Applied Radiology

The first step is high-quality tissue imaging using a 2D gradient-echo sequence and a single dose of gadolinium contrast. On the first-pass postgadolinium image, we look for the presence of contrast in the portal veins and a lack of contrast in the hepatic veins (Figure 1, top left). Approximately 60 to 90 seconds after the contrast injection, we image with fat suppression (Figure 1, bottom right), looking specifically look for such abnormalities as peritoneal disease, capsule-based disease of the liver, and certain neoplasms such as cholangiocarcinoma, which are well visualized on delayed imaging.

Approximately 15 to 20 minutes after the initial injection of gadolinium, we do high-definition MR angiography, using a 3D gradient-echo sequence and, typically, 20 mL of contrast (Figure 2). We routinely use power injection at a rate of 2 mL/sec. At that rate, the temporal resolution is good and it is possible to capture the hepatic arterial dominant phase, as well as specific enhancement characteristics of liver lesions.

Magnetic resonance angiography is especially useful when imaging patients with vascular anomalies. Figure 3 is an example of a replaced right hepatic artery. The source image on the top left shows the left hepatic artery arising from the celiac artery. The source image on the bottom left shows the take-off of the replaced right hepatic artery from the superior mesenteric artery, and a distal branch of the right hepatic artery. The combined multiple intensity projection (MIP) image shows both the left hepatic artery coming off the celiac artery, and a portion of the replaced right hepatic artery.

Detection and characterization

Although little has been reported or written about the theoretical roles for high-dose gadolinium MRI of the liver, its potential to improve the detection and characterization of liver lesions is very intriguing. One of the few reports on this topic came from Taupitz et al from the Charité University in Berlin. In 2000, they found that double-dose gadolinium contrast administration improved the definition of both enhancing nodules and the progression of enhancement in small hemangiomas when compared with single-dose contrast administration (M. Taupitz, personal communication). This finding may be important, as correct characterization of liver lesions is an essential part of evaluating patients with malignant disease. ^3-5 Double-dose contrast may be useful in improving characterization of lesions, such as hemangiomas, in patients being evaluated for possible malignancy. ⁵

High-dose gadolinium contrast may also be of value in improving lesion detection, particularly in assessing malignant lesions that are amenable to resection or other local therapy. The two most important of these lesions are colon cancer liver metastases, which typically are hypovascular, and hepatocellular carcinoma, which typically is hypervascular. ⁴

In imaging hypovascular colon cancer metastases, high-dose gadolinium contrast may improve the conspicuity of lesions by increasing enhancement of the ring. Appreciating ring enhancement is the single most important evaluation to make in correctly characterizing metastatic colon cancer metastases ^3,4 (Figure 4). Often, enhancement is best on delayed imaging, 90 seconds after contrast administration.

It is relatively easy to evaluate large lesions with a combination of T2-weighted, T1-weighted, and early and delayed postcontrast imaging. High-dose contrast, therefore, is likely to be most important in detecting small lesions, a crucial factor in selecting the type of therapy that is most appropriate for each patient.

In hepatocellular carcinoma, the tumors are usually small and hypervascular. Often, however, they must be imaged on the background of a heterogeneously enhancing cirrhotic liver. Although such lesions can be very difficult to see on precontrast T2- and T1-weighted images, they are easily appreciated on immediate post-gadolinium images. ⁶ The appearance of hepatocellular carcinoma as hyperintense lesions immediately following contrast administration once again raises the possibility that high-dose gadolinium studies may improve lesion detection and help guide patient management.

Figure 5 shows a typical small hepatocellular carcinoma. On T2-weighted precontrast imaging, it is difficult to appreciate the lesion at all, although it is slightly hyperintense on the T1-weighted sequences. On immediate post-gadolinium images, hepatocellular carcinoma shows very intense enhancement, surrounded by a hypoenhancing capsule around the lesion. On delayed post-contrast images, we appreciate rapid washout of the lesion and late capsular enhancement, which are very specific for hepatocellular carcinoma.

Pancreatic Imaging

In pancreatic imaging, just as in hepatic imaging, high-dose gadolinium contrast studies have the potential to improve the detection of both hypovascular and hypervascular lesions. For example, we would expect better definition of small hypovascular pancreatic ductal adenocarcinomas by showing improved enhancement of the surrounding pancreas.

We would expect even more consistently positive results when high-dose gadolinium contrast is used in the imaging of islet cell tumors, insulinomas in particular. These tumors are initially very small, in the range of 1 to 2 cm, and can be difficult to detect. It is likely that the use of high-dose gadolinium contrast media will improve the detection of small insulinomas, but this needs to be demonstrated in clinical studies.

Conclusion

High-dose gadolinium contrast plays a small role in hepatic and pancreatic imaging today. We use it primarily in studies that combine tissue imaging with MRA, but it is also valuable for the evaluation of 2 different organ systems in the same study. For example, high-dose gadolinium is useful when examining the liver, followed by MRA of the renal arteries.

In the future, high-dose contrast-enhanced MRI may fill a much larger and more important clinical role in the evaluation of malignant disease, enabling better detection and characterization of cancerous lesions in patients who are being considered for surgical management. Controlled clinical studies comparing low- and high-dose contrast MRI, as well as high-dose contrast-enhanced MRI and other imaging modalities, such as CT, are now needed to determine whether this potential will become a reality. *

Discussion

TG: Thank you very much for an excellent presentation. We have some time for discussion. Dr. Kramer?

LK: Can you distinguish between doses; that is, quantity of contrast versus injection rate, in trying to define margins, and having greater sensitivity?

RS: That's a good question. At this point in time, since most of the work that we've done has been with single-dose contrast, most people have been focusing on contrast injection rates. In the past, we used hand injection, and now we routinely use machine power injection at 2 mL/sec. At that rate, we have fairly good temporal resolution that we can get very nicely. We can capture the hepatic arterial dominant phase, and also the specific enhancement characteristics of liver lesions.

Going with a higher dose, I think we would still have to at least match that ejection rate. It's very important to maintain that very early enhancement, the hepatic arterial dominant phase. So with higher dose, it will still be very essential. We can't get away from fast injections. I think if anything, there may be more importance attached to fast injections.

HR: I had a question about your approach, which is dynamic in the sense of over several minutes, and is mostly driven by morphologic characteristics. Do you think there is any potential for parametric imaging of the earliest wash-in phases, washout of some of these lesions, region of interest, or maps as we've seen for brain or breast?

RS: Yes, I think that at this point in time I haven't appreciated that that is important. The reason is that in the liver, there are very specific enhancement characteristics. So we don't have to resort to temporal handling of contrast, because the various benign lesions and malignant lesions have very specific morphologic features in how they enhance. It's a very interesting topic for the same reasons that it's interesting in the brain and the breasts, in that angiogenesis and the pattern of blood supply, is crucial. But I think we have the advantage that there's better morphologic characteristics of the various liver lesions, so we haven't had to rely on dynamic handling or looking at enhancement rates and so on. But, as with many things, you borrow from one area to go to the next. One of the nice things about the relationship between brain and body imaging is that, generally speaking, many of the advancements have first come in imaging the brain. Then as things have become well developed in the brain, then other applications, like body, have developed. So it's interesting to see if the various approaches that are used in the brain can also be translated into, for instance, liver imaging. I think that may be very intriguing.

DR: I'd like to comment on that. I think we probably will see that coming, because just this year there was a paper showing that the risk of recurrence of hepatoma was a function of microvessel density, which we can probe with our dynamic contrast-enhanced methods. Therefore, we may not only have the ability to provide a specific diagnosis, but also to provide an estimate of prognosis or recurrence. So, these are intriguing functional methods that I think we are going to see applied more frequently to the liver in the future.

RS: Right, and along the same lines as angiogenesis, I think what is also very important is the response to various treatment methods. In those settings, looking at the different patterns of enhancement, different rates of enhancement may prove very important. So we're now starting a number of different studies, looking at angiogenic qualities of metastases, following chemotherapy and how it alters. I think that may be where physiologic information will become important.

MP: In the double-injection scheme, where you want to do some lesion characterization followed by vascular depiction, is there any effect from the first injection on the second injection that either makes it better or worse?

RS: Well, I think there certainly is some effect of the first injection on the second. But we've generally found that, most often, those kinds of effects are also worse. But we found that by waiting approximately 15 minutes, that enough of the gadolinium has left the system that there isn't much of an impact on the MRA. In addition, since we routinely now look at the images on PACS, and we page the source images, we found that the presence of contrast hanging around another structure becomes less important than it had been in the past, when we weren't using the PACS system. In fact, in some areas, as I think Martin may have shown, it may in effect be of some interest that you can see at 15, 20 minutes contrast in the renal collecting system. So you can use that as further information on how the renal collecting system appears. So that is the one positive thing that we've observed in that approach.

RL: You certainly have the option of retracting if it were a problem.

LK: We're doing more and more CTAs at our institution. We're picking up more and more asymptomatic liver metastases. After we do that, we go back and we get a conventional imaging study. In a lot of cases, it's MR, and we do it according to our standard protocols. We're not finding those metastases. We are now using the second-generation multidetector CTs with very high doses, and very rapid administrations of contrast. I suspect that you're correct, in that we may have to look at higher doses of gadolinium at higher rates of injection. But right now we don't have a lot of data on that. If we decide today to double the dose of contrast, are we doing research or is it practical non-research to go ahead and do right now?

RS: My approach is that you really have to show that there is a value. It is probably necessary to identify initial studies involving patients who have lesions already demonstrated on single dose, and then administer a double dose and compare the images. I think a perfect patient population for that are those with hepatocellular carcinoma, because it is quite often multifocal. I think that's a great way to evaluate initially, to see if it really is doing empirically what we think it should do, which is improving the demonstration of lesions.

I think, though, that the comparison between CT and MR always enters everything. When you talk about doing CTA, of course you could also do MRA for evaluating hepatic injection on MR rather than on CT. I think one of the compelling things that Martin said is that MR is fundamentally a much safer modality, that will always be the case, and the contrast agent is much safer.

We haven't compared it with CTA. But we've compared MR to CTAP. In fact, we stopped doing CTAP mainly because of the higher specificity of MR using the kind of protocol we're using. But I think you're right, insofar as if you are going to be looking at absolute detection of every lesion, you have to do something different. So either go with the interarterial injection on MR or start with higher doses and see if that makes a difference. I always prefer being noninvasive or less invasive to start with. It may be a good starting point to compare CTA with high-dose gadolinium, and look at patients with liver metastases who are being considered for surgical resection.