Results from a preliminary study have demonstrated that nearly one in 10 hepatocellular carcinoma (HCC) lesions were not detected on cross-sectional imaging using computed tomography (CT) or magnetic resonance imaging (MRI) when compared to lipiodol staining. Presenting these new data at the Society of Interventional Oncology (SIO) annual meeting (4–8 February, Savannah, USA), Emilio Cavazos-Escobar (University of Texas Health San Antonio, San Antonio, USA) emphasised that, although a “diagnostic discrepancy” does exist, the positive predictive value of cross-sectional imaging remained high, meaning lesions are more often than not correctly identified as HCC.
The motivation for their investigation arose after Cavazos- Escobar and colleagues began noticing unexpected lesions appearing on routine ‘stain-and-ablate’ procedures, he noted.
“Typically, the patient will receive cross-sectional imaging—either four-phase or three-phase CT or MRI with or without contrast. Then, if the intention is to treat the lesion, we schedule a clinic appointment where a targeted ultrasound is performed,” Cavazos-Escobar detailed. He continued that, in some cases, if the lesion is not apparent on ultrasound or is poorly conspicuous on noncontrast CT, they schedule the patient for a stain-and-ablate procedure to better delineate the lesion’s position and assess adjacent anatomical structures that may be involved.
To compare the diagnostic performance of cross-sectional imaging versus lipiodol staining, Cavazos-Escobar et al performed a retrospective review of 181 patients who underwent stain-and-ablate procedures between August 2020 and September 2025. Within the cohort of 181 patients, 282 lesions were identified by lipiodol staining. Out of the total of 282, 246 lesions were correctly detected and classified as LR-5—using the Liver Imaging Reporting and Data System (LI-RADS)—via cross-sectional imaging, for an overall sensitivity of 87.2%.
“Interestingly, 27 (9.6%) lesions were not detected altogether on cross-sectional imaging, and thirteen lesions were misclassified as false positives”, Cavazos-Escobar explained. “This meant that these were labelled LR-5, but when the patient was taken for lipiodol staining, there was no enhancement of the supposed tumor and therefore these were classified as not true lesions”. Additionally, nine lesions were misclassified as false negatives and were either classified as LR-3 or lower which eventually did stain with lipiodol.
Speaking at this year’s SIO annual meeting, Cavazos-Escobar relayed that the positive predictive value for cross-sectional imaging was 95%. “This means that when CT or MRI detects a lesion, there’s a good likelihood that it will be correctly identified as HCC,” he stated.
“It’s not uncommon to see patients whom we initially thought only had one lesion, inject lipiodol, and then all of a sudden, the patient has three lesions. We wanted to further investigate that discrepancy at our institution. There are some data available, but they’re very scarce, and there isn’t much research involving the use of lipiodol, so we’re hoping the present analysis meaningfully adds to the literature.”
Summarising, Cavazos-Escobar stated that a “diagnostic discrepancy does exist, which suggests that CT or MRI may underestimate tumour burden in some patients”.
