Introduction

As a rare form of primary liver cancer, combined hepatocellular cholangiocarcinoma (cHCC-CCA) is a mixture of the main constituents of hepatocellular carcinoma (HCC) and cholangiocarcinoma (CCA) [1]. The distinction between cHCC-CCA and HCC or CCA is largely dependent on the histologic components of these tumors. cHCC-CCA is receiving increasing attention in view of its unique histopathology and biological behavior, as well as the dilemmas in its diagnosis [2], and the 2019 WHO classification [3] of cHCC-CCA was updated with distinct changes from the 2010 classification. Although the imaging features of cHCC-CCA may be similar to those of HCC and/or CCA [4], imaging-based diagnosis of cHCC-CCA is always challenging because of its heterogeneous characteristics and features that overlap with those of HCC and CCA [5]. Currently, surgical resection is the curative treatment for cHCC-CCA, and the effectiveness of other treatments is still under debate [6]. Imaging misdiagnosis of HCC without histopathological confirmation can result in inappropriate treatments for cHCC-CCA [7]. All these factors may be problematic in treatment decision-making and in investigating the prognosis of patients with cHCC-CCA.

Alpha fetoprotein (AFP) and carbohydrate antigen 19-9 (CA19-9) are widely accepted as biomarkers for HCC and CCA [8, 9]. It has been suggested that AFP or CA19-9 can be utilized as an independent indicator of prognosis in HCC or CCA patients after curative treatment [10, 11]. However, the clinical value of either AFP or CA19-9 in patients with cHCC-CCA remains unclear. Therefore, the purpose of this study was to evaluate the clinical significance of AFP and CA19-9 in the context of imaging features in patients with cHCC-CCA.

Recently, more attention has been given to the substantial roles of clinicopathological features as prognostic indicators related to recurrence-free survival (RFS) and overall survival (OS) in patients with cHCC-CCA. A recent study reported that cHCC-CCA patients with normal AFP and CA19-9 levels had better outcomes, while elevated AFP and/or CA19-9, microvascular invasion (MVI), local extrahepatic invasion, and lymph node metastasis were independent predictors of poor long-term survival [12]. In addition, cHCC-CCAs with MVI might have poorer surgical outcomes with respect to early recurrence [13]. Therefore, a comprehensive analysis of prognostic factors for RFS and OS in relation to the clinicopathological and imaging characteristics in patients with cHCC-CCA was performed.

Materials and Methods

Patient selection

This retrospective study was approved by the Institutional Review Board of Zhongshan Hospital, Fudan University, and a written consent form was obtained from each subject before enrollment. In total, 187 consecutive patients with cHCC-CCA were diagnosed by postoperative pathology between June 2016 and November 2020. Of these, 26 patients were excluded because the category of cHCC-CCA with stem cell features is no longer used according to the updated 2019 WHO classification system. One additional patient was excluded due to the lack of an MRI scan within 30 days before hepatectomy. Finally, 160 patients with pathologically confirmed of cHCC-CCA were enrolled in this study. The flowchart of patient enrollment is provided in Fig. 1.

Fig. 1
figure 1

Flowchart of this study cohort. cHCC-CCA = combined hepatocellular carcinoma-cholangiocarcinoma, AFP = alpha fetoprotein, CA19-9 = carbohydrate antigen 19-9

Full size image

This study had some overlap with a previous study [13] in terms of patients. In the prior study, 113 patients with cHCC-CCA were included to predict MVI by conventional MRI findings. However, the current study enrolled cHCC-CCA patients confirmed by the updated 2019 WHO classification system, and investigated the clinical significance of AFP and CA19-9 in the context of imaging features and comprehensively the prognostic factors for RFS and OS in relation to the clinicopathological and imaging characteristics.

Clinical and pathological data evaluation

All clinical information of the cHCC-CCA patients, including age, sex, hepatitis B virus infection status, and levels of tumor markers such as serum AFP and CA19-9 within 7 days before curative resection, was retrospectively collected from the medical records. The cutoff values for AFP and CA19-9 were 20 ng/ml and 37 U/ml, respectively. The pathologic findings of the lesions included the number of tumors (single or multiple), tumor size (≤ 2 cm, 2–5 cm, ≥ 5 cm), tumor location (left, right, or caudate), proliferating antigen Ki-67, MVI, and lymph node metastasis. AFP and CA19-9 are widely considered to be biomarkers for HCC or CCA, but the clinical value of AFP and CA19-9 remains undefined in patients with cHCC-CCA. Therefore, all patients were divided into groups according to their AFP (AFP > 20 ng/ml and AFP ≤ 20 ng/ml) or CA19-9 (CA19-9 > 37 U/ml and CA19-9 ≤ 37 U/ml) levels. Then, the differences in clinical and pathological data were compared between the AFP groups or CA19-9 groups (Table 1).

Table 1 Clinical and pathological characteristics of patients with cHCC-CCA
Full size table

MRI

All patients were scanned with a 24-channel 1.5-T MR scanner (uMR 560, United Imaging Healthcare). Routine liver protocols consisted of transverse T2-weighted imaging (T2WI), T1WI, in-phase and opposed-phase sequences, and diffusion-weighted imaging (DWI, b value = 0, 50, and 500 s/mm2). Dynamic imaging was performed with a T1-weighted fat-suppressed sequence. The arterial phase was acquired when the contrast agent (gadolinium diethylenetriamine pentaacetic acid, Gd-DTPA; Magnevist, Bayer HealthCare) reached the ascending aorta after intravenous administration at a dose of 0.1 mmol/kg at a rate of 2 ml/s. The portal venous phase and delayed phase sequences were acquired at 70–90 s and 160–180 s, respectively. All the sequences with the detailed parameters are shown in a published paper [13].

Image feature interpretation

All MR images were retrospectively evaluated by two radiologists (C.W.Z. and C.Y., who have 12 and 14 years of experience in abdominal imaging, respectively). They were blinded to the clinical data, tumor markers, and pathological results but were aware that the patients had cHCC-CCA. A consensus was negotiated when there was disagreement between the two observers.

The following imaging characteristics of cHCC-CCAs were investigated on precontrast MR images: (a) the shape of the tumor (globular, lobulated, or irregular), (b) the definition of the tumor margin (smooth or nonsmooth), (c) intratumoral hemorrhage, (d) intratumoral fat deposition, (e) peritumoral bile duct dilatation, (f) hepatic capsule retraction, (g) restricted diffusion, and (h) target sign on DWI. In addition, the following dynamic enhancement features were assessed: (A) arterial phase: (a) global enhancement (defined as non-peripheral enhancement and enhancement area ≥ 50 %) or peripheral enhancement (defined as without central enhancement and enhancement area < 50%, including rim enhancement), and (b) arterial phase peritumoral enhancement (presence or absence, defined as enhancement surrounding the tumor border); (B) portal venous phase: (c) washout appearance (non-peripheral washout or peripheral washout), (d) enhancing capsule (presence or absence), (e) portal vein thrombus (presence or absence), and (f) corona enhancement (presence or absence, defined as enhancement adjacent to the tumor border); (C) delayed phase: (g) delayed enhancement (presence or absence); and (D) other imaging features: (h) nodule-in-nodule architecture (presence or absence) and (i) mosaic architecture (presence or absence) (Figs. 2 and 3). Most imaging features were classified into three categories: those favoring HCC features, those favoring CCA features, and those favoring malignancy features. In addition, comparisons of imaging features were also performed between the AFP groups (AFP > 20 ng/ml and AFP ≤ 20 ng/ml) and the CA19-9 groups (CA19-9 > 37 U/ml and CA19-9 ≤ 37 U/ml).

Fig. 2
figure 2

Images of a 58-year-old male patient with cHCC-CCA. There is an irregular tumor in the right lobe of the liver showing a target sign with peripheral hyperintensity and central hypointensity on diffusion-weighted imaging (a). T1-weighted imaging (b) shows heterogeneous hypointensity and hepatic capsule retraction (arrow). It presents arterial phase peripheral enhancement (arrow) and peritumoral enhancement (arrowhead) on contrast-enhanced T1-weighted imaging (c) with contrast agent Gd-DTPA, and peripheral washout appearance (arrows) on portal venous phase (d). Multiple recurrence lesions are found 15 months after surgery in this patient

Full size image
Fig. 3
figure 3

Images of a 62-year-old female patient with cHCC-CCA. There is an irregular tumor in the right lobe of the liver showing heterogeneous hyperintensity and peritumoral bile duct dilatation (arrow) on T2-weighted imaging (a). T1-weighted imaging (b) shows intratumoral hemorrhage (arrow). Arterial phase global enhancement (arrow) on contrast-enhanced T1-weighted imaging (c) with contrast agent Gd-DTPA and non-peripheral washout appearance on portal venous phase (d) is presented on this lesion

Full size image

Follow-up for RFS and OS determination after curative resection

One hundred and twenty-seven patients were followed up after initial hepatectomy. Recurrence was defined as intrahepatic or extrahepatic neoplasms observed by CT/MRI, positron emission tomography (PET)-CT, or pathological confirmation. The time to recurrence (TTR) and OS time were applied to evaluate the RFS and OS of the patients with cHCC-CCA. The differences in TTR and OS time were compared between pairs of subgroups according to clinical, pathological, and imaging features. Moreover, risk factors for RFS and OS in patients with cHCC-CCA after resection were also comprehensively investigated.

Statistical analysis

Statistical analyses were performed by using SPSS 26.0 (IBM). Data with a normal distribution are presented as the mean ± standard deviation, and the differences between the two groups were compared by using an independent sample t test. Data with a nonnormal distribution are expressed as the median and interquartile range, and the differences were compared by using the rank sum test. Moreover, categorical variables are shown as the number of cases and the percentage, and comparisons between these groups were performed by using the chi-square test or Fisher’s exact test. Kaplan-Meier survival curves and log-rank tests are utilized to compare the differences in RFS and OS after hepatectomy with in patients grouped according to clinical, pathological, and imaging features. Cox regression analyses are used to identify the independent prognostic factors, and the results are presented as hazard ratios with 95% confidence intervals (95% CIs). Differences with a p value of < 0.05 were considered statistically significant.

Results

Clinicopathological features of the study patients

One hundred and sixty patients with cHCC-CCA were enrolled in this study, with an average age of 55.5 ± 11.1 years old, and 124 (77.5%) patients were male. There were 89 patients with AFP > 20 ng/ml and 36 patients with CA19-9 > 37 U/ml. The comparisons of clinicopathological features in patients stratified by AFP or CA19-9 are detailed in Table 1. There were significant differences in age (54.0 ± 12.2 vs. 57.4 ± 9.3, p = 0.048), MVI (47.2% vs. 29.6%, p = 0.02), and tumor size (≤ 2 cm: 19.1% vs. 35.2%; ≥ 5 cm: 42.7% vs. 25.4%, p = 0.03) between the AFP > 20 ng/ml and AFP ≤ 20 ng/ml groups.

There was also a significant difference in lymph node metastasis (33.3% vs. 14.5%, p = 0.01) between the CA19-9 > 37 U/ml and CA19-9 ≤ 37 U/ml groups. No significant differences were found in sex, hepatic B virus infection, number of tumors, Ki-67, or tumor location (all p > 0.05) between the AFP groups or CA19-9 groups.

MRI characteristics of cHCC-CCAs

The comparisons of MRI characteristics of cHCC-CCAs are described in Table 2. The presence of some imaging features, such as nodule-in-nodule architecture (12.4% vs. 7.0%, p = 0.04), mosaic architecture (50.6% vs. 29.6%, p = 0.01), intratumoral hemorrhage (23.6% vs. 11.3%, p = 0.04), hepatic capsule retraction (32.6% vs. 9.9%, p = 0.001), arterial phase peritumoral enhancement (44.9% vs. 29.6%, p = 0.047), and portal vein thrombus (23.6% vs. 7.0%, p = 0.01), was significantly higher in the AFP > 20 ng/ml than in the AFP ≤ 20 ng/ml group. Unfortunately, none of the imaging characteristics were significantly different between the CA19-9 groups.

Table 2 MR Imaging features of cHCC-CCAs
Full size table

Prognostic factors for RFS in patients with cHCC-CCA

Of 127 patients followed up, 71 experienced recurrences after curative treatment during the observation period. The cumulative recurrence rates at 1, 3, and 5 years were 40.9%, 54.3%, and 55.9%, respectively. The median TTR was 15 (6–25) months. The survival analysis showed that CA19-9 > 37 U/ml (p = 0.03), arterial phase peripheral enhancement (p = 0.048), intratumoral hemorrhage (p < 0.001), mosaic architecture (p = 0.003), arterial phase peritumoral enhancement (p = 0.002), peritumoral bile duct dilatation (p = 0.03), hepatic capsule retraction (p = 0.01), and portal vein thrombus (p = 0.01) were predictors of poor RFS (Figs. 4 and 5). Further, the multivariate Cox regression analysis revealed that age (HR 1.031, p = 0.03), CA19-9 > 37 U/ml (HR 1.880, p = 0.04), arterial phase peritumoral enhancement (HR 2.287, p = 0.01), and delayed enhancement (HR 0.377, p = 0.02) were independent predictors of poor RFS in patients with cHCC-CCA (Table 3).

Fig. 4
figure 4

Kaplan-Meier survival curves showing the prognostic value of CA19-9 > 37 U/ml (a), intratumoral hemorrhage (b), arterial phase peripheral enhancement (c), and arterial phase peritumoral enhancement (d) for the recurrence-free survival of patients with cHCC-CCA after surgery

Full size image
Fig. 5
figure 5

Kaplan-Meier survival curves showing the prognostic value of portal vain thrombus (a), peritumoral bile duct dilatation (b), hepatic capsule retraction (c), and mosaic architecture (d) for the recurrence-free survival of patients with cHCC-CCA after surgery

Full size image
Table 3 Cox regression analysis of predictors for RFS and OS of patients with cHCC-CCA
Full size table

Prognostic factors for OS in patients with cHCC-CCA

In contrast, the 1-, 3-, and 5-year OS rates of the cHCC-CCA patients were 89.8%, 70.1%, and 66.9%, respectively. The median OS time was 25 (17–37) months. The survival analysis indicated that CA19-9 > 37 U/ml (p = 0.03), lymph node metastasis (p = 0.01), arterial phase peripheral enhancement (p = 0.002), mosaic architecture (p = 0.002), arterial phase peritumoral enhancement (p = 0.01), and portal vein thrombus (p = 0.04) were poor prognostic factors for OS (Fig. 6). Nevertheless, the multivariate Cox analysis identified arterial phase peripheral enhancement (HR: 2.391, p = 0.04) as a significant determinant for poor OS in cHCC-CCA patients (Table 3).

Fig. 6
figure 6

Kaplan-Meier survival curves showing the prognostic value of arterial phase peripheral enhancement (a), arterial phase peritumoral enhancement (b), portal vain thrombus (c), and mosaic architecture (d) for the cumulative overall survival of cHCC-CCA patients after surgery

Full size image

Discussion

Our study showed that neither HCC-favoring features (arterial phase global enhancement, non-peripheral washout, and capsular enhancement) nor CCA-favoring features (arterial phase peripheral enhancement, peripheral washout, and delayed enhancement) were significantly associated with elevated AFP or CA19-9. However, we also found that some characteristics (age, CA19-9 > 37 U/ml, arterial phase peritumoral enhancement, delayed enhancement, and arterial phase peripheral enhancement) were independent predictors of poor RFS or OS in cHCC-CCA patients.

In the present study, the incidence of MVI in the elevated AFP group was significantly higher than that in the normal AFP group (47.2% vs. 29.6%, p = 0.02). Similar to our study, a previous study revealed that a high AFP level (OR = 0.523, p = 0.04) was a potential predictive indicator of MVI in cHCC-CCA patients [13]. We also found that the frequency of tumor size ≥ 5 cm (42.7% vs. 25.4%) was higher in the group with elevated AFP levels. Some other researchers also indicated that elevated AFP and/or CA19-9 were independent poor prognostic factors in the tumor size ≥ 5 cm subgroup of cHCC-CCA patients [12]. Interestingly, it has been reported that high CA19-9 is significantly related to MVI (p = 0.01) and that elevated CA19-9 is an independent risk factor for lymph node metastasis (RR = 7.064, p < 0.001) in CCA [14, 15], but we detected that only the difference in lymph node metastasis between elevated and normal CA19-9 in cHCC-CCA patients was statistically significant. This may be explained by the mixture of disease elements in cHCC-CCA.

AFP and CA19-9 are well known as predictive risk biomarkers for HCC or CCA. In this study, the purpose of dividing patients with cHCC-CCA into two groups based on AFP or CA19-9 was to determine whether the HCC-favoring imaging features were present at a significantly higher frequency in the elevated AFP group and whether the CCA-favoring imaging characteristics were present at a significantly higher frequency in the elevated CA19-9 group, but the results showed that the patients with high AFP only displayed some ancillary features of HCC (such as nodule-in-nodule architecture, mosaic architecture, and intratumoral hemorrhage) and even some features of CCA (such as hepatic capsule retraction and arterial phase peritumoral enhancement). Unfortunately, none of the imaging characteristics were significantly different between the CA19-9 groups. Likewise, Shetty AS et al [4] illustrated that cHCC-CCAs manifested various imaging features that might be similar to those of HCC and/or CCA. cHCC-CCA is a biphenotypic tumor with a heterogenous mixture of both hepatocytic and cholangiocytic characteristics [16]. Therefore, the imaging findings of cHCC-CCA are complex and diverse and are not correlated with the AFP or CA19-9 levels.

The median RFS and OS were 15 and 25 months, respectively, which were highly consistent with the previous results (14.5 and 24.9 months) reported by Jiang et al [12]. We found that CA19-9 > 37 U/ml was an independent prognostic indicator for unfavorable RFS, but previous studies reported that CA19-9 could not independently predict the long-term prognosis of cHCC-CCA patients [17, 18]. In addition, our study showed that arterial phase peritumoral enhancement was an independent predictor of poor RFS. Arterial phase peritumoral enhancement presents compensatory hepatic arterial hyperperfusion surrounding the tumor due to portal branch microthrombosis [19]. Wang et al [13] suggested that irregular arterial peritumoral enhancement (OR 0.322, p = 0.001) was a predictor for MVI in cHCC-CCA, but not an independent prognostic factor for RFS. Furthermore, we also found that delayed enhancement was a significant determinant of RFS in cHCC-CCA patients. Delayed enhancement is known to be related to the fibrotic stroma in hepatic tumors, and delayed enhancement might be a prognostic factor in CCA [20]. Asayama et al [21] found that CCA patients with more than two-thirds delayed enhancement had a poorer postsurgical outcome than those with a small area of delayed enhancement.

In this study, arterial phase peripheral enhancement was identified as an independent predictor of poor OS. Similarly, Park et al [22] showed that the arterial phase global enhancement was correlated with better OS after surgery due to lesser amount of fibrotic stroma in cHCC-CCAs. However, intratumoral hemorrhage has been demonstrated to be a malignant feature of HCC because of being involved in tumor growth and metastasis [23]. Hu et al [24] suggested that intratumoral hemorrhage might be an independent poor indicator in HCC, but this was not observed in our study. Analyzing the reasons, we found that the incidence of intratumoral hemorrhage in HCCs was 44.1% (60/136) in Hu’s study, while the prevalence of intratumoral hemorrhage in cHCC-CCAs was only 18.1% (29/160) in our study. This may be related to the significantly higher incidence of intratumoral hemorrhage in HCCs than cHCC-CCAs. In addition, as there are few studies on cHCC-CCA, some results are based on HCC.

There are several limitations to this study. First, it is a single-center and retrospective study. It unavoidably has some selection bias and influencing factors. Therefore, we applied use multivariate analysis to make the results as objective as possible, but our results need to be validated in larger multicenter prospective studies. Second, the contrast agent Gd-DTPA was used for MRI in the present study, and some imaging features in the transitional phase (TP) and hepatobiliary phase (HBP) were not incorporated into this evaluation. Third, the 2019 WHO classification of cHCC-CCA has been updated, and the most significant change is that the category of stem cell features is no longer used. Kim et al [25] believed that there might be a few impacts on imaging-based diagnoses of cHCC-CCA according to the 2019 WHO classification system. However, 26 patients with stem cell features were excluded, but the enrolled patients were not classified in accordance with the latest system due to the lack of relevant pathological data. Finally, the probably malignant, not specific for HCC (LR-M) features defined in the 2017 version of the Liver Imaging Reporting and Data System (LI-RADS v2017) [26] were utilized to investigate the diagnostic accuracy and predict the prognosis of cHCC-CCA [7, 27, 28]. We plan to further differentiate cHCC-CCA and HCC or CCA by using LI-RADS v2018 [29].

In conclusion, the imaging features of combined hepatocellular carcinoma-cholangiocarcinoma are complex and do not correspond to the alpha fetoprotein or carbohydrate antigen 19-9 levels. In addition, some characteristics, including age, carbohydrate antigen 19–9 > 37 U/ml, arterial phase peritumoral enhancement, and delayed enhancement are independent predictors of poor recurrence-free survival. Furthermore, arterial phase peripheral enhancement is an independent predictor of poor overall survival in patients with combined hepatocellular carcinoma-cholangiocarcinoma.