Multiparametric Magnetic Resonance Imaging and Repeated Measurements of Blood-Brain Barrier Permeability to Contrast Agents

Summary

Breakdown of the blood-brain barrier (BBB) is present in several neurological disorders such as stroke, brain tumors, and multiple sclerosis. Non-invasive evaluation of BBB breakdown is important for monitoring disease progression and evaluating therapeutic efficacy in such disorders. One of the few techniques available for non-invasively and repeatedly localizing and quantifying BBB damage is magnetic resonance imaging (MRI). This usually involves the intravenous administration of a gadolinium-containing MR contrast agent such as Gd-DTPA, followed by dynamic contrast-enhanced MR imaging (DCE-MRI) of brain and blood, and analysis of the resultant data to derive indices of blood-to-brain transfer. There are two advantages to this approach. First, measurements can be made repeatedly in the same animal; for instance, they can be made before drug treatment and then again after treatment to assess efficacy. Secondly, MRI studies can be multiparametric. That is, MRI can be used to assess not only a blood-to-brain transfer or influx rate constant (K_i or K₁) by DCE-MRI but also complementary parameters such as: 1) cerebral blood flow (CBF), done in our hands by arterial spin-tagging (AST) methods; 2) magnetization transfer (MT) parameters, most notably T_1sat, which appear to reflect brain water-protein interactions plus BBB and tissue dysfunction; 3) the apparent diffusion coefficient of water (ADC_w) and/or diffusion tensor, which is a function of the size and tortuosity of the extracellular space; and 4) the transverse relaxation time by T₂-weighted imaging, which demarcates areas of tissue abnormality in many cases. The accuracy and reliability of two of these multiparametric MRI measures, CBF by AST and DCE-MRI determined influx of Gd-DTPA, have been established by nearly congruent quantitative autoradiographic (QAR) studies with appropriate radiotracers. In addition, some of their linkages to local pathology have been shown via corresponding light microscopy and fluorescence imaging. This chapter describes: 1) multiparametric MRI techniques with emphasis on DCE-MRI and AST-MRI; 2) the measurement of the blood-to-brain influx constant and CBF; and 3) the role of each in determining BBB permeability.

1. Introduction

The blood-brain barrier (BBB) is a structural and functional feature of the cerebral microvascular system that tightly regulates the flux of substrates into and out of the brain (1). The “barrier” in this system is mainly formed by the continuous series of tight junctions that join the endothelial cells of the brain microvessels and close the paracellular pathway. The lack of fenestrae and reduced numbers of caveolae within the endothelial cells also contribute to barrier function. Other constituents of the BBB complex are the basal lamina (BL), the pericytes that are embedded in the BL, and a surrounding cuff of astrocytic end-feet (1).

Spontaneous, mechanical or drug-induced thrombolysis after cerebral ischemia can lead to either partial or complete restitution of cerebral blood flow (CBF) resulting in reperfusion of the ischemic regions. However, if not begun within a very short time window after stroke onset, reperfusion can induce opening of the BBB and a different, more damaging lesion than that caused by permanent occlusion (2). This acute BBB damage can worsen over time, leading to hemorrhagic transformation (HT) which can be fatal (3). Furthermore, the risk of HT can increase as much as 10-fold with tPA treatment (4). Presently, apart from stroke onset time, no other reliable indicator(s) of the risk of HT are available. The heightened threat of HT associated with thrombolysis has, thus, restricted the widespread application of this vital therapy.

Some non-invasive techniques are available for characterizing acute stroke in humans. Traditionally, computed tomography (CT) has been the imaging modality of choice. More recently, multiparametric MRI, which combines two or more MRI techniques such as diffusion-weighted imaging (DWI) and perfusion-imaging (PWI), has been used to identify the penumbra region of the ischemic lesion. This region is characterized by a so-called perfusion-diffusion mismatch and is thought to be made up of tissue that is injured, but not irreversibly, and can be saved by prompt reperfusion (5). This approach has been employed to select patients for tPA therapy within and beyond the ‘3-hour-after-stroke-onset’ time window (6,7). Adding to the repertoire of multiparametric MRI, MR contrast agent (MRCA)-enhanced imaging has also been shown to delineate ischemic regions with acute BBB alteration (8,9). As alluded to above, detection of acute BBB opening in stroke is important since these regions have been shown to develop HT later either with or without tPA in both experimental models (3,10) and humans (11).

Despite the capability of MRI to detect areas of BBB opening, precise localization and quantification of such damage in stroke by dynamic contrast agent-enhanced MRI (DCE-MRI) was still lacking until several years ago. Originally, quantification of DCEMRI data was achieved by deconvolution to determine a transfer constant, K^trans, that was highly model-dependent (12). We have now developed a novel method for localizing and quantifying DCE-MRI data using Patlak plots to obtain the blood-to-brain transfer or influx rate constant (K_i) of Gadolinium- diethylenetriaminepentaacetic acid (Gd-DTPA) and tested the spatial and quantitative accuracy of these estimates with radiotracers and QAR (13). This 25-year-old Patlak plot technique, recently introduced by us to DCEMRI, has now been replicated successfully by other labs using various magnet and console combinations in both experimental models and human disease (see Note 1).

Patlak and co-workers developed a treatment of tracer uptake over a period of circulation by taking into account tissue and blood concentrations of the tracer at a number of points (minimally more than four) after injection (14,15). This multiple-time graphic method yields a curve of tissue uptake as a function of blood tracer level and time. The slope of the linear part of this curve represents K_i, which is the most useful and accurate of the several blood-brain transfer constants that can be linked to capillary permeability. These data and the Patlak plot approach for analyzing MRI findings were confirmed with the images and K_i results obtained with ¹⁴C-sucrose-QAR some minutes later in the same animals (13). This work was further extended to establish the spatial resolving power of DCE-MRI after a bolus MRCA injection by comparing these results to the QAR images of ¹⁴C-α-aminoisobutyric acid (AIB) distribution gained concurrently in the same rats (16).

Despite employing a very precise MRCA infusion technique, it is possible not to see contrast-enhancement in spite of BBB damage in stroke since adequate blood flow is required for the delivery of the MRCA to the tissue of interest. In acute stroke, CBF is the first function to be affected and low flow may result in little or no delivery of MRCA to the tissue. For this reason, an MRI technique for estimating the leakiness of the BBB that did not involve an MRCA was sought. In one series of experiments, the ability of magnetization transfer (MT)-MRI methods was tested against classical Gd-DTPA enhanced MRI to look for alternate quantitative MRI signatures that indicated BBB opening (17). The MT-based indices tested were: T₁ and T_1sat. The regions with increased T₁ and T_1sat changes accurately reflected BBB changes as confirmed by DCE-MRI, QAR and dual-contrast enhanced MRI (18). The MT-MRI could also be used to segment and measure tissue damage (17).

Together these techniques represent a powerful array of minimally invasive and non-invasive quantitative MRI methods for accurate localization and quantification of tissue and BBB damage in stroke (Fig. 1) and brain tumors. The multiparametric MRI approach can also be applied to study other pathologies such as multiple sclerosis, Parkinson’s disease and neurotrauma. The greatest advantage is that it can be repeatedly applied in the same subject within several hours on the same day or employed as follow up across different days. These measures are expected to aid in cerebrovascular disease characterization and its response to treatments.

Figure 1.

A representative data set showing multiparametric MRI-derived maps obtained before (I1t) and after reperfusion (R1t) from an animal subjected to 3 h of transient focal ischemia. Top (left to right): cerebral blood flow (CBF), apparent diffusion coefficient (ADCw) and transverse relaxation time (T₂) maps. Middle (left to right): spin–lattice relaxation times in the absence and presence of off-resonance saturation (T₁ and T_1sat, respectively), the forward rate of magnetization transfer (K_for), and magnetization transfer ratio (MTR) maps. Bottom row: A corresponding ISODATA segmentation theme map to demonstrate the ROIs that were selected to represent ischemic tissue with (red and green) and without (yellow) BBB disruption (left). To the right, are a corresponding ¹⁴C-AIB autoradiographic image and a cresyl violet stained histologic section that were used to confirm acute BBB damage and the region of ischemic damage, respectively. (From Knight et al. Magn Reson Med. 2005;54:822-832)

2. Materials

2.1. Middle Cerebral Artery Occlusion (MCAO) Model

1. Young adult Wistar rats, weighing 300 g.

2. Sterile surgical instruments, including microvascular clips (Codman, Raynham, MA), skin clamps.

3. Sterile 4–0 nylon suture, about 2 cm in length with its tip rounded using a heat source.

4. Sterile silk 4–0 sutures with needles.

5. Sterile 1 ml syringes (Becton Dickinson, Rutherford, NJ, USA).

6. Sterile 23 G needles (Becton Dickinson, Franklin Lakes, NJ, USA).

7. PE–50 catheters with beveled, rounded tips, about 100 cm long for venous and arterial lines. The other end of the catheter is connected to a blunted 23 gauge needle attached to a 1 ml syringe (Becton Dickinson, Sparks, MD, USA).

8. Betadine solution and alcohol swabs.

9. A small, animal hair clipper.

10. Sterile saline (500 ml) bags, each containing 50 IU heparin.

11. Anesthesia apparatus and small-animal nose cones for isoflurane anesthesia (Baxter Healthcare Corpn., Deerfield, IL) and a lidocaine (Abbott Labs, N. Chicago, IL) for irrigation.

12. A water-recirculated heating pad (Kent Scientific Corp., Torrington, CT, USA).

13. Sterile drapes, masks, gloves, gauge.

14. Operating microscope (Carl Zeiss, Germany).

15. A manual or programmable syringe pump (Model 944, Harvard Apparatus, South Natick, MA).

2.2. 9L Cerebral Tumor Implantation Model

1. Male Fischer-344 rats, weighing between 200-240 g (Charles River Breeding Laboratories, Wilmington, MA).

2. Stereotaxic equipment with two positioning holders for a dental drill and a Hamilton syringe each (David Kopf Instruments, Tujunga, CA).

3. Sterile -surgical instruments including scalpels, silk sutures, bone wax.

4. A sterilized 10 μL Hamilton syringe with a beveled tip.

5. A foot-operated dental drill (Foredom Series F, Bethel, CT).

6. Injectable anesthetics: Cocktail containing Ketamine (Abbott Laboratories, N. Chicago, IL; 80 mg/kg) and Xylazine (Phoenix Scientific Inc., St. Joseph, MO; 10 mg/kg).

7. A tube of ophthalmic ointment (AKWA Tears, Akorn Inc., Buffalo Grove, IL).

8. Sterile drapes, masks, gloves, gauge.

9. A source of immortalized glioblastoma cells (Dept. of Radiation Oncology, Henry Ford Hospital, Detroit, MI).

10. Eagle’s minimum essential medium (Baxter Healthcare, Deerfield, IL), supplemented with 10% fetal calf serum, 1% multivitamins and 1% nonessential aminoacids.

11. Trypsin (GIBCO, Grand Island, NY, USA; 0.05% stock solution diluted 1:1 with PBS).

2.3. Magnetic Resonance Imaging

1. A dedicated small animal magnet (Magnex Scientific, Inc.; Abingdon, UK) (see Note 2).

2. Magnet-compatible imaging sequences and consoles (Bruker Biospin MRI, Billerica, MA, USA).

3. Offline, post-processing hardware (SUN Microsystems, Inc., Santa Clara, CA) and software (Eigentool, Henry Ford Health System, Detroit, MI; C-program based Unix Shell scripts).

4. A commercially available or lab-made Gadolinium-based MR contrast agent such as Magnevist (Berlex, Montville, NJ, USA; 0.1 mmol/Kg).

3. Methods

Animal studies were performed in accordance with National Institutes of Health guidelines for animal research and were approved by the Institutional Animal Care and Use Committee.

3.1. Experimental Models

3.1.1. Middle Cerebral Artery Occlusion (MCAO) Model

In order to develop and test BBB permeability measurement techniques, it was necessary to utilize an animal model capable of producing a focal region of brain tissue with BBB opening. For this purpose we used a model of transient focal cerebral ischemia in rats via intraluminal occlusion of the MCA with a nylon filament (3,10).

1. Rats are anesthetized and placed supinely on the water-recirculated heating pad and the core temperature is kept at 37±1 °C throughout all surgical and MRI procedures using a feed-back regulated rectal probe.

2. The neck area is cleaned with alcohol swabs and betadine solution and shaved to expose skin.

3. A 2 cm longitudinal incision is made at the midline of the ventral aspect of the neck and the right common carotid artery (CCA), external carotid artery (ECA), and internal carotid artery (ICA) are exposed under the operating microscope. Care should be taken to avoid injuring the vagus nerve.

4. The ICA is further dissected to identify the pterygopalatine branch and intracranial ICA and a 5–0 silk suture is loosely tied at the origin of the ECA and the distal end of the ECA is ligated.

5. The CCA and ICA are temporarily clamped with microvasculature clips.

6. The 2 cm long, 4–0 surgical nylon filament, with a rounded tip round is introduced into the ECA lumen through a small puncture and the suture at the origin of the ECA tightened around the intraluminal nylon filament to prevent bleeding and the microvascular clips are removed.

7. The nylon filament is gently advanced from the ECA into the lumen of the ICA, a distance of 18.5 – 19.5 mm according to animal weight, until the tip of the filament blocks the origin of the middle cerebral artery (MCA) and the microvascular clip is released.

8. The skin incision is temporarily closed with either loose silk sutures or skin clamps. After MR quantification of occlusion effects, the rat is taken out of the magnet and reperfusion is performed at the desired time as indicted in step 9.

9. The skin sutures/clamps are opened and the occluding filament is withdrawn until its tip is visible in the lumen of the ECA and is no longer occluding the MCA or restricting flow in the ICA and the neck incision is closed.

3.1.2. 9L Gliosarcoma Brain Tumor Model

In brain tumors, BBB opening can be due to both BBB damage and increased angiogenesis which is typically seen at the periphery of solid experimental tumors. Owing to the tremendous leakiness of the vessels, it is also necessary to correct the Patlak model for MRCA backflux. The optimal experimental conditions that accurately reflect native leakage and indices of response to putative treatments were developed (19,20) using the 9L gliosarcoma model (21).

1. The tumor cells are maintained as exponential cultures in Eagle’s minimum essential medium ( see Subheading 2.2, #11) (21).

2. Immediately prior to each implantation, cells are trypsinized, resuspended and a final dilution of 2×10⁶ cells per mL is made in Eagle’s medium without serum.

3. The rat is anesthetized with an intraperitoneal injection of a cocktail of ketamine and xylazine and positioned in the stereotaxic head frame.

4. The scalp is cleaned with alcohol swabs and betadine and hair is shaved from the frontal surface of the scalp with electric clippers and ophthalmic ointment is applied to the rat’s eyes to keep the eyes moist.

5. An incision is made along the midline on the dorsal surface of the head to expose the frontal and temporal bones.

6. The position of the injection site is marked using one positioning arm of the dental drill: 2.5 mm anterior to the bregma and 2.0 mm to the right of the midline.

7. A 1.0-mm burr-hole for implanting tumor cells is drilled at this point through the skull without breaking the dura.

8. The Hamilton syringe fitted with a 26-gauge needle is filled with 5 μL of the cell suspension and fixed to the other stereotaxic positioning arm. Its tip is positioned over the burr-hole. The arm is then lowered to a depth of 3 mm and raised by 0.5 mm. The 5 μL volume is slowly injected to implant the tumor cells.

9. The syringe is left in place for 5 min and then gently retracted. The burr-hole is sealed with sterile bone wax and the skin sutured with sterile silk sutures.

10. Tumors are allowed to grow for up to 14-15 days post-implantation before MRI.

3.1.3. Femoral Arterial and Venous Cannulation

1. Anesthetize the rat, place it in a supine position on the heated rubber mat. Extend its legs and fix them to the operating base with adhesive tape.

2. Clean the skin on either left or right groin with alcohol swabs and shave hair in this region and apply betadine to the skin surface.

3. Make approximately 1 cm long, oblique (along the direction of extended leg) skin incision in the groin.

4. Expose the femoral artery and vein, separate them and clean the fascia.

5. For each vessel, a small length (~10 cm) 4–0 silk suture is put loosely around the proximal part and the ends of the suture are weighed down by a hemostat to pull the vessel taut. The distal end (approximately 1 cm apart) is ligated using a 4–0 silk suture with enough length for two more knots.

6. An opening (approximately one-half the vessel diameter) is made in the middle of this 1 cm segment of the vessel.

7. The beveled end of a PE-50 catheter is introduced into this opening, threaded about 1.5 cm proximally without kinking the vessel and the proximal suture is used to ligate the vessel over the catheter. Care must be exercised not to ligate beyond the catheter tip to avoid occluding the vessel.

8. The remaining part of the distal suture is used to tie the catheter to the vessel to prevent it from slipping out.

9. The catheters are tested for smooth blood flow. Blood samples can be now collected or the vessels can be used for physiological monitoring, drug injections etc.

3.2. Basic MRI Experimental Procedures

1. The rat is placed in a plastic holder following the surgical procedures. It can be either in a supine or prone position depending on the design of the MR coil. This holder should have provisions for supporting the animal’s body and for administering anesthetic gases if they are being used.

2. A feedback regulated water blanket and non-magnetic stereotaxic ear bars minimize head movement during imaging.

3. The animal holder is placed inside a 5 cm diameter quadrature driven transmit/receive birdcage coil tuned to the resonant proton frequency of ~300 MHz and the entire assembly is placed inside the MRI unit (see Note 2).

4. Scout images are obtained to adjust the position of the animal’s head until the central image slice is located at bregma.

5. Receiver and Transmitter coils are tuned to the frequency of the sample using the ‘wobble’ mode of the setup.

6. Shimming, and setting up central frequency, RF gain and receiver gain are done either automatically or manually.

3.2.1. Measurement of Cerebral Blood Flow (CBF)

Blood flow, an important function of the cerebral vascular system, can be measured at the local level by MRI. It is important that CBF be measured in studies of BBB permeability because the actual parameter determined in the latter experiments is an influx rate constant, K_i, depending not only on capillary permeability (P) and surface area (S) but also CBF. To get from K_i to PS product, the physiological expression of capillary permeability, the rate of blood flow is required. Hence, CBF is estimated in all of our MRI experiments by PWI using the arterial spin tagging (AST) technique as follows.

1. Cerebral perfusion can be estimated using AST with a variable tip-angle gradient-refocused echo (VTA-GRE) imaging technique (24).

2. In this procedure, the protons within the blood passing through the arteries of the neck are magnetically inverted thereby serving as an autologous label. The decrease in the net magnetization of the perfused tissue, a function of the rate of blood flow, is then regionally detected in the brain.

3. AST uses gradient and frequency offsets to create a plane of inverted spins in the neck of the animal approximately 18 mm below the center of the imaging plane. AST is performed by labeling inflowing arterial protons via a continuous wave (CW) RF adiabatic inversion pulse applied in the presence of a magnetic field gradient.

4. Labeling of inflowing arterial water protons is achieved via an axial gradient of ±0.3 kHz/mm using a CWRF pulse at a power of 0.3 kHz at a frequency offset of ±6 kHz followed by a VTA-GRE sequence with repetition time (TR)/echo time (TE)=11ms/5ms (24).

5. To allow for equilibration of inverted protons in the imaging plane, the inversion pulse is turned on for a period of 4.5 s preceding the production of each VTA-GRE image.

6. A corresponding set of control scans is acquired using gradient and frequency offsets that put the inversion plane 18 mm above the imaging plane (i.e. outside the head).

7. The images are acquired in sets of 4 with the frequency offset and gradient polarities permuted through all 4 combinations to remove any gradient asymmetries in the axial direction and to balance off-resonance MT effects. The images are acquired over a 32 mm field of view (FOV) and reconstructed using a 64×64 image matrix.

3.2.2. Magnetization Transfer (MT) Parameters

This involves the adiabatic transfer of magnetization between the free or mobile pool of protons, which in biological specimens is mostly water, and the bound or immobile pool of protons, mainly those associated with hydration layers of water surrounding large macromolecules and proteins. Of importance, changes in the MT parameters often correlate well with BBB opening, probably as the result of increased tissue water (i.e. vasogenic edema). The MT related parameters are M₀, M_0sat, T₁, T_1sat, K_for and the magnetization transfer ratio (MTR). Calculation of the apparent forward magnetization transfer rate, K_for, requires estimates of T₁ in both the absence and presence of off-resonance irradiation of the “bound” proton fraction (T₁ and T_1sat, respectively).

1. Estimates of T₁ and T_1sat are obtained using the Phase Incremented Progressive Saturation (PIPS) method (TR/TE= 40/7 ms, tip angle ~ 18°) (25).

2. The sequence is run in sets of two to produce estimates of M_0sat, T_1sat, M₀, and T₁. To estimate M_0sat and T_1sat, a CW RF saturation pulse of approximately 0.3 KHz amplitude, off-resonance by 6 KHz, is turned on for 4.5 s before the beginning of each image of the PIPS sequence, and for 30 ms (of the 40 ms TR) between each gradient-echo line in k-space to partially saturate the macromolecular proton pool.

3. The value of K_for is calculated from the equation:

   $${\mathrm{K}}_{\text{for}}=\frac{1}{{\mathrm{T}}_{1\text{sat}}}(1-\frac{{\mathrm{M}}_{0\text{sat}}}{{\mathrm{M}}_0})$$ (1)

   where K_for represents the product of 1/T_1sat and the magnetization transfer ratio (MTR = 1 - M_0sat/M₀), which is often used alone to examine MT effects (If the transfer of magnetization decreases, then T_1sat may have increased or MTR decreased; K_for from Equation 1, thus, provides a useful summary of MT effects). The total imaging time for both variations of the PIPS sequence is approximately 24 min.

3.2.3. Measurement of T₂

The biophysical meaning of T₂, the transverse relaxation time, is unclear, but changes in T₂ within areas of the brain often indicate underlying tissue injury or dysfunction with elevated T₂ values being highly correlated with increased tissue water content.

1. Sets of T₂-weighted image data are obtained using a multi-slice Carr-Purcell-Meiboom-Gill (CPMG) sequence (TR=1200 ms, TE=30, 60, 90 and 120 ms, with interleaved slice acquisition, 13 slices, slice thickness=1 mm). Total imaging time for the T₂WI series is approximately 13 min.

2. T₂ maps (quantitative images obtained by pixel-by-pixel processing of raw imaging data) are produced from a straight line least squares estimate of the slope from a plot of the natural logarithm of the normalized image intensity [ln(S/S₀)] versus TE.

3.2.4. Measurement of the Apparent Diffusion Coefficient of Water (ADC_w)

The apparent diffusion coefficient of water in brain is a function of the size and tortuousity of the extracellular space, the rate of water exchange between intracellular and extracellular fluid (ECF), the volume of intracellular water, and the rate of bulk flow of extracellular and edema fluid, if present.

1. A pulsed gradient spin-echo imaging sequence with progressively incremented diffusion weighting (b-value) is used to measure ADC_w (TR/TE = 1500/40 ms, b-value =0, 400, 800 s/mm², 128×64 image matrix and interleaved slice acquisition). The total scan time for this diffusion-weighted imaging (DWI) is approximately 10 min for the complete data set.

2. Maps of ADC_w are produced from a straight line least squares estimate of the slope from a plot of [ln(S/S₀)] versus gradient b-value.

3.2.5. Look-Locker MRI Quantification of T₁ after Gd-DTPA Injection

Gadolinium-DPTA changes the longitudinal relaxation rate (symbolized by ΔR₁), of the nearby water protons. The MRCA concentration is proportional to ΔR₁ (R₁ is the inverse of the longitudinal relaxation time, T₁) when this parameter is quantitatively determined.

1. Rapid quantitative estimates of brain tissue T₁ relaxation times are necessary to accomplish an estimate of MRCA tissue concentration as it varies with time, and can be obtained using an imaging adaptation of the Look-Locker method (26).

2. The technique employs the T₁ by Multiple Read-Out Pulses (TOMROP) sequence (27) to produce efficient and unbiased pixel-by-pixel estimates of T₁ (28).

3. Baseline high-resolution T₁-weighted spin-echo images are collected prior to Gd-DTPA injection. Then, after obtaining one or two baseline L-L T₁ estimates, Gd-DTPA is injected as a bolus into the femoral vein in <5 s and L-L data sets are acquired sequentially at approximately 2.5 min intervals for up to 30 min (13).

4. Data are obtained for five interleaved 2 mm thick, slices. At the conclusion of the L-L series, a final T₁-weighted multislice SE image set is obtained.

3.3. Estimating Transfer Constants by Patlak plots

1. The L-L T₁ maps over a series of time points are acquired as described above to estimate arterial plasma and tissue concentrations of the MRCA.

2. The plasma and tissue MRCA concentrations are determined by measuring MRCA induced T₁ changes in the superior sagittal sinus (SSS) and brain-tissue, respectively according to the relationships: ${\mathrm{C}}_{\mathrm{pa}}\left({\mathrm{t}}_{\mathrm{n}}\right)\propto \frac{\Delta {\mathrm{R}}_{1\mathrm{a}}\left({\mathrm{t}}_{\mathrm{n}}\right)}{(1-\mathrm{Hct})}$ and C_tis(t_n) ∝ ΔR_1tis(t_n), where ΔR_1tis(t_n) and ΔR_1a(t_n) are differences in R₁ (=1/T₁) measured in the tissue and sagittal sinus, respectively, at time point t = t_n and t = 0 (baseline or pre-CA time point) and Hct is the hematocrit (see Note 3)

3. Patlak plot technique is employed to estimate K_i and V_p+V_o by graphing $\frac{(1-\mathrm{Hct})\Delta {\mathrm{R}}_{1\mathrm{tis}}\left(\mathrm{t}\right)}{\Delta {\mathrm{R}}_{1\mathrm{a}}\left(\mathrm{t}\right)}$ versus $\frac{\underset{0}{\overset{\mathrm{t}}{\int }}\Delta {\mathrm{R}}_{1\mathrm{a}}\left(\mathrm{t}\right)\mathrm{dt}}{\Delta {\mathrm{R}}_{1\mathrm{a}}\left(\mathrm{t}\right)}$ as described below.

   The solution of the equations for the Patlak model of blood-brain exchange is given by the expression (15): $\frac{{\mathrm{C}}_{\mathrm{t}}}{{\mathrm{C}}_{\mathrm{pa}}}{\mathrm{K}}_{\mathrm{i}}\frac{\underset{0}{\overset{\mathrm{t}}{\int }}{\mathrm{C}}_{\mathrm{pa}}\left({\mathrm{t}}^{\prime }\right){\mathrm{dt}}^{prime}}{{\mathrm{C}}_{\mathrm{pa}}\left(\mathrm{t}\right)}+({\mathrm{V}}_{\mathrm{p}}+{\mathrm{V}}_{\mathrm{o}})$.

4. A graph of the ratio of the tissue CA concentration at the times of sampling (C_t) to the arterial plasma concentration at the respective times (C_pa) versus a concentration weighted time parameter referred to as “plot time” (integral of C_pa over time up to the time point of measurement divided by C_pa at that time) can be drawn to estimate K_i and V_p+V_o (Fig. 2). In MRI, linear least squares estimates of K_i and V_p+V_o are performed pixel-by-pixel to construct the corresponding maps (see Note 4).

Figure 2.

Representative Patlak plots of the ratio of brain tissue (C_tis) concentrations of Gd-DTPA to plasma [C_pa=C_a(t)/(1-Hct)] vs. concentration ratio-stretched time (t_stretch) for three brain regions of interest (ROI) from one rat. The Gd-DTPA K_i values were obtained from the slope of the lines. Among the regions with leaky capillaries, the influx appeared to be greater in the preoptic area (PoA) than the striatum. The slopes in these two regions were significantly different from zero and indicate an appreciable, but small, Gd-DTPA influx. The slope of the line was flat (and statistically not different than zero) for the contralateral ROI and indicates a normal K_i. The units on the abscissa are plasma concentration ratio-stretched time, not real time (which was about 24 min in this case). (From Knight et al. Magn Reson Med. 2005;54:813-821)

3.4. MRI Data Analysis and the Selection of Regions of Interest

1. The MRI data are processed offline using a system such as a SUN workstation.

2. A semi-automated segmentation algorithm is employed to minimize user bias in identifying tissue areas of MR abnormality indicative of processes such as BBB opening or hyperemia.

3. The MRI data are then reconstructed and baseline corrected using in-house software before applying the segmentation procedure. Most of the data from our studies were obtained from what is referred to as the central slice, 2.0 mm thick and extending from bregma +0.15 mm to −1.85 mm. Comparable image analyses techniques can be applied to adjacent slices if necessary and the data merged. One of the several available image segmentation protocols can be used to identify the regions of interest (ROI) (see Note 5).

3.5. MRCA Administration by Step-down Infusion (SDI)

To compensate for the exponential decay of blood MRCA level after a bolus injection, an experimental step-down infusion (SDI) technique that maintains the blood tracer level constant for the duration of DCE-MRI can be used (22). The SDI protocol improves the signal-to-noise ratio, enhances the spatial resolution of the images, and yields a more accurate blood concentration time course (the arterial input function or AIF) of the MRCA (23). For the usual bolus injection method, 60 μL of the stock solution is diluted to 0.1 ml with normal saline and injected via the femoral vein in 4-5 s. The SDI, however, is done with a syringe pump and a prescribed infusion schedule (Table 1).

Table 1.

The step-down infusion protocol developed and employed in the authors’ laboratories. (From Nagaraja et al. Magn Reson Imaging. 2007;25:311-318)

Rate	Pump speed (%)	Time		Duration		Volume
(ml/min)				(min)		(ml)
0.68	100	0-30 s	0.5		0.34
0.53	77.7	30-60 s	0.5		0.26
0.39	57.4	1-2 min		1.0		0.39
0.28	41.6	2-3 min		1.0		0.28
0.22	33.3	3-4 min		1.0		0.22
0.19	28.7	4-5 min		1.0		0.19
0.17	25.0	5-7 min		2.0		0.34
0.14	20.4	7-10 min		3.0		0.42
0.12	17.6	10-15 min	5.0		0.60
0.10	14.8	15-20 min	5.0		0.50
Total volume infused =					3.54 ml

1. For the infusate, 240 μL of the stock solution is diluted in 4.0 mL of saline (the osmolality of the Gd-DTPA preparation is very high (>1000 mOsm/kg), and this dilution brings it into the physiological range for infusion).

2. Approximately 3.5-3.6 mL of infusate is used per step-down experiment following the protocol given in Table 1. The rest of L-L T₁WI methods are identical to those explained above.

3.6. Estimating acute treatment efficacy in a brain tumor model

1. At the desired time-point after tumor implantation, the rat is anesthetized and a venous arterial and cannulations are performed as described above.

2. It is then positioned in the MRI holder and placed inside the magnet.

3. The scout images and T₂ images are obtained to locate and measure the tumor.

4. The chosen contrast media (normally, a large MRCA) is then injected and SE- and L-L T₁WI-images are obtained as described.

5. The drug being tested is injected at the appropriate dose after insuring the return to base-line of MRCA signal.

6. After waiting for drug action, one more MRCA injection is given and another set of SE- and L-L T₁WI- images are obtained.

7. The contrast enhancement before and after drug injections are compared quantitatively to assess drug effects (Fig. 3; see Note 6).

Figure 3.

Vascular parameters pre- (left, test) and post-dexamethasone (right, re-test) administration. A: Vascular volume v_D, B: Transfer constant K₁ [min⁻¹], C: Efflux constant k_b [min⁻¹], D: F-test for Model 3 vs Model 2, with a high value resulting in rejection of Model 2. Only those regions in C with high F-test values have valid results for the estimate of k_b. A widespread decrease in K₁ and kb are easily visualized. Less visible is a moderate decrease in v_D. Bright spots in the maps of v_D correspond to vascular pools. Note the decrease in the F-test from pre- to post-dexamethasone studies. (From Ewing et al. J Magn Reson Imaging 2008;27:1430-1438)

3.7. Sizing BBB Opening with Two Different MRCAs

1. The MCAO model with acute reperfusion is employed for this purpose. Therefore, perform the MCA occlusion and reperfusion following the protocols given in Subheading 3.1.1.

2. Measure, following the MRI protocols, the ischemic status of the tissue and the extent of reperfusion.

3. Quantify the spatial extent of the BBB opening by quantitative Patlak-K_i maps in the ischemic hemisphere using Gd-DTPA-enhanced L-L T₁WI. Wait for the Gd-DTPA signal to return to baseline.

4. Then quantify BBB opening and generate an identical data set using a larger MRCA such as Gd-BSA following the same protocols.

5. Compare the spatial enhancement patterns and K_i maps for the two contrast agents to determine the differences, and, thus, relative sizes of BBB opening in the ischemic tissue (Fig. 4; see Note 7).

Figure 4.

A representative collage of Gd-DTPA and Gd-BSA-Evans Blue (EB) enhancements during MRI, corresponding T_1sat maps and fluorescence images from one experiment. A – Gd-DTPA enhancement in a rat 24 hours after 3 hours of MCA occlusion. A large area of brightness is seen in the preoptic area and striatum. The small, bilateral, bright regions below this enhancement are parts of medium eminence and hence, naturally leaky. Contrast enhancement in such regions, ventricles (large arrows), and pial vasculature surrounding the brain were routinely observed. B – Subsequent Gd-BSA-EB enhancement in this rat. It is a much smaller area. Other normally leaky regions are also visible. In C, the T_1sat map for this slice is shown in gray scale. On the calibration bar on the left, hyperintense/bright areas indicate increased values. The ventricles and circumventricular organs appear bright on this map due to the inherent sensitivity of T_1sat to water/proton shifts. The low-magnification, reconstructed fluorescence image shown (D) has extravascular red fluorescence in the same regions of interest as the Gd-BSA-EB–enhancing area in B. The greenish-yellow hue in the rest of the vasculature is due to the combination of red (EB) and green (fluorescein isothiocyanate–dextran) fluorescence. This image was constructed by collecting the coronal brain section as a series of low magnification (X2.5) images and tiling them together. (From Nagaraja et al. Stroke. 2008;39:427-432).