| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Radiology, Children's Hospital, Harvard Medical School, 300
Longwood Ave., Boston, MA.
2 Department of Radiology, Division of Musculoskeletal Radiology, Massachusetts
General Hospital, Harvard Medical School, 15 Parkman St., Ste. 515, Boston, MA
02114.
3 Department of Biostatistics, Children's Hospital, Harvard Medical School,
Boston, MA.
4 Department of Medical Imaging, Hôpital Sainte Justine, Université
de Montréal, Montréal, QC, Canada.
5 Department of Surgery, Children's Hospital, Harvard Medical School, Boston,
MA.
Received April 21, 2003;
accepted after revision September 16, 2003.
Address correspondence to A. Kassarjian.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. The imaging studies and clinical records of all patients with infantile hepatic hemangiomas from two tertiary children's hospitals were reviewed. Univariate and multivariate stepwise logistic regression techniques were used to determine whether clinical presentation and imaging variables differentiated the type of treatment required.
RESULTS. Typical hemangiomas appeared as focal or multifocal T2-hyperintense spheres with centripetal contrast enhancement and dilated feeding and draining vessels. Three atypical patterns included focal mass lesions with central varix with or without direct shunts, focal mass with central necrosis or thrombosis, and massive hemangiomatous involvement of the liver with abdominal vascular compression. In general, patients with focal lesions without high flow needed no treatment, and those with central varix and direct shunts developed severe high-output cardiac failure that responded quickly to embolization. The pattern of massive replacement of liver was associated with hypothyroidism, abdominal compartment syndrome, and a high mortality rate. Multivariate analysis of 55 patients indicated that congestive heart failure was the only independent predictor of treatment (p = 0.005). The presence of a shunt was the only independent factor associated with embolization or surgery (p = 0.002).
CONCLUSION. Although the imaging features of infantile hepatic hemangiomas vary to some extent, MRI features are typical in most patients and certain imaging findings are predictive of the clinical course. MRI is the technique of choice in diagnosing infantile hepatic masses.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The purpose of this study was to describe imaging findings in patients with infantile hepatic hemangiomas and identify imaging predictors of the need for and the response to different types of therapy.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Imaging files of patients with infantile hepatic hemangioma, including sonography, CT, MRI, and angiographic studies, were concurrently reviewed by four radiologists, including three pediatric radiologists with extensive experience in imaging of vascular anomalies. The fourth radiologist was a senior radiology resident with extensive experience in the imaging characteristics of infantile hepatic hemangiomas. Imaging characteristics that were recorded included distribution of lesions (focal vs multifocal), vascularity of lesions (high flow vs low flow), echogenicity, presence of calcifications, signal intensity and enhancement pattern on MRI, presence of abrupt tapering of the aorta distal to the celiac axis (on sonography, MRI, and angiography), presence of abnormal flow voids (on MRI) or abnormal vessels (on sonography), and presence of shunts (direct arteriovenous or portovenous communications on angiography or sonography or arterialization of venous waveforms on Doppler sonography). Multifocality was defined as three or more separate lesions. For patients who did not undergo angiography or Doppler sonography, high-flow lesions were defined as those with imaging evidence of severe cardiomegaly, marked dilatation of the hepatic veins and the right atrium, or abnormal flow voids on MRI.
Institutional review board approval was obtained from both hospitals.
Univariate analysis was performed using Fisher's exact test for comparing proportions. Logistic regression analysis was performed to determine whether clinical presentation and imaging variables differentiated the need for treatment and the type of treatment provided. Nine candidate variables were considered: age, sex, presence of cardiac insufficiency, flow, aortic tapering, distribution of lesions, presence of a shunt, calcifications, and abnormal vessel findings on sonography. The same nine variables were tested to see whether any were predictors of mortality. A backward stepwise procedure was used to determine the final multivariate predictors using the likelihood ratio chi-square test to evaluate statistical significance [8]. All reported p values are two-tailed with an alpha level of 0.05 used to indicate significance. Data analysis was conducted with SPSS software (version 11.0, Statistical Package for the Social Sciences, Chicago, IL).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Clinical
The median age at presentation of the 55 patients was 30 days (range, 1
day–3 years 6 months). In general, patients received therapy in the
presence of high-output congestive heart failure, symptomatic mass effect, and
extensive or increasing disease. Patients without symptoms or with stable
lesions did not receive therapy.
Twenty-eight patients (51%) showed cardiac insufficiency. Six (15%) of 39 patients had laboratory evidence of hypothyroidism. Of the 55 patients, 13 (24%) did not receive treatment for their infantile hepatic hemangioma. Eleven (20%) were treated with steroids alone and 13 (24%) were treated with steroids and interferon alfa-2a. Eighteen (33%) of 55 did not respond adequately to medical therapy (steroids or interferon); of these, 13 (24% of all patients or 72% of those not responding to medical therapy) underwent transcatheter embolization and five (9% of all patients or 28% of those not responding to medical therapy) underwent successful surgical resection. Six patients died. Clinical and radiologic characteristics are presented in Table 1.
|
Among the six patients who died, one patient had a focal hemangioma, confirmed by angiography and open biopsy at age 3 weeks, after admission to the hospital with tachypnea. He had been cyanotic and floppy at birth, but responded to treatment with blood transfusions and was discharged home at age 6 days. After diagnosis of the hemangioma, he appeared to be doing well at home, but died unexpectedly at age 3 months. Postmortem examination confirmed a solitary hepatic hemangioma without evidence of high flow. The cause of death was not determined, although acute respiratory infection was noted.
The other five patients all had multifocal hemangiomas and received multiple therapies. One patient who also had hypothyroidism was treated with steroids, interferon, radiation, and vincristine without response. She died of sepsis and multiorgan failure. Postmortem examination showed hepatic venoocclusive disease and extensive hepatic hemangiomas. Another patient who failed medical therapy was referred to a liver transplantation center but died of multiorgan failure before transplantation. The third patient was treated with steroids, embolization, and radiation therapy but died of intracranial hemorrhage and subsequent hemorrhagic shock. The fourth was treated with steroids, interferon, embolization, radiation therapy, and abdominal fasciotomy for abdominal compartment syndrome but died of sepsis and multiorgan failure. The fifth patient was treated with steroids and interferon and appeared to be responding. This child died unexpectedly at home, probably because of pulmonary hypertension. Of the six infants who died, five had multifocal infantile hepatic hemangioma, four did not initially respond to medical therapy, and four had documented hypothyroidism. A fifth child had a clinical course and imaging findings similar to the others with hypothyroidism but was not assessed for thyroid function.
Imaging Findings
Forty-six patients (84%) underwent several types of imaging. Sonography was
the most common and usually the primary imaging technique. Results of the
imaging findings are listed in Table
1.
In general, the hemangiomas were well-defined hypoechoic lesions on sonography (Figs. 1A, 1B, 2, 3A, 3B). In two cases of diffuse hemangiomatous involvement of the liver, sonography showed the liver to be enlarged and heterogeneous in echo texture, but did not delineate discrete lesions. Calcifications could be seen with sonography in 16 (36%) of 44 patients. A few hemangiomas were well-defined hyperechoic masses in six (14%) of 44 patients. Overall, sonography showed discrete lesions in 42 of 44 patients for a sensitivity of 95% when compared with MRI or angiography. Although sonography showed abnormal vessels in 25 (60%) of 42 patients and the presence of shunting was confirmed in 10 (44%) of 23 who had Doppler imaging, the exact angioarchitecture was difficult to discern with sonography. Sonography was 90% sensitive in identifying direct shunts when compared with angiography in nine of 10 patients.
|
|
|
|
|
CT findings were dependent on the timing of the scan relative to the administration of contrast medium. Before contrast medium administration, most lesions were hypodense compared with the liver. Multifocal lesions were generally spherical and homogeneous in their density. They enhanced in a centripetal fashion, with the periphery of a lesion enhancing first. If the scan was delayed, the entire lesion generally enhanced uniformly. Focal lesions were more heterogeneous in their appearance with large hepatic arteries and hepatic veins and variable enhancement, usually centripetal but often sparing the center of the lesion. Like sonograms and MR images, CT studies showed evidence of high flow with dilatation of the proximal abdominal aorta, hepatic arteries, and hepatic veins and reduced size of the infrahepatic aorta. Some focal hemangiomas contained calcification.
Hemangiomas were consistently seen on MRI as well-defined spherical lesions that were hypointense relative to liver on T1-weighted sequences and strongly hyperintense on T2-weighted sequences (Figs. 4A, 4B, 4C and 5). Flow voids were visible either adjacent to or in some of the lesions in 19 (82%) of 23 of patients. None of the lesions was associated with surrounding edema. In the 14 patients who underwent MRI with gadolinium, the lesions showed centripetal enhancement. Delayed contrast-enhanced images showed homogeneous enhancement (in the absence of necrosis). All multifocal lesions had homogeneous signal intensity.
|
|
|
|
Focal lesions that contained necrotic or thrombosed regions or had central vascular spaces showed more heterogeneous signal intensities on both unenhanced and contrast-enhanced MR images as well as nonenhancing areas (Figs. 6A, 6B, 6C, 6D and 7). The solid nonthrombosed parts of the lesions showed intense homogeneous contrast enhancement as did the central varix, if present.
|
|
|
|
|
Six patients had diffuse hemangiomatous involvement of the liver that resulted in massive hepatomegaly with compression of the inferior vena cava and thoracic cavity to the point of causing respiratory compromise (Fig. 8). All six were hypothyroid, and innumerable well-defined spherical lesions were seen on MRI in each patient. The lesions had typical signal characteristics and homogeneous or centripetal enhancement as seen in multifocal hemangiomas in patients not known to be hypothyroid. In general, although these patients had high-flow lesions, the dilatation of the hepatic veins and right atrium was subjectively less marked than would be expected, given the mass of tumor present. The patients with hypothyroidism had morbidity related to tumor volume rather than cardiac insufficiency.
|
The angiographic findings in this group of patients with infantile hepatic hemangioma have been reported separately [9]. Briefly, five types of angiographic patterns were identified and classified according to the presence of abnormal (predominately low-flow) vascular channels, high-flow nodules, and direct shunts. Thirteen (87%) of 15 patients who were studied with angiography underwent embolization.
Four patients identified in our database search had a preliminary diagnosis of infantile hepatic hemangioma. Review of imaging studies revealed atypical features. All four of these patients underwent biopsy because of their imaging findings and were found to have other hepatic lesions. One patient was a 3-year-old girl with a history of left hepatic lobectomy for infantile hepatic hemangioma. Follow-up MRI showed multifocal liver lesions in the right lobe that were hypointense on T1-weighted images and hyperintense but inhomogeneous on T2-weighted images. Gadolinium-enhanced images showed central enhancement, unlike the typical peripheral enhancement seen in infantile hepatic hemangioma (Fig. 9). The age at presentation and the central enhancement were not consistent with infantile hepatic hemangioma, and a subsequent biopsy showed the lesions to be angiosarcomas.
|
Another patient had cutaneous and multifocal liver lesions that were hypointense on T1-weighted images and hyperintense on T2-weighted images. However, one of the lesions contained a fluid–fluid level consistent with hemorrhage (Fig. 10). Contrast-enhanced imaging showed only mild peripheral enhancement. The nodules did not have typical associated flow voids. The fluid level and mild degree of enhancement prompted a biopsy that showed metastatic neuroblastoma. A third patient, a 20-day-old boy, had a large varix with direct arterioportal shunts. However, both sonography and contrast-enhanced MRI failed to show an associated solid mass. The patient had been treated with steroids without response. Because of the lack of an enhancing mass and the lack of response to steroids, angiography was performed and revealed an arterioportal fistula, which was treated successfully by embolization. Finally, a 9-month-old boy had multifocal liver lesions that were multilobulated and necrotic on sonography. They were hypointense on T1-weighted images, had increased but heterogeneous signal on T2-weighted images and showed only mild contrast enhancement. Biopsy revealed hepatoblastoma.
|
Correlation of Clinical and Radiologic Findings with Treatment Outcomes
Of the 55 patients with infantile hepatic hemangioma, 42 (76%) received
medical, endovascular, or surgical treatment, although 13 did not receive any
treatment aside from observation (Fig.
11). We examined the data on these 55 patients to see which, if
any, of the imaging or clinical factors we recorded could predict the need for
treatment.
|
Table 2 presents the results of the logistic regression analysis to identify variables that distinguished patients who received treatment for infantile hepatic hemangioma from those who did not. The variables associated with the need for treatment in the univariate analysis were the presence of congestive heart failure (p < 0.001) and aortic tapering (p = 0.004). Ninety-six percent of patients with cardiac insufficiency (27/28) received therapy, but only 56% (15/27) of those without cardiac insufficiency received therapy (p < 0.001). Multivariate analysis revealed that the presence of congestive heart failure was the only independent predictor of the need for therapy (p = 0.005). The presence of cardiac insufficiency indicated a greater than 20-fold increase in the likelihood that a patient would receive therapy for infantile hepatic hemangioma (95% confidence interval [CI], 3.6–129.6). None of the imaging or clinical features was significantly associated with cardiac insufficiency.
|
We evaluated the patients who were treated for infantile hepatic hemangioma to ascertain whether clinical or imaging parameters could differentiate which patients could be adequately treated with medical therapy (steroids or interferon alfa-2a) and which would require additional therapy such as embolization or surgery (Table 3). In the patients who were treated adequately with medical therapy, we identified no significant predictors for the failure of steroids and subsequent need for interferon. The presence of a shunt was the only independent factor associated with the inability of medical therapy (steroids with or without interferon) to adequately control a patient's symptoms (p = 0.002). Patients with direct shunts were 20 times more likely to fail pharmacologic therapy for infantile hepatic hemangioma and thus require either embolization or surgery (95% CI, 2.8–141.8). Results of the univariate and multivariate analyses of the nine variables tested failed to show any explanatory predictors of death.
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Clinical
The patients with infantile hepatic hemangioma that are presented in this
series differ from most patients with infantile hepatic hemangioma with regard
to sex distribution, age at presentation, symptoms, need for treatment, and
mortality. Although infantile hepatic hemangiomas usually have a 3:1
female-to-male ratio, the ratio of females to males was 4:3 in this series. In
addition, although most infantile hepatic hemangiomas are asymptomatic lesions
that do not require treatment and have low morbidity and mortality, most
patients in this series were treated, but the mortality rate was 11%. These
differences are likely to reflect selection bias induced by the fact that the
patients in this study had been referred to tertiary children's hospitals with
established vascular anomalies programs.
Imaging
Infantile hepatic hemangiomas have variable imaging characteristics and
angioarchitecture. We have made some generalizations from this relatively
large series that may be useful in distinguishing typical hemangiomas from
lesions that require biopsy for diagnosis.
Multifocal hemangiomas are spherical lesions with homogeneous signal intensity on MRI, including decreased signal on T1-weighted sequences and increased signal on T2-weighted sequences. Typically, these lesions have vascular flow voids adjacent to or in some of the nodules and, except in small lesions, show evidence of high flow (dilated hepatic arteries and veins). They enhance either homogeneously or centripetally after contrast administration for MRI or CT studies. They may have associated shunts (direct arteriovenous or portovenous communications) and abnormal dilated hepatic veins [9, 12–14].
Focal lesions may have homogeneous or inhomogeneous imaging features. Centripetal contrast enhancement and focal signal abnormality are present in all lesions, but some have nonenhancing central areas. Focal hemangiomas may have low- or high-flow characteristics. The latter may contain direct arteriovenous or portovenous shunts. A central varix and anomalous draining hepatic veins may be present. This feature was used to diagnose one patient at 28 weeks' gestation, and the fetus was successfully treated by maternal administration of steroids [15].
Sonographic findings are variable, including hypoechoic or hyperechoic nodules and abnormal vascular channels and shunts. Sonography with Doppler interrogation is useful to diagnose most lesions and to follow patients after initial diagnosis [16]. However, a specific diagnosis of hepatic hemangioma may be difficult to make in the absence of enlarged vessels associated with the lesions. Individual nodules may be poorly seen in some patients with diffuse involvement of the liver.
MRI revealed discrete lesions in all the patients in this series who underwent MRI, but angiography and sonography sometimes fail to identify the discrete nodules. MRI, including dynamic contrast-enhanced and gradient sequences, is probably the best technique for diagnosis.
Lesions that present after age 1 year or that do not have the features described earlier, especially homogeneous solid components and centripetal enhancement, should undergo biopsy.
It remains to be seen how MDCT with its capability to perform high-quality MDCT angiography and more sophisticated MR angiography and MR venography with subtraction techniques will contribute to the diagnosis and characterization of atypical hepatic hemangiomas and whether these techniques will eventually be able to adequately identify and quantify any associated shunts. Angiography should be reserved for patients with symptomatic shunts in whom endovascular therapy is anticipated.
The presence of diffuse hemangiomatous involvement appears to be associated with a more complicated clinical course that may include hypothyroidism, probably because of the high levels of type 3 iodothyronine deiodinase activity produced by hemangiomas [17]. In our series, four of the six patients with hypothyroidism had a poor response to multiple therapies and died of complications including sepsis, multiorgan failure, pulmonary hypertension, and intracranial hemorrhage. They all had massive tumor volume resulting in compression of adjacent organs and vessels, rather than the more common pattern of high-output congestive cardiac failure. Another patient who had imaging appearances and a clinical course similar to these six patients died of hemorrhagic shock despite aggressive treatment including steroids, embolization, and radiation therapy. Unfortunately, thyroid function tests were unavailable in this patient. The other two hypothyroid patients are currently responding to medical therapy. Clearly, patients with extensive infantile hepatic hemangioma should have thyroid function assessment. Because patients in this series did not routinely undergo thyroid function studies, it is difficult to know the apparent relationship between lack of response to therapy and hypothyroidism.
Need for Treatment
Although steroids and, more recently, interferon and vincristine have been
the mainstays for treating hemangiomas, these therapies may be associated with
side effects including cardiomyopathy, spastic diplegia, and complications of
central venous catheters [18].
In light of these potential side effects, it would be useful to identify
factors predicting which hepatic hemangiomas are likely to require and respond
to medical treatment and which subset is likely to require additional
treatment methods including embolotherapy or surgery. Such knowledge could
potentially obviate side effects of therapy in some patients.
In our series, only the presence of cardiac insufficiency was an independent predictor of the need for therapy. Patients who developed cardiac insufficiency were more than 20 times more likely to receive treatment, which is not surprising, because the most common morbidity related to this lesion is high output cardiac failure. None of the recorded clinical parameters could predict response to therapy.
In the patients who required therapy, only the presence or absence of a shunt as shown on either sonography or angiography was an independent predictor of the inability of medical therapy to control symptoms. Patients with a shunt (arteriovenous, arterioportal, or portovenous fistula) were 20 times more likely to require embolization or surgery than those without shunts. Therefore, although clinical parameters seem to indicate which patients will require therapy, imaging findings appear to be the only predictors of the need for nonpharmacologic therapy such as embolization. Although this observation supports our clinical impression, it must be acknowledged that the presence of a shunt could have been missed in patients who did not undergo angiography or Doppler sonography.
The patients who died because of massive hepatic replacement and abdominal compartment syndrome probably would have been appropriate candidates for liver transplantation. None of these patients responded to pharmacologic treatment. Embolization was not deemed to be useful therapy because they did not generally have high-output congestive failure.
This study did not address the ability to identify such patients before clinical deterioration, but the patients presented a fairly consistent clinical course. They all had extensive multifocal hepatic disease at the time of diagnosis. All patients tested for thyroid function were hypothyroid. Patients had relatively mild cardiac volume overload in relation to their tumor volume, and they all showed progression of disease on appropriate pharmacologic therapy. Prospective data collection would be important to confirm this generalization, but it appears that patients with this constellation of findings should be considered for liver transplantation after a finite trial of pharmacologic therapy.
Our study has a number of limitations. Infantile hepatic hemangiomas—particularly the ones with atypical clinical or imaging characteristics—are rare, so the number of patients in this study is small. In addition, because the patients included in this study were seen over the course of two decades and imaged in multiple medical centers in two countries, the availability and quality of imaging studies reviewed were variable. Importantly, only 14 patients were imaged with gadolinium-enhanced MRI, a technique that we believe provides the most accurate diagnosis. This may have resulted in some patients having resection of their lesions because their diagnosis was uncertain, not because of poor response to pharmacologic therapy. Also, the effectiveness of interferon to treat hemangiomas was not known when some of the patients presented, which may have led to the use of embolization or surgery in some patients who might have responded to interferon had it been available.
Although we are confident of the diagnosis of hepatic hemangioma in all our patients, only nine of the diagnoses were confirmed pathologically. The remainder were presumed to be hemangiomas on the basis of imaging characteristics and either spontaneous involution (13 patients), response to steroids (10 patients), interferon (11 patients), or embolization (12 patients). Finally, because the institutions involved in this data collection have established vascular anomalies centers, the patients who were referred to us may have included a disproportionate number with atypical imaging or clinical features. Prospective multiinstitutional studies would correct these limitations.
In conclusion, although some variability appears in the imaging features of infantile hepatic hemangiomas, typical MRI features appear in most patients, and certain imaging findings are predictive of the clinical course. MRI is the technique of choice in diagnosing infantile hepatic masses.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  D. C. Rivard and L. H. Lowe Radiological Reasoning: Multiple Hepatic Masses in an Infant Am. J. Roentgenol., June 1, 2008; 190(6_Supplement): S46 - S52. [Full Text] [PDF]
|
|
|
|
|
|
C.-H. Liu, D. C. Rivard, A. C. Silva, K. J. Burkholz, and S. D. Beaty Imaging of Focal Hepatic Lesions: Self-Assessment Module Am. J. Roentgenol., June 1, 2008; 190(6_Supplement): S65 - S68. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Peredo-Pinto, J. Stroger Jr, and R. Nagasubramanian Index of Suspicion in the Nursery NeoReviews, February 1, 2007; 8(2): e85 - e87. [Full Text] [PDF]
|
|
|
|
 |
|
 K. M. Nord, J. Kandel, J. H. Lefkowitch, S. J. Lobritto, K. D. Morel, P. E. North, and M. C. Garzon Multiple Cutaneous Infantile Hemangiomas Associated With Hepatic Angiosarcoma: Case Report and Review of the Literature Pediatrics, September 1, 2006; 118(3): e907 - e913. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
