EASL Clinical Practice Guidelines

Resection

Surgery is the mainstay of HCC treatment. Resection and transplantation achieve the best outcomes in well-selected candidates (5-year survival of 60–80%), and compete as the first option in patients with early tumors on an intention-to-treat perspective [[176], [177]]. Hepatic resection is the treatment of choice for HCC in non-cirrhotic patients (5% of cases in the West, 40% in Asia) [[178], [179]], where major resections can be performed with low rates of life-threatening complications and acceptable outcome (5-year survival: 30–50%).

Modern standards of HCC resection in cirrhotic patients are defined by the panel as follows: expected 5-year survival rates of 60%, with a peri-operative mortality of 2–3% and blood transfusion requirements of less than 10% [[102], [157], [180], [181], [182]]. In fact, peri-operative mortality has decreased from 15% in the 1980s to 3–5% in the majority of referral units. Some centers have reported zero peri-operative mortality [[176], [183]]. Blood loss is significantly associated with patient outcome and may be controlled both by selecting patients with preserved liver functional reserve and by applying intermittent inflow occlusion during the hepatic parenchymal transection. Nowadays the selection of candidates for resection has been refined, and both the surgical technique – pre-resection imaging planning, ultrasonic dissector, intermittent Pringle maneuvre, low central venous pressure maintenance, etc. – and immediate post-operative management have been optimized. These strategies have led to a decrease in blood transfusion from 80% to 90% to less than 10% in two decades [183]. In addition, the implementation of anatomic resections according to Couinaud has ensured a surgical approach based on sound oncologic principles, although associated with modest decrease in early recurrence [184]. Anatomic resections aiming at 2 cm margins provide better survival outcome than narrow resection margins <1 cm [185] and are recommended only in case that the maintenance of appropriate function to the remnant liver volume is ensured. Retrospective studies linking anatomic resections and better outcome should be interpreted with caution, due to the propensity of performing wider interventions in patients with well-preserved liver function. Thus, caution should be exercised as the surgical effort is aimed at preservation of adequate hepatic reserve through tailoring of the procedures to individual patients and tumor characteristics – i.e. body size, central versus peripheral location of tumor nodule and solitary large HCC (versus infiltrating tumor types).

Selection of the ideal candidates involves an adequate assessment of the liver functional reserve and tumor extension. The refinement of assessment of liver function has moved from the gross determination of Child–Pugh class to a more sophisticated measurement of indocyanine green retention rate at 15 min (ICG15) [186] or hepatic venous pressure gradient (HVPG) ⩾10 mmHg as a direct measurement of relevant portal hypertension [187]. This concept of portal hypertension as prognostic factor in patients undergoing resection has recently been validated in Asia [182]. Surrogate measures of portal hypertension include two variables: platelet count below 100,000/mm3 associated with splenomegaly, being the spleen size the less among clinical parameters associated with portal hypertension [188]. Platelet count has been recently confirmed as independent predictor of survival in resected HCC cases [189]. In line with these considerations, although the extensive assessment of each component of portal hypertension (HVPG, esophageal varices, splenomegaly and platelet count) is recommended before surgery, platelet count remains the most accessible parameter of portal hypertension available. In practice, selection of patients with HVPG <10 mmHg or absence of surrogates of portal hypertension (esophageal varices, or splenomegaly with platelet count <100.000/mm3) lead to a resectability rate of less than 10% [99]. Expansion of these restrictive criteria by applying MELD score of ⩽10 needs to be prospectively validated with a survival end-point [189].

Some groups apply pre-operative portal vein embolization (PVE) of the branches supplying the portion of the liver to be resected in order to increase the residual liver volume if a major resection is envisioned [[183], [190]]. This approach is associated with a complication rate of 10–20% and occurrence of severe portal hypertension in 1% of cirrhotic patients [191]. However, the effectiveness of PVE in the frame of HCC in cirrhosis has not yet been properly tested in large controlled studies. Finally, an increasing number of data are collected on laparoscopic video-assisted hepatic resection, as an alternative non-invasive approach aimed at preventing liver deterioration compared to open approaches. The positive results reported for specific tumor locations in cohort series [192] need prospective comparison with traditional laparotomic resection before any change in current practice is made.

In patients properly selected according to liver functional status, the main predictors of survival are tumor size, tumor number presence of microsatellites and vascular invasion [176]. Tumor extension should be assessed by latest generation CT scan or MRI. Intraoperative ultrasonography (IOUS) enables the detection of nodules between 0.5 and 1 cm and is considered the standard of care for discarding the presence of additional nodules and to guide anatomical resections [193]. The Japanese Nationwide Survey has shown that a cut-off below 2 cm is an independent predictor of survival in a series of thousands of patients [194]. Five year survival rates for patients with HCC ⩽2 cm was of 66%, compared with 52% for tumors 2–5 cm and 37% for tumors >5 cm. Multinodularity also predicts survival, with 5-year survival rates after resection of single tumors of 57% and 26% for three or more nodules, respectively. Recently, some referral centers reported 5-year survival rates above 50% in patients undergoing resection for multiple tumors fulfiling Milan criteria (up to 3 nodules ⩽3 cm) not suitable for transplantation [[180], [181], [182]]. The positive results reported need further comparison of resection with loco-regional therapies prior to being adopted by these guidelines.

Vascular invasion is a known predictor of recurrence and survival, directly associated with histological differentiation, degree and size of the main nodule. Characteristically, microscopic vascular invasion involves 20% of tumors of 2 cm in diameter, 30–60% of cases in nodules 2–5 cm and up to 60–90% in nodules above 5 cm in size [176]. A more accurate observation of microvascular invasion has led to the identification of invasion of a muscular wall vessel or of more than 1 cm beyond the tumor edge as the 2 worst risk factors for prognosis [157]. Outcome of patients with single resected tumors varied from median survival ∼87 months for patients with no vascular invasion, 38–71 months for those with microvascular invasion with 0–1 risk factors, and 8–12 months for those with microvascular invasion and 2 risk factors or macrovascular invasion. This classification requires external validation [157].

Adjuvant treatments to prevent recurrence

Tumor recurrence complicates 70% of cases at 5 years, reflecting either intrahepatic metastases (true recurrences) or the development of de novo tumors [[161], [157], [180], [181], [182], [195], [196]]. These entities can be differentiated by means of comparative genomic hybridization, integration pattern of hepatitis B virus, DNA fingerprinting using loss of heterozygosity assays, or DNA microarray studies [197]. No clinical definition of both entities has been established, but the cut-off of 2 years has been adopted to grossly classify early and late recurrences [[149], [198]].

Several strategies to prevent and treat recurrence have been tested in the setting of randomized studies. Almost all published RCT have been conducted in Asia. Interferon is the most frequently evaluated drug so far. Different meta-analyses have evaluated the effect of adjuvant interferon treatment [[199], [200], [201]]. In one analysis including 13 studies (9 small RCT) there was a significant improvement in recurrence-free survival with interferon (estimated 3-year RFS of 54% versus 30% of placebo) [200]. Similar results were reported in other studies, in which different patient populations were studied. In the first Western RCT assessing interferon-alpha in 150 patients, negative results were obtained, but a positive trend in preventing de novo late recurrences was identified, providing the rationale for assessing this strategy in future research [181]. Considering the available information, the panel does not recommend adjuvant interferon due to the lack of significant patient numbers and partially conflicting data. Interestingly, recently miR-26 was identified as a potential marker predicting response to adjuvant interferon therapy [135]. Future studies in the adjuvant setting should include this type of molecular marker to more precisely categorize patients responding to adjuvant therapy.

Other strategies tested include chemotherapy, chemoembolization, internal radiation, immune therapies and retinoids. Adjuvant chemoembolization and chemotherapy do not bring any benefit in terms of prevention of relapse [202]. Internal radiation with 131I-labeled lipiodol showed a positive effect in a small trial and cohort study [[203], [204]]. Adoptive immunotherapy with activated lymphocytes with interleukin-2 reduced first recurrence in a trial with 150 patients (3-year recurrence: 33% versus 48% in the control group) [205]. A similar beneficial effect, described with retinoids and vitamin K2 preventing de novo tumors, has not been recently confirmed in the setting of two large RCT studies [[206], [207], [208]]. Overall, according to a recent Cochrane systematic review, 12 RCTs were identified with less than 1000 patients randomized leading to an unclear body of evidence for efficacy of any of the adjuvant and neo-adjuvant protocols reviewed [209]. Thus, none of these strategies are recommended in clinical practice.

Larger trials with a lower risk of systematic error will have to be conducted according to previously reported guidelines [149]. The primary end-point of the studies should be time to recurrence or overall survival. Due to the lack of proven effective treatments, it is justified to randomize patients to an untreated control arm. Selection of patients should be based on the BCLC staging system, and stratification prior to randomization should be done according to tumor size, number of nodules/satellites, and vascular invasion. Due to the nature of these investigations, multi-institutional studies are required. The positive results reported with sorafenib for advanced HCC warrant an international study in the adjuvant setting with this multikinase inhibitor.