Summary: diagnostic pathway and detection of FGFR2 fusions or rearrangements

Diagnosis rests on cross-sectional imaging followed by histological or cytological confirmation, because no radiological pattern is specific for cholangiocarcinoma. Guidelines recommend a core biopsy for pathology and molecular profiling before non-surgical treatment, molecular analysis in advanced disease suitable for systemic treatment, and profiling when first-line systemic treatment is initiated; parallel testing by next-generation sequencing is preferred over single-gene testing, and ESMO recommends tumour next-generation sequencing in advanced cholangiocarcinoma. FGFR inhibitors are recommended for FGFR2 fusions after at least one prior line of systemic therapy.

FGFR2 fusion detection is technically demanding because partners are diverse. In 226 cases tested by three methods, detection rates were 9.7% with RNA-based sequencing, 7.1% with DNA-based sequencing, and 10.2% with FISH, and against RNA-confirmed fusions DNA-based sequencing had a sensitivity of 71.4% and FISH 95.2%. In an interlaboratory ring trial, 13 of 16 centres (81%) passed the sequencing round, and partner-agnostic methods were judged superior. FoundationOne CDx is FDA approved for detecting FGFR2 fusions or rearrangements for pemigatinib, with an adjusted positive percentage agreement of 87.1% against an externally validated assay and 100% agreement on all measures against the F1 CLIA/CAP assay, although 27 of 107 FGFR2 rearrangement-positive participants enrolled in FIGHT-202 could not be evaluated with it because of insufficient DNA; the futibatinib label states that no FDA-approved test is available, and the lirafugratinib approval letter requires validation studies to support labeling of a companion diagnostic using blood or tissue. Circulating tumour DNA detected FGFR2 rearrangements in 5.3% of cholangiocarcinoma liquid biopsies against 7.6% of tissue samples, rising to 8.4% when tumour fraction was 1% or higher, and detected 84% of tissue-confirmed FGFR2 fusions or rearrangements in one study, whose authors advise confirming fusions with unidentified partners by tissue profiling. ESMO ranks FGFR2 fusions at ESCAT level IB, with an estimated prevalence of 5%-15% of intrahepatic cholangiocarcinomas, and notes that a DNA-based test might miss a fusion whose break point region is not fully covered by the gene panel, whereas RNA-based panels identify the fusion transcript and the fusion partner. In a multicentre Italian cohort, extended molecular profiling was performed in 79.9% of participants with advanced cholangiocarcinoma, and 26 (18.7%) of the 139 participants with an ESCAT I-III alteration received a corresponding tailored treatment.

Diagnosis: guideline recommendations on pathology, molecular biology, and staging

“Molecular analysis is recommended in advanced disease considered suitable for systemic treatment [I, A]” (opens the source at this quote in a new tab)
“Elevated CA 19-9 is associated with poorer prognosis and can be useful for assessing response to treatment [III, C]” (opens the source at this quote in a new tab)
“Thoraco-abdomino-pelvic CT remains the reference examination for lymph node and metastatic extension [III, A]” (opens the source at this quote in a new tab)
“FDG-PET is not recommended for imaging of the primary tumour. However, it may be considered where there is clinical suspicion of nodal metastases, distant metastases and disease recurrence [III, C; consensus = 100%].” (opens the source at this quote in a new tab)
“Radiological imaging should be carried out before ERCP or PTC in patients with jaundice [III, A]” (opens the source at this quote in a new tab)

Diagnosis: guideline recommendations on tissue acquisition for molecular profiling

“A core biopsy should be obtained for diagnostic pathology and molecular profiling before any nonsurgical treatment [III, A].” (opens the source at this quote in a new tab)
“In patients with d/pCCA without extraductal metastasis, PTC- or ERCP-guided biopsies should be carried out to obtain adequate tissue for diagnostic pathology and molecular profiling [III, A].” (opens the source at this quote in a new tab)

Diagnosis: composition of the molecular testing panel

“Parallel sequencing of several genes using focused next-generation sequencing (NGS) is preferred over single gene testing.” (opens the source at this quote in a new tab)
“Currently, the gene panel should include the respective coding DNA regions (target regions) of isocitrate dehydrogenase 1 (IDH1), human epidermal growth factor receptor 2 (HER2)/neu [v-erb-b2 avian erythroblastic leukaemia viral oncogene homologue 2 (ERBB2)] and BRAF to test for hotspot mutations, but the rapidly evolving landscape of drug targets and predictive biomarkers may soon necessitate larger panels.” (opens the source at this quote in a new tab)

Diagnosis: timing of molecular profiling in advanced disease

“Molecular profiling is recommended when first-line systemic treatment is initiated in patients with locally advanced, advanced or metastatic disease, particularly in those at high risk of progression or recurrence [I, A].” (opens the source at this quote in a new tab)

FDA-approved test requirement for FGFR2 fusion or rearrangement detection: pemigatinib

“Select patients for the treatment of locally advanced or metastatic cholangiocarcinoma with PEMAZYRE based on the presence of an FGFR2 fusion or rearrangement as detected by an FDA-approved test” (opens the source at this quote in a new tab)

FoundationOne CDx analytical validation for FGFR2 rearrangements

“The next-generation sequencing (NGS)-based FoundationOneCDx (F1CDx) was US Food and Drug Administration approved for detecting FGFR2 fusions or rearrangements.” (opens the source at this quote in a new tab)
“Adjusted positive, negative, and overall percentage agreements were 87.1%, 99.6%, and 98.3%, respectively, between F1CDx and evNGS. Compared with evNGS, F1CDx had a positive predictive value of 96.2% and a negative predictive value of 98.5%.” (opens the source at this quote in a new tab)

FoundationOne CDx clinical validation, limit of detection, and discordant cases

“The positive percentage agreement, negative percentage agreement, overall percentage agreement, positive predictive value, and negative predictive value all exhibited 100% agreement between the F1CDx assay and the F1 CLIA/CAP assay.” (opens the source at this quote in a new tab)
“The 80 enrolled patients who tested positive with both the F1 CLIA/CAP assay and F1CDx had a similar objective ORR of 37.5% (95% 2-sided exact CI, 26.9%–49.0%).” (opens the source at this quote in a new tab)
“Of the 107 FGFR2 rearrangement-positive patients enrolled in FIGHT-202, 27 could not be evaluated with F1CDx because of insufficient DNA.” (opens the source at this quote in a new tab)
“Using the empirical hit rate method, the analytical sensitivity was determined to be 10.75 chimeric reads or a tumor purity of 5.31%.” (opens the source at this quote in a new tab)
“F1CDx provided a reproducibility and repeatability of 100% for four of five samples, with one sample (sample number 4) demonstrating slightly lower reproducibility and repeatability of 95.5% and 90%, respectively (Supplemental Table S1).” (opens the source at this quote in a new tab)
“In our study comparison of F1CDx DNA versus RNA-based evNGS, only 2 of 130 patients who tested negative for FGFR2 rearrangement by F1CDx were found to be positive by RNA-based evNGS.” (opens the source at this quote in a new tab)
“Comparably, only 1 of 26 patients who tested positive by F1CDx was negative by RNA-based evNGS.” (opens the source at this quote in a new tab)
“Post hoc analysis of four select FGFR2-N/A rearrangements by RNA-based methods [RNA sequencing or evNGS (ArcherDX FusionPlex)] revealed the presence of in-frame fusions in each case, supporting this hypothesis.” (opens the source at this quote in a new tab)

Companion diagnostic status for futibatinib and lirafugratinib

“An FDA-approved test for detection of FGFR2 gene fusions or other rearrangements in patients with unresectable, locally advanced, or metastatic intrahepatic cholangiocarcinoma for selecting patients for treatment with LYTGOBI is not available.” (opens the source at this quote in a new tab)

“Conduct adequate analytical and clinical validation studies to support revisions to the labeling of lirafugratinib to specify the use of an FDA- authorized in vitro companion diagnostic device to accurately and reliably identify patients using blood and/or tissue samples who have FGFR2 fusions and other rearrangements and for whom lirafugratinib therapy is indicated.” (opens the source at this quote in a new tab)

Comparison of RNA-based NGS, DNA-based NGS, and FISH for FGFR2 fusion detection

“The detection rates for FGFR2 fusion/rearrangement were 9.7% (22/226) for RNA-based NGS, 7.1% (16/226) for DNA-based NGS, and 10.2% (23/226) for FISH. Among the 26 FGFR2 fusion/rearrangement-positive cases identified by any method, only 15 (57.7%) were positive by all three techniques, yielding a concordance rate of 95.1% (215/226).” (opens the source at this quote in a new tab)
“FISH demonstrated a sensitivity of 95.2% and specificity of 98.5%, compared to oncogenic FGFR2 fusions confirmed by RNA-based NGS, while DNA-based NGS exhibited a sensitivity of 71.4% and specificity of 99.5%, identifying FGFR2 mutations in 4 cases.” (opens the source at this quote in a new tab)

Limitations of break-apart FISH for FGFR2 rearrangement detection in routine practice

“Furthermore, the detection of FGFR2 rearrangements relied on break-apart probe FISH rather than RNA-based NGS. Although reliable for detecting structural rearrangements, it does not provide information on fusion partners or confirm whether the resulting transcript is in-frame, and may yield false-negative results for rearrangements involving partners in close proximity to FGFR2” (opens the source at this quote in a new tab)

Interlaboratory proficiency testing for FGFR2 fusion detection

“Thirteen of 16 (81%) centres successfully passed the NGS only and 3 of 5 (60%) centres passed the combined NGS + FISH round robin test. Identified obstacles were bioinformatic pipelines not optimised for the detection of FGFR2 fusions and assays not capable of detecting unknown fusion partners.” (opens the source at this quote in a new tab)
“Due to the marked heterogeneity of the genomic architecture of these fusions, fusion partner agnostic (i.e. open) methodological approaches that are capable of identifying yet unknown fusion partners are superior.” (opens the source at this quote in a new tab)

Partner-agnostic sequencing to detect known and novel FGFR2 fusions

“In view of the molecular diversity of cholangiocarcinoma, these results highlight the importance of incorporating DNA-based or RNA-basednext-generation sequencing assays into standard practice, to detect both known and novel FGFR2 fusions or rearrangements, and thus to identify all patients who might benefit from FGFR-targeted therapies.” (opens the source at this quote in a new tab)

Detection of FGFR2 rearrangements in circulating tumor DNA

“The overall prevalence of FGFR2 rearrangements in CCA LBx was 5.3% (Fig. 2A), lower than the 7.6% observed in tissue (P = 0.004). However, among LBx samples with TF ≥1% (525/1,215; 43%), the prevalence of 8.4% was comparable to tissue (Fig. 4A).” (opens the source at this quote in a new tab)

“Among patients with known FGFR2 fusions/rearrangements, FGFR1 fusions, FGFR3 fusions, or FGFR2 amplifications per tissue analysis, detection rates in ctDNA for these aberrations were 84%, 0%, 11%, and 59%, respectively.” (opens the source at this quote in a new tab)

“Sixteen patients (88.9%) were positive for FGFR2 fusion events in plasma.” (opens the source at this quote in a new tab)
“Additionally, increased ctDNA or the emergence of resistance mutations allowed earlier detection of disease progression compared with standard radiologic imaging methods.” (opens the source at this quote in a new tab)

Confirmation of ctDNA-detected FGFR2 fusions by tissue profiling

“The detection rate of FGFR2 fusions/rearrangements by ctDNA was 84% (98/117) among all tumor types and 85% (96/113) in cholangiocarcinoma (Figure 2A).” (opens the source at this quote in a new tab)
“Among 18 patients with FGFR2 fusions/rearrangements detected in tissue alone, 9 (50%) had a response to futibatinib” (opens the source at this quote in a new tab)
“Notably, FGFR2 fusions with unidentified partners and/or concurrent high-level FGFR2 amplifications in ctDNA may not be driver events; in these cases, FGFR2 fusions should be confirmed using tissue profiling.” (opens the source at this quote in a new tab)

Guideline recommendation for tumour next-generation sequencing in cholangiocarcinoma

“As for 2020 recommendations, ESMO recommends running tumour NGS in advanced non-squamous non-small-cell lung cancer, prostate cancer, colorectal cancer, cholangiocarcinoma, and ovarian cancer.” (opens the source at this quote in a new tab)

ESMO actionability ranking of FGFR2 fusions in advanced cholangiocarcinoma

“Based on the number of alterations classified as level I, it is recommended to carry out tumour NGS in patients with advanced CCA.” (opens the source at this quote in a new tab)
“Table 8 provides a list of genomic alterations level I/II according to ESCAT in advanced CCA.” (opens the source at this quote in a new tab)

Gene	Alteration	Estimated prevalence	ESCAT score	Drug class matched
FGFR2	“Fusions”	“5%-15% iCCA”	“IB”	“Pan-FGFR TKIs”
IDH1	“Mutations”	“8%-18% iCCA”	“IA”	“IDH1 inhibitors”

ESMO guidance on DNA-based and RNA-based sequencing for fusion detection

“ESMO recommends carrying out multigene NGS in patients with advanced cancers in countries where tumour-agnostic targeted therapies are accessible. Cost-effectiveness should be assessed at the local level and the decision to implement NGS should be taken accordingly. It is important for clinicians to ensure that fusions are integrated in the panel.” (opens the source at this quote in a new tab)
“The main point is also to stress the fact that clinicians must assess whether the NGS panel includes the detection of the fusions recommended in a specific disease.” (opens the source at this quote in a new tab)
“DNA sequencing has the advantage that DNA is more stable than RNA. However, DNA sequencing can only identify break points of translocations leading to gene fusions.” (opens the source at this quote in a new tab)
“For example, a test that does not detect a fusion might be false negative if the break point region is not fully covered by the gene panel employed for the diagnostic test.” (opens the source at this quote in a new tab)
“Several limitations of DNA-based fusion gene sequencing are overcome by RNA-based sequencing methods.” (opens the source at this quote in a new tab)
“RNA-based gene panels identify the transcript of the fusion gene resulting from a translocation, and provide data on the ‘expression’ of the fusion transcript, the fusion partner as well as the potential functionality (e.g. out-of-frame versus in-frame fusions).” (opens the source at this quote in a new tab)
“In contrast to DNA-based assays, it does not provide information on break points.” (opens the source at this quote in a new tab)

Diagnosis: guideline recommendations on second-line molecularly directed therapy

“Ivosidenib is recommended for the treatment of patients with CCA and IDH1 mutations who have progressed after ≥1 prior line of systemic therapy” (opens the source at this quote in a new tab)
“FGFR inhibitors are recommended for the treatment of patients with FGFR2 fusions who have progressed after ≥1 prior line of systemic therapy” (opens the source at this quote in a new tab)
“Pembrolizumab is recommended in patients with MSI-H/dMMR who have progressed on or are intolerant to prior non-immune checkpoint inhibitor-containing treatment” (opens the source at this quote in a new tab)
“Dabrafenib-trametinib is recommended for the treatment of patients with BRAFV600E mutations who have progressed after ≥1 prior line of systemic therapy” (opens the source at this quote in a new tab)

Diagnosis: imaging and tissue acquisition in intrahepatic cholangiocarcinoma

“iCCA is often seen at presentation as an intrahepatic mass and is incidentally found in 25–30% of patients” (opens the source at this quote in a new tab)
“Carbohydrate antigen 19-9 (CA19-9) is the primary serum biomarker used in CCA diagnosis, although it has subpar specificity and can be elevated in various conditions, such as biliary obstruction or pancreatic cancer.” (opens the source at this quote in a new tab)
“MRI may provide enhanced assessment of the primary mass, whereas CT imaging has superior detection of vascular enhancement and is, therefore, important in determining resectability” (opens the source at this quote in a new tab)
“Histopathological analysis of a biopsy specimen remains the mainstay for confirmation of an iCCA diagnosis.” (opens the source at this quote in a new tab)

Diagnosis: perihilar and distal cholangiocarcinoma

“Patients with pCCA and dCCA typically present with painless jaundice owing to underlying biliary obstruction.” (opens the source at this quote in a new tab)
“The sensitivity and specificity of MRCP to distinguish benign and malignant causes of hilar obstruction are 87% and 85%, respectively” (opens the source at this quote in a new tab)
“ERC has a diagnostic and therapeutic role in pCCA and dCCA as it enables detection of malignant strictures and acquisition of biliary brushings for cytology and fluorescence in situ hybridization (FISH) analysis.” (opens the source at this quote in a new tab)
“However, the sensitivity of biliary cytology for pCCA detection is low (43% according to one meta-analysis127). FISH analysis has a higher sensitivity than cytology (65% versus 19% in one series) and similar specificity for detection of CCA” (opens the source at this quote in a new tab)
“EUS–FNA has a higher sensitivity for detection of dCCA than detection of pCCA (81% versus 59%)” (opens the source at this quote in a new tab)

Diagnosis: histological confirmation and the limits of imaging

“As no specific CCA radiology pattern exists, histopathological or cytological analysis is mandatory to confirm the diagnosis1,28. This diagnosis is based on the WHO classification of biliary tract cancer showing an adenocarcinoma or mucinous carcinoma101, with tubular and/or papillary structures and a variable fibrous stroma102.” (opens the source at this quote in a new tab)
“CT is considered the standard imaging method for the preoperative assessment of both iCCA and pCCA; it provides a comprehensive evaluation of the primary tumour, the relationship with adjacent structures, and potential thoracic and abdominal spread” (opens the source at this quote in a new tab)

Diagnosis: staging and its limitations

“There is no widely used staging system for CCA, although it can be staged according to the American Joint Committee on Cancer (AJCC) TNM system103,104.” (opens the source at this quote in a new tab)
“Finally, the TNM classification misses relevant prognostic factors such as the presence of cancer-related symptoms (such as abdominal pain or malaise) or the degree of liver function impairment.” (opens the source at this quote in a new tab)

“Disease stage was defined as locally advanced versus metastatic as per the American Joint Committee on Cancer staging (version 7),21 according to which intrahepatic cholangiocarcinoma with multifocal or satellite liver lesions or metastases were classified as locally advanced in the absence of extrahepatic metastatic disease.” (opens the source at this quote in a new tab)

Diagnosis: yield of comprehensive genomic profiling in biliary tract tumours

“Reportable genomic alterations were detected in 88 patients (96.7%), whereas no reportable alteration was detected in three patients (3.3%).” (opens the source at this quote in a new tab)
“Potentially actionable genomic alterations were identified in 21 patients (23.1%). Expert panel-based therapeutic options were identified in 15 patients (16.5%), and genomically matched therapy was introduced in 7 patients (7.7%).” (opens the source at this quote in a new tab)
“Site-specific frequencies of potentially actionable alterations were highest in iCCA (17/46, 37.0%), followed by GBC (4/24, 16.7%), whereas no potentially actionable genomic alteration was identified in eCCA or ampullary tumors using this definition.” (opens the source at this quote in a new tab)

Diagnosis: uptake of and barriers to extended molecular profiling in routine practice

“EMP was performed in 79.9% of patients with advanced CCA. The rate of EMP increased significantly over time (2017–2023: +25.8%).” (opens the source at this quote in a new tab)
“Our results show that a larger proportion of patients with BTCs have had access to EMP in recent years, with a clear increase of 25.8%, from 43.2% in 2017 to 69.0% in 2023.” (opens the source at this quote in a new tab)
“EMP should at least include the identification of FGFR2 and NTRK rearrangements, IDH1 R132x and BRAF V600E mutations, microsatellite instability (MSI)/mismatch repair (MMR) status and HER2 amplification” (opens the source at this quote in a new tab)
“On the other hand, several limitations to the widespread application of EMP have been identified in daily clinical practice throughout Europe: complex pre-analytical steps, high rates of test failure in samples with low tumor cell content, prolonged turnaround time, increased costs and the need for a dedicated molecular tumor board to correctly interpret the biological data all make the implementation of EMP in BTCs challenging.” (opens the source at this quote in a new tab)
“All the analyses were performed on tissue samples, and no plasma samples were collected to perform molecular profiling.” (opens the source at this quote in a new tab)
“Of the 139 patients harboring a druggable alteration according to ESCAT I-III tiers, only 26 (18.7%) received one or more corresponding tailored treatments.” (opens the source at this quote in a new tab)

Diagnosis: misclassification as cancer of unknown primary

“Therefore, patients with CUP often face protracted diagnostic journeys, rapidly spreading metastases, and median OS ranging from two to eight months.” (opens the source at this quote in a new tab)