Professional Documents
Culture Documents
ABBREVIATION KEY
BBB ⫽ Blood-Brain Barrier
CBF ⫽ cerebral blood flow
CBV ⫽ cerebral blood volume
DCE ⫽ dynamic contrast-enhanced
DSC ⫽ dynamic-susceptibility
contrast
FDA ⫽ US Food and Drug
Blood-Brain-Barrier Imaging in Brain Administration
GBM ⫽ glioblastoma multiforme
Tumors: Concepts and Methods HIF-1␣ ⫽ hypoxia-inducible factor
IRF ⫽ impulse residue function
kep or kb ⫽ reverse transfer constant
Rajan Jain, MD, Brent Griffith, MD, Jayant Narang, MD, Tom Mikkelsen, MD,
Ktrans ⫽ forward transfer coefficient
Hassan Bagher-Ebadian, PhD, Siamak P. Nejad-Davarani, PhD, James R. Ewing, PhD, MSIVP ⫽ maximum slope of
and Ali S. Arbab, PhD, MD enhancement in the initial vascular
phase
MVCP ⫽ microvascular cellular
proliferation
ABSTRACT MVD ⫽ microvascular density
Malignant gliomas are often very heterogeneous tumors with complex vasculature, fre- nIAUC ⫽ initial area under the
quently exhibiting angiogenesis and increased vascular permeability. In vivo measurement normalized time-intensity curve
of the tumor vessel permeability can serve as a potential imaging biomarker to assess tumor nSDEP ⫽ normalized slope of the
delayed equilibrium phase
grade and aggressiveness. It can also be used to study the response of tumors to various
PCT ⫽ perfusion CT
therapies, especially antiangiogenic therapy. Central to the concept of permeability is a PS ⫽ permeability surface-area
thorough knowledge of the BBB and its role in brain tumors and angiogenesis. Much work product
has been done in the past to understand the structural/molecular composition of the BBB PSR ⫽ percentage signal-intensity
and the role it plays in various pathologic processes, including brain tumors. Various imag- recovery
rPSR ⫽ relative percentage signal-
ing techniques have also been used to evaluate BBB leakiness in brain tumors because
intensity recovery
higher tumor vascular leakiness is known to be associated with higher grade and malignant rCBV ⫽ relative cerebral blood
potential of the tumor and hence poor patient prognosis. These imaging techniques range volume
from routine postcontrast T1-weighted images to measurement of vascular permeability SDF-1 ⫽ stromal derived factor-1
using various quantitative or semiquantitative indices based on multicompartment pharma- TDLs ⫽ tumefactive demyelinating
lesions
cokinetic models. The purpose of this article is to discuss BBB anatomy; various clinically
TVA ⫽ total vascular area
available imaging techniques to evaluate tumor vascular leakiness (perfusion imaging), VEGF ⫽ vascular endothelial growth
including their advantages and limitations; as well as a brief discussion of the clinical utility factor
of measuring vascular permeability in brain tumors. We will also discuss the various perme- VEGFR-2 ⫽ vascular endothelial
ability-related indices along with the pharmacokinetic models to simplify the “nomenclature growth factor receptor-2
soup.”
Received August 1, 2011; accepted
after revision November 27.
From the Division of
INTRODUCTION multiple angiogenic cytokines. Tumors Neuroradiology, Department of
Radiology (R.J., B.G., J.N., A.S.A.)
The BBB consists of a complex of capillary larger than 1–2 mm1 cannot grow by rely- and Departments of Neurosurgery
endothelial cells, pericytes, and astroglial ing on passive transport of nutrients and (R.J., T.M.) and Neurology (S.P.N.-
D., J.R.E., H.B.-E.), Henry Ford
and perivascular macrophages, serving as oxygen. Further growth is supported by Health System, Detroit, Michigan.
an effective physical barrier to the entry of angiogenesis promoted by proangiogenic Please address correspondence to
Rajan Jain, MD, Division of Neuro-
lipophobic substances into the brain. How- factors; the most important among those is radiology, Departments of Radiol-
ever, in many brain tumors, as well as other VEGF.1-4 Animal studies indicate that neo- ogy and Neurosurgery, Henry Ford
Health System, 2799 West Grand
disease processes, the BBB becomes inter- angiogenesis and increased vascular per- Blvd, Detroit MI 48202; e-mail
rupted. It is this disruption and other meability are essential for the survival and rajanj@rad.hfh.edu
http://dx.doi.org/10.3174/ng.2120028
changes to the BBB in tumors that contrib- proliferation of tumor cells.5 An under-
ute to contrast enhancement on imaging standing of this process is essential because
and serve as a potential surrogate imaging it provides the physiologic basis for perfu-
marker. sion imaging in tumors. Tumor angiogen-
Malignant gliomas are hypervascular esis involves a multitude of controlled sig-
tumors with very heterogeneous and com- naling cascades and structural changes that
plex vasculature, which is regulated by occur in a defined order and continue until
PS Characterizes the diffusion of contrast agent from the blood • Usually measured using CT perfusion
vessels into the interstitial space due to deficient or leaky • Commonly used units: mL/100 g/min
BBB; it is a product of permeability and capillary surface
area
Ktrans Measures rate of flux of contrast agent from the intravascular • Usually measured using DCE MR imaging
compartment into the extracellular extravascular space techniques
and is dependent on both vascular permeability and • Commonly used units: minute⫺1
capillary surface area
• Because of dependency on both vascular
permeability and capillary surface area,
in states of very high permeability (eg,
very leaky tumors), Ktrans is only limited
by blood flow and hence predominantly
measures blood flow; whereas in states
of very low permeability, contrast agent
cannot leak out, so Ktrans measures
permeability
kb or kep Measures backflow of contrast agent from the extracellular • Usually measured using DCE MR imaging
extravascular compartment into the intravascular techniques
compartment • Commonly used units: minute⫺1
• Its measurement is based on a number of
factors including the volume of Ve, as
well as the interstitial pressure
trans
K
Ve ⫽
Kb
Note:—Ve indicates extravascular extracellular compartment.
limited to grading of tumors, because increased permeabil- which contrast leaks out of the vasculature via the following
ity is associated with immature blood vessels, which are relationship:
seen with angiogenesis, and with assessing response to ther- PS
⫺F,
apy because changes in permeability may provide a means E⫽1⫺e
of assessing treatment response, especially when antiangio-
genic therapy has been used.7,8 Measuring tumor vessel where PS is the permeability surface-area product and F is
permeability could also help in better understanding the flow. The PS product has the same dimensions as flow (mil-
mechanism of entry of therapeutic agents into the central Ⲑ
liliters/100 gram/minute), and thus the ratio PS F is dimen-
sionless. In physiologic terms, PS is the rate at which con-
nervous system, which could be very important for devel-
opment of methods to selectively alter the BBB to enhance trast agent flows into the extravascular tissues; it is related
drug delivery.9 to another commonly stated parameter of vascular leakage,
Ktrans, by the following equation:
Tumor Vascular Permeability: What are we measuring? trans⫽ExF,
K
Permeability is related to the diffusion coefficient of con-
trast agents in the assumed water-filled pores of the capil- where Ktrans is the forward transfer constant with, again,
lary endothelium. The diffusion flux of contrast agent the same dimensions as flow (unit commonly used is min-
across the capillary endothelium is dependent on both the
diffusion coefficient and the total surface area of the pores.
ute⫺1) (Table 1). It is easily demonstrated that if PS F ⬍⬍ 1
(or F ⬎⬎ PS, which is usually the case in high-grade leaky
Ⲑ
PS characterizes the diffusion of some of the contrast agent brain tumors because blood flow is usually very fast), then
from the blood vessels into the interstitial space due to Ktrans ⬵ PS. In normal cerebral vasculature, PS is negligible
deficient or leaky BBB and is used as a means of quantifying for all contrast agents presently in clinical use.
the “leakiness” of the regional vasculature (Table 1). PS is However, in brain tumors, estimates of Ktrans are com-
computed from the IRF. Contrast agent diffusion appears in plicated by the fact that there is backflow of the contrast
the IRF as a residual enhancement that occurs after the agent measured by kep or kb from the extravascular com-
initial impulse response and that decreases exponentially partment into the intravascular compartment, which is
with time. The IRF is used to estimate the first-pass fraction based on a number of factors including the volume of ex-
of contrast agent that remains in the tissue, the extraction travascular extracellular compartment and the interstitial
fraction, E.10 The extraction fraction is related to the rate at pressure (Fig 3 and Table 1).
Fig 4. Heterogeneity of tumor angiogenesis is manifest by differences in perfusion parameters in various regions. Different regions of interest show
high CBV and high PS, high CBV and low PS, and high PS and low CBV, suggesting that CBV and PS represent different aspects of tumor vasculature and
angiogenesis.
Understanding Angiogenesis and Its Correlation with T1-weighted acquisition, which is based on the fact that
Perfusion Parameters increase in the rate of T1 relaxation is proportional to the
As already mentioned, angiogenesis is an essential aspect of concentration of the contrast agent from which a time-
tumor growth because it provides tumors a means of main- concentration curve can be generated and is tracked during
taining blood supply. In the initial phase of angiogenesis, a longer time period (5–10 minutes); and 2) first-pass T2*-
vessel leakiness, which is measured by permeability (PS or based acquisition, which uses susceptibility-weighted imag-
Ktrans), increases more than the total number and area of ing to generate a time-concentration curve during the initial
blood vessels, which are measured by blood volume. How- 45– 60 seconds of circulation. Both DCE MR imaging tech-
ever, as the vessels mature, the total number and area of niques have advantages and disadvantages; however, the
blood vessels increase more than vessel leakiness, evolving major limitation is a nonlinear relationship of contrast
into a very heterogeneous tumor with regions showing dif- agent concentration with tissue signal intensity.
ferent mixtures of vessel characteristics and angiogenesis. Dual-echo gradient echo perfusion-weighted imaging18
This heterogeneity is manifest by differences in tumor blood is based on a simple 2-compartment kinetic model19 and
volume and permeability, which do not necessarily increase has been used to measure vascular permeability and to cor-
or decrease in tandem and probably represent 2 different rectly estimate blood volume in lesions/tumors with defi-
aspects of tumor vasculature (Fig 4). cient or absent BBB. Several studies have found a correla-
tion between increased vascular permeability and higher
Tumor Vascular Permeability: Dynamic Contrast- tumor grade.14,20-23 However, MR perfusion techniques
Enhanced Imaging Techniques have certain limitations because of the nonlinear relation-
While postgadolinium T1-weighted MR imaging gives a ship of the signal intensity with the contrast agent, both for
rough estimate of the disruption of BBB by providing a dynamic contrast-enhanced imaging with T1-weight-
snapshot in time and has been used in the past for quanti- ing24-27 and for dynamic susceptibility contrast imaging
tative estimation of permeability, dynamic imaging acqui- with T2- or T2*-weighting. In the latter case, if the contrast
sition provides a better estimate of vascular permeability.11 agent remains intravascular, the method is widely accepted
Various noninvasive dynamic contrast-enhanced imaging as a relative estimate of CBF and CBV, though there is a
techniques including MR imaging12-14 and CT perfu- possibility for artifacts because of difficulties in assessing
sion15,16,17 have been used for in vivo estimation of tumor the shape and timing of the arterial input function.28-34
vascular leakiness in brain tumors (Table 2). DCE MR im- However, in cases with leaky BBB, there is substantial leak-
aging data acquisition primarily involves 2 techniques; 1) age of contrast agent from the intravascular to extravascu-
Advantages • CBV and PS obtained • Routinely available • Purported to be the most reliable
in same exam
• Availability of AIF • Quick and easy acquisition • Fewer artifacts
• Robust data
• Absolute values
Limitations • Radiation exposure • Susceptibility artifacts • Flawed AIF
• Iodinated contrast • Non-availability of AIF • Difficult/complex modeling required
to derive indices
• Separate exam • Only rCBV estimates; permeability • Need to calculate baseline T1
estimates are questionable
• Limited coverage • Magnetic field inhomogeneities
Note:—AIF indicates arterial input function.
lar space and a strong and competing T1 contrast effect is hypothesized that low-molecular-weight contrast agents
often noticed in areas of pathology. leak relatively easily from the blood into the interstitium.
To minimize the competing T1 contrast, preloading with This might also lead to overestimation of the permeability,
contrast agent has been proposed,19 with some success. which will be influenced by and will approximate tumor
However, this approach does not allow an estimate of blood flow.40
Ktrans. An alternative approach has also been proposed to On the other hand, high-molecular-weight contrast
decrease the T1 effect,35 which uses a slower TR, lengthen- agents leak from the vessels and move through the intersti-
ing the TR of the experiment and undermining the estima- tium with relative difficulty and are hence flow-indepen-
tion of CBF, thus yielding only estimates of CBV and Ktrans. dent. Therefore, macromolecular permeability and vascular
A further refinement, allowing the estimate of blood volume volumes may be best measured by high-molecular-weight
and producing an index of transfer constant, has been sug- contrast agents.40 Even though CT and MR imaging con-
gested,19 and a dual-echo gradient echo sequence18 also trast agents do not differ much in their molecular weight,
shows some potential for an index of blood volume and different charges of nonionic CT contrast compared with
transfer constant. Despite the partial success of these rapid ionic MR imaging contrast may also be responsible for the
imaging studies, in contrast to PCT, there does not appear differences in permeability measured with these 2 modali-
to be an MR imaging technique that will reliably quantify ties. Blood pool contrast agents, particularly albumin-bind-
CBF, CBV, and PS/Ktrans in one experiment.17 PCT36,17 has ing agents (such as gadofosveset, ABLAVAR; Lantheus
been shown to be very robust due to the linear relationship Medical Imaging, North Billerica, Massachusetts; which
of contrast agent concentration with tissue attenuation. was recently FDA-approved for MR angiography of periph-
However, its clinical impact is limited due to the fact that it eral vascular disease) may solve some of the issues associ-
requires an additional examination by using iodinated con- ated with easy leakage of extracellular contrast agents into
trast and the risk of radiation exposure. On the other hand, the interstitium.
MR imaging has been the standard of care, and adding a Another controversial aspect of measuring permeability
DCE-MR imaging acquisition does not add significant with various perfusion imaging techniques has been the
burden. scanning time. It is presumed that delayed permeability due
to slow leakage of contrast from a leaky blood vessel may
Tumor Vascular Permeability Imaging: Limitations and not be accurately measured with the first pass of the con-
Controversies trast agent by using a 45- or 60-second scanning time20,41
Low spatial resolution has been one of the major limitations and can be measured only with longer acquisition times.16
of the available clinical imaging tools, which limits a de- However, there is no consensus or criterion standard for the
tailed assessment of the complex and heterogeneous vascu- optimal acquisition time. Gliomas, particularly high-grade
lar microenvironment.37,38 Another major limitation has gliomas, can have extremely variable and heterogeneous
been the use of the currently FDA approved low-molecular- blood flow due to the complex tumor vasculature, which
weight CT and MR imaging contrast agents (⬃0.5– 0.9 can influence permeability.38,39,42 Various other intratu-
kDa) for human subjects, which might limit the accurate moral factors that can also influence permeability include
differentiation of vascular permeability and blood vol- luminal surface area and interstitial, hydrostatic, and os-
ume.39 Permeability measurements depend on the leakage motic pressure across the endothelium. Slow blood flow
of particles from the blood to the interstitial space; it is and/or low osmotic gradients, which can occur in high-grade
tumors with a lot of vasogenic edema and also in the central derived from DCE-MR imaging have been successfully used
necrotic or hypoperfused parts of large tumors, can lead to a in the past in the evaluation of prostate, breast, cervical, and
larger component of delayed permeability, which may need pancreatic cancers.11,41-48 These parameters include onset
longer acquisition times for accurate estimation.17 time of contrast agent arrival, initial and mean gradient of
Another major limitation of permeability imaging is the upslope of the enhancement curve, maximum signal
complex multicompartment modeling needed for postpro- intensity, and washout gradient.46 Narang et al47 used
cessing calculations. Quantification of vessel permeability MSIVP, normalized MSIVP, normalized slope of the de-
with various perfusion techniques requires a 2 or more com- layed equilibrium phase, and nIAUC at 60 and 120 seconds
partment pharmacokinetic model with an arterial input (nIAUC60 and nIAUC120) to differentiate recurrent tumors
function, making these studies more complex than CBV from treatment-induced necrosis (Fig 5). The major disad-
estimation,11 and quantification is dependent on the imag- vantage with these is that these parameters lack a physio-
ing technique and the mathematic model43 that is used. logic basis limiting their growth as absolute quantitative
Steroid use is another confounder that can affect blood measures of tumor vascular leakiness. Despite that, these
volume and tumor-permeability measurements, affecting per- model-free metrics could have a more practical role to play
meability measurements more than blood volume,44,45 and, in the routine clinical setting because they do not depend on
hence, should be known before comparing results across dif- the technical expertise needed for the more complex model-
ferent patients and different time points in the same patient. based analysis. However, their direct comparison across
multiple centers or even individual examinations could be
Going Backwards with Tumor Vascular Permeability: limited due to the fact that these do not accurately reflect
Semiquantitative Indices contrast agent concentration in the tissue and hence can be
Accurate and absolute quantitative estimates of permeabil- influenced by scanner type and settings.
ity may not be easily possible due to complex multicompart- Previous authors have also successfully used similar in-
ment physiologic models needed to derive these metrics by dices such as relative PSR or signal-intensity enhancement-
using DCE MR imaging techniques as described above. On time curves as an indirect measure of vascular leaki-
the contrary, various model-free “semiquantitative” indices ness.48,49 Barajas et al48 showed lower relative PSR in
recurrent GBM compared with radiation necrosis by using these imaging features. Recent literature has tried to corre-
DSC MR perfusion imaging, suggesting a disrupted BBB late morphologic imaging features with gene expression in
that was more permeable to macromolecular contrast GBMs54-57; however, there has been little emphasis on cor-
agents; however, their measurements were not a direct esti- relating metabolic or physiologic imaging biomarkers with
mate of lesion leakiness. They also noted a large degree of gene expression. Perfusion parameters such as CBV and PS
overlap between the 2 groups, making rPSR a less robust estimates in GBMs have been shown to correlate positively
predictor of recurrent tumor. The same group also pub- with proangiogenic genes and inversely with antiangiogenic
lished the same methodology with stronger results by using genes. This correlation can help establish a genomic/molec-
rPSR to differentiate metastatic tumors from radiation ne- ular basis for these commonly used imaging biomarkers58
crosis, suggesting that rPSR may be a better prognostic and potentially add to our existing knowledge of their im-
indicator of tumor recurrence than rCBV.50 munohistologic bases.36,52
Most perfusion imaging studies have focused on blood vol- been used to evaluate post-treatment recurrent enhancing
ume estimates and correlation with survival prediction in lesions, most of these perfusion imaging techniques have
mixed populations of gliomas,62,63 whereas only a few used only blood volume estimates as a tool to differentiate
studies have explored tumor leakiness or permeability esti- recurrent high-grade tumor from radiation necrosis.
mates as prognostic factors.64-66 Mills et al65 used both However, recent studies have demonstrated the role of
CBV and Ktrans in 1 single experiment, though they found quantitative estimates of permeability in differentiating re-
the relationship of Ktrans with survival to be unexpected and current tumors from treatment-induced necrosis. In partic-
counterintuitive in high-grade gliomas because patients ular, a recent study by Jain et al72 demonstrated lower PS
with higher Ktrans showed better survival. Dhermain et al66 values in radiation necrosis compared with recurrent/pro-
recently showed that low-grade gliomas with microvascular gressive tumors (Fig 7). This is presumably due to leaky
leakage and contrast enhancement show poor progression- angiogenesis seen with high-grade recurrent tumors,
free survival compared with those without it. In a recently whereas vasculature associated with tumor-induced necro-
presented study67 and our unpublished data, we have eval- sis is not that leaky, despite having damaged endothelium
uated both rCBV and leakiness (PS) estimates as prognostic and hypoxia-related VEGF upregulation.72 Pseudoprogres-
factors in high-grade gliomas and found that both of these sion, on the other hand, has not yet been studied well with
parameters, which can be estimated with 1 single experi- permeability imaging, though some recent studies do show
ment by using PCT, correlated significantly with overall the utility of measuring rCBV to differentiate it from tumor
patient survival. However, after adjusting for World Health progression.73
Organization grade, PS estimates were not statistically sig-
nificant. However, within the grade III glioma group, pa- Assessing Treatment Response and Treatment Planning.
tients with grade III gliomas and lower PS (⬍1.7) had a
significantly better overall survival (P ⫽ .011) compared
Most of the imaging follow-up for brain neoplasms is based
with grade III gliomas with higher PS estimates, suggesting
on measurement of contrast-enhancing tumor. However,
that measuring tumor leakiness could provide additional
the increasing use of antiangiogenic agents and multitech-
prognostic information in high-grade gliomas.
nique treatment regimens for treatment of recurrent high-
grade neoplasms has raised doubts about just using con-
Differentiating Treatment Effects from Recurrent Tumor.
trast-enhancing tumor as a marker of treatment response.
A recurrent or progressive enhancing lesion in a treated While antiangiogenic agents inhibit angiogenesis and seem
brain tumor is fairly common in a busy neuro-oncologic to control tumor enhancement initially, infiltrative nonen-
practice, and differentiating treatment effects, whether due hancing tumor may continue to grow. Thus, measuring
to pseudoprogression (occurring within 12 weeks of only the enhancing lesion might overestimate the response.
chemoradiation therapy)68 or delayed radiation necrosis As such, other imaging criteria and methods have become
(usually occurring 3– 6 months post-treatment), from recur- more important in accurately assessing treatment response.
rent tumor based on conventional/morphologic imaging is Permeability measurements allow more accurate assess-
difficult. While both MR imaging49,69,70 and CTP71 have ment of treatment response following antiangiogenic treat-
CONCLUSIONS
Despite the limitations and lack of standardization of the
Fig 8. Treatment-response assessment. The above example shows a
clinically available imaging techniques to measure tumor
grade III glioma (A), before and after surgical resection. B, Follow-up vascular permeability, additional information obtained
studies after radiation therapy demonstrate a progressively increasing from permeability imaging may be used in a number of
lesion on postcontrast T1-weighted images. C, MR perfusion shows pro- important ways, including glioma grading and patient
gressively increasing Ktrans on DCE T1 maps, suggesting a recurrent pro- prognosis, as well as treatment guidance and assessing
gressive lesion, which was proved to be a recurrent tumor.
treatment response. With improvements of image acquisi-
tion and postprocessing software in the future, permeability
Fig 9. Tumefactive demyelinating lesion: a 55-year-old woman presenting with a single central necrotic enhancing lesion in the brain stem, mimicking
a high-grade neoplasm. PCT maps showed low PS in addition to low CBV, suggesting a non-neoplastic lesion, which was proved to be a tumefactive
demyelinating lesion on follow-up.