Cystic fibrosis-associated liver disease (CFLD) leads to important morbidity and mortality [
Dysfunction of the CFTR protein located on the apical pole of the cholangiocyte membranes [
Early recognition is important for diagnosis of fibrosis and prevents its complications, especially portal hypertension (PHT). Tools available for diagnosis of cystic fibrosis-associated liver disease (CFLD) are scarce [
The objective of the present cohort study was to describe the relevance of magnetic resonance cholangiography and transient elastography for assessment of cystic fibrosis-associated liver disease in adult patients.
A retrospective one-year cross-sectional cohort study was performed in our cystic fibrosis reference centre at Rouen University Hospital. All adult patients with CF, investigated by hepatobiliary MRI and by transient elastography for liver stiffness measurement (LSM) between July 2009 and July 2010, were included. We excluded patients in whom CFTR-related disorder was limited to one-organ dysfunction (i.e., congenital bilateral absence of vas deferens). Clinical and genetic characteristics were retrospectively collected from patient charts and included meconium ileus, pancreatic insufficiency, diabetes mellitus, body mass index,
LSM by transient elastography was measured by Fibroscan® (Echosens®, Paris, France, size M). Ten measurements were taken in 3 different sites in the hepatic right lobe through an intercostal space, with the patient’s right arm in maximal abduction. The results are expressed as a mean of 10 valid measurements. To be valid, the measurements had to meet two conditions: (i) an adequate success rate defined as the number of valid measurements divided by the total number of measurements performed greater than or equal to 60% and (ii) an interquartile range (IQR) of less than 30% of the median to take into account the variability of valid measurements. Results were expressed in kilopascal (kPa) and a correspondence table to the Metavir scoring system was used based on previous study of transient elastography in chronic biliary disease [
MRI examination was performed with 1.5 Tesla (Philips Achieva, Philips Medical Systems, Best, Netherlands) using a torso phased-array coil. No particular preparation was required. Fasting 4 hours before was required. All patients were placed in supine position with the upper abdomen centred on the coil. The following sequences were performed: (1) T1-weighted sequence, axial image (TR 183 ms, TE 2.3 ms, FOV 70 mm, slice thickness 7 mm, angle 55°, 152 × 432); (2) T2-weighted sequence, axial SPAIR (TR 4459 ms, TE 70 ms, FOV 76 mm, slice thickness 6 mm, angle 90°, 218 × 320); (3) T2-weighted sequence, axial HR (TR 1573 ms, TE 100 ms, FOV 79 mm, slice thickness 7 mm, angle 90°, 341 × 560); (4) T2-weighted sequence diffusion 2b (TR 1489 ms, TE 59 ms, FOV 90 mm, slice thickness 6 mm, 92 × 67); (5) 3D MR cholangiogram (TR 1341 ms, TE 574 ms, FOV 100 mm, slice thickness 2.4 mm, angle 90°, 221 × 560); (6) in and out phase sequence (TR 175 ms, TE 2.3 ms (in), 4.8 ms (out), FOV 40°, slice thickness 4 mm, angle 80°, 224 × 192). Abdominal radiologists (CSC and EK) reviewed all MRI results blinded to clinical or biochemical parameters and reached decisions by consensus. Native and 3D MIP (maximum intensity projection) reconstructions were analyzed on a workstation (EasyVision, Philips Medical Systems, Best, Netherlands). The following items were studied for each patient using a standardized scale: atrophy of either right or left hepatic lobe and/or hypertrophy of the caudate lobe, marked lobulations of liver surface, first-segment hypertrophy, splenomegaly (long axis superior to 12 cm), portal vein dilatation (diameter superior to 12 mm), splenic vein dilatation, intrahepatic or extrahepatic biliary duct irregularity (segmental strictures and dilatations), ascites, and steatosis. Pancreatic patterns were also noted: partial or total pancreatic fatty involution, Wirsung duct irregularity, and pancreatic cysts.
PHT was diagnosed in the presence of one or more of these signs: collateral circulation, portal vein dilatation, splenic vein dilatation, or splenomegaly. Hepatic dysmorphia was diagnosed in the presence of one or more of these signs: atrophy of either right or left hepatic lobe, hypertrophy of the caudate lobe, marked lobulations of liver surface, and first-segment hypertrophy.
We studied the results of LSM and hepatic MRI, according to the results of LFT or US.
Statistical analysis was conducted using SAS software version 9.3 (SAS Institute, Cary, NC, USA). SI units were used for all laboratory values with data summarized using mean ± standard deviation (SD) for continuous variables and number (%) for all recorded categorical variables describing the study population.
LSM are expressed in kPa as median (IQR). Student’s
Regarding concordance between study methods, agreement was evaluated by Kappa-test: 1 = perfect agreement, 1–0.8 = almost perfect agreement, 0.8–0.6 = good agreement, 0.6–0.2 = moderate and fair agreement, and less than 0.20 = slight agreement. To assess the diagnostic performance of LSM for prediction of PHT, the area under the receiver operating curve (AUROC) was calculated. Optimal LSM for prediction of PHT was identified by estimating sensitivity and specificity for various cut-offs.
Prevalence of abnormalities in MRI and LSM was compared regarding the presence or not of LFT and/or US abnormalities using chi-square test and Fisher’s exact test.
All statistical significance was taken at 95% confidence interval and
Of 64 adult CF patients followed up at our tertiary care center, 25 were included (Figure
Patients’ characteristics at study enrolment.
Patients’ characteristics | Studied patients |
Whole CF cohort |
|
---|---|---|---|
Gender (male/female) | 0.46 | 0.50 | 1 |
Median age (yrs [min.–max.]) | 25 |
30.1 |
0.12 |
Meconium ileus | 4 (16%) | 12 (21%) | 0.76 |
Pancreatic insufficiency | 22 (88%) | 49 (85.9%) | 1 |
Genotype | |||
|
11 (44%) | 20 (35.1%) | 0.46 |
|
12 (48%) | 31 (54%) | 0.63 |
2 other mutations | 1 (4%) | 6 (10%) | 1 |
Medium BMI (kg/m2 [min.–max.]) | 19.3 |
21.4 |
0.11 |
Alcohol | 0 (0%) | 0 (0%) | 1 |
Diabetes mellitus | 4 (16%) | 13 (22.8%) | 0.57 |
Dyslipidemia | 1 (4%) | 1 (1.7%) | 0.52 |
Antibiotics | 17 (68%) | 41 (72%) | 0.79 |
Antifungal drugs | 19 (76%) | 29 (51%) | 0.051 |
UDCA treatment | 10 (40%) | 14 (24%) | 0.19 |
Creatinine clearance ≥ 70 mL/min | 25 (100%) | 57 (100%) | 1 |
|
|||
FEV1 (%) | 67.6% |
71.2% |
0.33 |
BMI: body mass index; UDCA: ursodeoxycholic acid.
Flow chart for inclusion in our study.
At least one abnormal MRI sign was found in 13 patients (52%). We evidenced isolated hepatic dysmorphia, PHT, and isolated biliary tract abnormality in 3 patients (12%), 5 patients (20%), and 1 patient (4%), respectively (Figure
MRI results.
Diagnosis in MRI | Specific abnormalities in MRI |
|
Total number (%) |
---|---|---|---|
Abnormal MRI | 13 (52%) | ||
|
|||
Hepatic dysmorphia | Included lobulation of liver surface | 3 | 3 (12%) |
Included first-segment hypertrophy | 0 | ||
|
|||
Portal hypertension | Included splenomegaly | 5 | 5 (20%) |
Included portal vein dilatation | 2 | ||
Included splenic vein dilatation | 1 | ||
|
|||
Isolated biliary abnormalities | Intrahepatic | 1 | 1 (4%) |
Extrahepatic | 0 | ||
|
|||
Hepatic dysmorphia and portal hypertension | 0 | 0 (0%) | |
|
|||
Hepatic dysmorphia and biliary abnormalities | Included lobulation of liver surface and intrahepatic abnormalities | 1 | 1 (4%) |
|
|||
Portal hypertension and biliary abnormalities | Included portal vein hypertrophy and intrahepatic abnormalities | 1 | 1 (4%) |
|
|||
Hepatic dysmorphia, portal hypertension, and biliary abnormalities | Included splenomegaly, lobulation of liver surface, first-segment hypertrophy, and intrahepatic and extrahepatic abnormalities | 2 | 2 (8%) |
|
|||
Normal MRI | 12 (48%) | ||
|
|||
Total | 25 (100%) |
3D MRI cholangiogram shows intrahepatic and extrahepatic biliary duct irregularity, with choledochus stenosis.
Of the 25 patients included, 2 were excluded due to success rate of less than 60% in transient elastography, with mismatch between sites of measurement. Taking into account the median of the 10 highest measures for each patient, the median value of LSM was 5.7 kPa (3.4–9.9) in this cohort. Fibrosis scores were as follows: F0-F1 in 19 patients (82.6%), F2 in 3 patients (13%), and F3 in 1 patient (4.3%), and no patient reached the F4 stiffness threshold. All patients with a score greater than or equal to F2 were men (
In patients with evidence of PHT, mean LSM was 7.85 KPa (3.7–9.9) compared to 5 kPa (3.4–7.5) in patients with no evidence of PHT (
ROC curve for liver stiffness measurement predicting the presence of portal hypertension.
For diagnosis of CFLD, there was low concordance between transient elastography and MRI (
With a median of 2 LFT per patient, only 6 patients had at least 1 abnormal LFT (28%) during the study period.
Of the 60% of patients with routine US, 5 had abnormal US findings: 3 patients with signs of PHT and 2 with gallstones without evidence of any complications. All patients with abnormal US findings had at least 1 abnormal LFT.
FEV1 was not different between patients with or without PHT (
Table
Results of LSM and MRI according to the results of liver function tests and abdominal ultrasonography.
LFT or US abnormalities |
| |
---|---|---|
Yes ( |
No ( | |
50% F0-F1 ( |
94% F0-F1 ( |
0.04 |
50% ≥ F2 ( |
6% ≥ F2 ( | |
|
||
33.3% normal MRI ( |
58.8% normal MRI ( |
0.37 |
66.7% abnormal MRI ( |
41.2% abnormal MRI ( |
MRI: magnetic resonance imaging; US: ultrasonography; LFT: liver function test.
Noninvasive assessment of CFLD is an emerging topic with important implications [
The prevalence of liver disease varies considerably across studies, depending on the diagnostic methods used, from 27 to 35% [
Transient elastography for LSM is a promising tool in the field of hepatology. It has replaced LB in several chronic liver diseases including viral hepatitis [
Transient elastography however obviates this risk and multiple measurements may be appropriate in overcoming underestimation. We chose to apply the “rule of 10 valid measurements” on three sites but it may be worth considering other approaches including increased number of sites of measurement in such clinical conditions.
As others, we found median values of LSM to be quite low. Witters et al. reported a mean of 5.6 kPa in pediatric patients [
It is usual to consider that LSM is a valuable tool to distinguish low fibrosis (F0–F2) and F3-F4. LSM may be a valid approach to screen early CFLD. To do so, cut-off values for CFLD are mandatory. For the first time, Robertson et al. reported a cut-off at 8.83 kPa [
Our results show a link between increased LSM and PHT. This relationship is present even in patients without evidence of cirrhosis. We ruled out cirrhosis in the absence of F4 stage in LSM and in 75% of patients in the absence of dysmorphic liver feature on MRI. This suggests that noncirrhotic portal hypertension (NCPH) is secondary to other mechanisms. These mechanisms have already been described in primary biliary cirrhosis [
Early treatment of CFLD is of increasing importance in the management of patients with cystic fibrosis and prevention of severe liver damage. CFLD standard treatment is based on UDCA at a dose of 20–30 mg/kg/day [
Although it is a retrospective design on a small sample size, our study, based on noninvasive liver explorations using MRI and LSM, provides useful and additional information on assessment of CFLD. The results of such investigations are likely to impact clinical practice justifying further studies. Transient elastography for assessment of liver stiffness may soon play a central role in the noninvasive technique of portal hypertension screening. MRI seems essential for the CFLD diagnosis.
Cystic fibrosis
Cystic fibrosis-associated liver disease
Liver biopsy
Liver stiffness measurement
Liver function test
Magnetic resonance imaging
Kilopascal
Portal hypertension
Confidence interval
Ursodeoxycholic acid.
The authors declare that they have no competing interests.
C. Lemaitre, E. Billoud, M. Eliezer, M. Quillard, G. Riachi, E. Koning, H. Morisse-Pradier, and H. Montialoux contributed to acquisition of data and drafting of the paper. S. Dominique and C. Savoye-Collet contributed to acquisition of data, drafting of the paper, and critical revision. G. Savoye and O. Goria contributed to drafting of the paper and study supervision.
The authors are grateful to Nikki Sabourin-Gibbs, Rouen University Hospital, for her help in editing the paper.