Pathomolecular analysis, allowing the distinction of histological subtypes of HCC, is clinically useful as it is associated, at least in some subtypes, with specific prognosis, and possibly, response to systemic therapies (16, 24–26). Indeed, TP53 mutations are associated with macrotrabecular-massive HCC and poor prognosis, whereas CTNNB1 mutations are associated with microtrabecular HCC and better prognosis. In contrast, the prognosis of the SH subtype remains difficult to predict in the absence of an association with specific molecular genetic alterations of clinical interest and lack of a consensus pathological definition. Hence, we performed an exhaustive pathological analysis of SH-HCC based on one of the largest series of SH-HCC described in the literature, with 83 SH-HCC cases representing 28% of all HCC according to our new definition. This prevalence agrees with that found in previous studies (13–35%) (7–10). We also confirmed the association between SH-HCC and metabolic syndrome and, microscopically, with NAFLD in the underlying non-tumoral liver. Accordingly, the incidence of this subtype should grow in the future owing to the increased prevalence of metabolic syndrome leading to NAFLD, particularly in developed countries (3, 4). As suggested in other series (9, 12, 27), we also found excessive alcohol intake associated with metabolic syndrome and SH-HCC. The association of SH-HCC with metabolic syndrome and excessive alcohol intake could be explained by its carcinogenesis, with inflammation and lipotoxicity seeming to play a major role without specific genetic alterations (16, 18, 19).
Overall, SH-HCC tumors were smaller on average than non–SH-HCC tumors. They were well demarcated but less frequently encapsulated. As previously described, tumors had a yellow and/or white color, reflecting steatosis and fibrosis, respectively (7). Microscopically, SH-HCC consistently featured ballooning, fibrosis and inflammation. Ballooning was the most specific and sensitive criteria of SH-HCC. Importantly, ballooning was readily identifiable in all cases, cells. Compared to clear cells, ballooned tumor cells are usually larger, with indistinct cytoplasmic borders. In addition, fibrosis was highly specific when it had a pericellular distribution. While steatosis was also specific, its intensity was mild in most cases.
According to our results, the 50% threshold, currently used, does not seem enough relevant to distinguish SH-HCC from HCC NOS since SH-HCC cases were clinically homogeneous, whatever the extent of the SH component (7). Furthermore, in multivariate analysis, the percentage of SH contingent had no prognostic impact. Thus, according to our results, a minimum percentage of 5% SH component should be sufficient to retain the diagnosis, as some authors have already suggested (8, 10, 12). Then, we propose a new diagnostic definition based on ballooning (major criterion) with two other criteria (among steatosis, Mallory Denk bodies, fibrosis and inflammation) in at least 5% of the tumour area. According to this new definition, 28% of HCC corresponded to SH-HCC and formed a group with similar clinico-pathological features and prognosis to those observed in the initial SH-HCC group. In this study, we assessed the prognosis of patients with SH-HCC undergoing partial liver resection. We observed a similar OS and RFS as compared with non–SH-HCC. These results are in accordance with most of the studies that evaluated the prognosis of patients with SH-HCC mainly treated by liver transplantation (7–10, 27). Moreover, histoprognostic factors such as tumor differentiation, vascular invasion and satellite nodules were similar to those with non–SH-HCC.
Overall, 5% of HCC cases had intra-tumor steatosis covering at least 5% of the tumor surface, without any other criteria of SH-HCC. Steatotic HCC does not seem associated with metabolic syndrome but is more frequently associated with NAFL in background liver rather than NASH as compared with SH-HCC, which is more associated with NASH. This similarity between underlying non-tumoral liver and tumor could suggest common pathomolecular mechanisms in the process of liver carcinogenesis. Some other features (e.g., tumor size and macroscopic appearance) tended to be similar to non–SH-HCC, as proposed by other studies (12, 28).
On immunostaining, SH-HCC was distinguished from other HCC cases by a more frequent expression of inflammation markers, such as CRP and SAA. The concomitant expression of CRP and SAA was correlated with the number of SH criteria within the tumor. CRP was less specific than SAA in the diagnosis of SH-HCC. These results extend previous data and highlight the potential role of the inflammation pathway in the carcinogenesis of SH-HCC (12, 16, 19, 29, 30). The less common overexpression of glutamine synthetase agrees with the absence of CTNNB1 mutations in SH-HCC (16).
The excellent diagnostic agreement between biopsies and surgical resection in our cohort supports the use of biopsy to identify this subtype and to exclude differential diagnosis. Indeed, on biopsy, it may be challenging to distinguish SH-HCC and focal nodular hyperplasia with SH features (29, 31, 32). Reticulin stain, which is usually helpful for differentiating benign from malignant hepatocellular nodules, is less useful in this context because both lesions exhibit frequent preservation with a pericellular pattern (32, 33).
Because of the particular macroscopic and microscopic aspects of SH-HCC, imaging study would be of potential interest. A few imaging studies have assessed SH-HCC with a limited number of cases (34, 35). In a recent study, evaluation of the Liver Imaging Reporting and Data System did not differ between HCC subtypes, including SH-HCC (36). Thus, imaging studies of SH-HCC would be needed to evaluate the radiological characteristics of this subtype.
This was a monocentric and retrospective study in a referral centre for hepatobiliary pathology. It included only patients who were eligible for partial surgical resection, with an inherent good prognosis. Nevertheless, this homogeneous selection allowed for better highlighting the possible prognostic impact of a given histological subtype. The results from the subgroup analysis and preoperative biopsies need to be confirmed in larger series.
To conclude, our results suggest that the diagnosis of the SH-HCC subtype can be based on the association of ballooning as major criterion with at least two other minor criteria (among steatosis, Mallory Denk bodies, fibrosis and inflammation) regardless the extent of SH component. The prognosis of SH-HCC is similar to that of non–SH-HCC.