Gastrointestinal Imaging
Original Research
Hepatic Angiomyolipoma Versus Hepatocellular Carcinoma in the Noncirrhotic Liver on Gadoxetic Acid–Enhanced MRI: A Diagnostic Challenge
Audio Available | 
Share
OBJECTIVE. The purpose of this study is to describe the imaging characteristics of hepatic angiomyolipoma (AML) on gadoxetic acid–enhanced MRI and to identify imaging features that are helpful for differentiating it from hepatocellular carcinoma (HCC) in a noncirrhotic liver.
MATERIALS AND METHODS. We retrospectively identified 18 patients with pathologically proven hepatic AMLs who had undergone gadoxetic acid–enhanced MRI between 2008 and 2012. We randomly chose 36 patients with noncirrhotic liver who had a single HCC diagnosed radiologically during the same period. None of the HCCs was of the fibrolamellar variant. Two readers reviewed images in consensus to assess the lesion size, the presence of fat, signal intensity characteristics, enhancement profile, early draining veins, intratumoral vessels, and tumor capsules. The tumor-to-liver contrast ratios were measured. These features and the measurements were compared between the two groups.
RESULTS. AMLs are more commonly found in women (83.3%), whereas HCCs are more common in men (75%) (p < 0.01). The size of AMLs (3.4 cm) and HCCs (4.3 cm) did not differ significantly. Intratumoral fat was identified in both AMLs (50.0%) and HCCs (30.6%). The dynamic enhancement profile (arterial hypervascularity and hypointensity during the delayed phase) was similar qualitatively and quantitatively except for the portal phase. AMLs and HCCs differed significantly with regard to isointensity on DWI (16.7% vs 0.0%; p = 0.03), washout in the portal phase (61.1% vs 88.9%; p = 0.03), early draining veins (27.8% vs 2.8%; p = 0.01), intratumoral vessels (55.6% vs 22.2%; p = 0.03), and presence of capsule (11.1% vs 50.0%; p = 0.01).
CONCLUSION. On gadoxetic acid–enhanced MRI of noncirrhotic liver, AML is often indistinguishable from HCC on the basis of the enhancement profiles. Female sex and some imaging features including DWI could facilitate the differentiation.
Keywords: angiomyolipoma, gadoxetic acid, hepatocellular carcinoma, imaging, liver, MRI
Hepatic angiomyolipoma (AML) is a benign mesenchymal tumor that usually manifests as a solitary tumor in a noncirrhotic liver [1]. In general, hepatic AML is suggested when its fatty component is identified on CT or MRI [2, 3]. However, other hepatic tumors, such as hepatic adenoma, hepatocellular carcinoma (HCC), and, infrequently, even focal nodular hyperplasia, can also have a fat component [2, 3]. In addition, the fat content of hepatic AML can also vary from less than 10% to more than 90% of the tumor volume [4], and, in some instances, the intratumoral fat in AML cannot be easily recognized on CT or even on MRI [5, 6]. Moreover, hepatic AML is still an underrecognized entity, because AML is an uncommon tumor and only approximately 300 cases have been reported from 1976 to date [7]. Thus, AML is seldom included in the differential diagnosis and can easily be misdiagnosed as other hepatic tumors [8–10]. Consequently, it can lead to unnecessary treatment [8–10]. Among hepatic tumors, HCC in noncirrhotic liver shares many imaging features with AML, such as arterial hypervascularity and a fat component. Even though most HCCs occur in cirrhotic livers, up to 10–12% of all HCCs can occur in noncirrhotic livers [11, 12].
Liver MRI, especially when enhanced with gadoxetic acid, has been increasingly used to evaluate, detect, and characterize focal hepatic lesions [13, 14]. The enhancement characteristics of gadoxetic acid do not always coincide with those of extracellular contrast agents with which radiologists have traditionally been familiar [15–19], as has already been shown in both cholangiocarcinomas and hemangiomas [16–19]. This can be attributed to gadoxetic acid's relatively weak extracellular space distribution and inherently high background hepatocyte uptake, because the uptake of gadoxetic acid starts rapidly during the first pass and enhancement of the hepatic parenchyma occurs within 90 seconds after the contrast agent injection [20, 21]. Although the imaging findings of hepatic tumors, including AML, on extracellular contrast agent–enhanced MRI [22–24] are already well known, to our knowledge, there are still inadequate data regarding the characteristic features of AML on gadoxetic acid–enhanced MRI [9, 25].
Thus, the purpose of our study is to describe the imaging characteristics of hepatic AML on gadoxetic acid–enhanced MRI and to identify the helpful imaging features for differentiating it from HCC in a noncirrhotic liver.
| Materials and Methods |   |
|---|
This study was approved by the institutional review board of Asan Medical Center, and patient informed consent was waived because of the retrospective nature of this study.
A search of the electronic database of our institution was performed using the following keywords: “angiomyolipoma” AND “liver” AND [biopsy” OR “hepatectomy] AND “gadoxetic acid (Primovist, Bayer Schering Pharma AG, Berlin, Germany)–enhanced liver MRI” between January 2008 and December 2012. This search revealed 18 patients (three men and 15 women; mean age, 48.9 years; range, 25–74 years) who had both a histologic diagnosis of hepatic AML and had undergone gadoxetic acid–enhanced MRI examination before their biopsy or surgery. All of these patients were finally included in the study. One of the cases of AML included in this study has been described in greater detail in a previous publication [9]. The diagnosis of AML was made as a result of the pathology examination performed after sonography-guided percutaneous biopsy (n = 7) and surgery (n = 11).
To identify a comparison group, a search of our database using the keywords, “hepatocellular carcinoma” AND [“biopsy” OR “hepatectomy”] AND “gadoxetic acid–enhanced liver MRI” before biopsy or surgery during the same study period, revealed 480 consecutive patients. We then excluded 69 patients with multiple HCCs, because AML usually presents as a single hepatic tumor [26, 27]. Because AML usually occurs in noncirrhotic livers, we further excluded 320 patients with morphologic liver cirrhosis. To determine the presence of morphologic liver cirrhosis, MR images were reviewed by one radiologist with 8 years of clinical experience in abdominal imaging. The presence of cirrhosis was diagnosed according to typical radiologic features [28]. Consequently, a total of 91 patients who had a single HCC without definite evidence of morphologic cirrhosis were identified. To create a 1:2 match with the AML group, we randomly selected 36 patients with HCC from these 91 patients using a commercially available random number generator (QuickCalcs, GraphPad). The matching ratio of 1:2 instead of 1:1 was chosen because HCCs are more common than AMLs, and the former ratio would improve the statistical power [29]. Finally, the 36 patients (27 men and nine women; mean age, 54 years; age range, 39–82 years) with a single HCC in a morphologically normal liver and who had undergone gadoxetic acid–enhanced liver MRI were included in our study as the comparison group. Diagnosis of the 36 HCCs was made on pathologic examination after sonography-guided percutaneous biopsy (n = 1) and surgery (n = 35).
One of the authors, who was blinded to the pathology and radiologic diagnosis of the patients, searched the electronic medical records of our medical institution to record the patient demographics, clinical information regarding the presence of tuberous sclerosis [30], underlying liver disease, including hepatitis or alcoholic liver disease, and the preoperative or preprocedural diagnosis as listed in the MRI reports.
All MRI examinations were performed using a 1.5-T system (Magnetom Avanto, Siemens Healthcare) with dedicated six-channel torso array coils. The patients underwent unenhanced MRI (breath-hold dual gradient-echo T1-weighted imaging, breath-hold HASTE T2-weighted imaging, respiratory-triggered turbo spin-echo T2-weighted imaging, respiratory-triggered DWI, and unenhanced fat-suppressed spoiled gradient-echo T1-weighted imaging) and contrast-enhanced T1-weighted imaging. The specific parameters for each sequence are summarized in Table 1. After IV injection of 9–10 mL of gadoxetic acid (9 mL in patients with the timing of the arterial phase determined using a test bolus display method and 10 mL in patients using a real-time bolus method) at a rate of 1 mL/s followed by 20 mL of saline flush, contrast-enhanced dynamic and hepatobiliary phase MR images were obtained using a breath-hold 3D fat-suppressed spoiled gradient-echo T1-weighted sequence (volumetric interpolated breath-hold examination, Siemens Healthcare). Dynamic T1-weighted images were acquired in the arterial phase (determined using a real-time bolus or test bolus display method), portal phase (50 seconds after the initiation of contrast injection), delayed phase (3 minutes after the initiation of contrast agent injection), and the hepatobiliary phase (20 minutes after the initiation of contrast agent injection).
Two abdominal radiologists (one with 15 years of clinical experience and the other with 10 years of clinical experience) reviewed the MR images in consensus. The readers were blinded to the number of patients in each patient category, to the clinical information, and to the histopathologic diagnoses, although they were aware that the study population consisted of patients with either HCC or AML. All of the images were shuffled and randomly reviewed.
We evaluated the following findings: tumor size, location, and margin; the presence of fat; tumor signal intensity (SI) on T1-weighted imaging, T2-weighted imaging, and DWI with a b value of 900 s/mm2; the pattern of contrast enhancement; the presence of an early draining vein during the arterial phase [31]; the presence of prominent intratumoral vessels; and the presence of tumor capsules. A well-defined tumor margin was defined as a clear demarcation of the entire tumor on the hepatobiliary phase images; otherwise, it was defined as an ill-defined margin. The presence of fat in lesions was visually assessed as SI dropout on out-of-phase T1-weighted images compared with in-phase T1-weighted images (microscopic fat) or as SI dropout on fat-saturated T1-weighted images compared with in-phase T1-weighted images or a loss of SI at fat-water interfaces along the borders of tumors on out-of-phase T1-weighted images (macroscopic fat) [32]. On unenhanced and contrast-enhanced images, hepatic lesions were evaluated for predominant SI of the lesions and were visually compared with the SI of the surrounding liver parenchyma. On arterial phase images, an early draining vein was considered positive when a conspicuously dilated or nondilated vessel originating from the tumor with draining to the portal vein or hepatic vein was seen [31]. The presence of tumor capsules was assessed in the delayed phase images by identifying a thin linear-enhancing structure encasing the tumor. The continuity of tumor capsules was also assessed. The radiologists also attempted to determine whether there were other signs of tuberous sclerosis, such as renal AML [33].
Quantitative analysis of the lesions was also performed by the two radiologists. The readers performed ROI measurements of the SI of the reference hepatic lesions and the nontumorous liver parenchyma, respectively, at the same scan level on the arterial, portal, delayed, and hepatobiliary phase images. The SI of the liver was measured by averaging the SI values of three 1.5-cm-diameter circular ROIs positioned in areas devoid of large vessels, focal hepatic lesions, or artifacts. According to the lesion size, ROIs of each lesion were drawn as large as possible, while leaving a 1-mm peripheral margin of the lesion outside the ROI to prevent a partial volume averaging effect. The average SI values within ROIs were then noted. The mean of SI values measured on three consecutive slides was used as the representative SI for each target lesion. The SD of the background noise was measured in the largest possible rectangular ROI placed in the phase-encoding direction outside the abdominal wall. Tumor-to-liver contrast ratios were calculated as follows: (SITE − SILE) / SIN, where SITE is the SI of the tumor on contrast-enhanced images, SILE is the SI of the liver on contrast-enhanced images, and SIN is the SI of the background noise.
We compared the patient demographics, clinical data, preoperative or preprocedural radiologic diagnosis, and the qualitative and quantitative analysis results between AML and HCC. Among the qualitative analysis variables, the sensitivity and specificity of each significant MRI finding was calculated. For quantitative analysis, we compared the tumor-to-liver contrast ratio between AML and HCC on contrast-enhanced images. Statistical comparisons were performed using the Fisher exact test for categoric variables or the Mann-Whitney U test for quantitative variables.
Statistical analysis was performed using SPSS (version 21.0. for Windows, IBM-SPSS). A p value of less than 0.05 was considered to indicate a statistically significant difference.
| Results | |
|---|
No statistically significant difference was found in patient age at the time of the diagnosis with AML (mean [± SD] age, 48.9 ± 11.3 years; range, 25–74 years) and HCC (mean age, 55.8 ± 10.3 years; range, 30–82 years) (p = 0.73). AML was more frequently found in women (83.3%; 15/18), whereas HCC predominantly occurred in men (75%; 27/36) (p < 0.01). There was no patient with tuberous sclerosis according to the diagnostic criteria [30]. Although we excluded patients with morphologic liver cirrhosis on MRI, the presence of underlying liver disease, including hepatitis or alcoholic liver disease, was more frequent in patients with HCC than in those with AML (p < 0.01). In the HCC group, 28 patients had chronic hepatitis B, two patients had chronic hepatitis C, and three patients had alcoholic liver disease. In the AML group, only two patients had chronic hepatitis B. None of the HCCs was of the fibrolamellar variant.
On preoperative or preprocedural MRI reports, AML was included as a differential diagnosis for only 44.4% (8/18) of the patients with AML. For the remaining 55.6% (10/18) of these patients, AML was not suggested even as a differential diagnosis and, instead, HCC was suggested as the diagnosis. AML was more commonly mentioned for patients with fat-containing AML (77.8%; 7/9) than for those with AML not containing an identifiable fat component (11.1%; 1/9) (p = 0.02). For the 36 patients with HCC, HCC was suggested as a differential diagnosis for 100% (36/36) of them.
The MRI findings in AMLs and HCCs are shown Table 2. The tumor size of AMLs (mean, 3.4 cm; range, 1.0–7.8 cm) and HCCs (mean, 4.3 cm; range, 1.2–13.4 cm) at the time of diagnosis and the tumor locations did not differ significantly. In both AMLs and HCCs, the majority of the tumors had well-defined margins.
An intratumoral fat component was identified in both AMLs (9/18; 50.0%) and HCCs (11/36; 30.6%) (p = 0.23). Among the fat-containing lesions, 66.7% (6/9) of AMLs and 90.9% (10/11) of HCCs had microscopic fat that presented as SI dropout on out-of-phase T1-weighted images compared with in-phase T1-weighted images. The SI characteristics on T1-weighted and T2-weighted images did not differ significantly between the two groups. On DWI with a b value of 900 s/mm2, 16.7% (3/18) of the AMLs appeared isointense compared with the hepatic parenchyma, whereas all of the HCCs were hyperintense (p = 0.03).
The dynamic enhancement profiles after gadoxetic acid administration did not differ significantly between the two groups. Both AMLs (100%; 18/18) and HCCs (97.2%; 35/36) showed arterial hypervascularity. On delayed phase images, 83.3% (15/18) of the AMLs and 97.2% (35/36) of the HCCs presented as hypointense lesions, and hepatobiliary phase images revealed that all of the AMLs (18/18; 100%) and nearly all of the HCCs (34/36; 94.4%) were hypointense. In particular, the combination of arterial hypervascularity and hypointensity on the portal or delayed phase, which has been known to be a specific finding for the diagnosis of HCC, was common in both AMLs (15/18; 83.3%) and HCCs (35/36; 97.2%) (p = 0.10). However, on portal phase images, HCC was more commonly seen as a hypointense lesion (32/36; 88.9%) than AML (11/18; 61.1%) (p = 0.03).
Five of 18 AMLs (27.8%) presented with early draining veins on arterial phase images. In contrast, only 2.8% (1/36) of the HCCs were associated with early draining veins (p = 0.01). Early draining veins in all of the five AMLs were the hepatic veins, whereas early draining veins in the patient with HCC was the portal vein. Intratumoral vessels were more frequently seen in AML (10/18; 55.6%) than in HCC (8/36; 22.2%) (p = 0.03). Fifty percent (18/36) of the HCCs showed tumor capsular enhancement on the portal or delayed phase, whereas only 11.1% (2/18) of the AMLs had this finding (p = 0.01) (Figs. 1 and 2). All the tumor capsules both in AMLs and HCCs were continuous. None of the patients with AML showed other signs of tuberous sclerosis.
 View larger version (188K) | Fig. 1A —42-year-old woman with angiomyolipoma. A, Axial T1-weighted out-of-phase image shows small focus of signal loss (arrowhead, A) in peripheral portion of mass, compared with in-phase image (B), thus indicating fat component. |
 View larger version (167K) | Fig. 1B —42-year-old woman with angiomyolipoma. B, Axial T1-weighted out-of-phase image shows small focus of signal loss (arrowhead, A) in peripheral portion of mass, compared with in-phase image (B), thus indicating fat component. |
 View larger version (146K) | Fig. 1C —42-year-old woman with angiomyolipoma. C, Dynamic gadoxetic acid–enhanced MR images show intense enhancement and prominent intratumoral vessels, as well as early draining veins adjacent to tumor on arterial phase (arrowheads, C), washout of tumor on portal phase (D), and hypointensity on delayed phase (E). Mass was hypointense compared with increased signal intensity of surrounding hepatic parenchyma on hepatobiliary phase (F). |
 View larger version (147K) | Fig. 1D —42-year-old woman with angiomyolipoma. D, Dynamic gadoxetic acid–enhanced MR images show intense enhancement and prominent intratumoral vessels, as well as early draining veins adjacent to tumor on arterial phase (arrowheads, C), washout of tumor on portal phase (D), and hypointensity on delayed phase (E). Mass was hypointense compared with increased signal intensity of surrounding hepatic parenchyma on hepatobiliary phase (F). |
 View larger version (150K) | Fig. 1E —42-year-old woman with angiomyolipoma. E, Dynamic gadoxetic acid–enhanced MR images show intense enhancement and prominent intratumoral vessels, as well as early draining veins adjacent to tumor on arterial phase (arrowheads, C), washout of tumor on portal phase (D), and hypointensity on delayed phase (E). Mass was hypointense compared with increased signal intensity of surrounding hepatic parenchyma on hepatobiliary phase (F). |
 View larger version (164K) | Fig. 1F —42-year-old woman with angiomyolipoma. F, Dynamic gadoxetic acid–enhanced MR images show intense enhancement and prominent intratumoral vessels, as well as early draining veins adjacent to tumor on arterial phase (arrowheads, C), washout of tumor on portal phase (D), and hypointensity on delayed phase (E). Mass was hypointense compared with increased signal intensity of surrounding hepatic parenchyma on hepatobiliary phase (F). |
 View larger version (110K) | Fig. 1G —42-year-old woman with angiomyolipoma. G, DW image with b value of 900 s/mm2 reveals hyperintensity of tumor compared with surrounding hepatic parenchyma. |
 View larger version (177K) | Fig. 2A —67-year-old man with hepatocellular carcinoma. A, Axial T1-weighted out-of-phase image shows small focus of signal loss (arrowhead, A) in central portion of mass, compared with in-phase image (B), which means presence of fat. |
 View larger version (155K) | Fig. 2B —67-year-old man with hepatocellular carcinoma. B, Axial T1-weighted out-of-phase image shows small focus of signal loss (arrowhead, A) in central portion of mass, compared with in-phase image (B), which means presence of fat. |
 View larger version (171K) | Fig. 2C —67-year-old man with hepatocellular carcinoma. C, Mass shows strong arterial enhancement (C) and washout and tumor capsular enhancement on portal phase (arrowheads, D). Delayed (E) and hepatobiliary (F) phase images depict hypointensity of mass well. |
 View larger version (173K) | Fig. 2D —67-year-old man with hepatocellular carcinoma. D, Mass shows strong arterial enhancement (C) and washout and tumor capsular enhancement on portal phase (arrowheads, D). Delayed (E) and hepatobiliary (F) phase images depict hypointensity of mass well. |
 View larger version (166K) | Fig. 2E —67-year-old man with hepatocellular carcinoma. E, Mass shows strong arterial enhancement (C) and washout and tumor capsular enhancement on portal phase (arrowheads, D). Delayed (E) and hepatobiliary (F) phase images depict hypointensity of mass well. |
 View larger version (165K) | Fig. 2F —67-year-old man with hepatocellular carcinoma. F, Mass shows strong arterial enhancement (C) and washout and tumor capsular enhancement on portal phase (arrowheads, D). Delayed (E) and hepatobiliary (F) phase images depict hypointensity of mass well. |
 View larger version (145K) | Fig. 2G —67-year-old man with hepatocellular carcinoma. G, On DW image with b value of 900 s/mm2, mass looks hyperintense in comparison with liver. |
Among the aforementioned MRI findings, we selected the imaging findings that showed significant differences between AML and HCC to calculate the sensitivity and specificity (Table 3). The criteria included hypoor isointensity on DWI, no washout (i.e., hyper- or isointensity) in the portal phase, early draining veins, intratumoral vessels, and the absence of tumor capsular enhancement. Four of the five criteria (i.e., DWI, no washout in the portal phase, early draining veins, and intratumoral vessels) showed low sensitivity (16.7–55.6%) and high specificity (77.8–100%).
The tumor-to-liver contrast ratios reflecting the relative SI of tumors compared with the adjacent hepatic parenchyma of both AMLs and HCCs overlapped on contrast-enhanced images and thus also reflected the results of the qualitative analysis (Fig. 3). The mean number of voxels included in the ROIs was 1256 (range, 78–9390 voxels). The tumor-to-liver contrast ratios of the AMLs during the various phases (arterial phase, 22.9 ± 26.1; portal phase, −26.7 ± 25.6; delayed phase, −48.7 ± 36.1; hepatobiliary phase, −103.5 ± 63.3) and those of the HCCs (arterial phase, 10.0 ± 31.3; portal phase, −35.4 ± 31.8; delayed phase, −46.8 ± 45.4; hepatobiliary phase, −75.8 ± 49.0) did not differ statistically significantly (p = 0.16, 0.35, 0.89, and 0.9, respectively).
 View larger version (40K) | Fig. 3 —Graph of tumor-to-liver contrast ratios of angiomyolipoma (AML) and hepatocellular carcinoma (HCC) on each MRI phase. Data points are mean values. |
| Discussion | |
|---|
Because AML and HCC in a noncirrhotic liver shared many imaging features in this study, including the SI characteristics on both unenhanced imaging and gadoxetic acid–enhanced liver MRI, it was challenging to differentiate them. In particular, most AMLs showed arterial hypervascularity and wash-out on the portal phase or hypointensity on the delayed phase, which are known to be the typical imaging features for HCCs on extracellular contrast-enhanced liver MRI. However, we also found some helpful features for distinguishing AMLs from HCCs, including a female predominance without background liver diseases, isointensity in the portal phase, early draining veins, and intratumoral vessels. In contrast, the presence of a tumor capsule and hyperintensity on DWI favored the diagnosis of HCC over that of AML.
In more than half of the patients in our study, AML was misdiagnosed as HCC, which is consistent with the results of previous studies [8–10]. Given their similar MRI features, this is not a surprising result. The absence of an identifiable fat content can make it even more difficult to include AML as a differential diagnosis: AML was suggested as a possible diagnosis in only 11.1% of the AMLs without a detectable fat content. However, a considerable proportion of AMLs (38.9–50%), according to our study and a previous study [24], can present as focal hepatic lesions without detectable intratumoral fat. Therefore, the absence of fat in hepatic focal lesions does not necessarily preclude the possibility of AML.
The enhancement profiles of AML on gadoxetic acid–enhanced MRI overlapped with those of HCC qualitatively as well as quantitatively, except for the portal phase. These enhancement profiles on gadoxetic acid–enhanced MR images differed from those on MR images obtained with conventional extracellular contrast agents [34]. On extracellular contrast-enhanced CT and MRI, persistent enhancement maintained for up to several minutes after the contrast agent injection is one of the features distinguishing AML from HCC [34]. In contrast, when AML was evaluated using gadoxetic acid, hypointensity during the delayed phase was commonly noted in 88.3–92% of the AMLs, according to our and previous work [9, 25]. This finding reflects the unique property of gadoxetic acid, which is rapidly taken up by hepatocytes during the first pass and manifests as the enhancement of nontumorous hepatic parenchyma within 90 seconds after the contrast agent injection [20, 21]. This is in line with the phenomenon of pseudo wash-out, which is already known well in cholangiocarcinomas and hemangiomas [16–19]. Along with arterial hypervascularity and hypointensity during the hepatobiliary phase, the frequent appearance of hypointensity on delayed phase images potentially poses diagnostic challenges in the differential diagnosis of AML from HCC. Although gadoxetic acid can greatly improve the detection and characterization of focal liver lesions because of its distinctive properties as a dual-function contrast agent, we should keep in mind some potential disadvantages of gadoxetic acid in comparison with extracellular contrast agents. In some tumors, including AML, hemangioma, and cholangiocarcinoma, imaging features on delayed (or transitional) phases of gadoxetic acid–enhanced MRI in particular can be confusing [16–19]. In addition, the advantages of gadoxetic acid could be sometimes mitigated by the poor quality of arterial phase imaging due to weak arterial enhancement, inappropriate scan timing, and transient severe motion artifacts [35, 36]. Thus, we need to carefully determine which conventional MRI contrast agent or gadoxetic acid is more useful by weighing the benefits of the hepatobiliary phase and the aforementioned potential pitfalls of gadoxetic acid.
Some of the MRI features were different between the two tumors. Apart from the enhancement pattern of the other phases, the SI in the portal phase differed significantly between AML and HCC. This can be explained by the fact that tumor features on portal phase gadoxetic acid–enhanced MR images are similar to those on extracellular contrast-enhanced MR images, whereas they are different on delayed phase images [37]. Even though the frequency of washout on portal phase images was lower for AML than for HCC, 61.1% of the AMLs still appeared hypointense on the portal phase, which is consistent with the findings of previous studies [25]. This was reflected by the low sensitivity of this criterion. The early draining veins and intratumoral vessels can be helpful features for distinguishing AML from HCC, with high specificity but low sensitivity. Compared with studies that primarily used CT enhanced by extracellular contrast agents and that found frequencies of 80–83.3% for early draining veins and 80–100% for intratumoral vessels [24, 31], the frequency of these features was much lower in the present study. This difference can possibly be attributed to the weaker arterial enhancement of gadoxetic acid and the lower spatial resolution of MRI. The presence of tumor capsule can be an important clue to suggest the possibility of HCCs rather than AMLs, because this feature was found in only 11.1% of AMLs, in contrast to 50.0% of HCCs. This result was in agreement with those of previous studies [24, 31] reporting the absence of tumor capsule in AMLs; one study [25] did report a 45% frequency of tumor capsule in AMLs. The absence of tumor capsule in AML is consistent with the pathologic feature of AMLs as non-encapsulating tumors [24]. Enhanced tumor vessels in the peripheral portion, which are frequently noted in AMLs, are sometimes hard to differentiate from tumor capsules [24]. On DWI, 16.7% of the AMLs showed isointensity, whereas all of the HCCs showed high SI. DWI can provide additional information regarding the cellular environment of hepatic tumors and can consequently be helpful for differentiating benign from malignant tumors [38, 39]. However, the remaining 83.3% of the AMLs also showed high SI on DWI, just as the HCCs did, which was also found in a recent study [25]. Therefore, this criterion also had low sensitivity.
Although we found that some imaging features can be helpful to differentiate AML from HCC, most of the criteria showed limited sensitivities. In addition to the low level of awareness about AML and similar imaging findings, the limited sensitivity of the differential features of AML from HCC can make it easy to overlook AML as part of the differential diagnosis when one encounters an arterial hypervascular hepatic lesion. It would be helpful to increase the awareness of AML in this setting, especially when a patient is female and has no history of underlying liver disease.
Our study has several limitations. First, it included only a limited number of patients. Although our study included, to our knowledge, the largest number of AMLs to date, a study that consists of a sufficient number of AMLs evaluated with gadoxetic acid–enhanced liver MRI needs to be conducted. Second, because our study was retrospective, there was a selection bias. Because we included AML only when it was evaluated with gadoxetic acid–enhanced liver MRI, some AML cases with a prominent fat component and typical imaging features, which can be easily diagnosed as AML on CT, might not have been included in our study. However, considering MRI as a problem-solving tool, this situation can be similar to that in actual clinical practice. Third, we only compared the imaging features of AML and HCC. Although AML is supposed to be considered as a differential diagnosis in patients with hypervascular hepatic tumors, there are other hepatic tumors that should also be considered. Therefore, the sensitivity and specificity of our criteria could have been different if other hypervascular hepatic tumors had been included. Fourth, although we excluded HCCs in patients with liver cirrhosis on the basis of radiologic features, the assessment of liver morphologic features alone was inadequate for excluding the presence of chronic liver disease. This was proved by the results of our study showing that that most patients with HCC were finally confirmed to have chronic liver disease clinically, in spite of the absence of morphologic liver cirrhosis. Given this limitation, the possibility of HCC should be included in the absence of morphologic liver cirrhosis, especially in patients clinically at risk of HCC.
In conclusion, on gadoxetic acid–enhanced MRI of a noncirrhotic liver, AML is easily overlooked and misdiagnosed as HCC. AML can be considered as a differential diagnosis when we encounter a hypervascular focal hepatic lesion in a female patient without underlying liver disease, especially when it shows isointensity on the portal venous phase, the presence of intratumor vessels or draining veins, the absence of a tumor capsule, and isointensity on DWI.
Based on a presentation at the Radiological Society of North America 2013 annual meeting, Chicago, IL.
Supported by grant 2012R1A1A1012731 from the Basic Science Research Program through the National Research Foundation of Korea, funded by the Ministry of Education, Science, and Technology, Korea.
| References | |
|---|
