Optimizing curative management of hepatocellular carcinoma

The goal of curative management of hepatocellular carcinoma is to provide the best chance of remission. However, recurrence rates for both local and distant relapse are high. Patient subgroups at higher risk of these events can be identified based on histological patterns that are closely linked to specific molecular subtypes. Patient outcome has improved with more effective therapeutic strategies thanks to technological advances in surgical techniques and percutaneous ablation. The main goal of controlling the cause of liver disease is to decrease distant/late recurrence and prevent deterioration of hepatic function. Ongoing trials testing the combination of neoadjuvant and/or adjuvant regimens with these procedures as well as routine tumour molecular analysis may modify therapeutic algorithms for hepatocellular carcinoma in the future.


| INTRODUC TI ON
Increasing the rate of HCC patients eligible for curative procedures is key to improving the poor prognosis in these patients. This goal can be achieved by promoting surveillance of at-risk populations with chronic liver disease, in particular those with cirrhosis. Patients with very early-stage (BCLC 0) and early-stage (BCLC A) HCC have preserved liver function and solitary lesions or up to three nodules that are <3 cm in diameter. 1 These patients can benefit from resection, transplantation or percutaneous ablation (PA), with a median survival of between 50% and 70% at five years. 2 However, the term « curative» resection or ablation of HCC in patients with cirrhosis is misleading, since tumour relapse is frequent following these procedures. 3 It is therefore highly important to identify predictors of recurrence and promote continued improvement of procedures and therapeutic strategies to decrease this risk.
Early and late recurrence must be considered independently because they are associated with different risk factors. Early recurrence, arbitrarily defined as occurring within two years following a curative procedure, is related to the development of intrahepatic metastases and is linked to tumour burden including large size, an incomplete tumour capsule and vascular invasion. Late recurrence is more likeky to be due to a de novo carcinogenic process with risk factors related to the extent of patient's liver disease (presence of cirrhosis, persistence of the cause of hepatic insult) and high alpha-foetoprotein (AFP) levels. 4,5 Except for these easy-to-assess characteristics, information obtained by analysis of both tumour and non-tumour biopsy specimens refines the estimates of the risk of recurrence. Figure 1 summarizes these different prognostic factors and the potential actions that can be taken to optimize curative HCC management (except transplantation) in routine practice, which are developed in the present review. This includes advances in surgical or ablation procedures, control of the cause of liver disease and neoadjuvant/adjuvant approaches currently tested in clinical trials.

| REFINEMENT OF PROG NOS IS ACCORD ING TO PATHOLOG I C AL AND MOLECUL AR INFORMATION OBTAINED BY TUMOUR B IOPSY
There are three goals to obtaining a tumour biopsy for the therapeutic strategy of HCC: making a diagnosis, assessing the prognosis and predicting the response to therapy.
As recently described by Calderaro et al, 6 distinct morphological phenotypes of HCC have been found to be associated with the different genetic defects and biological pathways that drive tumour progression.
Next generation sequencing (NGS) using whole-exome, whole genome and RNA sequencing have highlighted the key mutations in the driver genes involved in liver carcinogenesis: somatic alterations in the TERT (2%-12%). 7 Moreover, integrative analysis of transcriptomic data together with genetic alterations has helped identify major molecular subgroups of HCC, G1 to G6. G1 to G3 subgroups are the 'proliferative' subclasses associated with chromosomal instability and TP53 mutations, whereas the G4 to G6 subgroups are 'non-proliferative' subclasses associated with chromosomal stability. G1 to G2 subgroups are associated with Hepatitis B Virus (HBV) infection and AXIN1 mutations with the G1 subgroup are enriched in stem cell features with a high serum AFP level.
G3 subgroups are enriched in FGF19 amplifications and TSC1/2 mutations together with dysregulation of cell cycle genes at the transcriptomic level. The G4 subgroup is associated with a transcriptomic profile close to that of the mature hepatocyte. Finally, G5 to G6 subgroups are strongly associated with the somatic mutations of CTNNB1, coding for β-catenin.
Since NGS is not routinely performed, different morphological phenotypes have been identified linked to the different molecular subclasses at the histological level, and can be classified as follows. 6 A CTNNB1 mutated HCC, observed in 20%-30% of resected HCC, is characterized by the activation of the WNT/ß-catenin pathway, with strong glutamine synthase immune-histochemical expression and corresponding to the G5 to G6 subgroup. These HCC are morphologically well-differentiated with microtrabecular and/or pseudoglandular architectural patterns, intratumour cholestasis and a lack of immune infiltration. This subclass has not been found to influence prognosis.
However, molecular profiling of tumour biopsies from advanced HCC has suggested an altered response to immune checkpoint inhibitors. 8 This could be important for future adjuvant/neoadjuvant strategies (see below).
The second phenotype is the macrotrabecular-massive (MTM) HCC, which is closely associated with the G3 molecular subgroup. This subclass is defined by an architectural pattern composed of large trabeculae (>6 cell-thick) of tumour cells, whatever their cytological aspect (hepatocytic, clear cells, undifferentiated) surrounded by endothelial cells, with empty spaces between these tumour clusters or macrotrabeculae. 9 It is important to note that the prognostic value of this phenotype has been validated in re- A third subtype is the steato-hepatitic subtype observed in about 10% of HCC, and belonging to the G4 molecular subclass. It is characterized by steatosis, cell ballooning, pericellular fibrosis and inflammatory infiltrate. These tumours are often well-differentiated and develop more frequently in patients with metabolic syndrome.
Although steato-hepatitic HCC is rarely associated with microvascular invasion and satellite nodules, no clear-cut relationship with prognosis has been established so far in biopsy samples.
The progenitor subtype, another HCC subtype, can be defined in biopsy samples by the immune-histochemical expression of markers of a biliary lineage such as CK19 and or EpCAM in more than 5% of cells. This subtype may represent 5% of HCC and is associated with TP53 and RPS6KA3 mutations as well as with G1 molecular subclasses. These HCCs express stemness-related markers and have been regularly linked to more aggressive behaviour with increased recurrence and shorter survival. 10 The immune microenvironment also influences prognosis independent from HCC histological subclass, since tumour infiltrating T lymphocytes were associated with a favourable outcome and a lower risk of relapse after surgery. 11 Finally, screening for microvascular invasion has been performed in biopsy specimens, 12 and two immuno-histochemical markers -PIVKA-II and H4K16 -were identified as a major prognosis factor.

| Advances in surgical treatments
The feasibility of liver resection depends on the quality of the parenchyma and the extent of resection, while oncological suitability depends on the biological features of the tumour.
Major hepatectomy (resection of 3 or more Couinaud segments) is only feasible in the absence of portal hypertension and with a MELD score <9. 13 Even when these criteria are met, preliminary portal vein embolization is recommended to assess the liver regeneration in patients with cirrhosis in whom right hepatectomy is planned. As growth of the future liver remnant is slower in patients with cirrhosis, chemoembolization is often performed first to prevent tumour progression while awaiting the benefit of portal embolization. Recently, simultaneous occlusion of both portal inflow and hepatic venous outflow of the planned resection target showed greater and more rapid growth of the future liver remnant than portal embolization alone, although the concomitant increase in liver function may be somewhat delayed compared to the increase in volume. 14 Until the most recent EASL guidelines, portal hypertension was considered to be a contraindication to hepatectomy, but these were revised based on several papers showing that resection can be safely performed in the presence of indirect signs of portal hypertension such as platelet count <100 000/µL and/or splenomegaly, as long as liver function remains normal (MELD score <9). 1 It may be possible to further broaden the criteria by incorporating other factors such as an assessment of liver/spleen stiffness. In the absence of clinically detectable ascites, the risk of liver decompensation after minor hepatectomy has been shown to be very low if liver stiffness as measured by elastometry is <12 kPa. 15 Measurements between 12 and 20 kPa make up a 'grey zone', and measurement of splenic stiffness and/or volume may increase the accuracy of the noninvasive assessment of resectability. open resection for HCC in 6812 patients. 16 Laparoscopy was associated with decreased operative blood loss, lower 30-day morbidity and mortality, and a shorter hospital stay, with a similar rate of R0 resection. There was a trend towards lower HCC recurrence and increased long-term survival with laparoscopy. Other innovative technologies, including intra-operative near-infrared (NIR) imaging fluorescence-guided surgery, are highly promising in the field of HCC. 17 Indocyanine Green (ICG), a vital dye that is extensively used to assess liver function, binds to plasma proteins and emits light with a peak wavelength of around 830 nm when illuminated with NIR light. ICG is excreted in bile, and tends to concentrate in HCC enabling intra-operative detection by the NIR camera. While the ultimate role of this technology must be determined, interest is particularly high among laparoscopic surgeons where there is a need for alternative means of tumour detection because of the absence of tactile feedback. Thus, indications for ablation can be extended to larger tumours (up to 8 cm) with centripetal energy devices, even infiltrative tumours with limited portal invasion (Vp1-3). 20 For standard indications (tumour <5 cm), the no-touch approach can be implemented with centripetal methods, especially multibipolar RFA, which involves inserting applicators outside the tumour (extratumourous method). The local recurrence rate in HCC up to 5 cm (including for those <3 cm) is markedly decreased with no-touch multibipolar RFA compared to monopolar RFA. 19 In addition, no-touch ablation allows safe and effective ablation of a wide spectrum of subcapsular tumours. Irreversible electroportation (IRE), which is also a centripetal ablative method, is the only nonthermal technique. IRE is currently a unique option for curative treatment of central HCC abutting the main bile ducts. In addition, because IRE spares the collagenic skeleton and the microvessels of surrounding non tumourous tissue, this technique appears more suitable than other thermal methods in fragile patients with poor liver function and severe comorbidities. 21

| Control of underlying liver disease
This step is pivotal because the goal is to preserve liver function and potentially decrease the rates of distant/late recurrences.

| HCV infection
It is not clear whether the benefit of viral clearance is related to a potential impact on the oncological process or a reduction in endstage liver disease (ESLD)-related risk of death independent from HCC management. Most of the available data for the impact of HCV eradication on the risk of HCC recurrence have been obtained by meta-analyses or meta-regressions performed in the interferon (IFN) era. However, although there was speculation on the potential benefits of IFN on HCC recurrence, based on its antiviral, anti-inflammatory and anti-angiogenic effects, randomized controlled trials did not demonstrate the efficacy of IFN-based adjuvant therapy in patients undergoing curative HCC procedures. 22 Moreover, the controversy on the potential harmful impact of direct antiviral agents (DAAs) on the risk of HCC recurrence has probably not only limited prescriptions in these otherwise priority candidates for antiviral treatment, but also prevented any definite conclusions from being drawn based on rigorous prospective data. 23 However the findings of prospective studies and the latest reports in large American populations 24 have challenged these hypotheses and recently provided further support of the safety of DAAs in patients who have achieved effective HCC remission. Finally, it is tempting to speculate that longer follow-up may be required to identify any differences in long-term HCC recurrence (more than 2-3 years). This is probably the time needed for the decreased inflammatory and/or fibrotic processes induced by viral suppression to affect liver carcinogenesis.
Overall, achieving a sustained virological response (SVR) in patients with HCV-related cirrhosis is not clearly associated with a modified risk of short-or medium-term tumour recurrence following a curative procedure. Nevertheless, HCV eradication favours optimal HCC management by preventing deterioration of liver function, leading to improved overall survival, whatever the antiviral regimen.
While awaiting confirmation from larger prospective studies with longer follow-up, patients with HCV-related cirrhosis, complicated or not by HCC who are eligible for curative procedures, must be prioritized for access to antiviral treatment.

| HBV infection
Maintained HBV virosuppression in patients with HBV-related cirrhosis has tertiary prophylactic properties in addition to increased overall survival (OS) and decreased HCC occurrence, and is independently associated with a reduced risk of late recurrence after local curativeintent treatment of HCC. At least three meta-analyses including more than 23 000 HBV-related HCC patients, mainly from Asia, showed that nucleoside analogue (NA) therapy significantly reduced the risk of recurrence after surgical resection and improved both disease-free and OS compared to no treatment. In patients with recurrent HCC, the use of HBV antiviral therapy was associated with both preserved liver function at recurrence and an increased proportion of patients eligible for curative HCC treatment. 25

| Other causes of chronic liver disease
Scientific evidence of the benefits of controlling non-virological factors involved in underlying chronic liver disease are scarce, but meaningful. 26 Abstinence from alcohol consumption and intensive care of diabetes mellitus has been related to an improved prognosis in HCC patients. Conversely, inadequate blood glucose maintenance in diabetic patients is a significant risk factor for recurrent HCC and poor survival after curative RFA therapy.

| Inclusion of patients eligible for curative procedures in ongoing neoadjuvant/adjuvant trials
The high rates of intrahepatic local HCC recurrence strongly influence patient prognosis, which could be improved by implementation of neoadjuvant and/or adjuvant strategies. However, many adjuvant therapies have failed to improve recurrence-free (RFS) or overall survival (OS), including sorafenib. 27 However, the results of these trials are limited by selection biases and tyrosine kinase inhibitors (TKI) side-effects.
Several studies have shown that surgery and particularly ablation procedures may significantly alter the immune microenvironment.
Therefore, it could be hypothesized that adding immunotherapy (Io) before and/or after these curative procedures could lead to improved RFS by inhibiting immune related pro-tumour effects. 28 In particular, the rationale for combining RFA and immunotherapy is based on boosting the immune response that is triggered by the necrosis induced by percutaneous treatment. Based on positive results in studies evaluating immune checkpoint inhibitors in advanced HCC, 29 which reported encouraging results and a fair safety profile, a number of studies are testing the combination of immune checkpoint blockade in addition to curative procedures. Table 1 summarizes ongoing Phase II or III trials testing neoadjuvant and/or adjuvant strategies before/following effective curative procedures (resection or PA). Most studies are designed to include patients with a high risk of relapse (multiple tumours or single >3 cm), with RFS as the main endpoint. All trials are being performed in academic centres worldwide and will recruit several hundred patients thus providing a potential opportunity for patients being managed in routine practice.

| PER S PEC TIVE S
Curative management of HCC patients will continue to improve as progress is made in prognosis, systemic therapy and innovation in surgical or ablative procedures. Molecular biology will allow personalized intervention strategies, as several non-tumour and tumour molecular prognostic signatures have been validated in both resected liver and biopsy specimens of small HCC, 30 although they must still be prospectively validated for use in clinical practice. Ultimately, combining precision medicine and new drugs with liver resection or percutaneous ablation in neoadjuvant and/ or adjuvant approaches might improve patient outcomes in the future.

D I SCLOS U R E S
Dr Nahon has received honoraria/grants from Abbvie, AstraZeneca, Bayer, Bristol-Myers Squibb, Gilead, Ipsen and Roche.

CO N FLI C T O F I NTE R E S T
The authors do not have any disclosures to report.

AUTH O R CO NTR I B UTI O N S
All authors contributed to the drafting and critical revision of the manuscript.