Hepatic MR imaging using IDEAL-IQ sequence: Will Gd-EOB-DTPA interfere with reproductivity of fat fraction quantification?

Abstract

BACKGROUND

Iterative decomposition of water and fat with echo asymmetry and least squares estimation quantification sequence (IDEAL-IQ) is based on chemical shift-based water and fat separation technique to get proton density fat fraction. Multiple studies have shown that using IDEAL-IQ to test the stability and repeatability of liver fat is acceptable and has high accuracy.

AIM

To explore whether Gadoxetate Disodium (Gd-EOB-DTPA) interferes with the measurement of the hepatic fat content quantified with the IDEAL-IQ and to evaluate the robustness of this technique.

METHODS

IDEAL-IQ was used to quantify the liver fat content at 3.0T in 65 patients injected with Gd-EOB-DTPA contrast. After injection, IDEAL-IQ was estimated four times, and the fat fraction (FF) and R2* were measured at the following time points: Pre-contrast, between the portal phase (70 s) and the late phase (180 s), the delayed phase (5 min) and the hepatobiliary phase (20 min). One-way repeated-measures analysis was conducted to evaluate the difference in the FFs between the four time points. Bland-Altman plots were adopted to assess the FF changes before and after injection of the contrast agent. P < 0.05 was considered statistically significant.

RESULTS

The assessment of the FF at the four time points in the liver, spleen and spine showed no significant differences, and the measurements of hepatic FF yielded good consistency between T1 and T2 [95% confidence interval: -0.6768%, 0.6658%], T1 and T3 (-0.3900%, 0.3178%), and T1 and T4 (-0.3750%, 0.2825%). R2* of the liver, spleen and spine increased significantly after injection (P < 0.0001).

CONCLUSION

Using the IDEAL-IQ sequence to measure the FF, we can obtain results that will not be affected by Gd-EOB-DTPA. The high reproducibility of the IDEAL-IQ sequence makes it available in the scanning interval to save time during multiphase examinations.

Key Words: Gadoxetate Disodium; Iterative decomposition of water and fat with echo asymmetry and least squares estimation quantification sequence; Fat fraction; Enhanced-Magnetic resonance imaging; R2*

Core Tip: The measurement of fat fraction was stable when using Iterative decomposition of water and fat with echo asymmetry and least squares estimation quantification sequence (IDEAL-IQ) before and after injection of hepatocyte-specific contrast agent. Such strong reproducibility makes IDEAL-IQ available in the interval for saving time during examination and optimizes the scanning protocol.

INTRODUCTION

The defining characteristic of hepatic steatosis (HS) lies in the intracellular accumulation of fat deposits within hepatocytes. Formerly regarded as a comparatively benign entity, it (HS) has the potential to advance towards the development of steatohepatitis, advanced fibrosis, and ultimately cirrhosis[1,2]. Fortunately, if intervention is performed in a timely manner, isolated steatosis is reversible, and a reduction in liver fat may diminish many of its associated risks. The fat fraction, which represents the quantitative ratio of fat mass to liver mass, serves as a pivotal parameter in the diagnosis and ongoing assessment of nonalcoholic fatty liver disease (NAFLD), the prevalence of which is approximately 25%[3]. Rapid radiological detection and diagnosis are essential in the management of patients with NAFLD[2,4,5].

Quantitative diagnosis and efficacy evaluation of HS have been studied clinically. Liver biopsy is not frequently used for HS diagnosis because of its invasive nature, and sampling errors can be caused by heterogeneous fat distribution[6]. Ultrasound detection is economical and convenient but is highly dependent on the procedure and lacks quantitative and objective criteria[7]. Computed tomography (CT) values cannot be used as a noninvasive technique in the evaluation of fat content, but the deposits of iron, copper, glycogen and amiodarone can increase the liver density and severely interfere with the accuracy of CT in the quantification of liver steatosis[8,9].

Currently, magnetic resonance spectroscopy (MRS) is commonly used as a standard fat quantification method[10-12]. However, the accuracy of this technique depends on the homogeneity of the magnetic field, and single spectral data cannot reflect the whole liver condition[11,12]. Recently, multipoint water-fat separation methods based on chemical shifts have been applied to quantitative analyses of liver fat[12-16]. Among these methods, the iterative decomposition of water and fat with echo asymmetry and least-squares estimation (IDEAL-IQ) not only generates an accurate evaluation of the FF (0%-100%) but also allows the quantitative analysis of the fat composition. This method utilizes a multiecho acquisition technique to facilitate T2* fitting, enabling the subsequent application of T2* correction. By eliminating the inherent T2* effect, this correction accounts for various factors that can impact the accuracy of FF measurements in chemical shift-based approaches. These factors include field inhomogeneity, T1 relaxation, noise bias, and eddy currents[16]. It facilitates the quantitative measurement of the whole liver and promising clinical application for testing the efficacy of drugs for hepatic disorders. The new quantitative scanning regimen not only facilitates accurate prognosis by monitoring dynamic changes in patients but also provides an objective basis for evaluating the treatment effect.

The primary objective of this study was to assess the potential interference of the liver-specific contrast medium Gd-EOB-DTPA on the measurement of hepatic fat content using the IDEAL-IQ sequence. Additionally, the study aimed to evaluate the reliability and resilience of this technique. By exploring these aspects, the researchers sought to identify and address the challenges associated with the widespread application of Gd-EOB-DTPA in clinical settings.

MATERIALS AND METHODS

Patients

In this institutional review board-approved study, patients with liver disease who visited the First Affiliated Hospital of Harbin Medical University between October 2015 and June 2016 were enrolled. All patients were diagnosed with hepatopathy using early imaging and subsequently screened with Gd-EOB-DTPA magnetic resonance imaging. The IDEAL-IQ sequence was used before and after injection to detect liver fat levels. In total, 79 patients were enrolled in this study. We excluded 14 cases for the following criteria: (1) Poor image quality caused by factors such as incompatible breath-holding; (2) Focal liver lesions exceeding 50% of the liver volume; and (3) Patients who underwent an abdominal operation. Finally, 65 patients were enrolled (48 males and 17 females with an average age of 54 years old and a range of 36 to 67 years old). Baseline patient characteristics and treatment history are shown in Table 1, and all the patients signed informed consent forms before the study.

Table 1 Baseline patient characteristics and treatment history.

Demographics	Parameter	No. of patients (%)
Age	< 60 yr	51 (65)
	> 60 yr	28 (35)
Sex	Male	48 (60)
	Female	17 (40)
Cause	HBV	20 (30)
	HCV	9 (14)
	HBV and HCV	2 (3)
	Alcohol	25 (38)
	Cryptogenic	9 (14)
Cirrhosis	Present	35 (54)
	Absent	30 (46)
Methods of diagnosis	Biopsy	15 (23)
	MR imaging	50 (77)
Tumor characteristics	Unifocal	43 (66)
	Multifocal	22 (34)
Diameter of lesion	< 3 cm	42 (65)
	> 3 cm	23 (35)
No treatment		45 (69)
Treated	Surgical treatment	9 (14)
	Internal medicine	11 (17)

Numbers in parentheses are percentages, and percentages were rounded. HBV: Hepatitis B virus; HCV: Hepatitis C virus.

Magnetic resonance imaging

A GE Signa HDxt magnetic resonance imaging (MRI) scanner (Milwaukee, United States) was used with 8-channel TORSOPA phased array coils. The patients fasted for 6 h before the test. Fat detection with the IDEAL-IQ sequence was performed four times before and after Gd-EOB-DTPA enhancement. The acquisition parameters were consistent at the four time points, indicating changes in the FF value and R2*. The scanning parameters were as follows: Breath-hold scanning, IDEAL-IQ (a three-dimensional volumetric imaging sequence): repetition time (TR)/first echo time (TE): 3.7 ms/1.7 ms, thickness 5 mm, receiver bandwidth 125 kHz, view: 44 cm× 40 cm, matrix: 256 × 256, flip angle 3°, NEX1. Six echoes in total were obtained for the T2* correction in two consecutive TRs with a 3-echo acquisition maintained for each. Two-dimensional parallel acceleration was used to maintain the scan time, approximately 21 s, within a single breath hold. The breath-hold double arterial phase LAVA enhanced scanning (BH Ax Dual LAVA+C) parameters were as follows: Turn over time 5.0 ms; flip angle 11°; bandwidth 125 kHz; matrix 256×152; view 44 × 40 cm; and the front and back range was 25% larger than the body surface with a thickness of 4-6 mm. The array spatial sensitivity encoding technique (ASSET) was used with a phase accelerating factor of 2.25 Ph. The injection speed of the contrast agent was 2.0-3.0 mL/s. The dosage was 0.025 mmol/kg, which is approved by the FDA, followed by injection of 20 mL saline at the same speed. After the contrast agent was injected, Dynamic-enhanced scanning using LAVA-Flex was employed to capture images during different phases, namely the arterial phase, portal phase, and equilibrium phase. Additionally, IDEAL-IQ was performed four times throughout the study, including a pre-contrast scan, scans between the portal phase (70 s) and late phase (180 s), a delayed phase scan (5 min), and a hepatobiliary phase scan (20 min). These scans were individually labeled as t1, t2, t3, and t4 in the subsequent statistical analyses and tables.

Data analysis

Two radiologists (with 5-7 years of experience in MR imaging of the abdomen) reviewed the image data of the patients in terms of the known pathological and comprehensive diagnosis. FF and R2* maps were autogenerated on the host scanner after each acquisition of the IDEAL-IQ sequence. All the images were independently evaluated by function tool 9.4 on an AW4.5 workstation (Advantage Workstation 4.5; GE Healthcare). Three regions of interest (ROIs) with a diameter of 16 mm in the right liver lobe were used to calculate the mean values (Figure 1). To ensure accurate measurements, regions of interest were carefully selected by excluding intrahepatic bile ducts, blood vessels, and focal liver lesions. The fat fraction detected in the images reflected the amount of fat present in the liver mass, which typically does not exceed 5% in healthy individuals. Similarly, employing the same approach, two ROIs with a diameter of 16 mm were delineated within the spleen for further analysis (Figure 1), one ROI was drawn in the spinal vertebral layer (Figure 2), and the FF and R2* values were recorded. The obtained anatomical images of the spine were considered the region of interest in the vertebral body bone marrow to avoid inclusion of the intervertebral disk. Finally, the FF and R2* values in the liver, spleen and spine determined by the two physicians were averaged to calculate the final result.

Figure 1 Fat fraction (%, upper row) and R2* (Hz, bottom row) mappings of the right liver lobe and spleen of a 57-year-old male patient with primary hepatocellular carcinoma. The mean FF and R2* values of the right liver lobe and spleen are shown on the images. FF: Fat fraction.

Figure 2 Fat fraction (%, upper row) and R2* (Hz, bottom row) mappings of the spine of a 47-year-old patient with hepatic hemangioma. The obtained anatomical images were referred to the region of interest in the vertebral body bone marrow to avoid inclusion of the intervertebral disk. FF: Fat fraction.

Statistics

The FF and R2* values at the four time points were evaluated, and the data were expressed as the means ± SDss and analyzed by ANOVA for repeated measurements. Furthermore, in each region, the values before and after injecting the contrast agent at each time point were compared, and Bland-Altman plots were applied to analyze the differences. P < 0.05 was considered statistically significant. SAS 9.13 software (SAS Institute Inc, China) was used to perform the statistical analysis.

RESULTS

For the 65 patients, the final FF and R2* values (mean ± SD, n = 65) of the liver, spleen and spine are shown in Table 2. The FF values of the liver, spleen and spine were compared before (time point of T1) and after injection of the contrast agent (T2, T3, and T4). The results showed no significant differences between them (T1-T2, P = 0.85; T1-T3, P = 0.24; T1-T4, P = 0.13), and the FF values were acceptable (Table 3). The R2* value increased from T1 to T2 from 43.14 to 46.21 s-1, respectively (P < 0.0001) in the liver and from 42.97 to 46.08 s-1 (P < 0.0001) in the spleen. Thereafter, the values reached 46.60 s-1 and 47.10 s-1 at t3 and t4 in the liver (P < 0.0001) and 46.38 s-1 and 46.76 s-1 in the spleen (P < 0.0001). R2* values in the spine were 121.75 s-1 at t1 and 124.90 s-1, 125.75 s-1 and 126.45 -1 at t2, t3, and t4, respectively (P < 0.0001). In summary, the R2* of the liver, spleen and spine increased significantly after injection (P < 0.0001), showing a T2* effect of Gd-EOB-DTPA (Table 4).

Table 2 Fat fraction and R2* (mean ± SD, n = 65) of the liver, spleen and spine at four time instants.

Index	T1	T2	T3	T4
Liver FF	6.0567 ± 3.1295	6.0622 ± 3.1026	6.0928 ± 3.0863	6.103 ± 3.1272
Spleen FF	2.0183 ± 0.7687	2.0363 ± 0.7531	2.0478 ± 0.7633	2.0652 ± 0.7494
Spine FF	25.6275 ± 6.6189	25.6422 ± 6.5966	25.6495 ± 6.5979	25.6648 ± 6.5941
Liver R2*	43.1359 ± 9.95	48.211 ± 10.0155	49.5981 ± 10.0038	52.0894 ± 10.0021
Spleen R2*	42.9711 ± 13.9805	46.0781 ± 13.7729	46.3787 ± 13.7615	46.7647 ± 13.796
Spine R2*	121.7495 ± 20.602	124.902 ± 20.3045	125.7506 ± 20.263	126.4444 ± 20.3439

T1: Pre-contrast; T2: Between portal phase (70 s) and equilibrium phase (180 s); T3: Delayed phase (5 min); T4: Hepatobiliary phase (20 min).

Table 3 The comparison of fat fraction (n = 65) before and after hepatocyte-specific contrast media injection.

Variation	DF	SS	MS	F	P value
T1-T2 (liver)	1	0.001	0.001	0.0319	0.8584
T1-T3 (liver)	1	0.0417	0.0417	1.3901	0.2399
T1-T4 (liver)	1	0.0685	0.0685	2.2825	0.1325
T1-T2 (spleen)	1	0.0103	0.0103	0.4911	0.4843
T1-T3 (spleen)	1	0.0328	0.0328	1.5601	0.2132
T1-T4 (spleen)	1	0.0703	0.0703	3.3418	0.0691
T1-T2 (spleen)	1	0.0069	0.0069	0.1703	0.6803
T1-T3 (spleen)	1	0.0157	0.0157	0.3859	0.5352
T1-T4 (spleen)	1	0.045	0.045	1.1053	0.2944

DF: Degree of freedom; SS: Variance; MS: Mean square; F: Statistics.

Table 4 The comparison of R2* (n = 65) before and after injection of Gadoxetate disodium.

Variation	DF	SS	MS	F	P value
T1-T2 (liver)	1	297.9598	297.9598	476.8081	< 0.0001
T1-T3 (liver)	1	377.59	377.59	604.2357	< 0.0001
T1-T4 (liver)	1	492.3481	492.3481	787.8766	< 0.0001
T1-T2 (spleen)	1	299.2445	299.2445	1035.896	< 0.0001
T1-T3 (spleen)	1	359.9598	359.9598	1246.074	< 0.0001
T1-T4 (spleen)	1	446.1213	446.1213	1544.34	< 0.0001
T1-T2 (spleen)	1	313.126	313.126	317.1906	< 0.0001
T1-T3 (spleen)	1	504.28	504.28	510.826	< 0.0001
T1-T4 (spleen)	1	694.3318	694.3318	703.3447	< 0.0001

DF: Degree of freedom; SS: Variance; MS: Mean square; F: Statistics.

The Bland-Altman plots (Figures 3-5) demonstrated good agreement of hepatic FF measurements between T1 and T2 (95% confidence interval: -0.6768%, 0.6658%), T1 and T3 (-0.3900%, 0.3178%), and T1 and T4 (-0.3750%, 0.2825%). These findings suggested considerable stability and repeatability of the IDEAL-IQ FF measurements under the impact of Gd-EOB-DTPA.

Figure 3 Agreement of fat fraction measurements between T1 (pre-contrast) and T2 (between portal phase and equilibrium phase). Bland-Altman plots indicating 95% of confidence interval in the hepatic fat fraction measurements (-0.6768, 0.6658).

Figure 4 Agreement of fat fraction measurements between T1 (pre-contrast) and T3 (delayed phase). Bland-Altman plots indicating 95% of confidence interval in the haptic fat fraction measurements (-0.3900, 0.3178).

Figure 5 Agreement of fat fraction measurements between T1 (pre-contrast) and T4 (hepatobiliary phase). Bland-Altman plots indicating 95% of confidence interval in the hepatic fat fraction measurements (-0.3750, 0.2825).

DISCUSSION

Unlike the R2* values, the FF values of the liver, spleen and spine obtained by the IDEAL-IQ sequence showed no significant differences at the four time points before and after the injection of Gd-EOB-DTPA. Our findings demonstrated that this technique for fat quantification can overcome the strong susceptibility effect induced by contrast media, including hepatocyte-specific contrast agents.

IDEAL-IQ is based on a chemical shift-based water and fat separation technique to obtain the proton density fat fraction (PDFF)[14,17,18]. The accuracy of water-fat separation is affected by many factors, including T2* attenuation and the multiple fat peak model. According to the chemical characteristics of triglycerides, the nine-peak fat model was formed. To remove the influence of T2*, a multiecho technique is used to predict the attenuation rate of R2*, and it is used in the calculation of the water and fat maps[16,19]. As a noninvasive liver fat analysis method, several result maps can be reconstructed in one scan, including the FF, R2*, water and fat phases. Moreover, to decrease the signal deviation between water and fat due to the T1 effect, a small flip angle pulse was used to ensure proton density (PD)[19,20]. The stability and repeatability of testing for liver fat using IDEAL-IQ was acceptable and achieved a high accuracy[12,20].

Approximately 50% of the hepatic cells absorbed Gd-EOB-DTPA, which is derived from a lipophilic EOB compound containing gadolinium. This absorption occurred through the organic anion transporting polypeptide (OATP) mechanism[21]. Although the contrast agent partially entered the hepatic cells via the intercellular space, it did not affect the water or fat composition but altered the magnetic field environment and T2* effect. However, such attenuation was eliminated by IDEAL-IQ. IDEAL-IQ imaging estimated a complex field map via six echo signals using the iterative least-squares method. A complex field map was used to differentiate water and fat and to obtain the dynamic 0 to 100% fat fraction. Amplitude reconstruction was used to quantify the fat fraction, and the phase error was removed. Finally, combined with the results of the two reconstructions, T2* was estimated[20,22]. Thus, elimination of the T2* effect by IDEAL-IQ was relatively stable. A few studies also evaluated the confounding effects of fat detection after injecting a contrast agent, especially SPIO, and found that IDEAL-IQ has stable performance[23]. As a liver-specific contrast agent, SPIO is absorbed by Kupffer cells into hepatic cells. The mechanism was similar to that of Gd-EOB-DTPA, and the difference was mainly in the amounts deposited in hepatic cells. Under T1 deviation, noise difference, T2* decay, a complicated fat resonance spectrum and vortex, the accuracy of the chemical shift in fat quantification was influenced differently[12,20]. However, IDEAL-IQ compensated for and neutralized the effects of interfering factors. Another study illustrated that the FF had no significant difference after injection with Gd-DTPA instead of with Gd-EOB-DTPA in our case[24]. Thus, the presence of contrast agent inside or outside cells does not alter the FF value theoretically.

As a reciprocal of T2*, R2* is related to the water and fat components[17,22]. Due to the application of small 3D flip angle acquisition with gradient echo flyback, the R2* value and FF value can be calculated simultaneously. Therefore, R2* is used to describe the decay of T2* caused by Gd-EOB-DTPA. In our study, the injection of a contrast agent slightly increased R2* due to transverse relaxation of Gd-EOB-DTPA. In the case of contrast agents with similar paramagnetic ions, the R2* influencing factors mainly included micro viscosity, hydrogen proton nuclear dispersion caused by susceptible echo, and the combining capacity with cells[23]. The combining capacity of Gd was lower than that of Gd-EOB-DTPA. However, the influence was limited, which requires further exploration. Being a widely employed liver-specific contrast agent, Gd-EOB-DTPA was similar to Gd in influencing T2* and the T1 relaxation time. At similar dosages, Gd-EOB-DTPA showed a higher relaxation time than Gd, and the paramagnetic effect greatly decreased the T1 relaxation time[25-27]. Nevertheless, IDEAL-IQ effectively addressed the confounding factors that could potentially disrupt accurate fat quantification. It rectified issues such as T2* decay, the distribution of multifat spectrum peaks, and generated separate water and fat images. Additionally, IDEAL-IQ provided valuable information through the fat fraction and R2* maps, further enhancing the reliability of fat quantification[27,28].

However, our study still has a few limitations. First, despite the accuracy and reliability of IDEAL-IQ, Gd-EOB-DTPA intake by hepatic cells was up to 50%, and the role of hepatic function still requires clinical investigation. Second, unstable breath holding resulted in a slight shift in the local scan layer and further deviation in the inconsistent ROI before and after contrast agent administration. However, we used the image calibration software of the AW4.5 workstation and drew ROIs in the same layer as much as possible. Third, in the four periods of IDEAL-IQ after contrast agent injection, the initial time was influenced by several factors. A deviation in the FF value and other parameters was caused by a failure to achieve the target time. Control of the scan time is important for further study.

CONCLUSION

The measurement of fat fraction was stable when using IDEAL-IQ to detect the FF values in patients scanned by EOB-MRI before and after injecting a hepatocyte-specific contrast agent. The R2* values in the liver, spleen and spine increased after injection of Gd-EOB-DTPA but had no significant effect on fat quantification. Such strong reproducibility makes IDEAL-IQ available in the interval for saving time during examination and optimizing the scanning protocol.

ARTICLE HIGHLIGHTS

Research background

Iterative decomposition of water and fat with echo asymmetry and least squares estimation quantification sequence (IDEAL-IQ) is based on chemical shift-based water and fat separation technique to get proton density fat fraction. Multiple studies have shown that using IDEAL-IQ to test the stability and repeatability of liver fat is acceptable and has high accuracy.

Research motivation

At present, there are few studies on comparing the fat content of different Gadoxetate Disodium (Gd-EOB-DTPA) enhanced organs at different periods. This research can further solve the problem of whether Gd-EOB-DTPA affects fat fraction measurement, and test repeatability of ideal-IQ at Different Periods, which can optimize clinical examination items.

Research objectives

The challenges associated with the widespread applications of liver-specific contrast medium Gd-EOB-DTPA, the purpose of this study is to evaluate whether the Gd-EOB-DTPA would interfere with the measurement of hepatic fat content that was quantified with IDEAL IQ sequence, and the robustness of this technique was also evaluated.

Research methods

In this study, IDEAL-IQ was employed to quantify liver fat content in a cohort of 65 patients who received injections of Gd-EOB-DTPA contrast material during imaging at 3.0T. Following the injection, IDEAL-IQ was performed four times, and measurements of the fat fraction (FF) and R2* were obtained at specific time points: pre-contrast, between the portal phase (70 s) and late phase (180 s), delayed phase (5 min), and hepatobiliary phase (20 min). These measurements allowed for the assessment of liver fat content at different stages of contrast material uptake and clearance.

Research results

The evaluation of the FF at four different time points in the liver, spleen, and spine revealed no significant differences. Furthermore, the measurements of hepatic FF demonstrated good consistency between T1 and T2, T1 and T3, as well as T1 and T4. On the other hand, the R2* values of the liver, spleen, and spine exhibited a significant increase following the injection of the contrast material.

Research conclusions

Using IDEAL-IQ sequence in the measurement of FF, we can get results that won’t be affected by Gd-EOB-DTPA. The high reproducibility of IDEAL IQ makes it available in the interval of scanning for saving time during multi-phase examination.

Research perspectives

Multi-center prospective studies with larger sample sizes are required to investigate whether the dosage of GD-EOB-DTPA affects IDEAL-IQ FF.

ACKNOWLEDGEMENTS

We would like to thank all the study participants and the staff of our hospital. Thanks the National Natural Science Foundation of China.