Arrival-Time Parametric Imaging in Contrast-Enhanced Ultrasound for Diagnosing Fibrosis in Primary Biliary Cholangitis

Liver biopsy is usually required for diagnosing fibrosis in primary biliary cholangitis (PBC), but contrast-enhanced ultrasonography (CEUS) is a possible alternative. The aim of this study was to investigate arrival-time parametric imaging (At-PI) in for diagnosing fibrosis in PBC. Forty-eight patients (male/female, 8/40; mean age, 60 ± 13 years) with PBC diagnosed by liver biopsy underwent CEUS during 2009–2019. Of these, 27 who also underwent shear wave elastography (SWE) were further analyzed. Perflubutane was intravenously injected and CEUS performed. Contrast dynamics of hepatic segment V and the right kidney were recorded and At-PI generated. The ratio of red indicating contrast arrival time <5 seconds to the entire liver contrast-enhanced area was calculated and compared with shear wave velocity (Vs) measured by SWE by fibrosis stage (F0–F3), bile duct loss score, cholangitis activity, hepatitis activity (HA0–HA3), and disease stage, as determined by liver biopsy. Ratio of red significantly differed between F0 and F2–F3 and between F1 and F2–F3. Using ratio of red to diagnose ≥F1 (≥F2), area under the receiver operating characteristic curve was 0.77 (0.92) (cutoff, 36.7% [47.1%]; sensitivity, 0.75 [0.92]; specificity, 0.82 [0.81]). At-PI was useful for diagnosing fibrosis, especially F2 or worse, in PBC, suggesting that At-PI can correctly diagnose fibrosis regardless of hepatic inflammation.

P rimary biliary cholangitis (PBC) is a chronic progressive cholestasis that affects primarily middle-aged women. 1,2 The precise causes of PBC are not known, but genetic and environmental factors are likely involved, and an autoimmune mechanism is implicated in the pathology. 3 Prognosis of PBC is associated with fibrosis stage, 4,5 which is usually determined by histopathology of liver biopsy specimens. However, histopathological diagnosis using conventional assessment systems (eg, the Scheuer scoring system 6 ) is associated with sampling errors due to heterogenous intrahepatic changes in PBC. Thus, Nakanuma et al 7 recently proposed a new histological staging and grading system (the Nakanuma classification 7 ). Nevertheless, biopsy is an invasive procedure 8 and cannot be used for serial follow-up of disease progression, and noninvasive prediction of liver fibrosis would be of great clinical utility.
Previously, we developed a novel contrast-enhanced ultrasound (CEUS) technique for quantitatively assessing changes in hepatic blood flow from portal venous dominant to arterial dominant and reported its efficacy for assessing liver fibrosis in chronic hepatitis C infection. 9 Here, we analyzed the findings of this technique against pathohistological findings by the Nakanuma classification to determine whether this technique can diagnose liver fibrosis in PBC.

Enrollment of Patients
Subjects were patients with PBC scheduled to undergo liver biopsy at Toho University Omori Medical Center between 2009 and 2019 (see Yoshimine et al for details [preprint], doi:10. 21203/rs.3.rs-129916/v1). Written consent was provided by the patient, or by a family member if the patient was unable to do so. Exclusion criteria were (1) current or past history of hepatitis C or hepatitis B; (2) regular daily alcohol intake of ≥60 g ethanol; (3) liver tumor or portal vein embolism; (4) egg yolk allergy; (5) pregnant and/or breastfeeding; (6) age 16 years or younger; (7) severe heart disease; and (8) severe pulmonary disease. Patients with regular daily alcohol intake were excluded because its potential influence on the findings of this study was not known. Patients with liver tumor or portal vein embolism were excluded because of the possible influence on the contrast agent arrival time. Exclusion criteria 4 to 8 were contraindications of perflubutane.
This prospective single-center study was conducted in accordance with the Declaration of Helsinki and was approved by the ethics committee at our institution (numbers 21-26, 26-227, M18272). Japan) with a 3.75-MHz convex array probe (PVT-375BT; Canon Medical Systems). Contrast-enhanced ultrasonography was performed at a right intercostal space. The mechanical index was 0.22-0.29, and the frame rate was 15-18 frames/s. Analyzed images showed the right kidney and liver parenchyma of the right hepatic lobe (segment V). Focus was set to 6-8 cm to cover the whole kidney. Patients fasted overnight before the CEUS examination. During the examination, participants were supine with the right arm above the head and held their breath. The recommended dose of perflubutane 10 (0.015 mL/kg; Sonazoid; GE Healthcare, Oslo, Norway) was administered as a bolus via the median cubital vein at 1 mL/s and flushed with 10 mL normal saline. Cine acquisition was started at the beginning of saline flush. Data generated for the first 40 seconds were saved as raw data. An independent ultrasonographer with over 26 years of experience performed US examinations and was blinded to patients' clinical characteristics.

Arrival-Time Parametric Imaging
The ultrasound system software generated At-PI images from stored video. By simply selecting the renal parenchyma as the region of interest, the system set the point at which 80% of the region of interest was contrasted as time 0 and sequentially calculated the arrival time in individual pixels of the hepatic parenchyma. A color map is superimposed on a B-mode image. The difference in arrival time between arterial and subsequent portal venous blood to the liver is reported as 5 seconds. 11 Arrival times 0 to <5 seconds are shown as red (and ≥5 to 10 seconds as yellow, indicating liver parenchyma supplied via the arterial and portal venous routes, respectively; Fig. 1).

Measurement of Contrasted Areas
Obtained At-PI data were evaluated quantitatively, as previously described. 12 Briefly, the ratio of red (ROR) pixel area (shorter arrival times) to the area of all contrast-enhanced pixels was calculated as the ROR, using ImageJ version 1.42 (National Institutes of Health, Bethesda, MD). Higher ROR indicates that the contrast agent arrival time in the liver is closer to that in the kidney, meaning a wider area of the liver parenchyma received the contrast agent from the arterial route, indicating an arterialportal shift in the blood flow balance toward arterial dominance in the liver. Two physicians calculated the ROR and were trained on use and interpretation of contrast agents in the liver.

Shear Wave Elastography
Virtual Touch Quantification (VTQ; Siemens Healthineers AG, Erlangen, Germany) has been used for acoustic radiation force impulse-based point shear wave elastography (SWE) in our hospital since October 2012. The VTQ elastography was therefore performed as additional analysis in all consenting patients. Shear wave velocity (Vs) was measured 6 times with an ACUSON S2000 US system (Siemens Healthineers AG), and the median value (m/s) was calculated. [13][14][15] The median value was considered to be representative of the VTQ measurements only if the interquartile range of all validated measurements was within 30% of the median value. We analyzed the correlation between median Vs and ROR range.

Serological Markers
Blood samples collected within 3 day before US examinations were used to analyze the correlation between the ROR range and various serological markers (aspartate aminotransferase  16 The AST/ALT ratio was examined because it is reported to be associated with progression of chronic liver disease 17 and PBC. 18

Liver Biopsies
After CEUS, liver biopsies were obtained using a semiautomatic biopsy instrument with a 16-gauge liver biopsy needle (Core II; InterV Clinical Products, Dartmouth, MA). Specimens obtained from the anterior segment of the right lobe under US guidance were fixed in 10% formalin, embedded in paraffin, sectioned, and stained with hematoxylin-eosin and azan for histologic evaluation. All liver specimens contained at least 15 portal tracts, including 11 complete portal triads.
The Nakanuma classification was used for histopathological assessment. 7 Fibrosis (F0-F3, item A) was assessed as follows: F0, almost no fibrosis or fibrosis in the portal tracts only; F1, fibrosis extending beyond the portal area, sometimes including incomplete septal fibrosis; F2, completely connected septal fibrosis or bridging fibrosis with variable lobular distortion; and FIGURE 1. Representative images for At-PI imaging. By simply selecting the renal parenchyma as the ROI, the system set the point at which 80% of the ROI was contrasted as time 0 and sequentially calculated arrival time in individual pixels of the hepatic parenchyma. We used red and yellow to display pixels arriving at 0 to <5 seconds and at ≥5 to 10 seconds, respectively. F3, cirrhosis (extensive fibrosis with regenerative nodules). Next, bile duct loss (BL0-BL3, item B) was assessed. Then, disease stage (stage 1-4) was determined based on the combined score of items A and B. We also evaluated cholangitis activity (CA0-CA3) and hepatitis activity (HA0-HA3) scores.
A single experienced pathologist blinded to patients' clinical characteristics evaluated all biopsy specimens. Liver biopsy and US imaging were performed on the same day.

Liver Pathological Factor
The ROR and Vs values obtained by US were analyzed with respect to liver pathological factors (fibrosis, bile duct loss, CA, HA, and disease stage).

Statistical Analysis
Parameter analysis of ROR and Vs values was performed for each liver pathological factor; box plots were used to analyze the ROR distribution and Vs values according to liver pathological factors. The Jonckheere-Terpstra test was used to evaluate trends, and the Kruskal-Wallis test was used for comparisons between groups. Receiver operating characteristic (ROC) curves were used to assess the diagnostic performance of ROR and Vs and to determine the optimum cutoff value and area under the ROC curve (AUROC) for diagnosing liver pathological factors. The correlation of ROR with Vs value and serological markers was analyzed using Spearman rank correlation coefficients. Logistic regression analysis was used to identify factors associated with progression of liver fibrosis. All data analyses were performed using Stata version 15 (StataCorp, College Station, TX). A P value less than 0.05 was considered statistically significant.

Patient Characteristics
This study enrolled 56 patients who consented to participate. After excluding 8 patients (2 with a disease other than PBC diagnosed by histopathology, and 6 who could not adequately hold their breath during At-PI), 48 patients comprising 8 men and 40 women aged 60 ± 13 years were included in the analysis. Fibrosis stage was 0 in 18 patients, 1 in 18 patients, 2 in 10 patients, and 3 in 2 patients. Clinical, biochemical, and histological characteristics of the patients with PBC (n = 48) analyzed in this study are summarized in Tables 1 to 4.  (Table 5).

Correlation Between At-PI and SWE
For additional analysis, 27 (4 men, 23 women) of the 48 patients consented to undergo SWE.
In all 27 patients who underwent SWE in this study, the interquartile range of all validated measurements was within 30% of the median value. The mean age was 61 ± 13 (range, 32-78) years. Fibrosis stage was F0 in 11 patients, F1 in 11, F2 in 5, and F3 in 0. ROR range (%) was not correlated with Vs ( P = 0.34, r = 0.19).

Diagnostic Capability of At-PI, SWE, and Serum Fibrosis Markers for Fibrosis Stage Based on ROC Curves
Results of the comparison between At-PI, SWE, and serum fibrosis markers in patients with PBC are shown in Table 6.

Arrival-Time Parametric Imaging
The cutoff value and AUROC for At-PI were, respectively, 47.1 and 0.92 for diagnosis of fibrosis stage F0-F1 and ≥F2, and 36.7 and 0.77 for stages F0 and ≥F1. When these values were used to diagnose cases of ≥F1 and ≥F2 on At-PI, the sensitivity and specificity were, respectively, 0.92 and 0.81 for ≥F2 and 0.67 and 0.89 for ≥F1 (Figs. 5A, B).

Shear Wave Elastography
The cutoff value and AUROC for SWE were, respectively, 1.40 and 0.80 for diagnosis of fibrosis scores F0-F1 and ≥F2, and 1.23 and 0.84 for F0 and ≥F1. When these values were used to diagnose cases of ≥F1 and ≥F2 on At-PI, the sensitivity and specificity were, respectively, 0.60 and 0.86 for ≥F2 and 0.75 and 0.82 for ≥F1.

Fibrosis-4
The cutoff value and AUROC for FIB-4 were, respectively, 2.78 and 0.85 for diagnosis of fibrosis scores F0-F1 and ≥F2, and 2.47 and 0.78 for F0 and ≥F1. When these values were used to diagnose cases of ≥F1 and ≥F2 on At-PI, the sensitivity and specificity were, respectively, 0.83 and 0.81 for ≥F2 and 0.60 and 0.94 for ≥F1.

AST/ALT Ratio
The cutoff value and AUROC for the AST/ALT ratio were, respectively, 1.05 and 0.60 for diagnosis of fibrosis scores F0-F1 and ≥F2, and 0.96 and 0.62 for F0 and ≥F1. When these values were used to diagnose cases of ≥F1 and ≥F2 on At-PI, the sensitivity and specificity were, respectively, 0.67 and 0.56 for ≥F2 and 0.77 and 0.50 for ≥F1.

Univariate and Multivariate Analysis of Predictive Factors for the Progression of Liver Fibrosis in Patients With PBC
Among the 4 diagnostic tools for fibrosis (AST/ALT ratio, FIB-4, SWE, At-PI) that were evaluated for ability to predict progression of liver fibrosis (liver fibrosis ≥F2) in patients with PBC, multivariate analysis identified At-PI as an independent predictor (Table 7).

DISCUSSION
Unlike other organs, the liver has a dual blood supply, receiving approximately 70% to 80% from the portal vein and approximately 20% to 30% from the hepatic artery. 19 Even with this dual blood supply, necrosis and desquamation occur in patients with chronic liver diseases such as PBC, resulting in the progression of liver fibrosis and leading to disease progression from chronic hepatitis to liver cirrhosis. As the disease progresses, the blood supply from the portal vein decreases, which is compensated by increased blood supply from the hepatic artery. [19][20][21][22] In other words, the arterial-portal blood flow balance in the liver shifts from portal vein dominance to the hepatic artery dominance with disease progression. The use of imaging to quantitatively assess such changes in the balance of arterialportal blood flow in the liver would be clinically useful as a noninvasive method for determining the severity of lesions in PBC and for predicting the development of esophageal varices and hepatocellular carcinoma.
Histopathological examination of biopsy specimens is currently required for diagnosis of PBC. However, the intrahepatic distribution of bile duct lesions is reported to be heterogenous in PBC, 23 which complicates the histopathological diagnosis of disease stage when a conventional scoring system is used. To address associated sampling error, the Nakanuma classification 7 was proposed and is becoming a standard histopathological diagnosis system in Japan. However, the invasiveness of biopsy remains a problem. Laparoscopic examination of the liver surface was previously shown to be useful for assessing the entire liver, 24 but this procedure is also invasive. Compared with liver biopsy and laparoscopy, ultrasonography is simpler, less invasive, and easier to perform repeatedly. Thus, in this study, we examined contrast-enhanced ultrasonography images, with a focus on hepatic arterialization. We showed that ROR was positively correlated with some serum fibrosis markers (ie, FIB-4 and AST/ALT ratio). ROC curve analysis of ROR for diagnosing ≥F1 and ≥F2 showed AUROC of 0.77 and 0.92, respectively, indicating that At-PI is useful for diagnosing fibrosis in PBC.
Only a few studies have investigated the noninvasive diagnosis of fibrosis in PBC. Yan et al 25 analyzed the results of an ultrasound-based technique (ie, 2D SWE) against histopathological findings in 157 patients with PBC, and ROC curve analysis showed AUROC values for histopathological diagnosis of ≥F2, ≥F3, and ≥F4 by the Scheuer scoring system of 0.88, 0.97, and 0.99, respectively, indicating good diagnostic performance. 25 Similarly, Park et al 26 examined point SWE findings against pathohistological findings in 41 patients with PBC and showed that the AUROC values for pathohistological diagnosis of ≥F2 and ≥F3 by the METAVIR classification were 0.81, and 0.91, respectively, indicating its good diagnostic performance. Vetter et al 27 used VTQ to perform SWE; histopathology of the liver using the Ishak score in 33 patients with PBC showed that liver stiffness increased with progression of fibrosis stage, particularly in biopsies with at least 10 portal fields (r = 0.7538, P = 0.0012). Mireen et al performed MRI using gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid as a contrast agent; histopathology of the liver showed AUROC values for ROC curve diagnosis of ≥F2, ≥F3, and ≥F4 by Ludwig classification of 0.76, 0.67, and 0.91, respectively, indicating good diagnostic performance. 28 However, none of those studies used the Nakanuma classification for histopathology, and thus, the results of those studies cannot be compared directly with the results of the present study. Nevertheless, the performance of our noninvasive technique using At-PI study seemed to be comparable to those of previously reported approaches.
SWE results have been reported to be influenced by liver inflammation. [29][30][31] Indeed, the Vs values in the present study differed significantly according to HA, suggesting that the SWE results were affected by liver inflammation. However, ROR values did not differ significantly according to HA. It is likely that At-PI can accurately diagnose liver fibrosis regardless of liver inflammation in patients with PBC, which is in good agreement with the findings of our previous study, which examined patients with hepatitis C. 32 Meanwhile, Namisaki et al 5 showed significantly poorer prognosis of PBC in patients with liver fibrosis with Nakanuma classification ≥F2 compared with those with liver fibrosis of F0-F1. Multivariate analysis of the 4 diagnostic tools (AST/ ALT ratio, FIB-4, SWE, At-PI) for liver fibrosis (≥F2) identified At-PI as an independent predictor and indicated the possibility that At-PI is useful for diagnosing progression to liver fibrosis (≥F2), which is associated with poor prognosis.
This study had several limitations. Our method might be affected by heart disease, which can alter the arrival time of con-trast agent to the liver; renal disorders, which possible change US signal kinetics in the kidney; habitual heavy drinking, which possibly alters hemodynamics; and portal vein embolism, which possibly disturbs the balance between arterial and portal blood flow. Accordingly, this method is not applicable to patients with these conditions. Patients cannot be examined by this method if the right hepatic lobe is difficult to visualize by sonography (eg, due to narrow intercostal spaces) or if they cannot hold their breath for 15-20 seconds. Perflubutane cannot be used in patients with egg allergy.