The diagnostic value of time-resolved MR angiography with Gadobutrol at 3 T for preoperative evaluation of lower extremity tumors: Comparison with computed tomography angiography.

PURPOSE: To evaluate the diagnostic value of time resolved magnetic resonance angiography with interleaved stochastic trajectory (TWIST) using Gadobutrol for preoperative evaluation of lower extremity tumors.
MATERIALS AND METHODS: This prospective study was approved by the local Institutional Review Board. 50 consecutive patients (31 men, 19 women, age range 18-80 years, average age 42.7 years) with lower extremity tumors underwent TWIST and computed tomography angiography (CTA). Image quality of TWIST and CTA were evaluated by two radiologists according to a 4-point scale. The degree of arterial stenosis caused by tumor was assessed using TWSIT and CTA separately, and the intra-modality agreement was determined using a kappa test. The number of feeding arteries identified by TWIST was compared with that by CTA using Wilcoxon signed rank test. The ability to identify arterio-venous fistulae (AVF) were compared using a chi-square test.
RESULTS: Image quality of TWIST and CTA were rated as 3.88 ± 0.37 and 3.97 ± 0.16, without statistically significant difference (P = 0.135). Intra-modality agreement was excellent for the assessment of arterial stenosis (kappa = 0.806 ± 0.073 for Reader 1, kappa = 0.805 ± 0.073 for Reader 2). Readers identified AVF with TWIST in 27 of 50 cases, and identified AVF with CTA in 14 of 50 (P < 0.001). Mean feeding arteries identified with TWIST was significantly more than that with CTA (2.08 ± 1.72 vs 1.62 ± 1.52, P = 0.02).
CONCLUSION: TWIST is a reliable imaging modality for the assessment of lower extremity tumors. TWIST is comparable to CTA for the identification of AVF and feeding arteries.

Introduction

Needle biopsy prior to surgical resection is widely accepted for assessment of lower extremity tumors [1]. However, preoperative imaging tests remain indispensable, because imaging assists to design surgical planning [2]. Preoperative vascular evaluation of tumors can even alter the operative approach [3]. The degree of arterial invasion is important in deciding whether tumor could be surgically resected [4].

Catheter digital subtraction angiography (DSA) is the diagnostic gold standard due to its high temporal and spatial resolution. Preoperative embolization can be performed concurrently in some cases to prevent hemorrhage during the definitive surgery. However, the performance of DSA is limited in practice as it is an invasive and expensive test relying upon the use of ionizing radiation [5], [6], [7]. Computed tomography angiography (CTA) and magnetic resonance angiography (MRA) are non-invasive alternatives to DSA for the depiction of tumors [8], [9], [10], [11]. With respect to MRA, time resolved MRA has been reported as superior to conventional contrast enhanced MRA in several respects: lower contrast agent dose, decreased venous contamination, and added dynamic information [12], [13], [14], [15]. TWIST (Time-resolved angiography With Interleaved Stochastic Trajectories, TWIST) imaging divided k-space into two regions – one central region responsible for overall image contrast and a peripheral, outer region responsible for image detail [16]. To accomplish greater temporal resolution, the central regions are sampled more frequently than the peripheral regions based on a technique called view sharing. While TWIST has been reported to yield excellent image quality and to be accurate in the diagnosis of peripheral arterial stenosis [11], its value in the assessment of lower extremity tumors has not been studied to the authors’ knowledge. The aim of the current study is to compare TWIST with CTA for the preoperative vascular evaluation of lower extremity tumors.

Material and methods

Patients

This prospective study was institutional review board approved. Inclusion criteria were: (1) a patient with a lower extremity mass; (2) tumor protocol MRI confirming presence of a mass; (3) a surgical plan for open biopsy or definitive surgery; (4) patient agreement to both CTA and TWIST. Exclusion criteria were: (1) patient GFR < 30 ml/m2/1.73 m2; (2) patient on dialysis; (3) pregnancy; (4) standard contraindications for MR or CT (i.e. pacemaker, paramagnetic foreign bodies, claustrophobia, etc.). From January 2013 to March 2015, 57 consecutive patients met the inclusion criteria and were enrolled in this trial. Informed consent was obtained from every patient. All 57 patients underwent TWIST. However, 2 patients with positive results of iodine allergy test did not undergo CTA. 5 cases were excluded because no tumor was ultimately found, thus 50 patients with lower extremity tumors were included in the final analysis.

TWIST examinations

All TWIST examinations were performed on a 3.0T whole-body MR scanner (Magnetom Skyra, Siemens medical solutions, Germany). Patients were placed on the scanner table in feet-first supine position. Two eight-element body array coils were used to cover the area of interest, and were combined with the posterior integrated multi-channel spine array coil. An 18-gauge IV line was placed in the right antecubital vein for contrast injection.

TWIST was performed in the coronal plane with the following parameters: TR/TE, 2.9/1.06 ms; flip angle, 25°; bandwidth, 700 Hz; FOV, 448mm × 358.4 mm; slice thickness, 1 mm; matrix, 450 × 360; spatial resolution, 1mm × 1mm × 1 mm; time resolution, 3.84s/frame; A&B, 15%/20%; GeneRalized Autocalibrating Partially Parallel Acquisition (GRAPPA) factor, 2; number of measurements, 25.

Gadovist (Gadobutrol, Bayer Pharma AG, Berlin, Germany) was utilized as the contrast agent in this study. Injection of contrast (0.1 mmol/kg body weight) was performed by an automatic injector (Accutron MR, MEDTRON, Germany) at a rate of 2.5 ml/s, followed by a 20 ml saline flush at the same rate.

CTA examinations

For each patient, the CTA examination was performed within 36 h after MRA. A 128-row CT scanner (Discovery HD 750, GE medical, America) was used with the following parameters: tube voltage, 100 Kv; tube current, 150 mA; pitch, 0.984:1; table speed, 55 mm/s; slice thickness, 0.625 mm; FOV, 50 cm. Iodinated contrast agent (Ultravist, Bayer, Germany, 1.2 ml/kg body weight) was administered by an electronic power injector (Stellant, MEDRAD, America) through an 18 gauge antecubital injection at a rate of 3 ml/s. The bolus-tracking technique was used. A region of interest (ROI) was positioned in the aortic bifurcation and image acquisition automatically started 5.5 s after the signal attenuation of ROI reached the predefined threshold of 120 Hounsfield Units (HU).

Image analysis

Imaging post-processing were performed on a dedicated Siemens workstation (Syngo.Via, Siemens Healthcare, Germany). Two readers with 10 years’ and 8 years’ experience viewed maximum intensity projections (MIP) images of CTA and arterial phase MIP images of TWIST for image quality evaluation. TWIST MIP were generated automatically by the scanner. CTA MIP images were reconstructed with a window setting of 600/300 (window width/window level). The readers were blinded to diagnosis and patient information and viewed cases in random order. The readers evaluated the MIP images according to a 4 point scale: 0 = non diagnostic image quality; 1 = poor image quality, observer not confident; 2 = fair quality, observer marginally confident; 3 = good quality, observer confident, 4 = excellent quality, observer highly confident. If the readers gave different scores for the same image, the average of the two scores was utilized for the analysis.

Arterial stenosis caused by tumor was evaluated by two readers with 8 years’ and 6 years’ experience, using both source images and reconstructed images. The readers were blinded to diagnosis and patient information, and rated stenosis as follows: 0 – Artery free from tumor; 1 – Artery displaced but not narrowed by tumor; 2 – Tumor narrowing artery less than 50%; 3 – Tumor narrowing artery ≥50%.

The presence or absence of AVF was determined by two readers with 10 years’ and 9 years’ experience in consensus. A fluoroscopic TWIST MIPs was examined to confirm the diagnosis of AVF. The two readers were asked to count feeding arteries in the MIP images of each tumor. The viewing interval between CTA and TWSIST for the same patient was at least 4 weeks in order to avoid recall bias.

Operation and pathology

For each patient, a definitive operation or open biopsy was performed within 48 h following the CTA examination. Mass resection was performed in 32 cases. An associated joint replacement was performed in 3 cases. An amputation at the level of the right hip joint was performed in one patient. 14 patients received incision biopsy without mass resection. Pathological correlation was available in all 50 cases (Fig. 1). In the present study, WHO classification of tumors of soft tissue and bone 2013 (4th edition) was used [17]. 38 of 50 cases were primary soft tissue tumors and 12 of 50 cases were bone tumors with soft tissue involvement. 18 cases were benign tumors, and 32 cases were malignant tumors.

Fig. 1

Pathology results.

Distribution of histopathology for the lower extremity tumors of the 50 evaluated patients.

Statistical analysis

The Wilcoxon signed rank test was used to determine differences in image quality of CTA and TWIST, and also used to compare number of feeding arteries identified by CTA and TWIST. Intra-modality agreement in assessing arterial stenosis was determined with a Kappa test. Kappa > 0.8 was considered as excellent agreement. 0.6–0.8 was considered good. 0.4–0.6 was considered fair. Kappa < 0.4 was considered as poor agreement. A chi-square test was used to determine the difference in AVF identified by CTA and TWIST. All data analysis was performed with SPSS (version 21.0, IBM, America). A p-value less than 0.05 was considered statistically significant.

Results

From January 2013 to March 2015, 50 consecutive patients (31 men, 19 women, age range 18–80 years, average age 42.7 years) with lower extremity tumors underwent TWIST and CTA, without any adverse events.

44 TWIST cases and 48 CTA cases were rated as excellent by both readers (Fig. 2). Image quality of TWIST was lower than that of CTA in 6 of 50 cases. Total image quality of TWIST was slightly lower than that of CTA, not reaching statistical significance (3.88 ± 0.37 vs 3.97 ± 0.16, P = 0.135).

Fig. 2

Image quality of TWIST.

(a) CTA image rated as excellent image quality. b: arterial phase TWIST image rated as excellent image quality. The arterial SNR is high with small branches clearly depicted.

Arterial stenosis caused by tumor is shown in Table 1. Intra-modality agreement between TWIST and CTA (see Table 2) was excellent for both readers (kappa = 0.806 ± 0.073 and 0.805 ± 0.073). 88.0% cases (44/50) were rated equally using the two methods.

Table 1

Arterial stenosis caused by tumor was rated: 0 –Artery free from tumor; 1 – Artery displaced but not narrowed by tumor; 2 – Tumor narrowing artery less than 50%; 3 – Tumor narrowing artery ≥50%.

Level	Reader 1		Reader 2
	TWIST	CTA	TWIST	CTA
0	20	22	19	21
1	24	20	25	21
2	5	7	5	7
3	1	1	1	1

TWIST = time resolved magnetic resonance angiography with interleaved stochastic trajectory, CTA = computed tomography angiography.

Table 2

Intra-modality agreement in determining arterial stenosis.

Test	Level	Reader 1				Reader 2
		CTA				CTA
		0	1	2	3	0	1	2	3
TWIST	0	20	0	0	0	19	0	0	0
	1	2	19	3	0	2	20	3	0
	2	0	1	4	0	0	1	4	0
	3	0	0	0	1	0	0	0	1
kappa		0.806 ± 0.073				0.805 ± 0.073

TWIST = time resolved magnetic resonance angiography with interleaved stochastic trajectory, CTA = computed tomography angiography.

Readers identified AVF at 27 cases using TWIST, while at 14 cases using CTA (54.0% vs 28.0%, P < 0.001). For 29 cases, the number of tumor feeding arteries identified by TWIST was equivalent to that by CTA (Fig. 3). For 18 cases, TWIST identified one or more feeding arteries than CTA (Fig. 4). The mean number of tumor feeding arteries identified by TWIST was greater than that by CTA (2.08 ± 1.72 vs 1.62 ± 1.52, P = 0.02).

Fig. 3

Multiple myeloma.

(a) TWIST of a tumor within the right thigh, with two feeding arteries (thin white arrows), and an arterio-venous fistula (thick white arrow). (b) CTA with excellent depiction of arterio-venous fistula (thick black arrow) and feeding arteries (thin white arrows). (c) Analogous findings as depicted on DSA. Here both feeding arteries (thin white arrow) are identified along with tumor stain from contrast administration.

Fig. 4

Malignant fibrous histiocytoma.

(a) CTA demonstrating the right superficial femoral artery displaced by a mass in the right distal thigh without evidence of stenosis. Increased number and abnormal morphology of small arterial branches are identified. Several feeding arteries (thin black arrows) and tumor stain are also demonstrated. (b) Stenosis degree of right superficial femoral artery rated with TWIST was equivalent to that rated with CTA. More feeding arteries (thin white arrows) are identified by TWIST. The feeding arteries in TWIST seem more distinct than those in CTA. (c) Arterio-venous fistula (thick white arrow) and abnormal enhancing nidi are well-seen in the later phase of TWIST imaging. Most of the draining veins corresponded to the feeding arteries.

Discussion

The sensitivity and specificity of MRA and CTA have been reported as comparable to that of DSA for the diagnosis of vasculopathy[18], [19], however, data regarding comparing these non-invasive techniques for preoperative assessment of lower extremity tumors is lacking. In the present study, TWIST using Gadobutrol at 3 T was evaluated for this purpose. With advances in MR hardware and software, time-resolved MRA can now be obtained with concurrently high spatial and temporal resolution [20], [21], [22]. The spatial resolutions utilized in this work were 1 mm isotropic for TWIST and 1mm × 1mm × 0.625 mm for CTA. Time-resolved CTA was not performed due to the current prohibitively high level of ionizing radiation exposure required [8], [23]. The main founding of the present study is that TWIST is comparable to CTA in identifying tumor feeding arteries and AVF, as well as assessing arterial stenosis caused by tumors.

Selection of contrast agent is important for contrast-enhanced MRA [24], including TWIST, because the concentration and relaxivities of the contrast agent affect image quality. Gadobutrol has a high relaxivity and is the only gadolinium (Gd)-based contrast agent approved for clinical use at 1 M concentration. Overall image quality was rated higher in examinations with gadobutrol compared with gadopentetate dimeglumine [25]. Morelli et al. [26] found that gadoburol contrast-enhanced MRA resulted in improved accuracy of renal artery stenosis assessments relative to gadoterate meglumine. Thus, gadobutrol was chosen as the contrast agent for TWIST in the present study.

In the present study, image quality of MRA was slightly lower than that of CTA, with the difference not statistically significant (P = 0.135). One possible explanation is that TWIST images relied upon mask subtraction, which was susceptible to patient motion. Fortunately, the lower extremity is not affected by respiration or bowel peristalsis like the chest and abdomen. Venous contamination was not observed in the arterial phase of TWIST in this study, which also contributed to the image quality of TWIST. Bolus-chase contrast-enhanced MRA is an alternative technique to TWIST in lower extremity imaging, but was not utilized in this study due to the potential for venous contamination [27], [28].

The spatial resolution of TWIST (1mm × 1mm × 1 mm) utilized in this study was proved sufficient for detection of small feeding arteries of tumor. The isotropic resolution also enables rotation of the 3D-MIP images of TWIST out of the coronal plane without significant vessel blurring, which is important for accurate identification of the tumor feeding arteries. Although the spatial resolution of CTA (1mm × 1mm × 0.625 mm) was even higher, CTA did not identify more feeding arteries that TWIST, one possible explanation was that non-subtracted CTA was used in the present study, whose vessel-to-background contrast was not high (Fig. 5).

Fig. 5

Intra-muscular hemangioma.

(a) An arterial abnormality is difficult to identify on the CTA MIP image. (b) Subtle perivascular enhancement (thick white arrows) and a dilated draining vein (thin white arrow) on TWIST are consistent with the presence of a vascular malformation.

Arterial invasion by tumor is an issue of particular concern to orthopedic oncologists. Excellent intra-modality agreement was observed between TWIST and MRA in the current study. TWIST seems an ideal substitute for CTA in the assessment of arterial stenosis caused by tumor. However, it is worth noting that TWIST is performed in the coronal plane, while CTA in the transverse plane. The detection of mild anterior-posterior stenosis with TWIST was slightly limited, thus underestimating stenosis degree relative to CTA.

Feeding arteries and AVF are usually of very small size, limiting their angiographic demonstration in some cases [29], [30]. With the development of more technically sophisticated MR hardware and software, spatial resolution and SNR has also improved. This is likely contributory to the ability of detecting feeding arteries. Furthermore, the conventional CTA had only one frame of imaging, while the TWIST contained multiple arterial and venous phases, enabling readers to view the dataset dynamically, similar to DSA, TWIST enabled readers to select the best arterial phase for evaluation of feeding arteries and AVF. The dynamic nature of the TWIST dataset also resulted in improved identification of AVF, whereas single phase CTA may fail to detect AVF duo to time-point missing.

One limitation of the current study is that only one MR contrast agent was utilized for TWIST. Additional verification of results is necessary using other agents. Second, subtracted CTA was not used duo to more radiation dose. Subtracted CTA was superior to non-subtracted CTA in vessel-to-background contrast, so it may identify more feeding arteries. Third, a relatively heterogeneous tumor population was also studied, including both primary soft tissue tumors and primary bone tumors with soft tissue components. Fourth, this is a single-center study with a small sample size, decreasing the statistic power.

Conclusion

TWIST is a reliable imaging modality for the assessment of lower extremity tumors and is comparable to CTA for the identification of arterio-venous fistulae and feeding arteries.