Targeted delivery of antibodies through the blood–brain barrier by MRI-guided focused ultrasound

Targeted delivery of antibodies through the blood–brain barrier by MRI-guided focused ultrasound

BBRC Biochemical and Biophysical Research Communications 340 (2006) 1085–1090 www.elsevier.com/locate/ybbrc Targeted delivery of antibodies through t...

620KB Sizes 0 Downloads 47 Views

BBRC Biochemical and Biophysical Research Communications 340 (2006) 1085–1090 www.elsevier.com/locate/ybbrc

Targeted delivery of antibodies through the blood–brain barrier by MRI-guided focused ultrasound Manabu Kinoshita, Nathan McDannold, Ferenc A. Jolesz, Kullervo Hynynen

*

Department of Radiology, Brigham and Women’s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA 02115, USA Received 22 November 2005 Available online 27 December 2005

Abstract The blood–brain barrier (BBB) is a persistent obstacle for the local delivery of macromolecular therapeutic agents to the central nervous system (CNS). Many drugs that show potential for treating CNS diseases cannot cross the BBB and there is a need for a non-invasive targeted drug delivery method that allows local therapy of the CNS using larger molecules. We developed a non-invasive technique that allows the image-guided delivery of antibody across the BBB into the murine CNS. Here, we demonstrate that subsequent to MRI-targeted focused ultrasound induced disruption of BBB, intravenously administered dopamine D4 receptor-targeting antibody crossed the BBB and recognized its antigens. Using MRI, we were able to monitor the extent of BBB disruption. This novel technology should be useful in delivering macromolecular therapeutic or diagnostic agents to the CNS for the treatment of various CNS disorders.  2005 Elsevier Inc. All rights reserved. Keywords: Dopamine D4 receptor; Non-invasive; Image-guidance; Focused ultrasound; Blood–brain barrier disruption

The central nervous system (CNS) is protected by a barrier system called the blood–brain barrier (BBB) from foreign toxic substances [1,2]. This almost impenetrable barrier, however, prohibits the delivery of many potentially effective diagnostic or therapeutic agents, causing a great disadvantage for the treatment of CNS diseases. The physiological role of the BBB to exclude larger molecules from the CNS should be temporarily turned down to allow the diffuse or local delivery of macromolecular therapeutic or diagnostic agents to the brain. Efforts to penetrate the BBB include direct catheter insertion, delivery by convection [3,4], changing the agent into a lipophilic substance or attaching it to carriers that can cross the BBB [1,5–7], and temporary enhancement of BBB permeability by the injection of hypertonic solutions into the carotid artery [8,9]. However, these techniques entail major disadvantages because they are either invasive or cause diffuse BBB break-

*

Corresponding author. Fax: +1 617 525 7450. E-mail address: [email protected] (K. Hynynen).

0006-291X/$ - see front matter  2005 Elsevier Inc. All rights reserved. doi:10.1016/j.bbrc.2005.12.112

down which preclude the delivery of macromolecular agents into specific sites. On the other hand, current advances in acoustic technology have made ultrasound not only a diagnostic but also a therapeutic tool. By using a focused ultrasound technique, acoustic energy can be concentrated into a focal spot with a diameter of a few millimeter, thus exposing ultrasound energy only to a relatively small targeted tissue volume. The combined use of focused ultrasound and magnetic resonance (MR) imaging facilitates image-guided target planning and real-time temperature mapping during sonications of various tumors as is demonstrated in clinical trials [10,11]. Using gas bubble-based ultrasound contrast agents in conjunction with ultrasound exposure, ultrasound cannot only cause thermal coagulative effects but also various other bio-effects such as temporary changes in cell membrane permeability [12] or disruption of the BBB [13–18]. Here, we present in vivo evidence that our MRI-guided focused ultrasound BBB disruption method makes possible image defined site-specific, local delivery of a macromolecular antibody to the murine CNS with minimal related tissue damage.

1086

M. Kinoshita et al. / Biochemical and Biophysical Research Communications 340 (2006) 1085–1090

Target selection and the extent of the BBB opening can be planned and monitored by the MR images. Currently, we demonstrated that using ultrasound transducers arranged in a phase array configuration in a larger surface area, it is possible to correct ultrasound wave distortions caused by the bone and focus ultrasound with therapeutic intensity through the intact skull [19]. Therefore, in the future it is possible to apply our novel image-guided, non-invasive method clinically to deliver antibodies and/or drugs to the CNS. This new paradigm may open a new research direction and can result in several diagnostic and therapeutic applications in the field of basic and clinical neuroscience. Methods Antibody. The antibody used in this experiment is a rabbit anti-human dopamine D4 receptor antibody, which recognizes the 3rd extracellular domain (Ac-176-185) of the human dopamine D4 receptor (Cat# 324405, Merck KGaA, Darmstadt, Germany) [20]. Cross-reactivity of this antibody for mouse brain was confirmed by standard immunohistochemistry with a dilution of 1:10,000. Ultrasound equipment. The ultrasound fields were generated by an in-house manufactured, focused, piezoelectric transducer with a 100mm diameter, an 80-mm radius of curvature, and a resonant frequency of 0.69 MHz. The half-maximum pressure amplitude diameter and length of the produced focal spot were 2.3 and 14 mm, respectively. The transducer driving equipment was similar to those previously reported [21]. Animal preparation. All of the procedures used in the animal experiments were approved by The Institutional Animal Committee. 15 SwissWebster mice (10 weeks), weighing 30–35 g, were used. Animals were anesthetized using a mixture of xylazine (10 mg/kg) and ketamine (70 mg/ kg). After a catheter for injection was placed in the tail vein, hairs over the skull were removed and the animal was placed in a supine position on the sonication table as in Fig. 1A. Sonication. The animals were prepared for the experiments as described above and placed on the system. The experimental protocol is summarized in Fig.1B. T1-weighted images were obtained to aid in the selection of target locations in the brain. After injecting the anti-dopamine D4 receptor antibody (50 ll in original concentration) via tail vein, sonication was performed with simultaneous injection of a 10 or 50 ll bolus of

microbubble-based ultrasound contrast agent OPTSION (Amersham Health, Princeton, NJ) which contains albumin coated microbubbles (mean diameter = 2.0–4.5 · 10 6 m; concentration = 5–8 · 108 bubbles/ ml). The sonication was delivered to one location in the gray matter with the center of the focal spot at approximately 2–3 mm deep from the surface in each brain. The sonication was pulsed with a burst length of 10 ms and a repetition frequency of 1 Hz. The duration of the whole sonication was 40 s. The peak acoustic pressure amplitude levels were kept constant over the duration of each sonication, however it ranged from 0.6 to 1.1 MPa (calibrated in water), depending on the sonication. The acoustic power output and the focal pressure amplitude as a function of the applied radiofrequency power were measured as described elsewhere [21]. As the ultrasound energy is attenuated during trans-skull delivery, the actual acoustic pressure in the mouse brain is expected to be lower. After the sonication and MR study was completed, trypan blue (80 mg/kg) was injected through the tail vein to mark and confirm the BBB disruption on the tissue blocks. Magnetic resonance imaging. The magnetic resonance imaging (MRI) scanner was a standard 1.5 T Signa system (General Electric Medical Systems, Milwaukee, WI). A 7.5-cm diameter surface coil was placed under the head. The sonications were performed through the hole of the coil that was filled with the bag containing degassed water (Fig.1A). A gradient echo sequence was used to aim the beam at the brain. Following the sonications, T1-weighted fast spin echo (FSE) images were obtained and repeated after injection of an intravenous (iv) bolus (10 ll) of gadopentetate dimeglumine MR contrast agent (Magnevist, Berlex Laboratories, Wayne, NJ) to detect and evaluate the opening of the BBB. The parameters for the MRI scans are listed in Table 1. Signal analysis. The MRI contrast enhancement was evaluated at each target location by averaging the signal intensity at the selected region of interest (ROI). The signal was normalized to the baseline value in the ROI before the contrast injection. ROIs selected at the contralateral hemisphere of the sonicated brain tissue served as controls. The average, normalized contrast enhancement observed in these control locations was subtracted from that in the targeted locations to estimate the impact of the sonications. Tissue preparation and immunohistochemistry. The animals were sacrificed approximately 3 h after sonication. The brains were immediately removed and frozen in an OCT compound (Tissue-Tek, Miles, Elkhart, IN). During tissue preparation, tissue damage was assessed and classified into 3 categories depending on the extent of hemorrhage where 0 indicates no-, 1 slight-, and 2 severe-hemorrhage (Fig. 2C). For hematoxylin–eosin and immunohistochemical staining, we prepared 10-lm sections cut parallel to the beam direction across the MRI slices. Immunohistochemical examination was carried out by staining the sections with a goat anti-rabbit IgG biotin labeled antibody (Vector Laboratories, Burlingame, CA). Signals were produced using the ABC method provided by the manufacture. Methyl green was used as a counter staining. The intensity of the immunohistochemical reaction was recorded on a 3point scale as 0 (no obvious positive signals), 1 (positive signals confirmed under a light microscope), or 2 (positive signals confirmed at the macroscopic level).

Results

Fig. 1. A diagram and the protocol for the BBB opening used in this experiment. (A) Mice in the supine position were placed on the sonication table in the MR scanner. The ultrasound beam was focused on the target in the brain through the intact skull. (B) After the injection of antibody, simultaneous sonication and OPTISON injection were carried out. For details, see Methods.

Using a mouse model and MRI-guided focused ultrasound-induced BBB disruption, we made an attempt to deliver a polyclonal antibody against the extracellular domain of the dopamine D4 receptor to the CNS. The antibody is originated from rabbit IgG. Therefore, the rabbit polyclonal antibody should be able to be detected with an anti-rabbit IgG antibody on the sections with successful BBB disruption. The experiment setup is illustrated in Fig. 1A, which is a setup similar to those of our previous studies in rabbits [13]. The whole setup was integrated into

M. Kinoshita et al. / Biochemical and Biophysical Research Communications 340 (2006) 1085–1090

1087

Table 1 MRI parameters used in the study Sequence

Purpose

Repetition time (ms)

Echo time (ms)

Matrix

Flip angle ()

Bandwidth (kHz)

Number of acquisitions

FSE T1-W

Target selection and contrast enhancement Tissue anatomy targeting

500

15

256 · 256

90

16

4

2

256 · 128

30

32

1

Gradient echo

8.2

FSE T1-W, fast spin echo T1-weighted sequence.

Fig. 2. MR monitoring of the BBB disruption and photographs of harvested brain showing the BBB disruption induced by focused ultrasound. (A) BBB opening was easily monitored by the leakage of MR contrast agent into the brain parenchyma. Leakage of the MR contrast agent into the brain parenchyma is observed at the MR images in 3 planes (AX, axial; COR, coronal; SAG, sagittal (arrows)). (B) The brain of the animal was harvested 3 h after sonication (0.8 MPa with 50 ll OPTISON). The location of the BBB opening was confirmed by trypan blue staining the affected area. Although no apparent damage was seen in successful cases, the signals from the injected antibody completely overlapped the area stained with trypan blue (compare Brain tissue and IH: immunohistochemistry). (C) Impact of the focused ultrasound on the damage was assessed. The tissue specimens were classified into three categories depending on the degree of BBB damage determined by the extent of hemorrhage; 0 indicates no-, 1 slight-, and 2 severe-hemorrhage.

a 1.5 T MR scanner for image guidance and monitoring. For sonication, each mouse was placed on the table in the supine position and then the brain was scanned. After selecting the target on the MR images, the target coordinates were converted to the ultrasound sonication position-

ing system. Antibody (original concentration 50 ll) and the ultrasound contrast agent OPTISON (10 or 50 ll) were injected into the tail vein during simultaneous sonication through the intact skull using a 0.69 MHz focused ultrasound transducer (Fig. 1B). Subsequently, a

1088

M. Kinoshita et al. / Biochemical and Biophysical Research Communications 340 (2006) 1085–1090

Table 2 Severity of tissue damage and BBB disruption by focused ultrasound Severity of tissue damage

0 1 2

Acoustic pressure in water (MPa) 0.6* (n = 3)

0.8* (n = 2)

0.8 (n = 4)

0.9 (n = 4)

1.1 (n = 2)

1/2 1/1 —

— 2/2 —

1/3 1/1 —

0/1 1/2 1/1

— — 2/2

Severity of tissue damage with various applied acoustic pressure by focused ultrasound is presented. The tissue specimens were classified into three categories depending on the tissue damage determined by the extent of hemorrhage; 0 indicates no-, 1 slight-, and 2 severe-hemorrhage. Each result is presented as (number of BBB disruption cases)/(total number of each group). Experimental groups with asterisk are groups sonicated with 50 ll OPTISON. Others were sonicated with 10 ll OPTISON.

gadolinium-based contrast agent was injected and the BBB breach was confirmed by post-sonication MR study. When successful BBB opening was achieved, the leakage of the MR contrast agent to the brain parenchyma was observed on the MR images (Figs. 2A and B). Finally, we injected trypan blue, which is unable to penetrate the intact BBB, to mark the site of the BBB disruption. Brains were harvested 3 h post-sonication. Successful BBB disruption was confirmed by the presence of a blue spot due to leakage of the dye into the brain parenchyma. In groups where the applied pressure amplitude was lower than 0.8 MPa, we observed no, or minute, hemorrhagic foci at the center of the ultrasound focus (Figs. 2B and C, Table 2). On the other hand, in groups where it exceeded 0.8 MPa, we noted major tissue damage in some animals and at an acoustic power of 1.1 MPa, all mice manifested major tissue damage (Table 2), suggesting that the optimum power for non-invasive murine BBB disruption ranges from 0.6 to 0.8 Mpa. After the attenuation of the beam through the mice skull is taken into consideration, this value is within the same order of that obtained in rabbits in our previous study [15]. When the sections were stained with anti-rabbit IgG antibody, positive signals were detected at the locations where the trypan blue staining was observed (Fig. 2B). Under a microscope, the signals emanated from locations within the sonicated focus such as the hippocampus (Fig. 3A) and small cells at the basal ganglia (Fig. 3B), sites characteristic for the localization of the dopamine D4 receptor [20,22]. On the contralateral side of the brain, no obvious staining was detected. These findings suggest that the anti-dopamine D4 receptor antibody was delivered only at sites where BBB disruption had been produced by sonication. When we compared the staining intensity and the MR signal change before and after sonication, we noted a good correlation (Fig. 4). These results suggest that MR imaging can be used for both target planning and monitoring of the BBB disruption. Discussion Today many molecular targeting drugs have been developed for various diseases, such as anti-HER2 monoclonal antibody; herceptin for breast cancer [23],

and anti-CD20 monoclonal antibody; Rituximab for malignant lymphoma [24]. There are also evidences suggesting that antibodies against the Ab can reverse cognitive deficits in early Alzheimer’s disease [25,26]. However, for in vivo clinical treatment, these promising large molecular agents are not applicable because of the presence of BBB. Antibodies with a molecular size larger than 150 kDa are easily blocked by the BBB. For example, the cerebrospinal fluid level of Rituximab in patients with CNS lymphoma is reported to be only 0.1% of that in the serum [27]. Several methods have been proposed to circumvent the BBB for drug delivery. In the case of delivering TF-CRM107, a transferrin receptor ligand-targeted toxin conjugate, to the target location for treating malignant gliomas, a method called ‘‘convection’’ was used. As this method requires the presence of a catheter inserted into the brain during the entire treatment, it involves invasive procedures and limits the number of treatment-eligible candidates [3,4]. Other methods have focused on modifying the agents to allow them to penetrate the BBB [1,5–7]. Although these techniques are effective in terms of trans-BBB delivery, they cannot provide site specificity, bringing up the possibility of causing unwanted toxic effects at locations outside of the treatment target. The method demonstrated in this report can overcome the above-mentioned problems. As the ultrasound energy is concentrated only around the target area, therapeutic agents can be delivered site specifically, sparing the surrounding tissue. More importantly, our previous studies have shown that this focused ultrasound-induced BBB disruption is transient and reversible [15]. This fact is greatly important opening the possibility of multiple or repeated use of this technique. Also MR imaging is an integral part of our system. The real-time monitoring of each step provides continuous closed-loop feedback information about the location and extent of the focal energy delivery. In addition, the BBB disruption magnitude had a good correlation with the signal intensity change on the MR images, suggesting that the BBB disruption can be predictable by the obtained MR images. In our rabbit studies [13], we occasionally noted a few extravasated erythrocytes around microvessels. This suggests that the mechanism underlying the BBB disruption involves physical damage

M. Kinoshita et al. / Biochemical and Biophysical Research Communications 340 (2006) 1085–1090

1089

Fig. 3. Light microscopic examination of the sonicated brain from the same animal from Fig. 2B. Detailed views of the sonicated and the contralateral side (control) of the hippocampus (A) and the basal ganglia (B) are presented. Although no obvious structural damage is observed (see HE), positive signals from the anti-dopamine D4 receptor antibody can be detected at the entire hippocampus (A) or at the basal ganglia (B). The strongly stained small cells in the basal ganglia are compatible with previous reports on the distribution of dopamine D4 receptors in the rat brain. In the untreated hemisphere, signals are weak or absent.

to components comprising the BBB. We have also seen enhanced active transportation of molecules through the BBB after treatment, leaving a possibility of a biophysiological effect created by microbubbles and ultrasound being involved in the opening [15]. Although further optimization is necessary to render our method clinically applicable, it represents a major step toward the use of antibodies and other macromolecular

therapeutic and/or diagnostic agents in the treatment of various CNS disorders. Acknowledgments This investigation was supported by the Shinya International Exchange Fund, the Osaka Medical Research Foundation for Incurable Diseases, the Osaka Neurological

1090

M. Kinoshita et al. / Biochemical and Biophysical Research Communications 340 (2006) 1085–1090

Fig. 4. Correlation between antibody staining intensity and the intensity of changes seen by MRI after BBB opening induced by focused ultrasound. The immunohistochemical reaction from the injected antibody was recorded as 0 (no obvious positive signals), 1 (positive signals confirmed by light microscopy), or 2 (positive signals confirmed at the macroscopic level) and the intensity of the MR changes after BBB opening was plotted as a function of staining intensity. There is a clear correlation.

Institute, NIH Grant (EB003268) and U41-RR 019703 NIH P01CA67165-03. References [1] W.M. Pardridge, Drug and gene delivery to the brain: the vascular route, Neuron 36 (2002) 555–558. [2] N.J. Abbott, I.A. Romero, Transporting therapeutics across the blood–brain barrier, Mol. Med. Today 2 (1996) 106–113. [3] D.W. Laske, R.J. Youle, E.H. Oldfield, Tumor regression with regional distribution of the targeted toxin TF-CRM107 in patients with malignant brain tumors, Nat. Med. 3 (1997) 1362–1368. [4] M. Weaver, D.W. Laske, Transferrin receptor ligand-targeted toxin conjugate (Tf-CRM107) for therapy of malignant gliomas, J. Neurooncol. 65 (2003) 3–13. [5] W.M. Pardridge, Drug and gene targeting to the brain with molecular Trojan horses, Nat. Rev. Drug Discov. 1 (2002) 131–139. [6] W.M. Pardridge, Blood–brain barrier genomics and the use of endogenous transporters to cause drug penetration into the brain, Curr. Opin. Drug Discov. Dev. 6 (2003) 683–691. [7] S.R. Schwarze, A. Ho, A. Vocero-Akbani, S.F. Dowdy, In vivo protein transduction: delivery of a biologically active protein into the mouse, Science 285 (1999) 1569–1572. [8] R.A. Kroll, E.A. Neuwelt, Outwitting the blood–brain barrier for therapeutic purposes: osmotic opening and other means, Neurosurgery 42 (1998) 1083–1099, discussion 1099–1100. [9] N.D. Doolittle, M.E. Miner, W.A. Hall, T. Siegal, E. Jerome, E. Osztie, L.D. McAllister, J.S. Bubalo, D.F. Kraemer, D. Fortin, R. Nixon, L.L. Muldoon, E.A. Neuwelt, Safety and efficacy of a multicenter study using intraarterial chemotherapy in conjunction with osmotic opening of the blood–brain barrier for the treatment of patients with malignant brain tumors, Cancer 88 (2000) 637–647. [10] C.M. Tempany, E.A. Stewart, N. McDannold, B.J. Quade, F.A. Jolesz, K. Hynynen, MR imaging-guided focused ultrasound surgery of uterine leiomyomas: a feasibility study, Radiology 226 (2003) 897– 905.

[11] K. Hynynen, O. Pomeroy, D.N. Smith, P.E. Huber, N.J. McDannold, J. Kettenbach, J. Baum, S. Singer, F.A. Jolesz, MR imagingguided focused ultrasound surgery of fibroadenomas in the breast: a feasibility study, Radiology 219 (2001) 176–185. [12] M. Kinoshita, K. Hynynen, A novel method for the intracellular delivery of siRNA using microbubble-enhanced focused ultrasound, Biochem. Biophys. Res. Commun. 335 (2005) 393–399. [13] K. Hynynen, N. McDannold, N. Vykhodtseva, F.A. Jolesz, Noninvasive MR imaging-guided focal opening of the blood–brain barrier in rabbits, Radiology 220 (2001) 640–646. [14] K. Hynynen, N. McDannold, N. Vykhodtseva, F.A. Jolesz, Noninvasive opening of BBB by focused ultrasound, Acta Neurochir. Suppl. 86 (2003) 555–558. [15] K. Hynynen, N. McDannold, N.A. Sheikov, F.A. Jolesz, N. Vykhodtseva, Local and reversible blood–brain barrier disruption by noninvasive focused ultrasound at frequencies suitable for transskull sonications, Neuroimage 24 (2005) 12–20. [16] N. McDannold, N. Vykhodtseva, F.A. Jolesz, K. Hynynen, MRI investigation of the threshold for thermally induced blood–brain barrier disruption and brain tissue damage in the rabbit brain, Magn. Reson. Med. 51 (2004) 913–923. [17] N. Sheikov, N. McDannold, N. Vykhodtseva, F. Jolesz, K. Hynynen, Cellular mechanisms of the blood–brain barrier opening induced by ultrasound in presence of microbubbles, Ultrasound Med. Biol. 30 (2004) 979–989. [18] N.I. Vykhodtseva, K. Hynynen, C. Damianou, Histologic effects of high intensity pulsed ultrasound exposure with subharmonic emission in rabbit brain in vivo, Ultrasound Med. Biol. 21 (1995) 969–979. [19] N. McDannold, M. Moss, R. Killiany, D.L. Rosene, R.L. King, F.A. Jolesz, K. Hynynen, MRI-guided focused ultrasound surgery in the brain: tests in a primate model, Magn. Reson. Med. 49 (2003) 1188– 1191. [20] K. Wedzony, A. Chocyk, M. Mackowiak, K. Fijal, A. Czyrak, Cortical localization of dopamine D4 receptors in the rat brain– immunocytochemical study, J. Physiol. Pharmacol. 51 (2000) 205– 221. [21] K. Hynynen, N.I. Vykhodtseva, A.H. Chung, V. Sorrentino, V. Colucci, F.A. Jolesz, Thermal effects of focused ultrasound on the brain: determination with MR imaging, Radiology 204 (1997) 247– 253. [22] M.C. Defagot, E.L. Malchiodi, M.J. Villar, M.C. Antonelli, Distribution of D4 dopamine receptor in rat brain with sequence-specific antibodies, Brain Res. Mol. Brain Res. 45 (1997) 1–12. [23] R. Bell, What can we learn from Herceptin trials in metastatic breast cancer? Oncology 63 (Suppl. 1) (2002) 39–46. [24] D.G. Maloney, O.W. Press, Newer treatments for non-Hodgkin’s lymphoma: monoclonal antibodies, Oncology (Williston Park) 12 (1998) 63–76. [25] C. Janus, J. Pearson, J. McLaurin, P.M. Mathews, Y. Jiang, S.D. Schmidt, M.A. Chishti, P. Horne, D. Heslin, J. French, H.T. Mount, R.A. Nixon, M. Mercken, C. Bergeron, P.E. Fraser, P. St. GeorgeHyslop, D. Westaway, A beta peptide immunization reduces behavioural impairment and plaques in a model of Alzheimer’s disease, Nature 408 (2000) 979–982. [26] I. Klyubin, D.M. Walsh, C.A. Lemere, W.K. Cullen, G.M. Shankar, V. Betts, E.T. Spooner, L. Jiang, R. Anwyl, D.J. Selkoe, M.J. Rowan, Amyloid beta protein immunotherapy neutralizes Abeta oligomers that disrupt synaptic plasticity in vivo, Nat. Med. 11 (2005) 556–561. [27] J.L. Rubenstein, D. Combs, J. Rosenberg, A. Levy, M. McDermott, L. Damon, R. Ignoffo, K. Aldape, A. Shen, D. Lee, A. Grillo-Lopez, M.A. Shuman, Rituximab therapy for CNS lymphomas: targeting the leptomeningeal compartment, Blood 101 (2003) 466–468.