Evidence-Based Imaging - part 10 pps

Goals of Imaging In patients with colorectal cancer imaging studies are acquired periodically in order to detect development of recurrent disease and to assess tumor burden and response to therapy. In the cirrhotic patient, the main goal of imaging is detection of developing complications, the most important of which is HCC. Many imaging modalities currently available have been used for detecting liver metastases, with variable success. Regardless of the technique used, the ability to detect a focal space-occupying lesion in the liver depends on the size of the tumor, the spatial and contrast resolution of the imaging method, the difference in contrast and perfusion between the tumor and background liver parenchyma, and the adequacy of the method used for displaying the images after acquired (10). All these factors affect the performance parameters of the various imaging techniques. A test is useful if sensitivity remains high at an acceptable specificity level. In a meta-analysis that studied the detection rate of liver metastases from gastrointestinal malignancies with multiple modalities, Kinkel et al. (3) suggest that, in order to be useful in clinical practice, the minimum accept- able specificity of imaging methods in this context should be 85%. Lower specificities would lead to excessive and unnecessary interventions such as biopsies, excessive complementary imaging tests, and follow-up exam- inations. When assessing cost-effectiveness of the imaging methods, other factors need to be considered: availability, cost, risks (such as radiation and use of toxic contrast agents), and potential benefit of tumor detection (i.e., likelihood of achieving long-term remission or cure with appropriate therapy). Overall Cost to Society On an individual level, cirrhosis results in impaired quality of life and indi- rect costs involving decreased productivity and lost days from work. The Centers for Disease Control and Prevention conservatively estimates U.S. expenditures in excess of $600 million annually on patients with HCC. In 2002, in the U.S., a total of 15,654 patients were discharged from hospitals with the diagnosis of HCC and 2522 patients died in the hospital with HCC. The mean length of hospital stay was 7.2 days with a mean cost of $32,193. This resulted in a total cost of $501,998,078. I. How Accurate Is Imaging in Patients with Suspected Hepatic Metastatic Disease? Summary of Evidence: Computed tomography (CT) and magnetic reson- ance imaging (MRI) are the most widely used techniques for evaluating the liver in the initial staging and follow-up of cancer patients. For detect- ing liver metastases, carefully performed CT and MRI studies with state- of-the-art equipment and interpretation by experienced radiologists afford similarly good results. Some studies showed a slight advantage for MRI (11,12) (moderate evidence). Others, including a multiinstitutional 524 B.C. Lucey et al. Chapter 28 Hepatic Disorders 525 study of 365 patients (13) (moderate evidence), have not. Computed tomography is usually preferred because it is more widely available and because it is a well-established technique for surveying the extrahepatic abdominal organs and tissues (such as the peritoneum and lymph nodes). However, MRI has an advantage in the characterization of focal lesions. Thus, MRI is commonly used as a problem-solving tool or for initial staging of a tumor. It is also preferred for patients who cannot receive intra- venous iodinated contrast material. Finally, concerns about the risk of radi- ation from repeated exposure to CT examinations make MRI a valuable alternative for children or young adults with malignancies. As mentioned previously, a comparison of the performance of CT vs. MRI for this and other indications needs to be reassessed periodically, considering the rapid evolution of both technologies and the increase in therapeutic options available. Kinkel et al. (3) reviewed a total of 111 studies that included over 3000 patients. At a specificity of at least 85%, the weighted sensitivities were ultrasonography (US) 55%, CT 72%, MRI 76%, and positron emission tomography (PET) 90% (moderate evidence). These data, however, need to be validated in prospective trials before broad conclusions can be drawn. Intraoperative ultrasonography (IOUS) has higher sensitivity than trans- abdominal ultrasonography, CT, and MRI (14,15). The role of FDG-PET and PET-CT will continue to expand, but cost constraints will limit their use to patients in whom the possible impact is greatest. Supporting Evidence: The most widely used imaging techniques today include US, CT, MRI, and, more recently, PET. There is extensive literature available regarding the relative merits and limitations of each of these modalities for detecting metastases of primary tumors from various organs. When analyzing the multiple studies published on this topic, several limitations are evident: insufficient definition of inclusion and exclusion criteria, incomplete reporting of methods used, and lack of a uniform standard of reference. Although the best standard of reference available is findings at laparotomy with bimanual palpation or intraoper- ative ultrasonography, this was used as the gold standard in only a minor- ity of studies (14,16,17). As indicated by van Erkel et al. (18), use of a suboptimal standard of reference results in underreporting of lesions and overestimation of detection rate. Another confounding factor is the varying method for reporting sensitivity numbers: per patient (detection of at least one lesion per patient) and per lesion (detection of all lesions per patient). Thus, it is important to continually scrutinize the results of all available current studies as evolving and improving technology can make results of prior studies redundant. Following is a review of the available data regard- ing the benefits and limitations of the various imaging techniques com- monly used for evaluating the liver in patients with colorectal cancer and other gastrointestinal primary malignancies. A. Ultrasonography Ultrasonography has the advantage of being widely available throughout the world, inexpensive, and essentially risk-free. The reported sensitivity of US for detecting liver metastases varies between 60% and 90% (3). 526 B.C. Lucey et al. Unfortunately, many of the published studies were performed in the 1980s (19,20) (limited evidence) and were largely limited to reporting sensitivity results on a per patient basis. More recently, the advent of US contrast agents has led several investigators to evaluate the use of US with current equipment. For detecting liver metastases, the sensitivity and specificity of US improve substantially with the addition of microbubble contrast agents. Microbubbles are essentially blood pool agents that augment the Doppler and harmonic US signal. In addition, some of these agents have a hepatosplenic specific late phase, which enables visualization of tumor foci in the liver that were otherwise undetectable (21). In a multicenter study, Albrecht et al. (22) found that the addition of a microbubble contrast agent increased the per patient sensitivity of US from 94% to 98% (not signifi- cant), while the per lesion sensitivity increased from 71% to 87% (highly significant, p < .05). Intraoperative ultrasonography has higher sensitivity than transabdom- inal US, CT, and MRI (14,15). Conlon et al. (14) compared MRI with IOUS in 80 patients with colorectal cancer metastases who underwent hepatic resection and found that IOUS findings added important information in 37 patients and changed the surgical approach in 14 patients. They con- cluded that IOUS provides valuable information prior to hepatic resection of colorectal cancer metastases. B. Computed Tomography Multiple factors pertaining to technique need to be considered when planning CT scans of patients with suspected metastatic disease and when examining reports that deal with this topic. The typical colorectal cancer metastasis is hypoattenuating and hypovascular relative to liver parenchyma. Thus, detectability is maximized by administering intra- venous contrast material and by acquiring the CT images during the time of peak enhancement of the liver parenchyma. This typically occurs during the portal venous dominant phase, which occurs approximately 60 to 80 seconds after the initiation of contrast injection. Ideally, hepatic parenchyma attenuation should increase by at least 50 Hounsfield units after the administration of intravenous contrast material. The addition of images acquired prior to the administration of intravenous contrast mate- rial or in the arterial-dominant or delayed phases of contrast enhancement are not routinely necessary when the indication for the scan is suspected hypovascular metastases. These are necessary when evaluating the cir- rhotic liver, when attempting to characterize a focal lesion, or when the primary tumor is one that is known to be associated with hypervascular metastases, such as neuroendocrine and carcinoid tumors, thyroid cancer, melanoma, breast cancer, or renal cell carcinoma (Fig. 28.1). Although specific protocols vary among institutions, most use a total load of 37 to 50g of iodine (23). Although as little as 30g have been used, detection of hypovascular focal lesions may be limited with this approach (24). In the patient with colorectal cancer who is being scanned to decide among the several therapeutic options available, the risk of overlooking a potentially resectable small liver metastasis needs to be outweighed vs. the benefit of limiting the amount of contrast material injected. In a carefully performed study, Valls et al. (25) used contrast-enhanced helical CT to detect liver metastases in 157 patients with colorectal Chapter 28 Hepatic Disorders 527 Figure 28.1. Importance of adequate technique for detecting computed tomogra- phy (CT) of metastatic disease to the liver. Noncontrast (A), arterial phase (B), and portal venous phase (C) CT images of a 57–year-old patient with breast cancer and abnormal results of liver function tests. There are multiple foci of hypervascular metastatic deposits seen exclusively in the arterial phase image (B). The appearance of the liver is near normal on the noncontrast (A) and portal venous phase (C) images. A B C carcinoma. Using intraoperative palpation and US as the standard of ref- erence, helical CT correctly depicted 247 (85.2%) of 290 metastases and had a 96.1% positive predictive value (moderate evidence). Surgical resection of the liver metastases was attempted in 112 patients and the authors achieved a 4-year survival rate of 58.6%. In their study, all false-negative interpretations occurred in lesions less than 1.5cm in diameter. Other studies that also used surgical findings and IOUS as the standard of refer- ence found similar high sensitivity and specificity (16), for detecting lesions as small as 4mm in diameter. Although with the multirow detector helical CT (MDCT) scanners that are now available it is possible to acquire CT images in multiple phases after administration of intravenous contrast material, it has not been con- vincingly demonstrated that detection of hypovascular metastases such as those from colorectal carcinoma is improved significantly by scanning in any phase other than the peak portal venous phase (16,26,27). The advent of MDCT has also brought about new paradigms related to CT technique. Although scanning with slice thickness of less than 1mm and often with isotropic voxels is tempting, there is debate as to what is the limit in thickness that achieves the performance that is adequate for demon- strating small metastatic lesions in clinical practice. Some studies have shown that scanning with a slice thickness of less than 5mm does not result in a significant improvement in sensitivity for detecting small lesions (28). Other investigators have obtained better results using thinner collimation (29). However, detection of even small lesions in the patient with cancer is important, since approximately 12% of lesions less than 1cm in diameter will prove to be metastatic in nature (30). The possible added benefit of images acquired with isotropic voxels remains to be deter- mined and will undoubtedly be the focus of multiple studies in the near future. Another CT technique that continues to be used at some institutions is CT during arterial portography (CTAP). This is an invasive technique that requires catheterization of the superior mesenteric or splenic artery for direct injection of contrast into the territory drained by the portal vein. This direct delivery of contrast into the porto-mesenteric circulation achieves the greatest degree of hepatic parenchymal enhancement and maximizes lesion detection with CT, with a sensitivity that exceeds 90% (17,31). The technique, however, is invasive and has a false-positive rate as high as 25% (17,31). This has led to decreased enthusiasm for this technique and its replacement with noninvasive CT and MRI methods using state-of-the-art equipment (32,33). C. Magnetic Resonance Imaging Magnetic resonance imaging of the liver for detecting metastases requires the acquisition of multiple sequences and administration of intravenous contrast material. Although the appearance of metastatic lesions on MRI is variable, the T1 and T2 relaxation times of metastases are prolonged rela- tive to normal liver parenchyma. In general, this results in hypointensity on T1-weighted sequences and hyperintensity on T2-weighted images (Fig. 28.2). T2-weighted MRI is also useful for characterizing focal lesions and differentiating nonsolid benign lesions such as cysts and hemangiomas 528 B.C. Lucey et al. Chapter 28 Hepatic Disorders 529 Figure 28.2. Typical appearance of hepatic metastasis on magnetic resonance imaging (MRI). T1-weighted (A), T2-weighted (B), and late arterial phase (C) MRI acquired in a patient with known colon cancer demonstrate a large metastatic deposit in the right hepatic lobe. The lesion is hypointense (relative to liver parenchyma) on the T1-weighted image, slightly hyperintense on the T2-weighted image, and demonstrates moderate enhancement after administration of gadolinium-DTPA. A B C from metastases. In multiecho T2-weighted scans, metastases become less intense when the echo time (TE) is increased from <120msec to 160msec or more. Conversely, cysts and hemangiomas typically remain hyperin- tense as the TE increases. For lesions with equivocal behavior, MRI can be used to measure the T2 value; the T2 of malignant tumors is approximately 90msec, while that of hemangiomas and cysts exceeds 130msec (34,35). However, metastases with liquefactive necrosis or cystic neoplasms may remain hyperintense on heavily T2-weighted images. Metastases can have a perilesional halo of high signal, indicating viable tumor, or demonstrate a doughnut or target appearance (36,37). For detection of liver metastases, a three-phase technique after adminis- tration of gadolinium is recommended; these phases are the arterial-dom- inant phase, the portal venous phase, and the hepatic venous or interstitial phase. Similar to CT, the detection of colorectal cancer metastases using MRI is maximized during the portal venous phase. In this phase, the lesions typically appear hypointense relative to the enhanced liver parenchyma and may exhibit variable degrees of enhancement (Fig. 28.2). In addition to lesion detection, this protocol also allows characterization of coexisting nonmetastatic focal lesions. This is important for staging recently detected malignant tumors, and has implications in determining the type of therapy to be offered. The reported sensitivity of MRI using multiple combinations of the sequences available varies between 65% and 95% (3,33,38–41), with a mean of approximately 76% (3) (moderate evidence). The administration of organ-specific contrast agents increases the lesion- to-liver contrast-to-noise ratio (CNR), thereby improving the conspicuity and detection rate of metastatic lesions. These include hepatobiliary agents such as mangafodipir trisodium (MnDPDP) (40) and gadobenate dimeg- lumine (Gd-BOPTA), and reticuloendothelial agents such as superpara- magnetic iron oxide (SPIO) particles (41). The available data regarding the need for these liver-specific agents is controversial, with some studies showing improved results (17,42) while others do not (3,43,44). In addition to a lack of consensus regarding the benefits associated with their use, these agents are generally considered costly and not widely available. Thus, a broad use of liver-specific contrast material for detecting liver metastases is not recommended at this time. D. Whole-Body Positron Emission Tomography Whole body PET performed with fluorine-18-fluorodeoxyglucose (18F- FDG) has also been used successfully for detecting extracolonic spread of colorectal carcinoma, including liver metastases. Although published studies have included small groups of patients, early results are encour- aging, with sensitivity and specificity exceeding 80% (45,46). Kinkel et al. (3) performed a meta-analysis study comparing the data available for detection of liver metastases from gastrointestinal tract neoplasms with noninvasive tests: US, CT, MRI, and PET. They reviewed a total of 111 studies that included over 3000 patients. At a specificity of at least 85%, the weighted sensitivities were US 55%, CT 72%, MRI 76%, and PET 90%. The strength of these data is moderate and they need to be validated in ran- domized trials before broad conclusions can be drawn. 530 B.C. Lucey et al. Chapter 28 Hepatic Disorders 531 II. What Is the Accuracy of Imaging in Patients with Cirrhosis for the Detection of Hepatocellular Carcinoma? Summary of Evidence: Screening for HCC in patients with cirrhosis is not easy. No one imaging modality dominates over the others. All imaging modalities have advantages and disadvantages with no one modality offer- ing both high sensitivity and specificity. The results of these individual studies often depend on the date of the study. This is primarily because of the rapid change in technology available in all imaging modalities. A reasonable consensus for screening includes biannual measurement of the AFP level. Annual sonography is the imaging modality most commonly used, as it is cheap, portable, and most widely available. If the AFP value increases and the sonogram does not show evidence of an HCC, either CT or MRI should be performed. Although MRI at present has marginally higher specificity than CT, the recent improvement in CT technology may change this soon (Fig. 28.3). Published sensitivities for MRI range from 48% to 87% (47–50). The CT sen- sitivities for these studies range from 47% to 71% without the use of com- puted tomography hepatic arteriography (CTHA) or CTAP. These reports conclude that MRI is certainly as sensitive and perhaps a little more so than CT. The use of superparamagnetic iron oxide (SPIO) has increased the sen- sitivity of MRI. Figure 28.3. Algorithm for imaging to detect HCC in a patient with cirrhosis. AFP, a-fetoprotein; f/u, follow-up. *6/12 means six months. The sensitivity of sonography for detecting HCC has been reported between 59% and 90% (51–55), with lower sensitivity for smaller lesions (55). Ultrasound may also lead to a high percentage of false-positive studies. Overall, there is little evidence to support the use of PET imaging in the detection of HCC. The value of PET in this patient population lies in detecting distant metastases, and PET may be useful in monitoring the response to treatment. Supporting Evidence A. Ultrasonography The 59% to 90% sensitivity of sonography cited above varies with lesion size, with the sensitivity for detecting lesions 2cm or less approaching 60%, with larger lesions having higher sensitivity (55). The sensitivity for detect- ing HCC also depends on patient selection. Screening a population at risk for developing HCC (i.e., chronic hepatitis carriers) is often performed dif- ferently from screening a population with documented cirrhosis. As a result, lesions missed by sonography in cirrhotic patients may be picked up by CT or AFP measurement, thus masking the false-negative cases that may be attributable to sonography (52). One major difficulty with sonog- raphy in the detection of HCC is the high percentage of false-positive studies. This is particularly difficult in the cirrhotic patient population as the risk of developing HCC is higher and therefore any focal geographic area of heterogeneity is concerning for HCC. This may lead to frequent per- cutaneous biopsy to obtain a definitive diagnosis with the attendant mor- bidity and mortality. Despite the difficulties of sonography, given the widespread availability, portability, and safety of the modality, sonography remains the imaging modality of choice for screening for HCC in cirrhotic patients. The time interval between sonograms remains controversial. There is no consensus as to when to perform repeat imaging; however, authors have suggested that annual or biannual interval imaging with sonography is the most effective approach to detecting HCC. There is great interest in the use of intravenous contrast agents for enhancing the value of sonography to detect and characterize liver lesions. There are many reports describing the value of these agents in patients with HCC (56–59). There is no doubt that these microbubbles demonstrate increased vascularity in HCC when used, increasing the color flow within HCC from 33% to 92% in one study (57); however, there is little published evidence to support the value of these agents in identifying HCC from degenerative nodules in patients with cirrhosis. Increased flow may be detected in other hepatic lesions also and not just in HCC after injection of the microbubbles. One potential use for the microbubbles is in the evalu- ation of patients following RFA. The results for contrast-enhanced sonog- raphy for detecting tumor recurrence post-RFA have been reported to be similar to those for CT (60). B. Computed Tomography Computed tomography has benefited even more than sonography from recent advances in technology. With the move from incremental CT to 532 B.C. Lucey et al. single-detector CT to multidetector CT, the ability to detect HCC in the cir- rhotic liver has improved. This difference in technology is the most impor- tant consideration when attempting to compare the results of studies performed to evaluate CT in the detection of HCC. This improvement allows for thinner slice collimation and improved image quality. Another technical parameter to consider is the use of dual-phase imaging. The liver has a dual blood supply from both the hepatic artery and portal vein. In normal livers, approximately three quarters of the blood supply comes from the portal system. In contrast, HCC depends more on the hepatic artery for blood supply. Therefore, ideally, imaging to detect HCC should include images obtained in the hepatic arterial phase, usually commenc- ing at 30 seconds after contrast administration. With the advent of multi- detector CT, imaging in dual phase became possible and this improved detection of HCC. When examining the reports available for detecting HCC in cirrhotic patients, it is important to differentiate between identifying patients with HCC and identifying lesions that represent HCC. This fact may change the sensitivity of an imaging modality greatly. The effect of this is clearly demonstrated in a study by Peterson et al. (61) evaluating patients pre–liver transplant for HCC, in which CT had a prospective sensitivity to detect patients with HCC of 59%. This fell to 37% when attempting to detect HCC on a lesion-by-lesion basis. Reported sensitivity for detecting HCC by CT varies greatly. Most recent reports yield sensitivities between 68% and 88% (5,62). These reports gen- erally refer to the percentage of patients in whom an HCC is found. Figures for detecting individual lesions are much lower. The value of some of these reports is always in some doubt, however, given the previously described rapid change in CT technology today. In an effort to improve detection of HCC using CT, CTAP is occasionally used. This involves placing a catheter into the splenic or superior mesenteric artery and directly injecting con- trast. Computed tomography hepatic arteriography (CTHA) has also been used, in which a catheter is placed directly into the hepatic artery. These techniques have yielded high sensitivities when used together. Makita et al. (63) found the sensitivity of CTAP alone to be 85.5%, CTHA alone to be 88.1%, and combined to be 95%. Specificity, however, suffers and the com- bined specificity reported by that group was only 54%. Similar findings have been reported by others (64,65) with sensitivities ranging from 82% to 97%, although the high number of false-positive studies with these tech- niques leads most authors to conclude that they have minimal role in the evaluation for HCC in cirrhotic patients, particularly given the relatively invasive nature of the procedures. C. Magnetic Resonance Imaging The MRI sequences used in the evaluation of the cirrhotic liver are the same as those used for the detection of liver metastases. The use of intravenous gadolinium is required in all cases. As with CT, the difficulty with MRI lies in differentiating early HCC from dysplastic nodules. As nodules change from regenerative to dysplastic to malignant, the T1 signal characteristics become more hypointense and the T2 signal characteristics become more hyperintense. As one moves along this spectrum, the primary blood supply of the mass changes from predominantly portal to predominantly hepatic Chapter 28 Hepatic Disorders 533 [...]... screening for HCC more cost-effective As with the studies based purely on detection of HCC, there is little consensus on the most cost-effective imaging modality to use to detect HCC While acknowledging that screening for HCC may not be cost-effective at all, if one is to perform imaging, which modality is most cost-effective is open to debate In a retrospective study, Gambarin-Gelwan et al (83) compared... (mm) Fat suppression T1 gradient-echo axial in and out of phase 200 4.6/2.3 80 7 Breath hold 192 ¥ 256 No Yes T2 dual echo, fast spin-echo 2350 40/140 90 6 256 ¥ 512 Yes No, respiratory triggered Precontrast T1 fat-suppressed gradient-echo 200 4.6 80 7 192 ¥ 256 Yes Yes Dynamic gadolinium 20 cc IV 3.5 1.7 10 2 192 ¥ 256 Yes Yes Precontrast T1 fat-suppressed gradient-echo 200 4.6 80 7 192 ¥ 256 Yes... technetium-99m (Tc-99m)-labeled dimercaptosuccinic acid (DMSA) may be of particular value in girls, for whom urethral obstruction is not an issue or for follow-up of well-documented vesicoureteral reflux Level II, moderate evidence, studies have shown an increase in the incidence (25–45%) of vesicoureteral reflux in siblings afflicted with the disease (29,30) For this reason, siblings under 10 years of... in children, particularly using sonographic and MRI contrast agents Imaging Case Studies A B Figure 29.1 Imaging case study for nephrolithiasis Woman with right flank pain underwent non–contrast-enhanced helical computed tomography (CT) that revealed a solitary right kidney with hydronephrosis (A) and an obstructing ureteral stone at the level of the mid-ureter (B) (arrow) Chapter 29 Imaging of Nephrolithiasis,... Med 1989;18(5):556–559 9 Smith RC, Verga M, McCarthy S, Rosenfield AT AJR 1996;166:97 101 10 Blake SP, McNicholas MM, Raptopoulos V AJR 1998;171(3):717–720 11 Schwartz BF, Schenkman N, Armenakas NA, Stoller ML J Urol 1999; 161(4) :108 5 108 7 12 Fielding JR, Silverman SG, Samuel S, Zou KH, Loughlin KR AJR 1998;171: 105 1 105 3 13 Sheley RC, Semonsen KG, Quinn SF Am J Emerg Med 1999;17(3):279–282 14 Boulay... performance of state-ofthe-art CT, MRI and PET-CT for detecting colorectal cancer metastases is highly desirable at this time 2 Need to develop an imaging modality that will differentiate dysplastic nodules from HCC 3 Need to identify HCC earlier Study design similar to the one shown for colorectal cancer metastases above is recommended—relates to entry 1, above 4 The role of PET and PET-CT in these populations... cost-effectiveness ratio . liver in the initial staging and follow-up of cancer patients. For detect- ing liver metastases, carefully performed CT and MRI studies with state- of-the-art equipment and interpretation by experienced. approximately 76% (3) (moderate evidence). The administration of organ-specific contrast agents increases the lesion- to-liver contrast-to-noise ratio (CNR), thereby improving the conspicuity and detection. gadobenate dimeg- lumine (Gd-BOPTA), and reticuloendothelial agents such as superpara- magnetic iron oxide (SPIO) particles (41). The available data regarding the need for these liver-specific agents