Background: Primary mediastinal large B-cell lymphoma (PMBCL) is an aggressive lymphoma affecting predominantly young female patients. Previous studies in this rare entity have focused on single genes or were limited in cohort size. Aims: To unravel the underlying genetic pathogenesis and its impact on outcome, we embarked on a comprehensive large-scale genetic investigation. Methods: Specimens of 486 previously untreated PMBCL patients were analyzed by paired tumor/normal whole-genome (WGS, n=14), whole-exome (WES, n=78) and targeted sequencing (TS, n=486). To understand the consequences of highly recurrent mutations in the chromatin-modifying gene ZNF217, we conducted functional and multi-omics analyses in CRISPR/Cas9 engineered cell lines. Results: WGS/WES revealed a complex genomic landscape in PMBCL with a median of 85 structural variants, a mutational burden of 5 mutations/Mb, 12 mutated coding candidate driver genes (CDG) and 4 focal somatic copy-number aberrations per sample (Fig.1a). Besides known targets, significant breakpoints were identified in genes previously not implicated in B-lymphomagenesis such as TOX and TP73 (36% and 21%). In addition, non-coding mutations clustered within the PAX5 enhancer region. With the identification of 50 recurrently mutated CDGs, we significantly expand the repertoire of known PMBCL drivers. The 10 most frequently mutated CDGs were SOCS1 (86%), B2M (67%), ITPKB (64%), ACTB (58%), STAT6 (58%), IGLL5 (56%), TNFAIP3 (53%), NFKBIE (49%), GNA13 (47%), and ZNF217 (36%), respectively. The operative mutational processes were attributed to aging, AID/APOBEC activity, defective MMR, and an unexpected infidelity of DNA-Polymerase ε. Next, we performed TS in 486 samples using a PMBCL-specific 106-gene panel. Recurrent lesions in 25 epigenetic modifiers were found in >90%, with ZNF217 being among the most frequently mutated genes (Fig.1b). After knockdown of ZNF217 in Karpas1106P and L428 cells, we demonstrated altered proliferation, migration, and apoptosis. Using mass spectrometry, we showed that ZNF217 is acting in a LSD1, CoREST and HDAC containing histone modifier complex. Accordingly, knockout of ZNF217 led to global changes in chromatin accessibility with an enrichment of differentially accessible motifs for crucial lymphoma-associated transcription factors, especially of the NF-κB, BATF/AP1, and IRF family, but also of CTCF, a major regulator of global 3D chromatin architecture. Resulting gene expression was characterized by changes in interferon-responsive genes and inflammation-associated transcription (Fig 1c). Clinical data were available for 329 cases, including 84 cases from clinical trials. Multivariate analysis using an IPI-corrected Cox regression model was performed. The estimated 5-year PFS and OS were 77% and 86%. Among the genetic lesions with the strongest association for poor outcome, we identified patients with mutatedCD58 having a significantly shorter survival (PFS: HR 2.96; p<.001; OS: HR 2.55; p=.006). In contrast, mutated DUSP2 indicated longer survival (PFS: HR 0.28; p=.002; OS: HR 0.15; p=.011) (Fig1d). Notably, DUSP2 mutated patients (25%) showed a similar outcome for CR rate, PFS and OS when comparing CHOP-like and intensified treatment regimens, suggesting no further benefit from treatment intensification in this very-low risk patient population. Image:Summary/Conclusion: Here, we present the genetic landscape of PMBCL highlighting a previously underappreciated role of chromatin modifying genes, identify novel treatment targets and provide a solid basis for guiding precision medicine approaches.