Introduction

Dynamic contrast enhanced magnetic resonance imaging (DCE-MRI) has been widely used to assess tissue microenvironment of cancer by measuring blood flow-related parameters such as perfusion, permeability or vascular volume. FDG PET is the modality of choice to assess abnormally high rates of glucose metabolism in cancer cells. However, the use of simultaneous MRI and PET to probe microenvironment changes has not been fully investigated. The purpose of this study is to demonstrate the synergistic role of FDG PET and DCE-MRI for assessing the metastatic status of regional lymph nodes in head and neck cancer.

Methods

Patients (n=20) with head and neck cancer who were scheduled for node dissection surgery were recruited in this study. All patients underwent one research PET/MR scan within one week prior to the planned surgery. The imaging study was conducted using a whole body 3T PET/MR scanner (Biograph mMR, Siemens Healthcare) at the Center for Biomedical Imaging, NYU School of Medicine. During the dissection surgery, the locations of removed lymph nodes were marked according to the conventional neck lymph node levels, in order to perform level-to-level comparison between PET/MR parameters and pathology. The PET acquisition started 1 min before injection of the PET tracer (10 mCi of 18F-FDG) and conducted in a dynamic fashion for 60 min. The attenuation correction map was estimated using the water and fat images generated from a T1-weighted Dixon gradient-echo sequence. PET images were reconstructed on the vendor platform incorporating the attenuation map which provides 127 axial slices with 344x344 matrix at 2x2x2 mm3 voxel size. A two compartmental model (1) was used to estimate the forward rate constant between blood and tissue compartment (K₁) is the reverse rate constant between blood and tissue compartments (k₂), and the rate constant representing trapping of FDG intracellularly after phosphorylation by hexokinase (k₃). While acquiring dynamic PET data, DCE-MRI data were acquired simultaneously using a golden-angle radial 3D gradient echo sequence with TR/TE=3.6/1.7ms, FA=10deg, Res=1x1x2mm. After acquisition of baseline images for 1 min, a single dose of Gd-DTPA (0.1 mM/kg body weight) was injected at 1 mL/s into an antecubital vein followed by saline flush, during which scanning was continued for another 9 min. The dynamic images were reconstructed using the GRASP method (2). MR contrast kinetic model parametric maps were computed using the modified Tofts model (3) with the arterial input function (AIF) estimated from a manually selected region of interest (ROI) in the cervical segment of the internal carotid artery. Estimated contrast kinetic parameters were the volume transfer constant between blood plasma and the interstitium (K^trans), the transfer constant from the interstitium to the blood plasma (K_ep), and the plasma volume fraction (V_p). Both PET and MRI data analyses, including co-registration between them, were performed using FireVoxel (https://wp.nyu.edu/firevoxel). ROIs were drawn on lymph nodes at the levels matched with the surgical pathology reports by a board certified head and neck radiologist. MRI and PET pharmacokinetic parameters were compared between pathologically proven metastatic nodes and negative ones. Statistical data analyses were performed using Graphpad Prism version 7.03 (GraphPad Software, La Jolla, CA, USA).

Results

Figure 1 shows a representative example of MRI and PET images with good alignment of two images from the simultaneous acquisition. The pharmacokinetic model analysis of dynamic PET data showed that metastatic nodes have significantly (p<0.001) higher rate of FDG phosphorylation (k₃) than normal nodes (Fig.2b). The vascular delivery of FDG, measured as K1, was higher in the metastatic nodes than in the normal nodes with no statistical significance (Fig.2a). The median Ktrans of MRI contrast agent was significantly (p=0.0192) lower in the metastatic nodes than in the normal nodes (Fig.2c). Assuming that the tumor vascular leakiness is high for both FDG and Gd-based contrast agent, the ratio of K₁ to K^trans was calculated as a surrogate marker for FDG cellular uptake including the expression of glucose transporters. As shown in Fig.2d, the median K₁/K^trans of the metastatic nodes was significantly (p=0.0222) higher than that of the normal nodes. The scatter plot of k₃ and K₁/K^trans (Fig.3) shows that the normal nodes are clustered in the region with low k₃ and low K₁/K^trans values, whereas the metastatic nodes tend to have high k₃ and high K₁/K^trans ratios.

Discussion

The combined DCE-MRI and FDG PET study has been successfully demonstrated in this study and can be used to assess the metastatic status of lymph nodes of head and neck cancer patients. Future studies will include the application of these kinetic parameters for detection of micro-metastasis in lymph nodes as well as extracapsular invasion.

Acknowledgements

NIH grant R21CA188217