Applied Radiology

The first technique to produce cholangiographic images by MRI was reported in 1991. However, it was not until the past three years that investigators have begun to extensively evaluate and refine the technique. MR cholangiopancreatography (MRCP) seems to be a very attractive alternative to imaging of the bile and pancreatic ducts, as it allows cholangiographic images to be obtained in a noninvasive manner.

Endoscope retrograde cholangiopancreatography (ERCP) has long been

considered the gold standard of cholangiographic imaging; however, it is not perfect. In the detection of choledocholithiasis, sensitivities are approximately 90%.2 ERCP also requires sedation, has potential complications, and is expensive. Complication rates for ERCP may be as high as 7%, and because of this, noninvasive MRCP is gaining attention. MRCP can be performed without sedation or contrast on any commercially available MR unit, at various field strengths.3 Given its potential usefulness, the question of whether MRCP is ready for widespread implementation remains. In this article, we will review the history of MRCP, discuss the various performance techniques which have been utilized, explain tips on how to perform and interpret MRCP, discuss findings of various disorders, discuss potential pitfalls, and review the role of MRCP.

Techniques

Patient preparation-Patient preparation for an MRCP examination is minimal. Some investigators have chosen to keep patients NPO for at least four hours to help reduce peristalsis and gastroduodenal fluid. Others have promoted the use of glucagon or buscopan to help reduce peristalsis, and recently, two studies4,5 have utilized a negative oral contrast agent to reduce the signal intensity from the bowel. Further investigation will be necessary to fully determine the role these agents should play.

Pulse sequences-At present, there is no clear consensus on the proper MRCP protocol. Initial techniques utilized gradient echoes with steady state-free precession (SSFP) or CE-FAST, which is very sensitive to slow flow and will only produce signal in tissues with very long T2 relaxation times. Therefore, only stationary fluid such as bile appears bright using this technique. These sequences have several limitations, however, including low signal-to-noise ratios, increased susceptibility to artifacts, high signal intensity of fat, difficulty in visualization of non-dilated ducts, sensitivity to motion (requiring long breath-holds), and a relatively limited area that can be scanned.1,6

The development of rapid acquisition with relaxation enhancement (RARE) techniques in 1986 was an important step in refining MRCP. Utilizing hybrid RARE techniques such as fast spin echo (FSE) or turbo spin echo (TSE), heavily T2-weighted images could be obtained rapidly. These sequences have several advantages over the gradient echo techniques, including improved signal-to-noise and contrast-to-noise ratios, better overall image quality, and less motion and susceptibility artifact.7,8

Recently, faster gradients have been developed which can now allow imaging of the entire biliary system in a single breath-hold. Those sequences are referred to as half-Fourier acquisition single-shot turbo spine echo (HASTE) or single-shot RARE. A few studies have evaluated this technique, demonstrating improved resolution and image quality over other breath-hold techniques.8,9,10 In a comparison study of three breath-hold techniques (gradient echo, FSE, and half-averaged single-shot RARE), the single-shot RARE sequence provided the best image quality and duct conspicuity. Contrast-to-noise ratios also were highest with the RARE sequence; however, FSE demonstrated higher signal-to-noise ratio. In this study, the FSE sequence was not optimized, however.8

Acquisition geometry-Most series have utilized 2D acquisition modes; however, some have shown excellent results with 3D acquisitions.11,12 Advantages of the 3D acquisition include improved resolution, though this sequence is long and, as such, is more susceptible to motion artifact than the 2D sequences.

Usual planes of imaging include axial, coronal, and oblique coronal, all oriented in an RAO projection to the bile duct bifurcation. When using 3D techniques or projection images, the coronal or oblique coronal plane should be utilized.

Images can be obtained as sequential slices, which can be reconstructed into 3D models or a single projectional image (figure 10). Slices less than or equal to 3 mm thickness allow for adequate reformations but must be weighed with the signal available on the scanner. If available, the use of torso phased-array coils will allow improved resolution.

Motion correction-Several options and breathing techniques have been examined as to their abilities to reduce motion artifact. Some researchers have been able to obtain long breath-holds from the patients, while others have had to segment the acquisition into several breath-holds. Increasing the number of excitations has been shown to allow shallow respiration, and respiratory triggering also has been attempted in order to suppress artifact. It was determined that resolution can be sacrificed to reduce imaging time and allow breath-holding in some instances, while increasing the resolution by increasing the number of excitations was indicated in order to reduce respiratory artifacts in other cases.2,9,10,12-17

Image interpretation

At our institution, we routinely review the individual source images acquired on a workstation. These images are critical in the identification of abnormalities. Most investigators have shown the source images demonstrate more abnormalities than the 3D reconstructions or single slice projection images.

Three dimensional reconstruction can be useful in displaying the relationship of structures and abnormalities. However, its current challenge is to gain

clinical acceptance by demonstrating cholangiographic images similar to those obtained by ERCP. Maximum intensity projection (MIP) algorithms are used to create 3D reconstructions. By rotating the image, the proper obliquities can be obtained to best depict anatomy and pathology. However, when using the MIP alorithm, valuable information such as visualization of smaller ducts may be lost during reconstruction, and stones may be obscured by the surrounding high intensity bile (figure 2). Axial source images which are projected into the coronal plane show significant degradation and misregistration. Therefore, source images obtained in the coronal plane provide the best coronal reconstructions.

Clinical results

Normal anatomy-Several studies have assessed the rate of visualization of normal ducts at MRCP. Non-breath-held TSE has demonstrated peripheral intrahepatic ducts in all dilated cases and 82% of non-dilated ducts.13 The main intrahepatic ducts were visualized in 97 to 100%, confluence of intrahepatic ducts in 86%, cystic duct in 59%, cystic duct insertion in 34%, and confluence of the CBD and pancreatic duct in 38% of the cases studied; the pancreatic duct was able to be visualized in the head in 66% and in the body in 31% of cases.13 The main pancreatic duct has been demonstrated using 2D FSE sequences in the head in 95% and body/tail in 96% of cases.2 In contrast, earlier studies utilizing a simple body coil showed a portion or all of the duct in only 65% of cases. Using respiratory gated 3D TSE, diagnostic pancreatograms were obtained in 67% of patients and these demonstrate a specificity of 69% in the confirmation of a normal pancreatic duct; this can be increased to 81% by utilizing source images.11 Using HASTE sequences and a phased-array coil, normal cystic ducts and branches of the pancreatic duct were demonstrated in 88% and 75% of patients, respectively.9

It seems that most MR techniques are capable of visualization of the extrahepatic bile ducts, central intrahepatic ducts, and portions of the pancreatic duct. Visualization of the cystic duct, ampullary region, and the pancreatic duct and its branches still remains problematic, however.

In several reports, measurements of the ducts with MRCP have underestimated the true size, as obtained at ERCP.14 This likely is secondary to lack of distention at MRCP and decreased signal at the periphery of the bile duct.

Anatomic variants-It is important to identify variations of normal anatomy in the biliary system, as the anomalous ducts can be injured during surgical procedures such as laparoscopic cholecystectomy.

Using non-breath-hold 2D FSE and a phased-array coil, MRCP proved useful for the detection of an anomalous course of the cystic duct. Sensitivities were 83%, 85%, 91%, and 86% for low insertion, medial insertion, parallel course to the hepatic ducts, and overall diagnosis of variant anatomy, respectively, with 100% specificity.15

The main anomaly of the pancreatic duct that is of clinical importance is pancreatic divisum (figure 3). Studies have reported up to 100% sensitivity (n=25) by using a combination of a body coil and torso multicoil for visualization.17 The majority of anomalous pancreatic ducts are best identified on axial source images. Other methods have had less success; on 3D TSE without routine use of axial images, only 67% (n=6) of anomalies were detected.11 Further studies utilizing newer techniques and a larger population will be necessary to truly assess the accuracy of MRCP.

Choledocholithiasis-Most gallstones will appear as a dark signal intensity structure on MRCP surrounded by high signal intensity bile. However, some stone compositions may have central areas of high signal. Also, soft stones and debris may not appear as low signal and can approach the signal intensity of bile. Experienced observers are now reporting sensitivities of greater than 90% (table 1) in the detection of cholelithiasis;2,12-14,16,17,19,20 however, larger studies using single-shot techniques are still needed to accurately assess the sensitivity.

There are several pitfalls which can lower the specificity in the diagnosis of choledocholithiasis. Presence of hemorrhage, protein, and debris all can alter the signal intensity of bile, mimicking the presence of a stone (figure 4). Strictures also may give the appearance of a stone. Examination of the duct in various locations may help in the differentiation. Small stones, especially in non-dilated ducts, also can lead to false-negative interpretation. Because of the narrowing of the CBD distally at the ampulla, the appearance of small stones impacted distally may be subtle (figure 5). If the duct is dilated, air may rise to the non-dependent portion, making distinction possible (figure 6). Therefore, obtaining proper history such as prior sphincterotomy or surgery is important. As with other abnormalities, evaluation of the source images is critical, as small stones may be obscured by the surrounding bile on MIP reconstructions.

Biliary strictures

MRCP is fairly accurate in diagnosing obstruction secondary to strictures and also as to the site (figure 7). Initial studies using SSFP demonstrated good sensitivity in detecting obstruction; however, visualization of the stricture and distal duct was limited.1,6,8 TSE techniques have demonstrated sensitivities of 86 to 100% in the diagnosis of obstruction and 93% in identification of the correct site.13,14,17 Up to 90% of strictures can be identified using 3D TSE.20 In their study, Macaulay et al showed no false-negative results, suggesting that, in the appearance of a normal duct, ERCP may be unnecessary.13 Distal obstructions can be difficult to diagnose and differentiation of benign from malignant strictures may not be possible. However, Guibaud et al demonstrated a sensitivity of 86% and specificity of 95% in diagnosing malignant obstruction.14

A potential problem in the MRCP evaluation of a stricture is that the duct distal to the narrowing may be difficult to see because of the collapsed nature and size of the duct. This can be problematic when evaluating length of strictures.13 The use of phased-array coils and single-shot techniques, which improve resolution, may allow better visualization of the distal stenosis.9

As with choledocholithiasis, several pitfalls exist. Anything which lowers the signal intensity of the bile, such as stone, debris, or hemorrhage, can give an appearance of narrowing. Ampullary masses and sphincter of Oddi dysfunction can lead to dilatation and may be difficult to diagnose on MRCP.

Pancreas

As indicated above, the pancreatic duct is more difficult to visualize than the bile ducts because of its smaller size, though at least a portion of the duct can be seen in a high percentage of individuals. This lack of visualization does not indicate a stricture and should be interpreted cautiously, especially in the absence of proximal dilatation. Three-dimensional TSE demonstrated an 87 to 100% sensitivity in detecting pancreatic duct dilatation. Also, HASTE sequences have been shown to identify 100% of dilated pancreatic ducts. However, only 50% of dilated branches off the main pancreatic duct may be visualized. The level of obstruction of the main duct is usually well demonstrated.9

Role of MRCP

In the preceding section, we discussed the accuracy of MRCP in the detection of various disorders of the biliary system and pancreas. However, exactly when to perform MRCP still needs to be defined. Potential clinical applications include the following:

Unsuccessful ERCP-MRCP has been shown to be useful in patients who underwent unsuccessful ERCP.21 Failure of ERCP may be due to many factors, including the inability to cannulate, postoperative biliary-enteric surgery, and significant duct obstruction. In these settings, MRCP seems a reasonable alternative compared to percutaneous cholangiography.

Preoperative evaluation-MRCP may provide useful information in patients undergoing laparoscopic cholecystectomy, demonstrating anomalous anatomy and choledocholithiasis.

Detection of choledocholithiasis-In patients with high suspicion of choledocholithiasis, ERCP should be performed regardless of the findings on MRCP. However, MRCP may be useful in the evaluation of patients with a low suspicion of choledocholithiasis or in patients considered to be at high risk for developing a complication during a diagnostic ERCP. Patients presenting with acute pancreatitis also may benefit from undergoing MRCP to exclude gallstone pancreatitis, as there is an increased risk of complications in performing ERCP in these individuals.

Strictures-In the evaluation of biliary and pancreatic strictures, ERCP still provides information unobtainable with MRCP. ERCP allows distention of the stricture, which may help differentiate benign from malignant causes, allow brushings to be obtained, and allow therapeutic procedures such as dilatation and stenting to be performed. However, one potential advantage of MRCP over ERCP is the visualization of masses associated with strictures. In tight strictures, ERCP may not be able to visualize the ducts proximal to the stricture. MRCP allows excellent visualization of the proximal dilated ducts and length of the stricture, and can guide surgical or percutaneous procedures.

Chronic pancreatitis-MRCP has shown the ability to visualize dilated ducts and to diagnosis chronic pancreatitis and its possible etiologies, such as pancreatic divisum. MRCP also is able to visualize intraductal stones. However, it is limited in its ability to demonstrate the side branches, in which dilatation can be an early change of chronic pancreatitis (figure 8).

Pancreatic function-MRCP has been reported to have the ability to evaluate for papillary stenosis and exocrine insufficiency.5 However, further studies will be necessary to corroborate these results.

Although the above scenarios seem to be reasonable, MRCP currently is being used mainly in a problem solving role. The goal of MRCP should be to replace diagnostic ERCP. At this point, the resolution of MRCP is still inferior to ERCP. Further technical developments such as HASTE imaging should allow MRCP to narrow the difference in resolution, while the shorter scan times will lower the cost. Further outcome studies will be necessary to confirm the equality of MRCP and ERCP. Refinement and optimization of the technique will need to be performed before widespread implementation of MRCP in order to assure acceptable accuracy rates. Despite this uncertainty, our clinical referrals are increasing. We believe that, while techniques are improving, MRCP is not yet ready to completely replace diagnostic ERCP. AR

Acknowledgment

Thanks to Kris McPeck, Heidi Wiedenfeld, and Dr. Ken Vitellas for their assistance in preparation of this manuscript.

References

1. Wallner BK, Schumacher KA, Weidenmaier W, Friedrich JM: Dilated biliary tract: Evaluation with MR cholangiography with a T2-weighted contrast-enhanced fast sequence. Radiology 181:805-808, 1991.

2. Reinhold C, Bret PA: Current status of MR cholangiopancreatography. AJR 166:1285-1295, 1996.

3. Pavone P, Laghi A, Catalano C, et al: MR cholangiopancreatography (MRCP) at 0.5 T: Technique optimisation and preliminary results. Eur Radiol 6:147-152, 1996.

4. Hirohashi S, Hirohashi R, Uchida H, et al: MR cholangiopancreatography and MR urography: Improved enhancement with a negative oral contrast agent. Radiology 203:281-285, 1997.

5. Matos C, Metens T, Deviere J, et al: Pancreatic duct: Morphologic and functional evaluation with dynamic MR pancreatography after secretin stimulation. Radiology 203:435-441, 1997.

6. Hall-Craggs MA, Allen CM, Owens CM: MR cholangiography: Clinical evaluation in 40 cases. Radiology 189:423-427, 1993.

7. Reinhold C, Guibaud L, Genin G, Bret PM: MR cholangiopancreatography: Comparison between two-dimensional fast spin-echo and three-dimensional gradient-echo pulse sequences. J Magn Reson Imaging 4:379-384, 1995.

8. Ichikawa T, Nitatori T, Hachiya J, Mizutani Y: Breath-held MR cholangiopancreatography with half-averaged single shot hybrid rapid acquisition with relaxation enhancement sequence: Comparison of fast GRE and SE sequences. J Comput Tomogr 20(5):798-802, 1996.

9. Miyazaki T, Yamashita Y, Tsuchigame T, et al: MR cholangiopancreatography using

HASTE (half-Fourier acquisition single-shot turbo spin echo) sequences. AJR 166:1297-1303, 1996.

10. Regan F, Fradin J, Khazan R, et al: Choledocholithiasis: Evaluation with MR cholangiography. AJR 167:1441-1445, 1996.

11. Soto JA, Barish MA, Yucel EK, et al: Pancreatic duct: MR cholangiopancreatography with a three-dimensional fast spin-echo technique. Radiology 196:459-464, 1995.

12. Barish MA, Yucel EK, Soto JA, et al: MR cholangiopancreatography: Efficacy of three-dimensional turbo spin-echo technique. AJR 165:295-300, 1995.

13. Macaulay SE, Schulte SJ, Sekijima JH, et

al: Evaluation of a non-breath-hold MR cholangiography technique. Radiology 196:227-232, 1995.

14. Guibaud L, Bret PM, Reinhold C, et al: Bile duct obstruction and choledocholithiasis: Diagnosis with MR cholangiopancreatography. Radiology 197:109-115, 1995.

15. Taourel P, Bret PA, Reinhold C, et al: Anatomic variants of the biliary tree: Diagnosis

with MR cholangiopancreatography. Radiology 199:521-527, 1996.

16. Chan YL, Chan ACW, Lam WWM, et al: Choledocholithiasis: Comparison of MR cholangiography and endoscopic retrograde cholangiography. Radiology 200:85-89, 1996.

17. Becker CD, Grossholz M, Becker M, et al: Choledocholithiasis and bile duct stenosis: Diagnostic accuracy of MR cholangiopancreatography. Radiology 205:523-530, 1997.

18. Bret PM, Reinhold C, Taourel P, et al: Pancreas divisum: Evaluation with MR cholangiopancreatography. Radiology 199:99-103, 1996.

19. Yamashita Y, Abe Y, Tang Y, et al: In vitro and clinical studies of image acquisition in breath-hold MR cholangiopancreatography: Single-shot projection technique versus multislice technique. AJR 168:1449-1454, 1997.

20. Soto JA, Barish MA, Yucel EK, et al:

Magnetic resonance cholangiography: Comparison with endoscopic retrograde cholangio-

pancreatography. Gastroenterology 110:289-597,

1996.

21. Soto JA, Yucel EK, Barish MA, et al: MR cholangiopancreatography after unsuccessful or incomplete ERCP. Radiology 199:91-98, 1996.

Dr. Fidler is Assistant Professor of Radiology and Section Chief of CT/MRI at the University of Nebraska Medical Center in Omaha, NE. Dr. Spritzer is an Associate Professor of Radiology at Duke University Medical Center in Durham, NC.