Pharmacokinetic analysis of different contrast agents on multiphase enhanced MRI for microvascular invasion: preoperative prediction in hepatocellular carcinoma

Abstract

Background

The preoperative diagnosis of microvascular invasion (MVI) for the solitary small hepatocellular carcinoma (sHCC) is crucial for the decision of surgical strategies.

Purpose

To compare the kinetic parameters and diagnostic effects of two contrast agents for preoperatively predicting MVI of sHCC on multiphase enhanced magnetic resonance imaging (MRI).

Material and Methods

Two groups of patients with known solitary sHCC underwent an enhanced MRI examination before hepatic resection: Data A (n = 61) patients underwent Gd-EOB-DTPA-enhanced MRI, and Data B (n = 41) patients had a normal contrast agent. The two sets of data were processed in the same way. Arterial peritumoral enhancement measured from multiphase enhanced MRI was analyzed using quantitative kinetic parameters, including initial signal enhancement (SE₁), peak signal enhancement (SE_peak), and calculation of the signal enhancement ratio (SE_R).

Results

The statistical analysis showed that the average SE₁ and SE_R (Data A) for the MVI-positive group were significantly higher (P < 0.05) than those in the MVI-negative group. The SE_R (Data B) and SE_peak showed no significant difference for either group. In Data A, the receiver operating characteristic analysis between the two groups had an area under the curve of 0.74 and 0.71 for SE₁ and SE_R, respectively, which was higher than that of Data B. The different contrast agents had the same enhancement curve trend.

Conclusion

Gd-EOB-DTPA-enhanced MRI had a better quantitative kinetic parameter analysis effect for arterial peritumoral enhancement on predicting MVI of sHCC in clinical practice.

Keywords

Hepatocellular carcinoma microvascular invasion semi-quantitative kinetic parameters different contrast agents multiphase enhanced MRI

Introduction

Hepatocellular carcinoma (HCC) is the second leading cause of cancer-related death worldwide (1). Vascular invasion is considered an important prognostic factor for tumor metastasis (2). Usually, macrovascular invasion (MVI) can be detected on imaging, such as contrast-enhanced computed tomography (CT) or magnetic resonance imaging (MRI) (3). By comparison, MVI can be difficult to detect on preoperative imaging, especially for small hepatocellular carcinoma (sHCC), which has been defined as a special type of HCC with a maximum tumor diameter ≤3 cm (4,5). The presence of MVI requires the use of a larger margin during anatomic resection for sHCC, as it results in poorer survival and local disease-control rates (6,7).

An accurate preoperative estimation of MVI presence can help surgeons choose appropriate surgical procedures for patients with sHCC based on their risk-benefit assessment (8). The preoperative diagnosis of MVI is crucial for the decision of surgical strategies (9). Dynamic contrast-enhanced magnetic resonance imaging (MRI) has a higher sensitivity in detecting HCC (81%) compared with CT (68%) and is widely used for preoperative imaging (10). Several studies have reported that certain imaging findings on multiphase enhanced MRI are useful for predicting MVI in HCC, including tumor size (7), peritumoral enhancement (3), and tumor margin (11). The latest articles recommend Gd-EOB-DTPA (11) as a hepatic special contrast agent to preoperatively diagnose the MVI of HCC by peritumoral hypointensity (12), radiological capsule (10,13), and the tumor margins in the axial and coronal hepatobiliary phases (11,14).

However, in all these studies, imaging features were extracted visually, which was limited by visual image grayscales (15), the lack of a pharmacokinetic analysis of the tumor, and peritumoral enhancement. A pharmacokinetic analysis has a high degree of visualization of quantitative indicators and is widely used in the breast, prostate, limbs, etc. (16). With traditional dynamically enhanced MRI, it is not easy to collect and analyze the liver blood status because the liver has simultaneous blood supply from veins and arteries (17). In this context, we applied a pharmacokinetic, semi-quantitative analysis to predict MVI and compared the difference between Gd-EOB-DTPA and extracellular contrast media (ECCM) in detail for the preoperative prediction of MVI in sHCC (18). To our knowledge, this is the first study to quantitatively preoperatively predict solitary sHCC with MVI based on a kinetic parameter analysis of multiphase enhanced MRI. This method provides new data for the preoperative diagnosis of MVI.

Material and Methods

Patients

This retrospective study was approved by the institutional review board at Harbin Medical University Cancer Hospital, and the requirement for informed consent was waived. Our pathology database was searched for all patients who underwent elective hepatectomy of HCC from February 2018 to October 2019. The inclusion criteria for our study were as follows: (i) histopathologically proven primary single sHCC (≤3 cm); (ii) enhanced MRI of the liver performed within one week before surgery; (iii) no macroscopic vascular invasion on MRI; (iv) no treatment before curative hepatectomy; that is, transarterial chemoembolization or radiofrequency ablation. Finally, Data A (patients who underwent a Gd-EOB-DTPA-enhanced MRI examination before surgery) comprised 61 patients (31 men, 30 women; mean age = 56 years) with known HCC enrolled in the study, and for comparison, Data B (patients who underwent a normal contrast agent-enhanced MRI examination before surgery) comprised 41 patients (33 men, 8 women; mean age = 54 years).

Pathological examination and MRI acquisition

Tumor size, number, and capsule condition were obtained by gross examination of the specimen. Histological type, differentiation grade, MVI, satellite nodules, and chronic liver disease were obtained under a microscope. The entire tumor was examined for small size HCC (≤3 cm). MVI was defined as the presence of a tumor in the vascular space of the surrounding hepatic tissue line by endothelial cells on microscopy (15).

An MRI examination was performed with a 3.0-T system (Ingenia; Philips Medical Systems, Eindhoven, The Netherlands) for all patients. A 32-channel phased-array coil was used. The scanning scale covered the area from the top to the lower edge of the liver. Each patient underwent a fat-suppressed dynamic three-dimensional volumetric interpolated breath-hold T1-weighted sequence (TR/TE1/TE2 = 3.6/1.32/2.3 ms, slice thickness gap = 5–2.5 mm, field of view [FOV] = 320 × 427 cm, matrix size = 200 × 250) before and after a bolus injection of the contrast agent (Data A: Primovist [Bayer Schering Pharma, Berlin, Germany], 0.1 mmol/kg body weight with a flow rate of 1 mL/s, followed by a 20-mL saline flush; Data B: gadoteric acid [Hengrui Pharmaceutical Company Ltd., Jiangsu, PR China], 0.2 mmol/kg body weight with a flow rate of 2 mL/s, followed by a 20-mL saline flush). The dynamically enhanced sequence included a pre-enhanced phase, arterial phase (20 s), portal venous phase (55 s), equilibrium phase (90 s), and delayed phase (180 s).

Data analysis

For each patient, the tumor region of interest (ROI) was evaluated by a radiologist with 15 years of abdominal diagnosis experience. Multiphase enhanced MRI data were analyzed using MATLAB (MathWorks, Natick, MA, USA) with an in-house software package. The peritumoral region was defined 1 cm away from the tumor margin (19). The peritumoral region in this study was 8 pixels thick, which was automatically dilated at a 1-cm radius from the labeled ROI by the topology algorithm shown in Fig. 1. A pictorial example of the peritumoral ROI algorithm is shown in Fig. 2, in which the same blue region is the tumor ROI and the growing red region represents the peritumoral ROI obtained with the cycling algorithm. Then, all the pixels on the peritumoral region were used for a pharmacokinetic, semi-quantitative analysis.

Fig. 1.

The peritumoral region of interest algorithm.

Fig. 2.

Pictorial example of the peritumoral region of interest algorithm.

For the kinetic analysis, we utilized the time signal intensity curve (SI(t)) from the multiphase enhanced MRI data (15). For each lesion, the average peritumoral region enhancement curve was used to calculate the kinetic parameters. The baseline signal intensity (SI₀) of each pixel was the signal intensity (SI(t)) of the precontrast time point. Three quantitative kinetic parameters, namely, initial signal enhancement (SE₁), peak signal enhancement (SE _peak ), and signal enhancement ratio (SE _R ), were calculated using SI(t) as follows (16):

S E_{1} = (S I_{1} - S I_{0}) / S I_{0},

(1)

S E_{peak} = (S I_{peak} - S I_{0}) / S I_{0},

(2)

S E_{R} = (S I_{1} - S I_{0}) / (S I_{last} - S I_{0}),

(3)

where SI₀, SI₁, SI _peak , and SI _last are the signal intensity at baseline, the first enhancement time point, the peak point, and the last time point, respectively. The detailed data analysis steps are as follows:

Step 1. DICOM data were loaded using MATLAB;

Step 2. The tumor ROI was evaluated by the radiologist;

Step 3. The peritumoral region was automatically dilated at a radius of 1 cm by the topology algorithm;

Step 4. The average time enhancement curve (SI(t)) was calculated from all pixels on the region;

Step 5. Quantitative kinetic parameters SE₁, SE_peak, and SE_R were calculated using Eqs. 1–3.

Student t tests were performed to examine whether there were significant differences between MVI-positive (MVI+) and MVI-negative (MVI–) groups for the calculated quantitative parameters. A receiver operating characteristic (ROC) analysis was performed to evaluate whether calculated parameters could be used to distinguish between MVI-positive and -negative groups. A P value <0.05 was considered statistically significant.

Results

The histopathological results revealed that 19 sHCC were positive for MVI, whereas 42 lesions were negative for MVI in Data A. In Data B, there were 10 MVI+ patients and 31 MVI– patients. The clinical information and histopathological findings are presented in Table 1.

Table 1.

Clinical information and histopathological findings.

Characteristics	Data A			Data B
Characteristics	MVI–	MVI+	P	MVI–	MVI+	P
Age (years)	56 ± 8.03	54 ± 8.68		53.00 ± 9.80	55.80 ± 8.93	0.428
Male/female (n)	29/13	2/17	0.086	25/6	8/2	0.645
Origin of liver disease
Hepatitis B/other (n)	35/7	18/1	0.356	27/4	9/1	0.676
AFP (ng/mL)	174.2 ± 369.5	279.6 ± 492.3	0.065	227.14 ± 371.83	351.40 ± 772.89	0.493
ALT (IU/L)	30.1 ± 29.4	54.7 ± 60.8	0.007	67.68 ± 39.42	68.74 ± 47.60	0.945
AST (IU/L)	33.7 ± 30.6	48.3 ± 41.7	0.104	42.71 ± 23.96	44.90 ± 27.49	0.810
Cirrhosis/none (n)	6/36	2/17	0.687	6/25	3/7	0.479
Child-Pugh A/B (n)	42/0	19/0	ND	29/2	9/1	0.708
Satellite node (n)	0	0	ND	0	0	ND
Histological grade
Good/moderate/poor (n)	7/28/7	2/11/6	0.732	6/23/2	1/9/0	0.529
Size	2.42 ± 069	2.57 ± 0.52	0.405	2.20 ± 0.74	2.75 ± 0.35	0.030

Values are given as n or mean ± SD.

AFP, alfa-fetoprotein; ALT, alanine aminotransferase; AST, aspartate aminotransferase; MVI, microvascular invasion; ND, not done; SD, standard deviation.

The ROIs of the peritumoral region for the different contrast agents are shown in Fig. 3. The red line indicates MVI+ patients, while the blue line indicates MVI– patients. There were no significant differences between the contrast agents based on a single image, so it made sense to quantify the nuances of pharmacokinetics. In the present study, the two groups were obtained from different patients, not different examinations of the same patient, so statistical comparisons of significant differences could not be made. Therefore, the area under the curve (AUC) was used to compare trends in the pharmacokinetic quantitative parameter diagnostic capability.

Fig. 3.

Peritumoral ROI examples. The red line indicates MVI positive, while the blue line indicates MVI negative. MVI, microvascular invasion; ROI, region of interest.

Fig. 4 shows boxplots of parameters for both MVI– and MVI+ groups: for Data A, the mean (± standard deviation [SD]) SE₁ was 0.36 ± 0.12 for the MVI– group and 0.46 ± 0.09 for the MVI+ group (Fig. 4a), SE_peak was 0.78 ± 0.24 for the MVI– group and 0.74 ± 0.18 for the MVI+ group (Fig. 4b), and SE_R was 0.42 ± 0.20 for the MVI– group and 0.56 ± 0.17 for the MVI+ group (Fig. 4c). The statistical analysis showed that the average SE₁ and SE_R for the MVI+ group were significantly higher (P < 0.05) than in the MVI– group. In Data B, the mean (± SD) SE₁ was 0.43 ± 0.23 and 0.69 ± 0.43 (Fig. 4d), SE_peak was 1.12 ± 0.40 and 1.30 ± 0.39 (Fig. 4e), and SE_R was 0.55 ± 0.30 and 0.68 ± 0.54 (Fig. 4f) for the MVI– and MVI+ groups, respectively. The statistical analysis showed that the average SE₁ for the MVI+ group was significantly higher (P < 0.05) than that in the MVI– group.

Fig. 4.

Box plots of parameters for both negative and positive groups with Data A and Data B. (a) SE₁, (b) SE_peak, and (c) SE_R for Data A; the average SE₁ and SE_R for the MVI– group were significantly lower (P < 0.05) compared to the MVI+ group. (d) SE₁, (e) SE_peak, and (f) SE_R for Data B; only the average SE₁ for the MVI– group was significantly lower (P < 0.05) than that of the MVI+ group. MVI, microvascular invasion.

To compare the different contrast agents for MVI, all the parameters are shown in Fig. 5. Data A and Data B showed similar trends, including the mean SE₁ and SE_R for the MVI+ group being significantly higher than in the MVI– group. The signal for Data A continued to grow, taking a longer time to peak compared to Data B. The negative group parameter values were similar to that of liver parenchyma. However, there was a great difference in the MVI+ group. The SE_peak in the MVI+ group was slightly lower than that of the MVI– group and liver parenchyma in Data A, but in Data B, the SE_peak of the MVI+ group was significantly higher than that of the MVI– group and liver parenchyma.

Fig. 5.

Parameter values of Data A and Data B for comparison. Although the different contrast agents showed a similar trend, Data A (Gd-EOB-DTPA-enhanced) was flatter, which may be affected by the slower injection rate.

ROC analyses were further performed to evaluate the diagnostic performance of the abovementioned significant quantitative parameters for predicting MVI. The AUCs, sensitivities, specificities, and positive and negative predictive values are detailed in Table 2. For Data A, the AUCs were 0.75 and 0.71 and the cutoff values were 0.42 and 0.54 for SE₁ and SE_R, respectively, while the corresponding AUCs were 0.71 and 0.62 with cutoff values of 0.65 and 0.57, respectively, for Data B (Fig. 6).

Fig. 6.

Comparison of the ROC curves. For Data A, by selecting the Youden index cutoff points (0.42 for SE₁ and 0.54 for SE_R), the calculated sensitivity was 0.68 and 0.58 and specificity was 0.74 and 0.76 for SE₁ (blue point) and SE_R (red point), respectively, and the corresponding AUCs were 0.75 and 0.71. For Data B, by selecting the Youden index cutoff points (0.65 for SE₁ and 0.57 for SE_R), the calculated sensitivity was 0.60 and 0.80 and specificity was 0.87 and 0.61 for SE₁ (blue point) and SE_R (red point), respectively, and the corresponding AUCs were 0.72 and 0.62. Therefore, Data A, with Gd-EOB-DTPA-enhanced imaging, had a better diagnosis for MVI. AUC, area under the ROC curve; MVI, microvascular invasion; ROC, receiver operating characteristic.

Table 2.

Diagnostic performance of the valuable parameters of quantitative analyses.

Data	Parameter	AUC	95% CI	Sensitivity	Specificity	PVP	NVP	Cutoff value
A	SE₁	0.75	0.62–0.85	0.68	0.74	0.54	0.84	0.42
A	SE_R	0.71	0.58–0.82	0.58	0.76	0.52	0.80	0.54
B	SE₁	0.71	0.55–0.84	0.60	0.87	0.60	0.87	0.65
B	SE_R	0.62	0.45–0.76	0.80	0.61	0.40	0.90	0.57

AUC, area under the curve; CI, confidence interval; NVP, negative predictive value; PPV, positive predictive value.

Discussion

It has recently been reported that radiological findings with current diagnostic imaging are useful for predicting MVI of HCC (15). However, the relationship between MVI of sHCC and the pharmacokinetic parameters of enhanced MRI have not been evaluated in detail. In the present study, we used a semi-quantitative analysis of kinetic parameters for preoperatively predicting MVI in patients with sHCC based on the peritumoral regions of different contrast agent-enhanced MRI images. Our results compared the pharmacokinetic parameters of an ECCM and a hepatospecific agent. To the best of our knowledge, this was the first study to establish a pharmacokinetic model of different contrast agents for the MVI prediction of sHCC. To ensure the accuracy of the data processing, some rules must be followed. First, an accurate ROI was important. A previous study (19) showed that >85% of MVI was found in the peritumoral region within 1 cm of the tumor boundaries. That was why the defined peritumoral region in our study was obtained by dilating the annotated tumor region at a radius of 1 cm and why a topology algorithm was used to eliminate errors in human subjective delineation. Second, the operation of the two groups of data had identical scan parameters, such as the dynamic phases, FOV, and voxel, except for the contrast agent and the injection speed, which effectively ensures the comparability.

Our results demonstrated that the kinetic analysis of peritumoral enhancement measured from dynamically enhanced MRI could be useful for quantitatively evaluating the MVI of sHCC. The key points are as follows. First, MVI is a histological feature of HCC related to aggressive biological behavior, and HCC is also characterized by angiogenesis (20). The parameters SE₁ and SE_R were higher in the MVI+ group than in the MVI– group, which was caused by tumor neovascularization, suggesting that the ROI located on the tumor was peripheral to the tumor presence. Second, the MVI– group parameter values were similar to that of liver parenchyma in both groups, indicating that there was no abnormal blood supply around the tumor. Third, the reason there was no statistically significant association between SE_peak and MVI might be that we did not collect enough time points, only four phases, so we could not truly obtain the true pharmacokinetic curve to reach the real peak, which requires further study.

In addition to these similarities in the kinetic analysis conclusions among the different contrast agents, there were some differences as well. First, a trend chart of quantization parameters (Fig. 4) showed that the hepatospecific agent curve was flatter than that of the ECCM. The reason was that the flow velocity of Data A was 1.0 mL/s, lower than that of the extracellular contrast agent (Data B) with a pressure 2.0 mL/s, which could lead to blood flow volume and signal intensity differences between them. Second, Data A had a larger AUC in Fig. 5, so the gadoxetate-enhanced liver MRI should be recommended for predicting MVI of HCC.

The statistical analysis showed that the SE₁ and SE_R of Gd-EOB-DTPA and the SE₁ of ECCM were related to MVI. Through the ROC curve, SE₁ had higher diagnostic efficiency, which shows that the arterial phase early enhancement rate of diagnosing MVI makes more sense, indicating that it could be used as the doctors’ key observation sequence for early MVI diagnosis. This was consistent with a previous article, which proposed that the arterial halo sign could predict MVI, but the differences distinguished by the naked eye are relatively rare; therefore, we needed computer-aided quantitative analysis methods to enhance our confidence in diagnosing MVI.

The present study has some limitations. First, the scale of our research was relatively small, especially the MVI+ group. Second, the contrast agent injection used different rates and concentrations in this study. In addition, we only analyzed the maximum slice of a cross-sectional area of the tumor, and the volume of interest should also be predicted in the future. Finally, our developed techniques were not tested on different datasets to evaluate the sensitivity and specificity obtained in this study.

In conclusion, there is a potential clinical application in using the kinetic analysis of dynamically enhanced MRI to predict MVI, and Gd-EOB-DTPA was more effective than the ordinary contrast agent in predicting MVI. The SE₁ parameter could be used as a new index for the quantitative evaluation of MVI.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received the following financial support for the research, authorship, and/or publication of this article: This research was funded Harbin Medical University Cancer Hospital Haiyan Funds (NO. JJZD2020-17).

ORCID iD

Zehong Lin

References

Lee

Park

, et al. Hepatocellular carcinoma: texture analysis of preoperative computed tomography images Can provide markers of tumor grade and disease-free survival. Korean J Radiol 2019;20:569–579.

Chen

S-L

Xiao

Xie

Z-L

, et al. The presence of microvascular invasion guides treatment strategy in recurrent HBV-related HCC. Eur Radiol 2020;30:3473–3485.

Kim

M-J

Jeon

. Prediction of microvascular invasion of hepatocellular carcinoma: usefulness of peritumoral hypointensity seen on gadoxetate disodium-enhanced hepatobiliary phase images. J Magn Reson Imaging 2012;35:629–634.

Huang

Liao

, et al. Prediction of microvascular invasion in hepatocellular carcinoma: preoperative Gd-EOB-DTPA-dynamic enhanced MRI and histopathological correlation. Contrast Media Mol I 2018;2018:9674565.

Kim

. Can microvessel invasion of hepatocellular carcinoma be predicted by pre-operative MRI? Eur Radiol 2009;2009:1744–1751.

Imai

Beppu

Nakayama

, et al. Preoperative prediction of poorly differentiated components in small-sized hepatocellular carcinoma for safe local ablation therapy. J Surg Oncol 2009;2009:121–126.

Yamashita

Y-I

Ima

Yusa

, et al. Microvascular invasion of single small hepatocellular carcinoma ≤3 cm: predictors and optimal treatments. Ann Gastroenterol Surg 2017;2018:197–203.

Chen

Zhao

, et al. Microvascular invasion (MVI) is a poorer prognostic predictor for small hepatocellular carcinoma. BMC Cancer 2014;14:1471–2407.

Imai

Yamashita

Y-I

Yusa

, et al. Microvascular invasion in small-sized hepatocellular carcinoma: significance for outcomes following hepatectomy and radiofrequency ablation. Anticancer Res 2018;38:1053–1060.

10.

Witjes

CDM

Willemssen

FOEJA

Verheij

, et al. Histological differentiation grade and microvascular invasion of hepatocellular carcinoma predicted by dynamic contrast-enhanced MRI. J Magn Reson Imaging 2012;36:641–647.

11.

Ariizumi

S-I

Kitagawa

Kotera

, et al. A non-smooth tumor margin in the hepatobiliary phase of gadoxetic acid disodium (Gd-EOB-DTPA)-enhanced magnetic resonance imaging predicts microscopic portal vein invasion, intrahepatic metastasis, and early recurrence after hepatectomy in patients with hepatocellular carcinoma. J Hepatobiliary Pancreat 2011;2011:575–585.

12.

Huang

Liang

Yang

, et al. Value of MR histogram analyses for prediction of microvascular invasion of hepatocellular carcinoma. Medicine (Baltimore) 2016;95:e4034.

13.

Zhu

Yang

, et al. Incomplete tumor capsule on preoperative imaging reveals microvascular invasion in hepatocellular carcinoma: a systematic review and meta-analysis. Abdom Radiol (NY) 2019;44:3049–3057.

14.

Chou

C-T

Chen

R-C

Lin

W-C

, et al. Prediction of microvascular invasion of hepatocellular carcinoma: preoperative CT and histopathologic correlation. AJR Am J Roentgenol 2014;203:W253–W259.

15.

Feng

S-T

Jia

Liao

. Preoperative prediction of microvascular invasion in hepatocellular cancer: a radiomics model using Gd-EOB-DTPA-enhanced MRI. Eur Radiol 2019;29:4648–4659.

16.

Zhou

Zhang

, et al. Quantitative analysis of lower leg muscle enhancement measured from dynamic computed tomographic angiography for diagnosis of peripheral arterial occlusive disease. J Comput Assist Tomogr 2020;44:20–25.

17.

Morelli

Michaely

Meyer

, et al. Comparison of dynamic and liver-specific gadoxetic acid contrast-enhanced MRI versus apparent diffusion coefficients. PLoS One 2013;8:e61898.

18.

Schalkx

van Stralen

Coenegrachts

, et al. Liver perfusion in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI): comparison of enhancement in Gd-BT-DO3A and Gd-EOB-DTPA in normal liver parenchyma. Eur Radiol 2014;24:2146–2156.

19.

Roayaie

Blume

Thung

, et al. A system of classifying microvascular invasion to predict outcome after resection in patients with hepatocellular carcinoma. Gastroenterology 2009;137:850–855.

20.

Chen

Z-H

Zhang

X-P

Wang

, et al. Effect of microvascular invasion on the postoperative long-term prognosis of solitary small HCC: a systematic review and meta-analysis. HPB (Oxford) 2019;2019:935–944.