Epithelioid angiomyolipoma of the kidney: Radiological imaging




To review the imaging findings of renal epithelioid angiomyolipomas.


Eight patients treated at two institutions were pathologically diagnosed as having epithelioid angiomyolipoma. All of them underwent computed tomography, and four underwent magnetic resonance imaging. The tumor size, existence of fat, heterogeneity, computed tomography attenuation, degree of enhancement, enhancement pattern and magnetic resonance imaging signal intensity were evaluated.


Intratumoral fat was not detected in any of the cases. On unenhanced computed tomography, the intratumoral attenuation was hyperattenuating in six of the seven patients who were examined using this modality. On T2-weighted images, the signal intensity of the solid component, cyst wall or septum was low in three of the four cases. Four of the eight cases were heterogeneous solid-type accompanied by hemorrhage, necrosis or hyalinization. One homogeneous solid-type lesion was large in size and was pathologically accompanied by neither hemorrhage nor necrosis. All three multilocular cystic types were pathologically accompanied by massive hemorrhage in the cystic component. One was accompanied by spontaneous perirenal hematoma.


The radiological appearance of most epithelioid angiomyolipomas has a tendency to be hyperattenuating on unenhanced computed tomography images, with low intensities on T2-weighted images. They can be heterogeneously solid, homogeneously solid or a multilocular cystic lesion with massive hemorrhage.

Abbreviations & Acronyms



corticomedullary phase


computed tomography


epithelioid angiomyolipoma


early excretory phase


fat-saturated T1-weighted image


fat-saturated T2-weighted image


Hounsfield units










multilocular cystic renal cell carcinoma


magnetic resonance imaging


mammalian target of rapamycin


not available


nephrographic phase




renal cell carcinoma


smooth muscle actin


T1-weighted image


T2-weighted image


tuberous sclerosis complex




Classic AML is well known as a benign mesenchymal neoplasm that is composed of varying amounts of dysmorphic blood vessels, smooth muscle cells and mature adipose elements.[1] Composed of epithelioid cells, polygonal cells and varying degrees of nuclear atypa with little or no fat cells, eAML is a new variant of AML.[1] A pathological differential diagnosis from high-grade RCC is sometimes challenging because of their similar morphologies. This tumor was first reported by Eble et al. in 1997.[2] In 1998, Pea et al. reported that some renal tumors previously reported as RCC reacted with HMB-45 and were reclassified as eAML.[3] Furthermore, approximately one-third of eAML have been found to present with malignant biological behaviors.[1, 4] As of 2004, the World Health Organization Classification of Renal Neoplasms has regarded eAML as a potentially malignant mesenchymal neoplasm.[5]

As a treatment strategy for eAML, surgical resection is justified because of the malignant potential of eAML. Thus, the ability to distinguish benign tumors from eAML is clinically important. Furthermore, the mTOR pathway was recently found to be activated in eAML,[6] and some studies have reported that mTOR inhibitors, such as sirolimus or temsirolimus, represent the best treatment option for patients with eAML.[7, 8] For the selection of an optimal treatment, the ability of imaging findings to suggest possible eAML is important. However, only a few articles have focused on the radiological appearance of eAML.[9-11] Herein, we retrospectively reviewed the imaging findings of eAML as obtained using CT and MRI.



Institutional review board approval was obtained at two institutions for the present retrospective study. Between June 1991 and September 2011, eight patients (6 women, 2 men) aged 27–61 years (mean 39.3 years) were pathologically diagnosed as having eAML at the two institutions. All the patients' medical records were retrospectively reviewed and summarized. The pathological proof of eAML was obtained by surgical resection. One patient had a clinical history of tuberous sclerosis complex. The pathological findings for three cases have been previously reported.[12] All eight patients underwent a CT examination, and four patients underwent an MRI examination.

Pathological diagnosis

eAML is composed of a proliferation of epithelioid cells with abundant cytoplasm, vesicular nuclei and often prominent nucleoli.[5] The pathological definition of eAML varies in the percentage of the epithelioid component, depending on the paper.[13, 14] For the present study, eAML was defined as tumors that were composed of predominantly epithelioid cells with abundant cytoplasm, vesicular nuclei and prominent nucleoli. The percentage of epithelioid component and fat component was evaluated by one uropathologist. Immunohistochemical staining for the antibodies of HMB45, SMA and cytokeratin was also used to confirm the diagnosis.

CT technique

All the images were obtained using a single-helical CT or multidetector-row CT (High Speed Advantage, LightSpeed 16; GE, Waukesha, WI, USA). The scanning parameters for imaging acquisition and reconstruction were as follows: 120 kVp; 200–400 mA, pitch of 0.9–1.3 and a reconstruction slice thickness of 5–7 mm. Thin-section images (slice thickness of 1–2 mm) were obtained in three cases. Four of the eight patients underwent triphasic dynamic-enhanced CT (unenhanced, CMP, NP and EEP), three patients underwent biphasic dynamic-enhanced CT (unenhanced, CMP and EEP) and the remaining patient underwent EEP imaging only. Intravenous contrast material (iohexol, Omnipaque 300; Daiichi Pharmaceutical, Tokyo, Japan) was injected at a dose of 80–120 mL at a rate of 2.5–3.5 mL/s for triphasic and biphasic dynamic-enhanced CT, and at a rate of 2.0 mL/s for one patient who underwent EEP imaging only. The scan delays were 30–40 s, 90–100 s and 150–180 s after the administration of CMP, NP and EEP.

MRI technique

All the images were obtained using 1.5-T systems (Signa Advantage, GE; Signa Excite, GE; MAGNETOM Avanto; Siemens, Erlangen, Germany). For the T2-weighted images, the following sequences were obtained: axial and coronal single shot fast spin echo images (TR/TE, 1100–1274.3/90–183; slice thickness, 4–7 mm; slice gap, 0–2 mm), and axial T2-weighted fast spin-echo images (TR/TE, 2600–4200/76–88; slice thickness, 7 mm; slice gap, 1–2 mm). Axial fat-saturated T2-weighted fast spin-echo images were also acquired (TR/TE, 2600–3800/93; slice thickness, 5–7 mm; slice gap, 1–2 mm). For the T1-weighted images, the following sequences were obtained: axial fat-saturated T1-weighted fast spin-echo images (TR/TE, 500/8; slice thickness, 8 mm; slice gap, 2 mm), axial in- and out-of-phase 2-D gradient echo images (TR/TE1/TE2, 150–210/4.2–4.6/2.1–2.3; slice thickness, 8 mm; slice gap, 0–2 mm) and axial fat-saturated 3-D gradient echo images (TR/TE, 3.8/1.8; slice thickness, 4 mm; slice overlapping, 2 mm). The field of view was 32–40 cm, and the matrix size was 256–320 × 192 – 256.

Imaging analysis

Two radiologists reviewed all the CT and MRI images without detailed knowledge of the pathological findings. The radiologists were only told that the final diagnosis was eAML. Tumor size, tumor location, existence of fat and calcification, heterogeneity, and the degree of enhancement were evaluated in all the cases. The CT attenuation of unenhanced, CMP and EEP images was measured in six cases: two using biphasic dynamic-enhanced CT, and four using triphasic dynamic-enhanced CT. The measurement was carried out using the higher CT attenuation area to avoid obtaining data for necrotic or cystic areas using a reviewer-defined region of interest. CT attenuation was not measured in two cases (cases 1 and 6) in which the films were available, but the original digital data required to measure CT attenuation were not available. For the detection of fat attenuation, a pixel analysis was also used for unenhanced CT images. In four cases, the signal intensities of tumors on T1-weighted and T2-weighted images, and the existence of a pseudocapsule on T2-weighted images were evaluated. In three cases, chemical fat suppression images were also evaluated for the detection of a microscopic fat component.

Based on the aforementioned information, each tumor was categorized as heterogeneous solid-type, homogeneous solid-type, unilocular cystic-type or multilocular cystic-type. Homogeneity was defined as being present when more than 90% of the area was occupied by the same attenuation value, as ascertained by visual inspection. Based on the findings of unenhanced CT, the tumor was classified as hyperattenuating, iso-attenuating or hypoattenuating by comparison with the renal parenchyma (cortex) in seven of the eight cases; unenhanced CT was not obtained in one case. In the six cases in which the digital data was available, the degree of enhancement (the difference between the attenuation value of the unenhanced scan and the CMP) was described at the most enhanced part of the tumor. According to previous studies,[15-18] the degree of enhancement of hypervascular clear cell type RCC was 78–133 HU, and that of hypovascular papillary RCC was 15–54 HU. Thus, a degree of enhancement higher than 80 HU was classified as marked, that between 50 and 80 HU as moderate, and less than 50 HU as weak. Regarding the enhancement pattern, when the difference in the CT attenuation between the CMP and EEP was less than 20 HU, it was defined as a persistent pattern. When the CT attenuation of the CMP was more than 20 HU greater than that of the EEP, it was defined as a washout pattern.


Table 1 summarizes the patient characteristics and the radiological findings of eAML. The tumor sizes ranged from 3.5 cm to 13 cm (mean 7.4 cm). The tumors were almost entirely composed of epithelioid cells (95–100%). The epitheloid tumor cells were positive for HMB45 and SMA, and negative for cytokeratin. Nuclear atypia was seen in all the tumors, occupying almost all of the tumor in seven cases and 70% in two cases (cases 1 and 7). An adipocytic component was not seen in five cases and constituted <5% of the tumor in three cases (cases 2, 4 and 7). Neither calcification nor a pseudocapsule was seen during the pathological examination in all the cases. In one case, a tumor thrombus was seen extending into the right renal vein (case 5). Another lesion showed a spontaneous perirenal hematoma and was associated with symptoms of left back pain (case 4). No cases of metastases or recurrences occurred in the present series.

Table 1. Characteristics of eight patients with epithelioid angiomyolipoma of the kidney
NoAge/sexSideLocationSize (cm)TSCPercentage of fat component at pathologyPseudo-capsule at pathologyUnenhanced CT attenuationHomogeneityAttenuation (HU)EnhancementT1WI or FST1WIT2WI or FST2WI
  1. †Tumor thrombus into the right renal vein, ‡Unenhanced CT was not taken. §The tumor had a massive hemorrhage and spontaneous perirenal hematoma.

No fat component was detected on unenhanced CT images, T1-weighted images, T2-weighted images or chemical fat suppression images in any of the cases. Calcification was not detected on CT, and a pseudocapsule was not detected on T2-weighted images in any of the cases. Four of the eight lesions were categorized as heterogeneous type (Figs 1, 2), one was categorized as homogeneous type (Fig. 3) and the other three lesions were categorized as multilocular cystic type (Fig. 4). None of the cases were categorized as unilocular cystic type.

Figure 1.

A 28-year-old man (case 1) with a heterogeneous solid-type lesion. (a) Unenhanced CT shows a hyperattenuating mass in the lower pole of the right kidney (arrows). No fat density was detected. (b,c) Dynamic enhanced CT shows a heterogeneously enhanced tumor on (b) corticomedullary phase and (c) early excretory phase images. (d) Hematoxylin–eosin staining specimens show pleomorphic tumor cells with large hyperchromatic nuclei and abundant eosinophilic cytoplasm. (e) The tumor cells were diffusely reactive with HMB45 antibody.

Figure 2.

A 36-year-old woman (case 4) with a heterogeneous solid-type lesion and spontaneous perirenal hematoma. (a) Unenhanced CT shows a hyperattenuating mass (long arrow) accompanied by a spontaneous perirenal hematoma (arrowheads) in the lower pole of the left kidney. No fat density was detected. (b) Dynamic enhanced CT shows a heterogeneous tumor on corticomedullary phase images. The enhancement was barely noted, because the tumor was nearly entirely occupied by a massive hemorrhage. (c) On a T2-weighted image obtained using MRI, the tumor was mostly low intensity.

Figure 3.

A 34-year-old woman (case 5) with a homogeneous solid-type lesion. (a) Unenhanced CT shows a hyperattenuating mass in the right kidney (arrows). No fat density was detected. (b,c) Dynamic enhanced CT shows a homogenously enhanced tumor on corticomedullary phase (b: arrows) and early excretory phase (c: arrows) images.

Figure 4.

A 27-year-old woman (case 6) with a multilocular cystic-type lesion. (a) Enhanced CT shows a multilocular cystic tumor (arrows). (b) On a fat-saturated T1-weighted image obtained using MRI, the intracystic component was occupied by a high-intensity signal corresponding to a massive hemorrhage, which was confirmed in the pathological specimen. (c) On T2-weighted image obtained using MRI, the wall of the multilocular cystic tumor appeared as a low intensity (arrows). (d) Hematoxylin–eosin stained specimen showing the wall of the multilocular cystic tumor occupied by epithelioid muscle component (arrows). Left upper pink area shows massive hemorrhage in the cystic component.

All four heterogeneous types were accompanied by hemorrhage, necrosis or hyalinization in the histopathological specimens. In one of these four tumors, the lesion was mostly occupied by hemorrhage, and was accompanied by a perirenal hematoma (Fig. 2). One homogeneous-type lesion was not accompanied by either hemorrhage or necrosis in the pathological specimen. All three cases of multilocular cystic types were pathologically accompanied by massive hemorrhage in the cystic spaces and had consistent, considerably thick walls composed of epithelioid cells with rupture resulting in hemorrhage.

Among the seven cases for which unenhanced CT findings were obtained, CT attenuation of the lesions was hyperattenuating in six (Figs 1, 2) and iso-attenuating in one based on visual inspection. In the quantitative analysis of the solid portions of six cases for which digital data was available, the CT attenuation was 48.5 HU (45–52 HU) in four hyperattenuating lesions (Fig. 1), 62 HU in another hyperattenuating lesion occupied mostly by hemorrhage and accompanied by perirenal hematoma (Fig. 2), and 37 HU in one iso-attenuating lesion. The average attenuation of the normal renal parenchyma was 33 HU. There was no pathological difference between the hyperattenuating and isoattenuating tumors.

Among the six cases in which digital data for the dynamic scans were available, the enhancement pattern was marked enhancement with a washout pattern in one, moderate enhancement with washout in one, moderate enhancement with prolonged enhancement in one, weak enhancement with prolonged enhancement in two and no enhancement because of massive hemorrhage in one. In the one case in which a dynamic scan was carried out without digital data, the enhancement pattern showed marked enhancement with a washout pattern on subjective evaluation. In the other case, dynamic CT was not carried out.

Among the four cases with MRI findings (two heterogeneous solid-type and two multilocular cystic-type), there were no findings suggesting fat in the lesions in any sequence. The signal intensities of the T2-weighted images or fat-saturated T2-weighted images were low compared with the renal parenchyma in the solid component or the cyst wall or septum in the four cases (Figs 2, 4), and iso-intense to normal renal parenchyma in one case. In the two multilocular cystic-types, a high-intensity area was visualized on T1-weighted images or fat-saturated T1-weighted images, corresponding to areas of hemorrhage (Fig. 4).


AML can be classified into at least three categories based on imaging and clinical aspects: classic, fat poor and epithelioid.[19] Classic AML is benign and the most common, containing a significant amount of fat, and thus can be easily diagnosed by detecting fat components on imaging. Fat-poor AML represents a substantial portion of benign AML and contains too few fat cells to be detected with unenhanced CT; such lesions are often inadvertently removed by surgery. Fat-poor AML are classified into two categories based on their imaging appearance on unenhanced CT: hyperdense AML (AML with minimal fat) and isodense AML.[20] eAML is a potentially malignant neoplasm with little or no fat component.[5] In the previous reports, the imaging findings were not well described or discussed; some merely described the lesions as “heterogeneous” or “cystic and solid”.[1, 9-11, 21-33] Our data showed that four lesions were heterogeneous solid-type lesions, one was a homogeneous solid-type lesion and three were multilocular cystic-type lesions.

Hyperattenuation on unenhanced CT is a characteristic finding of eAML. Attenuation on unenhanced CT images was hyperattenuating in six of the seven cases for which unenhanced CT images were available, and iso-attenuating in the remaining one case. The tumor component showed a low intensity on T2-weighted images in three of the four cases for which MRI findings were available. Several studies have reported that AML with abundant smooth muscle and little or no fat on pathological examination (AML with minimal fat) appeared as hyperdense and T2-low-intestity masses.[34-37] Thus, hyperattenuation on unenhanced CT images and low intensities on T2-weighted images in eAML might also be due to the significant amount of (in this case epithelioid) muscle component. The enhancement patterns were varied and non-specific. This variety of enhancement patterns in epithelioid AML is also the same as that for AML cases with no visible fat.[38]

The heterogeneous solid appearance was caused by pathological hemorrhage, necrosis or hyalinization in the present study. In previous studies, eAML were reported to be typically accompanied by hemorrhage and variable necrosis or cystic changes.[37] Thus, a hyperattenuating, heterogeneous enhancement might suggest a diagnosis of eAML. However, its differentiation from RCC might be difficult, because RCC often shows similar findings.[14]

One homogeneous-type lesion was hyperattenuating, similar to skeletal muscle on unenhanced CT, and showed moderate enhancement with a gradual enhancement of the enhancement pattern. Neither hemorrhage nor necrosis was noted in the pathological specimen. Although MRI findings were not obtained in this case, the CT results were similar to the findings for AML with minimal fat. However, the size of the tumor in this case was 7 cm, whereas the size of previously reported AML with minimal fat was approximately 3 cm (2.2–5.0 cm).[34, 37, 39] Furthermore, the average size of all eight eAML was 7.4 cm. Thus, eAML tends to be larger than AML with minimal fat. The larger size of eAML might be related to their malignant potential. A hyperattenuating, homogeneously enhancing mass with a relatively large size might suggest possible eAML.

The multilocular cystic-type eAML had hyperattenuating cystic walls or septa, and were accompanied by multiple cystic components with massive hemorrhage and necrosis. Some previous studies have also reported cystic dominant-type eAML as a result of necrotic cystic fluid or hemorrhage.[23, 29] Whether this type of lesion was originally a multilocular figure or the result of recurrent hemorrhage or necrosis is unclear. However, it is interesting that eAML, like clear cell RCC, also has a multilocular cystic-type entity. The radiological findings for multilocular type eAML differed from those for MCRCC with regard to the findings of massive hemorrhage, the enhancement pattern and the CT attenuation of the wall or septum. Although massive hemorrhage was seen in the cystic components of all four multilocular cystic-type eAML, there are few reports of MCRCC accompanied by massive hemorrhage in the cystic component on CT and/or MRI findings. To our knowledge, just two MCRCC cases have been reported to be accompanied by hemorrhage, but only in the partial portion of the cystic component.[40] Although the CT attenuation of the cyst wall or septum was hyperattenuating in all four multilocular cystic-type eAML, no reports of hyperattenuation of the wall or septum on CT images of MCRCC have been made.

One of our eight cases showed a spontaneous perirenal hematoma around the renal tumor. It is well known that classic AML often causes spontaneous perirenal hematoma, followed by RCC.[40] Other renal tumors that cause spontaneous perirenal hematoma include metastatic tumors of malignant melanoma, renal abscesses, ruptured renal cysts and pheochromocytomas.[41, 42] eAML should be included in the differential diagnosis of tumors with spontaneous perirenal hematoma.

The present study had several limitations. First, the number of cases was small. Thus, further evaluations using a large number of cases are necessary to confirm our findings. Second, some of our data were more than 10-years-old, and CT attenuation data were not available in these cases. Because the data were retrospectively collected from different hospitals, all the tumors were not examined using the same protocol. Third, as fat-sensitive imaging techniques, such as thin-slice section CT or chemical shift suppression images, were obtained in just three of the eight cases, the rate of fat detection in eAML should be further investigated. Fourth, the percentage of epithelioid component varies in definitions of eAML. In the present study, tumors with a high percentage of epithelioid component were studied. However, a difference in the percentage could influence the imaging appearance of eAML.

In conclusion, the radiological appearance of most eAML has a tendency to be hyperattenuating on unenhanced CT images, with low intensities on T2-weighted images, and could be a heterogeneously solid, homogeneously solid or multilocular cystic lesion with massive hemorrhage. eAML should also be included in the differential diagnosis of tumors with spontaneous perirenal hematoma.

Conflict of interest

None declared.