DCE-MRI in hepatocellular carcinoma-clinical and therapeutic image biomarker

Abstract

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) enables tumor vascular physiology to be assessed. Within the tumor tissue, contrast agents (gadolinium chelates) extravasate from intravascular into the extravascular extracellular space (EES), which results in a signal increase on T1-weighted MRI. The rate of contrast agents extravasation to EES in the tumor tissue is determined by vessel leakiness and blood flow. Thus, the signal measured on DCE-MRI represents a combination of permeability and perfusion. The semi-quantitative analysis is based on the calculation of heuristic parameters that can be extracted from signal intensity-time curves. These enhancing curves can also be deconvoluted by mathematical modeling to extract quantitative parameters that may reflect tumor perfusion, vascular volume, vessel permeability and angiogenesis. Because hepatocellular carcinoma (HCC) is a hypervascular tumor, many emerging therapies focused on the inhibition of angiogenesis. DCE-MRI combined with a pharmacokinetic model allows us to produce highly reproducible and reliable parametric maps of quantitative parameters in HCC. Successful therapies change quantitative parameters of DCE-MRI, which may be used as early indicators of tumor response to anti-angiogenesis agents that modulate tumor vasculature. In the setting of clinical trials, DCE-MRI may provide relevant clinical information on the pharmacodynamic and biologic effects of novel drugs, monitor treatment response and predict survival outcome in HCC patients.

Key Words: Dynamic contrast-enhanced magnetic resonance imaging; Perfusion magnetic resonance imaging; Hepatocellular carcinoma; Angiogenesis inhibitors; Clinical trials

Core tip: Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) enables tumor vascular physiology to be assessed. Within the tumor tissue, contrast agents extravasate from intravascular into the extravascular extracellular space, which results in a signal increase on T1-weighted MRI. These signal intensity-time curves can be deconvoluted by mathematical modeling to extract parameters that may reflect tumor angiogenesis. DCE-MRI allows us to produce highly reproducible parametric maps of quantitative parameters in hepatocellular carcinoma (HCC). In the setting of clinical trials, DCE-MRI may provide relevant clinical information of novel drugs, monitor treatment response and predict survival outcome in HCC patients.

INTRODUCTION

Hepatocellular carcinoma (HCC) is the sixth most common tumor and represents the third leading cause of cancer death worldwide[1]. It is the major cause of death in the cirrhotic patients, beside complications from portal hypertension[2]. The hypervascular nature and vascular in-out flow pattern of this tumor help differentiation of HCC from other tumors by non-invasive diagnostic criteria, without the necessity of tissue proof[3]. Like other malignant tumors, previous researches have reported the importance of angiogenesis with the development of HCC[4-8]. For example, Yamaguchi et al[9] found that vascular endothelial growth factor (VEGF) expression in HCC tissues may be related to the histological grade. Thus, various angiogenesis inhibitors have been developed to treat HCC. Among them, the multikinase inhibitor sorafenib was first approved and validated by two separate phase III trials conducted in Western and Asian countries, respectively[10,11]. Up to December 2013, there are at least 20 active phase III trials evaluating systemic treatments for advanced HCC (from clinicaltrials.gov-last visit 20^th December 2013), with most of studies using sorafenib as combination therapy.

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) have been used widely as biomarkers in many early phase clinical trials to evaluate the effects of anti-angiogenic drugs that modulate tumor vasculature[12-16], and to help effective drug selection and optimal drug dose decision[17]. For phase III clinical trials, DCE-MRI can serve as a surrogate biomarker to evaluate drug efficacy before the volumetric change of the tumor[18], and may be associated with progression-free survival and/or overall survival in these patients[19,20]. The enhancement patterns in HCC obtained by DCE-MRI are influenced by tumor angiogenesis and correlated with tumor microvessel density and VEGF expression[21]. Thus, suppression of tumor vascular permeability induced by anti-angiogenic agents can be reliably detected and quantified by DCE-MRI. Besides assessing anti-angiogenic agents, DCE-MRI can also be used in the evaluation of response of HCC after other treatments, including transarterial chemoembolization[22] and radiotherapy[23]. This review will attempt to summarize the current clinical application of DCE-MRI for HCC patients.

BASIS OF DCE-MRI

DCE-MRI images are obtained by injecting low-molecular-weight gadolinium chelated contrast agent into a vein with a constant rate[24]. The contrast agent is carried by blood flow into the tissue, causing increased signal intensity (SI) of the T1-weighted images due to the shortening of the longitudinal relaxation time of the tissue[25]. Within the tissue, the contrast agent passes from the arteries to the capillaries, and then extravasates to the extravascular extracellular space (EES). The rate of contrast agent extravasation to EES in the tumor tissue is determined by vessel leakiness and blood flow. Thus, the signal measured on DCE-MRI represents a combination of permeability and perfusion. DCE-MRI is sensitive to alterations in vascular permeability, extracellular space, and blood flow. To ideally record the signal change in the supplying blood vessel and within the tumor, a fast injection rate of the contrast agent captured with high temporal resolution is required[26,27].

This signal enhancement of liver perfusion can be quantified either with a semi-quantitative or quantitative analysis. The semi-quantitative analysis is based on the calculation of heuristic parameters that can be extracted from SI curves. In contrast, the quantitative analysis needs computational-based curve fitting algorithms using a bi-compartmental model with arterial input function. The parameters from both analysis methods have been shown to present correlation with tumoral angiogenesis[28].

SEMI-QUANTITATIVE ANALYSIS

Regarding the semi-quantitative analysis, different parameters that characterize the shape of the normalized SI-time curve can be extracted: (1) area under curve (AUC): expresses the amount of enhancement over a defined period of time (usually from starting increment of the SI-time curve to 60 or 90 s); (2) maximum of SI or Peak enhancement ratio (SI_maximun-SI_baseline/SI_baseline) of the enhancing curve; (3) wash-in Slope: determines the velocity of enhancement. It is calculated as the maximum change in enhancement per unit time, usually from 20% to 80% range of the increment curve; and (4) mean transit time (MTT): represents the mean time for blood to perfuse a region of tissue and is affected by the blood volume and blood flow in the region under analysis.

The semi-quantitative analysis is widely used because it is easy to calculate without the need of modeling. However, these heuristic parameters are highly affected by the gain factor of the acquisition systems, contrast media volume and injection rate, because the true concentration of contrast agent in the tissues is not estimated. Thus, differences in temporal resolution and injection rates can easily change the shape of SI curves, making comparison and quantification difficult[26,29]. Moreover, these descriptive parameters provide no physiologic insight into the behavior of the tumor vessels.

QUANTITATIVE ANALYSIS

On the other hand, the quantitative analysis is based on modeling the concentration change of the contrast agent using pharmacokinetic modeling techniques[30]. An initial conversion step of SI to concentration values is needed. Concentration vs time curves are then fitted using a bi-compartmental model (vessels and EES) with two vascular inputs (aorta and portal vein). The following parameters can be derived from a mathematical model[26,31]: (1) K^trans (forward volume transfer constant): determines the flux of the contrast agent from the intravascular space to the EES. It predominantly represents the vascular permeability in a permeability-limited (high flow) situation, but represents the blood flow into the tissue in a flow-limited (high permeability) situation; (2) K_ep (reverse reflux rate constant): expresses the return process of the contrast agent from the EES to the intravascular space; and (3) V_e (volume fraction of EES): an indirect measure representing the cellular density of the tissue.

These parameters require additional calculations to generate parametric maps obtained after a pixel-by-pixel curve fitting process of the region under analysis. Thus, they are more computationally technical to obtain than the semi-quantitative ones. After generating parametric maps, the mean or median values within region of interests are usually calculated to represent tumor microvasculature, but histogram analysis[32] or heterogeneity in parametric maps[33-35] may also provide additional information. For optimum parameter quantification, a high temporal resolution is required to record initial rapid uprising of the SI curve immediately after the contrast agent administration[36]. The accuracy of these parameters is influenced by curve fitting algorithms[37,38] and magnitude of motion artifacts[39].

MODEL SELECTION

Kety[40] first described the flow-limited tracer uptake in tissue, and since then several pharmacokinetic models have been proposed by Tofts et al[41], Brix et al[42] and Larsson et al[43]. All these models used single source of arterial input function. Because HCC receives major blood supply from hepatic arteries, a single-input two compartment model is commonly used in most articles. However, for liver parenchymal disease or metastatic hepatic tumors which are supplied by both hepatic arteries and portal veins, a dual-input one compartment model by Materne et al[44] is often used to obtain parameters including arterial blood flow, portal blood flow, hepatic arterial fraction, distribution volume and MTT. For example, several articles used DCE-MRI with Materne model to stage liver fibrosis[45,46]. Liver perfusion assessed by DCE-MRI revealed increased hepatic arterial fraction and distribution volume with increasing liver fibrosis[45,47,48].

Recently, a hepatocyte-specific contrast agent was developed and showed different characteristics from traditional gadolinium-based contrast agents. A new model was developed for analysis of hepatic uptake by DCE-MRI using this hepatocyte-specific contrast agent[49]. Depending on the mathematical model applied and physiological assumptions made, variants of such quantitative parameters are obtained. Hence, when applying tracer kinetic modeling to clinical studies, it is important to state the choice of kinetic model employed at the outset. Currently, there is no consensus as to which kinetic model is best suited to evaluate the liver and HCC, and the development of an international consensus is necessary to allow a wider use of this technique.

Different field strengths employed in the dynamic acquisitions for developing DCE-MRI analysis have been shown to have a direct effect on the results of the pharmacokinetic parameters[50,51]. The choice of contrast agent molecular properties[52] and the temporal resolution of the acquisition have a clear influence on the parameters. To standardize calculations, the acquisition should have enough temporal resolution (less than 2-5 s each image set, during at least 5 min), and voxel-wise statistical analysis is suggested.

CLINICAL APPLICATION OF DCE-MRI

DCE-MRI is helpful to differentiate HCC from colorectal metastasis[53]. The values of arterial, portal and total blood flow, and distribution volume were significantly higher in the HCC than in the metastatic group, whereas MTT was significantly higher in the metastatic group.

Miyazaki et al[54] demonstrated that a lower pretreatment distribution volume and high arterial flow fraction was associated with a better response to treatment in patients with neuroendocrine liver metastases treated using yttrium-90 (Y-90)-labeled octreotide (⁹⁰Y-DOTATOC).

DCE-MRI is emerging as a promising method for monitoring tumor response to treatment in HCC patients, and could be used an early imaging biomarker to predict survival outcome of patients. The data are summarized in Table 1.

Table 1 Summary of different hepatocellular carcinoma treatment, dynamic contrast-enhanced magnetic resonance imaging parameters and outcome.

Ref.	Case number	Treatment	Parameter	Time interval	Outcome measure	P value
Wang et al[55], 2004	7	Thalidomide	↓ Peak, ↓ Slope	8 wk	P vs NP	< 0.05
Liang et al[23], 2007	19	Radiotherapy	↓ Peak,↓ Slope	2 wk	R vs NR	< 0.05
Zhu et al[56], 2009	34	Sunitinib	↓ Ktrans	2 wk	P vs NP	< 0.05
Jarnagin et al[21], 2009	34 (26 ICC and 8 HCC)	Floxuridine (FUDR) and dexamethasone	High baseline AUC, ↓ Kep	2 mo	OSOS	0.002 0.013
Yopp et al[57], 2011	17 (14 and 3 HCC)	Floxuridine (FUDR)Bevacizumab	↓ AUC	2 wk	TTP	0.002
Hsu et al[58], 2011	31	Sorafenib, TG/uracil	High baseline Ktrans	-	P vs NP	0.008
Hsu et al[58], 2011	31	Sorafenib, TG/uracil	↓ Ktrans	2 wk	P vs NPOSPFS	0.003 0.015 0.030
Hsu et al[59], 2012	67	Vandetanib	↓ Ktrans	1 wk	Pre vs Post	NS

HCC: Hepatocellular carcinoma; ICC: Intrahepatic cholangiocarcinoma; TG: Tegafur; P: Progression; NP: Non-progression; R: Responder. NR: Non-responder; PFS: Progression-free survival; OS: Overall survival; TTP: Time to progression; Pre: Pre-treatment; Post: Post-treatment; AUC: Area under curve; NS: Non-significant.

Wang et al[55] evaluated thalidomide efficacy in seven patients with advanced unresectable HCC that had failed to respond to prior local therapy. When comparing the MRI parameters for the tumors before and during treatment, they found a statistically significant difference for the peak enhancement, the maximal enhancement, and the enhancement slope percentage between two groups of patients (four had progressive disease, three had stable disease/partial response) with different clinical outcomes.

Liang et al[23] investigated the changes of the hepatic parenchyma and tumors by DCE-MRI in 19 patients with advanced HCC who received radiotherapy for 50 Gy in 25 fractions. An increased slope and peak of the tumor at week 2 was associated with an improved local response (P < 0.05) (Figures 1 and 2). In the parenchyma, an increased slope at week 2 was associated with recurrence outside the radiation fields or with progression over distant sites (P < 0.05). These findings emphasized the value of DCE-MRI in the second week after the start of radiotherapy in predicting local tumoral responses or systemic metastasis of HCC after radiotherapy.

Figure 1 Forty-nine-year-old man with good local response. A: Isodose distribution and region of interests (ROIs): Left panel: Isodose distribution on center section of radiation treatment planning. Red: 45 Gy; orange: 40 Gy; yellow: 30 Gy; green: 15 Gy and blue: < 15 Gy. Right panel: ROIs on MRI before RT: red: tumor with strongest enhancement; yellow: non-tumor liver parenchyma receiving 30 Gy; green: non-tumor liver parenchyma receiving 15 Gy; blue: spleen; B: T1 weighted contrast-enhanced MRI before RT (left panel, the site corresponding to the right panel of (A) and after RT (right panel). Arrows indicate tumor margins; C: Time Intensity Curve of ROIs before RT (left panel) and at week 2 of RT (right panel). Red: tumor; yellow: 30 Gy; green: 15 Gy; blue: spleen. The curve of spleen is deviated after pause for respiration due to interference by lung perfusion. The initial spike due to refocusing artifact will not be counted into analysis. (From reference [23], reprint with permission).

Figure 2 Sixty-four-year-old woman with good local response and intrahepatic recurrence outside of RT fields. A: Isodose distribution and region of interests (ROIs): Left panel: isodose distribution on center section of radiation treatment planning. Red: 45 Gy; orange: 40 Gy; yellow: 30 Gy; green: 15 Gy; blue: < 15 Gy. Right panel: ROIs on magnetic resonance imaging (MRI) before RT: red: tumor with strongest enhancement; yellow: non-tumor liver parenchyma receiving 30 Gy; green: non-tumor liver parenchyma receiving 15 Gy; blue: spleen; B: T1 weighted contrast-enhanced MRI before RT (left panel, same site as the right panel of (A) and after RT (right panel). The intersect picture of MRI over right panel demonstrates no tumor progression in the RT field on the center section of treatment planning. Arrows indicate tumor margins and arrowhead, recurrent tumor outside the field of RT; C: Time Intensity Curve of ROIs before RT (left panel) and at week 2 of RT (right panel). Red: tumor; yellow: 30 Gy; green: 15 Gy; blue: spleen. The initial spike due to refocusing artifact will not be counted into analysis. (From reference [23], reprint with permission).

Zhu et al[56] conducted a phase II study of sunitinib, an anti-VEFG receptor tyrosine kinase inhibitor, in 34 patients with advanced HCC. They found significant decreases in K^trans and K_ep after treatment (P < 0.0001). The extent of decrease in K^trans was substantially higher in patients who experienced partial response or stable disease compared with that in patients with progressive disease or who died during the first two cycles of therapy. They concluded that rapid changes in tumor vascular permeability are potential determinants of response and resistance to sunitinib in HCC.

Jarnagin et al[21] reports the results of 34 patients (26 intrahepatic cholangiocarcinoma and eight HCC) who received hepatic arterial infusion with floxuridine and dexamethasone. Patients with high pretreatment AUC had a longer median survival than those with low AUC (P = 0.002). Besides, decreased K^trans and K_ep on the first post-treatment MR scanning both predicted survival. Hence, pretreatment and early post-treatment changes in tumor perfusion characteristics may predict treatment outcome ahead.

Yopp et al[57] evaluated 17 patients (14 intrahepatic cholangiocarcinoma and 3 HCC) treated with floxuridine and bevacizumab. Significant decreases in AUC and K^trans were noted in tumors after bevacizumab. Time to progression correlated inversely with changes in AUC after bevacizumab. Reductions in tumor perfusion were greater in tumors expressing markers of anti-hypoxia and VEGF.

In one study of locally advanced HCCs receiving sorafenib and cytotoxic therapy, conducted by Hsu et al[58], a decrease of K^trans by 40% or greater after 14 days of treatment was correlated with longer progression free survival (PFS) and overall survival (OS). Besides, percentage of K^trans change (difference between pre- and post-treatment) is an independent predictor of tumor response, PFS, and OS (Figures 3 and 4). In another study, Hsu et al[59] reported a randomized clinical trial of 67 HCC patients with vandetanib treatment, but no significant vascular change was found 1 wk after treatment. They explained that the steady-state concentration of vandetanib will be reached after at least 4 wk of treatment. Besides, the vascular features of heterogeneous nature of HCC due to tumor necrosis, arterio-venous shunting within the tumors, and the effects of prior local therapy, might preclude a reliable MRI measurement and comparison.

Figure 3 Representative dynamic contrast-enhanced magnetic resonance imaging findings in one advanced hepatocellular carcinoma patient. A: Post-contrast T1-weighted magnetic resonance imaging at baseline; B: After 14 d of study treatment; C: Corresponding color K^trans maps at baseline; D: After 14 d of study treatment. Hypervascular area was indicated by red color. The selected region of interest for K^trans measurement was indicated by white arrows. In this patient, the K^trans values at baseline and after study treatment were 798.6 × 10^-3/min and 206.6 × 10^-3/min, respectively; E: The initial area under the gadolinium concentration-time curves (IAUC) at baseline; F: After study treatment from the same patient. The IAUC values at baseline and after study treatment were 1526.2 mmol/kg × s and 1376.1 mmol/kg × s, respectively. (From reference [58], reprint with permission).

Figure 4 Representative dynamic contrast-enhanced magnetic resonance imaging K^trans color maps before treatment (day 0, left hand side) and day 14^th after treatment (right hand side) in two advanced hepatocellular carcinoma patients. Corresponding hypervascular area was indicated by red color. region of interests analysis is more sensitive based on hypervascular part than entire tumor, with mean values. K^trans is a good diagnostic biomarker in differentiation between stable disease (SD, upper row) and progressive disease (PD, lower row) in two patients with hepatocellular carcinoma. Difference of K^trans (∆K^trans ) between SD and PD measured on hypervascular part and entire tumor are both significant. (From reference [58], reprint with permission).

HCC EVALUATED BY PERFUSION CT

Similar to DCE-MRI, perfusion CT imaging of the liver is performed by acquisition of serial images after contrast bolus injection to obtain various perfusion indices, including regional tumor blood flow, blood volume, flow-extraction product, and permeability-surface area product. Previous reports have suggested that CT perfusion parameters can be used for quantifying tumor vascularity[60-64] and angiogenesis[65] in HCC, or as biomarkers to monitor response to chemoembolization[50], chemotherapy and a range of different targeted agents[66-68]. For example, in one study of locally advanced HCCs receiving bevacizumab and cytotoxic therapy, high pretreatment K^trans by perfusion CT indicated those patients with a RECIST response[67]. Their findings were comparable with the results investigated by Hsu et al[58]: in patients with locally advanced HCCs receiving sorafenib and cytotoxic therapy, high pre-treatment K^trans measured by DCE-MRI indicated those patients who did not develop progressive disease[58]. The main drawback of perfusion CT is radiation exposure, but recent advances in multidetector CT technology many help achieve acceptable radiation dose in HCC patients.

FUTURE DIRECTION

Dynamic contrast-enhanced MR imaging is a reproducible technique. According to previous studies, the reproducibility of K^trans is good to moderate (coefficient of repeatability ranges from about 15%-40%)[69,70]. This suggests that in a well-conducted study, a change of K^trans value of more than 40% is likely to indicate a significant drug effect[71]. The reproducibility of DCE-MR imaging parameters is influenced by lesion location, with the parameters being significantly more reproducible in the liver than in the lung[72]. However, current DCE-MRI technique lacks standardization across multiple MR platforms and institutions, making it difficult to implement the technique in a multicenter setting[17,73]. Besides, there is a need to establish clear thresholds for a significant response when using quantitative DCE-MR imaging parameters for assessment of therapy response.

CONCLUSION

DCE-MRI is an imaging technique that appears to provide quantitative and biologically relevant informations related to tumor vasculature and angiogenesis, which can inform novel drug efficacy, monitor treatment response and act as an imaging biomarker to predict treatment outcome and survival in HCC patients.