Washout appearance of hepatocellular carcinomas using standardized contrast-enhanced ultrasound (CEUS) including an extended late phase observation – Real-world data from the prospective multicentre DEGUM study

1 Background

Non-invasive diagnosis of hepatocellular carcinoma (HCC) in high-risk patients is based on the characteristic enhancement pattern of arterial phase hyperenhancement (APHE) followed by contrast washout (WO) in the portal venous or late phase upon contrast-enhanced ultrasound. The more recently developed standardised CEUS algorithms ESCULAP (Erlanger Synopsis for Contrast-enhanced Ultrasound for Liver lesion Assessment in Patients at risk) and CEUS LI-RADS® (Contrast Enhanced UltraSound Liver Imaging Reporting And Data System) are meant to improve the standardisation and quality of reporting CEUS findings in HCC-suspect liver lesions. With these CEUS algorithms, predefined criteria such as nodule size and contrast enhancement in the arterial phase, portal venous phase and late phase are used to categorize focal liver lesions according to their risk of being HCC lesion. Categories rank from definitely benign lesions to definite HCC [1–9]. For the diagnosis of definite HCC (LR-5) with CEUS LI-RADS®, a combination of APHE and subsequent contrast washout of mild degree and late onset (after≥60 seconds) is required. This definition is also taken up by the United States HCC guidelines by the American Association for the Study of Liver Diseases (AASLD) [10]. Nonetheless, recent studies suggest that a relevant proportion of HCCs does not display this diagnostic enhancement pattern and might therefore escape non-invasive diagnosis due to atypical washout and / or a lack of APHE [11–14].

Recently, we conducted a prospective multicentre real-life study funded by the German Society for Ultrasound in Medicine (DEGUM) to assess the diagnostic accuracy of standardised CEUS for the non-invasive diagnosis of HCC in high-risk patients [6–9]. Our results suggested that a relevant proportion of HCCs did not fulfil the diagnostic criteria defined as “typical enhancement pattern” of APHE followed by late-onset washout of a mild degree. Therefore, we conducted a sub-analysis in order to clarify whether specific features could be identified which might be related to an “atypical” contrast washout in CEUS.

2 Materials and methods

2.1 Study design

The design of the DEGUM CEUS HCC study has been described in detail recently [6–9]. Briefly, this prospective multicentre study was conducted in adult patients at risk for HCC according to national HCC guidelines with a solid focal liver lesion upon B-mode ultrasound. Histology or – in cases where histology was not possible – contrast-enhanced MRI or CT were used as a reference standard. All patients provided their written informed consent. The local Ethics committee approved the study (ethics vote 16_17B). Participating centres were equipped with password-protected, individualised accounts to enter anonymised clinical and imaging data via online entry forms.

For this sub-analysis comparing typical versus atypical contrast washout, the study collective was limited to those patients with histologically proven HCC. The design of this sub-analysis is shown in Fig. 1.

OPEN IN VIEWER

Fig. 1

illustrates the study design.

3 Results

3.2 Subgroups of HCC according to CEUS patterns

Figure 2 illustrates the subgroups according to CEUS patterns of the 316 HCCs. Typical CEUS examples of HCCs with typical and atypical washout are shown in Fig. 3.

OPEN IN VIEWER

Fig. 2

shows the proportions of typical and atypical washout in the subgroup of HCCs with APHE. APHE, arterial phase hyperenhancement. WO, washout. HCC, hepatocellular carcinoma. CEUS, contrast-enhanced ultrasound.

OPEN IN VIEWER

Fig. 3

shows CEUS examples of typical and atypical HCCs. A, typical HCC. 3A1, B-mode: solitary lesion of 21 mm in S. IV/VIII in a cirrhotic patient with chronic hepatitis C infection. 3A2-3A5, CEUS: 3A2, arterial phase hyperenhancement (APHE). 3A3, isoenhancement in the portal venous phase after 60 seconds. 3A4, isoenhancement in the late phase at 3 minutes. 3A5, slight washout with mild hypoenhancement in the late phase after > 4 minutes. 3B, atypical HCC. 3B1, B-mode ultrasound: hypoechoic lesion of 38 mm in S. VIII in a cirrhotic patient with primary biliary cholangitis (histology: G2 HCC). 3B2, colour mode: hypervasularisation of the lesion. 3B3-3B6, CEUS: 3B3, arterial phase hyperenhancement (APHE). 3B4, isoenhancement in the portal venous phase after 60 seconds. 3B5, isoenhancement in the late phase after 2 minutes. 3B6, sustained isoenhancement without contrast washout after > 6 minutes.

APHE was seen in 271/316 HCCs (85.8%). The pattern of “typical HCC” defined as APHE followed by late (onset > 60 seconds) and mild contrast washout was found in 158/316 HCCs (50%).

72 of the 271 HCCs with APHE (26.6%) showed early (onset < 60 seconds) and / or marked contrast washout; 41 did not show any washout appearance until after 6 minutes.

Figure 4 shows the relation between onset and intensity of washout in HCCs with APHE in direct comparison.

OPEN IN VIEWER

Fig. 4

shows the onset and intensity of washout in HCCs with arterial phase hyperenhancement. Early washout: onset≤60 seconds. APHE, arterial phase hyperenhancement. WO, washout. HCC, hepatocellular carcinoma. CEUS, contrast-enhanced ultrasound.

3.3 Subgroups of atypical HCCs

The characteristics of the subgroups of typical versus atypical HCCs are shown in Figs. 5, 6. Most HCCs were between 2 and 5 cm in size. In the subgroup of HCCs≤2 cm, marked or early washout was rarely seen, while a hyper-iso pattern occurred more often in this subgroup (Fig. 5).

OPEN IN VIEWER

Fig. 5

shows the relative proportion of HCCs stratified by size categories for the different CEUS patterns. Typical HCC, APHE+late-onset (>60 seconds), mild washout. APHE, arterial phase hyperenhancement. WO, washout. HCC, hepatocellular carcinoma. CEUS, contrast-enhanced ultrasound.

OPEN IN VIEWER

Fig. 6

shows the relative proportion of HCCs stratified by histological grading for the different CEUS patterns. Typical HCC, APHE+late-onset (>60 seconds), mild washout. APHE, arterial phase hyperenhancement. WO, washout. HCC, hepatocellular carcinoma. CEUS, contrast-enhanced ultrasound.

As to the histological grading, most lesions showed a moderate differentiation. The hyper-iso pattern seemed to be associated with well-differentiated tumours (41.9% G1), whereas poor differentiation seemed to occur mostly in the group showing marked washout (12.7%); however, there were no statistically significant differences in the grading between typical and atypical HCC.

Our prospective multicentre real-world data confirm that in high-risk patients (80% liver cirrhosis), the typical HCC pattern of APHE with late-onset (>60 seconds), mild washout (definite HCC LR-5 according to the CEUS LI-RADS® system) is found in only 50% of the HCCs, suggesting that with this definition, every second patient would escape definite HCC diagnosis with CEUS. In these cases, further imaging and / or biopsy would be necessary to establish a definite diagnosis.

With this sub-analysis of the prospective multi-centre DEGUM study, we tried to identify distinctive features of HCCs, which might be associated with this atypical contrast enhancement behaviour.

First of all, there is literary evidence that differences in tumour size and histological grading might be related to atypical contrast enhancement behaviour. As to the subgroup of atypical HCCs with APHE but no contrast washout (hyper-iso pattern), this feature was not related with tumour size, but seemed to occur especially in well-differentiated HCCs (although – probably due to the small sample size – statistical significance was not reached). Concerning the onset and intensity of washout, we found that about one third of HCCs with APHE and washout display either an early and/or marked washout, correlating with macro-invasion of the liver vessels and portal vein thrombosis.

These findings are in accordance with a study by van Herbay et al. reporting no correlation between histological grading and washout or APHE in 130 HCCs [16]. However, different studies report discrepant results, with some literature suggesting a relationship between moderate differentiation (G2) and typical enhancement patterns, whereas others find no such correlation ([17], n = 112 HCCs; [18], n = 372 HCCs in 346 patients). In this context, the reader has to bear in mind that in all studies, the proportion of moderately differentiated G2 HCCs considerably outweighs the G1 and G3 lesions. In our study collective, about 55% of both typical and atypical HCCs showed moderate differentiation. Thus, cautious interpretation is warranted as the different size of subgroups can lead to a certain bias.

Few studies have assessed the onset and intensity of washout so far; and again, there is considerable controversy in the literature. In retrospective single-centre study assessing CEUS in 114 HCC lesions in 91 patients with chronic liver disease (75.8% cirrhosis), Martie et al. found that a typical contrast pattern of APHE followed by washout was associated with larger tumour size (45/55 HCCs > 3 cm / 81.8% versus 34/59 HCCs < 3 cm / 57.6%) [19]. Yet, the study did not distinguish between early and late portal venous phase, and the extent of washout was not assessed. Moreover, histology was not available for all the lesions, but CT or MRI served as the reference standard in the other cases. Similarly, Fan et al. reported a correlation between contrast washout and both larger tumour size and lesser tumour differentiation [18]. In their study, contrast washout occurred in 79.3% of the HCCs (295/372) and was more often perceived in larger HCCs (20-30 mm) compared to small lesions≤20 mm. A later examination point after 120 seconds was not assessed; neither did the authors assess the extent of washout. Concerning the onset of washout, the authors argue that the greater the structural differences between the blood supply of a focal liver lesion and the surrounding parenchyma, the faster the contrast agents pass through the lesion [18]. Also, they suggest a change of venous drainage along with ongoing hepatocarcinogenesis and dedifferentiation as an explanation for the correlation of washout and tumour size and differentiation.

Also, there is considerable controversy in the literature concerning the possible association of tumour size and enhancement behavior. Some authors report that both APHE and washout could be related to larger tumour size [11, 16], whereas our results to not allow for such a conclusion. However, statistical significance for these associations is often restricted to certain subgroups, and both sample sizes of the subgroups and the range of tumour sizes assessed vary between the different studies.

Another aspect which might affect contrast enhancement behaviour is the issue of perfusion alterations, both through direct tumour infiltration of liver vessels and through accompanying portal vein thrombosis (bland thrombosis or tumour thrombosis). In our study, we found a significant correlation between atypical contrast enhancement and the presence of macro-invasion of the liver veins and portal vein thrombosis. These perfusion alterations due to tumour infiltration of the liver vessels occurred primarily in HCCs with early onset of washout and / or marked washout. As the ultrasound contrast agent SonoVue® used in Europe and the United States is a purely intravascular agent (contrary to contrast agents in MRI and CT), it can be expected that perfusion alterations associated with HCC might lead to a different contrast enhancement behaviour of these tumours. Yet, from a pathophysiological point of view, there seems to be no straightforward explanation of why portal vein thrombosis or infiltration of liver vessels by diffuse tumour growth should lead to a change in contrast washout. Rather, it seems plausible that washout might occur earlier and appear more pronounced, as this is also a feature of intrahepatic cholangiocellular carcinoma (iCCA), which is known to be associated with pronounced desmoplastic stromal reaction. Thus, diffuse tumour growth between or inside hepatic vessels might influence washout appearance upon CEUS. However, as these observations are not enough to explain the different contrast enhancement patterns, it is likely that atypical HCC is characterised by a different tumour biology altogether. As HCC is known to be a very heterogeneous tumour entity in general, it seems impossible to draw conclusions from single tumour features as to the contrast enhancement behaviour of a single nodule, although some properties seem to occur especially in those HCCs with atypical contrast pattern upon CEUS. With the emerging importance of targeted therapies for HCC, biopsy will gain further importance both from a diagnostic and from a therapeutic point of view; this will hopefully help us to better understand differences in tumour biology and their impact upon contrast enhancement behaviour of HCC lesions.

There are some limitations to our study. The sample size of some sub-groups was small, thus precluding statistical analysis. In addition, we did not compare the contrast enhancement patterns between CEUS and MRI or CT.

For MRI, several studies have shown that the use of liver-specific contrast agents can improve the diagnostic accuracy of the non-invasive diagnosis of HCCs especially in small lesions < 1 cm. [20, 21]. Similarly, the Kupffer-cell-specific ultrasound contrast agent Sonozoid, which is widely used in Asia but not available in Europe, has been demonstrated to have a high sensitivity and specificity for both detection and characterization of HCC [22].

In a retrospective single-centre study comparing CEUS and MRI with liver-specific contrast agent for the characterization of focal liver lesions in 83 patients with both benign and malignant liver lesions 42 HCCs), Beyer et al. found a similarly high sensitivity, specificity, positive and negative predictive value concerning lesion classification for CEUS, MRI, and a combination of both modalities [23]. However, in this study, histological findings were available in 53 lesions only, and the study was not conducted in cirrhotic patients or patients at risk for HCC [23].

A particular limitation of ultrasound-based techniques in general, which is often mentioned, is the dependency on the examiner’s clinical expertise [24, 25]. In our study, sufficient quality of examinations was ensured by the fact that only centres with a high level of expertise in abdominal ultrasound including CEUS according to DEGUM Level II-III (at least 6,000-10,000 ultrasound examinations) were invited to participate in the study.

Also, ultrasound-based techniques are often criticized for their insufficient sensitivity in case of small lesions, high depth location and / or patient-specific characteristics such as severe obesity and ascites. In our study, in the case of insufficient contrast enhancement in the (dorsal) liver parenchyma in the late phase, examiners were instructed to apply a second contrast bolus with subsequent examination of the late phase only. Nonetheless, there are rare cases where physical limitations make ultrasound-based diagnosis impossible. In these cases, a second imaging modality such as MRI or CT will be chosen, which may in some cases also be unsuccessful due to the same limitations. For these rare cases, biopsy will be the approach to choose.

Furthermore, some recent technical developments can be expected to further increase the diagnostic accuracy of CEUS especially in the case of small lesions. One of these options is the additional use of perfusion analysis, which may further help with a more objective analysis of contrast enhancement patterns and tumour neoangiogenesis relying on internal and external software. Besides its diagnostic value for the characterization of focal liver lesions, CEUS has also been demonstrated as a useful tool in differentiating reactive hyperemia from residual tumour tissue after percutaneous ablative interventions; in these cases, colour-coded perfusion analysis can be of additional value [26, 27].

Another technical refinement is the so-called high frame-rate (HiFR) CEUS. HiFR CEUS can be expected to further improve the assessment of dynamic tumour microvascularization, which has been shown to be of additional diagnostic value especially in small lesions [28–30].

In our patient collective of high-risk patients for HCC, approximately 80% of patients had liver cirrhosis. Whereas contrast enhancement patterns of HCC in the cirrhotic liver have been well studied, there is little evidence in the literature concerning typical CEUS features of HCC in non-cirrhosis. In a recent retrospective multicentre study in 96 patients with histologically proven HCC lesions in the absence of liver cirrhosis, Dong et al. found heterogeneous APHE in 75/96 HCCs (78.1%). Interestingly, all lesions in this study showed some form of washout appearance; in 27 cases /28.1%, onset of washout occurred early (<60 seconds) [26]. These findings are in accordance with our own results of a recent prospective multicentre study in 517 high-risk patients for HCC. In a sub-analysis focusing on the non-cirrhotic patients only (n = 47, among these 30 histologically proven HCCs), we found the typical CEUS pattern of arterial phase hyperenhancement and late-onset (>60 seconds), mild washout in 22/30 HCCs (73.3%). Interestingly, very late onset of washout > 4-6 minutes did not occur in HCCs in the non-cirrhotic liver [31].

Strengths of the study are the prospective, multi-centre real-life setting and the availability of histology as the gold standard in all lesions.

5 Conclusion

Mild contrast washout with late onset as defined as typical by the current imaging guidelines is observed in half of the HCCs in a real life setting.

Our results suggest that as only half of the HCCs display the enhancement pattern of APHE followed by late-onset (>60 seconds) and mild contrast washout as required for definite non-invasive diagnosis using the CEUS LI-RADS® criteria, the heterogeneity of HCC as a tumour entity is mirrored by a heterogeneity of enhancement patterns.

Besides measurable differences in nodule size, tumour differentiation and perfusion alterations, there might be a different tumour biology of atypical HCCs beyond those visible features. Examiners should be aware that especially in large tumours with diffuse growth pattern, portal vein thrombosis and infiltration of the liver vessels, washout can be atypical or even absent.