CT is a readily available noninvasive imaging modality with significant impact in PSC. Current multi-detector scanners generate images with high spatial resolution and high temporal resolution. High spatial resolution results from thin slices (~1 mm) and fast acquisition speed. Thin slices increase image sharpness and anatomic detail. Thin slices allow for the postprocessing of data sets using multiplanar reformatting (MPR), maximum intensity projection (MIP), and volume rendering techniques (VRT). These postprocessing algorithms produce nonaxial images displayed in coronal, sagittal, and nonorthogonal projections. VRT images can be rotated and tumbled in contiguous conventional and nonconventional projections for optimal anatomic display. For surgical planning, advanced software and an independent 3-D workstation can be used for lobar and segmental volumetrics and to display the anatomy of the hepatic veins, portal vein, and hepatic artery. High temporal resolution allows for bolus tracking of exogenously administered contrast with segmented time frames of image acquisition used to generate arterial, portal venous, and delayed phases of enhancement. Unfortunately, CT cholangiography with positive-contrast excretion into the bile ducts can no longer be performed. The contrast material used, Cholografin®, is no longer available in the United States.

Analogous to US, CT is often performed in patients with abdominal pain and jaundice. It is not uncommon for CT to be the first test to detect PSC. The CT findings of PSC, especially early in its course, can be subtle. Mildly dilated IHDs have a disconnected “dot-dash” pattern corresponding to end-on and longitudinally oriented distended duct segments separated by intervening soft tissue density strictures [42]. Even small, peripheral IHDs can be conspicuous within the background liver, being filled with low-density bile, which intrinsically increases the otherwise moderate contrast resolution of CT. The fibroinflammatory and fibroobliterative changes of PSC manifest as duct wall thickening, irregularity, and narrowing, with the degree of duct wall enhancement being variable and inconsistent [38]. Mural changes are most apparent at the level of the CHD. By CT, the CHD is large enough to be consistently demonstrated in patients without or with PSC. Low-density fat in the hepatic hilum delineates its outer wall, and low-density bile within the CHD lumen defines its inner wall. The CHD is discernible as a ring-like structure on axial images and is normally of uniform thickness ≤1.5 mm. In PSC, the CHD becomes irregular with wall thickening potentially ≥2.0 mm (Fig. 13.2) [38].

CCA can be suspected or detected by CT. Intrahepatic CCA can present as a mass lesion. Intrahepatic CCAs do have neovascularity. In larger intrahepatic CCAs, macroscopic neovascularity tends to be sparse, stringy, and peripheral. Mass-forming intrahepatic CCAs tend to be dominated by an abundant, central fibrous stroma with scant tumor cellularity. At arterial phase CT, these CCAs tend to have no discernible to mild peripheral enhancement with central iso- to hypodensity. During the portal venous and delayed phases, there can be centripetal enhancement with contrast retention in the extracellular matrix of the central fibrous tissue, which can be subtle [14]. These lesions tend to be rounded, somewhat poorly marginated in non-cirrhotic livers, but pseudoencapsulated in cirrhosis; they can be associated with overlying capsular retraction, adjacent dilated IHDs, and satellite nodules (Fig. 13.3a) [10, 37, 39]. With intrahepatic mass-forming CCAs, vascular encasement is common, but macroscopic thrombus is unusual [10]. The features of intrahepatic CCA can overlap with those of hepatocellular carcinoma (HCC), particularly poorly differentiated HCCs or larger HCCs with central necrosis.

Small intrahepatic CCAs can appear as arterial phase hypervascular nodules [8, 9]. These CCAs tend to accumulate contrast, and enhance during the portal venous and delayed phases of multiphasic imaging. This is compared to typical small HCCs which wash out and become hypodense during the portal venous and delayed phases. However, arterial phase hypervascular CCAs with subsequent washout do occur.

PSC patients with cirrhosis are at an increased risk of HCC, which is estimated to be up to 2 % per year [30]. This is probably related to the association of HCC and cirrhosis. Given the overlap of imaging features, HCC should be considered in PSC patients with cirrhosis.

Of perihilar CCAs, 70 % are of the periductal infiltrating morphologic subtype [14]. These can be difficult to demonstrate by CT and can appear only as a stricture. Although some features such as duct wall thickening >5 mm, stricture length ≥18–22 mm, shouldering, portal venous or delayed phase enhancement, and soft tissue stranding within portal fat planes suggest perihilar periductal infiltrating CCA, these findings are insufficient to reliably differentiate dominant benign strictures from malignant strictures in PSC [14, 38]. Of note, malignant lymphadenopathy is common in cases of perihilar infiltrating CCA [14].

Of perihilar CCAs, 12–22 % are of the mass-forming morphologic subtype [14]. Perihilar masses measuring 1–9 cm can occur with features analogous to intrahepatic mass-forming CCAs. Small lesions can be seen as hypervascular arterial phase nodules. Larger lesions tend to have less pronounced arterial phase rim enhancement and can have portal venous or delayed phase washin and contrast retention within the central fibrous stroma. Portal vein invasion with visible thrombus can be seen.

Distal CCAs are anatomically defined as involving the CBD between the cystic duct origin and the ampulla of Vater [29]. Approximately 89 % are periductal infiltrating, and 11 % are intraductal growing [19]. CBD dilatation is present in 96 % of cases. Imaging findings are usually limited to CBD dilatation with abrupt downstream narrowing, irregular wall thickening, and enhancement. Because these lesions tend not to be mass forming, only 11 % have associated main pancreatic duct dilatation. Main pancreatic duct dilatation occurs when the tumor extends into the downstream ampulla of Vater or into the surrounding pancreatic parenchyma [19].

Dynamic multiphasic abdominal MRI with an exogenous intravascular-extracellular contrast agent provides anatomic and enhancement characterization of PSC and its complications that are analogous to CT. An advantage of MRI is better contrast resolution compared to CT. A disadvantage of MRI is decreased spatial and temporal resolution compared to CT. Decreased spatial resolution and increased noise from physiological motion is also worse with MRI because of its relatively slower data acquisition time compared to CT. However, because of the differences in image content, CT and MRI are unpredictably complementary, and both are often used in cases of PSC.

Noncontrast MRI is used to generate two fundamentally different types of images. T2-weighted (T2W) images are based on differences in the micromagnetic environment of water-associated protons in fluid versus solid tissue. T2W MRI displays fluid as markedly hyperintense compared to an intermediate to hypointense soft tissue background. T1-weighted (T1W) images are derived from differences in the macromolecular environment of water-associated protons in fluid versus soft tissue. Using T1W MRI, fluid appears hypointense compared to mild to moderately hyperintense soft tissue. Because T1W images can be acquired faster, spatial resolution is better than with T2W scanning.

The inherently high contrast resolution of MRI can be augmented by intravenously administered exogenous contrast material. With the exception of hepatobiliary-specific agents, the pharmacokinetics of gadolinium-based MRI contrast is equivalent to iodinated CT contrast material. Intravenously administered gadolinium-based MRI contrast, which is not hepatobiliary specific, is used to generate a multiphasic dynamic series of T1W images that are analogous to multiphasic dynamic CT. Gadolinium-based agents increase the contrast resolution and signal-to-noise ratio, improving spatial resolution and lesion conspicuity. Because of its intravascular-extracellular distribution, gadolinium contrast demonstrates the same enhancement features of focal and diffuse pathology and of normal background anatomic structures as does iodinated CT contrast. As a result, a dynamic multiphasic T1W MRI series can be generated with arterial, portal venous, and delayed phases, with hypervascular lesions appearing hyperintense and hypovascular lesions being hypointense. With routine MRI scanning protocols, gadolinium contrast does not produce clinically significant changes in T2W images; postcontrast T2W scans are not obtained.

Using conventional contrast-enhanced MRI, the depicted features of PSC and its complications are the same as with CT (Fig. 13.3b). With multiphasic T1W MRI, the bile duct changes of PSC are shown as wall irregularity, thickening, and enhancement. Biliary obstruction is shown as duct dilatation accentuated by retained intraluminal bile that remains hypointense to the liver. Intrahepatic or perihilar mass-forming CCA can show arterial phase rim enhancement with centripetal washin during the portal venous and delayed phases. On T2W images, biliary obstruction is shown as duct dilatation accentuated by retained intraluminal bile that is hyperintense to the liver. Mass-forming CCA tends to be mild to moderately hyperintense compared to background hepatic parenchyma on T2W scans.

The initial detection and diagnosis of PSC by US, CT, and MRI are usually limited to previously undiagnosed patients presenting with unexplained abdominal pain and jaundice. When PSC and/or its complications are clinically suspected or established, MRC becomes an important noninvasive imaging modality. MRC is a heavily T2W MRI technique that generates high signal intensity from fluid bile. The intrinsic T2W hyperintensity of bile outlines the luminal morphology of normal and abnormal bile ducts against such an extremely hypointense background that solid tissue becomes indiscernible. Several sets of MRC images are acquired using different parameters to optimally demonstrate the biliary tree. Data sets can be directly obtained or indirectly produced by postprocessing in any anatomic plane for display. Directly acquired thick-slab coronal images with multiple obliquities around the sagittal axis and high resolution 3-D images reconstructed with postprocessing into a coronal rotational VRT data set result in images that are equivalent to invasive positive-contrast cholangiography (ERC and PTC). The multiprojectional and rotational features of MRC are optimal for the display of significant bile duct findings that could otherwise be obscured by the overlap of structures.

The MRC findings of PSC are the same as those described for ERC and PTC (Fig. 13.4) [25, 45]. Dave et al. reported a meta-analysis of the diagnostic performance of MRC compared to ERC and PTC [13]. Studies were selected only if they included a control group of patients with other hepatobiliary diseases. Of the manuscripts that fulfilled criteria for analysis, the overall prevalence of PSC among the study patients was 185/456 (41 %). MRC interpretations were compared to ERC or PTC as the reference standards. MRC had results comparable to conventional cholangiography with a sensitivity in detecting PSC of 86 % and a specificity of 94 %. In addition, three clinical scenarios were simulated to evaluate the impact of pretest probability on the results. When the pretest probability of PSC was 25 % (low clinical suspicion), the posttest probability of PSC given a negative MRC was 5 % (considered sufficient to exclude PSC). When the pretest probability was 75 % (high clinical suspicion), the posttest probability of PSC given a positive MRC was 98 % (considered sufficient to diagnose PSC). In what was considered the worst-case scenario, a pretest probability of 50 %, the posttest probability of PSC given a positive MRC was 94 %, and the posttest probability of PSC given a negative MRC was 13 %. MRC can be quickly performed in conjunction with dynamic multiphasic MRI providing additive information in cases of PSC and its complications [27, 34, 37].

In a retrospective study of 64 PSC patients, Ruiz et al. suggested that MRC features combined with multiphasic liver MRI findings can be used to predict PSC disease progression [34]. All patients had at least two MRCs separated by at least a 1-year interval with multiple scans performed in some patients. A semiquantitative method was used to systematically score both MRI and MRC findings to assess PSC disease severity. Scores from the first and last MRI and MRC were compared, with an interval increase in score considered disease worsening, no score change considered disease stability, and a decrease in score to be considered improvement. At mean follow-up of 4 years (range, 1–9), 58 % showed radiologic worsening, 42 % remained stable, and no patient showed improvement. Using data derived from the subgroup with interval worsening, two MRI progression risk score equations were developed, one for studies performed without contrast and another for studies performed with contrast. It was noted that nearly 90 % of patients with radiologic worsening had an elevated progression risk score, compared to a low progression risk score in nearly 85 % of patients with stable disease. In addition, over the study interval, 5/64 (8 %) patients were diagnosed with PSC-associated malignancies, CCA (n = 2), GBC (n = 2), and HCC (n = 1). Ruiz et al. concluded that risk score analysis could predict PSC disease progression and suggested that annual MRI and MRC were useful for PSC surveillance [34].