INTRODUCTION
Hepatocellular carcinoma (HCC) surveillance is associated with early tumor detection and improved survival in patients with cirrhosis.1 Surveillance is performed using semi-annual abdominal ultrasound with or without alpha fetoprotein (AFP); however, this strategy misses over one-third of HCC at an early stage.2 These data highlight a need for novel surveillance strategies with higher accuracy for early HCC detection. GALAD and Doylestown Plus are novel biomarker panels that combine multiple biomarkers with patient demographic and clinical characteristics; both demonstrated promising accuracy in phase II case-control studies;3,4 however, case-control studies can overestimate biomarker performance, highlighting a need for phase III cohort and nested case-control studies.5 Our study aimed to compare multiple biomarkers – including AFP, GALAD, and Doylestown Plus – in a nested case-control study of patients with cirrhosis.
METHODS
Study Population
We leveraged a previously described prospective cohort of patients with Child Pugh A or B cirrhosis who were enrolled into an HCC surveillance program between January 2004 and September 2006 and followed for a median of 2.0 years (Supplemental Methods).6 Serum and plasma were collected at each visit and stored at −80C, without interval thawing. Patients were followed with semi-annual surveillance until incident HCC, liver transplantation, death, or study termination. HCC was defined using AASLD criteria, and early-stage HCC was defined as BCLC stage 0 or A.
Statistical Analysis
Biomarker evaluation was performed using a prospective-specimen-collection, retrospective-blinded-evaluation (PRoBE) design.5 We performed a nested case-control study, with cases defined as patients with HCC and controls as those without incident HCC. Controls were required to have a lack of suspicious hepatic lesions for >1 year to minimize misclassification bias. All assays were performed blinded to HCC vs. non-HCC status, with biomarkers described in Supplemental Methods. GALAD combines gender, age, AFP, AFP-L3, and DCP; while, Doylestown Plus algorithm is comprised of age, logAFP, PEG-precipitated IgG, and fucosylated kininogen.3,4 Biomarker performance was evaluated using 1) area under the receiver operating characteristic curve (AUROC) and 2) patient-level sensitivity with screening-level specificity fixed at 90% - an acceptable level to minimize screening-related harms.7 Our primary outcome was any-stage HCC detection and secondary outcome was early-stage HCC detection. To calculate 95% confidence intervals for biomarker performance, we used a bootstrap procedure among 2000 datasets, constructed by random sampling with replacement. All analyses were conducted using R v4.0.3.
RESULTS
Patient Characteristics
Among 408 patients with cirrhosis, we conducted a nested case-control study with 29 HCC cases and 58 cirrhosis controls. Patient characteristics are detailed in Supplemental Table 1. Median age of patients was 52.0 years, and 60.9% were male. Median Child Pugh score was 7, and the most common cirrhosis etiologies were hepatitis C, alcohol-related liver disease and cryptogenic/NAFLD. Of HCC cases, 58.6% had BCLC 0/A HCC. Among HCC cases, four patients had biomarker assessment at HCC diagnosis, 10 within six months prior to diagnosis, and 13 patients at 6–12 months prior to diagnosis.
Biomarker Performance
Doylestown Plus had the highest accuracy for any-stage HCC detection, with an AUROC of 0.92 and sensitivity of 65.5% (47.6% – 92.6%) with specificity fixed at 90% (Supplemental Figure 1A). If restricted to samples within six months of HCC diagnosis, the highest sensitivity was observed with Doylestown Plus (78.6%, 95%CI 50.0% – 100%) and GALAD (71.4%, 95%CI 28.6% – 92.3%) – both exceeding sensitivity of AFP (57.1%, 95%CI 33.3 – 90.0) (Table 1). Results were consistent when restricting analyses to early-stage HCC (Supplemental Figure 1B), with the highest sensitivity observed for Doylestown Plus (63.2%, 95%CI 37.5% – 93.8%) and GALAD (57.9%, 95%CI 29.4% – 80.0%) with 90% specificity (Table 1).
Table 1:
Comparison of the patient-level true positive rate at the 10% screening-level false positive rate in the overall cohort
| Any-stage HCC | Early-stage HCC | |||
|---|---|---|---|---|
| Any time prior to diagnosis | 0–6 months prior to diagnosis | Any time prior to diagnosis | 0–6 months prior to diagnosis | |
| GALAD | 51.7 (95%CI: 29.–68.8) | 71.4 (95% CI: 28.6–92.3) | 57.9 (95% CI: 29.4–80.0) | 70.0 (95%CI: 18.2–100.0) |
| Doylestown Plus | 65.5 (95%CI: 47.6–92.6) | 78.6 (95%CI: 50.0–100.0) | 63.2 (95%CI: 37.5–93.8) | 80.0 (95%CI: 42.9–100.0) |
| AFP | 37.9 (95%CI: 23.3–68.2) | 57.1 (95%CI: 33.3–90.0) | 47.4 (95%CI: 25.0–75.0) | 70.0 (95%CI: 37.5–100.0) |
| AFP-L3% | 55.2 (95%CI: 28.1–73.1) | 61.5 (95%CI: 26.7–84.6) | 52.6 (95%CI: 17.6–75.0) | 55.6 (95%CI: 11.7–83.3) |
| DCP | 20.7 (95%CI: 0.0–41.4) | 7.1 (95%CI: 0.0–26.7) | 26.3 (95%CI: 0.0–52.6) | 10.0 (95%CI: 0.0–33.3) |
DISCUSSION
Our study results highlight the potential for novel biomarker panels including Doylestown Plus and GALAD to improve early HCC detection, with sensitivity >70% for any-stage HCC and >55% for early-stage HCC. The high accuracy of these biomarker panels likely relates to inclusion of multiple biomarkers – reflecting intra-tumoral heterogeneity – as well as demographics (age and gender) that are associated with higher HCC risk. Performance of these biomarker panels compares favorably to ultrasound alone, which had a sensitivity below 50% in a recent systematic review.2 Blood-based biomarkers also have high patient acceptance and are easy to implement in clinical practice so would likely also address surveillance underuse and thereby further increase surveillance effectiveness.8 While our study provides important nested case-control data derived from a prospective cohort, our results are limited by small samples sizes and wide confidence intervals. Phase III studies including Hepatocellular Carcinoma Early Detection Strategy (HEDS) and the Texas HCC Consortium studies are maturing and should provide further data for these novel biomarkers in the future.
Supplementary Material
Financial Source:
This study was conducted with support from NIH U01 CA230694 and R01 222900. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The funding agencies had no role in design and conduct of the study; collection, management, analysis, and interpretation of the data; or preparation of the manuscript.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflicts of Interest: Amit Singal has served as a consultant or on advisory boards for Genentech, Bayer, Eisai, Bristol Myers Squibb, Exelixis, AstraZeneca, FujiFilm Wako Diagnostics, Exact Sciences, Roche, Glycotest, GRAIL, and TARGET RWE. Neehar Parikh has served as a consultant or on advisory boards for Genentech, Bayer, Bristol Myers Squibb, Fujifilm Wako Diagnostics, Exact Sciences, and Eli Lilly. Jorge Marrero has served as a consultant for Glycotest. Anand Mehta has served as a consultant for Glycotest. None of the other authors have relevant conflicts to disclose.
REFERENCES
- 1.Singal AG, Pillai A, Tiro JA. Early Detection, Curative Treatment, and Survival Rates for HCC Surveillance in Patients with Cirrhosis: A meta-analysis. PLOS Medicine 2014; 11(4): e1001624. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Tzartzeva K, Obi J, Rich NE, Parikh ND, Marrero JA, et al. Surveillance imaging and alpha fetoprotein for early detection of hepatocellular carcinoma in cirrhosis: A meta-analysis. Gastroenterology 2018; 154(6): 1706–18.e1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Berhane S, Toyoda H, Tada T, Kumada T, Kagebayashi C, Satomura S, et al. Role of the GALAD and BALAD-2 serologic models in diagnosis of hepatocellular carcinoma and prediction of survival in patients. Clinical Gastroenterology and Hepatology 2016;14(6):875–86. e6. [DOI] [PubMed] [Google Scholar]
- 4.Wang M, Sanda M, Comunale MA, Harrera H, et al. Changes in the glycosylation of kininogen and the development of a kininogen-based algorithm for the early detection of HCC. Cancer Epi Biomarkers Prevention 2017; 26(5): 795–803. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Singal AG, Hoshida Y, Pinato DJ, Marrero JA, Nault J-C, et al. International Liver Cancer Association (ILCA) White Paper on Biomarker Development for Hepatocellular Carcinoma. Gastroenterology 2021 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Singal AG, Conjeevaram H, Fu S, Volk ML, Fontana RJ, et al. Effectiveness of Hepatocellular Carcinoma Surveillance in Patients with Cirrhosis. Cancer Epidemiology Biomarkers & Prevention 2012; 21(5): 793–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Singal AG, Patihandia S, Obi J, Fullington H, Parikh N, et al. Benefits and harms of hepatocellular carcinoma surveillance in a prospective cohort of patients with cirrhosis. Clinical Gastroenterology Hepatology 2021 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Woolen SA, Singal AG, Davenport MS, Troost JP, Khalatbari S, et al. Patient Preferences for Hepatocellular Carcinoma Surveillance Parameters: A MultiCenter Conjoint Study. Clinical Gastroenterology Hepatology 2021 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.