Methods
Patients
This was a single-center retrospective study conducted from January 2008 to December 2020. The medical files of all consecutive patients enrolled in this period with a diagnosis of HCC or iCCA (recorded from 2010 onward) were considered for the study. Data were collected prospectively through an internal database and analyzed retrospectively. Our workup included systematic assessment to detect underlying liver disease for HCC or iCCA patients, supported by noninvasive measurement of liver stiffness by transient elastography and controlled attenuation parameters. The diagnosis of HCC was by radiology following international guidelines8 or on histology in the absence of formal radiological criteria and in the absence of cirrhosis. HCC patients were classified according to the Barcelona Clinic Liver Cancer (BCLC) staging system. Those with a single nodule of >50 mm were designated as having BCLC stage AB disease. The treatment strategy was discussed during multidisciplinary concertation sessions, with recommendations based on the National Thesaurus of Digestive Cancerology (TNCD). Early tumors were treated with a curative approach or according to the stage migration concept.8 Advanced HCCs with sectorial portal thrombosis were sometimes treated with an intra-arterial procedure such as transarterial chemoembolization (TACE) based on expert recommendations.9 HCC patients who had transplantation as the first treatment were not included.
The diagnosis of iCCA was based on histology, and tumors were classified according to the eighth American Joint Committee on Cancer Staging System.10 Patients with a diagnosis of combined hepatocellular-cholangiocarcinoma were excluded. The treatment options, especially iCCA resection, were discussed during multidisciplinary concertation sessions in accordance with the National Thesaurus of Digestive Cancerology recommendations at that time. Advanced iCCA was treated with systemic chemotherapy, a cisplatin/gemcitabine (CisGem) regimen,11 or a combination of oxaliplatin, 5-FU and LV (FOLFOX) according to patient performance status (PS 0/1). If the PS was 2, the patient was treated with gemcitabine alone. Patients with unresectable iCCA without extrahepatic disease were sometimes treated with locoregional therapies [TACE or selective internal radiotherapy (SIRT)] in combination with systemic chemotherapy.
As part of this study, a histological review was performed to classify iCCA cases into the two main histopathological subtypes according to the size of the affected duct: small bile duct (SD), or large bile duct (mucinous) (LD). A radiological review was also performed on all iCCA cases, especially in histologically indeterminate iCCAs, to classify them according to gross appearance into the mass forming (MF) (mostly associated with SD iCCA) or periductal infiltrative (PDI) (+/− MF) subtype; the PDI subtype is exclusively seen with LD type iCCA. In addition to immunohistochemical study of tumor tissue to evaluate the protein expression of human epidermal growth factor receptor 2 (HER2) and to investigate the mismatch repair phenotype, we performed molecular profiling of iCCA patients to detect oncogenic molecular alterations targetable by therapies after gaining access to a molecular genetics platform in 2020. The platform employed tumor DNA and RNA targeted sequencing panels to detect activating mutations and chromosomal rearrangements, respectively. The Oncomine Focus Assay (Thermo Fisher Scientific, Waltham, MA, USA) was used to conduct concurrent DNA and RNA next-generation sequencing analysis from formalin-fixed paraffin-embedded samples, targeting 77 genes (mutations in 45 genes including BRAF, CDKN2A, EGFR, ERBB3, FGFR1, 2, 3, IDH1/2, KRAS, MET, NTRK1, 2, PIK3CA, PTEN, RAF-1, RET, TP53; fusions in 18 genes including EGFR, FGFR1, 2, 3, MET, NTRK1, 2, 3; copy number variations in 14 genes including EGFR, ERBB2 et 3, FGFR1, 2 et 3, KRAS, MET), and 15 genes that have various roles in the homologous DNA damage repair pathway (Oncomine BRCA Expanded panel). The current study was approved by the ethics committees of our institution. It followed the Good Clinical Practice guidelines and was conducted following the ethical principles of the Declaration of Helsinki.
Statistical analysis
Quantitative data were reported using the mean and standard deviation (SD); qualitative data were reported using the frequency and percentage. Quantitative data were compared between groups using Student’s t test for normally distributed data or the nonparametric Wilcoxon test otherwise; the chi-squared test or Fisher’s test was used for comparison of qualitative data. The Mantel-Haenszel chi-squared test was performed to compare ordinal scale data. Risk factors for HCC or iCCA were analyzed by univariate logistic regression analysis prior to multivariate logistic regression analysis. Items that were identified as significant in the univariate analysis were included in the multivariate model analysis. The multivariate results were reported using odds ratios (ORs) and 95% confidence intervals (CIs). Overall survival (OS) was defined as the time interval between the diagnosis of cancer and death or the time of last follow-up for patients who were still alive. Survival was compared between groups using the log-rank test. OS results were reported using median and interquartile range (q1, q3) and hazard ratios (HRs) and 95% CIs. All p-values were considered significant at α-level=0.05. All calculations were performed using SAS V9.4 statistical software (SAS Institute Inc., Cary, NC, USA).
Discussion
In this retrospective study from a liver unit, a comparative analysis found differences and similarities among HCC and iCCA patients. We found a higher proportion of HCC patients in this French cohort of patients with primary hepatobiliary tumors. Most patients with HCC were detected as opposed to iCCA patients. However, this trend may change over time, as the main risk factor for iCCA in this series was chronic liver disease with cirrhosis, which was found in more than one-third of the study participants.
iCCAs represent approximately 15% of primary liver cancers.1 The results of this study were in line with that finding, and other features associated with iCCA, such as female sex, multinodularity, and the presence of metastases, were identified, as in other series.4,13 The overrepresentation of women in the iCCA group compared to the HCC group in our series was found in iCCA patients without cirrhosis. Female sex was independently associated with iCCA diagnosis in a study by Lee et al.4 We also found an overrepresentation of females in iCCA in phase 2 and 3 trials evaluating targeted therapies for iCCA patients with specific genomic alterations (ClarIDHy14 (ivosidenib): 65%, FIGHT-20215 (pemigatinib): 58%, NCT0215096716 (infigratinib): 57%) and in the phase 3 TOPAZ-1 study17 (50%), in which 55% of the patients had iCCA. However, we found a similar sex distribution among cirrhotic patients in the HCC and iCCA groups, and there was a male predominance. The result is not surprising, as the risk factors associated with cirrhosis (notably alcohol) affect more men. Multinodularity and extrahepatic spread are classically associated with iCCA.4 Indeed, iCCA is characterized by lack of a capsule and a significant fibrous stroma,18 with effector cells such as tumor-associated macrophages,19 and activated myofibroblasts20 that are driven by cytokines and growth factors such as transforming growth factor-beta (TGF-β) secreted by tumor cells.
In our European series, we found that a large majority of HCC patients had cirrhosis. More than 30% of patients in the iCCA group had cirrhosis, which is a well-documented risk factor in that pathology.1,2 The finding is consistent with other studies.4,21 In an Italian multicenter series from expert centers, nearly half of the patients with iCCA (46%) had cirrhosis, and most of them were detected.22 Thus, there may have been an underestimation.21 Conversely, in phase 2/3 trials evaluating targeted therapies or the combination of chemotherapy plus immunotherapy as treatment for advanced iCCA, there have been few14 or no cirrhotic patients, or data on cirrhosis have not been available.15-17 The association of iCCA with cirrhosis raises several questions. Cirrhotic patients can develop either cancer, which highlights the absolute necessity of histological assessment if the radiological criteria for HCC are not fulfilled.23 Histological confirmation is even more important given the potential identification of therapeutic targets in cholangiocarcinoma (Table 6). Cirrhosis was the main risk factor for iCCA in our series, and as previously reported,24 cirrhosis, chronic hepatitis, alcohol use, and NASH were not associated with the iCCA subtype. The association of iCCA development with cirrhosis may allow earlier detection of this poor prognosis cancer13 and identification of a target population that can benefit from curative therapeutic options.25 Moreover, in previous studies, cirrhosis did not affect the prognosis of patients with iCCA.21,22 These results are not surprising, as liver function and tumor features (tumor size and metastatic spread) were comparable between our two iCCA subgroups with or without cirrhosis. Of course, these results should be interpreted with caution given the small sample size.
Our comparative analysis of HCC and iCCA found similar underlying liver diseases among cirrhotic patients. While cirrhotic patients are at risk of developing these two cancers, this result highlights the close link between these two entities and the importance of chronic inflammation related to hepatitis.1 Common nucleotide substitutions in HCC and iCCA related to chronic hepatitis have been described.26 The same study showed a similar gene expression profile (RNA-seq analysis) of biliary cancers and poorly differentiated HCC. The result suggests that the diseases involve activation of different oncogenic pathways but may have common cells of origin, although the last point remains controversial.5 Indeed, biliary tree cells harbor different cell types, including hepatocytes, liver progenitor cells (which have a biphenotypic orientation), and biliary epithelial cells (mature nonmucin-producing interlobular cholangiocytes, and cylindrical mucin-producing cholangiocytes located in LDs). In addition to this cellular diversity, mature hepatocytes maintain phenotypic plasticity and thus an ability to differentiate into cholangiocytes. Activation of the Notch pathway or Ras/MAPK cascade and Tp53 mutations has been implicated in the conversion of normal hepatocytes into malignant cholangiocytes.27,28 Thus, the similarities and heterogeneity of hepatobiliary tumors are substantial challenges that need to be considered.5 In addition to iCCA and HCC, there are other rare liver cancers with biliary phenotypes.29
Nearly 20% of the HCC patients in our series had nonalcoholic fatty liver disease (NAFLD), and nearly one-third in the iCCA group had NAFLD. The results are not unexpected since these diseases have been recognized as major causes of chronic liver disease, the incidence of which is increasing.30 Steatosis may be complicated by necrotic-inflammatory processes, which characterizes patients with NASH. Previous studies have shown a change in HCC tumor phenotype after TACE with the development of hepatobiliary cancers, suggesting the importance of necrosis in this process.31 Necroptosis, which is programmed cell necrosis resulting in plasma membrane disruption following osmotic shock, appears to be particularly important in NAFLD and NASH.32 Cells undergoing necroptosis release damage-associated molecular patterns that may shape the microenvironment through specific cytokine delivery by immune cells. A recent study underlines the influence of the microenvironment and necroptosis in biliary or hepatic cancers related to singular epigenetic regulators.33 Thus, specific oncogenes and the tumor microenvironment drive the liver cancer phenotype.
In line with other studies,4 our study found that therapeutic strategies differ between iCCA and HCC. A curative approach is more frequently used for HCC patients because one majority is detected. Some experts consider surgery the most effective treatment modality to achieve the best survival rate in HCC cases up to intermediate stage.34 Moreover, the latest version of the BCLC staging system no longer recommends TACE as the main modality for intermediate-stage disease treatment.35 In our series, stratification according to the BCLC system found tumors with different prognoses. However, BCLC stage AB HCC had a comparable prognosis with BCLC stage 0/A HCC, in contrast with the findings of other studies,36 probably owing to the different treatment modalities within each group. The poor OS seen with sorafenib treatment is comparable to that seen in other real-life cohorts,37 as the populations of patients have differed in trials.38 Moreover, there was no survival benefit following the use of any endovascular procedure, mainly TACE, as an alternative therapeutic option for advanced HCC. These results should be interpreted with caution in the absence of comparison using propensity score matching. Conversely, survival was longer in the sorafenib group treated with second-line TKI therapy, and survival with sorafenib would likely be better over time.38 Once again, the results should be interpreted with caution in the absence of comparison using propensity score matching; however, sequencing of systemic therapies can provide a survival benefit for HCC patients,39 although this strategy can be used in relatively few patients in the TKI era.40
Regarding the iCCA population, our series did not find a significant difference in prognosis between patients with SD iCCA and those with LD iCCA across all treatments despite there being more impaired liver function in the LD group. However, more than one-third of patients with LD iCCA were ineligible for specific treatment. Studies have shown that these two entities differ radiologically and histologically6 and regarding molecular alterations5 and treatment response.41 The LD iCCA subtype is characterized by higher rates of desmoplastic stroma and higher frequencies of TGF-β1 pathway gene alterations and Tp53 and KRAS mutations.41 LD iCCAs appear to have lower sensitivity to chemotherapy than SD iCCAs and they also reportedly show resistance to anti-PD1/PD-L1 immunotherapy41 related to the suppressive immune microenvironment in KRAS-altered tumors. Given the results of the TOPAZ-1 trial,17 which supports the combination of chemotherapy (CisGem regimen) plus anti-programmed death ligand 1 (PDL1) therapy (durvalumab), the management strategy for advanced iCCAs is about to change. This new combination may overcome such resistance. Furthermore, classification according to iCCA subtype is necessary to improve outcomes.
The difference in survival between HCC and iCCA patients is not unexpected and probably reflects the absence of screening in the iCCA group, which had a higher tumor burden than the HCC group (as has been previously reported), along with the difference in follow-up time. The difference was also likely related to the difference in curative approach rate in our series, as more patients in the HCC group were treated with curative strategies (32% vs. 20%). Moreover, the overall prognosis of advanced stage iCCA and HCC remains poor, together with comparable proportions of patients eligible for only palliative care. The situation is even more serious, as pointed out in a recent study conducted in all French hospitals, since most iCCA patients diagnosed during the period 2014–2015 only received supportive care.13
Chemotherapy with the CisGem regimen has been the only therapeutic approach for advanced biliary tract cancer (BTC) for many years and has produced modest results,11 as reflected by our series. Intensification of chemotherapy has not shown any benefit,42 and the use of new cytotoxics is under investigation. There is significant improvement with the combination of chemotherapy and immunotherapy.17 Moreover, the prognosis of BTC, especially iCCA, is likely to change thanks to personalized therapeutic approaches. In our study, 50% of the patients in whom genomic and transcriptomic studies were performed had a targetable molecular alteration (Table 6). Large-scale sequencing technologies have highlighted BTC heterogeneity along with various oncogenic alterations that may be targeted by therapies. These molecular alterations (mutations, rearrangements, and amplifications) are diverse and affect many cellular processes.5 The most commonly affected genes are those encoding the isocitrate dehydrogenase 1 (IDH1) and IDH2 enzymes (the mutations of which are mutually exclusive) involved in DNA repair mechanisms and epigenetic regulation and those encoding fibroblast growth factors, which are associated with the SD iCCA subtype (mutations seen in approximately <5–20% of patients). These aberrations also less commonly affect pathway kinase genes (BRAF, MET, and ERBB2), chromatin remodeling genes (ARID1A) and mismatch repair genes (MLH1 and MSH2, or the germline mutations of BRCA1 and BRCA2 that we identified in our patients). Several targeted therapies are now approved by the Food and Drug Administration and European Medicines Agency for patients with cholangiocarcinoma who harbor specific genomic alterations. The prescription of these agents is now facilitated and prioritized by the European Medical Oncology Society (ESMO) Scale for Clinical Actionability of Molecular Targets (ESCAT) system.43 Thus, tumor genetic testing should be performed as soon as the first systemic treatment is given and in the case of failure or progression.
This study suffers from limitations related to several factors: (1) It was a retrospective study, and we tried to reduce the relevant limitations by prospectively accessing data from our regular multidisciplinary collegial sessions. (2) We used histological review to classify the iCCA subtypes, as most of the biopsies were tumor biopsies and not surgical specimens. (3) There was a lack of data regarding the combination of anti-PD-L1 therapy (atezolizumab) plus anti-VEGF therapy (bevacizumab), which is the new standard of care for advanced HCC.44 The combination of atezolizumab and bevacizumab was not available in France until 2020. 4) Finally, there was a lack of propensity score matching for some comparisons among HCC patients.