CT: The key to the evaluation and management of hepatic malignancy
Dr. Federle is a Professor of Radiology and Chief of the Abdominal Imaging Division, University of Pittsburgh Medical Center, Pittsburgh, PA. Multidetector computed tomography (CT) plays a fundamental role in the evaluation of hepatic malignancies, whether primary tumors or metastatic disease. In a single study, conventional CT provides anatomic and morphologic information on the number, size, and distribution of liver tumors, while CT angiography (CTA) yields excellent depictions of hepatic arterial anatomy. Together, these imaging techniques accurately and noninvasively guide diagnosis, staging, and therapeutic decision making. Hepatic malignancies Hepatocellular carcinoma (HCC), once considered a rare condition in the United States, is now more common, due in large part to an increase in the number of patients with chronic hepatitis C. In fact, at the University of Pittsburgh Medical Center (UPMC), we have for the past 15 years seen more than 250 new cases of HCC annually. Fewer than 10% of patients with HCC can be treated with conventional hepatic resection, given their underlying chronic liver disease and the possibility of postsurgical hepatic insufficiency. Percutaneous ablation by ethanol injection and radiofrequency ablation (RFA) have shown promise, though large, multiple, or recurrent tumors are difficult to treat.1 Hepatic arterial chemoembolization may offer the best option for patients with unresectable, untransplantable, nonmetastatic HCC. Colorectal carcinoma, by comparison, is very familiar to all practicing radiologists. The second leading cause of cancer mortality in both men and women, it is responsible for more than 50,000 deaths in the United States each year.2 Colorectal cancer is largely preventable by proper screening and the removal of precancerous polyps; however, less than half of the U.S. population undergoes screening. Hepatic metastases develop in 50% to 60% of the 100,000 new cases of colorectal cancer in the United States each year and are the leading cause of mortality associated with this disease.3 Conversely, of the hepatic malignancies encountered in most radiology practices, metastases from colorectal carcinoma are the most common. Only 25% to 30% of patients with liver metastatic disease can be treated with surgical resection or RFA, the accepted forms of therapy.4-6 Even after surgical resection, hepatic metastases recur in 37% to 44% of patients with colorectal cancer, usually within 2 years.4-7 Primary tumor recurrence is more common with rectal cancer than with primary colonic cancer, and can be difficult to distinguish from postoperative fibrosis.8,9 The role of cross-sectional imaging in patients with cancer includes detection of the primary tumor; staging of the tumor; characterization of the tumor, particularly the differentiation of metastatic lesions from benign hepatic masses; follow-up after therapy to gauge tumor response and monitor for recurrence; guidance of percutaneous biopsy or ablation; and guidance of angiographic therapy, such as intra-arterial chemotherapy. As will be discussed in more detail later, the combination of positron emission tomography (PET) and CT is a major advance in the diagnosis and management of patients with colorectal cancer. The measurement of serum carcinoembryonic antigen (CEA) levels provides complementary information as well. Radiofrequency ablation One of the most exciting forms of treatment for primary and metastatic hepatic tumors is RFA. Figure 1 shows a patient with colon cancer that has metastasized to the liver. Before the procedure, the patient had an elevated serum CEA level of 7.0 ng/mL. Six months after intraoperative RFA, the CEAlevel was normal (1 ng/mL). Twelve months after surgery, the ablated lesion had shrunk and the CEA level, at 0.5 ng/mL, was almost undetectable. At 2 years, the CEA level remained stable at this very low level, and the liver showed no new metastases.
Radiofrequency ablation can be performed percutaneously but often is done in the operating room in partnership with a surgeon. This is particularly true in the case of superficial lesions, which are very difficult to manage percutaneously. The patient can experience a great deal of pain with the percutaneous approach, and it can be difficult to visualize superficial lesions and treat them without damaging surrounding tissues. Figure 2 shows a patient with esophageal adenocarcinoma that has metastasized to the liver. Awedge resection of the superficial lesion was performed surgically, and RFA was used to treat the deeper lesion. More than 2 years after initial detection and treatment of the liver metastases, there is no detectible residual or recurrent tumor.
At the University of Pittsburgh Medical Center, we have followed-up patients treated with RFA for hepatic tumors, many for as long as 5 years. Our first 68 patients included 15 with HCC, 39 with metastatic colorectal carcinoma, and 14 with other forms of metastatic cancer. To date, survival is 90%. There have been 3 deaths unrelated to cancer, and 4 cancer-related deaths. We have observed only a 5% local tumor recurrence rate. Distant liver and extrahepatic malignancies occurred in 22% of patients (unpublished data). In selecting patients for RFA of hepatic metastatic tumors, it is important to ensure that there are no extrahepatic lesions. Consider, for example, a patient with colorectal cancer who has both liver and pulmonary metastases. Resection of the primary tumor is virtually always indicated, but RFA, while effective at controlling the liver metastases, is unlikely to have an important influence on the patient’s ultimate clinical course. PET/CT The combined use of PET and CT is arguably the most important advance in oncologic imaging since the introduction of CT scanning itself.10,11 The hybrid PET/CT scanner looks like an oversized CT scanner but is, in fact, half CT scanner and half PET scanner. The patient moves through the scanner once during the acquisition of CT images, and again during the acquisition of PET images. Positron emission tomography has a sensitivity of nearly 100% for detecting primary colorectal cancer, although it is seldom used for that purpose. False-positive studies can occur, however, in inflammatory bowel disease and as a result of normal physiologic uptake of fluorodeoxyglucose (FDG).12,13 False-negative studies can sometimes occur with mucinous and well-differentiated carcinomas. Although PET has a sensitivity of less than 50% for the detection of metastases to the lymph nodes, it has a sensitivity of almost 90% for the detection of liver metastases.14-19 One of the most important roles for PET today is in the detection of recurrent or residual tumor in patients who have undergone resection of colorectal cancer. Conventional CT alone does not reliably accomplish this goal. Among patients judged by CT to have limited recurrent disease, only 25% are typically found to be curable at surgery. And among patients who are considered to have resectable hepatic metastases, only 50% are, in fact, suitable for surgical resection.14,16,18,20 By comparison, PET, particularly when combined with CT for precise anatomic localization, has an extraordinarily high sensitivity and specificity for recurrent colorectal carcinoma (90% to 100%, and 70% and 80%, respectively).21-24 Figure 3 shows a 70-year-old man with colorectal cancer who has undergone surgical resection, chemotherapy, and radiation therapy. Rising serum CEA levels raised suspicion for tumor recurrence, but the CT scan showed only a questionable pelvic soft-tissue density. A PET-only coronal image revealed an area of increased uptake in the pelvis, but it was difficult to localize the area or to determine its significance. The fused PET/CT image showed an intense focus of FDG uptake, clearly indicating an area of recurrent tumor. Avid FDG uptake on PET/CT can also distinguish between recurrent hepatic metastases and liver lesions associated with prior RFA (Figure 4).
The PET findings are not always definitive, however. To accurately diagnose the recurrence of colorectal carcinoma, it is important to take into account the serum CEA level; anatomical information provided by CT; and the potential for false-positive or false-negative PET studies. Moderately differentiated carcinoma comprises the majority of recurrent colorectal tumors and is usually associated with an elevated serum CEA level and a positive PET scan. A well-differentiated carcinoma, however, may produce an elevated or normal CEA level, and a positive PET scan. A poorly differentiated carcinoma may be associated with an increased or normal CEA, and a positive or negative PET scan. A mucinous carcinoma will often produce an increased CEA, but a negative PET scan. CTA Some patients who are not candidates for surgical resection or RFA might be candidates for chemotherapy. Systemic chemotherapy is often ineffective for liver metastases from colorectal cancer, however. Hepatic arterial infusion is a promising alternative that enables the delivery of maximum-dose chemotherapy to the liver tumors while shielding normal liver tissue and the rest of the body from its deleterious effects. Improved response rates and a survival benefit with this approach have been documented when compared with systemic chemotherapy.25 Moreover, intraarterial chemotherapy can be used alone or as an adjunct to surgical resection or RFA. The hepatic arterial infusion pump is inserted in the subcutaneous tissues. An attached catheter is tunneled through the rectus muscle, and its tip is inserted into the gastroduodenal artery. Before pump implantation, the surgeon needs an accurate assessment of the tumor burden and distribution (right or left lobe), and the hepatic arterial anatomy (the vessels supplying the tumor and their accessibility for catheterization or surgical pump placement). Multidetector CT can answer all of these questions accurately and noninvasively in a single study. Conventional CT provides anatomic and morphologic information on the number, size, and distribution of liver tumors, while CTA provides an excellent depiction of arterial anatomy and can replace conventional catheter angiography.26-29 Hepatic CTA can be used to characterize a variety of hepatic malignancies, both metastatic colorectal cancers and hypervascular tumors. The latter category comprises primary hepatocellular carcinoma and metastatic neuroendocrine tumors such as islet cell and carcinoid tumors. In general, hypervascular tumors respond very well to intra-arterial chemotherapy, whether with conventional agents such as cisplatin, or with experimental radioactive microspheres such as yttrium-90.30-33 Colorectal metastases generally respond to conventional oncologic agents such as floxuridine. Whereas chemotherapeutic agents are typically delivered episodically for HCC and neuroendocrine metastases, continuous intra-arte-rial infusions are more typical for metastatic colorectal carcinoma. The hepatic arterial distribution has numerous variations. In the conventional configuration, the common hepatic artery arises from the celiac axis and gives rise first to the gastroduodenal artery and then to the right and left hepatic arteries. However, it is very common for a replaced right hepatic artery to arise from the superior mesenteric artery, anatomy that may in some cases preclude hepatic arterial chemotherapy. The gastroduodenal artery may also arise from the right hepatic artery or the left hepatic artery. In such cases, the chemotherapy infusion would supply only one lobe of the liver. Figure 5 shows a small left hepatic artery arising from the common hepatic artery but proximal to the takeoff of the gastroduodenal artery. A tumor in the left lobe of the liver would not be perfused by chemotherapy from a pump placed in the gastroduodenal artery. As shown in Figure 6, certain anatomic variations preclude hepatic arterial pump placement. In this case, the left hepatic artery arises from the left gastric artery, and the right hepatic artery arises from the superior mesenteric artery.
A study by Kapoor et al26 evaluated the used of multidetector CTA with volumetric 3-dimensional rendering in the assessment of patients with metastatic colorectal cancer, prior to placement of a floxuridine infusion pump. Twenty-six patients had conventional angiography and/or surgical correlation of CTAfindings. Of these, 64% were found to have conventional anatomy and were judged to be good candidates for the intra-arterial pump. The others had variant anatomy, and 4 patients were excluded from pump placement by CTA. In a small percentage of cases, CTA provided inaccurate information. Figure 7 shows the case of a 68-year-old man who developed abdominal pain 14 weeks after initiation of chemotherapy by hepaticarterial pump infusion. CT angiography suggested good placement of the infusion catheter in the gastroduodenal artery. Conventional angiography, however, showed a small branch of the gastroduodenal artery to the duodenum that was not evident on CTA. In this patient, infusion of chemotherapy caused toxicity to the duodenum, and embolization of the gastroduodenal arterial branch was necessary.
CT protocol The CT protocol we use for the evaluation of patients for possible hepatic intra-arterial chemotherapy depends on the scanner that is being used. With the 4-detector-row scanner, we select a collimation of 1.25 mm, reconstruction interval of 0.5 mm, a table speed of 15 mm/sec, and a kV of 120. The mA is fixed at 300 or 350 or is determined by automatic exposure control. We use contrast media with a concentration of 350 or 370 mgI/mL, 120 mL delivered at 4 mL/sec. We acquire early arterial-phase images 18 to 20 seconds after contrast injection, “late” hepatic-arterial-phase images 35 seconds after contrast injection, and portal-venous-phase images at 60 seconds. With a 16-detector-row scanner, we use a collimation of 0.625 mm, a table speed of 13.75 mm/sec, a reconstruction interval of 0.65 mm, an mA of 670 and a kV of 140. Contrast medium concentration, volume, and injection rate are the same as for the 4-detector scanner. Conclusion We now have a variety of tools to treat hepatic malignancies, including surgical resection, radiofrequency ablation, and intra-arterial chemotherapy. Multidetector CT is the most fundamental tool in the diagnosis, staging, and determination of therapeutic options. New developments, including CTA and PET/CT, are essential for maximizing the value of these new therapies. REFERENCES
Discussion ELLIOT K. FISHMAN, MD: Mike, what are your protocols for doing these studies—contrast volumes, injection rates for, for example, both the hepatic arterial studies and the colon? Also, when do you use dual-phase acquisitions? MICHAEL P. FEDERLE, MD: It depends partly on the nature of the primary tumor. But I can give you that in some detail. It also is an evolving situation. One thing I hope we’ll talk about today is some of the newer contrast injectors. The type of injectors we are using right now facilitates the use of a preloaded syringe. We tend to use 125 mL of contrast for these studies, outside of the ones for which we are exclusively doing CT angiography, such as a pulmonary embolism study. For colorectal cancer, ordinarily we would not do arterial-phase imaging. But if we are working with a patient for whom intra-arterial chemotherapy is being considered, we will do an arterial-phase set of images in addition to the more conventional portal-venous set of images. So the arterial-phase images in this setting are true arterial, rather than the kind we would use for diagnosing hypervascular tumors. To do our timing, we would use a test bolus of contrast to time the maximum arterial arrival. That is usually approximately 18 to 20 seconds in most adults with good circulation time. After the arterial set of images, we would go back and do a portal-venous set of images. That timing is a little bit dependent on your injection rate and circulation time, and so forth. We generally do our injection rates at 4 or 5 mL per second. Hepatocellular carcinoma (HCC) is the model for primary hypervascular tumors, and metastatic neuroendocrine tumors are the most common secondary hypervasculars that we treat. For primary and metastatic hypervascular tumors, we would do a modification to the protocol, because we want to use the arterial-phase images both to detect and accurately stage the distribution of the hepatic tumors. That’s better achieved with a scanning delay of approximately 35 seconds. Then we would combine that with a follow-up portal-venous set of images as well. To reiterate, for planning for intra-arterial pump therapy for colorectal cancers, we generally do a true arterial-phase image, followed by a por-tal-venous phase. For hypervascular tumors, we generally do what has often been called a second hepatic arterial-phase or a portal-venous inflow phase image at about a 35-second delay, and then followed by the portal-venous or hepatic-parenchymal phase. ALEC J. MEGIBOW, MD, MPH, FACR: So Mike, what’s the level of clinician acceptance of the anatomy you display on the CTA studies? FEDERLE: Good question. The level of acceptance is actually very good. But in our hospital, we have clearly seen that the acceptance is dependent on giving them routinely excellent-quality images. Frankly, with the 4-detector scanner, we didn’t always give them great images every time. Some of those patients would then go on to conventional catheter angiography. With the 16-slice scanner, we can give them great CT angiograms in every case, and virtually none of the patients have to go on to conventional angiography. JULIA R. FIELDING, MD: I’m sure this comes up in everyone else’s hospital as well: the issue of IV placement and type of IV that each department allows for especially reactive rates of infusion. Central venous catheters in intensive care unit patients for PE studies are a separate thing, but perhaps when we are doing an outpatient or someone with not particularly good veins, I like to use a 20-gauge antecubital line or better. That’s routine in our hospital. But what do you do when the access is a problem? FEDERLE: Excellent question. I hope that we’ll see more discussion of this and more development of this issue. I’ve been badgering catheter manufacturers for years to develop and stand behind catheters that are meant for rapid bolus infusion. In fact, even though we all use them, there are virtually none on the market manufacturers will stand behind. We do the same thing. We place a fresh antecubital 20-gauge catheter for most of these patients, or 18-gauge if it looks as though the vein will easily accommodate it. But you can get warmed contrast material—at 4 mL per second—just fine through a 20-gauge catheter. Just within the last several months, I have been working with a manufacturer in evaluating a PICC (a peripherally inserted central catheter) line. We all know that many of our hospitalized patients have PICC lines, but none of the conventional PICC lines are suitable for rapid bolus injection. Well, a manufacturer has come out with a catheter that’s only slightly larger in external caliber than the standard PICC lines, and it’s warranted for rapid bolus power infusion. We now insert that whenever we have a question about the suitability of venous access in a patient, particularly when it’s very important to get a good-quality CT angiogram. We have had very good acceptance among our clinicians with this. We have used certain central catheters for power infusions for years, and will continue to use them. For instance, there is a relatively short—roughly 10 inches long—central venous catheter that has 2 or 3 lumens, all of which are at least 18gauge in size. We have used those for infusion. We have not done power infusions through the infusaports or other catheters that are designed primarily for venous access in patients for chemotherapy and so forth. I hope that gives at least an initial answer to that question. I do hope that we’ll have more discussion about this important topic, because that’s often a rate-limiting issue with these patients. FISHMAN: Mike, I have a question as well. Going beyond the anatomy, in terms of CT angiography, do you think that doing vascular maps and looking at specific patterns of lesions will be helpful for lesion discrimination? I know that you wrote an article on focal nodular hyperplasia (FNH) looking at that. How do you use that in practice? FEDERLE: I think we’re on just the threshold of that, Elliot. I do believe, as you’ve alluded to, that there are certain patterns of vascularity that are more common and perhaps indicative of either benign or malignant lesions. But I don’t think we know enough about that, because coming from a world of axial imaging, you often don’t appreciate the true course of vessels, particularly veins draining hepatic lesions. Even though we are all very experienced at turning 2-dimensional images into 3D pictures in our head, I think we’ve all experienced the phenomenon of really understanding more about the true morphology— the 3D morphology—of tumors, than we have in the past. So that remains to be explored. But I suspect there will be an understanding of certain morphologic characteristics that help us distinguish among various types of hepatic tumors. CHRISTOPH R. BECKER, MD: You mentioned the role of positron emission tomography (PET)/computed tomography (CT) for liver tumors, in particular for recurrent disease. Do you have any idea if PET/CT angiography may be of any value? FEDERLE: PET/CT angiography—that is a phrase I haven’t even imagined in my mind. Well, the easy answer is that I don’t know. PET/CT is evolving fairly rapidly. We’re now on our third generation of PET/CT scanners at the University of Pittsburgh. The initial scanners, both the PET part and the CT part, were quite slow; you couldn’t even do breath-held imaging on the initial scanners. The newer generation PET scanners have different types of gamma detectors, that are faster and they are being coupled with faster CT scanners, the 16-slice generation of CT scanners. So the easy answer is that I have no experience with PET/CTA and I don’t yet know whether that will be important. FISHMAN: Mike, you mentioned PET scanning. At this point, most sites do not give IV contrast, but I know that you have. What type of injection protocols do you use for doing the PET scan with IV contrast? FEDERLE: We do virtually all of our PET and PET/CT scans with both oral and IV contrast. The IV contrast is generally 125 mL of nonionic contrast, at about a 60% or 350 mgI/mL. It’s administered at 3 mL per second. The delay depends partly on which generation of PET/CT scanner we’re working with, in terms of how fast the CT scanner is. Generally, if it were a 16-slice scanner, you would probably use a scan delay of 80 seconds for the CT part, and maybe 70 seconds if you were using a 4-slice CT scanner as part of the PET/CT. FISHMAN: Who reads the PET/CTs in your institution? FEDERLE: You really want to get into the controversial things, don’t you? We do something that is quite inefficient, but I think it’s excellent, optimal patient care: we have trained nuclear medicine people looking at the PET part of it, and we (in my section, the abdominal images), render an interpretation of the CT part of it. On our monitors, we can look at both the PET part and the CT part. So we never send a report out of the department without consulting with each other. Trained body imagers read the CT part of it and nuclear medicine people read the PET part of it. FISHMAN: Then who puts the two pieces together? FEDERLE: We consult with each other so that we do not send out a report that is contradictory. So, for instance, I showed a slide of a patient with mucinous metastases from colorectal cancer. I think it would be a disservice to the clinicians and the patient if they sent out a report saying that the PET scan is normal. What they really said in that case was, “We don’t see any areas of FDG avidity. (However, the CT scan clearly shows a metastatic lesion within the liver.)” So that way we don’t look like we’re interpreting two different patients’ studies. FIELDING: I have a question. In my hospital, patients often undergo a very high-resolution CT prior to a partial hepatectomy. But the PET/CT would not be adequate for that. So we end up doing an extra CT that we don’t charge, which is the one that’s associated with the PET. We do charge for the one that we do the high resolution on. FEDERLE: Yes, we won’t resolve that issue this week, unfortunately, because it varies from state to state, payer to payer, etc. Reimbursement is a very big issue. I am on committee with the American College of Radiology that includes a number of us from the body imaging world and a number of people from the nuclear medicine world. We are attempting to come up with guidelines or training for PET/CT interpretation and also guidelines for interaction with the payers to come up with a policy that makes some sense. So it’s clearly a work in progress. Thank you. |
