Therapeutic yield of extensive molecular profiling in cholangiocarcinoma: a retrospective single-center study

Current available systemic therapies for advanced cholangiocarcinoma (CCA) are of limited effectiveness and prognosis is poor. Recently, introduction of next-generation sequencing (NGS) technologies led to a better understanding of the genetic pathophysiology and, consequently, identification of molecular alterations for targeted treatment. To determine the proportion of actionable alterations using extensive molecular profiling in a routine diagnostic setting and to study the effect of targeted treatment on disease control. Results of extensive molecular testing by either FoundationOne NGS or an in-house developed 96 cancer gene panel were retrospectively collected from patients with locally advanced or metastatic CCA diagnosed between 01/12/2018 and 01/08/2021 in a single center. Gene variants were classified according to ESCAT and correlated with efficacy endpoints. Of 125 patients included, 65 patients had an intrahepatic CCA (iCCA). FGFR2 fusions and IDH1/BAP1 mutations were more frequent in iCCA, while KRAS and SMAD4 mutations were predominant in extrahepatic CCA (eCCA). Targetable alterations (ESCAT tiers I–IV) were identified in 73,6% of patients. Overall survival was significantly better for higher tiers regardless of treatment. Thirteen patients (10.4%) received targeted treatment based on molecular profiling, with a median progression-free survival (PFS) of 7.3 months. Extensive molecular characterization led to the identification of targetable and potentially targetable alterations in a significant proportion of patients with locally advanced or metastatic CCA. We confirmed the association between higher ESCAT tier and benefit of a targeted treatment. Molecular analysis should therefore be considered in all patients fit enough for systemic treatment.


Introduction
Cholangiocarcinomas (CCAs) are rare and highly heterogeneous biliary malignant tumors that can arise at any site of the biliary tree (Banales et al. 2020). They represent the second most common type of primary liver cancer worldwide, after hepatocellular carcinoma (HCC) (Churi et al. 2014;Malencia et al. 2020). The incidence is highest in Southeast Asia, Latin America, and Eastern Europe, and demonstrates a female predominance (Athauda et al. 2020).
Despite the overall decline in cancer related death over the past few years (2015-2019), the incidence and mortality of CCA, particularly intrahepatic cholangiocarcinoma (iCCA), increases further (Rizvi et al. 2018;Cronin et al. 2022).

3
The current clinical classification is based on anatomic location: intrahepatic (iCCA), perihilar (pCCA), and distal cholangiocarcinomas (dCCA), as well as gallbladder and ampullar carcinoma. Often perihilar and distal are collectively referred to as extrahepatic cholangiocarcinoma (eCCA).
The two most common forms are pCCA and dCCA, accounting for approximately 50%-60% and 30%-40% of the cases, respectively, with iCCA representing only 10%-20% of the total (Rompianesi et al. 2021). Each subtype has a distinct epidemiology, biology, prognosis, and strategy for clinical management (Banales et al. 2020).
The overall survival is poor because of late detection and suboptimal response to standard therapy.
Surgery is the only curative treatment, but only a minority of patients (20-35%) have an early stage disease at time of diagnosis (Ioffe et al. 2021;Rizvi et al. 2018;Valle et al. 2021). Unfortunately, even for those patients, the risk of tumor relapse remains high, despite recent developments with capecitabine as adjuvant treatment (BILCAP study) (Primrose et al. 2019).
For patients with locally advanced or metastatic disease, the available systemic therapies are of limited effectiveness: the median overall survival with the current standard-of-care chemotherapy regimen in first line, cisplatin-gemcitabine based on the ABC-02 study (Valle et al. 2010) is less than 1 year. Recently, two phase III trials showed that adding a checkpoint inhibitor (durvalumab or pembrolizumab) to the standard chemotherapy backbone resulted in improved median overall survival with some patients achieving long-lasting responses (Kelley et al. 2023;Oh et al. 2022). Secondline options include FOLFOX, based on the results of the ABC-06 trial (Lamarca et al. 2021). The addition of liposomal irinotecan to 5-FU in second line has resulted in conflicting results and is currently not considered standard-of-care (Vogel et al. 2022;Yoo et al. 2021).
In recent years, the introduction of next-generation sequencing (NGS) technologies has opened new horizons for a better understanding of the genetic pathophysiology of CCA and, consequently, for the identification of molecular alterations for targeted treatments (Banales et al. 2020). Biliary tract cancers (especially iCCA) are known to present one of the highest frequencies of targetable molecular alterations across cancer types (Boilève et al. 2021;Verdaguer et al. 2022). Therefore, molecular profiling is now the standard of care in advanced and metastatic disease.
Several actionable targets have shown a degree of clinical success in patients with previously treated CCA, and so, recently several targeted therapies have been Food and Drug Administration (FDA), and some of them European Medicines Agency (EMA) approved for the treatment of advanced biliary tract cancer. Moreover, several tissue agnostic approvals are applicable to CCA harboring specific molecular alterations.
Approximately 13% of iCCA carry an isocitrate dehydrogenase (IDH1/IDH2) mutation, which is associated with a poor prognosis. More than 65% of these mutations are seen in woman and are almost never found in eCCA (< 1%) (Boilève et al. 2021). Ivosidenib was evaluated in a randomized, double-blind, phase 3 ClarIDHy study in patients with advanced IDH1 mutant CCA after progression on previous therapy. Benefit was seen with ivosidenib compared to placebo, with median PFS of 2.7 months vs 1.4 months (Abou-Alfa et al. 2020a; Athauda et al. 2020). Somatic mutations in the tumor suppressor genes BRCA1/2 are also reported and BRCA -mutated tumors often response to poly[ADP-ribose] polymerase (PARP) inhibition (Rizvi et al. 2018).
About 15-20% of biliary tract cancer (BTC) patients exhibit overexpression of HER2, also known as ERBB2, with a high preference for eCCA and gallbladder cancer. An ORR of 23% was seen in 39 HER2-positive patients treated with a dual anti-HER2 regimen, consisting of pertuzumab and trastuzumab in the BTC cohort of the MyPathway project, a nonrandomized, multicenter, open-label, phase IIa multiple basket trial (Meric-Bernstam et al. 2019). The phase II HERB trial evaluated the efficacy and safety of trastuzumabderuxtecan in patients with HER2-expressing unresectable or recurrent BTC and showed a 36.4% ORR (Ohba et al. 2022).
Other tissue agnostic approvals include malignancies positive for gene fusions involving NTRK (entrectinib and larotrectinib) or RET (selpercatinib). Finally, BRAF V600E mutations occur in 5% of biliary tract tumors and can be treated with a combination of dabrafenib plus trametinib (Rizzo et al. 2021;Subbiah et al. 2020Subbiah et al. , 2022. The complex molecular field of cholangiocarcinoma has opened new windows for improving the outcome of this therapeutically challenging rare disease (Verdaguer et al. 2022).
In this study, we investigated the proportion of targetable alterations in different CCA subtypes in Belgian patients presenting at a single center with a locally advanced or metastatic CCA. Different extensive molecular genetic tests were applied based on the dynamic evolution of genetic test strategies. We used the European Society for Medical Oncology (ESMO) Scale for Clinical Actionability (ESCAT) to classify identified variants and studied the impact of subsequent therapeutic choices and the effect on disease control.

Study design and population
All patients diagnosed with locally advanced or metastatic CCA between 1/12/2018 and 1/8/2021 who were treated at a single center (University Hospitals Leuven, Belgium) and whose tumor underwent NGS-based molecular analysis were included in this retrospective analysis.
Data collected from patients' electronic records included demographics, targeted treatments, response to targeted treatment, and survival status. The data were extracted in September 2021 from patient records and collected in an electronic database system, Microsoft Excel, in a pseudonymized manner. Also, patients with insufficient data were included in this study, with the lacking data reported as missing. The Research Ethics Committee UZ/KU Leuven review deemed that formal patient approval was not required owing to the retrospective, anonymous, and observational nature of this study.

Sample collection and NGS genomic profiling
Molecular profiling was performed either by FoundationOne CDx assay, FoundationOne Liquid CDx, or by an in-house capture-based next-generation sequencing panel of 96 frequently altered genes (Vanden Bempt et al. 2021).
For those tumors tested by the in-house panel, additional MSI immunohistochemistry (IHC) or polymerase chain reaction (PCR) was performed. In some cases, HER2 IHC or in situ hybridization (ISH) as well as NTRK IHC was performed too. Molecular alterations were classified along the ESMO ESCAT. The ESMO ESCAT defines six levels of clinical evidence for molecular targets according to the implications for patient management: tier I, targets ready for implementation in routine clinical decisions; tier II, investigational targets that likely define a patient population that benefits from a targeted drug, but additional data are needed; tier III, clinical benefit previously demonstrated in other tumor types or for similar molecular targets; tier IV, preclinical evidence of actionability; tier V, evidence supporting co-targeting approaches; and tier X, lack of evidence for actionability (Mateo et al. 2018).

Statistical analysis
Patient characteristics, age, gender, primary tumor site, underlying primary sclerosing cholangitis (PSC), and disease extent were recorded, as well as the detected genetic alterations. We investigated the potential associations between genetic alterations and survival. All patients with at least 2 months of follow-up were included in the survival analysis. The progression-free survival (PFS) and overall survival (OS) were estimated using the Kaplan-Meier method. The Cox proportional hazards model was used to calculate the hazard ratio and 95% confidence intervals. A p value of less than 0.05 was considered significant.

Patients' population
Overall, 125 patients with locally advanced or metastatic CCA were treated in our institution between 2018 and 2021 and underwent molecular analysis. Key demographics are presented in Table 1.
Sixty one (48.8%) patients were female and the median age of the cohort was 65 years.
Underlying primary sclerosing cholangitis was present in 4 patients.
Median overall survival for those patients with a follow-up of at least 2 months was 18.3 months with a median follow-up of 24.6 months.

Molecular analysis
An overview of all detected relevant genetic aberrations is provided in Table 1. 83 samples were analyzed using FoundationOne panel (62 tissue samples and 21 plasma) and 23 samples with the in-house capture sequencing-based assay. Additional PCR and/or IHC for the detection of MSI-H/dMMR was performed in 17 samples tested with the in-house panel. MSI-H/dMMR testing, NTRK IHC, HER2 IHC/ISH, or homologous recombination deficiency (HRD) (germline/ somatic) testing, was the only test performed in the remaining 19 samples. Molecular analysis failed quality control in 5 samples sent for FoundationOne analysis and 2 samples assessed by the in-house panel. 6/83 (7.2%) of samples and 2/23 (8.7%) of samples did not reveal one genetic aberration assessed by FoundationOne and in-house panel, respectively. Both FoundationOne CDx and FoundationOne Liquid CDx showed an equal amount of tiers I/II alterations (42.1% and 43%, respectively).
In patients with eCCA, KRAS and SMAD4 mutations, as well as ERBB2 amplification was most frequently observed ( Fig. 1).
Regardless of having received a targeted treatment, median overall survival for patients with ESCAT tiers I-II alterations was 24.9 months compared to 18.2 months for ESCAT tiers III-IV alterations and 12.7 months for others (p = 0.0431) (Fig. 2b).

Targeted treatments
Of the 125 patients, 13 (10.4%) received a targeted treatment based on the results of the molecular analysis. This includes 28% of patients with an ESCAT tier I alterations and 30% of patients with an ESCAT tier II alterations. Targeted treatments include immune checkpoint inhibitors for   (Fig. 3a).

Discussion
As a result of the identification of several targetable molecular alterations, the therapeutic landscape of biliary tract cancer is rapidly changing (Vogel et al. 2023 Here, we describe a large single-center cohort of patients with BTC who underwent molecular profiling of tumor DNA, either derived from tissue or from plasma. In contrast to what is commonly reported, the majority of patients in our cohort had intrahepatic cholangiocarcinoma. This finding might be due to selection bias, as iCCA is known to harbor the majority of the targetable alterations. However, recent global phase II/III trials in unselected biliary tract cancer patients included similar proportions of iCCA patients (Oh et al. 2022;Hack et al. 2021) suggesting that that this subtype is becoming the dominant one (Vithayathil et al. 2022).
With our study, we confirmed the marked differences in molecular drivers between iCCA an eCCA. FGFR2 fusions and IDH1 and BAP1 mutations were relatively more frequent in iCCA. In contrast, the profile of eCCA resembles that of pancreatic ductal adenocarcinoma with a higher frequency of KRAS and SMAD4 mutations. These results are in line with those of others (Bekaii-Saab et al. 2021;Churi et al. 2014;Ioffe et al. 2021;Valle et al. 2021;Verdaguer et al. 2022). Overall, we detected targetable alterations (ESCAT tiers I-IV) in 73.6% of patients. ESCAT tiers were of prognostic importance in our cohort, as overall survival was significantly better for higher tiers regardless of treatment. This emphasizes the need for caution when attributing differences in outcomes of patients that did or did not receive a targeted agent directly to personalized treatment, as a different disease biology might also play a role.
Of those patients who received a targeted treatment, those in ESCAT tier I derived significantly more benefit in terms of PFS compared to ESCAT tier II, validating the principle of the ESCAT system, as shown in other cancer types (Andre F et al. 2022). The most important contributor to this effect was the treatment of patients with dMMR/MSI-H tumors with checkpoint blockade. No differences in overall survival were noted, as our study lacked power to detect a significant effect.
In our study, only seven samples (6.6%) failed quality control, but there are data demonstrating a much higher failure rate of tissue biopsies in patients with CCA, with over 26% of samples inadequate for molecular profiling (Boliève et al. 2021). To overcome the drawbacks of tissue-based analysis, recent studies focus on the use of cell-free DNA (cfDNA). In cases when tissue is unavailable or of poor quality or quantity, cfDNA can provide genomic data. In theory, serial cfDNA analyses could guide further therapies by revealing resistance mechanisms or emerging targets. Finally, cfDNA may better capture intra-patient tumoral heterogeneity (Berchuck et al. 2022). Reports suggest a high level of reliability for IDH1 mutations (87%) and BRAF V600E mutations (100%) between cfDNA and tissue; but a disappointing concordance for FGFR2 fusions (18%) using the Guardant360 cfDNA NGS assay. The Illumina TruSight Oncology 500 circulating tumor DNA (ctDNA) assay has a higher reliability (FGFR2 fusions 87%) (Berchuck et al. 2022).
The limitations of our study are related to its retrospective design. Selection bias might affect the results, as certain subgroups of patients with BTC might more frequently undergo successful molecular profiling. Second, we could not compare outcomes of patients that received targeted treatment with those that did not, as such analyses are suffering from immortal time bias (Yadav et al. 2021). Third, as some patients only got liquid biopsy, there could be an underestimation of the number of FGFR2 fusions; as well as of the HER2-positive patients (particularly in gallbladder and iCCA) as not all patients received HER2 IHC.
In conclusion, we found targetable alterations in a significant proportion of patients with BTC and confirmed the association between higher ESCAT tier and benefit of a targeted treatment. NGS-based testing of all patients with locally advanced or metastatic BTC is therefore recommended, early in the course of the disease. However, cautious communication to patients about its potential is warranted. Several drivers are not targetable to date and the benefit of targeted treatment remains limited in some patients despite a druggable alterations, with some exceptions, such as in patients with MSI-H tumors treated with checkpoint inhibition.