An immune desert phenotype underlies therapy resistance in LBCL through distinct microenvironment-related miRNA profiles

Abstract

Abstract Large B-cell lymphoma (LBCL) is an aggressive hematological malignancy characterized by significant clinical and genetic heterogeneity, resulting in variable prognosis and treatment response. It has become increasingly evident that the elements of the lymphoma tumor microenvironment (TME) impact on cancer progression and drug resistance. The TME features may be heterogeneous in LBCL and related to specific molecular subtypes, such as those characterized by the dark-zone (DZ) gene signature. MicroRNAs (miRNAs), small non-coding RNAs that regulate gene expression, play a pivotal and multifarious role in cancer biology by directly influencing tumor cell behavior, regulating immune evasion, and shaping TME. In this study, we performed a transcriptome analysis by total RNA sequencing on tumor samples from 39 de novo LBCL patients treated with R-CHOP first-line treatment, 20 refractory/relapsed (R/R) patients, and 19 therapy-responders. Using the Molecular Functional Portraits bioinformatic tool, we classified tumors according to gene expression profiles into prognostically relevant TME subtypes. Notably, R/R patients were enriched for the “immune-desert” (ID) subtype, which is associated with the poorest clinical outcomes. Given that the high-risk DZ-signature subtype has been previously characterized by an immunosuppressive TME, we examined the overlap between the gene expression profiles of our ID samples and the DZ signature. Remarkably, 84 out of 104 genes composing the DZ signature showed similar modulation in ID samples, underscoring a strong correlation between the DZ signature and the ID phenotype. To investigate the role of miRNAs in shaping TME subtypes, a small-RNA sequencing was performed on samples from the same patient cohort. We found 124 miRNAs differentially expressed between ID patients and those from other TME groups. We observed a significant positive correlation (Spearman's R = 0.35; p=0.00012) between miRNA expression changes linked to TME subtypes (ID type vs other TME types) and those associated with treatment response (R/R vs responding patients), highlighting the connection between miRNA-mediated TME remodeling and clinical outcomes. Of note, four of the identified TME-related miRNAs—miR-130b-3p (upregulated in ID and R/R patients), miR-100-5p, miR-99b-5p, and miR-125a-5p (all downregulated in ID and R/R patients)— were also able to accurately discriminate R/R from responsive patients with an area under the ROC curve >0.67. These miRNAs have previously been implicated in modulating the tumor immune microenvironment across various solid and hematologic malignancies. Moreover, an anticorrelated target and pathways analysis shows that genes targeted by the identified TME-related miRNA signature were enriched in pathways critical for TME regulation and tumor immune evasion, such as the Notch and IL6-JAK-STAT3 signaling pathways. Deconvolution analyses are ongoing to identify which TME components differ between tumors with high versus low expression of this four-miRNA signature. This will allow a better understanding of the molecular mechanisms by which these miRNAs influence the tumor microenvironment and treatment response. We will further validate all these data in a larger cohort of LBCL patients as well as in cellular models of resistance to R-CHOP. Altogether, our results represent a comprehensive, multi-layered approach to understanding how microRNAs contribute to drug resistance by reshaping the TME in LBCL and may pave the way for novel combined therapeutic strategies, holding significant promise for advancing precision medicine.