Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Spread of cancer into the CSF is painful, debilitating and lethal. There is no effective therapy for overt disease (Chamberlain, 1994). Intravenous administration of antineoplastic agents often does not produce therapeutic concentrations in the CSF because these agents do not adequately penetrate the blood-brain barrier. The administration of antineoplastic agents directly into the thecal sac allows delivery of high concentrations of these agents to the tumor sites while keeping systemic drug exposure low. Rational drug therapy by this route depends on a thorough understanding of the pharmacokinetic parameters of the agent after i.t. administration, but these parameters are poorly defined for most drugs. Elimination of agents from the CSF is unique because of the presence of the blood-brain barrier and the constant unidirectional flow of CSF from the choroid plexus to the arachnoid granulations. In this study, we sought to describe the pharmacokinetics of i.t. mAbs in rats and monkeys, compare the CSF clearance of IgG and IgM mAbs and reduce the clearance of IgG mAb by inhibition of the rate of CSF production and bulk flow.

We studied two anti-ganglioside mAbs that are potential candidates for i.t. therapy of neoplasia (Larson et al., 1995).

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Animals and antibodies. All studies in rats were conducted using female athymic "nude" (Rnu/nu) rats, 6 to 12 weeks of age and weighing 92 to 180 g, that were obtained from the National Cancer Institute (Frederick, MD). The same breed is being used in ongoing experiments of the clinical effectiveness of intrathecally administered anti-ganglioside mAbs in the treatment of leptomeningeal neoplastic xenografts. Juvenile male rhesus monkeys (Macaca mulatta) weighing 5.8 to 7 kg were obtained from the Department of Pediatric Otolaryngology, University of Pittsburgh School of Medicine, which had purchased them from the Buckshire Corp. (Perkasie, PA). These monkeys had been treated with topical ciprofloxacin for experimental infection of the middle ear but were entirely healthy when they underwent the experiments in this study. All animal studies were approved by the institutional animal research and care committee.

Pharmacokinetic studies. mAbs were injected and CSF samples were taken through spinal i.t. catheters in the rat and through CM puncture in the monkey (Bergman et al., 1997; Muraszko et al., 1993). Rats (n = 4 or 5 per experiment) received a mAb dose of 1 mg/kg, and monkeys (n = 4) received a mAb dose of 2 mg/monkey. mAb was administered within 1 min to the rats and over 10 to 15 min with gentle barbotage to the monkeys. Catheters were flushed with a volume of air equal to the dead space of the catheter. The i.v. injections and blood samplings were made through the femoral vein of both rats and monkeys. One group of 5 rats received 50 mg/kg acetazolamide and 50 mg/kg furosemide i.v. over 60 to 90 min before mAb infusion. One group of 6 rats received a continuous infusion of antibody at 1 µg/hr through an intraventricular cannula connected to a subcutaneously implanted pump (Bergman et al., 1997). CSF samples were withdrawn by CM puncture on days 3 and/or 5. The animals were awake and mobile. One group of 5 rats received IgG mAb intravenously at a dose of 1 mg/kg. mAb concentration in CSF and serum was determined by ELISA.

ELISA. mAbs 3F8 and 3A7 were assayed in the CSF by ELISA using an antibody capture immunoassay. Wells of a 96-well microtiter plate were coated with 100 ng of GD2 (G-0776; Sigma Chemical, St. Louis, MO). Samples were added to the wells and labeled with goat anti-mouse FAb peroxidase conjugate (A3682; Sigma) and then stained with the TMBLUE substrate (Intergen-CDP, Milford, MA). The reaction was quenched with 1 N H₂SO₄, and absorbance at 450 nM was read on a microplate autoreader (model EL-311; Bio Tek Instruments, Winooski, VT).

Data analysis. Pharmacokinetic data were calculated using a software package for the statistical analysis of general nonlinear models, PCNONLIN (SCI Software Consultants, Lexington, KY).

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Pharmacokinetics after i.v. administration of mAb. The i.v. administration of mAb 3F8 in 4 rats resulted in no detectable concentration of the mAb in the CSF. Elimination from the serum in 5 rats was fit best by a single exponential curve with mean t_1/2 of 106.5 min (S.D., 27.6). The mean clearance was 0.19 ml/min (S.D., 0.12), and the mean volume of distribution was 183.5 ml/kg (S.D., 73.1).

Pharmacokinetics after i.t. administration of mAb. Elimination of the mAbs from the CSF of rats and monkeys is illustrated in figures 1 and 2, and the pharmacokinetic parameters are summarized in table 1. Elimination of both IgG and IgM mAb in the rat after i.t. administration was fit best by a biexponential curve. The IgM mAb had a significantly higher AUC (P = .03), significantly slower clearance (P = .04) and a longer half-life in the CSF that did not quite reach statistical significance (P = .07) compared with the IgG mAb. The administration of acetazolamide and furosemide before IgG mAb infusion resulted in a significant increase in the AUC (P = .01), decrease in the clearance (P = .03) and lengthening of half-life (P = .04). The ratio of CSF AUC to serum AUC was significantly higher after i.t. administration of the IgM than the IgG mAb (P = .03) and significantly higher after i.t. administration of the IgG mAb with i.v. acetazolamide and furosemide than the IgG mAb alone (P = .03).

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Mean concentration of mAb in CSF after i.t. administration of 1 mg/kg of either IgG mAb alone, IgM mAb alone or IgG mAb with i.v. acetazolamide and furosemide.

View larger version (11K): [in this window] [in a new window]   Fig. 2.   CSF concentration of IgG mAb 3F8 after i.t. administration of 2 mg to 4 separate monkeys.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

This study has demonstrated that mAbs were eliminated rapidly from the CSF despite being very large and polar. The clearance of IgM mAb 3A7 (molecular mass of 950 kDa) from rat CSF averaged 4.1 µl/min, indicating clearance by bulk flow of CSF. Bulk flow is equal to CSF production rate, which in the rat has been estimated to be 2.1 to 5.4 µl/min (Davson and Segal, 1996). Bulk flow of CSF through the arachnoid granulations transports all substances within the CSF directly into the dural venous sinus blood. The rate of bulk flow forms a minimum rate of clearance of any molecule in the CSF because the effective pore size of the arachnoid granulation is 7.5 µm (Collins, 1983; Pardridge, 1991; Welch and Pollay, 1961). Therefore, the measured disposition half-life of IgM mAb of 137.5 min represents the longest t_1/2 from rat CSF attainable after i.t. administration of any compound. Calculations based on published CSF production rates indicate that 78 to 182 min represents the theoretical maximum elimination half-life for any drug from the CSF of diverse mammalian species, and 182 min is the theoretical maximum in humans (Davson and Segal, 1996).

Clearance of the IgG mAb 3F8 (molecular mass of 150 kDa) from the CSF of both rat and monkey was significantly greater than bulk CSF flow. Clearance in the rat was 10.8 µl/min compared with a CSF production rate of 2.1 to 5.4 µl/min, and clearance from the monkey was 151 µl/min compared with a CSF production rate of 28 to 32 µl/min (Lux and Fenstermacher, 1975). Previous studies of elimination of IgG mAbs and other macromolecules from the CSF of monkeys have also demonstrated that the rate of elimination was faster than the bulk flow of CSF. An anti-CD7 IgG2A mAb linked to deglycosylated castor bean ricin A chain exhibited a mean ventricular half-life of 99 min, volume of distribution of 6.9 ml and clearance of 41.7 µl/min after intraventricular injection to rhesus monkeys (Hertler et al., 1989; Muraszko et al., 1993). An anti-human transferrin receptor IgG1 mAb coupled to recombinant ricin A chain exhibited an early-phase half-life of 84 min and a late-phase half-life of 10.9 hr, volume of distribution of 10.1 ml and clearance of 73 µl/min (Muraszko et al., 1993). The clearance of Escherichia coli L-asparaginase of 33.3 µl/min was found to be significantly greater than the clearance of the extracellular marker ¹¹¹In-DTPA (Riccardi et al., 1981). Intraventricular administration of -interferon yielded a clearance of 50 µl/min (Collins et al., 1990).

The route of elimination of IgG mAbs from the CSF separate from bulk flow through the arachnoid granulations is unknown. IgG antibodies are large polar protein macromolecules that do not diffuse across the tight junctions between the brain endothelial cells or choroid plexus epithelial cells that make up the blood-brain and blood-CSF barriers (Betz et al., 1989). They diffuse slowly and penetrate poorly from the CSF to the interstitium of the brain (Betz et al., 1989). They are neither attached to, taken up by nor metabolized by neurons or astrocytes. Active endocytosis and transcytosis through the choroid plexus epithelial cells represent one possible mechanism of elimination. Saunders (1992) summarized data indicating that plasma proteins were transferred to CSF by transcytosis through epithelial cells of the choroid plexus. Different proteins of similar size appeared to be transported across different areas of the choroid plexus and attained different steady-state CSF-to-plasma ratios, suggesting that there were specific receptor-mediated transfer mechanisms for proteins across the choroid plexus. Balin and Broadwell (1988) demonstrated that horseradish peroxidase conjugated to a lectin, wheat germ agglutin, administered into the ventricular CSF of mice was bound by the epithelial cell surface and transported intact through the cell to the capillary surface of the choroid plexus within 10 min after intraventricular injection. Similar mechanisms of active endocytosis and transcytosis may operate to eliminate intrathecally administered mAbs from the CSF and suggest that the antigen specificity of the antibody as well as the species of origin, antibody class and isotype may influence the fractional rate of clearance that is not due to CSF bulk flow. This may explain in part the differences in clearances for mAb obtained in our study, which used an IgG3 anti-ganglioside mAb; Muraszko et al. (1993) used an IgG1 anti-transferrin mAb, and Hertler et al. (1989) used an IgG2a anti-CD7 mAb.

This study demonstrated that elimination of IgG mAb from rat CSF was inhibited by prior i.v. administration of acetazolamide and furosemide. This resulted in a more than doubling of AUC of IgG in the CSF, a 58% reduction in the clearance and a 48% increase in half-life. Acetazolamide and furosemide are not known to interfere with transcellular transport, and their effectiveness in prolonging elimination of mAb was almost certainly due to inhibition of CSF production and bulk flow. Previous work has demonstrated that i.v. administration of either acetazolamide or furosemide results in 50% inhibition of the rate of CSF formation (Johanson et al., 1994; Davson and Segal, 1996; McCarthy and Reed, 1974). Both drugs interfere with the ability of choroid epithelium to transport chloride ions from blood to CSF but probably act via different mechanisms, and additive effects may be possible. Harnish and Samuel (1988) were able to increase the CSF concentration of sodium diatrizoate, an iodinated contrast medium, 2 hr after i.v. administration to rats by pretreatment with acetazolamide at 50 mg/kg.

Our pharmacokinetic data after the administration of i.v. IgG mAb 3F8 demonstrated that this mAb distributed in a space about the size of extracellular fluid volume of the rat, 183 ml/kg. Clearance from serum during the initial 360 min was 0.19 ml/min, which was much slower than hepatic blood flow in the rat (3.7 ml/min) (National Research Council, 1956), indicating low capacity of the liver to eliminate these antibodies. On the other hand, clearance of mAb from serum (0.19 ml/min) was 10-fold greater than clearance from CSF (0.011 ml/min). This higher clearance from serum and the much larger volume of distribution after i.v. than i.t. administration will account for the low serum concentrations of mAb observed after i.t. administration of mAb. The short disposition t_1/2 of 106.5 min for mAb in this study may be due to the limited sampling duration (6 hr) and description of the kinetics by a monoexponential process.

Radiographic imaging in the rat illustrated that both intraventricular and lumbar i.t. administration of 0.1 ml of Iohexol (Omnipaque; Sterling Pharmaceuticals Inc., Barceloneta, Puerto Rico) produced dispersion of the dye throughout most of the entire SAS within 1 min, but neither method immediately filled the entire CSF space (data not shown). The ventricles were well filled by intraventricular administration, but the lower thoracic and lumbosacral SASs were not. The cisternal spaces of the brain and cervical and thoracic SASs were well filled by i.t. injection, but the ventricles and lumbosacral SASs were not. The calculated volume of distribution of the central compartment in the rat studies was therefore of necessity less than the total volume of the CSF space, which has been reported to be 250 to 290 µl (Bass and Lundborg, 1973; Davson and Segal, 1996; Saunders, 1992). On the other hand, the CSF volume in the monkeys calculated after cisternal injection and barbotage of IgG was 12 ml, which compared very closely with published values of 12 to 13 ml (DiChiro et al., 1985; Muraszko et al., 1993).

The validity of our pharmacokinetic data was supported in several ways: (1) the congruence of the measurements of mAb concentration in the CSF and serum, (2) the ability to predict steady-state CSF concentrations after continuous infusion of mAb and (3) the similarity of IgG half-life in the CSF of rats and monkeys, as discussed below. (1) Slower clearance of the IgM mAb than the IgG mAb from the CSF of the rat was accompanied by a lower mean concentration and AUC of the IgM mAb in the serum. Slower clearance of the IgG mAb from the CSF after administration of acetazolamide and furosemide was also accompanied by a lower mean serum concentration and AUC. (2) Clearance calculated from the i.t. studies was used to predict steady-state mAb concentration after intraventricular infusion of 1 µg/hr mAb. The experimental values of 1.2 µg/ml on day 3 and 1.1 µg/ml on day 4 correspond very closely to the predicted CSF concentrations of 1.5 µg/ml. (3) Elimination of unmodified mAb from CSF of rats showed a t_1/2 of 58.5 min compared with t_1/2 of 53.8 min in rhesus monkeys. CSF production as a percentage of total CSF volume does not vary appreciably among the warm-blood species (Davson and Segal, 1996), thereby accounting for a similar rate of elimination by bulk flow of CSF.

Several findings require explanation. Elimination was monoexponential in the monkeys and biexponential in the rats because the i.t. injection of mAb was given slowly over 10 to 15 min in the monkey and rapidly over 1 min in the rat. CSF sampling time did not allow characterization of the initial distribution phase in the monkeys. We found that elimination of an IgG3 mAb from monkey CSF had a t_1/2 of 53.8 min, which closely matched the early phase half-life of 1.4 hr reported by Muraszko et al. (1993) for an IgG1 mAb-ricin A chain conjugate. In both studies, initial peak concentrations declined 50- to 100-fold within the first 6 to 10 hr after i.t. administration. However, Muraszko et al. (1993) took CSF samples over a prolonged period from 1 to 24 hr and found a late-phase half-life of 10.9 hr. This probably represented slow release of immunoconjugate from brain tissue and did not reflect removal by the eliminating organs. This observation is similar to what is known regarding slow release of gentamicin from kidney tissue contributing to a prolonged terminal t_1/2. The concentrations of immunoconjugate in the CSF during this terminal phase were <1 µg/ml and contributed only a very small amount to the AUC.

The V_c in the rat was 0.23 ml after i.t. administration of IgG mAb and 0.13 ml after i.t. administration of IgM mAb. Part of the difference is attributable to the smaller size of the animals used for the IgM experiment. The mean weight of the animals in the IgM experiment was 118 g compared with a mean weight of 153 g for the animals used in the IgG experiment. In addition, microaggregation of the large IgM molecules may have led to incomplete mixing of the IgM mAb throughout the CSF space. The V_c of intrathecally administered IgG mAb in the rats pretreated with acetazolamide and furosemide was 0.19 ml compared with 0.23 ml in rats who were not given these drugs. Both acetazolamide and furosemide are diuretics, and their ability to cause volume depletion probably contributed to the lower V_c.

In summary, the pharmacokinetic results of this study support the development of mAbs for i.t. therapy. The therapeutic index of these agents as reflected by the ratio of AUC CSF to AUC blood showed that i.t. administration of an IgG mAb produced an AUC in CSF that was 93 times greater than the AUC in blood. This difference was magnified 4.7-fold by prior administration of acetazolamide and furosemide, suggesting that enhancement of the therapeutic index might be possible through administration of these agents. Experiments are ongoing to test whether mAbs administered intrathecally will reach their target tissues in sufficient concentrations to be clinically effective.