Diagnosis efficacy of CEUS for hepatic inflammatory lesions

Abstract Purpose In this study, the efficacy of US/CEUS and clinicopathologic parameters in differential diagnosis of hepatic inflammatory lesions were evaluated. Methods This was a retrospective study in which CEUS imaging was performed on 182 patients. Among these patients, 44 patients had hepatic inflammatory lesions and 138 patients had malignant lesions. The ultrasound (US), CEUS, and clinicopathologic parameters with respect to differential diagnosis of hepatic inflammatory lesions were analyzed. Results Irregular lesion shape and unclear margin were commonly seen in hepatic inflammatory lesions by US/CEUS examination. Hypoenhancement in arterial phase (AP) and portal venous phase (PVP), and isoenhancement in delayed phase (DP) were more commonly found in inflammatory lesions rather than malignant lesions [9% (4/44), 68% (30/44), and 16% (7/44) vs 2% (3/138), 11% (15/138), 1% (1/138), respectively; P < .05]. The enhancement coverage was also a significant indicator for the differentiation of inflammatory lesions and malignant lesions (P < .05). History of hepatitis or cirrhosis, and higher serum alpha‐fetoprotein (AFP) level were indicators for malignant lesions, while liver parasites and higher body temperature were indicators for inflammatory lesions. When the US/CEUS findings were combined with clinicopathologic parameters, the diagnostic accuracy of inflammatory lesions could reach 93.3%, with sensitivity, specificity, positive predictive value, and negative predictive value of 63.64%, 96.03%, 84.85%, and 88.32%, respectively. Conclusion The US/CEUS findings combined with clinical characteristics can accurately differentiate hepatic inflammatory lesions and malignant lesions. The results of study will improve the diagnostic confidence for hepatic inflammatory lesions.


| INTRODUC TI ON
Hepatic inflammatory lesions include pyogenic liver abscess, parasitic liver abscess, inflammatory pseudotumor, and granulomatous inflammation. [1][2][3][4][5][6][7] The most common symptoms are fever and abdominal pain. It usually progresses rapidly, for example, liver abscess can develop a liquefaction center within 2 weeks. 8 However, due to the extensive use of antibiotics, hepatic inflammatory lesions usually exhibit atypical clinical features. Under the circumstances, radiological examinations often play an important role in the diagnosis of the inflammatory liver mass. Unenhanced ultrasound (US) and color Doppler ultrasonographic examination are widely used to screen liver lesions, but these techniques have limited performance in the characterization of inflammatory lesions because the inflammatory mass may present highly variable US findings depending on the pathological stage. In addition, the inflammatory lesions and malignant lesions share similar vascular pattern on color Doppler images, which may lead to misdiagnoses or even unnecessary surgery. [9][10][11][12][13][14][15][16][17][18] Contrast-enhanced ultrasound (CEUS) with intravenous bolus injection of microbubbles can reflect tissue perfusion and improve the display of the characterization of focal liver lesions (FLL), 19 especially for the differential diagnosis of benign and malignant FLL. CEUS is comparable to CT and MRI for the diagnosis of liver masses if an appropriate acoustic window is available. [20][21][22][23][24][25][26] However, diagnostic efficacy of CEUS for the hepatic inflammatory lesions is still debatable, because it has similar perfusion pattern with malignant lesion. Therefore, in this study, differential diagnostic value of US/CEUS findings and clinicopathologic parameters for inflammatory lesions diagnosis was retrospectively analyzed to evaluate the diagnosis efficacy of CEUS for hepatic inflammatory lesions. All ultrasound diagnosis was confirmed by histopathological examination of the percutaneous biopsy or surgical specimens. The study was approved by the local institutional ethics committee.

| Clinical characteristics
The demographics, body temperature, and history of cirrhosis or biliary calculi were recorded for each patient. All the patients underwent blood routine test, including white blood cell counts (WBC), eosinophil percentage (EOS %), hepatic function tests, parasitology test, and serum tumoral marker test including alpha-fetoprotein (AFP), carcinoembryonic antigen (CEA), and carbohydrate antigen 19.9 (CA19.9).

| US/CEUS examination
Conventional abdominal US and CEUS examinations were performed by four US doctors who all had more than 10 years' experiences in abdominal ultrasound and more than 5 years' experiences in CEUS.
During the conventional abdominal US examination, the size, shape, and internal echogenicity of the mass, the number of mass, and appearance of mass' margin were recorded.
Following the B-mode evaluation of the hepatic lesions, CEUS examinations were performed using Philips IU22 (Philips Healthcare) equipped with a C1-5 probe. For CEUS examination, a low mechanical index (MI) was used for continuous real-time imaging. Ultrasonic second-generation contrast agents, sulfur hexafluoride microbubbles (SonoVue ® ), were used. An intravenous bolus of 2.4 mL SonoVue ® was applied followed by a bolus of 5 mL saline flush.
The lesion was evaluated during three phases, 10 to 30 seconds (arterial phase, AP), 31 to 120 seconds (portal venous phase, PVP), and 121 seconds (delayed phase, DP) after SonoVue ® injection. The level of the enhancement of the lesion was compared to the adjacent liver parenchyma and describes as hypo-, iso-, or hyperenhancing ( Figure 1).

| Statistical analysis
Age, gender, history of hepatitis and/or cirrhosis, lesion number, lesion size, baseline ultrasound echogenicity, lesion margin, lesion shape, enhancement in AP, enhancement in PVP, enhancement in DP, the enhancement coverage, WBC, serum AFP level, CEA level, CA19-9 level, temperature, biliary calculi, parasites, and EOS% were treated as both continuous and dichotomous variables, using their respective medians for statistical analysis. Continuous variables were compared using Student's t test or non-parametric Mann-Whitney test. The diagnostic values of the clinicopathologic variables and US/CEUS findings were assessed using univariate regression analyses. The significant diagnostic factors (P < .05) were further subjected to a forward stepwise multivariate logistic regression to determine the independent diagnostic factors for differentiating malignant mass and inflammatory mass. All variables found to be significant on univariate analysis (P < .05) were entered into a step-down Cox proportional hazard regression analysis. SPSS 10.0 software package (SPSS Inc) was used for data analysis.

| Regression analyses of the clinicopathologic variables and US/CEUS findings
Univariate logistic regression analyses were performed to examine the diagnostic value of clinicopathologic variables and US/CEUS findings in differentiating malignant lesions and inflammatory lesions. The corresponding P value of each variable is listed in Table 1.
Using regression analyses, 13 variables were identified, including presence of hepatitis, cirrhosis, parasites, the appearance of lesion margin, the shape of the lesion, enhancement in AP, enhancement in PVP, enhancement in DP, the enhancement coverage, WBC, AFP, body temperature, and EOS%, which were significant predictors for inflammatory lesions or malignant lesions.

| Differentiating values of US/CEUS findings for inflammatory lesions and malignant lesions
Irregular lesion shape and unclear margin were more commonly found in inflammatory lesions. Irregular lesion shape and unclear margin were found in 75% (33/44) and 82% (36/44) of inflammatory lesions, respectively, whereas regular shape and smooth margin The enhancement coverage, which represented percentage area with isoenhancement or hyperenhancement in the arterial phase, was also a significant predictor for differentiating inflammatory lesions and malignant lesions (P < .05) (Figure 3).
The above-mentioned significant univariate predictors were further analyzed using stepwise multivariate logistic regression model to identify the independent differential diagnostic factors for malignant lesions and inflammatory lesions. Through multi-factor stepwise regression analysis, we identified three independent US/CEUS findings including enhancement in AP, enhancement in DP, and the shape of the lesion ( Table 2). The diagnostic accuracy for IL was 74.5% based on the three independent factors as ROC curve showed (Figure 4). The sensitivity, specificity, positive predictive value, and negative predictive value of this logistic model were 27.27%, 98.55%, 85.71%, and 80.95%, respectively.
Liver parasitic disease and higher body temperature were associated with inflammatory lesions. Liver parasites and higher body  The serum AFP level of patients with malignant lesions was higher than that of patients with inflammatory lesions (P < .05), whereas the WBC level and EOS% level of patients with inflammatory lesions were significantly higher than that of patients with malignant lesions (P < .05).
The above-mentioned significant univariate predictors were further analyzed using stepwise multivariate logistic regression model to identify independent differential diagnostic factors for malignant lesions and inflammatory lesions. Through multi-factor stepwise regression analysis, four independent clinicopathologic parameters including AFP, temperature, parasites, and WBC were identified.

| Diagnostic efficiency of US/CEUS findings and clinicopathologic parameters for differentiating inflammatory or malignant lesions
The enhancement in AP, enhancement in DP, shape, AFP, temperature, parasites, and WBC were included in our study as independent factors. The diagnostic accuracy was 93.3% based on the independent factors as ROC curve showed ( Figure 5). The sensitivity, specificity, positive predictive value, and negative predictive value of this logistic model were 63.64%, 96.03%, 84.85%, and 88.32%, respectively.

| D ISCUSS I ON
CEUS showed exquisite vascularity and tissue perfusion in realtime and excellent spatial resolution. 27 In this study, we performed phase. ICC and hepatic metastases often showed a rim-like hyper-/ isoenhancement with the enhancement coverage less than 50% in AP, followed by hypo-/isoenhancement during the PVP and LP. [28][29][30] The rim-like enhancement pattern also appeared in the inflammatory lesion due to the formation of necrotic center or peripheral inflammatory cell infiltration and granulation tissues. Although the Note: CEA > 3.4 ng/mL was considered as positive. CA19-9 > 22 U/mL was considered as positive. WBC > 10 × 10 9 /L was considered as positive. EOS% > 5% was considered as positive.

TA B L E 1 (Continued)
pathological changes of inflammatory lesions vary, the distribution of the enhancement within all inflammatory lesions was similar. The enhancement pattern of inflammatory lesions in AP was nonspecific.
Washout enhancement pattern was considered as the presence of hypoenhancement of the lesion in the portal or late phases preceded by hyperenhancement in the arterial phase. Lesions with washout enhancement pattern should be considered as malignant until proven benign. 31 However, in our study, it was found that the proportion of patients with isoenhancement in DP was more likely to be diagnosed with hepatic malignant tumor. When the lesion showed isoenhancement in DP, it obviously increased the confidence of IL. We could find 84% (37/44) typical of IL showed hypoenhancement in DP. The proportion of IL, which showed hypoenhancement in DP, was more than the malignant lesions, which meant that the IL washout was faster than the malignant lesion.
Washout enhancement in DP was more common in inflammatory Tumor markers are specific antigen used as a biomarker of malignant cellular transformation. [51][52][53][54][55] We included the clinically widely used tumor serum markers including AFP, CEA, and CA199 in this study and aimed to investigate whether tumor marker measurement would facilitate differential diagnosis of inflammatory lesion and malignant lesion. AFP was not found to be a significant factor as suggested by multivariable analysis; CEA and CA199 might be taken into account. Although chills and fever are not always typical in inflammatory lesions because of the extensive use of antibiotics, we found that higher body temperature indicated inflammatory lesions, especially when combined with parasites and WBC tests.
In this study, liver inflammatory lesions had rapid enhancement in the arterial phase and washout in the portal venous or the late phase, which is nonspecific and well-resembled malignant lesions.
Therefore, CEUS-derived temporal enhancement parameters could not be solely relied on for differential diagnosis of inflammatory lesion and malignant lesion. However, when combined with patient medical history and laboratory test, the diagnosis confidence was greatly improved.
The limitation to our study is that the number of patients with inflammatory lesions was small while that of patients with malignant lesions was large. Further research with larger numbers was needed.
In conclusion, the CEUS-derived temporal parameters combined with clinical characteristics can accurately differentiate hepatic inflammatory lesions and malignant lesions. Our method might be a potential way to improve the diagnostic confidence for hepatic inflammatory lesions.