Myomatous hepatic angiomyolipoma: imaging findings in 14 cases with radiological–pathological correlation and review of the literature

Abstract

Objective:

To display and analyse the imaging features of myomatous hepatic angiomyolipomas (MHAMLs).

Methods:

The imaging features (CT = 9; MRI = 10; ultrasound = 10; contrast-enhanced ultrasound = 4) of 14 patients with pathologically proven MHAMLs were reviewed retrospectively.

Results:

MHAMLs were surgically resected in the 14 patients (10 females and 4 males; age, 27–64 years; mean, 45 years), all of whom had negative hepatitis markers and were positive for the immunohistochemical stain homatropine methylbromide-45. The tumours were solitary and well defined, and ranged in size from 1.9 to 9.1 cm (mean, 5.7 cm). On dynamic contrast-enhanced CT, MRI and ultrasound scans, all tumours showed fast strong enhancement in the arterial phase and moderate washout in the portal venous and delayed phases, and the greater portions of the tumours were slightly lower than the surrounding hepatic parenchyma. In some cases, a small area of prolonged or increasing enhancement in the tumour was recognized in the delayed phase. Early draining vessels to the portal vein or hepatic vein could be seen in some cases. However, no capsular signs could be confidently identified in the delayed phase. Haemorrhagic cavities were recognized in two cases, and nodular low-intensity areas in the tumours on T₂ weighted imaging that showed slow and faint enhancement on dynamic scans were seen in two cases. However, no necrosis was identified.

Conclusion:

Dynamic enhanced imaging studies revealed some specific features of MHAMLs that distinguish them from other hypervascular hepatic tumours, especially when combined with clinical features. Familiarity with imaging and clinical features of MHAMLs could avoid unnecessary surgical resection of these generally benign tumours.

Advances in knowledge:

This article systematically describes the imaging features of MHAMLs.

Angiomyolipoma (AML) is considered a mesenchymal neoplasm and was classified under the group of perivascular epithelioid cell tumours in the World Health Organization classification of tumours in 2002.¹ The tumour consists of three components: smooth muscle cells (SMCs), adipose tissue and vessels. SMCs that are positive on homatropine methylbromide-45 (HMB-45) staining are considered a key component of the histological diagnosis of AMLs, as they have been postulated to be neoplastic mesenchymal cells. Hepatic AML (HAML) was first described by Ishak² in 1976 and was categorized into four subtypes: mixed (the most common type), lipomatous (>70% fat), myomatous (<10% fat) and angiomatous.^3,4 HAML is generally managed as a benign tumour clinically, although malignant cases are occasionally reported.^5–8 If the diagnosis of HAML can be comfortably made on imaging examinations, the tumour can be followed up using imaging modalities and an ultrasound-guided biopsy could be suggested. Surgical resection would then only be necessary in patients with symptoms, in those at risk of rupture or haemorrhage and in those with signs of malignancy.

Both the mixed and lipomatous types of HAMLs are relatively easy to diagnose with modern imaging techniques such as CT, MRI and ultrasound because of the mature fatty components of the tumour. The angiomatous type is also relatively easy to diagnose because of the malformed vascular components. However, the myomatous type of HAML (MHAML), which is rare, is often confused with other hypervascular tumours, such as hepatocellular carcinomas (HCCs), focal nodular hyperplasia (FNH) and adenomas, and an imprecise diagnosis could lead to unnecessary surgical resections. Some authors have described the imaging features of this rare benign tumour. For example, Jeon et al⁹ reported that the presence of early venous draining vessels, vein connections with prominent tumour vessels and absent capsular enhancement in contrast-enhanced triple-phase CT images were highly related to a diagnosis of MHAMLs. Therefore, familiarity with the imaging and clinical features of MHAMLs could facilitate a correct diagnosis and avoid unnecessary surgical resections. In this article, we describe our observations of 14 pathologically proven cases of MHAML.

METHODS AND MATERIALS

Patient selection

This retrospective study was approved by the ethics committee of our hospital. A total of 23 patients with a diagnosis of HAMLs were located from a search of the computer database at the hospital over a period of 7 years from March 2005 to April 2012. Among these patients, nine were excluded from the study for the following reasons: a pre-operative imaging examination had not been performed in two cases and seven had non-myomatous types of HAMLs. The remaining 14 patients (10 females and 4 males) all had histological data available and were strongly positive on HMB-45 staining.

CT studies

Multislice CT was performed in nine patients, eight of whom underwent 64-slice CT scans (LightSpeed™ Ultra; GE Healthcare, Milwaukee, WI, or Toshiba, Nasu, Japan), while one underwent an 8-slice CT scan (LightSpeed™ Pro; GE Healthcare). The scan parameters included 120 kV, 340 mA, a section thickness of 5 mm, reconstruction thickness of 1.25 mm and a field of view of 248 × 330 mm. An intravenous, non-iron-type contrast preparation (100 ml) was used in all patients. Enhanced CT was performed in the arterial, portal venous and delayed phases with delay times of 25, 65 and 120 s, respectively, after an injection of the intravenous contrast agent.

The CT scan was performed once in seven patients and repeated three times in two patients before surgical resectioning.

MRI studies

MRI was performed once in 10 patients using a 3.0-T MR superconductivity scanner (Signa Excite®; GE Healthcare). The scan parameters included an 8-channel body coil, a section thickness of 5 mm, layer spacing of 0.8 mm, T₁ weighted imaging (WI) with double-echo chemical shift imaging, T₂ WI and T₂ WI with fat suppression and diffusion-weighted imaging (b = 0, 800 mm² s⁻¹). The contrast agent used was gadolinium diethylenetriamine pentaacetic acid, 20 ml of which was injected through a 20-gauge intravenous cannula into the antecubital vein in a bolus fashion, followed by a flush of 10 ml of 0.9% sodium chloride solution. Enhanced MRI was performed using LAVA (liver acquisition with volume acceleration; GE Healthcare). Image acquisition started at 10 s after administration of the contrast agent. Each imaging acquisition period was 10 s long. The time interval between two imaging acquisitions was 10 s, and there were a total of six imaging acquisition periods.

Ultrasound and contrast-enhanced ultrasound examinations

Three ultrasound records were available in 10 patients and were performed with Logiq 9 and Logiq 700 ultrasound machines (GE Healthcare) equipped with a 4V1 vector transducer with a frequency range of 1.0–4.0 MHz, as well as an IU 22 ultrasound machine (Philips Medical Systems, Tokyo, Japan) equipped with a 375BT convex transducer with a frequency range of 1.9–6.0 MHz. The contrast agent used was SonoVue® (Bracco, Milan, Italy), a sulfur hexafluoride-filled microbubble preparation. A 2.4-ml bolus of this contrast agent was injected through a 20-gauge intravenous cannula into the antecubital vein, followed by a flush of 5 ml 0.9% sodium chloride solution. 10 of the 14 patients underwent two or more imaging checks.

Data analysis

All imaging data were directly interfaced with the picture archiving and communication system (Carestream Health, Inc., Onex Corporation, Toronto, Canada) and displayed by two abdominal radiologists (with 25 and 10 years' experience, respectively). The radiologists had knowledge of the pathological diagnosis, and the pre-operative image diagnosis was recorded. A consensus interpretation of imaging findings was accepted. The following items were analysed in the imaging scans: (1) the number of lesions; (2) the lesions' locations (left, right or caudate lobe); (3) margins (well defined or ill defined); (4) size; (5) the presence of tortuous tumour vessels (conspicuous dilated or non-dilated vessels) around or inside the tumour; (6) the presence of an early draining vein (considered a positive finding when a conspicuous dilated or non-dilated vessel originating from the tumour and draining to the portal vein, hepatic vein or inferior vena cava was seen); (7) peripheral rim enhancement (defined as a thin hyperattenuating/signal hyperintensity rind that surrounded at least half of the tumour in the delayed phase); (8) the enhancement pattern [either (i) a fast-in and fast-out mode (defined as tumour enhancement in the arterial phase, followed by hypoattenuation relative to the surrounding hepatic parenchyma in the portal venous or delayed phase) or (ii) a fast-in and slow-out mode (defined as tumour enhancement in the arterial phase, followed by hyperattenuation or isoattenuation relative to the surrounding hepatic parenchyma in the portal venous phase and, finally, hypoattenuation in the delayed phase)]; and, (9) the presence of capsules, fat, haemorrhage, cystic degeneration, calcification and necrosis.

RESULTS

Clinical features

The ages of the patients ranged from 27 to 64 years with a mean of 45 years. Two patients presented with symptoms of mild stomach pain or distension, but the tumours in the other 12 patients were without symptoms. Tests for hepatitis B surface antigen were negative in 10 and positive in 2 cases, and liver function tests and serum tumour markers such as alpha-fetoprotein (AFP), carcinoembryonic antigen and carbohydrate antigen 19-9 were within normal ranges in all 14 patients. None had associated tuberous sclerosis or renal AML.

10 of the 14 patients were examined with 2 or more imaging modalities, such as CT, MRI and ultrasound. 12 patients underwent surgical resection within 2 weeks after the diagnostic imaging studies, while 2 patients underwent surgery at 9 months and 2 years, respectively, after the initial imaging detection of the tumours. One of these patients was pre-operatively diagnosed with HCC (Case 9), and this patient underwent seven transarterial hepatic chemoembolization and radiation treatments during the following 2 years prior to surgical resection. The other patient (Case 1) was initially considered to have a benign hypervascular mass (Figure 1) and a follow-up with imaging modalities was suggested. After 5 months, a haemorrhagic cavity was detected in the mass, which, after another 4 months, turned out to be a small cystic area that was 1 cm in size. HCC was highly suspected at this time, and a surgical resection was performed.

Figure 1.

A 31-year-old female who had no abdominal discomfort. Enhanced CT (a–c) shows a fat-deficient mass with marked hyperattenuation, intratumoral vessels in the arterial phase and mild homogeneous hypoattenuation in the portal venous and delayed phases. A hypoattenuation area with values of about 20–30 HU turned out to be a haemorrhagic cavity, which was confirmed histologically after surgical resection on follow-up at 1.5 years (d–f).

Imaging features

The tumours were solitary in all 14 cases. Eight were located in the right lobe, five in the left lobe and one in the caudate lobe. The tumours ranged in size from 1.9 to 9.1 cm (mean, 5.7 cm), and their boundaries were well defined. Most were slightly lobulated, and some were circumscribed.

On dynamic enhanced studies, all tumours demonstrated remarkable enhancement in the arterial phase, which decreased to isointensity or slight hypointensity relative to the normal hepatic parenchyma. Areas of prolonged enhancement, which were visible as an irregular area of slowly increasing enhancement, could be confidently identified in the tumours in five cases (Figure 2j). Early draining veins connected to either the portal vein (n = 7) or the hepatic vein (n = 2) were recognized on dynamic CT scans and MRI (Figures 3 and 4d,f,h). Pseudocapsular enhancement in the delayed phase, which is a typical sign of HCC, could not be identified in any of the 14 cases. An incomplete or complete ring enhancement could be seen in the arterial or portal venous phases in nine cases, which were dilated vessels in three cases and peripheral prolonged enhancements in six cases. Haemorrhagic cavities ranging in size from 1.2 to 3.8 cm were found in two tumours with diameters of 4 and 8 cm. In Patient 1 (Figure 1), a haemorrhagic cavity was detected during the imaging follow-up, a small cystic area was identified 4 months later and surgical resection was performed owing to concerns about malignant transformation. In another (Case 4), a haemorrhagic cavity had been detected by the initial imaging studies and surgical resection was performed immediately to prevent eruption, although a confident diagnosis of HAML had been made preoperatively by CT, MRI and contrast-enhanced ultrasound (CEUS).

Figure 2.

A 43-year-old female in whom a hepatic mass was found at a health check-up. Imaging showed an unenveloped fat-deficient mass in the right liver lobe (a). Enhanced CT showed a mass with marked hyperattenuation, with intratumoral vessels in the arterial phase (b), mild hypoattenuation and prominent central vessels in the portal venous phase (c), hypointensity on T₁ weighted (WI) dual-echo imaging (d, e) and on T₂ WI (f) and T₂ WI with fat suppression (g), a hyperintense hepatic lesion which showed a hypointense signal on pre-contrast gadolinium diethylenetriamine pentaacetic acid–enhanced MRI (h) and arterial enhancement and washout with peripheral rim enhancement (i, j) on enhanced MRI, especially the nodular feature. (k) A well-demarcated hypoechoic mass in the right lobe of the liver measuring 7.8 cm in size.

Figure 3.

A 52-year-old male who had no abdominal discomfort. An early draining vein was observed in the arterial phase of contrast-enhanced MRI.

Figure 4.

A 58-year-old female who had no abdominal discomfort. MRI showed a hypointense tumour on T₁ weighted dual-echo imaging (a, b) with a hyperintense irregular area inside. On pre-contrast gadolinium diethylenetriamine pentaacetic acid (Gd-DPTA)-enhanced MRI (c), the hyperintense area can be seen inside the hypointense tumour. On dynamic Gd-DPTA-enhanced MRI (d–f), the early draining vein showed clearly (d). An unenhanced CT scan (g) showed a regular tumour with a hyperattenuation area inside. In the arterial phase of an enhanced CT scan, the early draining vein can be seen (h). Contrast-enhanced ultrasound of the hepatic lesion indicated an early-phase (time, 0.13 s after injecting the SonoVue®; Bracco, Milan, Italy) hyperechoic (j) and late-phase hypoechoic pattern. The curve in (i) indicated that the tumour showed a fast-in and slow-out feature compared with the hepatic parenchyma. Histological examination of the tumour showed that it was composed mainly of smooth muscle cells. (k) Blood vessels (haematoxylin and eosin stain ×10). (l) Homatropine methylbromide-45 positivity (immunohistochemical staining ×200).

In two cases, the hypointense signal area inside the tumour on T₂ WI/FS showed gradual enhancement on contrast-enhanced MRI. No signs of necrosis were identified in any of the masses. In Case 9, who had received multiple interventional therapies over a period of 2 years, only two sets of CT studies in the last 2 months prior to surgery were available. Between the two studies, interventional therapy was given once more, but no remarkable changes were observed.

The imaging characteristics of the 14 cases of MHAMLs are summarized in Table 1.

Table 1.

Radiological characteristics of the 14 cases of myomatous hepatic angiomyolipoma

Case	Ultrasound	CEUS	Plain CT	MRI	Enhancement mode	Peripheral rim enhancement	Early draining vein in the arterial phase	Tortuous tumour vessels	Fat	Haemorrhage
1	N	N	Low density	N	Fast-in and fast-out	A	A	A	A	P
2	N	N	N	T1WI hypointense, T2WI/FS hyperintense	Fast-in and fast-out	A	P	P	A	A
3	Hypoechoic	N	Low density	T1WI hypointense, T2WI/FS hyperintense	Fast-in and fast-out	P	P	P	A	A
4	Heterogeneous mixture of hyperechoic and hypoechoic; CDFI had blood signal	Fast-in and slow-out	Low density	T1WI hypointense, T2WI/FS hyperintense	Fast-in and slow-out	P	P	P	A	P
5	N	N	N	T1WI hypointense, T2WI/FS hyperintense	Fast-in and fast-out	A	A	A	A	A
6	Hypoechoic; CDFI had blood signal	N	Low density	N	Fast-in and fast-out	P	A	P	A	A
7	Hypoechoic; CDFI had blood signal	Fast-in and slow-out	Low density	N	Fast-in and fast-out	A	P	P	A	A
8	Heterogeneous mixture of hyperechoic and hypoechoic	N	N	T1WI hypointense, T2WI/FS hyperintense	Fast-in and fast-out	P	A	A	A	A
9	Hypoechoic	N	Low density	T1WI hypointense, T2WI/FS homogeneous hyperintense	Fast-in and fast-out	P	P	P	A	A
10	N	N	Low density	N	Fast-in and fast-out	A	A	A	A	A
11	Hyperechoic; CDFI had blood signal	Fast-in and slow-out	Low density	T1WI hypointense, T2WI/FS hyperintense	Fast-in and slow-out	P	P	P	A	A
12	Hypoechoic	N	Low density	T1WI hypointense, T2WI/FS hyperintense	Fast-in and slow-out	P	P	P	A	A
13	Hypoechoic	Fast-in and fast-out	N	T1WI hypointense, T2WI/FS hyperintense	Fast-in and fast-out	P	P	P	A	A
14	Hypoechoic	N	N	T1WI hypointense, T2WI/FS hyperintense	Fast-in and fast-out	P	P	P	A	A

A, absent; CDFI, colour Doppler flow image; CEUS, contrast-enhanced ultrasound; FS, with fat suppression; N, none; P, present; T1WI, T₁ weighted imaging; T2WI, T₂ weighted imaging.

Accuracy of pre-operative imaging and pathology of the tumour

The accuracy rate of the pre-operative imaging diagnosis of HAML by CT, MRI, ultrasound and CEUS in our study was 33.3% (3/9), 40% (4/10), 30% (3/10) and 25% (1/4), respectively. The pre-operative diagnoses included HCCs (n = 6), HAMLs (n = 5), hepatic adenomas (n = 2) and haemangioma (n = 1). All tumours were subsequently surgically resected. Their macroscopic and microscopic histological features were as follows. (1) The cut surface of the unencapsulated mesenchymal tumour had a whitish, leiomyoma-like appearance macroscopically, and the lesion was composed primarily of SMCs and thick-walled blood vessels and contained small amounts of adipose cells microscopically. (2) Abnormally large blood vessels were found inside or around the tumour at the tumour–background liver interface (n = 10). This feature was not found in tumours <4.0 cm in size but was present in 10 of 12 tumours ≥4.0 cm in size. (3) No capsules or haemorrhage were found in any of the tumours. (4) Immunohistochemical staining indicated that they were negative for hepatocytes, but the SMCs were strongly positive for a melanocytic cell-specific monoclonal antibody. The final histopathological diagnosis (HMB-45 positive) was consistent with HAML.

Follow-up

All 14 patients were followed up post-operatively for periods ranging from 2 months to 7 years (mean, 48 months). No signs of recurrence or metastasis were found in any of the patients.

DISCUSSION

The majority of HAMLs are benign tumours and without a significant risk of malignant transformation. Surgical removal should be limited to those patients with severe symptoms or a risk of rupture or massive haemorrhage, those with a rapidly growing mass and those in whom malignancy cannot be confidently excluded. However, if a diagnosis of HAML can be confidently made by imaging, the tumour can just be observed by imaging modalities. However, the correct pre-operative diagnostic rate for HAMLs by different imaging modalities has been reported to be only 0–23%,^10–12 with most cases misdiagnosed as HCC or haemangioma.¹¹ The myomatous type is the most difficult one to diagnose by imaging owing to a lack of mature fat components in the tumour. Moreover, the hypervascularity of MHAMLs means that they are often confused with other hypervascular tumours, such as HCC, FNH and hepatic adenomas. In the present study, the pre-operative diagnostic accuracy rate was not ideal, but MRI seemed to be the most accurate modality. Because MHAMLs are mainly composed of myoepithelial tissue and have a rich blood supply, a dynamic contrast-enhanced scan is the most important technique to show the characteristics of these tumours. Both CT and MRI produce good images, and the images could be analysed repeatedly. MRI is better than CT for its 6-period scan, which might best reflect the nature of the lesions. With CEUS, the result is mainly dependent on the operator's clinical experience.

Clinically, MHAML exhibits a female preponderance, and it could be found at any age, although it is mainly seen in adults. The female-to-male ratio in the present study was a little higher than the ratios reported in previous studies,^11,13–15 which may have been due to the small sample size of the study. The majority of the patients in our series were asymptomatic, in agreement with previous observations,^10,11,16 and were detected during health check-ups. MHAML is unrelated to hepatitis and is not associated with increases in serum tumour markers. Currently, the natural course and the growth velocity of the tumours have not been well characterized in long-term follow-up studies.^13,17 MHAMLs are usually solitary and located in the right lobe, and multiple lesions are extremely rare. The size of MHAMLs may increase dramatically,¹⁶ and they do have a malignant tendency.^5–8 In addition, bleeding in tumours has been reported.¹⁸ In Case 1 in the present study, a slightly haemorrhagic cavity was observed during the follow-up period.

Several imaging features may be useful in diagnosing MHAMLs:

1. 1. The dynamic enhancement characteristics of MHAMLs include early arterial enhancement. The density or signal in the portal venous and delayed phases is lower than in liver parenchyma, which is the reason why MHAML is often misdiagnosed as HCC. Other studies have also noted the same enhancement mode.^18–20 Yang et al²¹ found that tumours were inhomogeneously enhanced with enhanced MRI, and other studies have reported contrast enhancement in the arterial phase and slight hyperintensity or equal density in the portal venous phase,²² as a result of abnormalities in the vascular wall and lack of an arteriovenous fistula.

2. 2. A slight gradually enhancing area might suggest a benign fibrous vascular component. Hypointense signal areas inside the tumour on T₂ WI/FS had nodular slow enhancements on contrast-enhanced MRI. This feature may suggest some muscular component with fewer vessels.

3. 3. An early draining vein, to either the portal or hepatic vein, was observed in most cases, which is in agreement with the findings of Jeon et al.⁹ These authors reported that an early draining vein was seen in 80% of HAML cases but in only 7% of HCC cases, which was a statistically significant difference. However, Xiao et al²³ reported that an early portal draining vein was not seen in all cases. This suggests that this feature of MHAMLs needs to be confirmed in further studies with a larger patient sample.

4. 4. Peripheral rim enhancement was observed in most cases. This feature may be misdiagnosed as a pseudocapsule and more attention should be paid to it. Enhanced tumour vessels in the peripheral portion of the tumour form a peripherally enhanced rim. In the present study, this finding was specific to MHAMLs and was not seen in HCC.

5. 5. The absence of a tumour capsule is a characteristic sign of HCC but is not seen in MHAMLs, and it is the principal feature that differentiates the two tumours.

6. 6. Vascular malformation or aneurysmal haemorrhage is a feature of MHAMLs.

7. 7. Other CT or MRI features of MHAMLs include a well-demarcated homogeneous hypoattenuated/hypointense mass (due to the myomatous component), but necrosis or calcification are only rarely observed. These features, however, do not differ from those of HCC, and the imaging presentations of the two tumours overlap in this regard.

On pathological examination, abnormally large dilated vessels on the periphery or inside the lesion could lead to a diagnosis of MHAMLs. Some studies have suggested that abnormally large dilated vessels should be considered a typical histological feature of MHAMLs,^16,18,24 and might be important in recognizing, understanding and interpreting the findings of various imaging modalities. In the present study, abnormally large dilated vessels were present in 10 of the 12 tumours that were >4.0 cm. Other studies have found that this feature is usually present in tumours >2.5 cm.¹⁶ The peripherally enhanced rim feature is considered to be caused by the growth of the tumour into the surrounding parenchyma. Although none showed cystic degeneration or calcification, haemorrhage was seen in two patients. These findings also assist in distinguishing MHAMLs from other lesions.

HMB-45 is a useful immune marker and conclusively establishes the diagnosis of HAMLs.²⁵ When imaging does not confidently establish a diagnosis of either HCC or MHAML, an ultrasound-guided core needle biopsy using HMB-45 staining can be expected to establish a definite diagnosis and thus avoid unnecessary surgery.

Given the high incidence of HCC in China, this diagnosis should be put at the top of the list of differential diagnoses when radiologists encounter a hypervascular hepatic mass. Other diagnoses that need to be considered include fat-deficient liver tumours, such as hepatic adenomas or FNH (Table 2). The classic HCC enhancement mode is described as “fast-in and fast-out” in the setting of cirrhosis,²⁶ and the accompanying features include pseudocapsules and vascular invasions, such as portal thrombi, splenomegaly, lymphadenectasis and varicosity. In Western countries, hepatocellular adenoma is a rare tumour found in females on oral contraceptives or males on anabolic steroids. Evidence of minimal intratumoral bleeding is often present. Some are steatotic, and may be multifocal. The tumour usually exhibits slight hyperattenuation or isoattenuation in the delayed phase, sometimes with capsules but without cirrhosis. FNH is relatively subtle on non-contrast images and shows a rapid and intense enhancement during the arterial phase, followed by hyper- or isointensity in the portal venous phase.^27,28 Around 50% of cases of FNH have a central scar with late enhancement. Haemangiomas are the most common hepatic tumours and are usually characterized by peripheral and nodular enhancement in the arterial phase, followed by progressive centripetal filling in the delayed phase.

Table 2.

Helpful imaging characteristics for identifying and differentiating liver lesions

Features	MHAML	HCC	FNH	Haemangioma	Adenoma
Enhanced mode	Fast-in and fast-out	Fast-in and fast-out	Rapid enhancement in the arterial phase, followed by isointensity in the portal phase	Gradual enhancement	Slight hyper- or isoattenuation in the delayed phase
Early draining vein in arterial phase	P	A	A	A	A
Presence of tortuous vessels	P	A	A	A	A
Necrosis	A	P	A	A	A
Cystic degeneration	P	P	A	A	A
Haemorrhage	P	P	A	A	P
Calcification	A	A	A	A	A
Tumour capsule	A	P	A	A	A
Centre scar	A	A	P	A	A
Cirrhosis	A	P	A	A	A
AFP	–	↑	–	–	–
Bile duct dilation	A	P	A	A	A
Portal thrombi, splenomegaly, lymphadenectasis or varicosity	A	P	A	A	A
Lymphadenopathy	A	P	A	A	A

A, absent; AFP, α-fetoprotein; FNH, focal nodular hyperplasia; HCC, hepatocellular carcinoma; MHAML, myomatous hepatic angiomyolipoma; P, present.

‒, normal range; ↑, growth.

Our study does have some limitations. Firstly, it was a retrospective study, and the imaging protocols were not standardized during the 7-year period in which patients were seen at our hospital. Secondly, because of the rarity of MHAMLs, only a small number of cases were enrolled in the study.

In conclusion, MHAMLs should be considered if imaging shows an early draining vein, peripheral rim enhancement and the absence of a tumour capsule in a hypervascular hepatic tumour. Along with clinical features such as female gender, age around 30–50 years and a normal serum AFP, these imaging features may be helpful in discriminating MHAML from other hypervascular lesions in a non-cirrhotic liver.