Abstract
Polybromo-1 (PBRM1) loss of function mutations are present in a fraction of biliary tract cancers (BTCs). PBRM1, a subunit of the PBAF chromatin-remodeling complex, is involved in DNA damage repair. Herein, we aimed to decipher the molecular landscape of PBRM1 mutated (mut) BTCs and to define potential translational aspects. Totally, 1848 BTC samples were analyzed using next-generation DNA-sequencing and immunohistochemistry (Caris Life Sciences, Phoenix, AZ). siRNA-mediated knockdown of PBRM1 was performed in the BTC cell line EGI1 to assess the therapeutic vulnerabilities of ATR and PARP inhibitors in vitro. PBRM1 mutations were identified in 8.1% (n = 150) of BTCs and were more prevalent in intrahepatic BTCs (9.9%) compared to gallbladder cancers (6.0%) or extrahepatic BTCs (4.5%). Higher rates of co-mutations in chromatin-remodeling genes (e.g., ARID1A 31% vs. 16%) and DNA damage repair genes (e.g., ATRX 4.4% vs. 0.3%) were detected in PBRM1-mutated (mut) vs. PBRM1-wildtype (wt) BTCs. No difference in real-world overall survival was observed between PBRM1-mut and PBRM1-wt patients (HR 1.043, 95% CI 0.821–1.325, p = 0.731). In vitro, experiments suggested that PARP ± ATR inhibitors induce synthetic lethality in the PBRM1 knockdown BTC model. Our findings served as the scientific rationale for PARP inhibition in a heavily pretreated PBRM1-mut BTC patient, which induced disease control. This study represents the largest and most extensive molecular profiling study of PBRM1-mut BTCs, which in vitro sensitizes to DNA damage repair inhibiting compounds. Our findings might serve as a rationale for future testing of PARP/ATR inhibitors in PBRM1-mut BTCs.
Introduction
The prognosis of advanced/metastatic biliary tract cancers (BTCs) remains poor1, despite approval of novel therapies targeting FGFR2 gene fusions and IDH mutations2,3. Unfortunately, only a minor fraction of BTC patients are eligible for these specific treatments. In recent years the chromatin remodeling system has gained increased attention as a potential therapeutic target in several tumor types. Chromatin remodeling is a tightly regulated process controlling gene accessibility for transcription and thus regulating gene expression4,5. The Switch/Sucrose Non-Fermentable (SWI/SNF) chromatin remodeling complex represents a key component of chromatin remodeling6. Dysregulation of SWI/SNF due to loss of function mutations in genes—such as ARID1A, SMARCB1, and Polybromo-1 (PBRM1; also known as BAF180)—is present in up to 20% of malignancies7, but, to date, no targeted treatment related to the chromatin remodeling process has been approved.
In BTCs, PBRM1 mutations are present in 5–21% of cases8,9. In addition to chromatin remodeling, PBRM1 further participates in the repair of DNA double-strand breaks (DSB) via ATM phosphorylation10. Most genetic alterations of PBRM1 induce loss of function, consequently impairing DNA damage repair. Thus, PBRM1-mutated (mut) BTCs might be sensitive to agents targeting DNA damage repair. Indeed, the synthetic lethal effect of poly-(ADP-Ribose) polymerase (PARP) inhibitors (PARPi) has been suggested in an in vitro renal cell carcinoma model harboring PBRM1 mutations11. PARPi induces synthetic lethality in cells lacking the ability to repair DSBs, which was first demonstrated in Breast Cancer Gene 1 or 2 (BRCA1 or BRCA2)-mut tumors. Thus, PARPi has been approved for the treatment of BRCA-mut breast12,13,14, ovarian15,16,17, pancreatic18, and prostate cancer19. PARPi are currently under clinical investigation in several malignancies harboring mutations in genes involved in DNA damage repairs, such as RAD5120,21, PALB222, or ARID1A23,24. Moreover, combinations of PARPi with inhibitors of the DNA-damage repair protein ATR (Ataxia Telangiectasia and Rad3 related) are currently under investigation to overcome PARPi resistance25.
To the best of our knowledge, the genomic context of PBRM1 mutations in BTCs has not been investigated so far. Therefore, we aimed to characterize the molecular landscape of PBRM1-mut BTCs and investigate the potential therapeutic role of PARP/ATRi in PBRM1-mut BTCs.
Results
Patient characteristics
In total, 1848 BTC samples were centrally analyzed. Of these, 67.5% (n = 1249) were derived from primary tumor sites and 32.5% from metastatic lesions (n = 593). Altogether, 232 samples harbored different variants within the PBRM1 gene and 150 (8.1%) of these alterations were classified as pathogenic (frameshift: 51.4%; nonsense: 32.0%; splicing: 16.6%). The most frequently detected PBRM1 point mutation was the p.N258fs variant (n = 10, 6.6%; c.773delA (n = 3) and c.773dupA (n = 7)) followed by the p.R850* variant (n = 4, 2.6%; all c.2548 C > T; Fig. 1a). All pathogenic/likely pathogenic variants identified in the cohort are provided in Supplementary Table 1.