VOLUME 24 ⅐ NUMBER 20 ⅐ JULY 10 2006 JOURNAL OF CLINICAL ONCOLOGY REVIEW ARTICLE Dynamic Contrast-Enhanced Magnetic Resonance Imaging As an Imaging Biomarker Nola Hylton From the University of California, San ABSTRACT Francisco, San Francisco, CA. Submitted March 28, 2006; accepted Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is being used in oncology as a March 29, 2006. noninvasive method for measuring properties of the tumor microvasculature. There is potential for DCE-MRI to be used as an imaging biomarker to measure antiangiogenic effects of cancer treatments. This article Author’s disclosures of potential reviews the general methodology for performing DCE-MRI and discusses existing data and challenges to conflicts of interest are found at the applying DCE-MRI for treatment response assessment in clinical trials. end of this article. Address reprint requests to Nola J Clin Oncol 24:3293-3298. © 2006 by American Society of Clinical Oncology Hylton, PhD, Box 1290, University of California, San Francisco, San Francisco, CA 94143-1290; e-mail: Pharmacokinetic modeling of the DCE-MRI INTRODUCTION [email protected]. signal is used to derive estimates of factors related to © 2006 by American Society of Clinical Dynamic contrast-enhanced magnetic resonance blood volume and permeability that are hallmarks Oncology imaging (DCE-MRI) is a noninvasive imaging tech- of the angiogenic phenotype associated with most 0732-183X/06/2420-3293/$20.00 nique that can be used to measure properties of cancers. Tumor angiogenesis as measured immuno- DOI: 10.1200/JCO.2006.06.8080 tissue microvasculature. DCE-MRI is sensitive to histochemically by MVD, has been shown to be an differences in blood volume and vascular perme- independent prognostic indicator, and angiogenesis ability that can be associated with tumor angiogen- is a direct or indirect target of many new anticancer 4 esis, and thus DCE-MRI is a promising method and agents. Thus, there is great interest in developing potential biomarker for characterizing tumor re- DCE-MRI as a biomarker for angiogenic activity in sponse to antiangiogenic treatment.1-3 tumors. Although data suggest that DCE-MRI has DCE-MRI has been investigated for a range potential in this regard, there is a need to standardize of clinical oncologic applications including cancer techniques for both acquiring DCE-MRI data and defining how the imaging biomarker is quantified. detection, diagnosis, staging, and assessment of The accuracy of DCE-MRI relies on the ability treatment response. Tumor microvascular mea- to model the pharmacokinetics of an injected tracer, surements by DCE-MRI have been found to corre- or contrast agent, using the signal intensity changes late with prognostic factors (such as tumor grade, on sequential magnetic resonance images. Signal in- microvessel density [MVD], and vascular endothe- tensity changes can be rapid immediately after lial growth factor [VEGF] expression) and with re- (small molecular weight) contrast injection, and currence and survival outcomes. DCE-MRI changes thus the temporal sampling rate is important. How- measured during treatment have been shown to cor- ever, increasing the temporal sampling rate of MRI relate with outcome, suggesting a role for DCE-MRI has direct consequences on critical image character- as a predictive marker. istics such as spatial resolution, signal-to-noise ratio, With the accelerating pace of drug develop- and the volume of anatomy covered. The trade-offs ment, there is a desire to identify biomarkers that between temporal resolution and spatial resolution can be used to assess tumor biology in vivo and to for DCE-MRI are not clear, and are not easily tested. monitor the effects of treatment. The concept of an MVD measured histopathologically gives a partial imaging biomarker is very appealing. An imaging picture of the tissue microvasculature, but does not biomarker can be measured noninvasively and reflect its functional properties, including perme- repeatedly, and by evaluating the entire tumor in ability, that contribute to the DCE-MRI measure- vivo, can capture the heterogeneity of both the ment. Thus, it is not surprising that studies reporting tumor and its response to treatment. DCE-MRI is correlations between DCE-MRI parameters and a particularly attractive method because of the MVD have found only moderate associations. MVD intrinsic soft tissue contrast and anatomic detail is also a heterogeneous property of tumors. MVD provided by MRI in general, and the added ability measurement methods are limited by histopatho- of DCE-MRI techniques to measure properties of logic sampling and are generally hotspot values, the microcirculation. which are, by definition, localized. Associations have 3293 Information downloaded from www.jco.org and provided by INSTITUTE OF CANCER RSRCH on February 15, 2007 from 193.63.217.208. Copyright © 2006 by the American Society of Clinical Oncology. All rights reserved. Nola Hylton ⌬ been reported between independent MVD and DCE-MRI hotspot intensity ( S) to contrast agent concentration in the tissue (Ct), the measurements, although direct spatial correlation between the two precontrast tissue T1 value is needed. The tissue contrast as a function has generally not been attempted. Most of the evidence in support of of time Ct(t) also depends on the arterial blood plasma concentration DCE-MRI is based on correlation of imaging parameters with histo- as a function of time [Cp(t)], which varies depending on the mode of pathology and accepted prognostic factors such as tumor grade, met- injection (short v long bolus) and is affected by differences in cardiac astatic status, and clinical outcome, arguably the more important end output among subjects. Variability in Cp(t), also called the arterial points on which to establish the value of DCE-MRI. input function (AIF), can have a sizable effect on pharmacokinetic There are significant challenges to developing robust imaging parameters. To properly account for these effects, Cp(t) should be biomarkers. It requires both establishing that one or more functional measured for each patient, generally by including a large vessel in the measurements sensitively capture the biology of interest, and defining imaging field of view. C (t) and C (t) can be related through a gener- a measurement method that can be applied in a reliable and standard- t p alized kinetic model6: ized fashion. Specification of the measurement method can be com- plex, as in the case of MRI, where many experimental variables dC /dt ϭ Ktrans͑C – C /␷ ͒ ϭ KtransC Ϫ ␬ C influence the signal. Thus, optimizing and subsequently standardizing t p t e p ep t functional imaging measurement methods presents a significant task. where Ktrans is the volume transfer constant between the blood plasma and extravascular extracellular space (EES) per unit volume of tissue Ϫ1 ␬ DCE-MRI MEASUREMENT METHODS (min ), ep is the rate constant between the EES and blood plasma Ϫ1 ␷ (min ), and e is the volume of extravascular extracellular space per 6 Basic Principles and General Methodology unit volume of tissue. In a 1999 consensus publication, this set of DCE-MRI is performed by obtaining sequential magnetic reso- terms was recommended by an international group of investigators nance images before, during, and following the injection of a contrast developing DCE-MRI methodologies. The authors related the three trans ␬ ␷ agent. For human studies, the contrast agent is generally a small parameters, K , ep and e, to previously published terms and molecular weight gadolinium-containing compound such as gado- symbols and proposed that this set of kinetic parameters and symbols pentetate dimeglumine. T2* or susceptibility-weighted MRI can be be used universally to describe the uptake of low molecular weight used early after contrast injection (in the first few seconds) to observe gadolinium-based contrast agents that are in clinical use today. the transient first-pass effects of contrast agent, which provides infor- To adequately apply a two-compartment pharmacokinetic mation about perfusion. The T2* effect is measured as a rapid drop model to the MRI enhancement time course, measurement of the and subsequent recovery of signal intensity after bolus injection. Mea- intrinsic T1 value and AIF are needed. The need to measure T1 in- surement of first-pass T2* effects necessitates a rapid imaging method creases total scan time and may not be feasible in clinical practice. In that is generally performed over a single slice through the target tissue the absence of a baseline T1 measurement, an assumption of linearity and is thus of limited value in assessing disease morphology or extent. between signal intensity and gadolinium concentration can be made Dynamic T2* methods are less commonly used than dynamic T1- (removing the need for baseline T1), or the signal-intensity time curve weighted methods, outside of the brain. can be quantified using empirical quantitative measures such as the Dynamic T1-weighted imaging is used over a longer time course initial area under the curve, peak enhancement, time to peak enhance- (in the first several minutes) to observe the extravasation of contrast ment, or signal enhancement ratio (SER). It may also be infeasible to agent from the vascular space to the interstitial space, providing infor- include a large vessel in the field of view appropriate for measuring the mation about blood volume and microvascular permeability. The AIF. Expanding the field of view to accommodate a large artery may accumulation of contrast agent in the interstitium results in a signal compromise image resolution. The AIF requirement is often ad- increase on T1-weighted MRI. A subsequent wash-out effect can be dressed using average values measured in healthy control subjects observed if the vascular permeability is high and there is reflux of contrast agent back to the vascular space. from blood samples, which have been reported in the literature. Signal intensity will change in proportion to the contrast agent The relative merit of alternative approaches and the impact of concentration in the volume element of measurement, or voxel.