Recommendations on the use of magnetic resonance imaging in PSC-A position statement from the International PSC Study Group

TECHNICAL REVIEW

MRCP uses high-strength magnets and takes advantage of the high T2-weighted (T2w) signal intensity of bile relative to surrounding structures to provide detailed images of the biliary tree and the pancreatic duct. Because this imaging modality is dependent on T2w images, contrast agents are not required to obtain a cholangiogram. Patients complete the examination while fasting to reduce fluid in the surrounding enteric structures that can interfere with visualization of the duct anatomy.

MRI contrast agents are used to improve the detection and differentiation of mass lesions and inflammation and assess liver function. Most MRI contrast agents are gadolinium based, an element with strong paramagnetic properties. Based on their biodistribution after intravenous injection, currently available MR contrast agents can be classified as purely extracellular, or extracellular with a hepatocyte-specific component. Analogous to iodine-containing contrast agents used in computed tomography (CT), extracellular contrast agents for MRI (e.g., gadopentetate dimeglumine, Magnevist, gadobutrol, Gadovist, Bayer, Leverkusen, Germany; gadoterate meglumine, Dotarem, Guerbet, Villepinte, France) are well suited for assessment of the vascular system and lesion detection and characterization on the basis of tumor morphology and perfusion, resulting in a nonspecific enhancement behavior.7, 8

Hepatocyte-specific (hepatobiliary) contrast agents are also referred to as combined or bimodal agents, because they offer imaging properties of conventional extracellular and liver-specific gadolinium chelates. By chemical modification of the ligands with lipophilic side chains, partial hepatocellular uptake and subsequent biliary excretion is mediated, increasing the signal intensity of the liver, bile ducts, and hepatocyte containing lesions at T1-weighted (T1w) imaging. Lesion characterization thus depends not only on vascularity, but also on hepatocellular function.9, 10 The bimodal properties of hepatocyte-specific contrast agents allow for dynamic contrast-enhanced imaging as well as for image acquisition in the so-called hepatobiliary phase, including the acquisition of a contrast enhanced T1w MRC. Currently, two hepatocyte-specific contrast agents are available: Gd-BOPTA (gadobenate dimeglumine; MultiHance, Bracco Imaging, Milan, Italy) and Gd-EOB-DTPA (gadoxetate disodium; Primovist, Eovist, Bayer, Leverkusen, Germany). Hepatic uptake and biliary contrast elimination of these contrast agents depend on liver function and amount to approximately 50% in patients with normal liver and kidney function for Gd-EOB-DTPA and 3%-5% for Gd-BOPTA, respectively.11

Of note, kidney function and contraindications need to be considered when applying gadolinium-based MRI contrast agents. In addition to nephrogenic systemic fibrosis, it has recently been described that gadolinium deposits can be detected in the brain after repeated contrast-enhanced MRI in patients even with normal kidney function.12, 13 This seems to depend on the stability of the gadolinium chelate with the more unstable linear chelates of gadolinium associated with higher risk than the more stable macrocyclic formulations.14, 15 However, the long-term clinical meaning of these deposits remains unclear to date.14 In this context, risk should be balanced against the potential benefits of contrast-enhanced MRI in patients with PSC and adherence to recently published consensus recommendations is advisable regarding the technical aspects of MRI.8, 16

FUNCTIONAL MRI REVIEW: A NONINVASIVE MEASURE OF LIVER FUNCTION

Dynamic contrast-enhanced MRI for the assessment of hepatic perfusion uses gadolinium diethylene triamine pentaacetic acid (Gd-DTPA) or gadoxetic acid (Gd-EOB-DTPA).17, 18 Around 50% of Gd-EOB-DTPA is excreted by hepatocytes into bile canaliculi and the biliary system, allowing a two-compartment model measurement of liver perfusion and hepatobiliary function.19 Preclinical studies demonstrated the feasibility of dynamic hepatocyte-specific contrast-enhanced MRI to calculate functional parameters, such as hepatocyte extraction fraction and input-relative blood flow.20, 21 Calculating these parameters in patients with PSC, the segmental hepatocyte extraction fraction and input-relative blood flow was heterogeneously distributed throughout the liver and seemed to correlate with segmental biliary obstruction.20 This method is promising, but an external software is needed for the postprocessing and it seems too time-consuming for use in routine clinical practice.

In a recent MRI study, T1 mapping of the liver was shown to be a feasible technique to evaluate liver function on a global level and may be extrapolated on a segmental level in patients with PSC. T1 reduction correlated with liver enzymes, disease stage, and Mayo risk score.22 An emerging alternative method, which may be more applicable in the clinical setting, is the relative liver parenchymal contrast agent enhancement index, delivering quantitative information on contrast agent uptake as a sign of active inflammation or structural changes.23

MAGNETIC RESONANCE ELASTOGRAPHY REVIEW: A NONINVASIVE MEASURE OF LIVER FIBROSIS

Liver stiffness is a surrogate marker for fibrosis and can be measured by elastography. The two principle elastography techniques in clinical practice include magnetic resonance elastography (MRE) and shear-wave–based ultrasound techniques, such as transient elastography (TE). MRE involves delivering a shear wave to the patient through an external driver. The propagation of this shear wave is visualized with a special MRI sequence, and software algorithms use this information to generate an elastogram where regions are selected by a radiologist to determine the average stiffness (kilopascals; kPa).24 TE is an ultrasound shear-wave–based technology that can also measure liver stiffness. Both have been shown to correlate well with the histological stage of fibrosis and outcomes among patients with PSC.25-27 Use of MRE may offer several key advantages when compared to TE. For example, the performance of MRE is not influenced by obesity or anatomical constraints such as narrow intercostal spaces. Second, fibrosis in PSC can be patchy and MRE can assess more than 1,000 times the volume of liver than TE.28, 29 Next, MRE can be performed at the same time as MRCP without adding a significant amount of time or cost to the examination. This approach also allows for the identification of worsening strictures, which is important because the presence of a biliary obstruction may increase liver stiffness irrespective of the degree of fibrosis.30-32 To date, there are no head to head performance comparisons between TE or other shear-wave–based techniques and MRE among PSC patients, and MRE is not widely available in Europe.

In addition to MRE, diffusion-weighted imaging (DWI) can be used to quantify liver fibrosis and inflammation and may support the detection of liver tumors. However, it cannot discriminate fibrosis stages as well as MRE or TE and the technique has not been transferred into routine clinical practice.33-35

CLINICAL NEED AND EVIDENCE SUMMARY

Cholangiography is required to diagnose large-duct PSC.4, 5 Practice guidelines from all major societies support MRCP as the first diagnostic modality in patients with suspected PSC.4, 5, 36 The classic cholangiographic features of large-duct PSC include multifocal strictures and areas of dilatation or ectasia and ductal wall thickening involving the intra- and/or extrahepatic bile ducts (Figs. 1 and 2). T1w, T2w, and contrast-enhanced T1w images may demonstrate thickening of the wall of large bile ducts and signs of periductal inflammation.37 Peripheral intrahepatic duct obliteration (commonly referred to as pruning) can also be observed. Notably, the exclusive involvement of extrahepatic bile ducts is infrequent, and, when encountered, an alternative diagnosis should be considered.38 Retraction of the papilla can also be observed on MRCP.39 The cholangiographic features of PSC are not pathognomonic, and distinguishing between primary versus secondary causes of sclerosing cholangitis, particularly when IBD is absent, is challenging.40 For example, in the absence of pancreatic or other immunoglobulin G4 (IgG4)-related organ involvement, IgG4 associated cholangitis (IAC) can appear very similar to PSC.41, 42

Parenchymal changes may also be observed in PSC. For example, the liver can show segmental or lobular atrophy with compensatory hypertrophy attributed to chronic bile duct obstruction43 (Fig. 2). Patchy areas of peripheral parenchymal enhancement and gallbladder enlargement have been reported in early PSC, but the sensitivity and specificity of these observations in PSC is unknown.44

The diagnostic accuracy of MRCP reaches that of endoscopic retrograde cholangiopancreatography (ERCP; 83% and 85%, respectively).45 A meta-analysis including studies performed between 2000 and 2006 has shown a sensitivity of 0.86 and a specificity of 0.94 for the diagnosis of PSC.46 Furthermore, performing an initial MRCP (rather than ERCP) has shown to be a cost-effective diagnostic approach.47, 48 Moreover, MRI including MRCP is noninvasive, does not expose patients to radiation, and it can assess the liver parenchyma and surrounding structures.

When MRCP was performed in a population of patients with IBD (regardless of symptoms or liver tests), it was useful in detecting subclinical PSC and increased the prevalence of PSC by 3-fold.49 Hence, MRCP has the potential to detect early PSC. In the future, early detection of PSC could aid in prolonging patient transplant-free survival once effective treatment options are available and has the potential to identify patients for clinical studies at a time point when anti-inflammatory or -fibrotic drugs still have a chance of being effective. However, even using modern 3 Tesla (T) scanners, MRI still has limitations in the assessment of the distal part of the common bile duct and also in detecting subtle pathologies of smaller peripheral bile ducts.37, 50, 51 Indeed, approximately 10% of patients with PSC will have a normal cholangiogram (small-duct PSC), and a liver biopsy is required to establish the diagnosis.4, 5, 36, 52

Variations in MRI scanners and imaging protocols may lead to image heterogeneity across institutions. The working group's discussion of a common minimal standard for performing MRI in PSC took into consideration the technical capabilities, cost and duration of the scanning protocol, the increased risk of diagnosing CCA within the first year of diagnosing PSC, and the requirement to compare imaging studies from different centers in future studies. These recommendations are not based on strong evidence, but on multidisciplinary expert discussion, and represent the opinion of the majority of hepatologists and radiologists attending the workshops and participating in the subsequent discussions. For more information on MRI technical aspects, we would like to refer to the recent consensus statement of the European Society of Gastrointestinal and Abdominal Radiology.8

GUIDANCE STATEMENTS

CLINICAL NEED AND EVIDENCE SUMMARY

In patients presenting with symptoms of cholestasis or cholangitis, or worsening laboratory tests suggestive of biliary obstruction, endoscopic intervention can improve symptoms and may have a positive impact on disease progression.62 A preprocedural MRI/MRCP can provide valuable information to better guide an endoscopic or percutaneous intervention. This is exemplified by several clinical scenarios commonly encountered during the care of patients with PSC. First, it can provide a roadmap to better facilitate biliary drainage by: identifying the extent and location of strictures without the need to inject contrast in the biliary tree; recognizing the presence and location of atrophic segments so they can be avoided because stenting an atrophic lobe is unlikely to improve jaundice and is associated with infection63; targeting strictures associated with an abscess; assessing biliary anatomic variants (occurs in approximately 40% of the population); and postoperative anatomy (e.g., posttransplant anastomotic strictures).64 Second, T1w images may also aid in the detection of biliary stones, which often show a higher signal intensity on T1w images and may be found at higher frequencies in PSC, especially in prestenotic and dilated bile ducts.65 Third, it can allow more specific targeting of brushings and biopsies when there is concern for malignancy. Last, it can be used as a triage tool to determine whether an ERCP can be avoided.

MRI/MRCP can also detect disease-related complications. Chief among these is CCA (discussed below). However, MRI/MRCP can detect other malignancies associated with PSC, including hepatocellular carcinoma or gallbladder cancer. MRI/MRCP can distinguish between hepatic abscesses from malignant mimickers with accuracy greater than 95%, particularly when DWI is used.66 Furthermore, it can evaluate for the progression of biliary strictures.67 Endoscopic or percutaneous dilations are frequently performed in the setting of “dominant strictures,” and there is consensus that this should be done in patients with signs of bacterial cholangitis and new or worsening symptoms of cholestasis, such as pruritus or jaundice.4, 5 The term dominant stricture derives from ERCP studies and defines strictures of less than 1.5 mm in diameter in the common bile duct and less than 1 mm in the left or right and also common hepatic duct.68 However, the applicability of this definition and inter-rater agreement of so-called dominant strictures when MRI/MRCP is used is not well understood.

GUIDANCE STATEMENTS

CLINICAL NEED AND EVIDENCE SUMMARY

CCA may develop in approximately 10%-20% of individuals with PSC, and it is the leading cause of mortality.52, 69 The annual incidence of CCA is approximately 0.5%-2.0%, and nearly 30% of biliary cancers are detected within the first year of establishing a PSC diagnosis.52, 70 This illustrates that the development of a malignant biliary stricture can be a heralding event that brings patients to the attention of clinicians. When jaundice and other manifestations of a malignant biliary stricture occur, CCA is often detected at a late stage when curative therapy is no longer an option. For example, in a study that examined dominant strictures associated with CCA, median survival was only 6 months.71 Patients diagnosed with PSC late in life have an increased incidence of CCA when compared to their younger counterparts.72, 73 Early detection of CCA in this high-risk population is important. Protocols that utilize neoadjuvant chemoradiation, brachytherapy, and, ultimately, LT offer a potential cure among a select group of PSC patients with early-stage CCA.74 The presence of a mass lesion with delayed venous enhancement is nearly 100% specific for CCA (Fig. 4).36, 75, 76 Vascular encasement may also be present. Cross-sectional imaging is essential for delineating the size and extent of the tumor, which will, in turn, help determine patients' eligibility for curative therapies.74 Indeed, MRI/MRCP is more sensitive (89%) than CT (75%) and ultrasound (57%).75 Prospective studies have also reinforced that the sensitivity/specificity of MRI/MRCP (88%/85%) is better than CT (79%/79%) for CCA detection in PSC.77 The use of contrast media increases the sensitivity for CCA detection by 10% without diminishing the specificity when compared to a noncontrast MRCP.75 Hence, a contrast-enhanced MRI/MRCP is preferred when there is a concern for CCA. A mass lesion is often absent in early-stage CCA, and distinguishing benign from malignant strictures is difficult. Imaging features considered indeterminate for CCA include bile-duct wall thickening, irregularity or enhancement, marked biliary dilation, the presence of a so-called dominant stricture, or focal biliary dilation with ipsilateral lobe atrophy. The development of such features should prompt additional studies including ERCP with biliary brushings/biopsies.78-80

There is neither supporting nor refuting evidence to suggest that CCA screening with any test is associated with improved outcomes or cost-savings. However, MRI/MRCP is the preferred imaging modality to assess for CCA because of its improved ability to define biliary strictures.37, 75 Therefore, many practitioners take a rational approach to biliary tract cancer screening that involves the use of laboratory tests and imaging. Indeed, the majority of large-volume centers perform follow-up MRI/MRCP yearly or every other year.3, 69, 76, 79, 81 The rationale behind this strategy is to monitor for suspicious bile-duct strictures, new mass lesions, or gallbladder polyps to detect early malignancies and capture patients who might be eligible for curative therapy. Follow-up MRI/MRCP for CCA screening should be evaluated in prospective, multicenter studies.

GUIDANCE STATEMENTS

CLINICAL NEED AND EVIDENCE SUMMARY

Biomarkers that reflect PSC disease severity are needed for routine clinical practice and to serve as surrogate endpoints and stratification tools in clinical trials. Although evidence supporting MRI/MRCP as a biomarker in PSC is underdeveloped, this noninvasive tool has the potential to provide both structural and functional information that may hold prognostic relevance.

The radiological progression of PSC has been measured on MRI/MRCP, and nearly 60% of subjects were found to have radiographic disease advancement after 4 years in one study.67 Factors associated with radiographic progression include hepatic dysmorphy (e.g., hepatic atrophy), intrahepatic ductal dilation and the presence of portal hypertension (PH; when contrast was not used), or dysmorphy and parenchymal enhancement heterogeneity (when contrast was used).67 Whether these findings can be reproduced or predict clinical outcomes is unclear. However, the presence of PH can often be identified on cross-sectional imaging and is generally accepted as a marker of disease severity and, if detected, should prompt clinicians to institute screening measures such as assessing for varices.82 The presence and extent of arterial peribiliary hyperenhancement, in contrast to enhancement on other phases, was associated with a higher Mayo PSC risk score and may be a marker of active biliary inflammation.83 While in the early phases of development, several dynamic gadoxetate-enhanced MRI studies have shown that patients with PSC have a heterogeneously distributed liver function (compared to healthy controls) with delayed hepatobiliary excretion of this contrast agent. Delayed excretion correlated with liver tests, Mayo PSC risk score, and downstream biliary obstruction.22, 84-86 Liver stiffness, as measured by MRE, has been shown to predict hepatic decompensation in PSC and the optimal cutoff to predict cirrhosis was 4.9 kPa. This value is nearly identical to the values reported in TE-based PSC studies (recognizing that shear-based MRE measurements can be compared to Young's modulus-based TE measurements by dividing by a conversion factor of 3).25, 27 However, despite the potential of MRI/MRCP as a prognostic biomarker, there is a dearth of published information relating radiographic covariates and clinical outcomes, thereby limiting its contemporary use as a validated surrogate endpoint in clinical trials.

GUIDANCE STATEMENTS