O Kilickesmez MD; S Bayramoglu MD; E Inci MD; T Cimilli MD.
Conflicts of interest: None.
Dr Ozgur Kilickesmez, Department of Radiology, School of Medicine,Yeditepe University, Devlet Yolu Ankara Street 102/104, Kozyatagi 34752, Istanbul, Turkey. Email: email@example.com
The purpose of our study was to investigate the value of diffusion-weighted magnetic resonance imaging (DW-MRI) to discriminate benign and malignant focal lesions of the liver using parallel imaging technique. A total of 77 patients and 65 healthy controls were enrolled in the study. DW-MRI was performed with b-factors of 0, 500 and 1000 s/mm2, and the apparent diffusion coefficients (ADC) values of the normal liver and the lesions were calculated. The mean ADC value of the focal liver lesions were as follows: simple cysts (3.16 ± 0.18 × 10−3 mm2/s), hydatid cysts (2.58 ± 0.53 × 10−3 mm2/s), hemangiomas (1.97 ± 0.49 × 10−3 mm2/s), metastases (1.14 ± 0.41 × 10−3 mm2/s) and hepatocellular carcinomas (HCC) (1.15 ± 0.36 × 10−3 mm2/s). The mean ADC values of all the disease groups were statistically significant when compared with the mean ADC value of the normal liver (1.56 ± 0.14 × 10−3 mm2/s), (P < 0.01). There were also statistically significant differences among the ADC values of hemangiomas and HCC metastases (P < 0.01), and simple and hydatid cysts (P < 0.008). However, there was no statistically significant difference between HCC and metastases. The present study showed that ADC measurement has the potential to differentiate benign and malignant focal hepatic lesions. We propose to add DW sequence in the MR protocol for the detection and quantitative discrimination of hepatic pathologies.
Magnetic resonance imaging is considered the most accurate modality to image the liver for detection and characterization of diffuse and focal liver diseases and to discriminate benign from malignant tumors, reflecting its ability on the basis of various data acquired, such as T1, T2, and early and late post-gadolinium images.1,2 Characterization of focal liver lesions is very important because patients with known primary malignant neoplasms often have benign focal liver lesions, which must be differentiated from metastases. The lack of ionizing radiation with MR imaging and the safety of gadolinium chelates, as compared with iodinated contrast agents, are two important considerations for the preferential use of MR imaging over CT scanning in the investigation of liver disease.3 Furthermore, DWI has emerged as a new diagnostic tool with the ability to detect focal lesions and to discriminate malignant ones without the need for contrast material.4–7 We aimed to investigate whether DWI has the ability to detect malignancy and to discriminate metastases and hepatocellular carcinomas.
This was a retrospective study conducted at our institution between January 2006 and March 2007. A total of 77 patients (42 women, 35 men; mean age, 59 years) and 65 healthy controls (35 women, 30 men; mean age, 35 years) with completely normal liver MRI and laboratory findings were enrolled in the study. The research protocol was approved by the ethics committee of our institution. Written consent was obtained from all patients prior to commencement of the study.
Magnetic resonance imaging was performed on a 1,5 T body scanner (Avanto; Siemens, Erlangen, Germany) with a 33 mT/m maximum gradient capability using an eight channel phased-array body coil.
Before diffusion-weighted imaging, breath hold, axial 3-D gradient-echo T1-weighted (repetition time (TR), 5.32 ms; echo time (TE), 2.58 ms; flip angle (FA), 10°; matrix, 256 × 166; slice numbers, 96; slice thickness, 2.5 mm; interslice gap, 20%; field of view (FOV), 40 cm; averages, 1; acquisition time, 0:21 s; bandwidth, 300 Hz/Px), 2-D gradient-echo T1 in-phase and out-of-phase (TR, 128 msec; in-phase TE, 4.89 msec; out-of-phase TE, 2.38 msec; FA, 70°; matrix, 256 × 179; slice numbers, 30; slice thickness, 6 mm; interslice gap, 30%; FOV, 40 cm; averages, 1; acquisition time, 1:37 s; bandwidth 1/2, 390/410 Hz/Px), axial respiratory-triggered, turbo spin-echo T2-weighted sequence with fat saturation (TR, 1900 ms; TE, 76 ms; FA, 150°; matrix, 384 × 276; slice numbers, 29; slice thickness, 6 mm; interslice gap, 30%; FOV, 36 cm; averages, 1; acquisition time, 1:37 s; bandwidth, 260 Hz/Px), coronal T2-weighted half-Fourier single-shot turbo spin-echo (HASTE) (TR, 1100 ms; TE, 116 ms; FA, 150°; matrix, 256 × 204; slice numbers, 25; slice thickness, 6 mm; interslice gap, 30%; FOV, 35 cm; averages, 1; acquisition time, 0:28 s; bandwidth, 488 Hz/Px) sequences and then diffusion-weighted single-shot spin-echo echo-planar sequence with chemical shift selective fat-suppression technique; TR/TE, 4900/93; matrix, 192 × 192; slice numbers, 30; slice thickness = 6 mm; interslice gap, 35%; FOV, 45 cm; averages, 5; acquisition time, approximately 3 min, parallel acquisition techniques (PAT) factor, 2; PAT mode, parallel imaging with modified sensitivity encoding (mSENSE) was performed. Diffusion-weighted magnetic resonance imaging was performed with b-factors of 0, 500 and 1000 s/mm2.
Following DWI, contrast enhanced dynamic imaging was performed with axial 3-D gradient-echo T1-weighted MR sequence during and after administration of gadopentetate dimeglumine at a dose of 0.1 mmol/kg of body weight as a bolus injection with 20 s between each breath-hold acquisition (each breath hold lasted between 20 and 24 s). The final diagnoses of focal hepatic lesions were based on imaging findings and histopathological diagnosis. Hemangiomas were diagnosed with MR signal characteristics and typical contrast enhancement pattern. Simple and hydatid cysts were diagnosed with MR imaging findings and differentiation was made with indirect haemagglutination test findings. Of the six patients in the metastases group with known primary malignancies, the lesions were diagnosed according to the enhancement pattern and multiplicity of the lesions. The rest of the metastases group and all of the HCC group were diagnosed histopathologically.
The DWI datasets were transferred to an independent Workstation (Leonardo console, software version 2.0; Siemens) for post-processing, and the apparent diffusion coefficient (ADC) maps were reconstructed. To measure the ADC value of normal parenchyma we established round regions of interest (ROI) on four segments. Care was taken to exclude vessels and motion artifacts from the ROI. For each ADC value measurement, we applied five ROI measurements and accepted the average of the closest three measurements. In the patient group, a freehand ROI was defined for the lesions detected on the T2-weighted EPI image (b = 0), while referring to the conventional sequences for verification of the lesion boundaries. The ROI was then copied to the corresponding ADC map.
All statistical analyses were performed using SPSS (Statistical Package for Social Sciences) for Windows 10.0. The ADC values of cases are reported as the mean ± standard deviation. Variance analysis and paired samples test were also conducted for comparison of segments of abdominal organs. A P-value of less than 0.05 was considered to indicate a statistically significant difference.
Apparent diffusion coefficient values of all the patients who underwent conventional and diffusion-weighted MR examinations are listed as box plots in Figure 1.
The mean ADC value of the liver lesions were as follows (Table 1): simple cysts, 20 cases (3.16 ± 0.18 × 10−3 mm2/s); hydatid cysts, 13 cases (2.58 ± 0.53 × 10−3 mm2/s); hemangiomas, 15 cases (1.97 ± 0.49 × 10−3 mm2/s); metastases, 13 cases (1.14 ± 0.41 × 10−3 mm2/s); and hepatocellular carcinomas (HCC) 13 cases (1.15 ± 0.36 × 10−3 mm2/s). The mean ADC values of all of the disease groups were statistically significant when compared with mean ADC value of the normal liver group (1.56 ± 0.14 × 10−3 mm2/s), (P < 0.01). There were also statistically significant differences among the ADC values of hemangiomas and HCC metastases (P < 0.01), and simple and hydatid cysts (P < 0.008). However, there was no statistically significant difference between HCC and metastases.
Diffusion is the term used to describe the random (Brownian) motion of water molecules.8 With a very strong bipolar gradient pulse inserted into either a spin-echo pulse sequence (i.e. Stejskal-Tanner technique) or a gradient-echo pulse sequence, MR imaging can be made sensitive to the diffusion of water molecules in the tissue.6,9 Diffusion restriction increases in highly cellular tissues; in contrast, it decreases in low cellular tissues with large extracellular space or with broken-down cellular membranes.7
Studies have been published concerning the diffusion properties of focal hepatic lesions. Most of the studies revealed that ADC values of benign lesions (cysts and hemangiomas) were significantly higher than those of malignant lesions attributed to high cellularity of malignant masses.10–12
As with the previous studies, we found that hepatic cysts had the highest ADC because of their fluid content, with non-restricted motion of water molecules.4,10
Inan et al. found a statistically significant difference between simple and hydatid cysts based on diffusion signals and ADC values.13 The authors attributed this difference to viscous hydatid cysts consisting of scolices, hooklets, sodium chloride, proteins, glucose, ions, lipids and polysaccharides.14 Similarly, we found a statistically significant difference between simple and hydatid cysts.
Similar to previous studies, our study showed that qualitative and quantitative indices easily differentiated hemangiomas from cysts and malignant masses. Hemangiomas had lower ADC than cysts, probably related to vascular components of hemangiomas.15
Chan et al. reported that DWI could discriminate abscess and cystic tumors.6 In this study all the abscess cavities revealed significantly lower ADC values when compared with the necrotic portions of tumors without overlapping.
Metastases and HCC depicted low ADC values. ADC values in cases with metastases were slightly lower than HCC. This data is similar to findings by Taouli et al.12 and Demir et al.16, but are discordant with findings by Sun et al.17 The difference we found between HCC and metastases was not statistically significant.
Less frequently encountered focal liver lesions should also be studied with DWI for validation of the technique. The ADC value of the only focal nodular hyperplasia (1.43 × 10−3 mm2/s) detected revealed an ADC value similar to the normal liver parenchyma and higher than the malignant lesions (Fig. 6).
The reported ADC values of the lesions and normal liver varies substantially according to the choice of imaging parameters and there is no established protocol. Further studies to determine imaging parameters for routine imaging are required.
Studies have also been published concerning the use of DWI for diagnosis of diffuse liver diseases. These studies have shown that DWI may detect the degree of fibrosis non-invasively.18 In these studies ADC was lower when compared with the control group.
In our study, DW-MRI was performed without breath holding, thus allowing examination of severely ill, old, or obese patients who were unable to hold their breath for a long time. Our preliminary results suggest that DWI combined with parallel imaging makes the use of abdominal DW-MRI possible without breath holding in daily practise. Hence, there were ghost and motion artifacts that degraded the image quality, but were not severe enough to affect the diagnosis. Yoshikawa et al. stated that the parallel imaging technique may help to reduce artifacts and improve the reliability of measured values.19
In daily practise, besides the quantitative indices supporting the detection of malignancy, we had the benefit of using DWI for the simple detection of lesions, with high intensity against the suppressed back signal. Also, in some cases, the primary sites of the metastatic lesions (in two cases with stomach carcinoma), and additional metastatic lesions were detected more obviously when compared with conventional sequences. However, these findings should be validated in a large series.
Diffusion-weighted imaging may be added to routine abdominal imaging protocols as an additional sequence. Visual assessment of DW images has been shown to add confidence to lesion detection and characterization. The additional benefit of DWI is the ability to detect quantitative indices, which may be important in the assessment of disease response to treatment methods. Conventional assessment by measuring lesion size is insensitive to early treatment-related changes.20
The present study had several potential limitations. First, single-shot echo-planar imaging used with a higher b-value had a lower signal-to-noise ratio (SNR), resulting in image distortions. Second, the patient population was relatively small, and according to the small number of metastases we could not subgroup these lesions. Third, histopathological confirmation was not performed in all patients.
Even though, there was overlapping ADC values of metastases and HCC’s DMI is an excellent method to detect liver masses and to determine quantitative indices for discrimination of benign and malignant masses. We propose to add DWI to routine liver imaging protocols.