Early Detection of Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is the fastest growing cause of cancer-related death in the United States.
To address the magnitude of this problem, it is critically important to identify those at high risk for HCC and
institute effective surveillance strategies for early diagnosis. Liver cirrhosis is the main risk factor for HCC. Bi-
annual ultrasound and α-fetoprotein remains the surveillance modality most frequently used in patients with
cirrhosis, despite very low sensitivity and specificity. Our goals are to identify a blood-based model for risk
stratification in patients with cirrhosis, as well as an integrated blood-based and liver imaging model to
optimize early HCC detection in high-risk patients. During the first grant period, we developed a multi-center
prospective cohort of patients with cirrhosis under contrast MRI surveillance. Such cohort provides a unique
opportunity to study blood biomarkers and imaging features on clinical material from patients rigorously
classified as having a very early disease in a surveillance setting. Longitudinal collection of paired blood
samples and MRIs from these patients is particularly valuable in assessing how early blood markers and
imaging features become positive during the period when lesions are observed to obtain a diagnosis of HCC.
To date, 912 cirrhotic patients have been enrolled and 2590 paired blood samples and MRIs have been
collected. During follow-up, 63 patients developed HCC and 212 patients had detectable lesion(s). In parallel,
we have identified in plasma and exosomes, proteins and metabolites for HCC risk prediction and early
detection. We also developed quantitative imaging and artificial intelligence (AI)-based methods to analyze
imaging scans of patients with liver cancers. We demonstrated how voxel-wise enhancement pattern
mapping (EPM) can improve the contrast-to-noise ratio in CT scans. We extended this finding to MRIs for
patients with HCC, including patients in our prospective cohort. Differences in EPM signals from pre-
diagnostic MRIs to diagnostic MRIs may improve early detection and lesion characterization. Our AI-based
tools complement the EPM algorithm by providing high-throughput tools to process the thousands of MRIs
from our patient cohort in an efficient and accurate manner. In this competing renewal, we will extend the
existing cohort and further evaluate the performance of these novel blood and liver MRI markers. We will
determine longitudinal changes and evaluate their capacity to detect preclinical disease. We will identify the
panel of markers that best predict HCC development and that could therefore have utility in risk assessment
and early detection of HCC. This proposal achieves in one study two major goals: i) early detection and ii)
characterization of tumors when biomarker becomes positive. The impact is multiple: spare patients from
unnecessary imaging tests; identify high-risk patients and trigger the decision to perform MRI for surveillance
instead of ultrasound; detect lesions at an early stage allowing for curative treatment. Together, these clinical
applications would significantly reduce the cost of HCC surveillance and improve survival of HCC patients.

Grant Number: 5R01CA195524-09
NIH Institute/Center: NIH
Principal Investigator: LAURA BERETTA