FOS/FOSB rearrangements; defining the human bone tumour, osteoblastoma


Session type:


Matthew Fittall1,Nischalan Pillay2,Anna-Christina Strobl3,Annelien Verfaillie1,Jonas Demeulemeester1,Maxime Tarabichi4,Matthew Young5,Agamemnon Grigoriadis6,Peter Van Loo4,Peter Campbell5,Adrienne Flanagan2,Sam Behjati7
1The Francis Crick Institute,2University College London,3Royal National Orthopaedic Hospital,4The Francis Crick Institure,5Wellcome Trust Sanger Institure,6King's College London,7Wellcome Trust Sanger Institute



Osteoblastoma, and the related entity osteoid osteoma, are the most common benign bone-forming tumours. Large, inaccessible, and recurrent tumours can cause considerable morbidity. On occasion, there can be diagnostic uncertainty with osteosarcoma, a malignant tumour that requires multi-modal therapy. We sought to define the somatic changes that underpin osteoblastoma.


We analysed the transcriptome and whole genomes of 5 osteoblastomas and 1 osteoid osteoma and catalogued all somatic variants. FOS fusions were validated with Sanger sequencing. We used FISH and Immunohistochemistry (IHC) to validate the finding of FOS/FOSB rearrangements in 55 osteoblastomas, 17 angiosarcomas and 183 osteosarcomas. We analysed 55 osteosarcoma and 2652 pan-cancer whole genomes for similar rearrangements.


Amidst a paucity of somatic alterations, analysis of structural variants revealed breakpoints in the AP-1 transcription factor FOS in 5/6 cases(Fig.1), and its paralogue FOSB in the sixth case(Fig.2). All were validated with RNA-seq reads and FOS fusions were validated with Sanger sequencing. 

FOSB fusion brought expression under the control of the PPP1R10 promotor. FOS fusions were all between exon 4 and intronic or intergenic regions. FOS fusions resulted in the introduction of a stop codon within 30bp of the breakpoint.

In a validation cohort of 55 tumours, FISH identified FOSB or FOS breakapart signals in 1 and 48 tumours respectively (89%). IHC for the preserved N-terminus of FOS revealed marked immunoreactivity in all FOS rearranged cases, including the 3/6 FOS FISH negative cases with available material. Only 1/183 osteosarcoma cases had comparable FOS immunoreactivity. No osteosarcoma or pan-cancer whole genome harboured similar rearrangements.


We suggest that FOS or FOSB rearrangements define osteoblastoma and osteoid osteoma. The truncated FOS transcript and protein bears a striking resemblance to the retroviral oncogenic form of v-fos, known to induce osteosarcoma in mice. This is the first report of FOS alteration in a bone-forming tumour.