2 reads were present. For potential contaminants, the threshold of minimal number of reads was higher for each IMS, based on the literature. These results were coupled with data on pre-CAR T characteristics, treatment, and outcomes from our clinical database. We compared variables with Fisher's exact test, the t-test, logistic regression for OR and multivariable Cox regression analysis for survival for each IMS present in more than 10% of the samples. Multiple test correction was performed with the false discovery rate (FDR). Results: For the 83 pts whose samples were analyzed, median age at CART was 68 (32 – 81) years, 51 pts were male (61%), 44 pts had performance status (KPS) <90 (53%), 36 pts had high pre-CAR LDH (43%), median lines pre-CAR were 3 (min 2 max 7), stage at apheresis was III/IV in 62 patients (75%). Forty-three cases(52%), were germinal center (GC) derived and 30 (36%) non-GC, not otherwise specified (NOS) in 10 (12%) high grade (HG)BCL in 15 (18%). The median number of IMS per DLBCL sample was 9 (min 1 – max 73) vs 3 (min 0 max 90) in the controls, p=.02. No significant difference in IMS per sample was found between untreated DLBCL and pre-CAR DLBCL. The number of IMS per sample was not associated with age, sex, histology, LDH, KPS, stage at apheresis, PFS, or OS post CART. Detection of Escherichia, Erythrobacter, Panotea, Mycobacterium, actinomyces, Neisseria, Haemophilus, Prevotella, Veillonella, Bacillus was more frequent in pre-CAR DLBCL compared to controls (all p values and FDR q <.05). Detection of specific IMSs was associated with achieving complete remission (CR) post CAR T and these IMSs were independent variables in the logistic regression analysis (including Kps, lines pre-CAR, COO, high grade histology, LDH, stage at apheresis): Neisseria (36/83, 43%, OR 0.33, p=.04, FDR q=.3), Rhodococcus (19/83, 23%, OR 0.14, p=.01, FDR q=.3), Lautropia (9/83, 11%, OR 0.1, p=.04, FDR q=.3), Prevotella (26/83, 31%, OR 0.34, p=.05, FDR q=.3). Detection of Campylobacter, Bordetella, Achromobacter, Lymphcryptovirus, Veillonella showed a trend towards detection in patients achieving CR. The detection ofNeisseria (Logrank p=.005, FDR q=.05) andMoraxella (LogRank p=.002, FDR q=.09) was associated with shorter OS post CART, and the detection of Neisseria (HR OS 2.00, 95% C.I. 1.1 – 3.6, p=.02, FDR q=.09), Moraxella (HR OS 3.89, 95% C.I. 1.41-11.1, p=.011, FDR q=.09), Rothia (HR OS=2.2, 95% C.I. 1.15 – 4.12, p=.01, FDR q=.09) were independent predictors in the Cox regression analysis compared to the other mentioned risk factors. Conclusions: We detected genomic DNA from several bacterial taxa in biopsies of DLBCL. These were enriched compared to controls, and potential associations with treatment responses and outcomes were observed. The presence of Neisseria seems to be lymphoma-specific, associated with poor response to CART, and poor OS after CART. Further studies in larger independent cohorts are needed to confirm these results and to evaluate the interactions of IMS with the lymphoma TME.