Abstract
BACKGROUND:
Fibrolamellar hepatocellular carcinoma (FLHCC) is an unusual variant of hepatocellular carcinoma (HCC). Revealing the imaging features is important to the diagnosis of FLHCC.
OBJECTIVE:
The aim of this study was to investigate the imaging characteristics of FLHCCs.
METHODS:
This retrospective study included 29 patients with histopathologically proved FLHCC and 96 patients proved HCC. All patients underwent an ultrasound examination pre-operation.
RESULTS:
The average maximum diameters of the FLHCC and HCC lesions were 7.4±4.1 cm and 4.1±3.0 cm, respectively. On the ultrasound, 79.3% of the FLHCCs and 12.3% of the HCCs showed the internal hyperechoic area; 48.3% of the FLHCCs and 3.3% of the HCCs displayed a strip-like attenuation. Calcification was noted in 20.7% of the FLHCCs, while none in HCCs. On the contrast-enhanced ultrasound (CEUS), all FLHCC lesions and 87.7% of the HCCs displayed hyperenhancement in the arterial phase. An internal, unenhanced central scar appeared in all FLHCCs, while none in HCCs.
CONCLUSIONS:
The ultrasonographic features of FLHCC lesions indicate that they are relatively large masses showing the internal hyperechoic area or strip-like attenuation or calcification on the US and hypervascularity with an unenhanced central scar on the CEUS as compared with conventional HCC lesions.
Introduction
A fibrolamellar hepatocellular carcinoma (FLHCC) is an unusual variant of a hepatocellular carcinoma (HCC), which was first described by Edmonson in 1956. This type of carcinoma accounts for approximately 8% of the HCC cases [1, 2]. FLHCCs have distinct clinical and pathological differences with conventional HCCs, and they differ in terms of demographics, susceptibility factors, tumor markers, and prognosis. A diagnosis of FLHCC requires the integration of clinical information, imaging studies, and histological evaluation. Ultrasound (US) is a screening tool widely used in clinical practice, but the characteristics of FLHCCs shown on a conventional US have not been emphasized in prior investigations. A contrast-enhanced ultrasound (CEUS) is recommended for the evaluation of focal liver lesions due to its harmless contrast agent and real-time, multidimensional observation of lesions [3–5]. To our knowledge, there are few relevant studies of FLHCCs investigated with CEUS. In this study, we performed such an analysis and compared FLHCCs with conventional HCCs to reveal the imaging characteristics specific for FLHCCs. We hope this research will improve the diagnostic accuracy of FLHCCs using US.
Materials and methods
Patient population and clinical information
The institutional review board approved this retrospective study, and the requirement for patients’ written informed consent was waived (Approval number: B2013-073). We reviewed the pathology registry at our institution from January 2000 to January 2018, and 29 patients with a histopathological diagnosis of FLHCC were enrolled in this study. Participants included 19 males and 10 females (a mean age of 39.1±17.2 years old, with an age range of 15–77 years old), and the basic demographic and clinical characteristics of these patients are shown in Table 1. Among the 29 patients with FLHCCs, the primary symptoms included abdominal discomfort in 12 cases, abdominal masses in 4 cases, and weight loss in 2 cases. The other 11 cases were diagnosed as focal liver lesions on incidental US scans.
Clinical characteristics of the patients
Clinical characteristics of the patients
FLHCC, fibrolamellar hepatocellular carcinoma; HCC, hepatocellular carcinoma; AFP, alpha-fetoprotein (Normal range: 0–20 ng/mL at our institution).
Complete tumor resection was performed in all of the FLHCC cases by partial hepatectomy. All of the microscopic sections of the resected liver specimens were retrospectively reviewed for confirmation of the diagnosis of FLHCC by one pathologist with over 10 years’ experience in hepatic pathology. After surgery, all of the patients were followed up twice a year with imaging examinations and clinical consultation (the mean follow-up time was 6.8±4.5 years, with a follow-up time of 1–15 years).
As a control, we enrolled 96 patients with pathologically proved HCC from April 2015 to December 2015 at our institute. Participants included 76 males and 20 females (with a mean age of 55.8±10.6 years old, and an age range of 34–82 years old). The primary symptoms were abdominal pain and distention in 31 cases, anorexia in 11 cases, abdominal masses in 7 cases, fever in 2 cases, and jaundice in 2 cases; the other 43 cases were discovered upon cirrhosis screening with an abdominal US. All of the 96 cases underwent a partial hepatectomy. The same follow-up protocol was used for HCC patients (with a mean follow-up time of 3.3±1.4 years, and a time range of 3–4 years).
All of the FLHCC patients underwent conventional grayscale US and color Doppler imaging pre-operatively, and 15 of them who were enrolled after 2008 underwent CEUS, as well. All of the 96 patients with HCC underwent conventional US and CEUS. We used the commercially-available US systems Esaote Technos MP (Esaote Biomedica, Genoa, Italy), Philips iU22 (Philips Healthcare, Bothell, WA, USA), or GE Logiq E9 (General Electric Medical Systems, Milwaukee, WI, USA), with an abdominal convex array probe (with a frequency range of 1.9–6.0 MHz). The grayscale US features of the lesions, including the location, size, shape, echogenicity, and other morphological features, were recorded. The blood flow signal of the lesions on the color Doppler imaging, including whether the blood flow signal could be detected and the distribution of the blood flow signal, were recorded. The CEUS setting on the contrast imaging was used with a low mechanical index, which was 0.07 for the Esaote system, 0.06 for the Philips system, and 0.11 for the GE system. A bolus injection of 2.4 mL of the contrast agent (SonoVue, Bracco, Milan, Italy) was administered intravenously via an antecubital vein, followed immediately by a 5 mL saline flush. A second injection of the contrast agent was added at least 15 min later when patients had two or more hepatic lesions, or the dynamic imaging features of CEUS were not satisfactorily acquired. A timer was activated at the beginning of the injection of the contrast agent. The enhancement pattern of the target lesion was observed continuously for up to 6 min. The arterial, portal, and delayed phases were defined as 0–30 s, 31–120 s, and 121–360 s after the contrast medium injection. Consecutive 45 s cine clips were recorded at the beginning of the washin to reveal the arterial features through the peak enhancement, and clips of the portal venous phase and some repetitions at the delayed phase were stored. All of the US images of the lesions were reviewed retrospectively by two experienced radiologists with over 20 years of experience in consensus.
Imaging analysis
The conventional US features of liver lesions can be described as having the following features: (1) The echogenicity of the nodules, (2) the internal hyperechoic area, defined as an irregular hyperechoic area in the lesion (Fig. 1A), (3) strip-like attenuation, defined as an area in the center of the lesion with a strip-like echo defect (Fig. 1A), (4) calcification, defined as a strong echogenicity mass with acoustic shadowing, and (5) subcapsular thick blood vessels, defined as rich color signals at the periphery of the lesion on color Doppler imaging.
A 15-year-old boy with fibrolamellar hepatocellular carcinoma (FLHCC) in a non-cirrhotic liver. On the grayscale ultrasound, a large liver mass (large arrow) with a strip-like echo defect zone (thin arrow) in the center and internal hyperechoic area (white square) is shown (A). On the contrast-enhanced ultrasound (CEUS) (contrast-enhanced image on the left and corresponding B-mode on the right), peak enhancement in the arterial phase and an unenhanced central scar (arrow) is shown (B). The non-enhanced area (arrow) in the portal phase (C). Washout in the delayed phase (arrow) (D). The lesion (thin arrow) and the central scar (large arrow) in the artery phase (E) and portal phase (F) on magnetic resonance imaging (MRI). Microscopy (H&E, 50×) of the resected tumor showed that large eosinophilic, polygonal neoplastic cells in cords are separated by parallel sheets of fibrous stroma (arrow) (G).
The enhancement patterns of the hepatic lesions on CEUS were classified as follows: (1) Hyperenhancement, iso-enhancement, or hypo-enhancement, which were defined as the echogenicity of the lesion being higher than, similar to, or lower than that of the surrounding liver parenchyma at peak enhancement; (2) washout, the echogenicity of the lesion in the portal or delayed phases was lower than the liver parenchyma. The washout degree was classified into marked washout and mild washout. Marked washout was diagnosed when the liver nodule became virtually devoid of contrast. Mild washout was diagnosed when the liver nodule became less enhanced than the surrounding liver, but still demonstrated some degree of persistent contrast enhancement. When the enhancement of a lesion was similar to or had a higher degree than the surrounding liver parenchyma in the delayed phase, it was defined as no washout; (3) an unenhanced central scar, defined as a stellate or linear hypo-enhancement in the center of the lesion with an acoustic attenuation at the peak enhancement; (4) a non-enhanced area, defined as a round or irregularly non-enhanced area through the vascular phases on CEUS; and (5) non-enhancement, no enhancement of the lesion at all vascular phases on CEUS.
The diagnostic criteria for the HCCs in this study were made using the algorithm of the CEUS Liver Imaging Reporting and Data System (CEUS LI-RADS) issued by the American College of Radiology (ACR) [6]. The diagnosis of recurrence during the follow-up was defined as a newly detected focal liver lesion on the US and classified as CEUS LR-4 or LR-5 on CEUS and LR-4 or LR-5 on magnetic resonance imaging (MRI), and/or with an elevated alpha fetoprotein (AFP) level.
The mean values were used for the continuous variables. In the analysis of the patients’ characteristics and imaging features, the data were expressed as the means±SD. Differences in the proportions between the two groups were analyzed using the Pearson chi-square test and Fisher’s exact test. Parametric data were compared using the Student’s t test, and nonparametric data were calculated using the Wilcoxon test. All of the P values were derived from two-tailed tests, and a level of less than 0.05 was considered to be statistically significant. Statistical analysis was performed using IBM SPSS Statistics (version 22.0; IBM Corp, New York, USA).
Results
The conventional US findings
The characteristics of FLHCCs and HCCs shown in USs are summarized in Table 2. All of the 29 cases of FLHCCs were solitary, with 18 located in the right lobe and 11 located in the left lobe. The mean maximum diameter was 7.4±4.1 cm (with a range of 2.5–16.7 cm). On the grayscale US, 24 lesions (82.8%) were hypoechoic and 19 lesions (65.5%) were well defined. The internal hyperechoic area appeared in 23 lesions (79.3%). Nearly half of the lesions (48.3%) showed a strip-like attenuation. Calcification was found in six lesions (20.7%), and all were located in the center. Intrahepatic bile duct dilatation was seen in 1 lesion on the grayscale US.
Characteristic ultrasonography findings of FLHCC and HCC
Characteristic ultrasonography findings of FLHCC and HCC
FLHCC, fibrolamellar hepatocellular carcinoma; HCC, hepatocellular carcinoma; US, ultrasound; CEUS, contrast-enhanced ultrasound.
On color Doppler imaging, blood flow signals were detected in 26 lesions (89.7%) of the FLHCCs, and 17 (58.6%) of them had subcapsular thick blood vessels. Among them, 20 lesions showed an arterial Doppler spectrum and two showed a venous Doppler spectrum. The average resistance index (RI) was 0.69±0.13 (with a range of 0.58–0.84).
In the HCC group, there were a total of 122 lesions, of which 20 cases had two lesions and three cases had three lesions. There were 84 lesions located in the right lobe and 38 in the left lobe. The mean maximum tumor diameter of the HCC lesions was 4.1±3.0 cm (with a range of 1.0–16.0 cm). On the grayscale US, 90 lesions (73.8%) were hypoechoic and 45 lesions (36.9%) were well defined. The internal hyperechoic area and strip-like attenuation also occurred in 12.3% and 3.3% of the HCCs, respectively. No calcification was found in any of the 122 HCC lesions.
On the color Doppler imaging, blood flow signals were detected in 80 HCC lesions (65.6%), and 41 lesions (33.6%) appeared with subcapsular thick blood vessels. Among them, 75 lesions showed an arterial Doppler spectrum and five showed a venous Doppler spectrum. The average resistance index (RI) was 0.73±0.10 (with a range of 0.61–0.83).
The imaging findings are summarized in Table 2. On the CEUS, 15 FLHCC lesions (100%) were enhanced rapidly at the beginning of the arterial phase (with a mean washin time of 15.5±3.8 s, and time range of 8–22 s), showing a holistic synchronous hyperenhancement in 11 lesions (73.3%) and an enhancement from the periphery to the center in 4 lesions (26.7%). The majority of the lesions (86.7%) presented washout in the portal phase and 2 lesions (13.3%) in the delayed phase (with a mean washout onset of 85.8±35.8 s, and a time range of 40–125 s). All of the lesions presented a mild washout. An unenhanced central scar was seen in all lesions, and 4 lesions (26.7%) showed more than 1 non-enhanced area simultaneously (Fig. 1).
On the CEUS, all of the 122 HCC lesions were enhanced in the arterial phase (the mean washin time was 16.8±4.2 s, with a time range of 4–29 s), 107 lesions (87.7%) had hyperenhancement and 15 lesions (12.3%) had iso-enhancement. Washout was detected in 114 HCC lesions (93.5%) in the portal phase or in the delayed phase (the mean washout onset was 99.1±66.2 s, with a time range of 35–180 s) and all presented a mild washout. Additionally, 28 lesions (23.0%) displayed more than 1 non-enhanced area. However, no non-enhanced central scars were noted in the HCC lesions (Fig. 2).
A 65-year-old man with hepatocellular carcinoma (HCC) in a cirrhotic liver. On grayscale ultrasonography, a large liver mass (arrow) is shown (A). On the contrast-enhanced ultrasound (CEUS) (contrast-enhanced image on the left and corresponding B-mode on the right), peak enhancement in the portal phase (arrow) (B). The non-enhanced area (arrow) in the portal phase (C). Washout in the delayed phase (arrow) (D). The lesion (thin arrow) and the non-enhanced area (large arrow) in the artery phase (E) and portal phase (F) on magnetic resonance imaging (MRI). Microscopy (H&E, 100×) of the resected tumor showed that the thickened plates of malignant hepatocytes and an area of apoptotic and necrotic tumor cells are in the center (arrow) (G).
The imaging diagnoses of all 15 FLHCCs and 122 HCCs were classified using the CEUS LI-RADS criteria, as shown in Table 3.
Imaging diagnosis of all lesions according to CEUS LI-RADS criteria
FLHCC, fibrolamellar hepatocellular carcinoma; HCC, hepatocellular carcinoma; CEUS, contrast-enhanced ultrasound; LI-RADS, Liver Imaging Reporting and Data System.
The typical pathological findings in the resected FLHCCs and HCCs are listed in Table 4. In the 29 resected liver specimens of FLHCC, all had well-demarcated, lobulated borders from the surrounding liver parenchyma and were predominantly hard. The specimen sections were dark brown or brownish green with stripes of white fibrous tissue. Fibrosis infiltrated throughout the nodule was seen in all FLHCCs. A distinct central scar with irregular stellate or band-like firm fibrosis stroma was seen in 21 of 29 lesions (72.4%). Calcification was found in 12 of the FLHCCs (41.4%), and all were located in the center scar. Neither calcification nor a central fibrous scar was found in the conventional HCCs. Necrosis and hemorrhage both occurred in some cases of FLHCCs and conventional HCCs.
Typical pathological findings of FLHCC and HCC
Typical pathological findings of FLHCC and HCC
FLHCC, fibrolamellar hepatocellular carcinoma; HCC, hepatocellular carcinoma.
Microscopically, the FLHCC tumors were composed of large eosinophilic, polygonal neoplastic cells in sheets, cords, or trabeculae separated by parallel sheets of fibrous stroma. Vascular invasion of the portal or hepatic vein was found in 3 of the 29 FLHCCs (10.3%), and 7 cases (24.1%) of FLHCC were accompanied with regional lymphadenopathy, while 51 of the 96 HCC cases (53.1%) demonstrated vascular (portal or hepatic vein) invasion, and 4 cases (4.2%) of HCC had regional lymphadenopathy.
We confirmed pathologically that 4 cases (13.8%) of FLHCC and 45 cases (46.9%) of HCC had cirrhosis, including schistosomiasis cirrhosis in 1 of the FLHCC cases and alcoholic cirrhosis in 2 of the conventional HCC cases.
The recurrence of FLHCC was found in 9 cases (31.0%), among them, 7 cases within 1 year and 2 cases within 3 years, while the recurrence of HCC was found in 51 cases (53.1%), with 44 cases within 1 to 2 years and 7 cases within 3 to 4 years.
Discussion and summary
It was not until 1980 when Craig and Peters (1980) and Berman and Libbey (1980) reported that patients with FLHCC were distinct from those with HCC, after which the clinical importance of establishing the diagnosis was finally valued. FLHCC occurs primarily in young adult patients, with an age range of 10–35 years old (with a mean of 25 years old) without gender differences [7, 8]. In our patient population from a single center in China, the patients tended to be older (with a mean age of 39.1±17.2 years old, and an age range of 15–77 years old) at diagnosis than the ages previously reported, but still younger than patients with HCC (P < 0.001). In this study, no significant difference in gender was found between the FLHCC and HCC patients (P = 0.131). Similar to other reports, the FLHCC cases were not as sensitive as the HCC cases in regard to the AFP level (P = 0.016) [9–11]. In this study, the positive rate of the hepatitis B surface antigen in FLHCCs was lower than that in HCCs (P < 0.001), though it was higher than in an earlier study [12]. FLHCCs are commonly considered more suitable for liver resection than HCCs, as the majority of HCC patients are associated with liver cirrhosis [13–15]. The disease recurrence rate in the FLHCC patient population ranged as high as 33–100%, as previously reported, and there was a higher recurrence rate in patients with vascular invasion [16–18]. In this study, the FLHCC group showed a lower recurrence rate (31.0%) than the HCC group (53.1%) (P = 0.037). Seven patients (24.1%) with FLHCC had a recurrence in the first year after surgery, and all of the cases were confirmed with vascular invasion pathologically.
Histological differences are the most objective and widely accepted way to discriminate FLHCCs from HCCs. In contrast to typical HCCs, the tumor interstitium of FLHCCs is rich. The radiating fibrous septa compartmentalizes the tumor cells into zones and, if coalescent, results in a central fibrous scar. In this study, a central fibrosis scar was found in more than half of the cases (72.4%), while calcification was less common (41.4%) in FLHCCs; however, both center scars and calcification were rare in HCC cases (P < 0.001). As an earlier study showed, calcification, if it existed, was usually punctate, nodular, or stellate, and lay within the scar and was presented in 33–55% of the FLHCC cases [19]. Necrosis and hemorrhage occurred in both tumor types. Unlike in conventional HCCs, which have a vascular invasion in a high proportion, portal and hepatic vein invasion is less commonly seen in FLHCCs (10.3%) (P < 0.001). However, regional lymphadenopathy (24.1%) is significantly higher in FLHCCs than in conventional HCCs (4.2%) (P = 0.003). This distinctive characteristic of FLHCCs demonstrates the importance of systematic imaging examination for clinical staging.
We found that both US and CEUS images demonstrated frequent and characteristic features that correlated well with the gross pathologic features of FLHCCs. On the grayscale US, the FLHCC was often characterized as a well-defined hypoechoic nodule. In contrast to the multifocal lesions commonly seen in HCCs, the FLHCC was not accompanied by satellite lesions and was usually solitary (P = 0.004) [20]. Significant differences were found between FLHCCs and HCCs in the proportion of the internal hyperechoic area (P < 0.001) and strip-like attenuation (P < 0.001). In relation to the gross pathologic features, the internal hyperechoic area might result from the radiating fibrous septa that infiltrate the tumor, and the strip-like attenuation might be associated with the central fibrosis scar. It was reported that US was partially successful in identifying the central scar, with 33–60% of scars shown on US compared with those on computed tomography (CT) and pathologic analysis [19]. It was also reported that 90% of FLHCC lesions were found to have a hypointense central scar area on T2-weighted images of MRI [21]. Compared with conventional HCCs, in which calcification hardly occurs, calcification was identified in 20.7% of the FLHCCs in our study, and all were located in the central fibrosis scar, lower than that found in pathology (41.4%). It was reported that calcification was found in 35–68% of the FLHCC lesions imaged by CT [22]. This might imply that CT is more sensitive than US in demonstrating a low level of the calcification in the liver.
On the color Doppler imaging, FLHCCs tended to be characterized as having vascular-rich lesions containing subcapsular thick blood vessels with an arterial Doppler spectrum compared with HCC lesions (P = 0.011; P = 0.013). These results may have been caused by the relatively smaller size of the HCCs in these samples. It was reported that in relatively small HCC tumors, the sensitivity of color Doppler imaging in finding arterial neo-vascularity was low, and an abnormal flow was detected in less than 50% of the HCC lesions [23, 24]. An abnormally elevated resistive index of the arterial flow was detected in both the FLHCC and HCC lesions on the color Doppler imaging in our study.
CEUS has been very useful for characterizing liver tumors. Our study demonstrated that all of the FLHCC lesions presented rapid hyperenhancement in the arterial phase and a mild washout in the portal phase, similar to a typical consensus enhancement pattern of HCC lesions. No differences were found between them in regard to enhancement in the arterial phase (P = 0.317) or washout in the portal and delayed phases (P = 0.437) [25]. According to the CEUS LI-RADS criteria, all of the cases of FLHCC were classified as LR-5 and LR-M, which were diagnosed as probably or definitely HCC. Therefore, the CEUS LI-RADS criteria were appropriate for the clinical diagnosis of FLHCCs.
The dynamic enhancement patterns of hypervascularity in FLHCCs, which were consistent with previous reports on contrast-enhanced CT and MRI, demonstrated enlarged feeding arteries with a dense, often heterogeneous tumor blush [26, 27]. During the portal and late phases, CEUS showed discordance with CT and MRI images by demonstrating a mild centrifugal washout within the FLHCC lesions, while sustained enhancement was observed with CT and MRI images [28]. Such discordance has recently been described in up to 34% of the focal liver masses and might reflect the properties of the CT and MRI contrast agents, which, unlike microbubbles, diffuse into the interstitium and contribute to late enhancement [29, 30]. There was no statistical difference between the FLHCCs and HCCs in non-enhanced areas, which represented necrosis or hemorrhage in gross pathologic features (P = 1.000) on CEUS, showing that conventional US was substantially less useful for demonstrating necrosis or hemorrhage. The appearance of unenhanced central scars was the characteristic feature that differentiated the FLHCCs from conventional HCCs on CEUS (P < 0.001), which showed the same non-enhancement in contrast-enhanced CT or MRI images and was derived from the coalescent fibrosis stoma in the center [31].
There are some limitations to our study. First, it was a retrospective analysis. Second, the patient population with FLHCC was relatively small compared with that of HCC. Nevertheless, our sample size was relatively large compared with previously published studies. Third, there were only 15 FLHCC lesions that underwent a CEUS examination. With the widespread use of CEUS in clinical practice, the diagnostic accuracy of FLHCCs can be improved. Furthermore, quantification software provides a more objective way to compare the hemodynamics between different lesions in CEUS and perfusion analysis has been proven to be an effective way in diagnosing liver lesions and evaluating the treatment efficacy [32, 33]. This need further study and could be helpful in diagnosing FLHCC from other liver lesions. Last but not least, we used SonoVue as the contrast agent, which is a pure blood pool agent. Sonazoid, which is also the second generation contrast agents, exhibits an additional Kupfer phase [34]. Researches on the diagnosis of FLHCC by CEUS with Sonazoid need be further carried out and maybe valuable.
Conclusion
Patients with FLHCC are mostly young, and few have liver cirrhosis and hepatic viral infection compared with those with conventional HCC. On a grayscale US, a FLHCC is often characterized by a relatively large solitary mass with an internal hyperechoic area or strip-like attenuation or calcification in the center of the lesion compared with a conventional HCC. On the color Doppler imaging, the FLHCC lesions with a rich blood supply and subcapsular thick blood vessels can be more easily seen than those of HCC. On the CEUS, the FLHCC lesions present hyperenhancement in the arterial phase and a mild washout in the portal or delayed phases. The unenhanced central scar is a unique feature that helps distinguish FLHCC from HCC.
Footnotes
Acknowledgments
The research was supported by grants from National Natural Science Foundation of China (81571675; 81873897).