Authored date:2006-05-22

Introduction

Detection of local and regional lymph node metastases is important for determining therapy and prognosis in patients with various primary malignancies. Currently cross sectional imaging modalities like contrast enhanced* CT and MRI rely solely on size criterion as the primary yardstick for detection of malignant lymph nodes. However nodal size is not an accurate method, as not all malignant nodes are enlarged and conversely not all large nodes are malignant [1]. Thus a more robust and accurate technique is needed for staging lymph nodes which characterizes lymph nodes independent of their size. MRI enhancement with ultra-small super-paramagnetic iron oxide particle (USPIO) has shown to be an accurate and reliable imaging technique for detection of minimal nodal metastatic disease independent of lymph node size.
Commercial preparation of USPIOs like Ferumoxtran-10 (Combidex [Advanced Magnetics, Cambridge, Mass]; also known as Sinerem, AMI-7227, AMI-227, and BMS 180549) target the reticulo-endothelial system and have been specifically developed for MR lymphangiography. These agents are composed of an iron oxide crystalline core of 4.3–6.0 nm covered by low-molecular-weight dextran. Their T1 and T2 relaxations are 2.3 x 10⁴ and 5.3 x 10⁴ mol^-1 sec^-1 (20 MHz, 39°C), respectively, in 0.5% agar.
Upon intravenous injection, these contrast agents are transported to the lymph nodes binding to specific receptors within the nodal macrophages. If the lymph nodes are benign, the macrophages function normally and phagocytose the intravenously administered nanoparticles. Thus normal lymph nodes show a drop in signal intensity on MRI imaging after Ferumoxtran-10 administration due to the susceptibly from phagocytosed nanoparticles. In contrast, malignant infiltration results in lack of macrophages and hence these regions, due to the lack of iron oxide uptake, retain their high signal intensity. This is the underlying principle behind MR lymphangiography.

Imaging

Metastasis in normal-sized lymph nodes can be of the order of 4–9 mm [2], hence from the MR viewpoint it is very essential that the maximum voxel dimension should not exceed 2 mm for capturing even the least amount of metastasis in a lymph node. While this restriction was routine in the case of in-plane dimensions, through-plane dimensions were restricted in the case of multi-echo gradient echo (MEGE) sequences. MEGE sequences are very essential for MR lymphangiography as they facilitate the computation of T2* which is an indicator of malignancy. This problem has been alleviated with the development of the new works in progress T2*-weighted Multi-Contrast Gradient Echo (MCGE) sequence at Siemens Medical Solutions.

The Sequence

Currently the MCGE sequence supports a maximum of 12 contrasts with the facility to choose a bi-polar or mono-polar readout. While bi-polar readouts are standard for Gradient Echo sequences, they suffer from chemical shift artifacts that shift between echoes. This essentially destroys any inplane resolution advantage as they tend to spoil the computation of T2* in the regions that are chemically shifted. The mono-polar readout, however, maintains the chemical shift by using a gradient rewinder between readouts so that all the k-space lines are acquired in the same direction. The availability of integrated Parallel Acquisition Techniques (iPAT) in these sequences further saves acquisition time.

The Protocol

Currently the protocol has been optimized for pelvic studies and has been successfully tested on patients. Apart from the MCGE sequence, a VIBE and T2-weighted spin echo sequences are used for obtaining spatial and anatomical characterization of the nodes. The protocol is applied twice, once before the administration of the contrast agent and 24 hours after the administration of the contrast agent. The individual components of the protocol as currently tested on a 1.5T MAGNETOM Avanto are as follows:
1. A T1-weighted (T1W) 3D VIBE non breath-hold sequence with TE 1.9 ms, Flip Angle = 12º, TR 5 ms, BW 260 Hz/Px, FOV 340 x 340 mm and 1.3 x 1.3 mm and interpolated slice thickness of 2.68 mm acquired coronally to cover around 100 slices acquired in 2 minutes. These images are required for a MIP of the vasculature that will serve as a reference for location of the nodes.
2. A T2-weighted TSE non breath-hold sequence with TE 76 ms, Flip Angle 90º, TR 4000 ms, BW 160 Hz/Px, FOV 340 x 340 mm and slice thickness of 2.68 mm acquired axially to cover around 30 slices acquired in roughly 3 minutes. Though the T2-weighted images are not preferred for quantification of the metastasis, they are excellent in terms of signal- to-noise ratio (SNR) and devoid of blooming artifacts resulting from the susceptibility of iron, so node delineation and identification is easier on these images.
3. The T2*-weighted MCGE non breath-hold sequence with TE 8.8 ms to 70 ms (12 contrasts), Flip Angle 75º, TR 1800 ms, BW 130 Hz/Px, FOV 214 x 380 mm and slice thickness of 2 mm acquired axially to cover 45 slices in 6 minutes with iPAT factor of 2. All sequences use a 0 mm gap between slices.

Initial Experiences

After initially experimenting on an oil-water phantom, the protocol was tried on a patient. The T2* MCGE sequence was also run in the bi-polar mode for comparison. In the case of bi-polar mode the acquisition time was reduced roughly by a factor of two due to the absence of the rewinder between echoes. The images from the first echo and the corresponding T2* computed from the echoes are presented in Fig.1. It can be seen very clearly that for the same echo time and windowing that the image from the monopolar sequence is crisper and hence the computed T2* is more reliable. Fig. 2 shows pre- and post-contrast T2-weighted images. Note how the node is easy to discern, especially the partial fatty infiltration. Fig. 3 is a VRT rendering of the vasculature with segmented nodal information from the T2 images superimposed with color coded information from the T2*s. This has been achieved by the use of a works in progress syngo task card that has been optimized for the workflow of the current protocol.

Fig. 1  Comparison between bi-polar and mono-polar images. Images in top row are T2* maps and bottom row are single echo images at same echo time. Images on the left are from the bi-polar sequence and on the right are from the mono-polar sequence. Note how the images from the mono-polar sequence have more detail than those from the bi-polar sequence. This is due to the averaging effect of the alternating phase shifts in the bi-polar sequence.

Fig. 2  T2-weighted TSE axial slice showing partial fatty infiltration in the node. The image on the left is 
pre-contrast and on the right is post-contrast. Note that on the post-contrast image a portion of the node still remains bright while the rest of the node has become very dark. This is because the USPIO is not absorbed by the fatty infiltration. 
To differentiate fatty infiltration from metastases, the shape of the lymph node and the hypo intensity need to be taken into account. A fat suppressed T1 can also give clues on the fatty infiltration.

Fig. 3  Segmented nodes color coded on a VRT rendered vasculature. Information from segmented nodes is coded with unique values that are then combined with information from the T1-weighted 3D VIBE images to obtain the VRT rendering of the nodes on the vasculature. Benign nodes are coded green and malignant red.

Conclusion

With the usage of a mono-polar multi-contrast GRE sequence it has been demonstrated that the estimation of T2* is vastly improved, thereby improving the detection of metastatic lymph nodes.

Future Directions

With Tim (Total imaging matrix) technology, it is now possible to flexibly combine coils to increase the field of view for metastatic evaluation beyond a localized region. Implementation on 3T would offer interesting insights and challenges. While there will be an improvement in SNR and reduction in acquisition time, susceptibility artifacts will be more pronounced.
Optimization with respect to echo times and number of echoes is in progress. Reduction in number of echoes without any increase in noise would greatly help reduce the acquisition time, since with breath-holding in the case of abdominal studies time will be a factor. Addition of navigators on the other hand would also help in the case of abdominal studies.

Acknowledgements

We wish to acknowledge John Kirsch of Siemens Medical Solutions whose expert advice was crucial to the optimization of the protocols.

* This information about this product is preliminary. The product is under development and not commercially available in the US, and its future availability cannot be ensured.

References

[ 1 ] Harisinghani M.G., Weissleder R., Sensitive, Noninvasive Detection of Lymph Node Metastases, PLOS 2004, Vol.1., 
(3): 202–209. 
[ 2 ] Deserno W.M.L.L.G., Harisinghani M.G., et. Al., Urinary Bladder Cancer: Preoperative Nodal Staging with 
Ferumoxtran-10 enhanced MR Imaging, Radiology 2004; 233: 449–456.