Background: Current guidelines recommend ultrasound (US) for hepatocellular carcinoma (HCC) surveillance in cirrhotic livers. However, certain patient and disease characteristics may pose a technical challenge to US surveillance. We aimed to assess predictors of decreased sensitivity of US for detecting HCC.

Methods: At a single, high-volume liver transplant (LT) center in the U.S., all patients with HCC being evaluated for LT receive an abdominal US, which allows for comparison of US sensitivity to recent MRI or CT. In consecutive patients presenting 2012-2015, previously untreated LI-RADS 4 or 5 lesions found on cross-sectional imaging within three months of US were compared with US findings. Multivariable logistic regression models compared potential US sensitivities by patient and nodule characteristics.

Results: Of 536 patients completing a LT evaluation, 288 patients (54%) had no residual tumor after local-regional therapy (LRT) on CT or MRI and were excluded, and 39 (7%) did not meet imaging inclusion criteria. Median MELD of the study cohort (n=209) was 10.6 (IQR 8.8-14.2), median BMI was 28 (IQR 25-31), and 46% had previously received LRT. Moderate ascites was seen in 5% and 9% had severe ascites. Overall per-patient sensitivity of ultrasound compared to CT or MRI was 0.82 (95% CI 0.76-0.87). Patients with BMI ≥30 had an US sensitivity of 0.79 vs. 0.83 for BMI<30 (p=0.5). MELD and presence of moderate/severe ascites also did not affect US sensitivity. US sensitivity was decreased in patients with non-alcoholic steatohepatitis (NASH) vs. other etiologies (0.54 vs. 0.84, p=0.007). Compared to those with other etiologies and MELD < 10, patients with NASH and MELD ≥ 10 were less likely to have their HCC detected by US (sensitivity 0.44 vs. 0.85, p=0.008). Among 296 nodules in total in the cohort, the overall per-nodule sensitivity of US was 0.72 (95% CI 0.67–0.77). US was less likely to detect nodules 1-2cm in size than nodules ≥ 2cm, with a sensitivity of 0.62 vs. 0.79 for larger nodules (p=0.006).

Conclusions: US performed at a high-volume LT center demonstrates suboptimal sensitivity, failing to detect known HCC in nearly 20% and missing nearly 30% of all lesions. US was particularly insensitive in patients with NASH and for nodules < 2cm. HCC surveillance guidelines should consider cross-sectional imaging for patients with cirrhosis and concurrent metabolic risk factors, to promote earlier detection of HCC.

CITATION INFORMATION: Samoylova M, Mehta N, Roberts J, Yao F. Predictors of Ultrasound Failure to Detect Hepatocellular Carcinoma in Cirrhotic Livers. Am J Transplant. 2016;16 (suppl 3).

« Back to 2016 American Transplant Congress