Surveillance for Hepatocellular Carcinoma in Cirrhosis: End of Monopoly for Serum Alpha Fetoprotein
Hepatocellular carcinoma (HCC) is the final common pathway for all chronic liver diseases. HCC surveillance strategies are aimed at early detection to improve survival after a cancer diagnosis. The essence of HCC surveillance consists of 6-monthly ultrasonography of the liver, whereas serum measurement of alpha fetoprotein (AFP) is inconsistently recommended across different clinical practice guidelines. However, the sensitivity of ultrasonography for early-stage HCC was only 47%, and only improved to 63% when combined with AFP measurement.1 Another adversity is that the estimated use of imaging-based surveillance for HCC was only about 24%.2 Hence, novel scores or biomarkers are needed to improve the effectiveness of HCC surveillance. Des-γ-carboxy prothrombin (DCP), also known as protein-induced by vitamin K absence or antagonist-II (PIVKA-II), and lectin-bound alpha fetoprotein (AFP-L3%) are 2 other blood-based biomarker for HCC. The superiority in sensitivity and specificity for DCP alone or in combination with AFP has been described.3,4 In fact, an expert panel regarding the use of DCP/ PIVKA-II was formed in the Asia-Pacific region, where >70% of the global incident HCC cases is found.5 Consensus was reached that “PIVKA-II in combination with AFP improves the detection of HCC, including small-sized tumors (≤3cm) compared to either biomarker alone” and “PIVKA-II is valuable in the detection of HCC in AFP-negative HCC patients.”6 So far, the data in the literature mainly came from retrospective studies, and despite much enthusiasm in the East, studies in non-Asia countries/regions failed to demonstrate superiority of these novel biomarkers over AFP,7,8 partly owing to differences in study design and heterogeneous patient populations. The GALAD algorithm is a novel score that includes both DCP and AFP-L3% and consists of 5 parameters: Gender, Age, AFP-L3, AFP, and DCP, and has shown excellent accuracy in diagnosis of early HCC in patients with metabolic dysfunction–associated steatotic liver disease (MASLD).9 A similar score, the GAAD (Gender, Age, AFP and DCP) algorithm, involves all parameters of GALAD except AFP-L3 and demonstrates similar diagnostic performance as GALAD.10 The performance of these scores in non-Asian patients is unknown.
In this issue of Gastroenterology, in a multicenter study reported by Marsh et al,11 1558 patients with cirrhosis from 7 sites in the United States were recruited and prospectively followed up to evaluate the performance characteristics of the GALAD score and its components (AFP-L3 and/or DCP) to detect HCC within 12 months. Most patients were non-Hispanic White (80.6%) and the most common etiology of chronic liver disease was hepatitis C virus infection (42.4%), followed by MASLD (26.3%) and alcohol-related liver disease (14.7%). The majority of patients were in the compensated range of cirrhosis (73.9%). HCC surveillance was conducted using ultrasonography, computed tomography, or magnetic resonance imaging. Over a median follow-up of 2.2 years, 109 HCCs were detected (annual incidence rate of 2.4%). Serial samples (within 3 months, 6 months, and 12 months of HCC diagnosis) were available in 107 patients who were subsequently diagnosed with HCC and 1123 patients without subsequent diagnosis of HCC. With a panel of AFP, AFP-L3%, and DCP, the authors evaluated the performance characteristics of all submodels and individual biomarkers. The area under the receiver-operating characteristic curve for HCC within 12 months was 0.66 for AFP, 0.71 for DCP, 0.71 for GAA (gender + age + AFP), 0.74 for GAD (gender + age + DCP), 0.76 for GAAD, and 0.78 for GALAD. When focusing on the identification of early HCC within 12 months, the area under the receiver-operating characteristic curve (sensitivity and specificity) was 0.64 (39% and 82%) for AFP at the cut-off of 6.45 ng/mL and 0.77 (60% and 82%) for GALAD at the cut-off of −1.36. The data suggested incremental utility for each component to identify early HCC, supporting the combined use of all 3 biomarkers and clinical parameters, and verified the improved HCC detection with AFP-L3%, DCP, and the derived algorithms in patients with a different ethnic background.
The current article is the first prospective study to evaluate the role of GALAD at various timepoints preceding HCC diagnosis in patients with chronic liver disease. The cohort consisted exclusively of patients with cirrhosis, which led to higher event detection but the results are not generalizable to noncirrhotic patients. MASLD represented 26.3% in this cohort, which is helpful to fill the knowledge gap in the existing data that has been largely derived from viral hepatitis–related HCC in studies conducted in Asian countries/regions. Of note, the subgroup analysis showed less benefit for GALAD over AFP for nonviral etiologies compared with viral etiologies of HCC, but the mechanisms of which are unknown. The authors also evaluated the performance by setting various thresholds of sensitivity and specificity. Interestingly, the addition of AFP-L3% (GALAD) indeed has led to additional benefit when compared with GAAD, with a 5% increase in sensitivity of HCC detection within 12 months of diagnosis; in contrast to what was reported,12 which led to discontinuation of Elecsys AFP-L3% and the GALAD Algorithm developed by Roche Diagnostics.13 It is worth noting that the supplier of the biomarkers in this study was Fujifilm Wako Diagnostics, taking into consideration that different assays from various manufacturers have demonstrated unequal performance14 and the heterogeneous cut-off scores used.
In summary, Marsh et al11 validated the enhanced sensitivity of GALAD score – a combined panel of biomarkers and readily available clinical parameters – for early detection of HCC in cirrhosis compared with AFP. With available longitudinal samples, it would be of interest to explore the role of dynamic change in AFP-L3%, DCP, or the derived scores in the preceding 3–12 months in prediction of impending HCC. The specificity of GALAD was still not optimal. What were the reasons for elevated GALAD in those patients without subsequent diagnosis of HCC? Which component of GALAD was responsible for the falsely elevated GALAD – age, gender, AFP, AFP-L3, or DCP? Would a longer follow-up unveil undiagnosed HCC beyond the follow-up period of the current study? For those with HCC who did not have abnormal GALAD, were there any clinical phenotypes that can inform the risk of having falsely normal GALAD? Further stratification into subgroups with normal AFP and elevated AFP would be helpful to identify the best population for judicious use of HCC biomarkers.
