Multiple modalities have been used to diagnose hepatic AML. Laboratory tests, viral markers for hepatitis, tumor markers, and liver function have not been proven to be specific or helpful in the diagnosis of hepatic AML. Based on the large studies of hepatic AML [1,13], only 23% to 50% of the patients could have been reliably diagnosed before the operation.
It is because variable imaging appearances are due to the varying proportion of three components: vessels, smooth muscle cells, and adipose tissue.
In this study, fat component was demonstrated by chemical shift MRI techniques in three patients. Identification of signal drop on fat-saturated T1-weighted sequences or opposed-phase chemical shift pulse sequences showed 100% specificity for the intratumoral fat . It is well known that HCC sometimes showed a paradoxically high intensity in the hepatobiliary phase in Gd-EOB-DTPA-enhanced MRI ; conversely, hepatic AML never showed a high intensity. Early enhancement with delayed washout, mimicking HCC, was clearly detected in three patients, but the tumor border was irregular without capsular formation. Besides, early venous return in the arterial or portal phase was detected with various diagnostic imaging in three patients. Kassarjian et al.  reported a classification of hepatic hemangiomas with angiographic findings.
AML often contains a part of a hemangioma-like component inside the tumor. All three patients in this study demonstrated type 2 tumor, high-flow nodules, early filling of veins, and no visible direct shunts with no major vascular anomalies. We were able to identify early venous return in two patients only with CT angiography but in the remaining one with standard enhanced CT. The phenomena of early venous return might be useful in differentiation with the other hepatic tumors and caused early enhancement with delayed washout in the hemangioma-like component. Lately, an early draining vein has been reported to be seen in 73% of AML and was suggested to be useful for distinguishing AML from fat-containing HCC .
DW-MRI is sensitive to molecular diffusion and allows for tissue characterization by probing tissue microstructural changes [9,10]. We believe that examination of DW images in addition to routine abdominal MRI would enhance diagnostic performance of radiologists during evaluation of focal hepatic lesions.
Some studies [17,18] found that renal AML had significantly lower ADC than renal cell carcinoma, cysts, complicated cysts, and overall healthy parenchyma, which stated diffusion-restricting muscle and fat components as the causes for the decreased ADC of AML. Thus, we assessed ADC for liver AML. In this study, to obtain the true diagnosis of the tumor, we measured the ADC value as a quantitative measurement as an adjunct to DW-MRI . We retrospectively examined 240 patients with 195 malignant (HCC, n = 137; liver metastases, n = 44; intrahepatic cholangiocarcinoma, n = 14) and 45 benign liver tumors (hemangioma, n = 37; focal nodular hyperplasia, n = 8). The mean ADC (×10−3 mm2/s) of malignant tumors was 1.19 ± 0.30; for benign tumors, this value was 1.98 ± 0.47. Unfortunately, the values of AML were overlapping with those of other benign and malignant tumors, and we could not find any differences between the Japanese patients and non-Japanese patients.
It has been reported that ADC measurements at three different diffusion gradients may be a complementary tool in the differential diagnosis of malignant and benign tumors . More recently, it has been demonstrated that the ADC of the AML was significantly higher than that of fat-containing HCC (1.92 ± 0.29 × 10−3 mm2/s vs 1.33 ± 0.25 × 10−3 mm2/s, p < 0.001).
AFP is a well-known tumor marker of HCC. Preoperative AFP was positive in one patient who was diagnosed with HCC before hepatic resection. We have reported that preoperative AFP doubling time is a useful predictor of recurrence and prognosis after hepatic resection of HCC , but the AFP level in this patient did not increase before hepatic resection and continued to have an abnormal value postoperatively. Serial measurement is useful to distinguish nonspecific elevation of a tumor marker. In a patient without symptoms and risk factors for liver malignancy (such as chronic hepatitis B or C carrier, liver cirrhosis, alcohol abuse), with normal serum tumor markers and with imaging features suggestive of hepatic AML, conservative treatment with regular surveillance has been recommended . Recently, a malignant AML in the liver was reported. To obtain a definitive diagnosis of a liver tumor mimicking AML, tumor needle biopsy is advocated. But if the tumor is located at the surface of the liver, laparoscopic exploration and biopsy is a preferred approach to avoid seeding of tumor cells.