The occurrence and mortality rates of cholangiocarcinoma have risen in recent years, particularly for intrahepatic cholangiocarcinoma[30–32]. Historically, surgery was the only viable treatment for intrahepatic cholangiocarcinoma, but this was not feasible for most advanced cases. Currently, many targeted drugs have gained regulatory approval for ICC patients, such as pemigatinib for patients with cholangiocarcinoma harboring FGFR2 fusion and tibsovo for patients with cholangiocarcinoma harboring IDH1 mutation. As FGFR2 fusion/translocation has been identified as a common event and is present at much higher frequencies in ICC [6–12], studies have emphasized on FGFR2. In our study, through FISH and/or NGS, we identified 30 cases harboring FGFR2 fusion/translocation. Clinicopathologically, patients harboring FGFR2 fusion/translocation did not show age/gender or tumor grade and tumor differentiation preponderance. Nevertheless, in other studies, researchers found that ICC patients with FGFR2 gene fusions were younger with a female preponderance [33]. Notably, we found that cases with FGFR2 gene fusion/translocation tended to have MF in gross classification and small ducts in histological classification, suggesting that the presence of FGFR2 gene fusion/translocation may have prognostic utility. Based on mucin productivity and immunophenotype, ICC was classified into large ducts and small ducts. Small duct-type ICC is generally characterized by little mucin production and exhibits immunoreactivity to N-cadherin and/or NCAM. More importantly, IDH mutation and FGFR2 translocation are restricted to small duct-type ICC, and small duct-type ICC often has prognostic utility [34, 35]. Most importantly, we observed that FGFR2 fusion/translocation tended to be enriched in special types of small duct ICC- cholangiolocarcinoma (CLC). These cases represent similar histological features: tumor cells show a small cuboid with a high nuclear/cytoplasmic ratio, oval nucleus, pale cytoplasm, lack of mucus, small atypia, and mostly in a well-differentiated state, with loose formation in the hyalinized collagen fibrous stroma characterized by angular small ducts, cords, or branching arrangements. Immunohistochemistry displayed MUC5ac negativity, MUC6 negativity or sporadic positivity, and CD56 was often positive. CLC arises in small intrahepatic ductules and only accounts for ~ 10% of ICC[36, 37], characterized by low-grade cytologic atypia, anastomosing cords and glands resembling cholangioles or canals of Hering[38]. In our study, 82.8% (24/29) of FGFR2 fusion/translocation ICCs displayed CLC with a similar immunotype. Because of the unique gross and morphological features of FGFR2 fusion cases, we described this distinct group as “FGFR2 fusion subtype ICC”.
Most notably, our genomic studies of ICC highlight the particularly important role of FGFR2 mutation, including site mutation and short in-frame deletion. We noted five types of mutations in a total of four patients, three site mutations (p. R203C, p. P253R and p. C382R), on in-frame deletion (p. N631_M640del) and one frame-shift (p.I548 Wfs*8), both in the TK domain of FGFR2. Different from FGFR2 fusion/translocation, half of tumors with FGFR2 mutation represented gross classification with MF and PI mixed, and all of these gross mixed cases showed large duct (LD) type in histological classification. Additionally, in contrast to FGFR2 fusion/translocation, FGFR2 mutation shows high stage predominance, not low stage predominance. The FGFR2 site mutations in extracellular domain are known oncogenic mutation that activates FGFR signaling[21] and such as FGFR2 F276C has been reported as target stie in ICC[15, 16, 20]. All site mutations we found were located in the FGFR2 extracellular space, but according to the pan-cancer study results, FGFR2 mutation sites distributed over all the exons. Although activating FGFR2 mutations are very rare in ICC, future tumor-agnostic studies with FGFR site mutations have the potential to widen the range of FGFR inhibitors.
It is worth mentioning that we also identified two TK domain FGFR2 deletions that have never been reported in other studies; one harbored an EX14 in-frame deletion (FGFR2 p. N631_M640del), while the other harbored an EX12 out-frame deletion (FGFR2 p. I548 Wfs*8), which results in premature termination of the protein at exon 12 during translation. FGFR2 extracellular domain in-frame deletions are known to cause autosomal dominant congenital craniosynostosis syndromes during growth [26–29] and cause oncogenicity during cancer development in ICC [15], whereas the function of FGFR2 TK domain deletions was not known until now. Data from a consortium of multicenter studies in multiple tumors also identified FGFR2 TK domain deletions, and we also identified FGFR2 deletions in multiple tumors. FGFR2 genetic short in-frame deletion occurred in patients with ICC at a higher frequency than in others and was followed by lung squamous carcinoma. FGFR2 deletion mutations are widely distributed, ranging from exon 4 to exon 17. Both in ICC and in lung squamous carcinoma, the deletion mutation was enriched in FGFR2 EX7, whereas in ICC, mutations predominated in L258-D304 del, and in lung squamous carcinoma, they predominated in R251-V274 del. Notably, in ICC, there was also a deletion in the FGFR2 TK domain, which proved that the TK domain deletion was not a random event in our cohort. In ICC, data from the consortium of multiple centers showed that p. K545del was located in the FGFR2 TK domain. Site mutations in the FGFR2 TK domain contribute to the molecular brake and induce clinical resistance to FGFR inhibitors; these site mutations include p. V564, p. E565, p. N594, p. L617, p. K641 and p. L659 [20, 39]. p. K545del and p. N631-M640del were very close to these polyclonal resistance mutation sites, and whether the deletion in the TK domain functioned as drug-resistant or drug-sensitive to FGFR inhibitors is unknown.
To assess the function of different FGFR2 mutants and determine their responsiveness to FGFR2-targeted drugs, we selected representative mutants (FGFR2 p.H167-N173del, p.I288-D304del, p.N631-M640del, p.C382R, p.K545del and p. I548 Wfs*8), constructed expression vectors of these mutants and introduced these mutants into NIH3T3 cells for assaying their oncogenic abilities. We found although all FGFR2 mutants we selected have the propensity to promote tumorigenesis and metastasis of ICC, not all of the mutants are capable of rendering ICC cells sensitive to targeted therapy. It has been found that although all activating FGFR2 mutants can transform and activate the receptor, the mechanisms of kinase activation are completely different. In the case of FGFR2 gene fusion, the tyrosine kinase domain of FGFR2 is retained but the C-terminal tail is replaced by a partner gene, which usually contains an intracellular dimerization domain that promotes FGFR2 dimerization and kinase activation[13]. In the case of FGFR2 EIDs, the intracellular domain of FGFR2 is intact, but changes in the extracellular domain of FGFR2 cause cysteine residues to be lost or gained. The changes in cysteine residues disrupt inhibitory intramolecular disulfide bonds or form abnormal intermolecular disulfide bonds, which promote FGFR2 dimerization and kinase activation[40]. Truncation of FGFR2 in the post kinase region is sufficient to drive ligand-independent growth and represent an alternative mechanism of FGFR2 activation, it is distinct from fusions [41–43].
In summary, our study investigated FGFR2 genetic changes in ICC and other cancers. We found that most FGFR2 fusions/translocations in ICC were in a subtype known as cholangiolocarcinoma and we called this group "FGFR2 fusion subtype ICC". We observed that only 1.38% (4/290) of ICC patients had FGFR2 mutations and that FGFR2 in-frame deletions occurred more frequently in ICC than in other cancers. FGFR2 fusions rarely occurred with other driver genes, but FGFR2 in-frame deletions and site mutations often co-occurred with TP53 mutations in ICC. Different FGFR2 mutations had varying responses to FGFR-selective small molecule kinase inhibitors in ICC cells. Our findings offer insight for future treatments for patients with FGFR2 genetic alterations in ICC.