EASL Clinical Practice Guidelines

Surveillance

Surveillance consists of the periodic application of a diagnostic test to subjects at risk for developing a given disease. Its usefulness and applicability are influenced by several factors, such as the incidence of the surveyed disease in the target population, the availability of efficient diagnostic test(s) at bearable costs and their acceptability by the target population, and the availability of treatments and their effectiveness [46]. The aim of surveillance is to obtain a reduction in disease-related mortality. This is usually achieved through an early diagnosis (stage migration) that, in turn, enhances the applicability and cost–effectiveness of curative therapies. Stage migration, however, cannot serve as a surrogate for the main end-point, which is patient survival.

HCC is a condition which lends itself to surveillance as at-risk individuals can readily be identified because of the presence of underlying viral hepatitis or other liver diseases. In fact, in the Western world, HCC arises in a cirrhotic background in up to 90% of cases [47], and cirrhosis itself is a progressive disease that affects patient survival. The presence of cirrhosis then influences the chances for anti-tumoral treatment and affects their results, thus rendering early diagnosis of HCC even more crucial. Moreover, many available treatments can have an adverse impact on cirrhosis, and the exact cause of death, which could be either the underlying disease or HCC, cannot be clearly defined in some instances. For this reason, a reduction in overall mortality represents a more appropriate end-point to assess the efficacy of surveillance.

Target populations

Cirrhotic patients

Decision analysis and cost–effectiveness models suggest that an intervention is considered cost-effective if it provides gains of life expectancy of at least 3 months with a cost lower than approximately US$ 50,000 per year of life saved [48]. Cost–effectiveness studies indicate that an incidence of 1.5%/year or greater would warrant surveillance of HCC in cirrhotic patients [49], irrespective of its etiology [[10], [17], [50], [51]]. It may also be possible to identify cirrhotic patients at low risk of developing HCC [[52], [53], [54]] and hence exclude them from surveillance, thereby saving costs although this approach is not proven yet. Conversely, the presence of advanced cirrhosis (Child–Pugh class C) prevents potentially curative therapies from being employed, and thus surveillance is not cost-effective in these patients [[1], [55]]. As an exception, patients on the waiting list for liver transplantation, regardless of the liver functional status, should be screened for HCC in order to detect tumors exceeding conventional criteria and to help define priority policies for transplantation. Finally, although it seems intuitive that surveillance might not be cost-effective above a certain age cut-off, the lack of data prevents the adoption of any specific recommendation.

Non-cirrhotic subjects

Patients with chronic HBV infection are at risk of HCC development even in the absence of cirrhosis. In these cases, the recommended cut-off of annual incidence above which surveillance should be recommended cannot be applied. The cut-off of annual incidence in these patients is ill-defined, albeit expert opinion indicates that it would be warranted if HCC incidence is at least 0.2%/year [[56], [57]]. Thus, cost–benefit modeling is needed in this scenario. The incidence of HCC in adult Asian or African active HBV carriers or with a family history of HCC exceeds this value, whereas HCC incidence ranges from 0.1% to 0.4%/year in Western patients with chronic HBV infection [[58], [59]]. Viral load also appears to increase the risk of developing HCC. In Asian patients, serum HBV-DNA above 10,000 copies/ml was associated with an annual risk above 0.2%/year [18].

Unfortunately, there is scanty and sometimes contradictory information on the incidence of HCC in patients with chronic hepatitis C without cirrhosis. Data from Japan would suggest that patients with mild fibrosis have a yearly HCC incidence of 0.5% [51]. A recent study from the United States has pointed out that HCC does occur in patients with chronic hepatitis C and bridging fibrosis in the absence of cirrhosis (Metavir F3) [12]. The fact that the transition from advanced fibrosis and cirrhosis cannot be accurately defined led the EASL guidelines to recommend surveillance also for patients with bridging fibrosis [1]. This panel also endorses such a policy. In this respect, transient elastography appears to be a promising tool able to stratify patients at different HCC risks [[14], [60]].

Information about the incidence of HCC in patients with non-viral chronic liver disease without cirrhosis, such as non-alcoholic and alcoholic steatohepatitis, autoimmune liver disease, genetic hemochromatosis, α1-antitripsin deficiency, and Wilson disease is limited [[23], [24], [25], [61]]. However, available evidence suggests that HCC usually arises in these contexts once cirrhosis is established [1]. Certainly, patients with metabolic syndrome or non-alcoholic steatohepatitis leading to cirrhosis should undergo surveillance [62], whereas the risk of HCC development is not established in non-cirrhotic individuals.

Treated viral chronic hepatitis

Recent advances in therapy have led to relatively high rates of viral clearance or suppression among those patients being treated for chronic hepatitis B or C. Successful treatment, leading to sustained virological response in chronic hepatitis C, and HBeAg seroconversion or sustained HBV-DNA suppression in chronic hepatitis B, decreases, but does not eliminate the risk of HCC [[63], [64], [65], [66]]. Surveillance should be offered to treated patients with chronic hepatitis B who remain at risk of HCC development due to baseline factors, or to those with HCV-induced advanced fibrosis or cirrhosis, even after achieving sustained virological response.

Surveillance tests

Tests that can be used in HCC surveillance include serological and imaging examinations. The imaging test most widely used for surveillance is ultrasonography (US). US has an acceptable diagnostic accuracy when used as a surveillance test (sensitivity ranging from 58% to 89%; specificity greater than 90%) [[67], [68]]. A recent meta-analysis including 19 studies has showed that US surveillance detected the majority of HCC tumors before they presented clinically, with a pooled sensitivity of 94%. However, US was less effective for detecting early-stage HCC, with a sensitivity of only 63% [69]. In contrast, in a recent Japanese cohort including 1432 patients, careful US surveillance performed by highly skilled operators resulted in an average size of the detected tumors of 1.6 ± 0.6 cm, with less than 2% of the cases exceeding 3 cm [70].

The widespread popularity of US also relies on the absence of risks, non-invasiveness, good acceptance by patients and relatively moderate cost. Nonetheless, US detection of HCC on a cirrhotic background is a challenging issue. Liver cirrhosis is characterized by fibrous septa and regenerative nodules. These features produce a coarse pattern on US, which may impair identification of small tumors. Because of these limitations, the performance of US in early detection of HCC is highly dependent on the expertise of the operator and the quality of the equipment. Thus, special training for ultrasonographers is recommended. The recent introduction of US contrast agents has not proven to increase the ability of US to detect small HCC tumors [71].

There are no data to support the use of multidetector CT or dynamic MR imaging for surveillance. Practical experience suggests that the rate of false-positive results that will trigger further investigation is very high and non-cost-effective. These circumstances are overcome in the setting of the waiting list for liver transplantation where CT scan or MRI are alternatives to US. These techniques should be also considered when obesity, intestinal gas, and chest wall deformity prevent an adequate US assessment. Even in these circumstances, radiation risk due to repeated exposure to CT scan and high cost of MR make debatable their use in long-term surveillance.

Serological tests that have been investigated or are under investigation for early diagnosis of HCC include alpha-fetoprotein (AFP), des-gamma-carboxy prothrombin (DCP) – also known as prothrombin induced by Vitamin K Absence II (PIVKA II) – the ratio of glycosylated AFP (L3 fraction) to total AFP, alpha-fucosidase, and glypican 3 [[12], [72]]. AFP is the most widely tested biomarker in HCC. It is known that persistently elevated AFP levels are a risk factor for HCC development and can be used to help define at-risk populations [73]. Of note is that AFP has been mostly tested in the diagnostic mode rather than for surveillance. This is relevant, since its performance as a diagnostic test cannot be extrapolated to the surveillance setting. As a serological test for surveillance, AFP has a suboptimal performance. One randomized study [74] and one population-based observational study [75] reached opposite results. The latter study provides rationale for testing AFP in special populations or health care environments when US is not readily available [75]. However, when combined with US, AFP levels are only able to provide additional detection in 6–8% of cases not previously identified by US. Reasons for the suboptimal performance of AFP as a serological test in the surveillance mode are twofold. Firstly, fluctuating levels of AFP in patients with cirrhosis might reflect flares of HBV or HCV infection, exacerbation of underlying liver disease or HCC development [76]. Secondly, only a small proportion of tumors at an early stage (10–20%) present with abnormal AFP serum levels, a fact that has been recently correlated with a molecular subclass of aggressive HCCs (S2 class, EpCAM positive) [[77], [78], [79]]. When used as a diagnostic test, AFP levels at a value of 20 ng/ml show good sensitivity but low specificity, whereas at higher cut-offs of 200 ng/ml the sensitivity drops to 22% with high specificity [80].

All other serum markers have usually been evaluated, alone or in combination, in a diagnostic rather than surveillance setting. Moreover, their diagnostic performance has often been assessed at an HCC prevalence remarkably higher than that expected in the context of surveillance [81]. In the latter setting, DCP, measured with a first generation assay, did not offer substantial advantages with respect to AFP [82]. In addition, DCP levels have been associated to portal vein invasion and advanced tumoral stage, a fact that prevents the usage of this marker for early detection [82]. A similar situation occurs with AFP-L3 fraction levels [83]. At present, none of these tests can be recommended to survey patients at risk of developing HCC. Several markers, such as fucosylated proteins, are currently under investigation [84].

In conclusion, US can be seen as the most appropriate test to perform surveillance. The combination with AFP is not recommended, as the 6–8% gain in the detection rate does not counterbalance the increase in false positive results, ultimately leading to an about 80% increase in the cost of each small HCC diagnosed [[69], [85]].

Surveillance efficacy

Two randomized controlled trials have been published on HCC surveillance. In one population-based study cluster randomization (randomizing entire villages) was performed comparing surveillance (US and AFP measurements every 6 months) versus no surveillance in a population of Chinese patients with chronic hepatitis B infection, regardless of the presence of cirrhosis [86]. Despite suboptimal adherence to the surveillance program (55%), HCC-related mortality was reduced by 37% in the surveillance arm as a result of increased applicability of resection in detected cases. The other AFP-based surveillance study carried out in Qidong (China) in high-risk individuals (males, HBsAg+) did not identify differences in overall survival [74].

Other types of evidence include population and non-population-based cohorts and cost–effectiveness analysis, which mostly reinforce the benefits of regular US schemes [[55], [69], [87], [88], [89], [90], [91], [92], [93]]. However, these studies are heterogeneous as far as stage and etiology of liver disease, and surveillance protocols. Moreover, almost all suffer from methodological biases such as lead-time bias (apparent improvement of survival due to an anticipated diagnosis) and length time bias (over-representation of slower-growing tumors). While the latter is unavoidable in this type of study, lead-time bias can be minimized using correction formulas. When this was done, the advantage of surveillance remained [94].

Surveillance interval

The ideal interval of surveillance for HCC should be dictated by two main features: rate of tumor growth up to the limit of its detectability, and tumor incidence in the target population. Based on available knowledge on mean HCC volume doubling time [[87], [88], [89]]; a 6-month interval represents a reasonable choice. Considering, though, that inter-patient variability is so huge, a shorter 3-month interval has been proposed by Japanese guidelines [[90], [95]]. However, the unique randomized study comparing 3 versus 6-month based programs failed to detect any differences [91]. On the other hand, cohort comparisons of 6 versus 12-month schemes provide similar results [[52], [92]], while retrospective studies identified better performance of the 6-month interval in terms of stage migration (small HCC amenable for curative treatments) [96] and survival [97]. Meta-analysis of prospective studies has shown that the pooled sensitivity of US-based surveillance decreases from 70% with the 6-month program to 50% with the annual program [69].

Finally, cost–effectiveness studies have shown that semi-annual US-based surveillance improves quality-adjusted life expectancy at a reasonable cost [98]. In light of available knowledge, a 6-month scheduled surveillance appears the preferable choice. Further trials in this setting would be difficult to implement.

Recall policy

Recall policy is crucial for the success of surveillance procedures. It consists of a defined algorithm to be followed when surveillance tests show an abnormal result. This definition must take into account the ideal target of surveillance, i.e. the identification of HCC at a very early stage (2 cm or less), when radical treatments can be applied with the highest probability of long-term cure [99]. In case of HCC, abnormal US results are either a newly detected focal lesion or a known hepatic lesion that enlarges and/or changes its echo pattern [100].

Pathology studies show that the majority of nodules smaller than 1 cm, that can be detected in a cirrhotic liver, are not HCCs [101]. Thus, a tight follow-up is recommended in these cases (Fig. 2). An accepted rule is to consider any nodule larger than about 1 cm as an abnormal screening result warranting further investigation [56]. These new nodules should trigger the recall strategy for diagnosis with non-invasive or invasive (biopsy) criteria, as described in the section of diagnosis. If a diagnosis cannot be reached with non-invasive criteria due to atypical radiological appearance, then biopsy is recommended. If even biopsy provides inconclusive results, then a tight follow-up every 4 months is recommended. A second biopsy can be considered in case of growth or change in the enhancement pattern. Upon detection of a suspicious nodule, the recommended policy is to evaluate the patient in a referral center with appropriate human and technical resources [56].



Fig. 2 Diagnostic algorithm and recall policy. ∗One imaging technique only recommended in centers of excellence with high-end radiological equipment. ∗∗HCC radiological hallmark: arterial hypervascularity and venous/late phase washout.

View Large Image | View Hi-Res Image | Download PowerPoint Slide