Multidisciplinary management in the treatment of intrahepatic cholangiocarcinoma

A 63-year-old woman who was a former smoker with a past medical history of hypertension and gastroesophageal reflux disease initially presented with upper abdominal pain. Her family history was notable for breast cancer in her mother, lung cancer in her father, and renal cell carcinoma in her sister. An ultrasound showed a heterogenous mass in the left lobe of the liver measuring 8.7 × 7.0 × 5.1 cm that was abutting the common bile duct and concerning for a neoplasm (Figure 1A). On laboratory testing, her alpha fetoprotein (AFP) was elevated (15.7 ng/mL), carbohydrate antigen 19-9 (CA 19-9) was normal (<15 U/ml), and carcinoembryonic antigen (CEA) was slightly elevated (0.6 ng/ml). She underwent an ultrasound-guided biopsy that demonstrated cytokeratin 7 (CK7)-positive, poorly differentiated adenocarcinoma with nonmucinous gland formation and papillary architecture within sclerotic stroma. Given that the biopsy was positive for CK7 with negative hepatocellular (hepatocyte-specific antigen, arginase, glypican), CDX2, TTF1, and synaptophysin markers, the mass was diagnosed as an intrahepatic cholangiocarcinoma (iCCA). A computed tomography (CT) scan of the chest, abdomen, and pelvis did not show any extrahepatic metastatic disease but did show a central left hepatic lobe mass in segment 4a/4b that measured 7.7 × 6.7 cm with calcifications suggestive of iCCA (Figure 1B,C). A CT scan also revealed potential tumor thrombus within the middle hepatic vein and distal left portal vein branches, extrahepatic (periportal, gastrohepatic, peripancreatic, portacaval) lymphadenopathy, left intrahepatic biliary ductal dilation, and common bile duct dilation.

The patient was started on gemcitabine, cisplatin, and nanoparticle albumin-bound paclitaxel (nab-paclitaxel). After 3 months of chemotherapy, the patient's AFP increased to 43.8 ng/ml and her CA 19-9 increased to 21.4 U/ml. On repeat CT scan, the size of the tumor was stable, but there was suspected intraductal extension toward the central inferior aspect of segment 4b. Given her suboptimal response to chemotherapy, radiation oncology was consulted. Approximately 4 months after starting chemotherapy, the patient underwent yttrium-90 radioembolization (Y90 RE) to the left hepatic hemiliver and subsequently was resumed on a gemcitabine, cisplatin, and nab-paclitaxel regimen (Figure 2A). A CT scan 5 months after starting treatment and 1 month after Y90 RE demonstrated a stable left hepatic lobe mass with interval necrosis. However, this effect was mostly seen in the tumor in the left lobe of the liver, whereas there was still some residual arterial enhancement along the right side of the mass because where the tumor extended into the right lobe was not treated given concern of toxicity to the remaining liver. After nine cycles of chemotherapy and the Y90 RE treatment, re-staging CT scans did not demonstrate any metastatic disease, and the tumor in the left lobe of the liver had a seemingly good response to the Y90 RE (Figure 2B). In addition, the periportal, portacaval, and gastrohepatic lymphadenopathy had decreased in size, and there was no new or progressive adenopathy.

At this point, the patient was taken to the operating room, and an extended left hepatectomy, cholecystectomy, and extensive lymphadenectomy, including skeletonizing of the hilum, left hepatic artery, bile ducts, and common hepatic artery, was performed. On postoperative day 5, the patient was tachycardic, a CT scan showed a large fluid collection in the resection bed, and the patient was brought back to the operating room. A bile leak from a small dehiscence along the left hepatic duct staple line was found and oversewn. The patient underwent endoscopic retrograde cholangiopancreatography postoperatively with placement of a biliary stent and, on cholangiogram, there was no evidence of a continued bile leak. The patient was discharged home on postoperative day 9. Final pathology demonstrated a poorly differentiated cholangiocarcinoma (CCA), small duct type, that had approximately 30% focal necrosis/70% persistent viable tumor with lymphovascular invasion and perineural invasion; carcinoma extended focally to the cauterized resection margin/edge, and there were no metastatic lymph nodes (n = 0 of 3).

Postoperative circulating tumor DNA (ctDNA) levels remained low but slightly positive, and her AFP came down to a normal range. Because of the close resection margin, it was recommended that the patient proceed with chemoradiation to the resection margin with adjuvant capecitabine. After completion of her chemoradiation, her ctDNA level was zero. Five months after completion of her radiation therapy, her ctDNA level was positive. A magnetic resonance image showed new liver lesions in the right lobe concerning for metastatic disease. Given the magnetic resonance imaging findings and positive ctDNA, there was high suspicion of recurrence of a fibroblast growth factor receptor-2 (FGFR2) fusion CCA. She tested positive for an FGFR2–AHCYL1 fusion iCCA and is currently on an FGFR inhibitor (pemigatinib) through a clinical trial.

CCA can be divided into extrahepatic CCA (eCCA) and iCCA.¹ The patient in this case scenario had an iCCA. For resectable iCCA, upfront surgery with adjuvant capecitabine is the recommended therapy. However, this patient presented with a locally advanced (extrahepatic lymphadenopathy), poorly differentiated tumor and thus was treated with upfront chemotherapy. Neoadjuvant chemotherapy can sometimes shrink the iCCA, thereby helping with the technical and anatomical considerations of the resection. Upfront chemotherapy can potentially downsize/downstage iCCA tumors to help facilitate resection. Le Roy et al. reported that 53% of patients with locally advanced iCCA were converted to resectable disease with neoadjuvant chemotherapy; these patients and had overall survival (OS) and recurrence-free survival (RFS) rates similar to those of patients who presented with initially resectable disease. Among patients with downsized/downstaged disease, 31% had an R0 resection, and 67% had an R1 resection.² In another study, 36% of patients with iCCA were able to be downsized/downstaged with upfront chemotherapy; these individuals had an improved survival versus patients who were unable to undergo surgery. Overall, one half of patients who underwent resection had an R0 resection.³ In addition, neoadjuvant chemotherapy can give the patient's tumor a test of time to declare its biology. If the patient progressed quickly on chemotherapy, then an unnecessary large operation could be avoided. The use of neoadjuvant chemotherapy was particularly indicated for this patient, who had several risk factors for early recurrence/progression (i.e., extrahepatic lymphadenopathy, poor differentiation, possible vascular invasion). Ideally, preoperative systemic therapy can help select patients for surgery who will benefit the most from an oncologic perspective.

Currently, gemcitabine/cisplatin is first-line therapy for advanced iCCA based on the ABC-02 clinical trial (ClinicalTrials.gov identifier NCT02170090). That phase 3 study included 410 patients with locally advanced or metastatic CCA, gallbladder cancer, or ampullary cancer. Patients who received treatment with cisplatin and gemcitabine had a median OS of 11.7 months, whereas patients in the gemcitabine cohort had a median OS of 8.1 months (p < .001). Progression-free survival (PFS) was 8 and 5 months in the cisplatin plus gemcitabine cohort and the gemcitabine cohort (p < .001), respectively.⁴ Based on the phase 3 ABC-02 trial, cisplatin and gemcitabine have been the standard-of-care first-line therapy for metastatic or locally advanced biliary tract cancers (BTCs) for the past decade.

Despite recent advances, patients with iCCA still have poor OS. Consequently, there have been several attempts to identify more efficacious systemic chemotherapy. The recently published TOPAZ-1 phase 3 trial (ClinicalTrials.gov identifier NCT03875235) compared cisplatin and gemcitabine combined with durvalumab.⁵ The data demonstrated that the addition of durvalumab improved OS, PFS, and the objective response rate among patients with unresectable or metastatic BTCs.⁵ In September 2022, durvalumab was approved for combination therapy with gemcitabine/cisplatin for patients with locally advanced or metastatic BTCs. Abraxane (nab-paclitaxel) in combination with gemcitabine is currently used as first-line treatment for metastatic pancreatic adenocarcinoma. Nab-paclitaxel may enhance the delivery of gemcitabine by depleting the surrounding stroma associated with pancreatic adenocarcinoma.^{6, 7} Because BTCs, including iCCAs, are also tumors rich in stromal tissue, it has been hypothesized that the addition of nab-paclitaxel will have a similar effect in BTCs. To this end, the Southwest Oncology Group (SWOG) 1815 and S1815 phase 2 and 3 clinical trials (NCT03768414) were designed to evaluate the role of nab-paclitaxel in iCCA treatment. Specifically, the SWOG 1815 phase 2 clinical trial evaluated the use of gemcitabine, cisplatin, and nab-paclitaxel in 62 patients with advanced BTCs (63% had iCCA). The addition of nab-paclitaxel led to a partial response rate of 45% and a disease control rate of 84%. The median OS for patients was 19.2 months, and the PFS was 11.8 months, both of which were improvements over historical controls.⁶ In the current clinical case, in light of these emerging data, the patient was placed on a combination of gemcitabine, cisplatin, and nab-paclitaxel. Of note, the currently ongoing phase 3 trial compares gemcitabine and cisplatin with or without nab-paclitaxel among patients with advanced BTCs (ClinicalTrials.gov identifier NCT03768414). At the recent 2023 American Society of Clinical Oncology Gastrointestinal Cancers Symposium, preliminary data from 441 patients (67% with iCCA) were presented. Although there was no statistically significant difference in median OS between the two regimens, on exploratory subset analyses, the addition of nab-paclitaxel versus gemcitabine/cisplatin alone demonstrated a trend toward improved OS among patients with locally advanced disease (median OS, 19.2 vs. 13.7 months, respectively; p = .09).⁸ Given the approval of durvalumab by the US Food and Drug Administration and data from the phase 3 SWOG 1815 trial, future patients with locally advanced or metastatic iCCA may benefit from the addition of nab-paclitaxel; however, more data are needed.

After Y90 RE treatment and subsequent resection, the patent was started on adjuvant capecitabine. This decision was based on the BILCAP trial (EudraCT number 2005-003318-24), a phase 3 clinical trial that randomized patients with BTCs to either adjuvant capecitabine or observation after curative resection. Although the study did not meet its primary end point (OS in the intention-to-treat analysis), the data did demonstrate that OS was 51.1 months in the capecitabine arm versus 36.4 months in the observation arm.⁹ In turn, the data suggested that an adjuvant regimen with capecitabine may be appropriate for patients with CCA who undergo resection—especially individuals with high-risk tumor characteristics.

Targeted therapy of genetic aberrations in FGFR-2, isocitrate dehydrogenase-1, B-Raf V600E, human epidermal growth factor receptor-2, or neutrophic tyrosine receptor kinase are emerging as important treatment strategies for patients with iCCA. In fact, molecular profiling is now the standard of care among patients with iCCA and is important to identify potential genetic aberrations that can be targeted in these patients. Among patients with iCCA, isocitrate dehydrogenase-1 mutations and FGFR-2 fusions are the most common perturbations, occurring in 15%–20% and 10%–15% of patients, respectively.¹⁰ The patient in the current clinical scenario had an FGFR-2 fusion. In preclinical studies, the knockdown of FGFR-2 can decrease cell growth and colony formation of CCA cells. In addition, inhibition of FGFR-2 enhances the suppressive effect of gemcitabine on cell migration and invasion.¹¹ Selective FGFR inhibitors have demonstrated efficacy among patients with CCA that contain FGFR-2 genetic aberrations, and several agents (i.e., pemigratinib, futibatinib, infigratinib) have been approved for patients with previously treated locally advanced or metastatic BTCs. A phase II study evaluating pemigatinib among patients with previously treated, locally advanced or metastatic CCA demonstrated a median PFS of 6.9 months and an OS of 21.1 months among patients with FGFR fusions.¹² Other studies examining patients with advanced BTCs who were treated with standard second-line therapy options reported a median OS of 5–9 months.^13-15 The FOENIX-CCA2 phase 2 trial (ClinicalTrials.gov identifier NCT02052778) was recently published and evaluated the use of futibatinib among 103 patients who had previously treated, locally advanced or metastatic iCCA with an FGFR2 genetic aberration. That study demonstrated a median PFS of 9 months and median OS of 21.7 months; the disease control rate was 83%, with an objective response rate of 42%.¹⁶ There are several ongoing clinical trials focused on comparing FGFR inhibitors with standard chemotherapy (gemcitabine/cisplatin) to determine whether these agents should be used as first-line therapy in appropriately selected patients (ClinicalTrials.gov identifiers NCT04093362, NCT03656536, and NCT03773302). Given the recurrence in our patient, molecular profiling was performed, and an FGFR-2 fusion was identified. Therefore, the patient was enrolled in a clinical trial to receive pemigatinib, an FGFR-2 inhibitor.

The ctDNA level was considered in the care of this patient. ctDNA consists of tumor DNA fragments that are released into the blood when tumor cells undergo apoptosis. ctDNA has the potential to be used for postoperative cancer surveillance, guidance of neoadjuvant treatment, as well as identification of genetic aberrations for targeted molecular therapy. Although some data from small studies suggest strong concordance with the primary and metastatic tumor, other studies have demonstrated a lack of concordance in some patient populations. Therefore, further studies are needed to validate and standardize the use of ctDNA.¹⁷

Locoregional treatments to the liver are a crucial part of multidisciplinary management to treat iCCA. In the current case, the patient had a suboptimal response to the first three cycles of chemotherapy yet did not develop extrahepatic metastatic disease. In turn, liver-directed therapy was considered. Although a hepatic artery infusion pump (HAIP) was discussed, the multidisciplinary team decided to proceed with preoperative Y90 therapy. Intra-arterial therapy with transarterial chemoembolization or Y90 has been used to treat patients with CCA. Although there are no data to support the superiority of Y90 over transarterial chemoembolization for patients with CCA, prospective studies have demonstrated better outcomes and lower toxicity for patients with hepatocellular carcinoma.^{18, 19} Given the better toxicity profile, Y90 has become a preferred intra-arterial therapy for many patients with liver malignancies.

With the Y90 RE procedure, the arterial supply of the tumor is identified using angiography, and small glass or resin embolic particles containing radio nucleotide Y90 are injected into the artery. These Y90 particles emit β-radiation, which allows for a higher dose of radiation to be delivered directly to the tumor without systemic toxicity.²⁰ Retrospective data demonstrate that Y90 treatment for iCCA can down-size the tumor to allow for surgical resection.²¹ In a retrospective study of 136 patients with unresectable iCCA (58% chemotherapy-naive) who received Y90 RE, the median OS was 14.2 months, and 8.1% of patients were converted to operative candidates.²² The recently published multicenter MispheC phase 2 clinical trial (ClinicalTrials.gov identifier NT01912053) evaluated the use of Y90 and cisplatin/gemcitabine as first-line therapy for patients with unresectable iCCA; at 3 months, there was a 41% response rate and a 98% disease control rate. At a follow-up of 36 months, the median PFS was 14 months. OS was 75% at 12 months and 45% at 24 months (median OS, 22 months). In addition, 22% of patients were down-staged and were able to undergo a surgical intervention, with an R0 resection margin in 20% of cases.²³ Given these data, the decision was made to proceed with Y90. The patient in the current case had a good response to the Y90 treatment, with decreased size and enhancement of the tumor, but the treatment effect was localized to the portion of the tumor in the left hemiliver that represented the targeted portion of the tumor. Because of concerns about toxicity, a rim of tissue on the right side of the lesion remained essentially untreated. Given the response to Y90 RE and disease control on chemotherapy, the decision was made to proceed with surgery.

After resection, there was concern that the cauterized edge of the specimen represented a close surgical margin. Therefore, after 6 months of adjuvant capecitabine, additional chemoradiation with concurrent xeloda was administered to the surgical margin. Data from one meta-analysis demonstrated a benefit in OS related to adjuvant radiation therapy among patients who had undergone an R1 resection.²⁴ In a separate retrospective study, 70 patients with iCCA were categorized into three groups according to margin width and treatment with adjuvant therapy (i.e., <1-cm resection margin plus adjuvant radiation to the margin vs. <1-cm resection margin and observation vs. ≥1-cm resection margin and observation). There were no differences in OS or disease-free survival (DFS) between patients who had <1-cm resection margin plus adjuvant radiation to the margin and those who had a ≥1-cm resection margin plus observation. However, both of these groups had improved OS and DFS compared with patients who had a <1-cm resection margin yet did not receive radiotherapy.²⁵ Therefore, patients who had close or positive surgical margins after resection likely benefit from adjuvant chemoradiotherapy.

Successful treatment of CCA relies heavily on a curative-intent resection. Patients who present with resectable disease should proceed with surgery and adjuvant capecitabine. Unfortunately, even after a successful oncologic resection, the recurrence rate ranges from 42% to 70%.^{9, 26, 27} Given our patient's high-risk disease and the high incidence of recurrence among patients with iCCA, the patient was started on neoadjuvant chemotherapy with plans to re-assess resectability after several cycles of systemic therapy. After the patient had completed three cycles of chemotherapy, the CT scan showed stable tumor size and no metastatic disease. However, there was concern for intraductal extension into the left hepatic duct near the bifurcation of the right and left hepatic ducts, as well as abutment/involvement of the middle and left hepatic veins. In addition, her AFP level was increasing. Overall, this clinical picture was consistent with a suboptimal response to treatment. From a surgical and anatomic perspective, the goal was to halt progression of the disease while administering additional cycles of neoadjuvant chemotherapy, as well as to improve response with liver-directed therapy. After Y90 RE treatment and completion of neoadjuvant chemotherapy, an improved response with a decrease in size and enhancement of the iCCA on CT scan was noted. At this time, the patient was brought to the operating room for an extended left hepatectomy, cholecystectomy, and portal lymphadenopathy.

There are several important elements of the surgical resection for iCCA that need to be considered, including anatomic resection (AR) versus nonanatomic resection (NAR), margin status, and lymph node dissection. Although the beneficial effect of AR (major resection) versus NAR (minor resection) for hepatocellular carcinoma has been established, its role relative to iCCA has been questioned.²⁸ In one retrospective study of 1023 patients with iCCA who underwent curative resection, the short-term and long-term outcomes were examined relative to major versus minor hepatectomy.²⁸ Patients who underwent a major hepatectomy had a higher risk of postoperative complications (48.4%) versus minor hepatectomy (27.2%; p < .001). In the propensity-matched analysis, patients who underwent a major versus minor hepatectomy had equivalent OS and RFS (median OS: 38 vs. 37 months, respectively; p = .556; median RFS: 20 vs. 18 months, respectively; p = .635). In a different study, DFS and OS among 702 patients with iCCA who underwent AR versus NAR were examined.²⁹ In that study, the two cohorts of patients had a similar risk of complication (AR vs. NAR, 26.6% vs. 25.1%; p = .634), yet patients who underwent AR versus NAR generally had better 1-year, 3-year, and 5-year DFS and OS. When the AR and NAR cohorts were stratified by disease stage, the benefit of AR was only seen in patients who had stage IB or II (without microvascular invasion) disease. Our patient had a stage II tumor with microvascular invasion. Based on this study, there was no difference in DFS or OS between AR and NAR for the cohort of patients on subanalysis that matched our own patient (stage II with microvascular invasion).

Although the role of AR versus NAR remains debated, achievement of an R0 margin should remain a universal goal of curative-intent resection for iCCA. Unfortunately, given the size and location of many iCCA tumors, achieving an R0 margin can often be challenging. In fact, in one review of 583 patients with iCCA who were treated at one of 12 major hepatopancreatobiliary centers, one in six had an R1 resection.³⁰ Patients with an R1 resection had a higher risk of recurrence and shorter OS versus patients with an R0 resection; a wider R0 margin (5–9 mm vs. 1–4 mm) was associated with improved RFS and OS. The impact of margin status on long-term outcome is likely confounded by the finding that patients who undergo an R1 resection are more likely to have worse tumor biology (i.e., larger, bilateral tumors, perineural invasion, etc.). To this point, a retrospective study of 1105 patients with iCCA from the International Intrahepatic CCA Study Group database evaluated the prognostic significance of margin status relative to overall tumor burden. Of note, patients who had low or medium tumor burden had better survival as the margin width increased; however, patients who had high tumor burden did not derive the same survival benefit relative to surgical margin status.³¹ Ultimately, an R0 resection margin likely cannot overcome poor tumor biology. The current patient had both a high tumor burden as well as close surgical margins on final pathology.

In addition to primary tumor considerations, evaluation of the nodal basin has been a topic of much interest in the treatment of iCCA. Lymph node status is an important long-term prognostic factor for patients with iCCA.³² Specifically, the presence of lymph node metastases not only adversely affects OS but also influences the relative effect of tumor number and vascular invasion on survival.³³ Therefore, assessment of the nodal basin should be performed at surgery. Lymphadenectomy should involve areas including the periportal lymph nodes (station 12) as well as other nodal stations, depending on the location of the lesion in the liver.³⁴ Of note, patients who have lymph node metastases beyond station 12 have worse OS versus individuals who have lymph node metastases confined only to station 12.³⁵ The National Comprehensive Cancer Network Guidelines and the eighth edition of the American Joint Committee on Cancer AJCC Cancer Staging Manual both recommend a lymphadenectomy with at least six lymph nodes evaluated to accurately stage patients with iCCA.^{36, 37} Multiple studies have demonstrated that only about one half of patients who undergo resection for iCCA have pathologic examination of even one lymph node.³⁸ In addition, only about 15% of patients have the recommended six nodes identified on final pathologic assessment.³⁹

Although surgical resection was performed in the current patient, another surgical option to treat patients with advanced iCCA is the placement of an HAIP. HAIP has been associated with a partial response in 27%–59% of patients and disease stabilization in 40%–73% of patients.⁴⁰ In one retrospective study of patients with multifocal iCCA, surgical resection was compared with intra-arterial therapy (HAIP, transarterial chemoembolization, or transarterial radioembolization).⁴¹ There was no difference in OS or PFS among patients who had surgery versus transarterial therapy; a subgroup analysis did demonstrate, however, that patients who had HAIP therapy had improved OS and PFS versus those who underwent only surgery (OS, 39 vs. 20 months, PFS, 9 vs. 5 months, respectively).⁴¹ A recent phase 2 trial from Memorial Sloan Kettering Cancer Center (ClinicalTrials.gov identifier NCT01862315) evaluated the use of HAIP with floxuridine and gemcitabine/oxaliplatin in 38 patients with unresectable iCCA.⁴² Overall, 58% of patients achieved an objective radiographic response, 84% achieved disease control, and four patients had a marked response with downstaging to allow for resection.

Unfortunately, iCCA is an aggressive cancer that is often diagnosed at a late stage. Despite curative-intent resection, patients often suffer from a high risk of recurrence and poor survival. Management of iCCA is complex and requires coordination and collaboration among medical, surgical, and radiation oncology teams. In general, patients who present with resectable disease undergo curative-intent resection, including margin-negative resection and lymphadenectomy, with strong consideration of adjuvant capecitabine based on the BILCAP trial. Patients with locally advanced/unresectable or metastatic disease should receive systemic therapy (often with gemcitabine and cisplatin with or without durvalumab) and also should be considered for local liver-directed therapies in the setting of no extrahepatic disease. An emerging understanding of the molecular underpinnings of iCCA will continue to facilitate a more targeted approach to treating this disease.

The authors declare no conflicts of interest.