Validation Studies of Dynamic Contrast Enhanced MRI

There are numerous imaging and modeling strategies available for the application of DCE-MRI. The results of so many studies are difficult to compare due to the wide variety of existing mathematical models, each having their own assumptions and constraints (Tofts 1997).

The DCE-MRI approach is widely used to draw inferences into microvascular parameters such as microvascular permeability, blood volume, and tissue perfusion. Confirmed insight into the whereabouts, and amount of the tracer, at the tissue level would unquestionably help to determine the most accurate and precise way to model the signal changes in DCE-MRI. This can be accomplished by correlating DCE-MRI with quantitative transmission electron microscopy (TEM) methods. TEM can be used, in combination with energy dispersive X-ray spectrometry (EDXS) microanalysis to assess the subcellular content and location of heavy metals like gadolinium and iron (Elster 1989; LoPachin and SaubermanN 1990; Taherzadeh et al. 1998). This method analyses the characteristic X-ray patterns, produced from the heavy metal based MRI contrast agents when an electron beam passes through the tissue. Where the concentration is high enough elemental distribution maps can be produced (LoPachin and Saubermann 1990). A previous study has shown the feasibility and necessity of combined MRI/EDSX studies (Hawkins et al. 1990). This was done to examine whether Gd-DTPA leaked out of the cerebral vasculature of Guinea pigs with experimental allergic encephalo-myelitis, a multiple sclerosis-like disorder that causes breakdown of the BBB. Neither ultrastructural MRI contrast agent quantification nor elemental map

Myelitis Mri Contrast

Fig. 2.3. Clinical imaging of brain tumor permeability. Tj-weighted 3D imaging is used dynamically during a bolus injection of paramagnetic contrast agent (gadolinium-based). A representative slice from a 3D data set is shown at 6-s intervals (left) revealing signal enhancement after the fourth frame, which progresses in a metastasis (right, anterior portion of brain). Constructing the ratio of tumor enhancement to reference blood vessel enhancement (from the sagittal sinus) and plotting as a function of time indicates the rate of such progressive enhancement (right, upper). Kinetic modeling allows construction of a pixel-by-pixel map of the permeability, Kps (right, lower), clearly delineating the tumor (arrow).

Dynamic Enhancement (6 s / 3D)

Fig. 2.3. Clinical imaging of brain tumor permeability. Tj-weighted 3D imaging is used dynamically during a bolus injection of paramagnetic contrast agent (gadolinium-based). A representative slice from a 3D data set is shown at 6-s intervals (left) revealing signal enhancement after the fourth frame, which progresses in a metastasis (right, anterior portion of brain). Constructing the ratio of tumor enhancement to reference blood vessel enhancement (from the sagittal sinus) and plotting as a function of time indicates the rate of such progressive enhancement (right, upper). Kinetic modeling allows construction of a pixel-by-pixel map of the permeability, Kps (right, lower), clearly delineating the tumor (arrow).

ping were performed, due to the equipment limitations. However, this study showed the feasibility of combining both methods in the study of microvas-culature. With the rapid advances made in MR and EDSX imaging technologies over the last 5 years, the merging of the macro and microscopic technologies is feasible.

There are cheaper and simpler microscopic imaging methods (e.g., light microscopy), other than the EDSX electron microscopy methods. Macromolecu-lar contrast agents containing Gd have been histo-chemically assessed using light microscopy (Saeed et al. 1998). Van Dijke et al. (2002) demonstrated via light microscopy post-in vitro streptavidin staining that even macromolecular contrast agents permeate out of the intravascular space over a period of 1 h in a breast cancer model. However, no current micro scopic imaging method can image heavy metals such as gadolinium in small chelates. There are more sensitive biochemical methods, other than EDXS, for measuring heavy metal content. For example, inductively coupled plasma (ICP) atomic emission spectrometry, polarized X-ray fluorescence excitation analysis (FEA), and HPLC. However, these procedures lack the ability to localize the contrast agent precisely within the tissue architecture (Elster 1989). The tissue specimen must be first homogenized, digested, or vaporized, before they can be analyzed by the ICP-AES, FEA, or HPLC instruments.

In an experiment to lo calize the subcellular lo cation of Gd-DTPA, New Zealand white rabbits implanted with vx2 tumor cells (Park et al. 1998) were evaluated using a combination of DCE-MRI and analytical electron microscopic methods (Noseworthy et al. 2002). Following DCE-MRI and parametric mapping of microvascular permeability, portions of muscle and tumor were examined under a field emission scanning electron microscope (FESEM). Using quantitative energy dispersive X-ray spectroscopy gadolinium was detected in the lumen of blood vessels in both tumor and muscle tissues, as expected, at concentrations similar to that observed in plasma collected prior to euthanasia. Interestingly at 5 min post-euthanasia, a time when the extra and intravas-cular spaces are theoretically in equilibrium, with respect to Gadolinium concentration, there was a four-fold larger Gd accumulation in the tumor extra-vascular space (Fig. 2.4). And, contrary to all reports on the in vivo behavior of small molecular weight Gd-based contrast agents, Gd was detected within the endothelial cells lining blood vessels of both tumor and muscle. This study showed that the current mathematical models for fitting DCE-MRI data may be oversimplified. The nature of the intracellular signal was speculated to arise from Gd being transported within vesiculo-vacuolar organelles (VVOs), which are known to transport molecules across vascular endothelial cells between intra- and extravas-cular spaces, or within transendothelial cell channels (Dvorak and Feng 2001). With the existence of an intracellular pool of Gd it is possible, therefore, that current DCE-MRI models could be overestimating blood volume or extravascular volume. In other experiments examining high molecular weight iron (Fe) particles (Ultrasmall superparamagnetic iron oxide particles, USPIOs), Fe was localized in only the intravascular space (unpublished observation).

A form of validation is also offered by comparing imaging derived parameters of microvascular characterization with histological assessment of microvascular density (MVD), assessed by staining with Factor VIII or CD-31. Critics of this approach point to sampling errors in comparing "hot spots" of immunohistochemical staining under light micros

T1-weighted SPGR

Electron Micrograph (vasculature)

T1-weighted SPGR

Electron Micrograph (vasculature)

Oesophage Marteau Piqueur
X-Ray Spectra

Intracellular (ic) (vascular endothelia)

Tumor extravascular (ev)

Tumor intravascular (iv)

Intracellular (ic) (vascular endothelia)

Tumor extravascular (ev)

Fig. 2.4. Rabbit vx2 tumors were analyzed using DCE-MRI to produce maps of permeability. Regions exhibiting hypervascular permeability were biopsied, cryopreserved, and analyzed using energy dispersive X-ray microanalysis. Gd concentration in intravascular (iv), extravascular (ev), and endothelial intracellular (ic) spaces were assessed at 5 min post injection. Gd was found to be four times higher in the extravascular space, relative to intravascular. In addition, Gd was localized within vascular endothelial cells (ic)

Fig. 2.4. Rabbit vx2 tumors were analyzed using DCE-MRI to produce maps of permeability. Regions exhibiting hypervascular permeability were biopsied, cryopreserved, and analyzed using energy dispersive X-ray microanalysis. Gd concentration in intravascular (iv), extravascular (ev), and endothelial intracellular (ic) spaces were assessed at 5 min post injection. Gd was found to be four times higher in the extravascular space, relative to intravascular. In addition, Gd was localized within vascular endothelial cells (ic)

copy with regional assessments derived from macroscopic imaging. Furthermore, the MVD assessment will provide a count of all visible vessels, not simply "functional" or perfused vessels. Nonetheless, coarse agreement between microvascular characterization derived from imaging with MVD is usually demonstrated (e.g., van Dijke et al. 1996).

Other attempts at validating dynamic DCE-MRI-derived parameters have taken advantage of permeability reducing pharmaceuticals such as Avastin (a monoclonal antibody to the VEGF molecule) (FerrarA 2002). Demonstrating "sensitivity to change," or reduction of permeability post-treatment with such an agent has been interpreted as confirming both the mode of action or biological effect of the pharmaceutical as well as providing some form of experimental utility validation for the microvascular characterization afforded by DCE-MRI (Pham et al. 1998; Brasch et al. 1997; Gossmann et al. 2002; Roberts et al. 2002a,b).

0 0

Post a comment