Functional liver imaging

Bernard Van Beers

Department of Radiology

Beaujon University Hospital Paris Nord

Laboratory of Imaging biomarkers

UMR 1149 Inserm - University Paris Diderot France

Functional imaging

• Morphological imaging

– Based on size, shape and signal intensity

– Often qualitative

• Functional imaging

– Based on function

• Perfusion

• Diffusion

• Elastography

– Often quantitative: imaging biomarkers

• Rationale

– Provide information that can not been easily observed

Biomarkers Definition Working Group, 2001

Imaging biomarkers: RECIST criteria

• RECIST: response evaluation criteria in solid tumors

• Measurement of tumor diameter at CT

– Complete response: disappearance of the lesions

– Objective response: decrease ≥ 30%

– Stable disease

– Progressive disease: increase ≥ 20%

• Used since more than 10 years to assess response to treatment in drug development studies

Therasse P et al. JNCI 2000; 92: 205-216Jain RK et al. J Clin Oncol 2013; 266: 812-821

Limitations of RECIST criteria

• RECIST : semi-quantitative score with arbitrary cutoffs

• Decrease in size is not always observed because tumor tissue may be completely replaced with necrosis or fibrosis, especially when targeted treatments are used

Chun YS et al. JAMA 2009; 302: 2338-2344

Colorectal liver metastases treated with chemotherapy and bevazucimab

Ronot M et al. Oncologist 2014; 19: 394-402

Size criteria in HCC treated with sorafenib

Ronot M et al. Oncologist 2014; 19: 394-402

Size criteria in HCC treated with sorafenib

Limitations of mRECIST/EASL

• 2D measurements in very heterogeneous tumors

March JuneApril

Hanahan D et al. Cell 2011; 144: 646-674

Beyond RECIST

• Functional biomarkers

– FDG PET: metabolism

– Dynamic contrast-enhanced CT/MRI: angiogenesis

– Diffusion MRI: cellularity

– MR elastography: visco-elasticity

Response to treatment: volumetric assessment of ADC and enhancement

Bonekamp S et al: Radiology 2013; 268: 431-439

HCC after TACE

Bonekamp S et al: Radiology 2013; 268: 431-439

Volumetric ADC increase ≥ 25% and portal venous enhancement decrease ≥ 65% 3 – 4 weeks after TACE are better predictors of survival than RECIST, mRECIST and EASL criteria

Improvement of diagnostic performance with functional MRIrelative to RECIST

• Shift from morphological to functional parameters

• Shift from manual one-dimensional to automatic three-dimensional approach• Tumor heterogeneity is taken into account: texture analysis• Reproducibility is improved

• Multiparametric approach

Bonekamp D et al. Eur J Radiol 2014

Early diffusion and perfusion changes after TACE of HCC

Diffusion and perfusion changes are already observed at MR imaging one week after TACE

Boustany G et al. 2015

Pharmacokinetic modeling for liver perfusion

• Enhancement depends on local perfusion/ permeability but also on arterial input function

• Pharmacokinetic modeling improves the assessment of results between patients and in longitudinal studies

• BUT• Long acquisition times: respiratory artefacts• Specific image acquisition scheme that is suboptimal for

morphological analysis

t

utk

pportalpaarterialatissue dueuCkuCktC0

)(

112)()()(

Liver

Parenchyma

Ctissue(t)

vd

k1aCartery(t)

Endothelium

Liver: one-compartment model and

dual input with delays

k1a the hepatic artery inflow

k1p the portal vein inflow

k2 the reflux rate

vd=k2/(k1a+k1p) the distribution volume

Tc=1/k2 the mean transit time

Annet, Radiology 2003

Intravascular

Space

k1p

Cvein(t)

k2

Hepatic perfusion: kinetic model

tv

K

plasma

trans

tissuee

trans

etCKtC

)()(

Extravascular

Extracellular

Space

Ctissue(t)

ve

KtransCplasma(t)

Endothelium Kety : one-compartment model

Ktrans reflects the blood flow and/or

the endothelial permeability

depending on whether the perfusion

is flow- or permeability-limited

Tofts, J Magn Reson Imaging 1999

Leach, Brit J Cancer 2005

Intravascular

Space

Tumor: kinetic model

dC tissue (t) / dt = Ktrans Cplasma - k Ctissue

Perfusion MRI changes after treatment in liver metastases

Improvement of disease free survival in patients with liver colorectal metastases treated with chemotherapy and bevacuzimab when perfusion increase < 40% after one week and perfusion decrease > 40% after 10 weeks

De Bruyne S et al. Br J Cancer 2012

ADC: distinction between benign and malignant lesions

• High ADC in benign lesions with high fluid content such as hemangiomas

• No significant difference in ADC between benign hepatocellular lesions and malignant tumors

Doblas S et al. Invest Radiol 2013;48: 722-728

Garteiser P. et al. Eur Radiol 2012; 22: 2169-2177

Visco-elastic properties

Areas under ROC curves

• AUROCADC = 0.71

• AUROC Gl = 0.76

• AUROC malignancy index = 0.84

Imaging of liver chronic liver disease

• Inflammation and fibrosis: microscopic findings

• Changes

– Stiffness

– Diffusion

– Perfusion and hepatocyte transport

– T1 relaxation time

– Susceptibility

Mechanical waves captured by

sequence sensitive to movement

Wavelength depends on elasticity

Attenuation depends on viscosity

Dynamic elastography

Elastography

• Elastography: most established method to stage liver fibrosis

• Three techniques with increasing diagnostic performance

– Transient elastography

– Shear wave elastography

– MR elastography

TE

SSI

3D MRESandrin L et al. IEEE Trans Ultrason Ferro Freq Contr 2002

Huwart L et al. NMR Biomed 2006

Bavu E. et al. Ultrasound Med Biol 2011

MR elastography

Elastography: reliability and reproducibility

• Reliability

– Transient elastography

• 85% of patients with BMI < 30 kg/m2

• 71% of patients with BMI ≥ 30 kg/m2

– Shear wave elastography

• 90 % of patients with BMI < 30 kg/m2

• 73% of patients with BMI ≥ 30 kg/m2

– MR elastography

• 95%, can be performed in obese patients and patients with ascites

• Reproducibility

– Ultrasound elastography: interobserver agreement: 85%

– MR elastography: better reproducibility than ultrasound elastography

Van Beers BE et al. Semin Liv Dis; in press

Ultrasound elastography: accuracy

• Meta analyses and comparative studies

• Transient elastography

– AUROC 0.84 for ≥ F2

– AUROC 0.94 for F4

• Shear wave elastography

– Two-dimensional shear wave elastography: better results for ≥ F2

Bavu E. et al. Ultrasound Med Biol 2011

F ≥2

MR elastography: accuracy

• MR elastography

– AUROCs > 0.9 for ≥ F2, ≥ F3 and F4

– Higher accuracy than transient and shear wave elastography

Huwart et al. Gastroenterology 2008Cui et al. Hepatology 2016Imajo K et al. Gastroenterology 2016Joon JH et al. Radiology 2014

Prognostic biomarker and cost-effectiveness

• Patients with high baseline stiffness values but also patients with increasing values during follow-up have impaired prognosis and survival

• Ultrasound elastography is cost-effective in patients with chronic HCV and HBV infection and NAFLD to diagnose advanced fibrosis and cirrhosis

• MR elastography is more cost-effective than liver biopsy to confirm advanced fibrosis

Corpechot C et al. Gastroenterology 2014Tapper EB et al. Am J Gastroenterol 2015Zhang E et al. Eur Radiol 2015

F1

F4

Potential indications for MR elastography

• Discordant results between ultrasound elastography and serum biomarkers

• Follow-up after treatment

• Assessment of cirrhosis severity and portal hypertension

• Diagnosis of non-alcoholic steatohepatitis: NASH

F1 F3

Loomba et al. Hepatology 2015; 61: 1239-1250

Cirrhosis and portal hypertension

• Portal hypertension: HPVG, endoscopy

• Elastography is an non-invasive alternative to HVPG measurements

• Feasibility of ultrasound elastography often limited, especially > 10 mm Hg

• MR elastography is promising

Garcia-Tsao G et al. Hepatology 2010Ronot M et al. Eur Radiol 2012Procopet B et al. J Hepatol 2015

MR elastography of portal hypertension

0 20 40 60 80 1000

20

40

60

80

100

Gl 84 Hz

Gd 84 Hz

Gl 56 Hz

Gd 56 Hz

G* 56 Hz

G* 84 Hz

1 - Specificity %

Sen

sitiv

ity %

Ronot M et al. Eur Radiol 2014

Viscosity (Gl) of the spleen is a marker of portal hypertension

Baveno VI consensus workshop on portal hypertension

• Stiffness > 15 kPa: compensated advanced chronic liver disease

• Stiffness > 20 kPa: significant portal hypertension (HVPG ≥ 10 mm Hg)

• Stiffness < 20 kPa associated with platelet count > 150,000 can safely avoid screening of esophageal varices by endoscopy

de Franchis R et al. J Hepatol 2015

Increase in liver stiffness is non specific

• Fibrosis

• Portal hypertension

• Liver congestion

• Bile duct obstruction

• Inflammation

– Acute flares in hepatitis B virus infection

– NASH

de Franchis R et al. J Hepatol 2015

Diagnosis of NASH

• Liver ultrasound showing steatosis

• Elevated aminotransferase levels

• Liver biopsy

– Diagnosis of NASH

– Diagnosis of F3 fibrosis: advanced fibrosis, best prognostic factor

Angulo P et al. Gastroenterology 2015

Alternatives to diagnose liver fibrosis: DW-MRI

• Decrease of ADC with increasing stage of fibrosis

• Diagnostic performance > MRE

• DW-MRI is influenced by steatosis and perfusion changes

Lewin M et al. Hepatology 2007Wang QB et al. hepatology 2012Leitao H et al. Radiology 2016

Gadoxetic acid-enhanced MR imaging

• Hepatocyte uptake through OATP1B1/B3, biliary efflux through MRP2, sinusoidal backflux through MRP3

• In NASH and liver fibrosis : decrease of OATP1B1/B3 et MRP2, increase of MRP3

Van Beers BE et al. J Hepatol 2012Giraudeau C et al. Eur Radiol 2016

Gadoxetic acid-enhanced MR imaging

F0 F2 F4

AUROC > 0.9

• Decrease of signal intensity in NASH and liver fibrosis

• AUROCs of gadoxetic acid-enhanced MRI in fibrosis: 0.80 - 0.85

Feier D et al. Radiology 2013Choi YR et al. Invest Radiol 2013

Gadoxetic acid-enhanced MR imaging

• Texture analysis of signal enhancement during the hepatobiliary phase

• AUROC > 0.9

Leporq B et al. 2016

F0 F2 F4

Dynamic gadoxetic acid-enhanced MRI

Lagadec M. et al. Radiology 2015Giraudeau C. et al. Eur Radiol 2016 Leporq B et al. ISMRM 2015

Conclusions

• Functional imaging may add important quantitative information

• Liver tumors

– Predictive of tumor aggressiveness and response to treatment

– Diffusion and perfusion measurements

– Taking whole tumor volume and its heterogeneity into account

– Multiparametric

• Chronic liver disease

– Predictive of disease stage and response to treatment

– Elastography and perfusion/hepatocyte transport measurements