Keywords

1 Introduction

Liver resection (LR) and transplantation (LT) are curative surgical options for patients with hepatocellular carcinoma (HCC). The latter, in particular, is the best treatment option for selected patients with early-stage disease. HCC is the fifth most common cancer type and the third most common cause of cancer-related death worldwide [1, 2]. LT could provide a long-term survival for HCC patients fulfilling the Milan criteria, similar to the LT outcomes in patients without cancer. Despite the restrictive selection policy, the recurrence rate of this tumor drastically affects patient survival in up to 25% of cases [3]. It is well known that the timing of recurrence after LT can represent a key predictor of survival, and that a recurrence occurring within 24 months (early intra- and/or extra-hepatic HCC recurrence) is frequently associated with a worse prognosis [4].

Early recurrence is defined as recurrence within 2 years after resection of HCC. From a pathophysiological perspective, this clinical entity might result from circulating HCC cell clones engrafting and growing in a target organ after LT or from non-detectable extra-hepatic metastases being present before LT.

Late recurrence (beyond 2 years after surgery) is more a result of new malignant clones and is related to underlying liver conditions (e.g., liver cirrhosis) [5]. The natural history of HCC is quite heterogeneous and it is possible to distinguish two models of tumor progression, depending on whether the disease develops in a cirrhotic or a non-cirrhotic liver [6].

2 Clinical Setting and Risk Factors

The primum movens in HCC development is an insult by pathogenic noxious agents (e.g., viruses, alcohol, aflatoxins, etc.) which damage the liver parenchyma. Immediate elimination of the insult can result in restoration of the normal hepatic tissue structure; if instead the causative agent cannot be completely removed, an inflammatory process ensues with fibrotic evolution of the liver and hepatic cell regeneration with nodular arrangement of the cells, and consequent formation of cirrhotic pseudonodules. This leads to subversion of the normal hepatic architecture which will inevitably result in a functional alteration of the liver. Subsequently, a series of genetic mutations involving genes involved in cell apoptosis and proliferation will follow, which will cause the bypass of the physiological mechanism of hepatic senescence. These events will give rise to altered cell lines, with the ability to invade vascular structures and produce distant metastases, and therefore to the onset of permanent lesions that over time can evolve into neoplasia. Specifically, the regenerative pseudonodules will become dysplastic and then neoplastic [7].

Despite much evidence supporting the recommendations of scientific societies around the world, the clinical manifestations of HCC are often late, with a diagnosis that is generally made in the advanced stage of the disease, thus significantly limiting treatment options. Based on genomic profiling and next-generation sequencing, two distinct subgroups of HCC (proliferative and non-proliferative) have been identified, each with “core” genomic alterations and/or oncogenic pathways and distinct histopathologic features and clinical outcomes. Proliferative HCC is an aggressive phenotype with moderate to poor cellular differentiation, high alpha-fetoprotein (AFP) levels, frequent vascular invasion, high tumor recurrence, and poor outcomes. Non-proliferative HCC is a less aggressive phenotype with well- to moderately differentiated tumor at histologic examination, lower AFP levels, and better outcomes [8].

However, translation of this knowledge into clinical practice and therapeutic decision-making has been challenging given the substantial intra- and inter-tumor genetic heterogeneity with some mutations present only in specific regions within the tumor, such that a single biopsy specimen is likely to be insufficient in accurate molecular stratification. To calculate the risk of post-LT HCC recurrence, comprehensive machine-learning algorithms, based on serial imaging, AFP, locoregional therapies, treatment response, and post-transplant outcomes, were recently proposed that demonstrated higher levels of accuracy than other risk scores for optimizing the allocation of donor organs [9].

3 Diagnostic Tools and Oncologic Monitoring

HCC is a neoplasm for which an adequate and careful surveillance program is essential, to diagnose the tumor at an early stage and be able to use a greater number of therapeutic strategies, thus improving patient outcomes [10]. Patients at high risk for developing HCC are those with chronic HBV or HCV infections or even cirrhotic patients. Surveillance should be carried out in cirrhotic patients who have eradicated HCV infection following antiviral therapy; in all cirrhotic patients, regardless of etiology, in Child-Pugh class A and B, and in those in class C awaiting transplant, while no increase in survival was observed for the other class C patients. HBV-positive non-cirrhotic patients, in whom the annual incidence of HCC is greater than 0.2% per year, should also be included in a surveillance program. Similarly, surveillance is recommended in patients with chronic HCV infection and bridging fibrosis in the absence of cirrhosis [11]. Liver ultrasound (US) is the most appropriate means of surveillance available to us, with acceptable diagnostic accuracy when used as a surveillance test (sensitivity between 58% and 89%; specificity >90%). US is less effective in detecting HCC in the early stage, with a sensitivity of only 63% when used for diagnostic purposes. The widespread use of US screening is also due to good patient compliance, low cost, non-invasiveness and the absence of risks. On the other hand, this method is highly dependent on the skill of the operator and the characteristics of the patient (body mass index ≥33, presence of ascites or bloating). With regard to serological markers, the AFP assay is certainly the most widely used in HCC. Persistently high values represent a risk factor for HCC development and can be used to define high-risk populations [12]. However, several studies have confirmed the importance of AFP in diagnosis but not in surveillance, since increased AFP levels in cirrhotic patients could be associated not only with HCC development, but also with a possible exacerbation of the underlying disease or reactivation of HBV or HCV infection; moreover, only a small proportion of early-stage tumors (10–20%) have high serum AFP values, a fact recently associated with a particularly aggressive subclass of HCC. As part of the surveillance, it was found that the combined use of AFP and US leads to an increase in sensitivity by about 6–8% compared to US alone, causing, however, an increase in false positives and costs. The diagnostic accuracy for HCC of this serological test was confirmed by retrospective case-control studies which considered the cut-off of 10–20 ng/mL to be more efficient for diagnosis, with sensitivity around 60% and specificity 80%.

Other tumor markers such as des-gamma-carboxy prothrombin (DCP) also known as vitamin K absence-induced prothrombin, the glycosylated AFP fraction (L3 fraction) on total AFP, alpha-fucosidase, and glypican [13] were evaluated, alone or in combination, more for diagnosis and prognosis than for surveillance. In particular, DCP levels have been associated with portal venous invasion and an advanced tumor stage, similarly to the levels of the AFP-L3 fraction. At present, none of these serological tests can be recommended for the surveillance of patients at risk of developing de novo HCC. The ideal recommended surveillance interval is six months, since a quarterly interval does not translate into any clinical benefit and vice versa an annual interval, while reducing costs, is associated with fewer diagnoses of HCC at an early stage and lower survival [14].

Recurrence is generally established by radiologic evidence of new tumor on computed tomography (CT) or magnetic resonance imaging (MRI), and systemic treatments are warranted when it presents as or becomes systemically spread [2]. In the last decade, many HCCs have been diagnosed based on imaging features alone in patients at high risk by using the typical radiologic features at dynamic imaging, without histopathologic evaluation. For this reason, correlation of imaging findings with specific molecular traits of HCC has gained substantial interest in recent years [15].

Diagnostic imaging (CT and MRI) is crucial for obtaining the standard of care for HCC evaluation, and they have a role in the preoperative assessment to predict HCC recurrence after LT and or liver resection. Very recently, the role of gadoxetate-enhanced MRI in differentiating proliferative from non-proliferative HCCs was analyzed, demonstrating that the majority of proliferative HCCs showed rim arterial enhancement, defined as irregularly shaped rim-like peripheral enhancement, and a large hypo-enhancing central component (50% or more area) may correlate with larger necrotic areas. Therefore, rim arterial enhancement was shown to predict a proliferative HCC, with reduced overall survival and an increased rate of intra- and extra-hepatic metastases [15].

4 Clinical Decision-Making and Surgical Management

The following preoperative CT and MRI findings are specific independent risk factors for postoperative early HCC recurrence as indicators of microvascular invasion [2]:

  • tumor size;

  • multifocality;

  • hypointensity on T1-weighted MRI sequences;

  • corona enhancement, hypointensity on hepatobiliary phase MRI;

  • peritumoral hypointensity on hepatobiliary phase;

  • non-smooth tumor margin;

  • incomplete tumor capsule;

  • satellite nodule;

  • mosaic architecture;

  • absence of fat in the mass;

  • macro-vascular invasion.

The size and number of tumors, which together represent tumor burden, are important prognostic factors for HCC: as tumor size increases, HCCs tend to have a higher frequency of vascular invasion, extrahepatic metastases and a decrease in patient survival. Multifocal tumors can represent multiple independent HCCs occurring simultaneously (multifocal HCC) or intrahepatic metastases from a primary HCC. The availability and success of curative treatment options, such as liver resection or transplantation, are highly dependent on the size and number of HCCs. Indeed, liver resection for HCC <3 cm improves long-term patient survival [2]. Intrahepatic metastases develop through two different routes. Small satellite nodules around the primary tumor form when cancer cells enter portal venules that drain from the primary tumor and spread into the surrounding parenchyma. Metastatic nodules outside the drainage area, including other segments or the contralateral lobe, develop through the systemic circulation of cancer cells. Multifocal tumors can have variable histological grades and other features, while all metastatic tumors of a single HCC are considered to have advanced lesions with advanced tumor grade. The prognosis of patients with intrahepatic HCC metastases tends to be worse than those with multifocal HCC.

Poorly differentiated HCCs tend to show lower signal intensity on hepatobiliary phase MRI than well-differentiated or moderately differentiated HCC [15]. Vascular invasion is more common in larger or higher histological grade HCCs. Cancer cells involve the portal venous system more frequently than the hepatic veins. Macrovascular invasion is related to poor prognosis because it provides tumor cells with the pathway to access the portal or systemic circulation, which can result in intrahepatic or systemic metastases. Thus, vascular invasive HCCs have frequent multifocality and a higher relapse rate after LR, ablation therapy, or LT [16] (Fig. 23.1).

Fig. 23.1
Three computed x-ray images of liver carcinoma. In image a, an arrow points to a cell mass in the right lobe. In image b, a dashed line encircles the upper part of the right lobe surgically resected. In image c, an arrow points to a cell mass near the surgical margin.

(a) Patient with typical hypervascular hepatocellular carcinoma (HCC) in the right hepatic lobe (arrow) on axial arterial phase contrast-enhanced computed tomography (CT) image. (b) The same patient underwent hepatic resection (circle), as shown on the axial portal venous phase contrast-enhanced CT image performed some weeks after surgery. (c) Six months later, axial contrast-enhanced magnetic resonance imaging in the arterial phase showed recurrent HCC along the hepatic resection margins (arrow)

Full size image

Surgical resection is regarded as the first-line treatment option for HCC patients with well-preserved liver function. Nevertheless, almost 70% of HCC patients develop tumor recurrence within 5 years after surgery [5]. Currently, the therapeutic options of HCC recurrence include LR, particularly for isolated hepatic and extra-hepatic metastases, and the following range of therapies:

  • transarterial chemoembolization or embolization (TACE or TAE);

  • radiofrequency and microwave thermal ablation (RFTA and MWTA);

  • multi-target tyrosine kinases inhibitor (sorafenib).

Limited but unresectable HCC recurrence in selected patients can be treated with locoregional therapy, which may include TACE, TAE, MWTA, and RFTA, with potential survival improvement, considering their repeatability or potential combination in a multimodality approach [17].

5 Conclusion

Managing recurrent HCC is a challenging area, as reflected by the highly heterogeneous conditions and treatment strategies. Since molecular classification and data from molecular analytics are not yet incorporated into the clinical practice guidelines for HCC management or prediction of recurrence, the evaluation of patients with HCC should include preoperative CT or MRI, which can be used to effectively predict early recurrence and preoperatively stratify these patients. Machine learning techniques with deep learning approaches to extract hidden qualitative and quantitative data from clinical images (including texture analysis) are increasingly being studied in oncology. However, they are challenged by repeatability and reproducibility, and they need a large volume data for adequate stratification.