Dilated cisternae chyli: a sign of uncompensated cirrhosis at MR imaging

In order to retrospectively determine the frequency of dilated cisterna chyli (CC) on MR images in patients with cirrhosis, and to assess its value as a simple diagnostic imaging sign of uncompensated cirrhosis. Study population included 257 patients (149 with pathologically proved cirrhosis and 108 control subjects without the history of chronic liver diseases) who had 1.5 T MR imaging. Cirrhosis patients were divided into compensated and uncompensated groups. Three independent observers qualitatively evaluated the visibility of CC 2 mm or greater in transverse diameter, identified as a tubular structure with fluid signal intensity. CC diameters greater than 6 mm were defined as dilated. Statistical analysis was performed by Student t-test and interobserver agreement via intraclass correlation coefficient. CCs with diameter 2 mm or more were recorded in 113 of 149 (76%) cirrhotic patients and 15 of 108 (14%) control subjects (P < 0.001). Dilated CCs were significantly more frequent in uncompensated than compensated cirrhotic patients (54% vs 5%, P < 0.001). The sensitivity, specificity, accuracy, and positive predictive value of dilated CC for uncompensated cirrhosis were 54%, 98%, 80%, and 96%, respectively. Dilated CC can be used as a simple and specific sign complimentary to other findings of uncompensated cirrhosis.

About 80% of the lymph formed in the liver leaves via hilar lymphatics and enters the thoracic duct at or near the cisterna chyli (CC) [1]. In cirrhosis, there is an increased hepatic production of lymph, which may cause distension of the thin-walled lymphatic channels, particularly if there is increased resistance at the venolymphatic junction such as due to elevated systemic venous pressure [2]. Dilatation of porta hilar lymphatics and CC in patients with portal hypertension has been shown by lymphangiography, endoscopic ultrasound, and at autopsy [3][4][5][6][7].
The CC is commonly visible on MR images as high-signal intensity (SI) filled structure on heavily T2-weighted images, with delayed enhancement several minutes after the intravenous administration of gadolinium contrast agent [8]. In our current practice at MR imaging, we often noted an enlarged CC in patients with uncompensated cirrhosis. Thus the purpose of this study was to determine the frequency of dilated CC on MR images in patients with cirrhosis, and to assess its value as a simple diagnostic imaging sign of uncompensated cirrhosis.

Patients
From July 2005 through December 2006 at our institution, hepatic MR imaging reports, clinical MR imaging requests, and patient charts were searched by one author (S.K.V) to identify patients with cirrhosis and patients without chronic liver diseases. This retrospective review was conducted according to a protocol approved by the Institutional Review Board for retrospective review of radiology, clinical, and pathology records. Informed consent was not required. These cases were then cross-referenced to the histopathologic records to select patients with pathologically proved cirrhosis. Patients were excluded if they had a history of hepatic resection, cholecystectomy, or a malignant hepatic tumor larger than 2 cm, as these may affect the morphology of cirrhotic liver. In patients having more than one MRI study, the first MRI study was used for initial analyses.  [9,10]. Uncompensated cirrhosis was defined as the presence of at least two of the following five criteria: ascites, encephalopathy, hyperbilirubinemia >3 mg/dL, hypoalbuminemia < 3.2 g/dL, and prothrombin time >3 s [11].
Patients in the control group had undergone MR imaging for reasons other than chronic liver diseases (e.g., suspected liver cyst or hemangioma, benign disease of other organs). Reports indicated benign hepatic lesions including hemangioma, cyst, focal nodular hyperplasia in 59, pancreatic and biliary system abnormality in 19, liver metastasis in 11, normal in 10, or undetermined due to insufficient clinical data in 9.
Cirrhosis patients had been referred for MR imaging to evaluate the severity of cirrhosis and portal hypertension, to acquire preoperative studies before liver transplantation, or to evaluate suspected hepatocellular carcinoma.  which also enhance on delayed images, were distinguished from lymphatic channels due to their enhancement on early phase images, and their SI lower than fluid on heavily T2-weighted single shot FSE images. Images were reviewed on an independent imaging workstation. Each reviewer individually chose images for measurement on the basis of clarity of the landmarks. CCs with minimum transverse diameter of 2 mm constituted the study group. A cut-off point of 6 mm was used to define dilated CC [12]. Maximum transverse diameters at its widest part were recorded on axial heavily T2-weighted single shot FSE using electronic caliber.
Presence or absence of ascites was noted. The growth rate of CC was expressed as diameter doubling time [13] for the available serial MRI follow-up.  Table 1). Dilated CCs with maximum transverse diameter (range 6-13 mm; mean 7 mm) were seen in 2 of 43 (5%) in compensated and 57 of 106 (54%) in uncompensated cirrhosis patients (P< 0.001) ( Fig. 1) (Table 1). None of the control patients had dilated CC. Between two patients with compensated cirrhosis with dilated CCs (diameters 11 and 6 mm; mean 8.5 mm), one had associated evidence of right-sided cardiac decompensation (Fig. 2) and the other had prior history of pancreatitis. The sensitivity, specificity, accuracy, and positive predictive value of dilated CC for uncompensated cirrhosis were 54%,98%, 80%, and 96%, respectively. Among the 25 patients with serial MRI, the diameter of the CC on first MR images ranged from 3 to 11 mm (mean 5.6 mm) and 4 to 12 mm (mean 6.6 mm) on last MRI (mean follow-up 282 days) (not significantly different, P = 0.13). Child-Pugh classification was C in 23 of 25; 2 other patients progressed from Child B to Child C based on the increase in the bilirubin and development of ascites. In these two patients, CC diameter on first MRI ranged from 4 to 5 (mean 4.5 mm) and 5 to 6 (mean 6.6 mm) on last MRI (mean follow-up 278 days) (Fig. 3). Interobserver reliability assessment Intraclass correlation coefficient (ICC) assessment as described by Shrout and Fleiss [14] indicated excellent to almost perfect agreement. ICC for CC measurement in compensated cirrhosis, uncompensated cirrhosis, and control subjects were 0.99, 0.97, and 0.98. ICC for the measurement of dilated CC in compensated cirrhosis, uncompensated cirrhosis, and control subjects were 1.0, 0.97, and 0.97.

Discussion
The results of the present study confirm the ability to identify CC by MRI in healthy volunteers and in patients with cirrhosis. Further, dilated CC had high specificity and positive predictive value for uncompensated cirrhosis, present in 54% of patients with uncompensated cirrhosis with excellent interobserver agreement.
Few previous studies have reported the imaging appearance of CC [15,16]; the frequency of seeing the CC on lymphangiography ranged from 30% to 53% [4], but only 1.7% on CT [17] On MRI, Erden et al. [18] detected CC in 96% of patients on heavily T2-weighted images. Unlike these investigators, we used a threshold diameter of 2 mm for identifying CC because of potential variability in identifying structures smaller than a pixel. This probably accounted for our lower rate of visible CC of 76% in cirrhotic patients and 14% of control subjects.
Lymphatics from the lower portion of the body converge in the lumbar region of the abdominal cavity to form the CC, which extends for about 6 cm just to the right of the abdominal aorta. At the level of the 12th thoracic vertebra, the CC narrows and becomes the thoracic duct. In patients with portal hypertension, hepatic hilar lymphatics become distended and obstruct the hepatic venules. This phenomenon of increased lymph production in patients with advanced liver diseases is caused by disturbance in the drainage of vascular flow from the sinusoid to the central or terminal hepatic veins associated with lobular distortion [7,19,20].
We speculate that in uncompensated cirrhosis any rise in pressure in the portal system will bring about an increase in liver and splanchnic lymph production, which is a factor responsible for CC dilatation [21,22]. Little is known about the growth rate of such lymphatic channels [3]; however, in our study we did not find any significant change of diameter of CC among 25 patients over the period of less than 1 year.
It should be noted that the CC was visible in 14% of the control subjects. This may relate to the spectrum of normal physiology, but it is possible that some of these patients had increased thoracic duct lymph flow due to other conditions such as pancreatic and biliary obstruction, state of hydration (e.g., right-sided cardiac decompensation with elevated hepatic venous pressure) [23]. None of the control patients had CC wider than 6 mm, in contrast to 5% of those with compensated cirrhosis and 54% with uncompensated cirrhosis. The 54% (57 of 106) sensitivity of dilated CC for determining uncompensated cirrhosis is relatively low. However, its high specificity (98%; 149 of 151) renders this sign as complementary to other findings of uncompensated cirrhosis that may be more sensitive [24,25].
We acknowledge the following limitations of our study. We did not compare MRI with another gold standard, but our interobserver agreement for detection rate of CC among the readers was quite high, with acceptable interobserver agreement. We were not able to confirm our imaging findings by conventional direct or indirect portographic methods in patients with portal hypertension. Conventional portography is invasive and therefore was not performed in most patients. We could not evaluate changes in visibility of the CC in relation to progression from compensated to uncompensated cirrhosis, since only short-term follow-up was possible. A further prospective study would be necessary to determine whether the visibility or diameter of CC has prognostic significance or can be used to monitor disease progression.
Our control patients were chosen on the basis of imaging technique and period of imaging identical to that of the patients with cirrhosis. We did not match sex, race, or socioeconomic status between groups. However, we are not aware of any data to suggest that CC differs with respect to these variables. Since the control population was drawn from among people examined with MR imaging performed because they were suspected of having disease, this population may not be representative of the general population or of patients without cirrhosis who are at risk for cirrhosis.
Another limitation was that the diameter of CC was determined on the basis of MR imaging, not by means of histological confirmation. Pathologic proof, however, would not be practical in the clinical setting.
In conclusion, although CC is commonly visible in subjects without liver disease or with clinically compensated cirrhosis, dilated CC found at MR is specific for uncompensated cirrhosis.