| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters to the Editor |
Department of Radiology, University of Tsukuba, Institute of Clinical Medicine, 1-1-1 Tennodai, Tsukuba 305-8575, Japan, e-mail: itaiy@md.tsukuba.ac.jp
Editor:
We read with great interest the article by Dr Semelka and colleagues in the April 2000 issue of Radiology (1). They reported an intense perilesional enhancement of metastatic liver tumor on early gadolinium-enhanced magnetic resonance (MR) images that strongly correlated with histopathologic hepatic parenchymal changes, including peritumoral desmoplastic reaction, inflammatory cell infiltration, and vascular proliferation.
We have been performing dynamic computed tomography (CT) and MR imaging for many years, and perilesional (extratumoral parenchyma) enhancement of the tumor is one of the important aims (2,3). We have questions and comments about the figures and patients. First, why was the precontrast image not shown in figure 1? Needless to say, a comparison of an enhanced image with a precontrast image is essential. Second, there was a definite enhanced rim in figure 2 (example of no perilesional enhancement) that might be interpreted as being within the tumor. Third, wedge-shaped perilesional enhancement was depicted in figure 3b. However, was there any wedge enhancement stated in table 2?
There are several factors to clarify in the investigation of perilesional enhancement. Is the enhancement located outside of the tumor, in the peripheral part of the tumor, or both? When does the enhancement start, and how long does it last? Is the pattern of enhancement circumferential (ring or irregular) or wedge-shaped? Regarding the first factor, it is difficult to determine the exact location unless the margin of the tumor is distinct on the precontrast image and unless exactly the same levels are depicted on pre- and postcontrast images.
We performed a CT-pathologic correlation study of metastatic liver tumor that was exposed to the hepatic capsule by using helical CT during infusion hepatic arteriography; we confirmed perilesional enhancement (3). Perilesional enhancement can be caused mainly by an increased supply of arterial flow, a draining of tumoral blood (4), or an increased interstitial space in the hepatic parenchyma due to fibrotic and desmoplastic changes (2,4). In the current study, the third item was ruled out. However, the second item could not be excluded completely with a 20-second image. Close–time interval single (same)-level dynamic study is most useful to analyze the hemodynamics of liver tumor and the surrounding parenchyma (2,4).
Since blood flow, either supplying or draining, may affect an area larger than that affected by histopathologic change in a vessel, quantitative imaging-pathologic correlation seems difficult, especially when imaging is performed with bolus injection of contrast material. However, an extraordinarily high correlation (r = 0.99) was noted between the thickness of enhancement and the thickness of the histopathologic tumor border (ie, 0.1-mm histopathologic change exactly correlated with a 1–2-mm-wide enhancement) (1). Still, there was no description of whether the corresponding areas of the largest enhanced parts were histopathologically examined. At least in a case of wedge-shaped enhancement, histopathologic study should be performed in the corresponding area, since wedge-shaped enhancement occurs in a part that contacts the tumor and that mostly correlates with portal and/or hepatic venous obstruction or arterioportal shunt.
In our experience with histopathologic study (3) and peritumoral sparing in fatty liver (reflecting an area of portal flow absence) (5), the change in width was not necessarily homogeneous in a lesion. If the tumor border thickness is uniform, or if the mean thickness correlated with the thickness of enhancement, the difference in diameter on early post- and precontrast images should be used after it is divided by a factor of 1–2 (near 1 for marked wedge-shaped enhancement and 2 for uniformed circumferential enhancement) according to the uneven width of both sides of the enhanced areas.
Dr Semelka and colleagues represented the disappearance of perilesional enhancement after chemotherapy in a patient (seemingly during chemotherapy and not satisfying the entry requirement). However, with single-level dynamic CT and/or multiphase dynamic CT and/or MR imaging, changes of perilesional enhancement pattern are occasionally encountered, even in patients who are repeatedly examined in short intervals without any treatment. Summation of small differences in the injection speed, image timing, circulation state, and other unknown factors may be responsible for this phenomenon. We may know the general rule from examining only one patient, but we are concerned that this rule often leads to an erroneous conclusion.
In conclusion, Dr Semelka and colleagues performed imaging-pathologic correlation and presented an interesting concept of tumor border thickness. However, their data were not properly used to lead to a definite conclusion.
REFERENCES
Department of Radiology, University of North Carolina School of Medicine, CB 7510, 101 Manning Drive, Chapel Hill, NC 27599-7510, e-mail: richsem@med.unc.edu
My colleagues and I appreciate the interest of Drs Itai and Irie in our article (1). We also have extensive experience, with 10 years of experience performing gadolinium-enhanced MR imaging of the liver in relatively large numbers of patients and writing research publications (1–17). The various aspects of early enhancement features are intriguing, and our understanding of them is still incomplete.
In an effort to clarify some of the information in the article and to expand this discussion with our clinical experience and theories, I will begin by defining the difference between perilesional and lesional enhancement of the liver tumors on images obtained immediately after gadolinium enhancement. In general, lesional enhancement shows relatively sharp external borders (with the notable exception of invasive neoplasms, such as some infiltrating hepatocellular cancers). When enhancement is lesional with no contribution of perilesional enhancement, the lesion size on precontrast images and images obtained immediately after the administration of contrast material is essentially identical. This information is reflected in table 2 (1). Examples of lesional enhancement patterns are ring (generally metastases) (2, 4–6), diffuse homogeneous (adenoma, focal nodular hyperplasia, some metastases, some hemangiomas, and some hepatocellular carcinomas) (2,4–9), peripheral nodular with a discontinuous ring (hemangioma) (2,9), and diffuse heterogeneous (generally hepatocellular carcinoma) (2,4,7).
Perilesional enhancement is defined as enhancement that occurs beyond the confines of the tumor in the surrounding hepatic parenchyma (1,2,4,8–12). This can be appreciated as enhancement of the liver lesion on images obtained immediately after contrast enhancement that is larger than the dimensions of the lesion shown on nonenhanced images (1, table 2). Perilesional enhancement can be either wedge-shaped or circumferential, with wedge-shaped enhancement often well defined (10,11) and circumferential enhancement often ill defined (1,2). Figures 1 and 3 (1) illustrate the ill-defined circumferential pattern. In general, both types of perilesional enhancement are best seen or only seen on images obtained immediately after the administration of gadolinium-based contrast material. The wedge-shaped pattern may persist on more delayed images. It is crucial that the timing of contrast material administration is both precise and reproducible to capture the hepatic arterial–dominant phase in the demonstration of perilesional enhancement (2,3). Correct timing is characterized by contrast enhancement in the hepatic arteries and portal veins and a lack of contrast enhancement in the hepatic veins.
The sources of this perilesional enhancement are varied and not fully understood. My belief is that, as Drs Itai and Irie described, wedge-shaped enhancement is often due to the compression of the subtending portal vein (10,11), but it may also result from the same process that causes indistinct circumferential perilesional enhancement. This latter enhancement pattern, by chance, happened to be the most common perilesional enhancement pattern that we observed in our small group of patients in the current study (1). The indistinct circumferential pattern is most often only transiently observed on images obtained in the hepatic arterial–dominant phase and does not persist on later postcontrast images.
It is interesting to note which tumors commonly exhibit this pattern. In our clinical experience, the most common tumors that show an indistinct circumferential pattern are colon cancer metastases (1,2,6), followed by pancreatic ductal adenocarcinoma (2,6), hepatic lymphoma (13), and breast cancer. Often, these tumors are not hypervascular but hypovascular; therefore, the pattern does not reflect a hepatic arterial sump effect from hypervascular tumor that draws increased blood flow to the liver subsegment. This finding is supported by the fact that this perilesional enhancement is generally not present in truly hypervascular lesions in which a hepatic arterial sump effect should be most prominent (5,6). We have also observed lack of perilesional enhancement in hypervascular metastases in our extensive experience with neuroendocrine metastases (5,14), carcinoid metastases, and renal cell cancer metastases.
In the present study (1), we demonstrated that increased perilesional enhancement is associated with histologic changes in the margins surrounding the tumor, which our pathologist has termed the histologic tumor border (defined as the histologically altered liver parenchyma surrounding the tumor, not the outer portion of the tumor). The perilesional enhancement on MR images was thicker than the histologic tumor border; therefore, enhancement was not confined to the tumor border alone but also extended into the surrounding histologically normal liver. This finding is what we described in our study, and although we considered it clear originally, I hope this description provides further clarification.
The previous discussion does not address in itemized form the issues raised by Drs Itai and Irie, as we wanted to provide an overview on the subject of lesional and perilesional enhancement. Many of their points are incorporated into this discussion. Responses to other issues that they brought up are the following: First, to minimize the number of illustrations, we had opted not show the nonenhanced image in figure 1. Second, they are correct that there are differences in the thickness of circumferential perilesional enhancement around the lesion, but we selected a representative dimension to measure. Third, we did not imply an exact 0.1-mm tumor border to a 1-mm perilesional enhancement, but there was a direct proportional relationship that had a high correlation.
We have not performed studies, as Drs Itai and Irie have described, that demonstrate changing patterns of perilesional enhancement from study to study, when no therapeutic measures have intervened. We agree that, likely, the most common cause of these changes is variation in the timing of contrast material administration, with changes in the patient’s cardiac output likely the second most common cause. The timing of contrast material administration is crucial, and, in this regard, we are in accord with Drs Itai and Irie. Changes in the perilesional enhancement observed between studies, when images are not acquired in the exact hepatic arterial–dominant phase, are likely artifactual. A real component to this variation may reflect variations in the biologic activity of the liver lesions and variations in their release of humoral angiogenic substances. We consider the concluding assertions of Drs Itai and Irie unwarranted, since our study findings do demonstrate a correlation between perilesional enhancement and the thickness of the histologic tumor border, which is what we observed.
We appreciate the interest that Drs Itai and Irie have in our work; more research should be performed to study early enhancement features of liver lesions—findings that we believe may provide important information on the biologic activity of liver tumors. Any differences between our findings and those of Drs Itai and Irie are a reflection that our understanding of liver lesion biology is still in its seminal stages.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
