fax: (33) 1 49 28 22 59
Diffusion-weighted magnetic resonance imaging for the assessment of fibrosis in chronic hepatitis C†
Article first published online: 30 JUL 2007
Copyright © 2007 American Association for the Study of Liver Diseases
Volume 46, Issue 3, pages 658–665, September 2007
How to Cite
Lewin, M., Poujol-Robert, A., Boëlle, P.-Y., Wendum, D., Lasnier, E., Viallon, M., Guéchot, J., Hoeffel, C., Arrivé, L., Tubiana, J.-M. and Poupon, R. (2007), Diffusion-weighted magnetic resonance imaging for the assessment of fibrosis in chronic hepatitis C. Hepatology, 46: 658–665. doi: 10.1002/hep.21747
Potential conflict of interest: Nothing to report.
- Issue published online: 24 AUG 2007
- Article first published online: 30 JUL 2007
- Manuscript Accepted: 26 MAR 2007
- Manuscript Received: 19 NOV 2006
Liver biopsy is the gold standard for assessing fibrosis but has several limitations. We evaluated a noninvasive method, so-called diffusion-weighted magnetic resonance imaging (DWMRI), which measures the apparent diffusion coefficient (ADC) of water, for the diagnosis of liver fibrosis in patients with chronic hepatitis C virus (HCV). We analyzed 20 healthy volunteers and 54 patients with chronic HCV (METAVIR: F0, n = 1; F1, n = 30; F2, n = 8; F3, n = 5; and F4, n = 10) prospectively included. Patients with moderate-to-severe fibrosis (F2-F3-F4) had hepatic ADC values lower than those without or with mild fibrosis (F0-F1; mean: 1.10 ± 0.11 versus 1.30 ± 0.12 × 10−3 mm2/s) and healthy volunteers (mean: 1.44 ± 0.02 × 10−3 mm2/s). In discriminating patients staged F3-F4, the areas under the receiving operating characteristic curves (AUCs) were 0.92 (±0.04) for magnetic resonance imaging (MRI), 0.92 (±0.05) for elastography, 0.79 (±0.08) for FibroTest, 0.87 (±0.06) for the aspartate aminotransferase to platelets ratio index (APRI), 0.86 (±0.06) for the Forns index, and 0.87 (±0.06) for hyaluronate. In these patients, the sensitivity, specificity, positive predictive value, and negative predictive value were 87%, 87%, 72%, and 94%, respectively, with an ADC cutoff level of 1.21 × 10−3 mm2/s. In discriminating patients staged F2-F3-F4, the AUC values were 0.79 (±0.07) for MRI, 0.87 (±0.05) for elastography, 0.68 (±0.09) for FibroTest, 0.81 (±0.06) for APRI, 0.72 (±0.08) for the Forns index, and 0.77 (±0.06) for hyaluronate. Conclusion: This preliminary study suggests that DWMRI compares favorably with other noninvasive tests for the presence of significant liver fibrosis. (HEPATOLOGY 2007.)
Liver biopsy is currently the gold standard for staging hepatic fibrosis. However, it is an invasive procedure with risks of complications. Furthermore, it has poor patient acceptance and shows possible interpretation and sampling errors.1, 2 Therefore, alternative noninvasive approaches have been developed to assess liver fibrosis in a wide range of chronic liver diseases, especially in patients with chronic hepatitis C virus (HCV) infection. These include routine biochemical and hematological tests, serum markers of connective tissue, and scoring systems using a combination of clinical and/or laboratory tests.3–8 However, such methods fail to quantify liver fibrosis in about 50% of patients.9 Other promising attempts in the assessment of liver fibrosis rely on imaging techniques. Recently, the measurement of liver stiffness with ultrasound transient elastography has been validated to detect significant fibrosis in patients with chronic hepatitis C.10 However, transient elastography cannot be applied to patients with ascites, narrow intercostal spaces, and overweight status. In recent years, magnetic resonance imaging (MRI) techniques, including diffusion-weighted magnetic resonance imaging (DWMRI), have been developed to characterize diseased tissue. DWMRI measures the apparent diffusion coefficient (ADC) of water, a parameter that is dependent on the tissue structure. Several reports on the measures of DWMRI in normal and cirrhotic livers suggest a lower ADC value in the latter,11–13 which is thought to reflect a restriction of water diffusion in fibrotic tissue.
This prospective study was aimed at assessing the performance of DWMRI for the diagnosis of liver fibrosis in patients with chronic HCV infection with reference to other noninvasive methods: FibroTest, the aspartate aminotransferase to platelets ratio index (APRI), the Forns index, the serum hyaluronate concentration, and transient elastography.
Patients and Methods
Twenty healthy adult volunteers with no evident hepatic pathology were recruited after oral informed consent. They consisted of 9 women and 11 men and had a mean age of 31.4 ± 7 years (range: 22-49 years).
Sixty consecutive patients were prospectively enrolled, but only 54 of these were eligible for a complete analysis because transient elastography was actually not carried out in 6 patients (because of organizational failure). The 54 patients included 37 men and 17 women with a mean age of 46.4 ± 12 years (range: 25-75 years). The inclusion criteria were as follows: a chronic HCV infection defined by positive serum anti-HCV antibodies and detectable serum HCV RNA, the availability of a liver biopsy sample at least 10 mm long, the absence of clinical evidence for cirrhosis, and the oral informed consent of the patients. The exclusion criteria were as follows: a coinfection with human immunodeficiency virus or hepatitis B virus, other causes of liver diseases, hepatocellular carcinoma, and uninterpretable liver biopsy examination results. Our institutional review board approved the study.
MRI was performed before the liver biopsy examination, on the same day, with a 1.5-T superconducting magnetic resonance (MR) system (MAGNETOM Maestro Class Symphony, Siemens Medical Systems, Erlangen, Germany) with a 33 mT/m maximum gradient capability. A phased-array abdominal coil was made from a 6-element phase array settled on the anterior part of the patient and 4 elements of the spine coil integrated into the scanner table on the posterior part of the patient. Single-shot echoplanar diffusion-weighted imaging was synchronized with the patient respiration by means of the PACE (Prospective Acquisition Correction) navigator-triggered technique and was performed in all patients with 4 b values: 0, 200, 400, and 800 seconds/mm2. Parallel imaging (the GRAPPA [Generalized Autocalibrating Partially Parallel Acquisition] technique with an iPAT [integrated Parallel Acquisition Techniques] factor of 2) was used to reduce both the susceptibility artifacts and acquisition time within the expiratory period. The signal acquisition time within the expiratory period did not exceed 1400 milliseconds (duration of the single-shot acquisition), and the total acquisition time depending on the respiratory period of the patient remained less than 2 minutes for all patients. A frequency-selective fat saturation was additionally applied to cancel chemical shift artifacts. The MR sequence parameters were as follows: a TR of 1400 milliseconds, a TE of 76 milliseconds, a 128 × 128 matrix size, 3 excitations, a 36-40–cm field of view, 11 slices with a 8-mm slice thickness, and a bandwidth of 1302 Hz/pixel. The diffusion gradients were applied in 3 orthogonal directions, and quantitative ADC maps were automatically calculated on a voxel-by-voxel basis with commercially available software (Syngo2004A, Siemens Medical Systems). The liver signal intensity was recorded as the mean of values generated by the placement of 3 separate circular regions of interest (ROIs) of a 0.5-cm diameter over the right hepatic lobe on the ADC mapping images in 5 different slices. Care was taken to exclude vessels from the ROIs.
Measurements of liver stiffness by transient elastography were also performed before the liver biopsy examination, on the same day, with a FibroScan (Echosens, Paris, France) as previously described.10 Ten validated measurements were performed on each patient. The success rate was calculated as the number of validated measurements divided by the total number of measurements. The results were expressed in kilopascals (kPa). The median value was considered representative of the elastic modulus of the liver. Only procedures with 10 validated measurements and a success rate of at least 60% were considered as reliable and were retained.
Surrogate Serum Fibrosis Markers.
The following parameters were determined for blood sampled on the day of liver biopsy: serum aspartate aminotransferase, alanine aminotransferase, γ-glutamyl transpeptidase (GGT), total bilirubin, platelet count, serum hyaluronate concentration, cholesterol, prothrombin index, a2-macroglobulin, apolipoprotein A1, and haptoglobin. Aspartate aminotransferase, alanine aminotransferase, GGT, and total bilirubin were assessed with an automated Olympus AU640 and were calibrated on a calibrator for automated systems (Roche Diagnostics). Apolipoprotein A1, alpha2-macroglobulin, and haptoglobin levels were determined with a BNII immunonephelemeter (Dade Behring, Marburg, Germany). Serum hyaluronate was assessed with an enzyme-linked protein-binding assay (hyaluronic acid test kit, Corgenix, Peterborough, United Kingdom).
The APRI index was calculated as follows: aspartate aminotransferase (× upper limit of normal) × 100/platelet count (109/L).5
The Forns index was calculated according to the following formula: 7.811 − 3.131 × ln[(platelet count (109/L)] + 0.781 ln[GGT (IU/L)] + 3.467 ln[age (years)] − 0.014 [cholesterol (mg/dL)].4
The laboratory followed the preanalytical and analytical recommendations required to obtain FibroTest results.3 The transferability of the obtained results was validated for FibroTest calculations with BioPredictive Co. The FibroTest score (f) was calculated as follows: f = 4.467 × log[α2-macroglobulin (g/L)] − 1.357 × log[haptoglobin (g/L)] + 1.017 × log[GGT (IU/L)] + 0.0281 × [age (years)] + 1.737 × log[bilirubin (μmol/L)] − 1.184 × [apolipoprotein A1 (g/L)] + 0.301 × sex (female = 0, male = 1) − 5.540.
Liver Histology and Quantification of Liver Fibrosis.
Liver biopsy specimens were fixed in formalin and embedded in paraffin. All biopsy specimens were analyzed by an experienced pathologist blinded to the results of DWMRI, transient elastography, and blood tests. The mean size of the liver biopsy was 20.2 mm (range: 10-35 mm), and 20 patients had liver biopsies larger than 25 mm. Liver fibrosis and necroinflammatory activity were evaluated semiquantitatively according to the METAVIR scoring system. Fibrosis was staged on a 0-4 scale as follows: F0 = no fibrosis, F1 = portal fibrosis without septa, F2 = portal fibrosis and few septa, F3 = numerous septa without cirrhosis, and F4 = cirrhosis. The activity was graded as follows: A0 = none, A1 = mild, A2 = moderate, and A3 = severe. The presence of hepatic steatosis was also assessed and considered severe when greater than 30%.
Quantitative variables were compared with the Mann-Whitney test; trends were tested with the Jonckheere-Terpstra (JT) test.14 The intrareader and extrareader variability for ADC determination were evaluated by the calculation of intraclass correlation coefficients. The coefficient of variation of the ADC values was calculated as the ratio of the standard error of the mean to the mean of the 15 ROIs studied in each individual.
The diagnostic performance of DWMRI, transient elastography, FibroTest, APRI, the Forns index, the serum hyaluronate concentration, and their linear combinations was assessed with receiver operating characteristic (ROC) curves. An ROC curve is a plot of the sensitivity versus 1 − specificity for all possible cutoff values. The most commonly used index of accuracy is the area under the receiving operating characteristic curve (AUC), with values close to 1.0 indicating high diagnostic accuracy. We also provided the standard deviation (SD) of the AUC as a precision index, using equivalence between the Wilcoxon statistic and the area under the curve.15 Optimal linear combinations of biomarkers were investigated for improving the diagnosis.16
Optimal cutoff values for DWMRI were chosen to maximize the sum of the sensitivity and specificity, and the positive and negative predictive values were computed for these cutoff values.
All statistical analyses were performed with the R software (cran.r-project.org).
The characteristics of our 54 patients with chronic HCV infection and fitting inclusion criteria are summarized in Table 1. The fibrosis grade distribution was as follows: F0, n = 1; F1, n = 30; F2, n = 8; F3, n = 5; and F4, n = 10. Severe steatosis was found in 9 patients.
|Characteristic||n = 54|
|Sex (men)||37 (60.6%)|
|Age (years)||46.4 ± 12|
|Body mass index (kg/m2)||24.7 ± 3.7|
|Histologic activity (METAVIR)|
|Fibrosis score (METAVIR)|
Evaluation of Fibrosis by DWMRI and METAVIR Scores.
In the volunteer group, the ADC values ranged from 1. 26 to 1.64 × 10−3 mm2/s (mean: 1.44 ± 0.02 × 10−3 mm2/s). The mean coefficient of variation was 2.4%, with a 95% confidence interval of [1.8-2.8]. The interreader and intrareader correlation coefficients were estimated to be 0.80 (range: 0.54-0.89) and 0.75 (range: 0.40-089), respectively. The ADC values of HCV patients ranged from 0.81 to 1.64 × 10−3 mm2/s, with a mean individual coefficient of variation of 2.5% and a 95% confidence interval of [1.8-3.6] (Fig. 1A). The ADC decreased as the fibrosis score increased (JT test, P < 0.0001). Patients with moderate-to-severe fibrosis (F2-F3-F4) had hepatic ADC values lower than patients with or without mild fibrosis (F0-F1; mean: 1.10 ± 0.11 × 10−3 versus 1.30 ± 0.12 × 10−3 mm2/s, P < 0.0001) and healthy volunteers (P < 0.0001). Figure 2 shows the diagnostic value (ROC curves) of ADC values for stages F3-F4 (Fig. 2A) and for stages F2-F3-F4 (Fig. 2B). The corresponding AUCs are shown in Table 2. For an ADC value cutoff level of 1.21 × 10−3 mm2/s for stages F3-F4, the sensitivity, specificity, positive predictive value, and negative predictive value were 87%, 87%, 72%, and 94% respectively. The ADC values decreased with the degree of necroinflammation (A0-A1, 1.3 ± 0.1 × 10−3 mm2/s, versus A2-A3, 1.2 ± 0.1 × 10−3 mm2/s, P < 0.01). The difference in ADC according to the fibrosis score (F0-F1 versus F2-F3-F4) remained after standardization on the degree of necroinflammation (P < 0.005), and there was no modification effect through inflammation between ADC and fibrosis (test for the interaction term: P = 0.63). Images from patients with METAVIR score of F1 and F4 are shown in Fig. 3.
|AUC||METAVIR F3-F4||METAVIR F2-F3-F4|
|ADC (10−3 mm2/s)||0.92 ± 0.04||0.79 ± 0.07|
|Transient elastography (kPa)||0.92 ± 0.05||0.87 ± 0.05|
|FibroTest||0.79 ± 0.08||0.68 ± 0.09|
|APRI||0.87 ± 0.06||0.81 ± 0.06|
|Forns index||0.86 ± 0.06||0.72 ± 0.08|
|Hyaluronate||0.87 ± 0.06||0.77 ± 0.06|
Comparison of DWMRI with Transient Elastography and Serum Markers.
Figure 1B shows box plots of transient elastography with values ranging from 2.8-45 kPa. Transient elastography increased with increasing fibrosis score (JT test, P < 0.0001). The AUC for stages F3-F4 (Fig. 2A) were 0.92 (±0.04) for MRI, 0.92 (±0.05) for elastography, 0.79 (±0.08) for FibroTest, 0.87 (±0.06) for APRI, 0.86 (±0.06) for the Forns index, and 0.87 (±0.06) for hyaluronate (Table 2). The AUC curve for stages F2-F3-F4 (Fig. 2B) were 0.79 (±0.07) for MRI, 0.87 (±0.05) for elastography, 0.68 (±0.09) for FibroTest, 0.81 (±0.06) for APRI, 0.72 (±0.08) for the Forns index, and 0.77 (±0.06) for hyaluronate (Table 2). The AUC values were not statistically different between these noninvasive tests.
Combinations of DWMRI, Transient Elastography, and Serum Markers.
We evaluated linear combinations of ADC values, log(transient elastography), and APRI to improve the diagnosis of the fibrosis stage. The best diagnostic value was obtained with a combination of the ADC value and transient elastography [1.3 log(transient elastography) − 2.3 ADC], which yielded only a slight improvement in the AUC: 0.88 (±0.05) for stages F2-F3-F4.
For discriminating F0-F1 versus F2-F3-F4, the best cut points were less than 1.24 for ADC and greater than 8.7 for transient elastography. DWMRI and transient elastography agreed on the diagnosis of F0-F1 versus F2-F3-F4 in 40 patients (74%). For the 14 patients about whom they disagreed, DWMRI agreed with liver specimen biopsy examination results in 8 cases (6 F0-F1 and 2 F2-F3-F4), and transient elastography agreed with liver examination results in 6 cases (3 F0-F1 and 3 F2-F3-F4). For the diagnosis of F ≥ 2, liver examination confirmed DWMRI in 43/54 cases (79.6%), transient elastography in 41/54 cases (76%), and concordant MRI elastography results in 32/40 cases (80%).
For discriminating F0-F1-F2 versus F3-F4, the best cut points were less than 1.21 for ADC and greater than 12.9 for transient elastography. DWMRI and transient elastography agreed on the diagnosis of F0-F1-F2 versus F3-F4 in 44 patients (81.5%). For the 10 patients about whom they disagreed, DWMRI agreed with liver biopsy examination in 5 cases (4 F0-F1-F2 and 1 F3-F4), and transient elastography agreed with liver examination in 5 cases (4 F0-F1-F2 and 1 F3-F4). For the diagnosis of F ≥ 3, liver examination confirmed DWMRI in 47/54 cases (87%), transient elastography in 48/54 cases (89%), and concordant DWMRI elastography results in 41/44 cases (93%).
We prospectively and blindly assessed the performance of DWMRI for the evaluation of the severity of liver fibrosis in patients with a chronic HCV infection. The diagnostic performance of DWMRI for liver fibrosis was found to be good and at least similar to that shown by transient elastography, APRI, the Forns index, FibroTest, and hyaluronate as reported in previous studies. The combination of ADC and transient elastography resulted in the best diagnostic performance for significant fibrosis (F ≥ 2).
The development of noninvasive imaging methods such as DWMRI for the assessment of liver tissue damage represents a major advance in the management of liver diseases. Cross-sectional imaging findings of cirrhosis are generally capable of detecting advanced diseases on the basis of signs of portal hypertension with good sensitivity and specificity.17 However, these findings have a limited usefulness for the detection of fibrosis. Diffusion-weighted imaging, which is widely used in brain imaging for the evaluation of acute ischemic stroke, has become possible in the abdomen with the advent of the echo-planar MRI technique, which allows fast imaging times to be performed, thus minimizing the effect of gross physiologic motion from respiration and cardiac movement.18, 19 Diffusion is the motion of water molecules in biologic tissues, so-called Brownian motion, which is induced by heat.12 Because in vivo random motions not only include water diffusion but also other random microscopic motions, such as microcirculation or perfusion in the capillary network, ADC, rather than the diffusion coefficient, is used to evaluate water diffusion. Collagen fiber is the main component of hepatic fibrosis. The protons contained in this tissue are less abundant than those in water and are tightly bound. Previous studies support the idea that ADC values are lower in a cirrhotic liver than in a normal liver.11–13 However, an overlap has been reported between normal and cirrhotic livers.12, 19 To improve the signal intensity during DWMRI of the liver, we used a parallel imaging technique and combined multiple b values into a single acquisition, using diffusion gradients along 3 directions. This technical optimization of DWMRI allowed us to have additional sample points for generating the isotropic diffusion and ADC maps with a subsequent improvement in the signal intensity. Actually, diffusion tensor imaging and parallel imaging could be used during DWMRI of the liver without compromising the measurement of hepatic ADC.20
A few previous studies have addressed the question of ADC values in the liver. Differences in MR equipment and in echoplanar sequence parameters make it difficult to compare these studies with this one. In this respect, the choice of the b value plays a critical role. This value is a function of the amplitude and duration of the diffusion gradient and of the time allowed for the proton to diffuse between the 2 successive gradient pulses. Therefore, its choice is a compromise between adequate diffusion strength and image quality. According to the signal attenuation equation, Sb = S0 exp(−b × ADC), only 2 measurements are in theory required to evaluate ADC: one in the absence of a diffusion gradient (S0) and one in the presence of a diffusion gradient with a given b value (Sb). However, ADC evaluation is consolidated by the use of several b values, which allow a linear regression analysis to be performed from the derived equation, ln Sb = ln S0 − b × ADC.11, 21, 22
As far as a normal liver is concerned, our findings on volunteers (mean ADC value: 1.44 ± 0.02 × 10−3 mm2/s) are consistent with those reported by Taouli et al.20 and Oner et al.,23 that is, 1.52 ± 0.21 × 10−3 and 1.61 ± 0.45 × 10−3 mm2/s, respectively, with b values of 0 and 500 seconds/mm2. In a noncirrhotic liver, Müller et al.24 reported mean ADC values of 1.39 ± 0.16 × 10−3 mm2/s, using 8 b values from 0-454 seconds/mm2, and Kim et al.21 reported mean ADC values of 3.69 ± 1.84 mm2/s for b < 100 seconds/mm2, 1.20 ± 0.43 × 10−3 mm2/s for b < 410 seconds/mm2, and 1.02 ± 0.25 × 10−3 mm2/s for b < 850 seconds/mm2. In a cirrhotic liver, previous studies suggested ADC values lower than those in a normal liver. Müller et al.24 reported ADC values of 0.9 and 1.20 × 10−3 mm2/s (only 2 patients in the study), and Kim et al.21 reported mean ADC values ranging from 2.71 ± 1.62 × 10−3 mm2/s for b < 100 seconds/mm2 to 1.11 ± 0.48 × 10−3 mm2/s for b < 410 seconds/mm2 to 0.88 ± 0.26 × 10−3 mm2/s for b < 850 seconds/mm2. Taouli et al.12 reported mean ADC values of 1.09 ± 0.46 with b values of 0 and 400 seconds/mm2 and 1.37 ± 0.52 × 10−3 mm2/s with b values of 0 and 500 seconds/mm2. Aubé et al.25 reported a mean value of 2.05 ± 0.79 × 10−3 mm2/s with b values of 0 and 200 seconds/mm2. Only 2 previous studies attempted to evaluate ADC values in hepatic fibrosis. The first one, by Boulanger et al.,19 using 5 b values ranging from 50-250 seconds/mm2, reported a lack of a significant difference between the control subjects and HCV patients. The second one, by Koinuma et al.,13 using b values of 0 and 128 seconds/mm2, found a correlation between the ADC values and fibrosis scores with mean ADC values of 3.45 ± 0.33 × 10−3 mm2/s in a normal liver, 2.45 ± 0.25 × 10−3 mm2/s in chronic hepatitis, and 1.98 ± 0.32 × 10−3 mm2/s in cirrhosis. However, both studies involved relatively small b values, an experimental condition under which ADC values are likely to be overestimated because the diffusion signal intensity is contaminated by perfusion.21, 22
Consistent with the study by Koinuma et al.,13 our results showed that the ADC values decreased as the fibrosis scores increased, and the difference between patients without or with mild fibrosis (F0-F1) and patients with significant fibrosis (F2-F3-F4) was statistically significant. Interestingly, the ADC values of the 6 patients with missing elastography data were consistent with this analysis: 3 F1 patients with a mean ADC value of 1.33 ± 0.10 × 10−3 mm2/s, 2 F2 patients with a mean ADC value of 1.25 ± 0.08 × 10−3 mm2/s, and 1 F3 patient with an ADC value of 1.20 ± 0.11 × 10−3 mm2/s. On the other hand, the fact that the mean ADC value was significantly higher in the volunteer group than that in the F2-F3-F4 patients may support a potential application of DWMRI as a detection tool. This point would deserve a specific study. We also found a significant relationship between the ADC values and necroinflammation scores. Besides fibrosis, it seems that ADC values might also reflect the intensity of inflammation or necrosis and decrease with the alteration of the tissue structure. On the other hand, steatosis could also affect the ADC value. However, only 9 of our patients had severe steatosis, and further studies are required to address this question. The disagreements sometimes found between the DWMRI and liver biopsy results for fibrosis assessment could be explained by biopsy sampling errors. As a matter of fact, Bedossa et al.2 reported that fibrosis was correctly categorized in only 75% of 25-mm-long biopsy specimens and in only 65% of 15-mm-long biopsy specimens. In addition, the quantification of liver fibrosis using DWMRI provides an evaluation of the whole liver, enabling a more accurate assessment of disease severity in cases in which the fibrosis is not uniform across the entire organ.
The diagnostic performance of DWMRI is at least similar to that of the other noninvasive tests. Interestingly, the combination of DWMRI with the other noninvasive tests results in a slight improvement of AUC values for stages F2-F3-F4. Compared with the diagnostic performances of transient elastography, FibroTest, APRI, the Forns index, and serum hyaluronate reported in the literature, the diagnostic performance of AUC values was at least similar.3–5, 26 Furthermore, the diagnostic performances of all noninvasive tests were better for patients with severe fibrosis (F ≥ 3) than for patients with significant fibrosis (F ≥ 2). In terms of practical care, one may prefer a method that is sensitive for F3-F4 patients because these must be offered treatment. While the issue is debated, F2 patients may also benefit and should therefore be detected. Figure 2 shows that DWMRI and transient elastography have much the same sensitivity and specificity characteristics for F0-F1-F2 classification versus F3-F4 classification, whereas DWMRI seems to be somewhat less sensitive than transient elastography with respect to F0-F1 classification versus F2-F3-F4 classification. However, these interpretations remain speculative because of the limited number of patients involved in the study.
It is known that limitations of serum surrogate markers to determine the severity of fibrosis are acute inflammation, hemolysis, Gilbert syndrome, cholestasis, and renal failure.3–5 Limitations of transient elastography should also be mentioned and concern obese patients and patients with ascites, even if clinically undetected. MRI also has classical contraindications (pacemakers and claustrophobia) but could be useful when other tools have failed. In addition, DWMRI benefits from the intrinsic advantages of MRI, such as the ability to perform conventional liver MRI at the same time to detect not only fibrosis but also hepatocellular carcinoma and signs of portal hypertension. In addition, DWMRI could potentially be used to follow treated liver fibrosis. However, the cost and availability of MRI are the main limitations of the technique.
There is a combination of new imaging techniques to further improve the diagnostic efficacy of DWMRI. For example, proton spectroscopy at 1.5 T identifies various metabolites (choline and aliphatic hydrocarbons among others) that may be altered in both steatosis and fibrosis.27 Perfusion MRI can detect the early alterations in the vascular architecture of fibrosis and cirrhosis.28, 29 Recently, the synergistic effects of gadolinium chelate and superparamagnetic iron oxide contrast agents in MRI for the hyperintense reticulation detection of fibrosis have achieved an accuracy greater than 90% for the diagnosis of fibrosis.30, 31 Another promising method is MR elastography, which measures the viscoelastic parameters of the liver according to the stage of fibrosis.32, 33 All these imaging-related technologies require further research to be incorporated into validated management algorithms.34
In conclusion, the performance of DWMRI compares favorably with other noninvasive tests for the presence of significant liver fibrosis in patients with chronic hepatitis C infection. We suggest that DWMRI could be used as an adjunct to routine imaging to assess the degree of liver fibrosis in patients for whom liver MRI is performed.
- 14Nonparametric Statistical Methods. 2nd ed. New York: Wiley; 1999., .
- 34Diffusion-weighted imaging of the liver: technical challenges and prospects for the future. Magn Reson Med Sci 2005; 31: 175–186., , , , , , et al.