Background – Hotspot mutations in SF3B1 result in a neomorphic protein that causes aberrant splicing of hundreds of transcripts, impacting many cancer-associated pathways. The widespread changes make SF3B1 an attractive clinical marker and potential therapeutic target, but clinical and biological implications remain controversial. This study investigates the genetic profile, survival outcomes, and biological impacts of SF3B1 mutations in breast cancer.

Methods – We analyzed a de-identified clinical cohort from the Tempus Database consisting of 420 breast cancer patients (pts) with SF3B1 mutations who underwent somatic NGS testing via the Tempus xT assay. SF3B1 hotspot mutation frequencies were assessed. Real-world overall survival (rwOS) was defined as the time from biopsy to death from any cause. Hazard ratio (HR) was calculated using Cox proportional hazards (CoxPH) model, and p-values were calculated using the Wald test.

Results – In accordance with previous studies we observed an enrichment of SF3B1 mutations in HR+ HER2- disease and an enrichment of the K700E hotspot. While it is known that distinct mutations within a gene can result in oncogenicity and targetability differences, the impact of distinct SF3B1 hotspots is unknown. To address this, we’ve utilized an innovative genome editing approach to achieve high efficiency knock-in of mutations in isogenic cell lines. Strikingly, all SF3B1 hotspot mutations negatively impact cell growth in-vitro and tumor formation and growth in orthotopic xenograft models. To evaluate proliferative indices in our clinical cohort, MKI67 gene expression was compared between patients with SF3B1 hotspot mutations (N=182) and wild-type SF3B1 (N=10,460) by Kruskal-Wallis test, and found to be decreased (p<0.001) in pts with SF3B1 hotspot mutations. Further, we observed differential tolerability of introduced hotspot mutations both in terms of mutation incorporation and stability. SF3B1 mutant cell lines revert to wild type SF3B1 status through deletion of the mutation, with elimination of R625 mutations occurring more quickly than K700E. Concordantly, we observed a median mutant variant allele fraction of 0.2 (IQR 0.09, 0.3) in our patient cohort, indicating these mutations exist subclonally. Importantly, in HR+ HER2- pts, we observed significantly improved rwOS (HR=0.67, 95%CI[0.45, 1.00], p=0.035) in pts with SF3B1 hotspot mutations (N=82) compared to wild type (N=3,474) using a multivariate CoxPH model adjusted for race, sex, and age at diagnosis.

Conclusions – Improved rwOS and the deleterious biological impacts observed support the utility of SF3B1 as a therapeutic target and emphasize the need to understand the role of these mutations in cancer. Long read RNA sequencing and splice junction analyses are ongoing and will provide mechanistic insight on the selective enrichment of certain hotspots and be used to identify both pan-SF3B1 and hotspot-specific therapeutic targets.

VIEW THE PUBLICATION