ª Springer Science+Business Media New York 2014
Abdominal Imaging
Abdom Imaging (2014) DOI: 10.1007/s00261-014-0100-6
Perilesional enhancement of liver cavernous hemangiomas in magnetic resonance imaging Marta S. C. Sousa,1,2 Miguel Ramalho,1 Vasco Here´dia,2 Anto´nio P. Matos,1 Joa˜o Palas,1 Yong Hwan Jeon,3 Diana Afonso,4 Richard C. Semelka3 1
Department Department 3 Department 4 Department 2
of of of of
Radiology, Radiology, Radiology, Radiology,
Hospital Garcia de Orta, 2801-951 Almada, Portugal Hospital Espı´ rito Santo, E´vora, Portugal University Of North Carolina at Chapel Hill, Chapel Hill, NC, USA Hospital Beatriz Aˆngelo, Loures, Portugal
Abstract Objective: To evaluate on magnetic resonance imaging (MRI) the occurrence rate of temporal perilesional parenchymal enhancement (PPE) associated with hepatic hemangiomas in a large consecutive series and to determine which aspects are associated with this observation. Materials and methods: Institutional review board approved this retrospective study. A computerized search of the MRI database was performed for consecutive patients between January 2008 and January 2012. The study population included 513 liver hemangiomas in 224 patients (104 males and 120 females; mean age of 55.2 ± 13.5 years; age range 24–89 years). Two readers independently reviewed the frequency of PPE, size, speed of enhancement and location of each hemangioma. Marginal models with generalized estimating equation were used. Wald test was applied to verify if the model coefficients were significant. Results: 80/513 (15.6%) hemangiomas showed PPE. The incidence of PPE was significantly higher (p < 0.05) in hemangiomas with Type1 speed of enhancement (51/80, 63.8%) than in those with Type2 or Type3. 66/80 (82.5%) hemangiomas with PPE were subcapsular (p < 0.05). Conversely, the majority (280/433, 64.7%) of hemangiomas without PPE were deep in location (p < 0.001). Lesser proportion of hemangiomas with PPE was located in segment IVa (p < 0.05). Conclusion: PPE is not uncommonly seen along with hepatic hemangiomas. This appearance is most frequently observed in rapidly enhancing small lesions with a subcapsular location.
Correspondence to: Miguel Ramalho; email: [email protected]
Hemangiomas are the most common benign tumors in the liver with a reported prevalence of 0.4–20% according to several studies [1]. Hemangiomas range in size from few millimeters to more than 10 cm in diameter, with most ranging from 0.5 to 5 cm [2]. Hemangiomas are multiple in more than 50% of cases and are typically discovered between the fourth and fifth decades of life [3, 4]. At MRI, the majority of hepatic hemangiomas display pathognomonic pre and postgadolinium imaging features that usually enable the correct diagnosis [5]. However, an atypical appearance of hemangioma is frequently encountered in daily practice. It has been shown that perilesional parenchymal enhancement (PPE) depicted on arterial phase images, previously thought to be a feature associated with a malignant lesion, has been found in benign liver masses, in particular hemangiomas with an incidence that can be as high as 26% [6–11]. PPE is appreciated as a transient increased contrast enhancing area adjacent to a lesion in the arterial phase [6, 11]. Previous studies have shown that PPE is most commonly encountered in rapidly enhancing small hemangiomas [9, 11–13]. The pathophysiological mechanism of temporal PPE is not completely clarified yet, but has been proposed an association with arterio-venous fistula [6, 8–13]. One small study [8] reported that hemangiomas with PPE were located less than 2 cm from the liver capsule, speculating an association between temporal PPE adjacent to hemangiomas with their location in the liver. However this has not been documented in a study involving a large consecutive population. Therefore, our purpose was to evaluate the occurrence rate of temporal PPE enhancement associated with hepatic hemangiomas in a large consecutive series, and to determine which aspects are associated with this observation.
M. S. C. Sousa et al.: Perilesional enhancement of liver hemangiomas on MRI
Materials and methods Patients Institutional review board approval was obtained for this retrospective study with waived informed consent. A computerized search of the database of the main institution was performed using the keywords ‘‘hemangioma’’ and ‘‘angioma’’ for consecutive patients between 1/2/2008 (mm/dd/year) and 1/2/2012 that underwent abdominal magnetic resonance imaging (MRI) examination. Inclusion criteria for this study were patients who were at least 18 years of age, who underwent abdominal MRI, with one or more hepatic nodule(s) that met the criteria for liver hemangiomas, as defined below. If a patient had multiple follow-up MR studies, the examination with the best arterial phase and least imaging artifacts was selected. Criteria used for the diagnosis of hemangioma were based on the pathognomonic combined appearance on pre and post-contrast images as follows: (i) a well defined hepatic lesion with moderately high T2-weighted signal intensity and moderately low T1-weighted signal intensity; plus (ii) a discontinuous ring of peripheral nodules or complete or near complete flash-filling enhancement on post-gadolinium arterial phase images; and (iii) centripetal progression or uniform enhancement in the late phases [5]. The inclusion criterion for adequate arterial phase timing represented images acquired during an imaging window starting in the late arterial phase (predefined as the presence of contrast in the renal arteries and renal veins), and ending in the hepatic arterial dominant phase (predefined as the presence of contrast in the portal vein and the absence of contrast in the hepatic veins) [14]. A total of 251 examinations were reviewed. Patients were excluded if they had any of the following: history of previous chemotherapy or prior liver resection or hepatic intervention (n = 2), portal vein thrombus (n = 3), atypical enhancement pattern (n = 3), and the absence of post-gadolinium dynamic evaluation or inadequate arterial phase (n = 19). The study population included 513 liver hemangiomas in 224 patients (104 males and 120 females; mean age of 55.2 ± 13.5 years; age range 24–89 years). The use of MRI was at the discretion of the referring physicians for all examination.
MR technique and image analysis All subjects underwent MRI of the abdomen with a 1.5-T system (GE-Signa HDx, GE Healthcare) using a phasedarray torso coil. As part of our standard abdominal MRI protocol, the following sequences were performed: axial unenhanced T1-weighted dual-echo in-phase/out-of-
phase (125 msec /4.3 or 2.1 msec, 80 flip angle) 2D gradient-echo (GRE) sequences; coronal T2-weighted halfFourier single-shot fast spin-echo sequences (1894.7/83.9); axial T2-weighted half-Fourier single-shot fast spin-echo sequences (1800/89.9) with and without fat suppression; and axial pre and post-gadolinium fat-suppressed 3D GRE (4.17/1.98, 12 flip angle) at the arterial, venous and interstitial phases. Intravenous gadoterate meglumine (Dotarem, Guerbet, Paris, France) was administered as a powerinjected (Medrad, Pittsburgh, PA) bolus of 0.1 mmol/ kg at 2 ml/s in all subjects, followed by a bolus of 20-ml saline flush. The arterial post-contrast sequences were acquired with an empirical delay time calculated to initiate acquisition at 20 s after the start of contrast injection. The portal venous sequence was acquired at 60–80 s after the start of injection, and the interstitial/equilibrium sequences were acquired at 3–4 min.
Image analysis All examinations were reviewed and interpreted by two readers: M. Ramalho, a radiologist with 8 years of experience in abdominal imaging, and M. S. C. Sousa, a senior (4th year) radiology resident. All images were reviewed by using a PACS Barco monitor and DICOM viewing software (Centricity-Ris, GE Healthcare). The frequency of PPE, the speed of lesion enhancement, location and size of each hemangioma were measured independently. Disagreements between the reviewers were resolved by consensus with a third radiologist (V. Here´dia) with 7 years of experience in abdominal imaging. PPE was defined as an area of transient increased enhancement either in conjunction or not, with early opacification of vascular structures adjacent to the tumor, as displayed during the early phase of dynamic MRI. PPE was diagnosed when images showed circumferential or wedge-shaped or homogeneous enhancement in the liver parenchyma adjacent to the hemangioma in the arterial phase of dynamic MRI and faded to isointensity or notably less hyperintense in the portal phase. Speed of lesion enhancement was appreciated at arterial phase and subjectively separated in three types (Fig. 1): (i) Type1, with approximately 100% of intratumoral enhancement extent; (ii) Type2, with 60–99% of intratumoral enhancement extent; and (iii) Type3, with less than 60% of intratumoral enhancement extent. The percentage of enhanced tumor volume was estimated by evaluation of contiguous sections that demonstrated this lesion. The location of hemangiomas was defined as subcapsular when positioned at 10 mm or less from the hepatic capsule and deep located when positioned at more than 10 mm from the liver capsule. The two readers
M. S. C. Sousa et al.: Perilesional enhancement of liver hemangiomas on MRI
Fig. 1. Axial fat-suppressed T2-weighted half-Fourier single-shot fast spin-echo images (A, C, E) and gadolinium-enhanced fat-suppressed T1-weighted 3D gradient-echo MR images obtained at the arterial phase (B, D, F). Patient 1 (A, B) shows three well defined hepatic hemangiomas with moderately high T2-weighted signal intensity (arrows, A). Type1 speed of enhancement (approximately 100% of intratumoral enhancement extent) is seen in the lesion present on segment VIII (arrow, B). Note the adjacent perilesional enhancement in this hemangioma with a deep location. Pa-
tient 2 (C, D) shows one hemangioma with moderately high T2-weighted signal intensity with is seen on segment II (c). Arterial phase image (D) shows Type2 speed of enhancement with 60–99% of intratumoral enhancement extent. Patient 3 (E, F) shows two well-defined hepatic hemangiomas with moderately high T2-weighted signal intensity (E) are seen in segment VIII, demonstrating the typical discontinuous ring of peripheral nodules on the arterial phase (F), which was considered to be a Type3 speed of enhancement (less than 60% of intratumoral enhancement extent).
M. S. C. Sousa et al.: Perilesional enhancement of liver hemangiomas on MRI
Table 1. Statistical analyses Complete modele
Simple model PPE Liver segmentationa II III IV A IV B V VI VII VIII Lesion sizeb Small Medium Large Speed of lesionc T2 T3 Subcapsular
OR
P value
95% CI
0.98 0.62 0.08 0.68 0.29 0.78 0.84 0.61
0.980 0.629 0.026 0.711 0.233 0.798 0.855 0.608
[0.15; 6.45] [0.09; 4.4] [0.01;0. 74] [0.09; 5.31] [0.04; 2.21] [0.12; 5.16] [0.12; 5.7] [0.09; 4.12]
0.76 0.46 0.15
0.282 0.035 0.018
[0.46; 1.26] [0.22;0. 95] [0.03;0. 72]
0.05 0.07 7.66
Abdominal Imaging
Abdom Imaging (2014) DOI: 10.1007/s00261-014-0100-6
Perilesional enhancement of liver cavernous hemangiomas in magnetic resonance imaging Marta S. C. Sousa,1,2 Miguel Ramalho,1 Vasco Here´dia,2 Anto´nio P. Matos,1 Joa˜o Palas,1 Yong Hwan Jeon,3 Diana Afonso,4 Richard C. Semelka3 1
Department Department 3 Department 4 Department 2
of of of of
Radiology, Radiology, Radiology, Radiology,
Hospital Garcia de Orta, 2801-951 Almada, Portugal Hospital Espı´ rito Santo, E´vora, Portugal University Of North Carolina at Chapel Hill, Chapel Hill, NC, USA Hospital Beatriz Aˆngelo, Loures, Portugal
Abstract Objective: To evaluate on magnetic resonance imaging (MRI) the occurrence rate of temporal perilesional parenchymal enhancement (PPE) associated with hepatic hemangiomas in a large consecutive series and to determine which aspects are associated with this observation. Materials and methods: Institutional review board approved this retrospective study. A computerized search of the MRI database was performed for consecutive patients between January 2008 and January 2012. The study population included 513 liver hemangiomas in 224 patients (104 males and 120 females; mean age of 55.2 ± 13.5 years; age range 24–89 years). Two readers independently reviewed the frequency of PPE, size, speed of enhancement and location of each hemangioma. Marginal models with generalized estimating equation were used. Wald test was applied to verify if the model coefficients were significant. Results: 80/513 (15.6%) hemangiomas showed PPE. The incidence of PPE was significantly higher (p < 0.05) in hemangiomas with Type1 speed of enhancement (51/80, 63.8%) than in those with Type2 or Type3. 66/80 (82.5%) hemangiomas with PPE were subcapsular (p < 0.05). Conversely, the majority (280/433, 64.7%) of hemangiomas without PPE were deep in location (p < 0.001). Lesser proportion of hemangiomas with PPE was located in segment IVa (p < 0.05). Conclusion: PPE is not uncommonly seen along with hepatic hemangiomas. This appearance is most frequently observed in rapidly enhancing small lesions with a subcapsular location.
Correspondence to: Miguel Ramalho; email: [email protected]
Hemangiomas are the most common benign tumors in the liver with a reported prevalence of 0.4–20% according to several studies [1]. Hemangiomas range in size from few millimeters to more than 10 cm in diameter, with most ranging from 0.5 to 5 cm [2]. Hemangiomas are multiple in more than 50% of cases and are typically discovered between the fourth and fifth decades of life [3, 4]. At MRI, the majority of hepatic hemangiomas display pathognomonic pre and postgadolinium imaging features that usually enable the correct diagnosis [5]. However, an atypical appearance of hemangioma is frequently encountered in daily practice. It has been shown that perilesional parenchymal enhancement (PPE) depicted on arterial phase images, previously thought to be a feature associated with a malignant lesion, has been found in benign liver masses, in particular hemangiomas with an incidence that can be as high as 26% [6–11]. PPE is appreciated as a transient increased contrast enhancing area adjacent to a lesion in the arterial phase [6, 11]. Previous studies have shown that PPE is most commonly encountered in rapidly enhancing small hemangiomas [9, 11–13]. The pathophysiological mechanism of temporal PPE is not completely clarified yet, but has been proposed an association with arterio-venous fistula [6, 8–13]. One small study [8] reported that hemangiomas with PPE were located less than 2 cm from the liver capsule, speculating an association between temporal PPE adjacent to hemangiomas with their location in the liver. However this has not been documented in a study involving a large consecutive population. Therefore, our purpose was to evaluate the occurrence rate of temporal PPE enhancement associated with hepatic hemangiomas in a large consecutive series, and to determine which aspects are associated with this observation.
M. S. C. Sousa et al.: Perilesional enhancement of liver hemangiomas on MRI
Materials and methods Patients Institutional review board approval was obtained for this retrospective study with waived informed consent. A computerized search of the database of the main institution was performed using the keywords ‘‘hemangioma’’ and ‘‘angioma’’ for consecutive patients between 1/2/2008 (mm/dd/year) and 1/2/2012 that underwent abdominal magnetic resonance imaging (MRI) examination. Inclusion criteria for this study were patients who were at least 18 years of age, who underwent abdominal MRI, with one or more hepatic nodule(s) that met the criteria for liver hemangiomas, as defined below. If a patient had multiple follow-up MR studies, the examination with the best arterial phase and least imaging artifacts was selected. Criteria used for the diagnosis of hemangioma were based on the pathognomonic combined appearance on pre and post-contrast images as follows: (i) a well defined hepatic lesion with moderately high T2-weighted signal intensity and moderately low T1-weighted signal intensity; plus (ii) a discontinuous ring of peripheral nodules or complete or near complete flash-filling enhancement on post-gadolinium arterial phase images; and (iii) centripetal progression or uniform enhancement in the late phases [5]. The inclusion criterion for adequate arterial phase timing represented images acquired during an imaging window starting in the late arterial phase (predefined as the presence of contrast in the renal arteries and renal veins), and ending in the hepatic arterial dominant phase (predefined as the presence of contrast in the portal vein and the absence of contrast in the hepatic veins) [14]. A total of 251 examinations were reviewed. Patients were excluded if they had any of the following: history of previous chemotherapy or prior liver resection or hepatic intervention (n = 2), portal vein thrombus (n = 3), atypical enhancement pattern (n = 3), and the absence of post-gadolinium dynamic evaluation or inadequate arterial phase (n = 19). The study population included 513 liver hemangiomas in 224 patients (104 males and 120 females; mean age of 55.2 ± 13.5 years; age range 24–89 years). The use of MRI was at the discretion of the referring physicians for all examination.
MR technique and image analysis All subjects underwent MRI of the abdomen with a 1.5-T system (GE-Signa HDx, GE Healthcare) using a phasedarray torso coil. As part of our standard abdominal MRI protocol, the following sequences were performed: axial unenhanced T1-weighted dual-echo in-phase/out-of-
phase (125 msec /4.3 or 2.1 msec, 80 flip angle) 2D gradient-echo (GRE) sequences; coronal T2-weighted halfFourier single-shot fast spin-echo sequences (1894.7/83.9); axial T2-weighted half-Fourier single-shot fast spin-echo sequences (1800/89.9) with and without fat suppression; and axial pre and post-gadolinium fat-suppressed 3D GRE (4.17/1.98, 12 flip angle) at the arterial, venous and interstitial phases. Intravenous gadoterate meglumine (Dotarem, Guerbet, Paris, France) was administered as a powerinjected (Medrad, Pittsburgh, PA) bolus of 0.1 mmol/ kg at 2 ml/s in all subjects, followed by a bolus of 20-ml saline flush. The arterial post-contrast sequences were acquired with an empirical delay time calculated to initiate acquisition at 20 s after the start of contrast injection. The portal venous sequence was acquired at 60–80 s after the start of injection, and the interstitial/equilibrium sequences were acquired at 3–4 min.
Image analysis All examinations were reviewed and interpreted by two readers: M. Ramalho, a radiologist with 8 years of experience in abdominal imaging, and M. S. C. Sousa, a senior (4th year) radiology resident. All images were reviewed by using a PACS Barco monitor and DICOM viewing software (Centricity-Ris, GE Healthcare). The frequency of PPE, the speed of lesion enhancement, location and size of each hemangioma were measured independently. Disagreements between the reviewers were resolved by consensus with a third radiologist (V. Here´dia) with 7 years of experience in abdominal imaging. PPE was defined as an area of transient increased enhancement either in conjunction or not, with early opacification of vascular structures adjacent to the tumor, as displayed during the early phase of dynamic MRI. PPE was diagnosed when images showed circumferential or wedge-shaped or homogeneous enhancement in the liver parenchyma adjacent to the hemangioma in the arterial phase of dynamic MRI and faded to isointensity or notably less hyperintense in the portal phase. Speed of lesion enhancement was appreciated at arterial phase and subjectively separated in three types (Fig. 1): (i) Type1, with approximately 100% of intratumoral enhancement extent; (ii) Type2, with 60–99% of intratumoral enhancement extent; and (iii) Type3, with less than 60% of intratumoral enhancement extent. The percentage of enhanced tumor volume was estimated by evaluation of contiguous sections that demonstrated this lesion. The location of hemangiomas was defined as subcapsular when positioned at 10 mm or less from the hepatic capsule and deep located when positioned at more than 10 mm from the liver capsule. The two readers
M. S. C. Sousa et al.: Perilesional enhancement of liver hemangiomas on MRI
Fig. 1. Axial fat-suppressed T2-weighted half-Fourier single-shot fast spin-echo images (A, C, E) and gadolinium-enhanced fat-suppressed T1-weighted 3D gradient-echo MR images obtained at the arterial phase (B, D, F). Patient 1 (A, B) shows three well defined hepatic hemangiomas with moderately high T2-weighted signal intensity (arrows, A). Type1 speed of enhancement (approximately 100% of intratumoral enhancement extent) is seen in the lesion present on segment VIII (arrow, B). Note the adjacent perilesional enhancement in this hemangioma with a deep location. Pa-
tient 2 (C, D) shows one hemangioma with moderately high T2-weighted signal intensity with is seen on segment II (c). Arterial phase image (D) shows Type2 speed of enhancement with 60–99% of intratumoral enhancement extent. Patient 3 (E, F) shows two well-defined hepatic hemangiomas with moderately high T2-weighted signal intensity (E) are seen in segment VIII, demonstrating the typical discontinuous ring of peripheral nodules on the arterial phase (F), which was considered to be a Type3 speed of enhancement (less than 60% of intratumoral enhancement extent).
M. S. C. Sousa et al.: Perilesional enhancement of liver hemangiomas on MRI
Table 1. Statistical analyses Complete modele
Simple model PPE Liver segmentationa II III IV A IV B V VI VII VIII Lesion sizeb Small Medium Large Speed of lesionc T2 T3 Subcapsular
OR
P value
95% CI
0.98 0.62 0.08 0.68 0.29 0.78 0.84 0.61
0.980 0.629 0.026 0.711 0.233 0.798 0.855 0.608
[0.15; 6.45] [0.09; 4.4] [0.01;0. 74] [0.09; 5.31] [0.04; 2.21] [0.12; 5.16] [0.12; 5.7] [0.09; 4.12]
0.76 0.46 0.15
0.282 0.035 0.018
[0.46; 1.26] [0.22;0. 95] [0.03;0. 72]
0.05 0.07 7.66
