Ultrasound in Med. & Biol., Vol. 39, No. 9, pp. 1527–1535, 2013Copyright � 2013 World Federation for Ultrasound in Medicine & Biology

Printed in the USA. All rights reserved0301-5629/$ - see front matter

/j.ultrasmedbio.2013.02.012

http://dx.doi.org/10.1016

d Original Contribution

TISSUE DOPPLER IMAGING AND TISSUE STRAIN IMAGINGFOR THE EVALUATION OF HEPATIC FIBROSIS IN PATIENTS

WITH CHRONIC HEPATITIS B

YING WANG,* ZHIFENG WU,y YAN JU,z LI CAO,x LIN SHI,* FA TONG,* XUAN JIANG,*and CHANGREN ZHU

x

*Department of Ultrasonography, Clinical Medical College of Yangzhou University, Jiangsu, China; yDepartment ofRadiotherapy, Clinical Medical College of Yangzhou University, Jiangsu, China; zDepartment of Echocardiography, ClinicalMedical College of Yangzhou University, Jiangsu, China; and xDepartment of Pathology, Clinical Medical College of Yangzhou

University, Jiangsu, China

(Received 18 September 2012; revised 16 February 2013; in final form 20 February 2013)

AtherapyWest Rcom.cn

Abstract—We studied the feasibility of evaluating the stages of liver fibrosis with tissue Doppler imaging (TDI) andtissue strain imaging (TSI) for patients with chronic hepatitis B virus infection. One hundred ten patients weredivided into two groups: normal adult group (n 5 38) and chronic liver disease group (n 5 72, patients infectedwith HBVs). The chronic liver disease group was divided into three subgroups on the basis of the Scheuer scoringsystem and clinical evidence: mild fibrosis (S0 and S1, n5 11), moderate fibrosis (S2 and S3, n5 27) and cirrhosis(S4 and clinically typical cirrhosis, n5 34) groups. TDIwas performed for a chosen oblique section. Four regions ofinterest (ROIs), A–D, were chosen in the hepatic parenchyma based on the direction of propagation from the heartto the liver. Strain rate curves were obtained on the basis of TDI and TSI findings. Strain peak rates (SPRs) of allROIs and the differences in times to SPRs for the four ROIs (TA-B, TB-C and TC-D) in the hepatic parenchyma weremeasured with TDI and TSI. Strain rate curves were analyzed for each ROI. The strain rate curves for the normaladult group were synchronous, whereas those for the chronic liver disease group were asynchronous. SPRs of theROIs gradually decreased with the progression of liver fibrosis. The SPRs of ROI B significantly correlated withchronic liver disease severity (r5 0.991, p, 0.05). Areas under the curve (AUCs) of the ROI A and ROI B SPRs atthe moderate fibrosis and cirrhosis stages were 0.86 ± 0.06, 0.81 ± 0.56 and 0.90 ± 0.65, 0.92 ± 0.04, respectively.The AUC of the SPRs of ROIs A and B correlated better than the platelet/age/phosphatase/a-fetoprotein/aspartateaminotransferase (PAPAS) index for advanced fibrosis. The differences in time to SPRs among the peaks of thefour ROIs (TA-B, TB-C and TC-D) gradually increased with the progression of liver fibrosis. TDI and TSI withquantitative measurements using tissue Doppler analysis software (TDIQ, GE Medical Systems, Horten,Norway) provided reliable information for evaluating non-invasive liver fibrosis in patients with chronic hepatitisB. (E-mail: wuzhifeng2@yahoo.com.cn) � 2013 World Federation for Ultrasound in Medicine & Biology.

Key Words: Tissue Doppler imaging, Tissue strain imaging, Quantitative assessment, Strain peak rates, Liverfibrosis, Cirrhosis.

INTRODUCTION

Hepatic fibrosis with chronic hepatitis B viruses (HBVs)is associated with a higher risk of hepatocellular carci-noma (Di Bisceglie 2004; Thompson et al. 2008). Patientprognosis strongly depends on histologic severity. Atpresent, liver biopsy is the gold standard for the diagnosisand staging of liver fibrosis (Theise 2007). However, liver

ddress correspondence to: Zhifeng Wu, Department of Radio-, Clinical Medical College of Yangzhou University, 98 Nantongoad, Yangzhou, 225001, China. E-mail: wuzhifeng2@yahoo.

1527

biopsy is an invasive procedure accompanied by the riskof various complications, such as pain, hemorrhage andbile peritonitis; thus, it is not suitable for routine clinicalfollow-up (Al and Shiffman 2007). Furthermore, a liverbiopsy sample represents only a small part of the wholeliver, and sampling bias is possible (Ratziu et al. 2005;Vuppalanchi et al. 2009). Therefore, to assess the severityof liver fibrosis, non-invasive, tolerable and reliablemethods are urgently needed. With recent developmentsin imaging technology, the role of ultrasound in liverdisease has significantly increased.

Nodular hepatic surface, coarse echotexture andsignal attenuation, intrahepatic vessel contours and

1528 Ultrasound in Medicine and Biology Volume 39, Number 9, 2013

spleen length are useful parameters in the diagnosis ofcirrhosis (Oberti et al. 1997; Zheng et al. 2003). However,the diagnostic accuracy of routine ultrasound is lowerbecause digestive gas can interfere with ultrasound inves-tigation (Aube et al. 1999). Through the use of Dopplerultrasound signs, including the new hepatic vein spec-trum, approximately 90% of patients with compensatedchronic liver disease can be correctly diagnosed (Aubeet al. 2004). Doppler parameters might not be suitablefor assessment of the stage of liver fibrosis by clinicians(Bernatik et al. 2002). Transient elastography showspromise in diagnosing cirrhosis and assessing diseasestage (Stebbing et al. 2010); however, transient elastogra-phy images are invisible during the measurement stage.With the development of ultrasound contrast agents andspecific imaging contrast techniques, contrast-enhancedultrasound has been reported to perform well and hasgreat potential in the evaluation of liver fibrosis (Liuand Lu 2010; Li et al. 2010); however, the drawbacksof contrast-enhanced ultrasound are that it requirescontrast agent injections and depends on operator skilland access to the relevant technology (Bonekamp et al.2009). Acoustic radiation force impulse imaging hasbeen reported to perform better diagnostically in the eval-uation of chronic liver disease severity (Kim et al. 2010;Boursier et al. 2010). Nevertheless, it is not adequate forestimating the early stages of fibrosis (Haque et al. 2010).Tissue strain imaging (TSI) is based on tissue Dopplerimaging (TDI), a myocardial examination technique. Arecent study using TSI found significant differences instrain values between normal adults and adults withchronic hepatitis and cirrhosis (Hotta et al. 2007).However, the authors did not compare their imagingwork with liver biopsy results to study differencesbetween the stages of liver fibrosis. Moreover, the posi-tions of the regions of interest (ROIs) for each groupwere unclear. Supersonic shear imaging, used asa complementary tool for B-mode ultrasound, couldcomplement morphologic information for fibrosis stagingand hepatic lesion imaging (Muller et al. 2009; Bavu et al.2011). Studies on the evaluation of fibrosis in chronichepatitis C or non-alcoholic fatty liver diseases areavailable.

The purpose of this study was to assess the value ofusing TDI and TSI to evaluate hepatic fibrosis stages inpatients with HBVs.

METHODS

PatientsThis study included 41 healthy volunteers without

hepatic disease history and with normal ultrasound find-ings (normal adult group) and 74 patients patients in-fected with HBVs (chronic liver disease group) who

had been admitted to the Clinical Medical College ofYangzhou University between February 2010 and April2012. On the basis of the Scheuer scoring system for liverfibrosis staging (S0, S1, S2, S3 and S4) (Scheuer 1991,1995) and clinical evidence, the chronic liver diseasegroup was categorized into three subgroups: mildfibrosis (S0 and S1), moderate fibrosis (S2 and S3) andcirrhosis (S4, liver biopsy and clinically typicalcirrhosis) groups. Forty-nine patients in the chronic liverdisease group underwent liver biopsy or surgery, andtwenty-five patients with liver cirrhosis were diagnosedon the basis of clinical criteria. The clinical criteria forliver cirrhosis were as follows: (i) previous episodes ofcirrhotic decompensation (encephalopathy, icterus, vari-ceal bleeding or ascites formation); (ii) morphologicsigns of liver cirrhosis on ultrasound examination(Aube et al. 1999, 2004); and (iii) endoscopic signs ofportal hypertension (Wiechowska-Kozlowska et al.2008). Each subject who participated in the studyprovided informed consent. The study protocol con-formed to the Declaration of Helsinki and was approvedby our institutional ethics committee. The clinical andlaboratory assessment findings are summarized inTable 1.

Tissue Doppler imaging and tissue strain imagingexamination

All TDI and TSI examinations were performed bya sonographer blind to the patients’ clinical data andrelated information, using a GE Vivid E9 ultrasoundmachine (GE Medical System, Horten, Norway) witha 2.5-MHz transducer. An oblique anterior approachalong the costal margin was employed with a 2.5-MHztransducer on the abdomen at the subxiphoid deviationto the right, and the second hepatic portal (confluenceof the hepatic veins) was visualized. Sector scans wereused when the confluence of the hepatic veins started todisappear; the right cardiac atrioventricular junctionwas visualized in the plane of the oblique quadra-cavity. The angle between the transducer and the bodywas approximately 45� (Fig. 1). TDI and TSI were per-formed within 3 d before liver operations or liver biop-sies, which were guided by sonography and employedan 18-gauge needle. To reduce measurement error anddetermine examiner variability, every recording wasrepeated three times. During TDI and TSI examination,the patients were asked to hold their breath as long aspossible, and dynamic images lasting longer than 30seconds were stored digitally by the ultrasound systemin AVI format and written into computer discs.

Analysis of tissue Doppler and tissue strain imagesThe strain rate from the ROI placed on the right

atrioventricular junction included the three peak

Table 1. Clinical and laboratory data (n 5 115)

Datum Normal adult group (n 5 41)

Chronic liver disease group

Mild fibrosis (n 5 11) Moderate fibrosis (n 5 28) Cirrhosis (n 5 35)

Age (y) 49.14 6 2.80 40.00 6 10.02 47.79 6 2.05 53.10 6 1.61Sex (male/female) (29/12) (7/4) (19/9) (26/9)Alkaline phosphatase (U/L) 74.07 6 30.22 92.64 6 15.71 93.36 6 11.24 117.17 6 33.85Platelet count (3109/L) 190.02 6 40.64 171.45 6 39.64 149.68 6 38.54 74.69 6 14.97Aspartate aminotransferase (U/L) 26.83 6 8.28 69.73 6 18.87 88.14 6 20.10 64.97 6 23.08a-Fetoprotein (ng/mL) 1.15 6 0.56 1.64 6 0.89 2.99 6 1.77 4.67 6 2.37

Mild fibrosis 5 S0 and S1; moderate fibrosis 5 S2 and S3; cirrhosis 5 S4 and clinically typical cirrhosis.Values are expressed as the mean 6 standard deviation.

Fig. 1. Chosen oblique plane underlying B-mode sonography.Four regions of interest (ROIs), A–D, were chosen, and the firstROI (A) was placed on the liver edge perpendicular to the atrio-ventricular junction; the other three ROIs (B–D) were placed

end to end, forming a straight line in the same manner.

Tissue Doppler and tissue strain imaging d Y. WANG et al. 1529

curves: one negative strain rate and two positive strainrates (one during systole and one during diastole). Thefirst positive strain rate, referred to as the strain peakrate (SPR), which was higher than the second positivestrain rate, was selected to compare differencesbetween the groups. The strain rate (SR) was the rateat which the deformation occurred (deformation pertime unit). Higher SR values (more deformation pertime unit) indicated better flexibility. In a previousstudy (Rosen et al. 2009), the time to SPR in differentregions was used to evaluate the extent of myocardialdyssynchrony. Times to SPR for different regions thatwere the same or only slightly different were definedas synchronous. Dyssynchrony may be the result ofliver fibrosis.

Another sonographer who was blind to the patients’clinical data and who did not conduct the TDI and TSIexaminations analyzed all TDI and TSI data offline.Each dynamic plane included four ROIs in the hepaticparenchyma based on the direction of propagationfrom the heart to the liver. Four ROIs (A–D) with diam-eters of 6 mm were chosen, and the first ROI (A) wasplaced on the liver edge perpendicular to the atrioven-tricular junction; the other three ROIs (B–D) wereplaced end to end, forming a straight line in the samemanner (Fig. 1). All ROIs on the liver edge were placedat approximately the same depth in all patients. Thestain rate curve of each ROI was obtained, and eachROI provided a value with which to assess the flexibilityof the liver tissues, which was verified and automati-cally calculated using Tissue Doppler Imaging Quantifi-cation software (TDIQ, GE Medical Systems, Horten,Norway). Tissue strain images were acquired in theTDI mode; the stain rate curves of the ROIs wereanalyzed, and the SPRs of the ROIs were measuredbased on liver movement imaging curves (TDI andTSI) using TDIQ. Strain rates were calculated ass21 5 (V2 2 V1)/d$cos q, where V2 2 V1 5 differencein velocities; d 5 distance between the two points of thevelocity measurement; and cos q 5 cosine of the anglebetween the ultrasound beam and the direction of move-

ment (Pavlopoulos and Nihoyannopoulos 2008; Dandeland Hetzer 2009). In practice, the ROI was an areadefined manually by the operator as an average of thestrain rate measurements, which provided a spatialaverage of the computed values (Pavlopoulos andNihoyannopoulos 2008). TDIQ was installed in a GEVivid E9, which was easy to operate and automaticallyprovided a series of quantitative parameters. The SPRsof the four ROIs for the different groups were recordedand compared. The differences among the time to SPRsof the four ROIs were denoted as TA-B, TB-C and TC-D,respectively. TA-B, TB-C and TC-D were measured usingTDIQ, which analyzed the patterns among the differentgroups. Moreover, to reduce measurement error, everyrecording was repeated three times.

Histologic staging of chronic liver diseaseUltrasound-guided percutaneous liver biopsy was

carried out using a suction technique with an 18-gaugeneedle. Liver biopsy samples longer than 12 mm were

1530 Ultrasound in Medicine and Biology Volume 39, Number 9, 2013

considered adequate (Pagliaro et al. 1983). Liver tissuesamples obtained by nodulectomy or liver biopsy werefixed in formalin and embedded in paraffin. Slices (4mm thick) were prepared with hematoxylin and eosinstaining and assessed under an optical microscope bytwo pathologists blind to the patients’ clinical data. Liverbiopsy specimens were analyzed and scored according tothe Scheuer scoring system (Scheuer 1991, 1995).

Clinical and laboratory assessmentsRelevant clinical data (Table 1) were obtained less

than 3 d before liver operation or biopsy. TDI and TSIresults for inpatients with biochemical data werecompared with other prediction scores. The platelet/age/phosphatase/a-fetoprotein/aspartate aminotrans-ferase index (PAPAS index)(Seto et al. 2011) was basedon five common clinical parameters: age, alkaline phos-phatase (ALP) level, aspartate aminotransferase (AST)level, a-fetoprotein (AFP) level and platelet count. Allfive parameters reported in previous studies were associ-ated with significant fibrosis in patients with chronichepatitis B (Fung et al. 2008). The PAPAS index wascalculated as

Log ðindex1 1Þ5 0:02551 0:00313 age ðyÞ1 0:14833 log ðALP ½U=L�Þ1 0:0043 log ðAST ½U=L�Þ1 0:09083 log ðAFP ½ng=mL�11Þ2 0:0283 log ðplatelet count ½109=L�Þ

Statistical analysisThe data are expressed as means 6 standard devia-

tions. The continuous variables appeared to be normallydistributed. Differences between the normal adult andchronic liver disease groups and differences among thethree subgroups (mild fibrosis, moderate fibrosis andcirrhosis) were analyzed using Student’s t-test. Receiveroperating characteristic (ROC) curves were constructed,and the area under the curve (AUC) was calculated usingHanley and McNeil’s (1982) non-parametric method.The SPRs and PAPAS index for chronic liver diseasewere selected to determine diagnostic accuracy. Correla-tions between the fibrosis and SPR groups were assessedwith Spearman rank correlation coefficients. Statisticalanalysis was performed with SPSS 17.0 for Windows(SPSS, Chicago, IL, USA). A p value, 0.05 was consid-ered statistically significant.

RESULTS

Three healthy volunteers and two patients in thechronic liver disease group were excluded from the study;their TDI and TSI results could not be interpreted becauseof fatty liver. Another 110 patients were examined via

TDI and TSI, and the strain rate curves for all ROIs inthe liver parenchyma were effectively obtained. TheSPRs for each ROI in the participants in the normal adultand chronic liver disease groups were acquired, and thedifferences in time to SPRs of the four ROIs weremeasured.

Characteristics of strain rate curvesThe strain rate curves of all ROIs (A–D) in the

normal adult group were synchronous (identical motiontendency, Fig. 2), and the time to SPRs of the four ROIswere the same or slightly different. However, the stainrate curves in the chronic liver disease group were asyn-chronous, and the time to SPRs of the four ROIs obvi-ously differed (Fig. 2). Although TA-B, TB-C and TC-Ddid not significantly differ between the normal/mildfibrosis group and moderate fibrosis/cirrhosis group (p. 0.05), these values gradually increased with theprogression of liver fibrosis (Fig. 3, Table 2). The differ-ences in the SPRs of the four ROIs between the normaladult group and chronic liver disease group are illustratedin Figure 2, and the changes in TA-B, TB-C and TC-D withprogression of liver fibrosis are illustrated in Figure 3.

Correlations between SPRs of TSI and stages of liverfibrosis

The TDI and TSI quantitative parameters in thedifferent groups are summarized in Table 2. The SPRsof ROIs C and D did not significantly differ among thedifferent groups (p . 0.05); the SPRs of ROI A did notsignificantly differ between the normal adult and mildfibrosis groups (p . 0.05). However, the SPRs for ROIB in the normal adult and chronic liver disease groups(mild fibrosis, moderate fibrosis and cirrhosis) decreasedstepwise with increasing chronic liver disease severity (p, 0.05) (Fig. 2, Table 2). The SPRs of ROI B correlatedwith chronic liver disease severity (r5 0.991, p, 0.05).The SPRs of ROI Avaried according to stage between themild and moderate groups and between the moderate andcirrhosis groups (p , 0.05). SPRs of the ROIs for thechronic liver disease group are illustrated in Figure 2and listed in Table 2.

Comparison of the ROC curves of the PAPAS index andSPRs of ROIs A and B

The PAPAS index can predict and exclude signifi-cant fibrosis, which is reportedly useful in the diagnosis

Fig. 2. Off-line measurements of real-time imaging were analyzed by TDI-Q. The SPRs of ROIs A, B, C and D are shownfor the normal group (a), mild fibrosis group (b), moderate fibrosis group (c) and cirrhosis group (d). The values of 1, 2, 3

and 4 represent the SPRs of ROIs A, B, C, and D, respectively.

Tissue Doppler and tissue strain imaging d Y. WANG et al. 1531

of advanced fibrosis with HBV infection (Seto et al.2011). In our study, the SPRs of ROIs A and B werethe most useful for differentiating advanced fibrosis(Fig. 4). Additionally, the SPRs of ROI B were the mostuseful for differentiating stages of liver fibrosis, includingthe early stage. In Figure 5 are the ROC curves of thePAPAS index and the SPRs of ROIs A and B at the

moderate fibrosis and cirrhosis stages. The area underthe ROC curves for the PAPAS index and the SPRs ofROIs A and B were 0.73 6 0.09, 0.86 6 0.06 and 0.906 0.07, respectively, at the moderate fibrosis stage and0.80 6 0.06, 0.81 6 0.06 and 0.92 6 0.04, respectively,at the cirrhosis stage. The areas under the ROC curvefor the SPRs of ROIs A and B, especially ROI B, were

Fig. 3. Change in time to strain peak rate (SPR) with progression of liver fibrosis. TA-B , TB-C and TC-D increased graduallywith the progression of liver fibrosis. Points represent differences in time to SPRs of the four ROIs in the different groups.

Lines through the points represent standard deviations.

1532 Ultrasound in Medicine and Biology Volume 39, Number 9, 2013

larger than that for the PAPAS index. The cutoff value,sensitivity and specificity of the PAPAS index and theSPRs of ROIs A and B in predicting liver fibrosis at themoderate fibrosis and cirrhosis stages are listed inTable 3.

DISCUSSION

It is difficult to accurately determine the cause ofa deformation, as it is usually assumed that deformationsresult from external loads (Dandel et al. 2009), bodyforces (such as gravity or electromagnetic forces) ortemperature changes within the body. Strain is one wayto explain deformations. Deformation analysis dependson finite strain theory, which states that the undeformedand deformed configurations of the continuum are signif-icantly different and are clearly distinct (Kwan et al.1990). The common cases are elastomers, plastically de-forming materials, fluids and biological soft tissues

Table 2. Quantitative parameters of TSI and th

ParameterNormal adultgroup (n 5 38)

Mild fibrosis(n 5 10)

Moderate fibro(n 5 27)

SPR (s-1)A 6.72 6 0.82 6.50 6 0.48 5.79 6 0.51B 3.56 6 0.72 3.04 6 0.59 2.19 6 0.18C 0.62 6 0.55 0.50 6 0.41 0.43 6 0.27D 0.29 6 0.39 0.16 6 0.03 0.14 6 0.08

TA-B(ms) 12.00 6 7.59 13.80 6 7.48 19.67 6 5.38TB-C(ms) 12.68 6 8.35 13.20 6 7.04 22.63 6 5.61TC-D(ms) 11.84 6 7.10 13.50 6 8.32 18.30 6 4.49PAPAS index 1.61 6 0.50 1.64 6 0.17 1.90 6 0.35

PAPAS index5 platelet/age/phosphatases/a-fetoprotein/aspartate aminotranfibrosis 5 S0 and S1; moderate fibrosis 5 S2 and S3; cirrhosis 5 S4 and clinValues are expressed as means 6 standard deviations. The p values in the la

with the Student’s t-test of variance.* p , 0.05 compared with the normal group.y p , 0.05 compared with the mild fibrosis group.z p , 0.05 compared with the moderate fibrosis group.

(including liver tissue). Normal liver tissue has a homoge-neous texture with hepatic portal veins scatteredthroughout the parenchyma. Liver cirrhosis is a chronicdegenerative disease characterized by fibrous tissue thatcovers the lobes, parenchymal degeneration and infiltra-tion of the lobules with fat. The essential features aresimultaneous parenchymal necrosis, regeneration anddiffuse fibrosis, resulting in the disorganization of lobulararchitecture (Guyot et al. 2006). ‘‘Classic’’ Doppler echo-cardiography measures the high-frequency, low-ampli-tude signals from rapidly moving red blood cells,enabling quantification of blood flow velocities. TDIuses the same physical principle; that is, it measures thelow-frequency, high amplitude signals of myocardialtissue motion (Van de Veire et al. 2008). The first descrip-tion of echocardiographic strain was derived from TDIvelocity data using the Doppler equation to convert ultra-sound frequency shifts into velocity information along thescan lines (Gorcsan and Tanaka 2011). A moving object

e PAPAS index in the 110 included cases

sis Cirrhosis(n 5 35)

p value

Normal/mild

Mild/moderate

Moderate/cirrhosis

4.70 6 0.93 0.41 0.001y ,0.001z

1.57 6 0.41 0.036* 0.044y ,0.001z

0.38 6 0.30 0.60 0.53 0.510.13 6 0.08 0.31 0.39 0.4620.34 6 7.90 0.51 0.01y 0.7024.86 6 6.05 0.86 ,0.001y 0.1419.03 6 4.46 0.63 0.03y 0.522.28 6 0.35 0.83 0.03y ,0.001z

sferase index; SPR5 strain peak rate; TSI5 tissue strain imaging; Mildically typical cirrhosis.st column were obtained by comparing the two groups in the same ROI

Fig. 4. Strain peak rates (SPRs) of regions of interest (ROIs) Aand B at different stages of liver fibrosis (mild fibrosis, moderatefibrosis and cirrhosis) in patients with hepatitis B virus. Barsrepresent mean SPRs of ROIs A and B. Lines through the bars

represent standard deviations.

Tissue Doppler and tissue strain imaging d Y. WANG et al. 1533

does not necessarily undergo deformation if every part ofitmoveswith the samevelocity.Deformation occurswhendifferent elements of the object move at different veloci-ties and the object has to change shape during its move-ment (Pavlopoulos and Nihoyannopoulos 2008). Theterm strain, whichmeans ‘‘stretching,’’ is used in echocar-diography to describe ‘‘deformation’’ (D’Hooge et al.2000). The strain rate is the rate at which deformationoccurs. Strain rate is a temporal derivative of strain anddescribes the rate of shortening or lengthening of an object(Dandel et al. 2009). TSI can provide a visual representa-tion of the local contraction and expansion of the myocar-dial wall, eliminating the effects of overall cardiacmovement and movements of the diseased regions thatare pulled by surrounding healthy tissues (Dandel andHetzer 2009). TSI will likely be the method of choicefor assessing hepatic fibrosis in routine clinical practice(Hotta et al. 2007).

In this study, we found that the SPRs of the ROIs(A–D) were normal and similar to the strain rate for the

Fig. 5. Area under receiver operating characteristic (ROC) cuscoring system. The ROC curves of the strain peak rates (SPand the PAPAS index predicted moderate fibrosis (a) and cirr

evaluating liver

ROI placed on the right atrioventricular surface in thenormal adult group, whereas the SPRs of the ROIs (A–D) were irregular and short in chronic liver disease. Thestrain rate curves tended to be more irregular withincreasing severity of chronic liver disease (Fig. 2), whichis why we selected the first and highest positive SPR toevaluate the stages of liver fibrosis. The SPRs of ROIsA–D, especially that for the SPRs of ROI B, decreasedgradually. Additionally, we found that the severity offibrosis significantly correlated with the decreased SPRsof ROI B measured by TDI and TSI (Table 2). Thisfinding was likely related to the difference in the degreeof fibrosis among the groups. The amount of fibrotictissue resulting in the fibrotic changes was most accu-rately reflected by the SPRs of ROI B, as measured byTDI and TSI, which were reasonably related to theseverity of liver fibrosis. The SPRs of ROI A did notsignificantly differ between the normal adult and mildfibrosis groups because ROI Awas too close to the heartto be greatly affected by heart movement in both groups.The SPRs of ROIs C and D did not significantly differentamong any of the groups because these values ap-proached zero. The location of ROI B chosen in this studywas the optimal location with which to analyze changesin the strain rates among the different groups. InFigure 4, it can be observed that the variations in theSPR of ROI B for the moderate fibrosis group were lesssignificant than those for the other groups. A possiblereason for this phenomenon was that the diagnostic accu-racy of SPR B for moderate fibrosis was higher than thatfor mild fibrosis. Compared with moderate fibrosis, livercirrhosis led to changes from a homogeneous texture toa heterogeneous texture, and according to the laws ofmechanics, object movement changed from regular toirregular; thus, variation in the measured value may belarger in the liver cirrhosis group.

rve calculated using tissue Doppler imaging and anotherRs) of region of interest (ROI) A, the SPRs of ROI Bhosis (b). The SPRs of ROI B were the most useful forfibrosis.

Table 3. Accuracy of diagnosis of the severity of liver fibrosis

Parameter

Moderate fibrosis Cirrhosis

Cutoff value Sensitivity (%) Specificity (%) Cutoff value Sensitivity (%) Specificity (%)

SPR of ROI A (s-1) ,6.20 85.20 80.00 ,5.05 68.60 96.30SPR of ROI B (s-1) ,2.49 96.30 80.00 ,1.90 85.70 100.00PAPAS index .1.69 74.10 70.00 .1.95 85.70 70.40

PAPAS index 5 platelet/age/phosphatase/a-fetoprotein/aspartate aminotransferase index; SPR 5 strain peak rate.

1534 Ultrasound in Medicine and Biology Volume 39, Number 9, 2013

Many predictive models of fibrosis, including theAST/platelet radio index (APRI) and the FIB-4 index,are based on patients with chronic hepatitis C (Waiet al. 2003). The PAPAS index was based on five commonclinical parameters: age, ALP, AST, AFP and plateletcount. All five parameters have been shown in previousstudies to be associated with significant fibrosis inchronic hepatitis B (Fung et al. 2008). To compare thequantitative parameters of TDI and TSI (SPRs of ROIA and B) with the PAPAS index in evaluation of theseverity of liver fibrosis, we calculated the area underROC curves. The SPRs of ROIs A and B, as measuredby TDI and TSI, were superior to the PAPAS index.This finding is reasonable because the previously re-ported PAPAS index values were obtained bymultivariateanalysis with indirect parameters, whereas the SPRs ofROIs A and B measured with TDI and TSI were deter-mined from direct evaluation of the liver.

The time to peak or onset of systolic velocity hasbeen measured to assess dyssynchrony of the heart (Yuet al. 2007). In our study, changes in the strain rate curvesof the strain values in hepatic parenchymawere obviouslysimilar (Fig. 2). These important observations had predic-tive value for the differences in time to the SPR derivedfrom TDI and TSI. Although the differences in time toSPRs from ROIs A to D (TA-B, TB-C and TC-D) betweenthe mild fibrosis group and moderate fibrosis groupwere statistically significant (p , 0.05), cutoff valuesfor TA-B, TB-C and TC-D were not obtained among themild, moderate and cirrhosis groups. However, the courseof the differences in time to SPR of the four ROIs (TA-B,TB-C and TC-D) was also related to liver fibrosis stage.Therefore, the greater differences in TA-B, TB-Cand TC-D may be the result of more severe liver fibrosis.

Our study has some limitations. First, 35 patients inour study were diagnosed with liver cirrhosis usingwidely accepted clinical criteria, and the others werediagnosed with needle biopsy of the liver, which createdthe possibility of differences in histologic stage based onsampling error. Second, the quantitative parameters(SPRs) of TDI and TSI that varied in our study mightbe affected by ROI position. To minimize the variabilityin quantitative parameters, one physician performedall TDI examinations and TSI analyses in our study.

However, this might have resulted in another limitationin that we were not able to compare results betweendifferent operators. Finally, we did not compare othernon-invasive methods with TDI and TSI; thus, a futurestudy is warranted to compare the different imagingmethods.

In conclusion, the SPRs measured by quantitativeanalysis of TDI and TSI reflected the severity of liverfibrosis. TDI and TSI provided a useful non-invasivemethod to evaluate liver fibrosis stage in patients withHBVs.

REFERENCES

Al KB, ShiffmanM. Percutaneous liver biopsy in clinical practice. LiverInt 2007;27:1166–1173.

Aube C, Oberti F, Korali N, Namour MA, Loisel D, Tanguy JY,Valsesia E, Pilette C, Rousselet MC, Bedossa P, Rifflet H,Maiga MY, Penneau-Fontbonne D, Caron C, Cales P. Ultrasono-graphic diagnosis of hepatic fibrosis or cirrhosis. J Hepatol 1999;30:472–478.

Aube C, Winkfield B, Oberti F, Vuillemin E, Rousselet MC, Caron C,Cales P. New Doppler ultrasound signs improve the non-invasivediagnosis of cirrhosis or severe liver fibrosis. Eur J GastroenterolHepatol 2004;16:743–751.

Bavu E, Gennisson JL, CouadeM, Bercoff J, Mallet V, FinkM, Badel A,Vallet-Pichard A, Nalpas B, Tanter M, Pol S. Noninvasive in vivoliver fibrosis evaluation using supersonic shear imaging: A clinicalstudy on 113 hepatitis C virus patients. Ultrasound Med Biol2011;37:1361–1373.

Bernatik T, Strobel D, Hahn EG, Becker D. Doppler measurements: Asurrogate marker of liver fibrosis? Eur J Gastroenterol Hepatol2002;14:383–387.

Bonekamp S, Kamel I, Solga S, Clark J. Can imaging modalities diag-nose and stage hepatic fibrosis and cirrhosis accurately? J Hepatol2009;50:17–35.

Boursier J, Isselin G, Fouchard-Hubert I, Oberti F, Dib N, Lebigot J,Bertrais S, Gallois Y, Cales P, Aube C. Acoustic radiation forceimpulse: A new ultrasonographic technology for the widespreadnoninvasive diagnosis of liver fibrosis. Eur J Gastroenterol Hepatol2010;22:1074–1084.

Dandel M, Hetzer R. Echocardiographic strain and strain rate imaging:Clinical applications. Int J Cardiol 2009;132:11–24.

Dandel M, Lehmkuhl H, Knosalla C, Suramelashvili N, Hetzer R. Strainand strain rate imaging by echocardiography: Basic concepts andclinical applicability. Curr Cardiol Rev 2009;5:133–148.

D’Hooge J, Heimdal A, Jamal F, Kukulski T, Bijnens B, Rademakers F,Hatle L, Suetens P, Sutherland GR. Regional strain and strain ratemeasurements by cardiac ultrasound: Principles, implementationand limitations. Eur J Echocardiogr 2000;1:154–170.

Di Bisceglie AM. Issues in screening and surveillance for hepatocellularcarcinoma. Gastroenterology 2004;127:S104–S107.

Fung J, Lai CL, Fong DY, Yuen JC, Wong DK, Yuen MF. Correlation ofliver biochemistry with liver stiffness in chronic hepatitis B and

Tissue Doppler and tissue strain imaging d Y. WANG et al. 1535

development of a predictive model for liver fibrosis. Liver Int 2008;28:1408–1416.

Gorcsan JR, Tanaka H. Echocardiographic assessment of myocardialstrain. J Am Coll Cardiol 2011;58:1401–1413.

Guyot C, Lepreux S, Combe C, Doudnikoff E, Bioulac-Sage P,Balabaud C, Desmouliere A. Hepatic fibrosis and cirrhosis: The(myo) fibroblastic cell subpopulations involved. Int J BiochemCell Biol 2006;38:135–151.

Hanley JA, McNeil BJ. The meaning and use of the area under a receiveroperating characteristic (ROC) curve. Radiology 1982;143:29–36.

Haque M, Robinson C, Owen D, Yoshida EM, Harris A. Comparison ofacoustic radiation force impulse imaging (ARFI) to liver biopsyhistologic scores in the evaluation of chronic liver disease: A pilotstudy. Ann Hepatol 2010;9:289–293.

Hotta N, Ayada M, Okumura A, Ishikawa T, Sato K, Oohashi T,Hijikata Y,Murayama S, IdoM, Kakumu S. Noninvasive assessmentof liver disease: Measurement of hepatic fibrosis using tissue strainimaging. Clin Imaging 2007;31:87–92.

Kim JE, Lee JY, Kim YJ, Yoon JH, Kim SH, Lee JM, Han JK, Choi BI.Acoustic radiation force impulse elastography for chronic liverdisease: Comparison with ultrasound-based scores of experiencedradiologists, Child-Pugh scores and liver function tests. UltrasoundMed Biol 2010;36:1637–1643.

KwanMK, Lai WM,Mow VC. A finite deformation theory for cartilageand other soft hydrated connective tissues: I. Equilibrium results.J Biomech 1990;23:145–155.

Li N, Ding H, Fan P, Lin X, Xu C, Wang W, Xu Z, Wang J. Intrahepatictransit time predicts liver fibrosis in patients with chronic hepatitisB: Quantitative assessment with contrast-enhanced ultrasonography.Ultrasound Med Biol 2010;36:1066–1075.

Liu GJ, Lu MD. Diagnosis of liver cirrhosis with contrast-enhancedultrasound. World J Radiol 2010;2:32–36.

Muller M, Gennisson JL, Deffieux T, Tanter M, Fink M. Quantitativeviscoelasticity mapping of human liver using supersonic shearimaging: Preliminary in vivo feasibility study. Ultrasound MedBiol 2009;35:219–229.

Oberti F, Valsesia E, Pilette C, Rousselet MC, Bedossa P, Aube C,Gallois Y, Rifflet H, Maiga MY, Penneau-Fontbonne D, Cales P.Noninvasive diagnosis of hepatic fibrosis or cirrhosis. Gastroenter-ology 1997;113:1609–1616.

Pagliaro L, Rinaldi F, Craxi A, Di Piazza S, Filippazzo G, Gatto G,Genova G, Magrin S, Maringhini A, Orsini S, Palazzo U,Spinello M, Vinci M. Percutaneous blind biopsy versus laparoscopywith guided biopsy in diagnosis of cirrhosis: A prospective, random-ized trial. Dig Dis Sci 1983;28:39–43.

Pavlopoulos H, Nihoyannopoulos P. Strain and strain rate deformationparameters: From tissue Doppler to 2 D speckle tracking. Int J Car-diovasc Imaging 2008;24:479–491.

Ratziu V, Charlotte F, Heurtier A, Gombert S, Giral P, Bruckert E,Grimaldi A, Capron F, Poynard T. Sampling variability of liverbiopsy in nonalcoholic fatty liver disease. Gastroenterology 2005;128:1898–1906.

Rosen BD, Fernandes VR, Nasir K, Helle-Valle T, Jerosch-Herold M,Bluemke DA, Lima JA. Age, increased left ventricular mass, andlower regional myocardial perfusion are related to greater extentof myocardial dyssynchrony in asymptomatic individuals: Themulti-ethnic study of atherosclerosis. Circulation 2009;120:859–866.

Scheuer PJ. Classification of chronic viral hepatitis: A need for reassess-ment. J Hepatol 1991;13:372–374.

Scheuer PJ. The nomenclature of chronic hepatitis: Time for a change.J Hepatol 1995;22:112–114.

SetoWK, Lee CF, Lai CL, Ip PP, Fong DY, Fung J, Wong DK, YuenMF.A new model using available clinical parameters to predict signifi-cant liver fibrosis in chronic hepatitis B. PLoS One 2011;6:e23077.

Stebbing J, Farouk L, Panos G, Anderson M, Jiao LR, Mandalia S,Bower M, Gazzard B, Nelson M. A meta-analysis of transient elas-tography for the detection of hepatic fibrosis. J Clin Gastroenterol2010;44:214–219.

Theise ND. Liver biopsy assessment in chronic viral hepatitis: Apersonal, practical approach.Mod Pathol 2007;20(Suppl 1):S3–S14.

Thompson CJ, Rogers G, Hewson P, Wright D, Anderson R, Jackson S,Ryder S, Cramp M, Stein K. Surveillance of cirrhosis for hepatocel-lular carcinoma: A cost-utility analysis. Br J Cancer 2008;98:1166–1175.

Van de Veire NR, De Sutter J, Bax JJ, Roelandt JR. Technologicaladvances in tissue Doppler imaging echocardiography. Heart2008;94:1065–1074.

Vuppalanchi R, Unalp A, Van Natta ML, Cummings OW,Sandrasegaran KE, Hameed T, Tonascia J, Chalasani N. Effectsof liver biopsy sample length and number of readings on samplingvariability in nonalcoholic fatty liver disease. Clin GastroenterolHepatol 2009;7:481–486.

Wai CT, Greenson JK, Fontana RJ, Kalbfleisch JD, Marrero JA,Conjeevaram HS, Lok AS. A simple noninvasive index can predictboth significant fibrosis and cirrhosis in patients with chronic hepa-titis C. Hepatology 2003;38:518–526.

Wiechowska-Kozlowska A, Bialek A, Starzynska T, Milkiewicz P.Endoscopic ultrasonography in the assessment and management ofcomplications of liver cirrhosis. Pol Merkur Lekarski 2008;24:134–139.

Yu CM, Sanderson JE, Marwick TH, Oh JK. Tissue Doppler imaging: Anew prognosticator for cardiovascular diseases. J Am Coll Cardiol2007;49:1903–1914.

Zheng RQ, Wang QH, Lu MD, Xie SB, Ren J, Su ZZ, Cai YK, Yao JL.Liver fibrosis in chronic viral hepatitis: An ultrasonographic study.World J Gastroenterol 2003;9:2484–2489.