麻豆传媒

AMSTERDAM -- Although biopsy remains the gold standard for diagnosing liver fibrosis, imaging tests increasingly appear to be a viable way to garner equivalent information with less patient discomfort and risk, researchers said here.

In presentations at the meeting of the European Association for the Study of the Liver, scientists from across Europe reported on the strengths and weaknesses of various imaging modalities as tools for routine clinical practice.

There was no clear winner among transient elastography, magnetic resonance elastography (MRE), real-time shear wave elastography (RTSWE), and acoustic radiation force impulse (ARFI) imaging, but all appeared to be nearly as accurate as liver biopsy in quantitative assessment of fibrosis and for predicting outcomes such as death and cirrhotic decompensation.

The role of liver imaging for these purposes in the U.S. has recently come to the fore with the FDA's clearance last week of the Fibroscan transient elastography device. Fibroscan is the established leader in noninvasive fibrosis imaging and, according to its French manufacturer, Echosens, the U.S. is the last major market to approve its device.

All these forms of elastography work by setting up shear waves in the liver. Patterns of propagation of these waves correspond to the degree of liver stiffness, which in turn correlates with the level of fibrosis. All but MRE use ultrasound to generate the waves.

Studies presented here evaluated one or more of these technologies against another, with or without liver biopsy as a reference standard, and in a variety of patient populations.

Transient Elastography Versus Biopsy

Perhaps the most direct assessment was reported by Juan Macias, MD, of Hospital Universitario de Valme in Seville, Spain. He reported a retrospective analysis of 297 patients coinfected with HIV and hepatitis C virus (HCV) who had been tested with liver biopsy as well as transient elastography, with these tests performed within a year of each other. The study period covered 2005 to 2011.

Findings indicated that fibrosis stage as established from biopsies and liver stiffness measurements from transient elastography were equally accurate in predicting overall mortality and decompensation of cirrhosis.

Kaplan-Meier curves for patients with stage F4 fibrosis (overt cirrhosis) and for those with elastography measurements in the highest quintile (21 kPa and above) were nearly identical through up to 6 years of follow-up, for both all-cause death and for decompensation of cirrhosis, Macias reported.

Point estimates of the increased risk for these outcomes were somewhat higher in models based on biopsy findings than in the elastography-based analyses, but the error bars in the latter were markedly smaller.

For example, the risk of decompensation doubled with each increase in fibrosis stage (hazard ratio 2.00, 95% CI 1.32 to 3.00), whereas each 5-kPa increase in liver stiffness corresponded to a hazard ratio of 1.42 (95% CI 1.31 to 1.55).

"The noninvasive nature of [transient elastography] should favor its use instead of liver biopsy when the only issue is predicting the clinical outcome of liver disease in HIV-HCV coinfection," Macias told attendees.

ARFI Versus Transient Elastography

Acoustic radiation force impulse imaging is another up-and-coming imaging method for liver disease. Like transient elastography, it uses ultrasound to generate mechanical waves within the liver, but the nature of the waves and the interpretation of the resultant patterns differs.

Derek Bardou of CHU Angers in Angers, France, noted that the two technologies have been compared head-to-head in previous studies, with pooled data suggesting that ARFI is less accurate.

But transient elastography has a significant drawback -- it doesn't work on obese patients. Bardou pointed out that the previous analyses were all conducted on a per-protocol basis, such that patients for whom the transient elastography attempt failed to yield usable results were excluded.

He argued that a more stringent "intent-to-diagnose" analysis would be a better reflection of the utility of the two methods in routine practice.

From 2009 to early 2013, he and his colleagues used both methods on a total of 267 patients with chronic, noncancerous liver disease (patients with cirrhotic complications or sepsis were excluded) who also underwent liver biopsies. Areas under the receiver-operating characteristic (AUROC) curves for classifying patients' liver disease stage were calculated for both test types, with biopsy results serving as the reference standard.

The researchers found that, on a per-protocol basis, AUROC values with ARFI were indeed lower -- indicating poorer accuracy -- than those seen with transient elastography. In this analysis, Bardou and colleagues excluded 6.7% of patients in whom transient elastography could not be performed. ARFI failed in fewer than 1%.

But in the intent-to-diagnosis analysis involving all 267 patients, there was no significant difference in AUROC values for the two methods.

Bardou added that whole-liver results with ARFI were more accurate than findings only in the right lobe, the "classical" way to perform ARFI, he explained.

RTSWE Versus Transient Elastography Versus Biopsy

Another study reported here sought to validate real-time shear wave elastography as an alternative -- not necessarily superior -- to liver biopsy.

Giovanna Ferraioli, MD, of Italy's University of Pavia, presented findings from 88 patients with chronic liver disease of varied origin and 33 healthy controls.

Patients underwent both RTSWE (using the ElastPQ system) and transient elastography as well as biopsy. The controls had only the noninvasive testing.

RTSWE, in this study, involved a fixed "sample box" located a maximum of 70 mm below the Glisson's capsule within the liver. Patients held their breath for 2 to 4 seconds and 10 images were collected, with the median stiffness value in kPa used as the final result. As the name suggests, and unlike transient elastography, RTSWE delivers readings almost immediately. In some studies, it has appeared to be more accurate as well.

Both imaging methods showed stiffness values that progressed upward with the degree of fibrosis ascertained with the biopsies. RTSWE yielded somewhat more detail, in that the median values for each patient group stratified according to fibrosis stage (F0/1 to F4) tracked steadily higher. Transient elastography results for patients with F2 fibrosis, on the other hand, were nearly identical to those with F0/1 disease (5.45 versus 5.5 kPa).

Ferraioli and colleagues found that, as expected, RTSWE values in the healthy controls were lower than in patients with liver disease (median 3.3 kPa, interquartile range 3.7 to 4.0).

Transient elastography readings tended to be higher (median 3.8 kPa, interquartile range 4.5 to 5.0) and overlapped in the controls with those from patients with liver disease (median in F2 patients 5.45, interquartile range 4.3 to 8.0).

RTSWE "compares favorably" with transient elastography, Ferraioli concluded.

MR Elastography Versus Biopsy

Use of MRI equipment to analyze liver stiffness is an even newer approach. It, too, can be used to generate vibrations that propagate through the liver. Rocio Gallego-Duran, also of the Hospital Universio de Valme, reported on a validation study in which artificial neural networks were used to generate elastography values from MRI scans.

Her study involved 63 patients with biopsy-confirmed non-alcoholic fatty liver disease, including 32 with non-alcoholic steatohepatitis (NASH) and 25 with significant fibrosis.

The first 22 of these patients were used as a "training cohort" for fine-tuning the software settings to match biopsy results as closely as possible. The resulting model was then tested in the remaining 41 patients, serving as a validation cohort.

For diagnosing NASH, the model showed sensitivity of 77% and specificity of 90%, Gallego-Duran reported. Positive and negative predictive values were 89% and 79%, respectively.

The model was not quite as good at diagnosing fibrosis. With the best-performing cutoff values, sensitivity was 87% but specificity was only 63%. As a result, the positive predictive value was just 59%, although the negative predictive value was a respectable 89%.

Gallego-Duran told attendees that the MRI-based technique holds some potential advantages over the ultrasound-based methods. Because it produces high-resolution images of the entire liver, it may provide a fuller picture of liver disease and can also reveal other types of liver injury. Patients' body fat also is not an issue for image quality, as it is for transient elastography, she said.