Replacing a crystal ball with a calculator in predicting liverdisease ou

Scott L. Frie

Division of Liver Diseases, Icahn School of Medicine

493

ness with the risk of decompensation. MELD correlates perfectly well with outcomes. Still, the method

gy 2014 vol. 60 j 905906

FocusJournal of HepatoloHospital/University College London builds upon their previouslydescribed method of collagen proportionate area, or CPA, which

determining risk of decompensation, as reported by others aswell [7]. Because collagen continues to accumulate even whenthe biopsy stage is cirrhosis [8], this and related morphometricmethods can continuously quantify collagen accumulation acrossthe full spectrum of disease with no upper limit.

To avoid the reliance on biopsy, non-invasive methods toassess liver structure or function offer an obvious advantagebecause of their markedly reduced risk and ability to acquire datarepeatedly over time. Among these, both transient- and magnetic

Received 23 January 2014; accepted 23 January 2014qDOI of original article: http://dx.doi.org/10.1016/j.jhep.2013.12.023, http://dx.doi.org/10.1016/j.jhep.2013.12.016. Address: Division of Liver Diseases, Box 1123, Mount Sinai School of Medicine,1425 Madison Ave., Room 1170C, New York, NY 10029, United States. Tel.: +1 212659 9501; fax: +1 212 849 2574.E-mail address: scott.friedman@mssm.eduThe study by Tsochatzis and colleagues from the Royal Free reinforces the importance of quantitative collagen content inSee Articles, pages 93

There is growing optimism about the potential to stabilize orreverse advanced liver disease thanks to remarkable progress inthe treatment of chronic viral hepatitis. Striking evidence of cir-rhosis reversibility in patients treated with antiviral therapiesfor HBV [1] or HCV [2] raises the prospect of exploiting pathwaysof brosis regression to treat other chronic liver diseases, in par-ticular alcoholic and non-alcoholic fatty liver diseases. In theselatter illnesses, however, one must rely on evidence of improve-ment in either histology or other indices of liver structure orfunction, rather than the more straightforward prospect of mea-suring viral suppression as a determinant of improvement. Itsworth noting that initially in early antiviral trials, regulatoryagencies still required evidence of histologic improvement inassociation with viral suppression before relying solely on viralendpoints as an index of efcacy.

It is also important to remember that improved histologyalone does not guarantee improved outcomes, and the goal ofany effective therapy is to enhance health and/or survival, notto improve a biopsy score. In that spirit there is intensied inter-est in identifying measures that can accurately predict clinicaloutcomes, whether they require invasive (i.e., biopsy) or non-invasive assessment. In fact, there has been scant evidence todate that standard liver biopsy staging systems correlate rigor-ously with outcomes, in part because traditional scoring systemsrely on semi-quantitative staging rather than exact quanticationof brosis content.

Two complementary studies in this months issue of the Jour-nal address these key questions, the rst by Tsochatzis et al.,which uses a quantitative measure of collagen content in liverbiopsies to establish a correlation with clinical outcomes, andthe second by Asrani et al., which exploits the emerging technol-ogy of magnetic resonance elastography to correlate liver stiff-tcomes

dman

at Mount Sinai, New York, NY, United States

9 and pages 948954

quanties collagen content in liver biopsy specimens, rather thanrelying solely on semi-quantitative scoring systems [3]. In thisnew study, however, the investigators have directly comparedthe accuracy of this method with the scoring systems of eitherLaennec [4], Kumar [5] or Nagula [6] in the same patient cohortto determine their relative value in predicting outcomes among69 patients who were followed for a mean of 65 months follow-ing biopsy. The CPA requires staining of liver biopsy specimenswith sirius red, followed by digital morphometric quanticationof stained tissue, expressed as a percentage of total area [3]. Inaddition to specialized software with a one-time cost of 7800Euros, the CPA requires an additional 10 min per sample of thepathologists (or technicians) time to gather data. In this study,a number of histologic variables were quantied, including septalthickness, nodule size, as well as brosis stage for each of thescoring systems, yet CPA was the only biopsy variable indepen-dently associated with clinical decompensation, and was compa-rable to MELD in this respect. These data reinforce the value ofquantitative assessment of collagen content as a more accuratedeterminant of prognosis than standard staging systems. Inter-estingly, this kind of digital assessment is routinely used in ani-mal models that test anti-brotic drugs, yet clinicalpathologists have been slow to embrace the technology.

Of course, the obvious criticism is that CPA, while more accu-rate than standard staging systems, still requires invasive liverbiopsy via either the trans-jugular or percutaneous route, and,moreover, that the tissue specimen must be sufciently large tominimize sampling variability. Furthermore, many patients withapparent cirrhosis by clinical and or imaging assessment are unli-kely to undergo biopsy if the etiology is already known, and onecould not justify performing a liver biopsy simply to predict prog-nosis by CPA, particularly if the MELD is already abnormal, since

resonance-elastography (MRE) are methods that quantify tissuestiffness as a reection of brosis content and/or inammationand edema. Hundreds of studies have evaluated these technolo-gies, and the United States Food and Drug Administrationrecently approved transient elastography for management ofpatients with chronic viral hepatitis and fatty liver disease, in linewith the technologys earlier widespread acceptance elsewhere inthe world. MRE offers the theoretical advantage over transientelastography of sampling the entire liver rather than a smallregion of interest. In the cross-sectional study of 430 patientsreported by Asrani et al., mean liver stiffness by MRE was signif-icantly higher in those with decompensated liver disease than inthose who were compensated. More importantly, in 167 patientsfrom this study followed prospectively for a mean of 27 months,

Conict of interest

The author is a consultant for several companies in the area ofdiagnostics and therapeutics for fatty liver disease and/or brosis,including Abbvie, Alnylam, Angion, Boehringer Ingelheim, BristolMyers Squibb, Conatus, Echosens, Exalenz, Fibrogen, Genentech-Roche, Gilead, Intercept, Immuron, Intermune, Galectin Thera-peutics, Kinemed, Nimbus, Nitto Denko, Novartis, Sano Aventis,Takeda, Tobira, and Vaccinex.

References

[1] Marcellin P, Gane E, Buti M, Afdhal N, Sievert W, Jacobson IM, et al.

Focusthose patients with a mean value of >5.8 kPa had an almostve-fold hazard of decompensation.

How would one use this information to impact clinical deci-sion-making or improve outcomes? At the least, accurate strati-cation of decompensation risk by MRE could be used to increasethe frequency of clinical follow-up, seek evidence of esophagealvarices, or avoid disease-specic therapies that might provokedecompensation. In addition, MRE could help identify a subgroupof patients at high risk who might benet most from effectiveanti-brotic therapies once they are available.

But the holy grail of non-invasive diagnostics is not simply topredict risk of decompensation. Rather, an ideal diagnostic mustalso correlate with improved outcomes following therapeuticintervention in order to justify its use in place of biopsy in clinicaltrials or practice. In addition to elastography, other methodsundergoing evaluation for this purpose include functional breathor clearance tests [9,10], serum markers [11], hepatic venouspressure gradient measurement [12], Doppler/ultrasound andCT/MRI. Ultimately, a combination of tests may yield the greatestprecision in quantifying risk and documenting response totherapy.

The prospect of effective therapies for both fatty liver disease[13] and hepatic brosis [14] has heightened the sense of urgencyin identifying non-invasive measures that correlate with clinicaloutcomes. The importance of progress in addressing this chal-lenge was evident in the jointly sponsored AASLD-FDA endpointsconference held in September, 2013, to help clarify endpoints forclinical trials that evaluate therapies for NAFLD (http://www.aasld.org/additionalmeetings/Pages/aasldfdanash.aspx). Awritten summary of the meeting recommendations is anticipatedsoon. Together with studies like those of Tsochatzis et al. andAsrani et al., these recommendations will help us replace a crystalball that prophesizes the prognosis of liver diseases with a calcu-lator that can quantify its consequences.906 Journal of Hepatology 201Regression of cirrhosis during treatment with tenofovir disoproxil fumaratefor chronic hepatitis B: a 5-year open-label follow-up study. Lancet2013;381:468475.

[2] DAmbrosio R, Aghemo A, Rumi MG, Ronchi G, Donato MF, Paradis V, et al. Amorphometric and immunohistochemical study to assess the benet of asustained virological response in hepatitis C virus patients with cirrhosis.Hepatology 2012;56:532543.

[3] Manousou P, Dhillon AP, Isgro G, Calvaruso V, Luong TV, Tsochatzis E, et al.Digital image analysis of liver collagen predicts clinical outcome of recurrenthepatitis C virus 1 year after liver transplantation. Liver Transpl2011;17:178188.

[4] Kim MY, Cho MY, Baik SK, Park HJ, Jeon HK, Im CK, et al. Histologicalsubclassication of cirrhosis using the Laennec brosis scoring systemcorrelates with clinical stage and grade of portal hypertension. J Hepatol2011;55:10041009.

[5] Kumar M, Sakhuja P, Kumar A, Manglik N, Choudhury A, Hissar S, et al.Histological subclassication of cirrhosis based on histological-haemody-namic correlation. Aliment Pharmacol Ther 2008;27:771779.

[6] Nagula S, Jain D, Groszmann RJ, Garcia-Tsao G. Histological-hemodynamiccorrelation in cirrhosis-a histological classication of the severity ofcirrhosis. J Hepatol 2006;44:111117.

[7] Sethasine S, Jain D, Groszmann RJ, Garcia-Tsao G. Quantitative histological-hemodynamic correlations in cirrhosis. Hepatology 2012;55:11461153.

[8] Goodman ZD, Becker Jr RL, Pockros PJ, Afdhal NH. Progression of brosis inadvanced chronic hepatitis C: evaluation by morphometric image analysis.Hepatology 2007;45:886894.

[9] Everson GT, Shiffman ML, Hoefs JC, Morgan TR, Sterling RK, Wagner DA, et al.Quantitative liver function tests improve the prediction of clinical outcomesin chronic hepatitis C: results from the hepatitis C antiviral long-termtreatment against cirrhosis trial. Hepatology 2012;55:10191029.

[10] Lalazar G, Pappo O, Hershcovici T, Hadjaj T, Shubi M, Ohana H, et al. Acontinuous 13C methacetin breath test for noninvasive assessment ofintrahepatic inammation and brosis in patients with chronic HCVinfection and normal ALT. J Viral Hepat 2008;15:716728.

[11] Schmeltzer PA, Talwalkar JA. Noninvasive tools to assess hepatic brosis:ready for prime time? Gastroenterol Clin North Am 2011;40:507521.

[12] Ripoll C, Groszmann R, Garcia-Tsao G, Grace N, Burroughs A, Planas R, et al.Hepatic venous pressure gradient predicts clinical decompensation inpatients with compensated cirrhosis. Gastroenterology 2007;133:481488.

[13] Corrado RL, Torres DM, Harrison SA. Review of treatment options fornonalcoholic fatty liver disease. Med Clin North Am 2014;98:5572.

[14] Friedman SL, Sheppard D, Dufeld JS, Violette S. Therapy for broticdiseases: nearing the starting line. Sci Transl Med 2013;5:167sr161.4 vol. 60 j 905906

Replacing a crystal ball with a calculator in predicting liver disease outcomesConflict of interestReferences