Abstract
Background/Aim: To investigate the accuracy of contrast-enhanced ultrasound (CEUS) in differentiating benign and malignant portal vein thrombosis (PVT) complicating hepatocellular carcinoma (HCC), compared to diffusion-weighted magnetic resonance imaging (DWI). Patients and Methods: Forty-nine patients with HCC who had PVT were enrolled. The quantitative and qualitative parameters of CEUS were analysed. We examined the diagnostic performance of CEUS compared with DWI. The relationships between CEUS parameters and biomarkers were also assessed. Results: All qualitative CEUS parameters (e.g., arterial-phase enhancement, washout in the venous phase, vessel occlusion, and expansion) were significantly more common in malignant thrombosis than in benign thrombosis (p<0.05). Among the quantitative CEUS parameters, increased area under the time-intensity curve, prolonged time for full width at half maximum, and prolonged rise time were observed in malignant thrombosis, compared to benign thrombosis (p<0.05). CEUS and DWI performed similarly in discriminating between malignant and benign thrombi (p>0.05). Several CEUS parameters exhibited significant correlations with the tumour marker and stage (p<0.05). Conclusion: CEUS was useful for characterisation of PVT and showed a consistent high diagnostic accuracy, compared to DWI.
Malignant portal vein thrombosis (PVT) is a common complication in patients with concurrent hepatocellular carcinoma (HCC). The presence of malignant PVT in HCC worsens the stage, even for patients with small nodular HCCs. In addition, PVT may newly develop in patients whose HCC has been resected or is fully necrotised. Therefore, the characterisation of PVT is of considerable clinical importance to determine the tumour stage and therapeutic strategy (1). Although the reference standard for PVT is histopathological examination, portal vein biopsy is an invasive procedure with an associated risk of bleeding; the usefulness of this biopsy can be limited due to sampling errors that cause false-negative results. As a result, multiple diagnostic imaging studies, such as ultrasound (US), computed tomography (CT), and magnetic resonance imaging (MRI) are often used for thrombus discrimination in clinical practice (2-4). Because of its superb spatial- and contrast resolution, MRI is one of the best imaging modalities. Diffusion-weighted imaging (DWI) is a magnetic resonance (MR) technology that has been recently applied in abdominal imaging; it measures the microscopic diffusion of water molecules in tissues. Although several investigators have reported the use of DWI in determining the nature of the PVT, their findings have been incongruent (2, 5, 6).
Contrast-enhanced ultrasonography (CEUS) improved visualisation of vascular structures and revealed tiny vessels in tissues. In the recent 15 years, efforts in differentiating benign and malignant PVT using CEUS were made and the results were inspiring (4, 7). Although comprehensive meta-analysis describing the performance of CEUS in the characterisation of PVT has emerged recently, there are no comparative studies focusing on the diagnostic accuracy of different imaging modalities (8). A recent study showed that the diagnostic performance of CEUS was superior to the performances of dynamic CT and MRI for small liver tumours (9). However, there has not been a report comparing the diagnostic ability using a quantitative scale between CEUS and MRI for the assessment of PVT. CEUS could provide the quantitative analysis parameters to interpret the perfusion flow within portal vein lesion, thereby providing a more sophisticated and accurate evaluation of the contrast distribution and increased confidence in characterisation of the thrombus. To our knowledge, this study is the first quantitative analysis of CEUS comparing the diagnostic performance of CEUS versus MRI to assess the nature of the PVT.
The purpose of the present study was to investigate the quantitative time-intensity curve (TIC) parameters and qualitative CEUS characteristics to differentiate between benign and malignant PVT in patients with HCC, compared to the use of DWI.
Patients and Methods
Patients. Fifty-one patients with HCC who exhibited a PVT in liver dynamic CT or MRI were consecutively enrolled at a tertiary referral centre between September 2017 and January 2019. Among these 51 patients, two were excluded due to the presence of isolated PVT without HCC. After diagnosis of PVT using CT or MRI, patients were subsequently assessed with CEUS within 2 weeks of the previous imaging. Patients with contraindications for US contrast agents (e.g., history of cardiac shunt, pulmonary hypertension, or hypersensitivity to contrast agent) were excluded. HCCs were established based on the diagnostic criteria of the European Association for the Study of the Liver. The modified Union for International Cancer Control (mUICC) and Barcelona Clinic Liver Cancer (BCLC) staging systems were used for HCC staging (10, 11). At the time of diagnosis of HCC with PVT, all patients had undergone assessment of tumour markers, such as alpha-fetoprotein (AFP) and protein induced by vitamin K absence or antagonist-II (PIVKA II). The Ethics Committee of our institution approved the study protocol (OC19OESI0017). Written informed consent was obtained from each patient at enrolment.
CEUS imaging technique. The contrast agent used in this study was Sonazoid (GE Healthcare, Oslo, Norway), which consists of a phospholipid shell containing perfluorocarbon microbubbles. The recommended clinical dose for imaging of liver lesions is 0.015 ml of encapsulated gas per kg of body weight. B-mode sonographic scans were obtained to identify PVT using a Philips iU22 unit (Philips Medical Systems, Seattle, WA, USA) with a C5-1 curvilinear transducer. The contrast agent was injected manually as a bolus over 2 s, followed by a 10-ml flush of saline solution. The whole vascular phase consisted of the arterial (10-40 s after the injection), portal venous (60-90 s), and late (3 min) phases. Time zero was defined as the initiation of the saline flush (12). Continuous imaging was recorded for each patient from the beginning of the saline flush and obtained for approximately 5 min at a low mechanical index setting of 0.2. The second examination was performed in the same manner at the Kupffer phase after 15 min. CEUS image data were saved to the US hard disk system and then transferred to a personal computer for further quantitative analyses with advanced US quantification software (QLAB 8.1; Philips Medical Systems).
MR imaging technique. MR imaging of the liver was performed using a 3.0-T magnetic resonance system (Magnetom Skyra; Siemens Healthcare, Erlangen, Germany) with 18-channel body-array coils. The following imaging sequences were performed: unenhanced fast spin-echo-, fat-suppressed single-shot fast spin-echo T2-weighted imaging, and gradient-echo T1-weighted imaging with in-phase and opposed-phase sequences. Thereafter, a dynamic study was performed using T1-weighted 3D multiplanar spoiled gradient-echo volume interpolated breath-hold examinations with fat saturation in the axial plane (section thickness, 2.6 mm; repetition time/echo time, 3.7/1.4; flip angle, 13°; matrix size, 384×512). After unenhanced images had been acquired, gadoxetic acid (Primovist; Bayer Healthcare, Berlin, Germany) was injected at a dosage of 0.1 ml/kg body weight, with a flow rate of 1 ml/s through the antecubital vein, followed by a 20-mlL saline chaser. Three-dimensional gradient-recalled echo imaging was performed in the arterial phase (30-35-s delay with the bolus-tracking technique), portal venous phase (65-80-s delay), transitional phase (180-s delay), and hepatobiliary phase (20-min delay). DWI was performed with echo-planar imaging; b values of 0, 400, and 800 s/mm2 were applied with an apparent diffusion coefficient (ADC) map that was generated automatically.
Analysis of CEUS and MR images. Image analyses were performed quantitatively and qualitatively by one radiologist to ensure consistency of the measurements taken. To compensate for minor breathing artefacts, all sequences underwent motion compensation before the start of the analysis. Time-intensity curves (TICs) were generated from a region of interest (ROI) placed over the lesion and were sufficiently large to encompass the entire thrombus. Respiratory movement artefacts were eliminated by automatic adjustments. The raw data were fitted to a drift-diffusion model. The wash-in slope (i.e., the maximum wash-in velocity of the contrast medium; unit, dB/s), time to peak intensity (i.e., time to maximum enhancement; unit, s), peak intensity (i.e., maximum intensity of the curve; unit, dB), area under the TIC (AUC, area under the TIC that was proportionate to the total volume of blood flow in the ROI; unit, dB), mean transit time (i.e., corresponding to the centre of gravity of the perfusion model; unit, s), time for full width at half maximum (FWHM, time between half amplitude values in each side of the maximum; unit, s), and rise time (RT, time from injection until the peak of enhancement; unit, s) were obtained using QLAB software. For each ROI, the analysis was repeated three times; the mean value of perfusion parameters was obtained to minimise the transitional distance caused by respiration and ensure the accuracy of the analyses. Qualitative CEUS parameters included the presence or absence of arterial-phase enhancement, presence or absence of the venous-phase washout, effect on the vessel (i.e., occlusive or nonocclusive), and presence or absence of vessel expansion. A presumptive CEUS diagnosis was made according to the most recent (2017) version of the CEUS Liver Imaging Reporting and Data System (12).
MR images were evaluated from the picture archiving and communication system by one radiologist, who did not perform the CEUS examination and was blinded to the CEUS results. For qualitative analysis, the reader was asked to mark the signal intensity (SI) of the thrombus on the diffusion weighted (DW) images with a b-value of 800 s/mm2 and to classify the thrombus into one of the following categories: hyperintense, isointense, or hypointense, relative to the liver. For quantitative analysis, oval ROIs, which included at least two-thirds of the thrombus area, were drawn directly on the corresponding ADC image to obtain ADC values.
Reference standards for characterisation of thrombosis. Stringent criteria were defined for assigning a benign or malignant nature to the thrombus. The reference standard included a combination of typical imaging findings or histologic confirmation. For the imaging reference standard, two independent reviewers categorised thrombi as benign or malignant using pre-established criteria, which were commonly used as reference standards and were in agreement with each other (2, 5). The presence of enhancement was used to differentiate malignant thrombi from benign thrombi. Clear evidence of enhancement on CT and MR images during the arterial phase of dynamic imaging was defined by enhancement on contrast-enhanced images when compared with the baseline images (≥20 Hounsfield units on CT images and obvious enhancement on subtraction MR images).
In addition, rapid progression of thrombosis (i.e., within 3 months) on follow-up CT or MR imaging was considered to indicate a malignant thrombus. In benign thrombosis, the thrombi were stable or showed a reduction in their extent; alternatively, recanalisation of occluded vessels was observed, with or without anticoagulation therapy, on follow-up imaging at 12 months.
Statistical analysis. Values are expressed as means±standard deviations and as proportions with percentages, as appropriate. Chi-squared tests or Fisher's exact tests were used to assess categorical data. Independent Student's t-tests or Mann-Whitney U-tests were used to assess continuous data. Diagnostic accuracy was evaluated using receiver operating characteristic curve analysis with a calculation of the area under the receiver operating characteristic curve (AUROC). The optimal cut-off value was chosen according to the Youden index. The diagnostic performance of CEUS and DWI in the evaluation of PVT was presented as sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy. Associations between CEUS parameters and clinical markers were assessed using point-biserial correlations, Cramer's V, and Spearman correlations. The intra-class correlation coefficient (ICC) was calculated to evaluate intra-observer variability (13). The ICC was interpreted as follows: <0.6=poor; 0.6-0.79=moderate; >0.8-1=excellent agreement. All statistical analyses were conducted using SAS software (version 9.3, SAS Institute, Cary, NC, USA). A p-value <0.05 was considered statistically significant.
Results
Patient characteristics. In total, 49 patients were included in this study. The baseline characteristics of patients are summarized in Table I. At the time of diagnosis of PVT, 41 of 49 (83.7%) patients had viable parenchymal HCC; eight patients previously had prior HCC but had no viable or remnant HCC, following resection or ablation of HCC. During the median follow-up period of 6.5 months (range=1-25 months), 19 (38.7%) patients survived. Among the 49 patients, 45 had both CEUS and MRI scans, including four patients who had CEUS and CT scans. Based on the CT or MRI findings, there were 14 benign thrombi and 36 malignant thrombi in 49 patients, including one patient with simultaneous benign and malignant PVT. Because of non-interpretable quantification, mainly due to the patient's condition, CEUS results were obtained for 49 thrombi in the first vascular phase imaging and for 44 thrombi in the second Kupffer phase imaging. No side-effects of the sonographic contrast agent were noted during US examination (Figure 1).
Qualitative and quantitative analysis of CEUS. The qualitative and quantitative parameters measured at the PVT are listed in Table II. All qualitative parameters were significantly different between malignant and benign tumours (<0.05). Arterial-phase enhancement, washout in the venous phase, vessel occlusion, and expansion were significantly more common in malignant thrombosis than in benign thrombosis. In both the vascular and Kupffer phases, it was possible to differentiate malignant and benign thrombosis using the following parameters: AUC, FWHM, and RT. In malignant thrombosis, larger AUC and prolonged FWHM and RT were observed, compared to benign thrombosis. In the receiver operating characteristic curve analysis, the following parameters showed increased discrimination or accuracy in both vascular and Kupffer phases: AUC [AUROC 0.782 (0.614-0.949), p=0.001 and AUROC 0.789 (0.607-0.971), p=0.002, respectively] and RT [AUROC 0.761 (0.605-0.917), p=0.001 and AUROC 0.694 (0.507-0.880), p=0.042, respectively] (Table III).
Diagnostic performance of CEUS compared to DWI. The sensitivity, specificity, PPV, NPV, and accuracy of CEUS and DWI for the discrimination between malignant and benign thrombi are shown in Table IV. Since there was no statistically significant difference between CEUS and DWI, the overall diagnostic accuracy of CEUS was similar to that of DWI. There was also a strong correlation between DWI and CEUS (p<0.0001) (Table V). A higher signal intensity on DW images or a lower ADC value tended to occur more often in malignant thrombosis than in benign thrombosis at the presumptive CEUS diagnosis. Representative images are shown in Figures 2 and 3.
There was discrepancy in the assessment of three PVT lesions between MRI and CEUS findings. Among them, one lesion without diffusion restriction on MRI had been judged to benign PVT; their CEUS diagnosis were classified as malignant PVT, due to arterial-phase enhancement and venous phase washout of the contrast agent. During the overall follow-up period, this lesion was progressed to malignant PVT, consistent with the CEUS finding (Figure 4). The remaining two lesions with diffusion restriction on MRI had been judged to be malignant PVT; their CEUS diagnoses were classified as benign PVT, due to lack of arterial-phase enhancement or venous phase washout of the contrast agent. All of these lesions were progressed to malignant PVT, consistent with the MRI findings.
Baseline demographic and clinical characteristics of patients.
Relationship between CEUS parameters and biomarkers (tumour marker or stage). The relationship between CEUS parameters and biomarkers is shown in Table VI. Among qualitative parameters, arterial enhancement, vessel occlusion and vessel expansion were correlated with the mUICC stage in both vascular and Kupffer phases (p=0.013, p<0.0001, and p=0.0001; p=0.001, p<0.0001, and p=0.017, respectively). Arterial enhancement, vessel occlusion and expansion were more often noted in patients with higher mUICC stages than in patients with lower mUICC stages. Among the quantitative parameters, AUC exhibited a statistically significant correlation with PIVKA II in both vascular and Kupffer phases (p=0.004 and p=0.031, respectively). With an increasing PIVKA II level, a larger AUC was more frequently noted. AUC, mean transit time, FWHM, and RT were significantly associated with mUICC stage in both vascular and Kupffer phases (p=0.021, p=0.047, p=0.008, and p=0.030; p=0.005, p=0.003, p=0.001, and p=0.012, respectively). Patients with higher mUICC stage exhibited larger AUC, prolonged mean transit time, FWHM, and RT, compared to patients with lower mUICC stage.
Reproducibility of perfusion parameters. ICCs for quantitative time-intensity curve parameters ranged from 0.32 to 0.89. The ICCs for wash-in slope and RT indicated poor agreement. Other parameters showed moderate to excellent agreement (Table VII).
Discussion
The present study showed that some qualitative and quantitative CEUS parameters could be used to discriminate benign and malignant PVT in patients with HCC. The values for AUC, FWHM, and RT in CEUS findings significantly differed between benign and malignant PVT. However, FWHM showed insufficient accuracy in the receiver operating characteristic curve analysis and RT showed poor intra-observer agreement, contrary to the other parameters. Therefore, based on its consistent reliability, we propose the use of AUC for discrimination of benign and malignant PVT in clinical practice. In addition, among the quantitative CEUS parameters, AUC exhibited the best correlation with clinical markers.
Flow diagram of patient selection.
The ability of CEUS to differentiate between malignant and benign PVT has been previously reported (4, 7, 8, 14, 15). Unfortunately, the prior studies only performed qualitative analyses of CEUS images, which were subjective and prone to interobserver variation. Objective signal quantification has improved the accuracy and sensitivity of CEUS, reducing limitations such as operator dependence (16). US contrast agents remain within the vascular bed and behave in a manner similar to that of red blood cells. Thus, the calculation of tissue perfusion and microvascularity indices is much simpler, compared with agents used for CT and MR imaging that diffuse into the extravascular space (17). Whereas interstitial contrast agents for MRI may show pseudoenhancement, the accurate washout produced by the intravascular US microbubble contrast is a critical feature for distinguishing malignant hepatic masses from benign masses. Thrombi are typically similar in nature to the hepatic masses from which they originate (4, 18). Similarly, in our study, qualitative analysis showed arterial-phase enhancement and venous phase washout of the contrast agent in malignant thrombi. Quantitative analysis showed that AUC, a blood volume-related parameter, was increased in malignant thrombi. Considering that AUC means the total volume of blood flow, this finding indicated that malignant thrombi showed significantly higher regional blood volume than benign thrombi. However, time-related parameter results did not show statistical significance.
We used Sonazoid, a combined blood-pool and Kupffer cell agent. Sonazoid-CEUS has a unique Kupffer phase imaging in addition to vascular imaging, because Sonazoid microbubbles are phagocytosed by Kupffer cells (liver-specific macrophages), enabling parenchyma-specific liver imaging (19). Kudo et al. devised a method termed defect reperfusion US imaging, which confirms the flow of blood into the defect (20). Combined Kupffer and arterial phase images are obtained by Sonazoid reinjection at the Kupffer phase. This innovative method was developed based on the two favorable properties of Sonazoid, namely, the demonstration of real-time vascular images and stable Kupffer phase images tolerable for repeated scanning. Because of information on both intratumoral hemodynamics and reticuloendothelial function in the same cross-sectional image, Sonazoid-CEUS with defect reperfusion US imaging has dramatically changed the management of HCC (21). In our study, the results obtained by Kupffer phase imaging had greater discrimination accuracy and a stronger correlation with the clinical markers than those obtained by vascular phase imaging. Although it is difficult to explain these findings, it can be speculated that information about arterial vascularity, i.e. reinjection method, increases the diagnostic accuracy of PVT, as in HCC.
Qualitative and quantitative analysis of CEUS.
Our findings support the observations of previous studies in which DW signal intensity and ADC measurements were found to reliably distinguish between benign and malignant PVT in patients with HCC; the vascularity of the thrombus, as well as the presence of malignant cells, may have affected both DW signal intensity and ADC values of the thrombi (2, 22-24). Furthermore, our results showed that CEUS findings correlated well with DWI-MRI findings; CEUS findings may be more accurate, especially for the diagnosis of malignant PVT with infiltrative HCC. This result is potentially related to continuous real-time monitoring of contrast diffusion, which provides easy detection of the thrombus and maximum enhancement, compared to DWI, which is affected by molecular diffusion and tissue capillary perfusion.
Benign portal vein thrombosis in a 61-year-old man who presented with nonviable hepatocellular carcinoma (HCC). A filling defect (arrow) is noted in the right portal vein branches to the compact lipiodolised HCC (arrowhead) at segment VI on portal phase (B) contrast-enhanced MRI, which shows no enhancement on the subtraction image (A). The thrombus appears isointense on DWI (b=800 s/mm2) (C) and exhibits a high apparent diffusion coefficient (ADC) value on the corresponding ADC map (D). Compact lipiodolised HCC exhibits a lack of enhancement and no diffusion restriction, indicating a nonviable tumour. (E) Screenshot taken from QLAB illustrating localisation of the region of interest (ROI) and determining the perfusion indices. Three ROIs were drawn on the thrombus. Left portion of the image shows contrast-image mode imaging and right portion shows standard (B-mode) imaging. Bottom: Quantitative time-intensity curve (TIC) curves. Smooth curves are fitted curves, while non-smooth curves are original curves. The y-axis of the time-intensity graph corresponds to the intensity or mean echo (decibels), whereas the x-axis corresponds to the absolute time (s). The TIC curve shows a gradual increase in signal intensity. The thrombus was stable, without anticoagulation therapy, on follow-up CT.
Malignant portal vein thrombosis in a 61-year-old man with infiltrative hepatocellular carcinoma (HCC). Arterial phase contrast-enhanced MRI (A) and subtraction image (B) show infiltrative HCC (asterisk) in the left hepatic lobe and enhancing thrombus in the left portal vein (arrow). The malignant thrombus appears hyperintense on diffusion-weighted magnetic resonance imaging (DWI) (b=800 s/mm2) (C) and isointense to the liver on the corresponding apparent diffusion coefficient (ADC) map (D). (E) Screenshot taken from QLAB illustrating the localisation of the region of interest (ROI) and determining the perfusion indices. The ROI was drawn on the thrombosis. Left portion of the image shows contrast-image mode imaging and right portion shows standard (B-mode) imaging. Bottom: Quantitative time-intensity curve (TIC) curves. Smooth curves are fitted curves, while non-smooth curves are original curves. The y-axis of the time-intensity graph corresponds to the intensity or mean echo (decibels), whereas the x-axis corresponds to the absolute time (s). The TIC curve shows an early peak of enhancement, followed by washout.
A 76-year-old woman with PVT. A filling defect is noted in the main portal vein (arrow) on portal phase (B) contrast-enhanced MRI, which shows no enhancement on the subtraction image (A). The thrombus appears isointense on DWI (b=800 s/mm2) (C) and exhibits a high apparent diffusion coefficient (ADC) value on the corresponding ADC map (D), suggesting benign thrombus. (E) Screenshot taken from QLAB illustrating localisation of the region of interest (ROI) and determining the perfusion indices. Three ROIs were drawn on the thrombus. Left portion of the image shows contrast-image mode imaging and right portion shows standard (B-mode) imaging. Bottom: Quantitative time-intensity curve (TIC) curves. Smooth curves are fitted curves, while non-smooth curves are original curves. The y-axis of the time-intensity graph corresponds to the intensity or mean echo (decibels), whereas the x-axis corresponds to the absolute time (s). The TIC curve shows an early peak of enhancement, followed by washout. CEUS diagnosis was classified as malignant PVT. During the overall follow-up period, this lesion was progressed to malignant PVT.
Significant quantitative time-intensity curve parameters of CEUS for characterization of PVT.
Diagnostic performance for characterization of PVT.
Association between DWI and CEUS.
The relationship between CEUS parameters and clinical finding (tumor marker or stage).
CEUS is a promising tool that can be used as a non-invasive approach without the added risks of ionizing radiation and nephrotoxicity, which would impair renal perfusion and increase the risk of nephrogenic systemic fibrosis (25). Furthermore, DW MRI is a procedure that can often miss a thrombus in small portal venous branches and needs long times of breath-hold acquisitions, sometimes not feasible in cirrhotic patients. Accordingly, CEUS can serve as an alternative tool for the assessment of patients who are unable to undergo MRI. Recent studies suggest that a combination of CEUS and MRI might reduce the rate of false-negative findings in the assessment of focal liver lesions (26). Similarly, the combination of the two imaging modalities is expected to show higher accuracy for diagnosis of PVT.
Our study had a few limitations. First, histopathologic confirmation of the thrombus was available only for a small proportion of patients (2%). While histological assessment is the ideal reference standard to determine the nature of the thrombus, there are practical limitations to the pathologic correlation of the nature of PVT. In addition to the risk of bleeding due to coagulation defects that accompany liver cirrhosis, inadvertent biopsy in cases of infiltrative HCC could lead to misdiagnosis; an insufficient amount of biopsy material could also lead to misdiagnosis. In addition, there is increasing reliance on imaging features for the characterisation of PVT, to replace open laparotomy or percutaneous biopsy. Second, heterogeneity of measurements is a common limitation in all attempts to use CEUS to quantify organ perfusion, as previously reported by Leen et al. (27). To minimise this heterogeneity, CEUS was performed by one radiologist to ensure consistency for all measurements taken and we aimed to locate ROIs at similar depth and distance, as recommended by Averkiou et al. (28). In addition, three ROIs were drawn for each experimental time point, and the results were averaged to minimize heterogeneity of the measurements. A third potential limitation is recall bias, as the reader who performed the retrospective review was also involved in performing CEUS examinations. To minimise this recall bias effect, we ensured a minimum of 6 months' interval between CEUS examinations and subsequent retrospective review.
Intra-class Correlation for intraobserver variability, accounting concordance of quantitative time-intensity curve parameters.
In conclusion, our study found that the AUC, obtained by CEUS, may be a useful parameter for the diagnosis of PVT. CEUS showed a diagnostic accuracy comparable to that of DWI-MRI for PVT.
Acknowledgements
The Statistical consultation was supported by the Department of Biostatistics of the Catholic Research Coordinating Center.
Footnotes
Authors' Contributions
YRS conceived the idea for this study, wrote and edited the manuscript. JHK drafted the manuscript. JHK, SHY and SWN conducted the data acquisition. YRS, JHK and YJL interpreted the data. All Authors read and approved the final manuscript.
Conflicts of Interest
The Authors declare no conflicts of interest regarding this study.
- Received June 11, 2020.
- Revision received July 2, 2020.
- Accepted July 3, 2020.
- Copyright© 2020, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.