We thank Dr. Ghittoni et al. for their interest in our recent study published in HEPATOLOGY.1 Our study raised major concerns about the use of contrast-enhanced ultrasound (CEUS) as the only dynamic imaging technique for noninvasive diagnosis, because nearly 50% of the intrahepatic cholangiocarcinoma (ICC) lesions showed a homogeneous contrast uptake followed by washout.
The aim of our study was clearly stated in the article: to assess the CEUS features of ICC in the setting of cirrhosis. Thus, we never attempted to evaluate the diagnostic accuracy of CEUS for ICC diagnosis because the only way to conduct such an ambitious assessment is by a prospective study, as we conducted recently for hepatocellular carcinoma (HCC).2 Ghittoni et al. wonder how many other patients might have received a false positive diagnosis of HCC because of the application of the noninvasive criteria that, in the case of CEUS, we have proven may be associated with misdiagnosis. Because of our prospective validation of the noninvasive diagnostic criteria for HCC,2 other groups have repeatedly shown that the strict application of the noninvasive criteria is associated with 100% specificity.3–5 We revealed that CEUS may suggest the malignant nature of the hepatic nodule, but should not pretend to define its final diagnosis. In any case, after potential misdiagnosis by CEUS, staging will be needed prior to treatment, and at that point, our data suggest that magnetic resonance imaging will depict a pattern that is not HCC-specific.6 Upon this finding and according to the American Association for the Study of Liver Diseases guidelines, a pathological confirmation should be applied, because any data raising concerns should prompt the request for a biopsy.
The authors have illustrated their concerns with an interesting case report that raises the potential false diagnosis of HCC by CEUS in histologically proven ICC. Unfortunately, Ghittoni et al. have not shown the images of contrast-enhanced computed tomography scan that is reported to have a weak arterial enhancement. As such, it should have raised some doubt about its HCC origin and likely deserved a biopsy according to the guidelines.
The final conclusion of Ghittoni et al. is that biopsy is the most accurate option for confident diagnosis of every liver nodule. We agree that biopsy has a pivotal role in the diagnostic work-up of liver nodules in a cirrhotic liver. However, the use of biopsy as the preferred diagnostic modality in this setting also has several limitations. First, location of the tumor, clotting disorders, and ascites may prevent needle insertion, and the procedure is not free of risks. It is surprising that in the study of Caturelli et al., the biopsy had an applicability of 100% in a cohort of 294 patients with cirrhosis with nodules <20 mm where no restriction was applied.7 Experienced interventional radiologists acknowledge that there are patients in whom biopsy cannot be obtained, and this is one of the concerns with the data published by Caturelli et al.7 Second, biopsy of small liver nodules in cirrhotic livers is flawed by false negative results due to sampling error or to the unfeasibility of confidently distinguishing between dysplastic changes and well-differentiated HCC. In our experience, biopsy of nodules <20 mm has a false negative rate of around 30% for HCC diagnosis.2 Caturelli et al. claimed that our diagnostic accuracy with biopsy is too low and that their experience was far more positive.8 Nevertheless, they also reported a false negative rate near 11%,7 regardless of the low prevalence of non-HCC diagnosis in their population, which may suggest a selection bias in avoiding nodules highly unlikely to be HCC and/or a potential overdiagnosis of HCC for any reason. Surprisingly, no ICC was diagnosed, despite the authors including a cohort of 294 patients with cirrhosis with newly detected nodules <20 mm. Finally, expert pathologists on liver neoplasia (International Consensus Group for Hepatocellular Neoplasia) have recently highlighted the difficulties for confidently distinguishing dysplastic changes from well-differentiated HCC, showing an interobserver agreement quantified by kappa index among highly expert pathologists of only 0.49 for diagnosing HCC in liver nodules <20 mm obtained by resection. Accordingly, they agreed that biopsy should only be requested in those nodules that do not exhibit the specific vascular profile on dynamic imaging.9
In summary, we hope that these comments have helped to clarify some of the controversies raised by Ghittoni et al. Noninvasive diagnosis of HCC by imaging has repeatedly proven its accuracy, but only if state-of-the-art technique and proper expertise are in place. Our study only pointed out concerns regarding the use of CEUS as the only diagnostic modality for characterization of liver nodules detected within a cirrhotic liver. If imaging is not conclusive for HCC or the clinical background may suggest an alternative diagnosis, biopsy should be mandatory.