Testing for AFP in combination with ultrasound improves early liver cancer detection

KEYWORDS:

• Screening
• liver cancer
• ultrasound
• alpha-fetoprotein
• biomarkers

Hepatocellular carcinoma (HCC) is the fastest rising cause of cancer-related death in the United States and a leading cause of death in patients with cirrhosis [1]. Tumor stage at diagnosis is of prognostic and therapeutic significance. Patients with early-stage HCC are eligible for curative therapies such as surgical resection or liver transplantation and can achieve 5-year survival rates of nearly 70% [2]. More advanced HCCs, on the other hand, have a poor prognosis and are primarily managed with palliative measures. Given that symptomatic presentation is almost universally indicative of advanced tumor burden, effective implementation of HCC surveillance during its subclinical course is critical. Several studies have demonstrated that HCC surveillance is associated with significant improvements in early tumor detection, curative treatment receipt, and overall survival in patients with cirrhosis [3].

Although the American Association for the Study of Liver Disease (AASLD) and European Association for the Study of the Liver (EASL) agree on the need for HCC surveillance in at-risk patients including those with cirrhosis, they differ in recommended surveillance tests. Both societies recommend abdominal ultrasound every 6 months but disagree about using the serum biomarker, alpha-fetoprotein (AFP), as an adjunct surveillance test [4,5]. The recently updated AASLD guidelines recommend ultrasound with or without AFP, because the authors believed it was not possible to determine if ultrasound alone or ultrasound with AFP would lead to a greater survival benefit based on current data [4]. In contrast, EASL continues to recommend ultrasound alone for surveillance and discourages the use of AFP due to reports of minimal improvement in sensitivity with potential increase in cost [5].

A number of factors must be examined when considering ultrasound as a surveillance modality. Ultrasound has many advantages including being cheap, readily available, and noninvasive. It does, however, face a number of shortcomings. First, increasing data highlight the variability in ultrasound performance between patients and centers. Ultrasound is well known to be operator dependent, with image quality affected by the experience and skill of the person completing the exam [6]. In the United States, ultrasound exams are performed by trained technicians and frozen ultrasound images are interpreted later by radiologists, whereas in other parts of the world, physicians perform the surveillance exams themselves. Second, ultrasound quality may be impacted by patient characteristics. A recent study found ultrasound quality was inadequate for HCC evaluation in one in five patients, with poor ultrasound quality seen more commonly in obese patients, those with increased liver nodularity, and patients with cirrhosis due to nonalcoholic steatohepatitis (NASH), amongst other characteristics [7]. Third, poor quality or indeterminate exams can lead to screening-related harms including additional use of cross-sectional imaging, such as computed tomography (CT) or magnetic resonance imaging (MRI), and increased patient anxiety [8]. These factors lead to inconsistency in screening adequacy between sites and patients. With the rise of metabolic syndrome and increasing incidence of NASH cirrhosis, ultrasound quality concerns and failure are anticipated to become more problematic. While interventions such as using expert ultrasound operators, developing specific surveillance protocols, and standardizing reporting methods may improve ultrasound quality and reduce surveillance failure, they would not address the patient-related issues. Therefore, surveillance strategies other than, or in addition to, ultrasound should be considered. While CT- or MRI-based surveillance are not as operator dependent and may improve sensitivity for early tumor detection, their routine use is limited by cost-effectiveness and issues such as contrast or radiation exposure. Therefore, use of serum biomarkers, such as AFP, which are thought to be more objective, provides a potential solution to this growing issue.

A randomized controlled trial comparing ultrasound alone and ultrasound with AFP was attempted as part of a French multicenter study [9]. However, there were high rates of AFP contamination in the ultrasound-alone group, precluding any meaningful comparisons. This high rate of contamination not only reflects providers’ hesitance to abandon AFP but also suggests that a randomized controlled trial will not be possible and we will be forced to rely on comparisons within cohort studies. A recent meta-analysis of cohort studies evaluating the benefit of using AFP with ultrasound for HCC surveillance in patients with cirrhosis found a significant increase in sensitivity for early-stage HCC detection when AFP was used with ultrasound (63%, 95% CI 48%–75%) compared to ultrasound alone (45%, 95% CI 30%–62%) [10]. This observed increase in sensitivity was consistently noted across included studies despite a wide variation in reported sensitivities for ultrasound. Specifically, the benefit of AFP was observed in prospective studies, studies in the United States, and studies conducted after the year 2000. Although studies did not directly evaluate downstream benefits, such as improved survival, it is reasonable to infer that increased early tumor detection would translate into a survival benefit given that curative therapies are only available for early tumor stage. The benefit of adding AFP may be particularly notable in patients with increased ultrasound echogenicity and nodularity [11].

Of note, the increased sensitivity for early HCC detection observed in the meta-analysis was accompanied by a small but statistically significant decrease in specificity. Though a rise in AFP levels can be a sign of HCC, it is necessary to recognize that AFP can also be elevated in other clinical scenarios, including other malignancies, or undergo benign variability, contributing to false-positive and indeterminant results. AFP-related false-positive results have been found to be more commonly associated with specific characteristics, including hepatitis C (HCV) infection and elevated alanine aminotransferase (ALT) levels [12,13]. Therefore, the decrease in specificity with AFP may be less noteworthy as the epidemiology of cirrhosis changes to reflect increasing treatment of HCV infection and the rising prevalence of NASH. Further, there are several methods to decrease AFP-related false-positive results, improving its value when implemented in clinical practice. The AFP cutoff for upper limit of normal, usually set at a single threshold of 20ng/mL, can be tailored to cirrhosis etiology, with higher cutoffs in viral than nonviral cirrhosis [12,14]. Alternatively, interpreting AFP trends over time rather than as individual values and using algorithms to adjust for high baseline AFP levels can also better guide clinical interpretation of this biomarker [15,16]. Therefore, further evaluation is needed to discern whether the observed decrease in specificity translates clinically into a significant increase in physical harms (e.g. further diagnostic imaging or biopsy), financial harms (e.g. direct costs of testing or indirect costs of missed work), or psychological harms (e.g. anxiety or depression related to false-positive results). A single-center study characterizing screening-related harms suggested AFP-related physical harms due to false-positives are lower than those related to ultrasound, likely due to providers using clinical judgement when interpreting positive AFP levels but not being able to do so with ultrasound results; however, these results still require validation in other practice settings [8].

The benefit of combining AFP with ultrasound is a proof of concept for the potential value of biomarkers but unlikely to be the long-term solution. Using the two tests, ultrasound and AFP, in combination still misses up to one-third of HCCs at an early stage, underscoring the continued need for improved biomarkers. Several other biomarkers such as des-gamma carboxyprothrombin (DCP) and lectin-bound alpha-fetoprotein (AFP-L3) have been proposed and are undergoing phase III biomarker evaluation [17]. Given the heterogeneity between and even within tumors, a single biomarker is unlikely to be sufficient and a panel of biomarkers may be necessary to achieve adequate sensitivity for early tumor detection. One of the best studied biomarker panels to date, GALAD, combines three biomarkers (AFP, AFP-L3, and DCP) with patient age and gender and has been shown to have high accuracy (AUROC ~0.90) in large case–control studies; however, this panel still requires validation in large cohort studies prior to routine use [18]. While awaiting these data evaluating novel biomarkers, including biomarker panels, we should use the best of our currently validated surveillance tests.

In summary, ultrasound appears to be suboptimal for detecting HCC at an early stage when used alone. The addition of AFP to ultrasound significantly increases the sensitivity for HCC detection at an early stage, which can translate to increased curative treatment eligibility and improved survival. The increase in sensitivity is accompanied by a small, albeit statistically significant, decrease in specificity which could potentially be ameliorated with optimized AFP thresholds, longitudinal evaluation, and clinical interpretation. Additional evaluation is needed to determine whether the decrease in specificity relates to a clinical increase in physical, financial, or psychological harms. Until more optimal imaging modalities or biomarkers are available, we believe AFP should be used with ultrasound in all patients with cirrhosis undergoing HCC surveillance.

Conflicts of Interest

Amit Singal has been on advisory boards and consultant for Wako Diagnostics, Roche, Exact Sciences, and Glycotest.

Declaration of interest

A. Singal has been on advisory boards and consultant for Wako Diagnostics, Roche, Exact Sciences, and Glycotest. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose

Additional information

Funding

This work was conducted with support from Division of Cancer Prevention, National Cancer Institute [RO1 CA222900]. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.