Search

Rituximab is a monoclonal antibody directed against the CD20 B-cell antigen and is approved for the treatment of B-cell lymphoma. We investigated the pharmacokinetics of rituximab following intravitreal administration to assess the feasibility of treating primary intraocular lymphoma. Intravitreal injections of rituximab 0.1 ml (1 mg) were performed in rabbits. Drug concentrations in the aqueous and vitreous humor were measured at intervals from 2 to 17 days after administration. The half-life of the total amount of rituximab in the two compartments was calculated to be 4.7 days. The aqueous and vitreous humor drug levels decayed in parallel maintaining an average ratio of approximately seven. Fitting the data to a two-compartment model yielded a clearance from the aqueous humor of 1.2 microl/min. The clearance was less than the reported rate of aqueous humor outflow indicating that elimination by this route could have been sufficient to account for the disappearance of the drug from the eye. The duration of time over which sustained levels of rituximab were achieved suggest that intravitreal administration warrants further investigation as an approach to treating vitreous and anterior chamber infiltrates in patients with primary intraocular lymphoma.

Object. Convection-enhanced delivery (CED), the delivery and distribution of drugs by the slow bulk movement of fluid in the extracellular space, allows delivery of therapeutic agents to large volumes of the brain at relatively uniform concentrations. This mode of drug delivery offers great potential for the treatment of many neurological disorders, including brain tumors, neurodegenerative diseases, and seizure disorders. An analysis of the treatment efficacy and toxicity of this approach requires confirmation that the infusion is distributed to the targeted region and that the drug concentrations are in the therapeutic range. Methods. To confirm accurate delivery of therapeutic agents during CED and to monitor the extent of infusion in real time, albumin-linked surrogate tracers that are visible on images obtained using noninvasive techniques (iopanoic acid [IPA] for computerized tomography [CT] and Gd—diethylenetriamine pentaacetic acid for magnetic resonance [MR] imaging) were developed and investigated for their usefulness as surrogate tracers during convective distribution of a macromolecule. The authors infused albumin-linked tracers into the cerebral hemispheres of monkeys and measured the volumes of distribution by using CT and MR imaging. The distribution volumes measured by imaging were compared with tissue volumes measured using quantitative autoradiography with [14C]bovine serum albumin coinfused with the surrogate tracer. For in vivo determination of tracer concentration, the authors examined the correlation between the concentration of the tracer in brain homogenate standards and CT Hounsfield units. They also investigated the long-term effects of the surrogate tracer for CT scanning, IPA-albumin, on animal behavior, the histological characteristics of the tissue, and parenchymal toxicity after cerebral infusion. Conclusions. Distribution of a macromolecule to clinically significant volumes in the brain is possible using convection. The spatial dimensions of the tissue distribution can be accurately defined in vivo during infusion by using surrogate tracers and conventional imaging techniques, and it is expected that it will be possible to determine local concentrations of surrogate tracers in voxels of tissue in vivo by using CT scanning. Use of imaging surrogate tracers is a practical, safe, and essential tool for establishing treatment volumes during high-flow interstitial microinfusion of the central nervous system.

We measured zidovudine concentrations in blood, muscle, and brain extracellular fluid (ECF) by microdialysis and in serum ultrafiltrate and cerebrospinal fluid (CSF) samples during a continuous intravenous infusion (15 mg/kg/h) and after bolus dosing (50-80 mg/kg over 15 min) in nonhuman primates to determine whether CSF drug penetration is a valid surrogate for blood-brain barrier penetration. Recovery was estimated in vivo by zero net flux for the continuous infusion and retrodialysis for the bolus dosing. In vivo recovery was tissue-dependent and was lower in brain than in blood or muscle. Mean (+/-S.D.) steady-state blood, muscle, and brain zidovudine concentrations by microdialysis were 112 +/- 63.8, 105 +/- 51.1, and 13.8 +/- 10.4 microM, respectively; and steady-state serum ultrafiltrate and CSF concentrations were 81.2 +/- 40.2 and 14.1 +/- 8.0 microM, respectively. Brain ECF penetration (microdialysis brain/blood ratio) and CSF penetration (standard sampling CSF/serum ratio) at steady state were 0.13 +/- 0.06 and 0.17 +/- 0.02, respectively. With bolus dosing the mean (+/-S.D.) zidovudine area under concentration-time curve (AUC) normalized to a dose of 80 mg/kg was 577 +/- 103 microM. h in blood, 528 +/- 202 microM. h in muscle, and 108 +/- 74 microM. h in brain (brain/blood ratio of 0.18 +/- 0.10) by microdialysis. Serum ultrafiltrate AUC was 446 +/- 72 microM. h and the CSF AUC was 123 +/- 4.7 microM. h (CSF/serum ratio of 0.28 +/- 0.06). In conclusion, recovery was tissue-dependent. CSF and brain ECF zidovudine concentrations were comparable at steady state, and the corresponding AUCs were comparable after bolus injection. Thus, zidovudine penetration in brain ECF and CSF in nonhuman primates is limited to a similar extent, presumably by active transport, as in other species.

Regional delivery of drugs can offer a pharmacokinetic advantage in the treatment of localized tumors. One method of regional delivery is by intra-arterial infusion into the basilar/vertebral artery network that provides local access to infratentorial tumors, which are frequent locations of childhood brain cancers. Proper delivery of drug by infused solutions requires adequate mixing of the infusate at the site of infusion within the artery lumen. Our mixing studies with an in vitro model of the vertebral artery network indicate that streaming of drug solution is likely to occur at low, steady infusion rates of 2 ml/min. Streaming leads to maldistribution of drug to distal perfused brain regions and may result in toxic levels in some regions while concurrently yielding subtherapeutic levels in adjacent regions. According to our model findings, distribution to both brain hemispheres is not likely following infusion into a single vertebral artery even if the infusate is well-mixed at the infusion site. This outcome results from the unique fluid flow properties of two converging channels, which are represented by the left and right vertebral branches converging into the basilar. Fluid in the model remains stratified on the side of the basilar artery served by the infused vertebral artery. Careful thought and planning of the methods of intravertebral drug infusions for treating posterior fossa tumors are required to assure proper distribution of the drug to the desired tissue regions. Improper delivery may be responsible for some noted toxicities or for failure of the treatments.

We have developed a pharmacokinetic model for monoclonal antibodies (mAb) to aid in investigating protocols for targeting small primary tumors or sites of metastatic disease. The model describes the uptake of systemically-administered antibody by a prevascular spherical tumor nodule embedded in normal tissue. The model incorporates plasma kinetics, transcapillary transport, interstitial diffusion, binding reactions, and lymphatic clearance. Antigen internalization can easily be incorporated. Simulations obtained from a three-dimensional finite element analysis are used to assess errors in predictions from earlier models in which the influence of the normal tissue was collapsed into a boundary condition at the tumor surface. The model employing a Dirichlet boundary condition substantially overpredicted the mean total tumor mAb concentration at all times. Although the model with a concentration-dependent flux (composite) boundary condition underpredicted mAb concentration, the discrepancy with finite element results is only notable at early times. Sensitivity analyses were performed on mAb dose and on the coefficients for mAb diffusion in the tissue regions, since reported antibody diffusivity values have varied over 30-fold. The results of the study suggest that mAb diffusivity and mAb binding site density in tumors should have major influences on optimizing doses and scheduling of mAb administration in tumor targeting protocols.

Purpose. We examine the theoretical basis for calibrating microdialysis probes in vivo for pharmacokinetic experiments in which the extracellular analyte concentrations vary in time.

Purpose: Because intraperitoneal (i.p.) therapy may provide a therapeutic advantage and because hyperthermia enhances carboplatin (CBDCA) cytotoxicity, we evaluated the feasibility, toxicity, and pharmacokinetics of CBDCA given via continuous hyperthermic peritoneal perfusion (CHPP) in patients with small-volume residual ovarian cancer. Patients and Methods: Six patients underwent optimal cytoreductive procedures (residual disease ≤5 mm) as initial treatment of stages II and III epithelial ovarian adenocarcinoma. All patients received a 90-min CHPP at a CBDCA dose of 800–1200 mg/m2, with the perfusate being recirculated rapidly from a reservoir through a heat exchanger, resulting in i.p. temperatures of 41–43 °C. Plasma, perfusate, and urine samples were collected and platinum was quantified by flameless atomic absorption spectrophotometry. Results: At no time did any patient's core temperature exceed 40 °C. Peak perfusate platinum concentrations were 8- to 15-fold higher than peak ultrafilterable plasma concentrations. The permeability-area product was extremely high and variable (14–90 ml/min), resulting in a regional advantage of 1.9–5.3. The percentage of the dose absorbed ranged widely from 27% to 77%. Dose-limiting hematologic toxicity was observed at a dose of 1200 mg/m2 and this was associated with a CBDCA AUC in plasma of 11 mg min ml−1. Conclusions: CHPP with CBDCA was safely given to three patients at a dose of 800 mg/m2, and dose-limiting hematologic toxicities observed at 1200 mg/m2, correlated with the plasma CBDCA exposure established when lower doses of CBDCA are given systemically. The pharmacokinetic data are consistent with the expected effect of vigorous mixing on the exposed peritoneal surface area. Variable drug absorption and clearance make the prediction of systemic exposure highly uncertain. These findings may have important implications for novel therapies given i.p.

Monoclonal antibodies (MAbs), produced for specific tumor antigens, can be linked with radioisotopes or metabolic toxins and administered intraperitoneally (i.p.) to treat metastatic cancer located on the peritoneum. Despite their specific binding properties, these proteins distribute to the serosal surface of all tissues surrounding the cavity in the same manner as other serum proteins. Recent data have raised a problem of access of the solution containing the MAb to significant portions of the peritoneal surface. If the MAb does arrive at the surface of the tumor, it penetrates via diffusion and convection. The rapidity and depth of penetration of the MAb are very dependent on the binding characteristics of the MAb to the tumor cells. Current data indicate that tumors often have a large interstitial space relative to normal muscle, and this can accelerate both diffusion and convection. However, a highly permeable tumor vasculature in the absence of lymphatic drainage has also been shown to produce interstitial pressure gradients from the center toward the periphery of the tumor, setting up a potential outward flow which may be a significant barrier to the movement of MAbs into the nodule. While theoretical mechanisms of diffusion, convection, and binding are well established, there is still a great need for in vivo data.

Pharmacokinetics is an important property of a drug. It determines the temporal profiles of drug and metabolites in blood and tissues, which in turn drive the magnitude and temporal pattern of response following drug administration. Poor pharmacokinetic properties of a drug limit its clinical utility. Too rapid elimination normally necessitates frequent administration and inconvenience for the patient; too slow elimination creates potential problems when trying to remove drug in the event of intoxication. Poor oral absorption may also limit the use of this common and convenient route of administration. For these and other reasons it is important to try to ensure that new drugs under development are likely to have desirable pharmacokinetic features at potential therapeutic doses, prior to actual administration to humans. In addition, accurate prediction of human pharmacokinetics from preclinical studies, often coupled with in vitro human receptor occupation data, helps to design Phase I studies and to calculate the appropriate drug doses that will be needed. This chapter contains a brief discussion of the three preclinical approaches to this prediction: allometry, microdosing, and physiologic modeling, incorporating in vitro and physiologic data. All three approaches involve scale-up of one form or another.

The origin and magnitude of potential errors in whole-blood sampling are prredicted on the basis of a mathematical model. The model describes the kinetics of solute metabolism, breakdown, and interphase distribution (i.e., partitioning and exchange between formed elements and plasma) within a blood sample during sample withdrawal and storage. The model is applied to the determination of the integral over time of solute concentration in the plasma (area-under-the-curve, or AUC) from a sample withdrawn through an arterial or venous catheter. Errors in AUC determination can be substantial and are strongly dependent on the duration of sampling (T), the rate constants for solute degradation processes, the rate constant for solute exchange between the formed elements and the plasma (ke). ,nd the equilibrium ratio for distribution of the solute between formed elements and plasma (R). When the value of the dimensionless group keT/R is small, little solute exchanges between plasma water and formed elements before the two phases of the blood are separated. When keT/R is large, the solute distribution is close to equilibrium at all times. In these two keT/R limits, the contribution of solute redistribution to sampling error is small. Sizable errors resulting from redistribution are associated with intermediate values of keT/R, even in the absence of metabolism and despite rapid separation of the phases at the end of the withdrawal period. Chemical conversion within either of the blood phases introduces additional sampling error under most circumstances.

For many compounds (neurotrophic factors, antibodies, growth factors, genetic vectors, enzymes) slow diffusion in the brain severely limits drug distribution and effect after direct drug administration into brain parenchyma. We investigated convection as a means to enhance the distribution of the large and small molecules 111In-labeled transferrin (111In-Tf; M(r), 80,000) and [14C]sucrose (M(r), 359) over centimeter distances by maintaining a pressure gradient during interstitial infusion into white matter to generate bulk flow through the brain interstitium. The volume of distribution (Vd) containing > or = 1% concentration of infusion solution increased linearly with the infusion volume (Vi) for 111In-Tf(Vd/Vi, 6:1) and [14C]sucrose (Vd/Vi, 13:1). Twenty-four hours after infusion, the distribution of 111In-Tf was increased and more homogeneous, and penetration into gray matter had occurred. By using convection to supplement simple diffusion, enhanced distribution of large and small molecules can be obtained in the brain while achieving drug concentrations orders of magnitude greater than systemic levels.

High-flow microinfusion provides a means for delivering macromolecules to large volumes of brain in easily obtainable time intervals. Slowly degraded approximately 180-kDa macromolecules, delivered at a constant volumetric flow rate of 3 microliters/min into homogeneous brain tissue (e.g., gray matter), would penetrate to a 1.5-cm radius in 12 h. The predicted concentration profile is relatively flat until it declines precipitously at the flow front. Hence, tissues are dosed rather uniformly, providing control over the undesired toxicity that may occur with alternative methods that depend on large concentration gradients for tissue transport. The penetration advantage of high-flow (convective) over low-flow (diffusive) microinfusion has been assessed at fixed pharmacodynamic effect. A 12-h high-flow microinfusion of a macromolecule degraded with a characteristic time of 33.5 h would provide 5- to 10-fold increases in volume over low-flow infusion and total treatment volumes > 10 cm3. Slower degradation rates would result in larger treatment volumes; more rapid degradation rates would reduce the volume but still favor convective over diffusive administration. This technique may be applicable to a variety of diagnostic and therapeutic agents such as radioimmunoconjugates, immunotoxins, enzymes, growth factors, and oligonucleotides.

Pegylated liposomal formulations contain lipid conjugated to polyethylene glycol. The disposition of encapsulated drug is dictated by the composition of the liposome, thus altering the pharmacokinetic (PK) profile of the drug. Allometric scaling is based on a power-log relationship between body weight (W) and drug clearance (CL) among mammals and has been used to compare the disposition of nonliposomal drugs across species. The objectives of this study were to use allometric scaling to: (1) compare the disposition of pegylated liposomal drugs across speciesand determine the best scaling model and (2) predict PK parameters of pegylated liposomal drugs in humans. The PK of pegylated liposomal CKD-602 (S-CKD602), doxorubicin (Doxil®), and cisplatin (SPI-077) were compared. PK studies ofS-CKD602, Doxil®, and SPI-077 were performed at the maximum tolerated dose (MTD) in male and female mice, rats, dogs and patients with refractory solid tumors. The allometric equation used to evaluate the relationship between W and CL in each species was CL = a(W)m (a = empirical coefficient; m = allometric exponent). Substitution of physiological variables other than body weight, such as factors representative of the mononuclear phagocyte system (MPS) were evaluated. Dedrick Plots and Maximum Life-Span Potential (MLP) were used to determine scaling feasibility. Standard allometry demonstrated a relationship between clearance of S-CKD602, Doxil®, and SPI-077 and body, spleen, liver, and kidney weights, total monocyte count, and spleen and liver blood flow. However, using scaling to predict CL of these agents in humans often resulted in differences >30%. Despite a strong correlation between body weight and MPS-associated variables with CL among preclinical species, the use of the equations did not predict CL. Thus, new methods of allometric scaling and measures of MPS function need to be developed.

Spatial solute concentration profiles resulting from in vivo microdialysis were measured in rat caudate-putamen by quantitative autoradiography. Radiolabeled sucrose was included in the dialysate, and the tissue concentration profile measured after infusions of 14 min and 61.5 min in an acute preparation. In addition, the changes in sucrose extraction fraction over time were followed in vivo and in a simple in vitro system consisting of 0.5% agarose. These experimental results were then compared with mathematical simulations of microdialysis in vitro and in vivo. Simulations of in vitro microdialysis agreed well with experimental results. In vivo, the autoradiograms of the tissue concentration profiles showed clear evidence of substantial differences between 14 and 61.5 min, even though the change in extraction fraction was relatively small over that period. Comparison with simulated results showed that the model substantially underpre-dicted the observed extraction fraction and overall amount of sucrose in the tissue. A sensitivity analysis of the various model parameters suggested a tissue extracellular volume fraction of approximately 40% following probe implantation. We conclude that the injury from probe insertion initially causes disruption of the blood-brain barrier in the vicinity of the probe, and this disruption leads to an influx of water and plasma constituents, causing a vasogenic edema.

The current mortality rate for lung cancer exceeds 85%, as it has for the last 3 decades. This statistic reflects the utility of the major diagnostic tool that has been used during this period to diagnose lung cancer: the chest X-ray. The overwhelming majority of new cases of lung cancer that are detected with chest X-rays involve individuals who already have regional or distant metastatic disease. Because the systemic treatment of this disease has not improved greatly, patients with metastatic disease rarely are cured. This article reviews the issues involved with the development of sputum-based cellular diagnostics for early stage lung cancer. The biomarker, heterogeneous nuclear ribonucleoprotein A2/B1, is the lead marker for this approach. It has been used in several studies in independent cohorts that have suggested that its overexpression in bronchial epithelial cells is associated highly with the development of lung cancer. This marker is detectable 1 year or more prior to the detection of lung cancer by chest X-ray. Finding this early airway-confined phase of lung cancer may allow for the evolution of new management approaches for very early stage lung cancer. Research activities, such aerosolized chemoprevention, are discussed.

✓ A model was developed to determine whether protein-based chemotherapeutic agents can cross the blood-brain barrier and successfully treat brain tumors. The human small-cell lung carcinoma N417D was grown as a solid tumor in the nude rat brain, and diphtheria toxin (DT) was administered intravenously as therapy. Because rat cells lack functional DT receptors and are 1000 to 10,000 times less sensitive to DT than human cells, a therapeutic window exists between the implanted human tumor and the nude rat host. The pharmacokinetic and pharmacodynamic characteristics of DT were defined. Within 6 hours, more than 90% of the initial DT concentration was removed from the blood. The blood-to-tumor transfer constant Ki for DT in small N417D tumors was 0.49 µl/gm-min, one-fourth to one-fifth the reported values for permeability to proteins in other experimental tumor models. Despite the toxin's short plasma half-life and the relatively intact blood-tumor barrier, DT administered intravenously as a single dose significantly extended animal survival. Untreated nude rats developed solid parenchymal tumors and died in 11 to 16 days (median 15 days). When administered at 0.1 µg/animal, DT increased the median survival time to 19 days (p < 0.0016) while 1.0-µg doses extended median survival times to 26.5 days (p < 0.0002). A higher dose of DT (3.0 µg) had no further beneficial effect on survival (26.1 days). Blood-brain barrier constraints to successful monoclonal antibody-based therapies of brain tumors may have been overestimated since antibody conjugates have plasma half-lives longer than DT, and the permeability of N417D tumors to DT is equal to or less than the permeability of other experimental tumors to large proteins. Recently developed immunotoxins that have the higher potency of DT and a therapeutic window as wide as DT has in this nude rat/human tumor paradigm may be effective in treating brain tumors despite limited blood-tumor permeability.

The pyrimidine analogue gemcitabine (2', 2'-difluorodeoxycitidine, dFdC) is active against pancreatic cancer, and its high clearance (CL(tb)) and low incidence of local toxicity make it an excellent candidate for evaluation as intraperitoneal (IP) therapy. We designed a dosing schema that used multiple sequential exchanges of a peritoneal dialysate containing dFdC in an effort to produce prolonged IP dFdC exposure.As part of a study involving multi-modality therapy for advanced pancreatic adenocarcinoma, patients were treated with four 6-h IP dwells of dFdC (50 mg/m(2) in 2 l) over a 24-h period. A second 24-h cycle of IP dFdC therapy was repeated 1 week later. Each exchange of dialysate contained 50 mg/m(2) dFdC in 2 l of commercial 1.5% dextrose dialysis solution. Plasma and peritoneal fluid were analyzed by HPLC to determine concentrations of dFdC and its inactive metabolite 2', 2' difluorodeoxyuridine (dFdU). Clinical data were recorded to note drug toxicity and response.Nine patients underwent IP dFdC therapy, and eight were able to receive two cycles. There were no recorded significant toxicities. Low plasma dFdC concentrations (1 microg/ml) were present transiently in seven of nine patients, and dFdC was not detectable in the plasma of the other two. Plasma dFdU concentrations were low but increased gradually until 12 h and then declined little if any. IP dFdC concentrations declined rapidly, and dFdC was seldom measurable prior to administration of the next scheduled 6-h dwell. dFdU concentrations in peritoneal fluid were very low (0.5 microg/ml) throughout treatment. The mean area under the concentration versus time curve (AUC) for dFdC in peritoneal fluid was 182 microg/ml x h, which was approximately 70x the AUC of dFdC reported in the ascites of a patient undergoing systemic dFdC therapy.IP dFdC was well tolerated, and no significant toxicities were noted. The rapid decrease in peritoneal dFdC concentrations and low concentrations of IP dFdU imply almost total absorption of IP-administered dFdC. Little, if any, dFdC could be detected in plasma, but the steady-state plasma dFdU concentrations also imply absorption and inactivation of virtually all IP-administered dFdC. These findings are consistent with the known high CL(tb) and low incidence of local toxicity of dFdC and argue for its further evaluation as a drug for IP therapy.

Purpose Sustained-release intravitreal drug implants for posterior segment diseases are associated with significant complications. As an alternative, subconjunctival infusions of drug to the episclera of the back of the eye have been performed, but results in clinical trials for macular diseases showed mixed Results To improve understanding of transscleral drug delivery to the posterior segment, the distribution and clearance of gadolinium-diethylene-triamino-penta-acetic acid (Gd-DTPA) infused in the subconjunctival or intrascleral space was investigated by means of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). Methods In anesthetized rabbits, catheters were placed anteriorly in the subconjunctival or intrascleral space and infused with Gd-DTPA at 1 and 10 muL/min. Distribution and clearance of Gd-DTPA were measured using DCE-MRI. Histologic examination was performed to assess ocular toxicity of the delivery system. results. Subconjunctival infusions failed to produce detectable levels of Gd-DTPA in the back of the eye. In contrast, intrascleral infusions expanded the suprachoroidal layer and delivered Gd-DTPA to the posterior segment. Suprachoroidal clearance of Gd-DTPA followed first-order kinetics with an average half-life of 5.4 and 11.8 minutes after intrascleral infusions at 1 and 10 muL/min, respectively. Histologic examination demonstrated expansion of the tissues in the suprachoroidal space that normalized after infusion termination. Conclusions An intrascleral infusion was successful in transporting Gd-DTPA to the posterior segment from an anterior infusion site with limited anterior segment exposure. The suprachoroidal space appears to be an expandible conduit for drug transport to the posterior segment. Further studies are indicated to explore the feasibility of clinical applications.

Publisher Summary This chapter discusses allometry, physiological pharmacokinetics (PK), and the use of in vitro systems to predict drug metabolism in experimental animals and human study participants. Allometric principles are used to answer a variety of questions relevant to the application of preclinical PK data to the design of dose regimens for humans. Allometric scaling to predict distribution volume from studies in three animal species—rat, dog, and monkey—has been found to be reasonably satisfactory. The distribution and disposition of a drug in the body result from a complex set of physiological processes and biochemical interactions. A physiological PK model was developed to predict the deamination of cytosine arabinoside in humans from enzyme parameters determined from homogenates of human tissue. Interpretation is required for extrapolating in vitro results to the pharmacokinetically relevant rate of metabolism by the liver in the body. Issues such as drug binding in the culture medium and in plasma, distribution between plasma and red blood cells, and penetration into liver slices must be considered. The prediction of in vivo intrinsic clearance of cytochrome P450 (CYP) substrates in the rat from studies using microsomal preparations and isolated hepatocytes is elaborated in the chapter.

In vivo microdialysis relies principally on diffusion as the mechanism for solute exchange between the tissue and the probe perfusate. Mathematical models for describing microdialysis have been formulated based on known concepts of diffusion through tissue and the membranes with which probes are currently constructed. The models predict that the rate of exchange is strongly influenced by processes that remove the analyte from the tissue interstitium. This presents an opportunity to derive additional information about the clearance processes from microdialysis measurements. Experimental and mathematical approaches that have been developed to date for exploiting this possibility are presented in a revised framework suitable for both hydrophilic and lipophilic analytes.

Object. The activity of γ-aminobutyric acid (GABA), the principal inhibitory neurotransmitter, is reduced in the hippocampus in patients with complex partial seizures from mesial temporal sclerosis. To provide preliminary safety and distribution data on using convection-enhanced delivery of agents to treat complex partial seizures and to test the efficacy and safety of regional selective neuronal suppression, the authors infused muscimol, a GABA-A receptor agonist, directly into the hippocampus of nonhuman primates using an integrated catheter electrode. Methods. Ten rhesus monkeys were divided into three groups: 1) use of catheter electrode alone (four monkeys); 2) infusion of escalating concentrations of muscimol followed by vehicle (three monkeys); and 3) infusion of vehicle and subsequent muscimol mixed with muscimol tracer (three monkeys). Infusions were begun 5 days after catheter electrode placement and continued for 5.6 days before switching to the other agent. Head magnetic resonance (MR) images and electroencephalography recordings were obtained before and during the infusions. Brain histological studies and quantitative autoradiography were performed. Neurological function was normal in controls and when muscimol concentrations were 0.125 mM or less, whereas higher concentrations (0.5 and 1 mM) produced reversible apathy and somnolence. Fluid distribution was demonstrated on MR images and muscimol distribution was demonstrated on autoradiographs throughout the hippocampus and adjacent white matter. Conclusions. Targeted modulation of neuronal activity is a reasonable research strategy for the investigation and treatment of medically intractable epilepsy.

A finite-element (FE) method is used to numerically solve a pharmacokinetic model that describes the uptake of systemically administered antibody (mAb) in a prevascular spherical tumor nodule embedded in normal tissue. The model incorporates plasma kinetics, transcapillary transport, lymphatic clearance, interstitial diffusion in both the normal tissue and tumor, and binding reactions. We use results from the FE analysis to assess previous predictions that employed either a Dirichlet boundary condition (b.c.), or an approximate, composite (Dirichlet and Neumann) b.c. at the tumor surface. We find that the Dirichlet b.c. significantly overpredicted the mean total tumor mAb concentration. In contrast, the composite b.c. yielded good agreement with FE predictions, except at early times. We also used the FE model to investigate the influence of the approximately 30-fold difference in the values of mAb diffusion coefficient measured by Clauss and Jain (Cancer Res. 50:3487–3492, 1990) and Berk [et_al.] (Proc. Natl. Acad. Sci. U.S.A. 94:1785–1790, 1997). For low diffusivity, diffusional resistance slows both mAb uptake by and efflux from the tumor. For high diffusivity at the same mAb dose, more rapid uptake produces earlier and higher peak mAb levels in the tumor, while the efflux rate is limited by the dissociation of the mAb–tumor antigen complex. The differences in spatial and temporal variation in mAb concentration between low and high diffusivities are of sufficient magnitude to be experimentally observable, particularly at short times after antibody administration. © 2002 Biomedical Engineering Society.

Regional delivery of drugs can offer a pharmacokinetic advantage in the treatment of localized tumors. One method of regional delivery is by intra-arterial infusion into the basilar/vertebral artery network that provides local access to infratentorial tumors, which are frequent locations of childhood brain cancers. Proper delivery of drug by infused solutions requires adequate mixing of the infusate at the site of infusion within the artery lumen. Our mixing studies with an in vitro model of the vertebral artery network indicate that streaming of drug solution is likely to occur at low, steady infusion rates of 2 ml/min. Streaming leads to maldistribution of drug to distal perfused brain regions and may result in toxic levels in some regions while concurrently yielding subtherapeutic levels in adjacent regions. According to our model findings, distribution to both brain hemispheres is not likely following infusion into a single vertebral artery even if the infusate is well-mixed at the infusion site. This outcome results from the unique fluid flow properties of two converging channels, which are represented by the left and right vertebral branches converging into the basilar. Fluid in the model remains stratified on the side of the basilar artery served by the infused vertebral artery. Careful thought and planning of the methods of intravertebral drug infusions for treating posterior fossa tumors are required to assure proper distribution of the drug to the desired tissue regions. Improper delivery may be responsible for some noted toxicities or for failure of the treatments.

Due to physiological barriers within the eye, which limit penetration of many drugs from the systemic circulation into the vitreous, the most common method of treating retinal disease is direct intravitreal injection. However, this common procedure may be inappropriate for a wide range of drugs as it may lead to highly variable concentrations potentially causing higher toxicity for tissues inside the eye and limiting therapeutic effect. A recent procedure is to use surgically implanted drug release device, called implant here, in the vitreous of the eye that allow controlled release of drug over a sustained period of time. For constant release of drug over 15 hours, a substantial reduction in peak drug concentration is predicted near the retina. When compared with the implant, a doubling of drug concentration would be expected for more than 3 hours near the retina for the intravitreal injection.

Purpose: Our previous work had shown that retinoic acid (RA) inhibits cell growth and induces apoptosis in estrogen receptor-positive (ER-positive) MCF-7 and T-47D human breast carcinoma cells, but not in ER-negative human breast carcinoma cells MB-231 and MB-453. The purpose of this work was to determine whether these differences might be due to differences in uptake and metabolism of the drug between ER-positive and ER-negative cells. Methods: We measured RA uptake in cultured human breast cancer cells and determined its metabolism by high-pressure liquid chromatographic analysis. Results: The two ER-positive cell lines reached maximum RA uptake at about 2 h, followed by a sharp decline, so that most RA had disappeared from the cells and from the medium by 24 h and was found as oxidation products in the culture medium. In contrast, the two ER-negative cell lines showed a pattern of lower accumulation without the sharp increase and subsequent steep decline, so that by 24 h there was more RA in these cells and their culture medium than in the RA-responsive ER-positive cells, even though at 2 h the ER-negative cells had taken up less RA than the ER-positive cells. Kinetic analysis of the uptake of RA in MCF-7 cells was consistent with rapid movement across the cell membranes and the actual rate determined by diffusion of albumin-bound retinoid to the cells. Conclusions: This study is the first to demonstrate profound differences in RA accumulation and confirms previous results on different rates of RA metabolism between ER-positive and ER-negative human breast cancer cells. The findings reported here, therefore, may introduce additional elements to be considered in the design of new drugs for cancer chemoprevention and therapy.

Currently, most new lung cancers are found only after symptoms or signs reflect the occurrence of regional or distant metastases. Delayed awareness of lung cancer accounts for its 87% five year mortality rate. New insights may lead to improved methods of early lung cancer detection. Experimental intervention proposed for early cancer management generally entails the oral administration of chemoprevention agents. In this chapter, we discuss an alternative delivery strategy using direct aerosol delivery. The model chemoprevention agents for this discussion are the vitamin A analogues. Preliminary clinical reports suggest benefit in reducing the frequency of new lung cancers for previously resected non-small cell lung cancer patients as well as treated head and neck cancer patients. Chronic oral administration of retinoids is problematic due to the frequency of debilitating side effects. Alternatively, by delivering retinoids directly to the airways as an aerosol, the first-pass drug exposure is to the target cell population and the amount of drug available to the systemic circulation may be greatly reduced. We hypothesize that this approach to retinoid delivery will result in a significantly greater therapeutic index. Direct delivery of the drug to epithelial surfaces afflicted by early cancer may also be a general strategy for improving the therapeutic index with other chemoprevention agents.

OBJECTIVE The goals of this clinical trial of intraventricular 454A12-rRA therapy were to identify dose-limiting toxicities, to evaluate the pharmacokinetics of single-dose intraventricular 454A12-rRA, and to detect antitumor activity. METHODS We performed a pilot study of intraventricular therapy with the immunotoxin 454A12-rRA in eight patients with leptomeningeal spread of systemic neoplasia. The immunotoxin 454A12-rRA is a conjugate of a monoclonal antibody against the human transferrin receptor and recombinant ricin A chain, the enzymatically active subunit of the protein toxin ricin. Patients were treated with single doses of 454A12-rRA ranging from 1.2 to 1200 micrograms. RESULTS The early phase half-life of 454A12-rRA in ventricular cerebrospinal fluid (CSF) averaged 44 +/- 21 minutes, and the late phase half-life averaged 237 +/- 86 minutes. The clearance of the immunotoxin was faster than the clearance of coinjected technetium-99m-diethylenetriamine penta-acetic acid, averaging approximately 2.4-fold greater. No 454A12-rRA degradation was detected by Western blot analysis of ventricular CSF for a period of 24 hours, and bioactivity was retained in CSF paralleling the concentration of immunotoxin. No acute or chronic drug toxicity was identified in patients who received less than or equal to 38 micrograms of 454A12-rRA by intraventricular injection. Doses more than or equal to 120 micrograms caused a CSF inflammatory response that was associated with transient headache, vomiting, and altered mental status. This acute syndrome was responsive to steroids and CSF drainage. No systemic toxicity was detected. In four of the eight patients, a greater than 50% reduction of tumor cell counts in the lumbar CSF occurred within 5 to 7 days after the intraventricular dose of 454A12-rRA; however, no patient had their CSF cleared of tumor, and clinical or magnetic resonance imaging evidence of tumor progression was demonstrated in seven of the eight patients after treatment. CONCLUSION Tumoricidal concentrations of the immunotoxin 454A12-rRA can be attained safely in the CSF of patients with leptomeningeal tumor spread.

Both theory and clinical studies demonstrate that drug concentrations in the peritoneal cavity can greatly exceed concentrations in the plasma following intraperitoneal administration. This regional advantage has been associated with clinical activity, including surgically documented complete responses in ovarian cancer patients with persistent or recurrent disease following systemic therapy, and has produced a survival advantage in a recent phase III trial. Two pharmacokinetic problems appear to limit the effectiveness of intraperitoneal therapy: poor tumor penetration by the drug and incomplete irrigation of serosal surfaces by the drug-containing solution. We have examined these problems in the context of a very simple, spatially distributed model. If D is the diffusivity of the drug in a tissue adjacent to the peritoneal cavity and k is the rate constant for removal of the drug from the tissue by capillary blood, the model predicts that (for slowly reacting drugs) the characteristic penetration distance is (D/k) 1/2 and the apparent permeability of the surface of a peritoneal structure is (Dk) 1/2. The permeabilityarea product used in classical pharmacokinetic calculations for the peritoneal cavity as a whole is the sum of the products of the tissue-specific permeabilities and the relevant superficial surface areas. Since the model is mechanistic, it provides insight into the expected effect of procedures such as pharmacologic manipulation or physical mixing. We observe that large changes in tissue penetration may be difficult to achieve but that we have very little information on the transport characteristics within tumors in this setting or their response to vasoactive drugs. Enhanced mixing is likely to offer significant potential for improved therapy; however, procedures easily applicable to the clinical setting have not been adequately investigated and should be given high priority. Clinical studies indicate that an increase in irrigated area may be achieved in many patients by individualizing the dialysate volume and consideration of patient position. [J Natl Cancer Inst 1997;89:480-7]

✓ Hyperosmolar blood-brain barrier disruption (HBBBD), produced by infusion of mannitol into the cerebral arteries, has been used in the treatment of brain tumors to increase drug delivery to tumor and adjacent brain. However, the efficacy of HBBBD in brain tumor therapy has been controversial. The goal of this study was to measure changes in vascular permeability after HBBBD in patients with malignant brain tumors. The permeability (K1) of tumor and normal brain blood vessels was measured using rubidium-82 and positron emission tomography before and repeatedly at 8- to 15-minute intervals after HBBBD. Eighteen studies were performed in 13 patients, eight with glioblastoma multiforme and five with anaplastic astrocytoma. The HBBBD increased K1 in all patients. Baseline K1 values were 2.1 ± 1.4 and 34.1 ± 22.1 µl/minute/ml (± standard deviation) for brain and tumor, respectively. The peak absolute increases in K1 following HBBBD were 20.8 ± 11.7 and 19.7 ± 10.7 µl/minute/ml for brain and tumor, corresponding to percentage increases of approximately 1000% in brain and approximately 60% in tumor. The halftimes for return of K1 to near baseline for brain and tumor were 8.1 ± 3.8 and 4.2 ± 1.2 minutes, respectively. Simulations of the effects of HBBBD made using a very simple model with intraarterial methotrexate, which is exemplary of drugs with low permeability, indicate that 1) total exposure of the brain and tumor to methotrexate, as measured by the methotrexate concentration-time integral (or area under the curve), would increase with decreasing infusion duration and would be enhanced by 130% to 200% and by 7% to 16%, respectively, compared to intraarterial infusion of methotrexate alone; and 2) exposure time at concentrations above 1 µM, the minimal concentration required for the effects of methotrexate, would not be enhanced in tumor and would be enhanced by only 10% in brain. Hyperosmolar blood-brain barrier disruption transiently increases delivery of water-soluble compounds to normal brain and brain tumors. Most of the enhancement of exposure results from trapping the drug within the blood-brain barrier, an effect of the very transient alteration of the blood-brain barrier by HBBBD. Delivery is most effective when a drug is administered within 5 to 10 minutes after disruption. However, the increased exposure and exposure time that occur with methotrexate, the permeability of which is among the lowest of the agents currently used clinically, are limited and the disproportionate increase in brain exposure, compared to tumor exposure, may alter the therapeutic index of many drugs.

A technique is described for plasma sampling by continuous membrane plasmapheresis performed on blood flowing through an extracorporeal arteriovenous shunt. The plasmapheresis sampler in the shunt employs replaceable commercial planar membranes 2.5 cm in diameter. Validation tests were conducted for 0.6-μm pore diameter microporous membranes with several low-molecular-weight, nonmetabolized solutes that either rapidly equilibrate between plasma and formed elements or remain extracellular. Ex vivo tests were performed for bolus intravenous administration to rabbits. The technique yielded values for time-averaged plasma concentrations comparable to those obtained with serial blood and continuous blood withdrawal methods. The new technique should be particularly advantageous when the distribution of the solute of interest between plasma and formed elements of the blood undergoes significant changes during the sampling interval as a result of binding, exchange, or metabolism in the formed element phase.

Background: Pegylated liposomal formulations contain lipid conjugated to polyethylene glycol. These formulations of anticancer agents have been designed to prolong the drug circulation time, increase tumor delivery, and improve the therapeutic index. The disposition of encapsulated drug is dictated by the liposomal carriers, thus altering the pharmacokinetic (PK) profile of the drug. Allometric scaling is based on a power-log relationship between body weight (W) and drug clearance (CL) among mammals and has been used to compare the PK of small molecule drugs across species. However, it has not been used to compare PK of liposomes or nanoparticle agents. The objectives of this study were to use allometric scaling to: 1) compare the disposition of pegylated liposomal drugs across species and determine the best scaling model, and 2) predict PK parameters of pegylated liposomal drugs in patients. Methods: PK studies of pegylated liposomal CKD-602 (S-CKD602), pegylated liposomal doxorubicin (PLD), and pegylated liposomal cisplatin (SPI-077) were performed at the maximum tolerated dose (MTD) in male and female rats and dogs and in patients with advanced solid tumors as part of phase I studies. Plasma samples were analyzed for sum total (encapsulated + released) drug. The standard allometric equation, CL = a(W)m (a = empirical coefficient; m = allometric exponent) was used to evaluate the relationship between W and CL in each species. The relationship between measures of the mononuclear phagocyte system (MPS) and CL were also evaluated, by substituting the MPS-associated factors for W (liver weight, spleen weight, spleen blood flow, liver blood flow, and monocyte count). Dedrick Plots were used to determine scaling feasibility. Results: Standard allometric scaling using W showed that CL scaled across all agents: [S-CKD602 (R2 = 0.92), PLD (R2 = 0.90), and SPI-077 (R2 = 0.98).] However, the percent (%) difference from scaled predicted and observed PK parameters in humans were still > 30%. All MPS associated factors showed correlation with the CL of all three agents (R2 > 0.90). Based on R2 values, total monocyte count displayed the strongest correlation (R2 = 0.99) with drug clearance across the three species for all three agents. Conclusions: The model that scaled CL the best across all species was the standard allometric method when using potential measures of the MPS, including liver weight, spleen weight, and total monocyte count compared to W. The selection of PK parameters, physiological variables, and pegylated liposomal agent used in this study all contribute to the determination of the optimal scaling method. However, a consistent and predictive allometric model with new methods of allometric scaling and measures of MPS function need to be developed for pegylated liposomal agents. This potential new method could then be used to facilitate the development of future liposomal and nanoparticle agents. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 373. doi:10.1158/1538-7445.AM2011-373

Hair taken from rats dosed with 203 Hg-labeled methylmercury was fed to previously untreated rats in order to determine if the organo-mercurial was available for release from the hair matrix within the gut lumen and for subsequent systematic absorption. Cumulative fecal excretion data were consistent with an absorption of about 80% of the ingested methylmercury. The relative amounts of methylmercury and of its metabolite, inorganic mercury, in the feces indicated that the percentage of the parent compound released from hair within the intestine equaled or exceeded the estimated bioavailability. Radioactivity in tissues of animals killed 42 hr following hair consumption confirmed that mercury absorption had occured.

✓ Intra-arterial carotid artery chemotherapy for malignant gliomas is limited by focal injuries to the eye and brain which may be caused by poor mixing of the drug with blood at the infusion site. This inadequate mixing can be eliminated in animal models with diastole-phased pulsatile infusion (DPPI) which creates 1-ml/sec spurts during the slow blood flow phase of diastole. Before treatment with intracarotid cisplatin, 10 patients with malignant gliomas were studied to determine whether intravascular streaming occurs after intracarotid infusion in humans, and if so, if it is reduced with DPPI. Regional cerebral blood flow (rCBF) studies were performed by intravenous injection of H215O and positron emission tomography. This was followed by supraor infraophthalmic internal carotid artery (ICA) injections of H215O with either continuous infusion or DPPI. Local H215O concentration in the brain was determined and the images of radiotracer distribution in the continuous infusion and DPPI studies were compared to the rCBF images. Intravascular streaming of the infusate was identified by a heterogeneous distribution of the infused H215O in brain compared to rCBF. Extensive and variable intravascular streaming occurred in three patients who received infusions into the supraophthalmic segment of the ICA. Some brain areas received up to 11 times the expected radiotracer delivery, while other regions received as little as one-tenth. This streaming pattern was markedly reduced or eliminated by DPPI. In the five patients who received infraophthalmic infusions, a minimally heterogeneous distribution of the infusate was detected. The authors conclude that extensive intravascular streaming accompanies supraophthalmic ICA infusions in patients. The magnitude of streaming can be substantially reduced or eliminated with DPPI. Those who perform intra-arterial infusion should consider using DPPI to assure uniform drug delivery to brain.

Publisher Summary This chapter discusses overall concepts of quantitative microdialysis, including some difficulties of application. The major factors that may influence the proper quantitative use of the microdialysis probe include: (1) microdialysate concentrations obtained in vitro depend strongly on microvascular transport and metabolism in the extracellular space surrounding the probe, as well as diffusion; (2) microdialysis probes may establish concentration gradients about themselves that may be of sufficient extent to engage in mass transfer with anatomical regions that are outside the desired sampling or infusion volume; (3) in in vivo application, the permeability of the common dialysis membranes used in current probes usually plays only a minor role because most of the resistance to mass flow into or out of the probe arises from the tissue and not the membrane; and (4) estimates of extracellular concentration obtained from microdialysis data by the analyses are spatially averaged quantities.

A mathematical framework was developed to provide a quantitative basis for either in vivo tissue or in vitro microdialysis. Established physiological and mass transport principles were employed to obtain explicit expressions relating dialysate concentration to tissue extracellular concentration for in vivo applications or external medium concentrations for in vitro probe characterization. Some of the important generalizations derived from the modeling framework are: (i) the microdialysis probe can perturb the spatial concentration profile of the substance of interest for a considerable distance from the probe, (ii) for low molecular weight species the tissue is generally more important than the probe membrane in determining the dialysate-to-tissue concentration relationship, (iii) metabolism, intracellular-extracellular and extracellular-microvascular exchange, together with diffusion, determine the role of the tissue in in vivo probe behavior, and, consequently, (iv) in vitro "calibration" procedures could be useful for characterizing the probe, if properly controlled, but have limited applicability to in vivo performance. The validity of the proposed quantitative approach is illustrated by the good agreement obtained between the predictions of a model developed for tritiated water ([3]H2O) in the brain and experimental data taken from the literature for measurements in the caudoputamen of rats. The importance of metabolism and efflux to the microvasculature is illustrated by the wide variation in predicted tissue concentration profiles among [3]H2O, sucrose and dihydroxyphenylacetic acid (DOPAC).

✓ Four patients with malignant cerebral gliomas received 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU) into the internal carotid artery (ICA) while the ipsilateral jugular drainage was pumped extracorporeally through a hemoperfusion cartridge containing a nonionic adsorbant resin. Each patient received 220 mg/sq m BCNU, infused over 45 minutes through a toposcopic catheter positioned with the tip in the ICA beyond the origin of the ophthalmic artery. Jugular blood was pumped extracorporeally at 300 ml/min through a large-bore catheter in the jugular bulb. Plasma samples were obtained for BCNU measurement at frequent intervals from the right atrium. During a separate treatment, 6 weeks before or after the hemoperfusion treatment, the same dose of BCNU was infused into the ICA and atrial samples were obtained on a similar schedule. Hemoperfusion of the jugular blood during intracarotid infusion reduced the systemic exposure by 56% to 87% and increased total body clearance of BCNU by two- to eightfold. The calculated pharmacokinetic advantage (brain:body exposure ratio) was between 21 and 55:1 when the combined treatment was used.

✓ Focal injury to the brain or retina is a frequent complication of drug delivery to the internal carotid artery (ICA) and may be due to poor mixing of the drug with blood at the infusion site. Rhesus monkeys were studied to determine whether phased drug delivery during diastole from a modified pulsatile angiographic injector would improve drug mixing in vivo. A radiolabeled flow tracer, carbon-14-iodoantipyrine (14C-IAP), was injected into the ICA of three monkeys in 80-msec pulses, each ending at least 50 msec before the end of local diastole. Local isotope concentration in the brain was determined by quantitative autoradiography. The ratio of highest to lowest concentration was 1.86 ± 0.26 (mean ± standard deviation) in the frontoparietal cortex, 1.65 ± 0.42 in the frontoparietal white matter, 1.89 ± 0.28 in the temporal cortex, and 1.39 ± 0.17 in the basal ganglia. These results were similar to recordings in three control animals that received intravenous 14C-IAP to demonstrate complete drug mixing ( 1.37 ± 0.12, 1.41 ± 0.11, 1.70 ± 0.08, 1.22 ± 0.24, respectively), and contrasted to findings in five animals which received continuous intracarotid infusions to demonstrate standard ICA drug delivery (4.54 ± 2.07, 2.94 ± 1.45, 5.43 ± 3.57, 3.60 ± 2.90, respectively). Pulsed intra-arterial infusion during diastole provides a technically simple method for improving intravascular drug mixing, and results in drug delivery to tissue capillaries that is proportional to blood flow.

A flow-limited multicompartment model simulates the distribution and disposition of phenol red in the dogfish after intravenous administration. Glucuronide conjugate, as well as parent compound, was found in urine and bile, but not in plasma, kidney, or liver tissue. An apparent 4-hr lag in phenol red appearance in the gall bladder was simulated using two stirred tanks connected in series to represent the bile ducts. The model should facilitate use of the dogfish in pharmacokinetic studies of drugs, environmental contaminants, and other xenobiotics.

The rate of disappearance of fluorouracil from peritoneal fluid has been experimentally measured and mathematically modeled. The experimental data were obtained following the instillation of 50 ml of dialysis fluid which contained an initial fluorouracil concentration ranging from 24 microM to 12 mM. The rate of disappearance was strongly dependent upon concentration. A distributed model has been formulated which incorporates concepts of diffusion with saturable metabolism and nonsaturable capillary uptake in the tissue surrounding the peritoneal fluid. This model successfully describes the experimental observations and also suggests that the effective penetration depth into tissue is highly dependent upon concentration.

Techniques have been developed for isolated perfusion of chemotherapeutic agents in patients with glioblastoma. Three catheters that facilitate crossing the carotid siphon have been developed; two are based on an everting or toposcopic principle, and one uses microjets for deflectability and improved mixing. Blood from the ipsilateral jugular vein is aspirated at high volumes (300 ml/min) for extracorporeal circulation through an adsorption column (for recovery of carmustine) or dialysers (for recovery of cisplatin). Preliminary experience in 10 patients suggests that high doses of chemotherapeutic agent can be administered using these catheters, with reduced retinal and systemic toxicity.