MR Imaging of Brain Tumors: Toward Physiologic Imaging

A few short years ago, the introduction of contrast agents for MR imaging led to much excitement about the possibilities for evaluating the presurgical workup of patients with brain tumors. Tumor detection became much more feasible, blood brain–barrier (BBB) breakdown could be detected, and an assessment of the tumor vascularity (ie, its blood volume) became possible. Nonetheless, the mere observation of contrast enhancement does not provide answers to all the interesting questions facing the clinician helping patients with brain tumors. Where is the most malignant region within a tumor? Can we better predict the patient's prognosis, or better follow up on a patient after therapy? One reason these questions are difficult to answer is that contrast enhancement represents the combination of two parallel physiologic processes. Intravenously injected contrast agent arrives in the blood pool of a tissue, reflecting the tumor vascularization, and, in the presence of a BBB breakdown, it leaks into the interstitial space, reflecting the microvascular permeability.

Subsequently, several strategies have been developed to separate these two parameters to achieve more useful, quantifiable information on tumor biology and behavior. Most work has been focused on quantification of the tumor's vascularity, or relative cerebral blood volume (rCBV), a parameter historically used in nuclear medicine studies such as positron emission tomography (PET) and single-photon emission computed tomography (SPECT) (1). rCBV has been studied extensively, and elevations in this parameter have been found to have a good specificity for tumor malignancy. PET and SPECT, however, are limited by poor spatial resolution, which results in insensitivity to small lesions; they are also insensitive to low-grade intraaxial brain tumors. Furthermore, PET still lacks sufficient availability to serve as a viable clinical approach for routine application.

MR imaging with T1-weighted first-pass techniques to assess a contrast agent's transit through the brain has been applied to this problem, but the low blood volume of cerebral tissue (less than 10%) makes it difficult to appreciate changes in signal intensity after contrast injection. Dynamic contrast-enhanced MR imaging gained popularity following the introduction of echo planar–capable systems with high-performance gradients. Fast imaging with high sensitivity to changes in magnetic susceptibility allows assessment of the T2* effect on contrast-enhanced images, which is much stronger than the T1 effect and extends over a wider area. Consequently, several studies have been performed using some kind of T2* echo-planar MR imaging to quantify the rCBV of tumor tissue (2, 3). These slightly different protocols for rCBV measurement require a standardized approach in order to make the results comparable across studies. One worthwhile approach is taken by Roland Bruening et al in their article Effects of Three Different Doses of a Bolus Injection of Gadodiamide: Assessment of rCBV Maps in a Blinded Reader Study in this issue of the AJNR (page 1603). The authors focus on one aspect of first-pass dynamic T2*-weighted MR imaging, the contrast agent dose. Their conclusions are valid and in agreement with anecdotal experience in day-to-day practice that the bolus injection of 0.2 mmol/kg body weight is an adequate dose of gadodiamide for the measurement of rCBV.

Unfortunately, life is not that simple and T2*-weighted imaging is influenced by parameters other than just the contrast agent dose. Thus, other factors influence the calculation of rCBV, including whether a spin-echo (SE) echo-planar sequence or a gradient-recalled (GRE) echo-planar sequence is used. Recently, from the famous brain/vein debate in fMR imaging, it has been found that GRE sequences could be dominated by large vessel contributions. On the other hand, using an SE sequence (or an asymmetric SE sequence) increases the relative contribution of small to large vessels, which may be more relevant in determining tumor type. In any case, one would expect a different dose dependence for an SE echo-planar sequence, which trades off the sensitivity of the GRE sequence against improved microvascular specificity. Quite important for echo-planar sequences and T2* measurements is a standardized echo time. Different dose response characteristics might be expected with varying echo times. Other sequence parameters may affect the T1 contribution, confounding the T2* effects. T1 saturation occurs with a shorter repetition time and higher flip angle; the more T1 effect the greater the contamination, particularly in the second half of a contrast agent bolus curve. Thus, there are many parameters to consider when attempting to standardize the acquisition of rCBV.

Furthermore, going back to the two contributing aspects of contrast enhancement, there is more to physiologic imaging of brain tumors than just rCBV calculations. One must not forget about the BBB breakdown in tumors. The resulting extravasation of contrast agent during the observation of the first pass leads to errors in the calculation of rCBV, as mentioned by the authors. While T1 effects cause the contrast agent curve not to return completely to baseline, the leakage of contrast agent into the interstitium yields a return above baseline. Thus, measures have to be taken to correct for this effect, particularly in high-grade tumors. In addition, instead of just correcting for the effect of BBB breakdown and the contrast agent extravasation into the interstitum, the microvascular permeability can potentially be used as an independent variable in its own right. The rate of transendothelial diffusion is related to the integrity of the vascular wall in general and the BBB in particular, and these are reflections of tumor angiogenesis. Indeed, the feasibility of the approach to use dynamic contrast-enhanced MR for the noninvasive assessment of tumor microvascular characteristics has recently been demonstrated (4). Such a technique can be used not only to assess tumor characteristics, but potentially to monitor new cancer treatments, like angiogenesis inhibitors.

Where do we have to go with the physiologic imaging of brain tumors? The trend is definitely toward standardization of established techniques like T2*-weighted dynamic contrast-enhanced MR imaging. But standardization should also include other parameters that might confound these measurements. This will then allow comparison of results from different centers. Physiologic imaging should incorporate many tools, not only one physiologic parameter like blood volume. We should be evaluating the vascular permeability of a tumor, a reflection of tumor angiogenesis, as well as cellular metabolic profiles obtained using MR spectroscopy, and integrating this information within the context of our anatomic maps. In this way, the information we contribute can be used by physicians and in clinical trials in a meaningful way.