Abstract
Introduction
In recent years ultrasound (US) elastography became increasingly important for evaluation of tissue’s stiffness. Application examples for clinical use of US elastography including transient, strain and shear wave elastography are summarized in the EFSUMB guidelines and recommendations on the clincal use of elastography [5, 13]. Strain elastography (SE) is a common and widely available method and can be used effectively in percutanous diagnostics of the liver in addition to its applications in breast, prostate, gastrointestinal tract and thyroid imaging. Subject of a few recent studies was an evaluation of SE for percutanous diagnostics of parenchymal changes in liver tissue (e.g. fibrosis, cirrhosis). Furthermore, efficiency of SE in differentiating malignant from benign focal liver lesions (FLL) was assessed, which is usually perfomed by contrast-enhanced modalities such as computertomography (CT), magnetic resonance imaging (MRI) and contrast-enhanced ultrasound (CEUS) [1, 12]. Depending on the tumor’s etiology sensitivity values of 60–100% were obtained for strain elastography in detection and characteriziation of malignant liver lesions. Those values proved to be marginally lower compared to published sensitivity values of at least 80% for relatively well studied CEUS [7, 16].
Strain elastography was performed percutanously in the majority of studies leading to a potential falsification of the qualitative and semi-quantitative measurements by overlapping of ascites, subcutanous fat, organs (e.g. colon) etc. Kato et al. were one of the author groups to publish results of intraoperative elastography measurements placing the US probe on the surface of a complete mobilized liver. However, analysis in this study was performed qualitative only classifying liver tumors in 4 groups (A-D) by subjective impression. A semi-quantitative analysis was absent [14].
Aim of this study was to evaluate the efficiancy of intraoperative strain elastography (Io-SE) compared to well established intraoperative CEUS (Io-CEUS) [18] for dignity assessment of liver tumors during surgery by obtaining semi-quantitative measurements on the liver surface in order to avoid false measurements by overlaying structures. Postoperative histopathology was used as goldstandard.
Material and methods
A prospective data aquisition of DICOM US data of 100 patients with 116 liver tumors, who underwent liver surgery for tumor between October 2010 and March 2016, was carried out.
Intraoperative ultrasound (Io-US) was performed by one experienced examiner (more than 10 years of experience, DEGUM III) under sterile conditions (LOGIQ, E9, GE). Manual probe guidance was carried out by the operating surgeon. A complete mobilization of the liver was necessary. Multifrequency linear probes (ML-6-15) of 6–15 MHz and 6–9 MHz (9 L) in virtual convex mode were used. Loops >5 seconds were digitally stored on the US machine and in PACS. B-mode, SE and CEUS were documented during surgery.
Strain elastography
For high quality perfomance of strain elastography a constant alternation of compression and relief is necessary [13]. By application of an combined autocorrelation method an US grey scale or color coded elastography image can be created. A quality index of at least one red and seven green marks max. can be used for optimisation of compression performance on the used US machine (LOGIQ, GE) and faciliates the comparability of results.
A retrospective reading of elastography loops was performed qualitative by evaluation of the dominant intratumoral color pattern and defining three groups of pattern. Group A consisted of lesions with mainly blue coded areas (>50%) in the tumor’s center suggesting malignancy. Lesions with inhomogenous elastographic pattern displaying no dominant color were summarized to Group B. Group-B-lesions’ dignity was assessed in combination with B-mode impression. Group C lesions displayed more than 50% red, yellow or green coded areas suggesting the lesion to be soft and therefore benign.
Semi-quantitative measurement of the tumors’ stiffness was performed by (Q-analysis) using a scale from 0–6 for quantification of stiffness (0 = soft, 6 = indurated). 2 regions of interest (ROIs) sized 5–10 mm were placed in the lesion’s center, 5 ROIs were placed in the surrounding liver tissue as reference stiffness values for comparative purposes. Five particular measurements per ROI were carried out during Q-analysis. A representative mean value was obtained. Intratumoral stiffness values >2.5 were rated as maligant after application of a ROC-analysis and Youden index to the ROC-analysis results. Stiffness values in the lesion’s center lower 2.5 suggested dominant soft tissue. Consequently these lesions were classified as benign. Intraoperative application examples from our clinical routine are displayed in Figs. 1–6.
Io-CEUS
Io-CEUS was performed by use of contrast harmonic imaging (CHI) technique after bolus injections of 5–10 ml sulphurhexaflourid microbubbles (SonoVue, Bracco, Italy) [8]. Microbubbles injection was followed by injection of 10–20 ml sodium chlorid applicated by the anesthesiologist in a cental venous line. CEUS was recorded over 5 minutes storing cine loops >5 seconds. First, localisation and characterisation of already known liver lesions was performed following a scan of the complete mobilized liver for detection of further unknown liver lesions and marking possible ablation or resection margins. Informed consent of all patients was obtained for Io-CEUS. A positive ethics committee votum for this sudy was obtained.
Analysis of Io-CEUS was performed by evaluation of wash-in- and wash-out-kinetics of a lesion in arterial, portal venous and late phase. Examples are shown in Figs. 4–6. Wash-out beginning in portal venous until late phases and/or arterial rim hypervascularisation were considered criteria of malignancy depending on different tumors’ etiologies [6, 19].
The results of Io-SE and Io-CEUS were correlated to the results of histopathology. Histopathology was documented in all cases and served as goldstandard.
Statistics
A statistical analysis of elastography data was performed using a ROC-analysis, t-test with paired samples (SPSS, Version 24, IBM) and Youden-index (Microsoft Office Excel, Version 2011).
Results
Prospective data acquisiton of intraoperative US data of 100 patients with 116 liver tumors was carried out. All patients underwent liver surgery for tumor disease betwenn October 2010 and March 2016. 65 patients were male, 35 female. Mean age was 60.5 years (2–81 years SD±12.8 years). The size of liver lesions ranged from 0.5 to 8,8 cm (mean 2.42 cm SD±1.44 cm). Postoperative histopathology showed 105 malignant and 11 benign lesions. 35 hepatocellular carcinoma, 10 cholangiocellular carcinoma, 58 metastasis of different primary cancer vs. 3 complicated cysts, 3 adenoma, 2 cholangioma, one partially thrombosed hemangioma, one granuloma and one dystrophic fibrosis were found.
Qualitative analysis was performed by reading of elastographic color pattern. Liver lesions were divided into three groups A – C based on the dominant color in SE mapping. 66 lesions showed predominantly blue coded areas (>50%) suggesting the tumor to be indurated and therefore malignant. These lesions were scored into Group A. Group B consisted of 37 liver tumors with inhomogenous elastographic pattern. These lesions were considered either benign or malignant depending on the impression of B-mode findings.13 liver lesions showed a predominantly red-yellow-green color pattern suggesting the lesions to be soft and therefore benign. Results are shown in Table 1.
Semi-quantitative analysis of Io-SE by Q-analysis showed intratumoral indurations of 3.2 on average in ROC-analysis (range 0.54–5.83 SD±1.19). The surrounding liver tissue showed an average of 3.4 in Q-analysis, the values ranging from 1.8–5.98 (SD±1.04). 76 out of 105 malignant lesions showed central tumor indurations of 2,5 or higher. 29 histologically proven malignoma showed no central indurations suggesting the lesions to be benign. 4 out of 11 benign lesions showed central indurations too (>2.5) and were elastographically considered malignant. Two lesions with high stiffness values in Q-analysis but Group C – color pattern were correctly characterized as hepatic adenoma and benign cyst, but B-mode was necessary for characterisation. 5/11 benign lesions showed no intratumoral indurations (<2.5). The surrounding liver tissue displayed high stiffness values in 82 cases (mean >2.5). In 34 cases no significant indurations of liver tissue were detected (<2.5). Table 2 contains an overview of these results.
Using t-test for paired samples there was found a difference according to tendency between malignant and benign liver lesions (p = 0.06, 95% CI), but no significant differences were found neither between stiffness of tumors of different etiologies nor between intratumoral and “surrounding tissue“ stiffness values (p = 0.33, 95% CI). After application of ROC-analysis and Youden index a cut-off value of 2.5 for detection of malignoma was assumed resulting in a sensitivity of Io-SE of 72.4%. Specificity was 63.6% respectively.
103 out of 105 malignant liver lesions showed a contrast enhancement suspicious of malignancy. Two lesions (1,9%) could not definitely be characterized by Io-CEUS as malignant. Differential diagnostic possibilities included atypical hemangioma or cyst and hepatocellular carcinoma in both cases. Three out of 11 histologically proven benign lesions (partially thrombosed hemangioma, fibrosis, granuloma) were wrongly considered malignant by Io-CEUS. Eight benign lesions could correctly be characterized as benign by Io-CEUS. Sensitivity of Io-CEUS for detection of malignant liver lesions was 98% respectively, specificity 72.7%. These results are summarized in Table 3.
Discussion
In this study, 100 patients with 116 FLLs were examined intraoperatively using strain elastography and Io-CEUS on the liver surface. Dignity of the liver lesions was assessed. Assuming a cut-off value of 2.5 for malignant liver lesions sensitivity of SE was 72.4% respectively. Using Io-CEUS sensitivity of 98% could be obtained aside from two false negative results. These results suggest an advantage of Io-CEUS compared to Io-SE for dignity assessment of FLLs.
Strain elastograpy for dignity assessment of liver lesions has already been studied. However, the majority of studies was purely limited to a qualitative SE analysis of color pattern and different groups of pattern would be defined depending on the tumor etiology [14]. Sensitivity rates up to 95% were reported. Only a few studies provided results of semi-quantitative SE measurements. Onur et al. and Cesario et al. examined malignant and benign liver lesions using semi-quantitative SE. In comparison to the results of this study, both groups reported statistically significant differences in stiffness measurements of benign and malignant FLLs. However, this might be due to this study’s limitation of having a small patient group with benign lesions [3, 17].
Onur et al. reported an average stiffness value of 2.82 for malignant and 1.45 for benign FLLs in semi-quantitative SE. In this study a marginally lower cut-off value of 2.5 was found for differentiation between malignant and benign lesions. It must be noted though that the surrounding liver tissue in this trial displayed a mean value of 3.44 and was evaluated as indurated therefore, so the different results need to be considered relatively [17].
Io-US using CEUS is considered a standard procedure for an optimal curative liver resection in patients with FLLS [4, 22]. Findings of Io-CEUS of this study showing a sensitivity of 98% for detection of hepatic malignoma are consistent with the findings of current literature. In 2013 Friedrich-Rust et al. published sensitivity rates of 79–100% for detection of malignant liver lesions in percutanous CEUS. This metaanalysis reviewed 45 trials [7]. A large trial of the German Society for Ultraosund in Medicine (DEGUM) including more than 1300 patients reported sensitivity rates for CEUS of 95.8% confirming the already mentioned findings [19]. Furthermore Loss et al. could verifiy the importance of Io-CEUS for detection of more, preoperatively unknown liver lesions leading to relevant therapeutical changes [11, 15].
This advantage of CEUS over SE appears to be of high importance as SE is less suitable for a fast screening of the whole mobilized liver due to methodology. However, SE offers the advantage of being a non-invasive method independent of any contrast agents which is highly available on most US machines.
In this study Io-CEUS failed to characterize two hepatocellular carcinoma correctly as malignant (1,9%). Differential diagnostic possibilities included atypical hemangioma and cyst. As these lesions were resected anyway due to intended hemihepatectomy postoperative histology examination could be performed and proved malignancy in both cases. Failure of Io-CEUS had no therapeutical relevance. The findings are consistent with findings of Bernatik et al. who discussed problems of CEUS in another DEGUM multicenter trial [2]. In their study CEUS classified 39/1349 (0.9%) lesions wrong. Compared to this study’s results the rate of false diagnoses of Bernatik et al. was slightly lower, but the variable sizes of patients included must be considered.
Main limitation of this study was that the patient groups (benign and malignant tumors) varied greatly in size due to strict indications for resection of benign liver lesions. Therefore bigger studys with higher case numbers or multicenter trials are needed. Furthermore, no standarized procedure has been established for SE application so a quality marker of seven green marks only allowed a comparison of all included patients. An inter-institutional comparison of data, e.g. for follow-up, is limited. As strain elastography was the only available method on the used US machine for this analysis more suitable or comparable methods such as ARFI etc. could not be applied.
In sum, results of this study show an advantage of Io-CEUS compared to Io-SE in detection and characterisation of liver tumors during surgery. However, these findings lead to the idea of combined use of both methods so advantages of both can further improve surgical approach.
In addition Wiggermann et al. proved the benefit of percutanous strain elastography in monitoring radiofrequency ablation of hepatic tumors [20]. As there has been an increasing demand in intraoperative tumor ablations in addition to surgical resections, SE could be a valuable method for intraoperative monitoring of ablation as no contrast agent application is needed.
Recent innovations in CEUS were reported by Hackl et al. Tested in mice targeted contrast agent for liver metastases of a human colon cancer cell line could detect lesions >0.8 mm [9]. In clinical applications of targeted CEUS smallest lesions <10 mm could be detected eventually leading to a change in therapy. Individual patient therapy could be improved and curative liver tumor surgery could be optimized.
Conclusion
Strain elastography is a valuable and effective, non-invasive method for intraoperative characterisation of FLLs and assessment of dignity using described criteria. Central tumor indurations of 2.5 and higher were considered as suspicious of malignancy as shown by own ROC-analysis. In this study routinely used CEUS during liver surgery proved to be superior to SE showing a higher sensitivity in detecting and characterizing malignoma. In conclusion, routine use of strain elastography in addition to well established and surgically favored CEUS should be considered for optimization and indiviual adaption of elective liver surgery.