Intraventricular and Intrathecal Therapy

Morris D. Groves

Arthur D. Forman

Victor A. Levin

Tumor cells from almost any solid or hematologic malignancy can metastasize and come to rest in the meninges, the choroid plexus, or the ependymal lining of the CNS. When this occurs, because of the proximity of these structures to the cerebrospinal fluid (CSF) pathways, tumor cells can further disseminate throughout the CNS via CSF flow pathways. Owing to the anatomy that protects the CNS from many blood-borne toxins, it is a unique therapeutic challenge to effectively treat cancer in the meninges and the spinal fluid using systemically administered agents.

DRUG DELIVERY

The delivery of many anticancer drugs from the bloodstream to the brain¹^,² or CSF is quite restricted because of a variety of factors. Drug-related factors include a drug’s level of protein binding,³ its molecular weight,⁴ polarity, and lipid solubility.²^,³^,⁵ Physical factors include the architectural properties of the blood-brain barrier (BBB) and the blood cerebrospinal fluid barrier (BCSFB) such as the tight junctions between the endothelial cells of brain capillaries and the epithelial cells of the choroid plexuses. The tight junctions limit paracellular diffusion of polar compounds. Furthermore, the cells creating the BBB and the BCFSB can contain an enriched content of degradative enzymes.⁶ Finally, ATP-dependent pumps such as the P-glycoprotein system, multidrug resistance proteins (MRPs), and organic and inorganic ion transporters can mediate the efflux of some anticancer drugs away from the brain and the CSF into the blood.⁵^,⁷^,⁸^,⁹^,¹⁰^,¹¹

Table 8-1 CSF:Plasma (or Serum) Drug Ratios After Intravenous Administration in Rhesus Monkeys or Humans

Drug	CSF:Plasma Ratio	References
Triethylenethiophos-phoramide	1.0	¹³
Thiotepa	1.0	¹³
Tiazofurin	0.28	¹⁴
6-Mercaptopurine	0.27	¹⁵
5-Fluorouracil	0.155	¹⁶
Arabinosyl-5-azacytidine	0.15	¹⁷
Cytosine arabinoside	0.06-0.22	^18,19
Vincristine	ND-0.05	^18,20
Spiromustine	0.047	²¹
Cisplatin	0.029-0.03^a	^22,23
Carboplatin	0.028^a	²³
Oxaliplatin	0.02^a	²³
Etoposide 0.003-0.05	^b	^24,25
MTX	0.01-0.04	^26,27
Interferon-α	0.033	²⁸
Daunomycin	ND	¹⁸
L-Asparaginase	ND	¹⁸
Trastuzumab	0.0023-0.02^c	^29,30
Idarubicin	0.08-0.15 (two animals)	³¹
Idarubicinol (active metabolite)	0.019
Daunorubicin	0.04-0.12	³¹
Daunorubicinol	0.024 ± 0.019
Topotecan (lactone)	1.0^d	³²
Rituximab	0.001	³³
Hydroxyurea	0.24^e	³⁴
Cyclophosphamide	0.20(0.00-1.1)	³⁵
Ifosfamide	1.2 (0.4-1.6)^f
^aConcentration of active drug. ^bNonprotein-bound fractional ratio. ^cOne log increase in CSF levels for patients undergoing cranial irradiation or with neoplastic meningitis. ^dAble to achieve 1 ng/ml × 8 hr after plasma topotecan AUC values of 120-200 ng/ml/hr. ^e80 mg per kg oral dose. ^fLowest levels in patients receiving dexamethasone. ND, not detectable.

The BBB and the BCSFB are not identical. Data suggest a more relaxed tight junction architecture in the BCSFB, which correlates with differential diffusion capacities between the BBB and BCSFB.¹² Table 8.1 shows the CSF:plasma ratios for some of the drugs studied in humans and rhesus monkeys.
A CSF:plasma ratio below 0.05 signifies nonspecific leakage of drug. From the table, it can be seen that many drugs normally achieve CSF:plasma ratios below 0.05 in the setting of a normal BBB and BCSFB. If meningeal neoplasia occurs, sites of blood-CSF leakage develop, and drugs can more readily leak from blood into CSF.

CSF PHYSIOLOGY

An understanding of CSF physiology will help in understanding intraventricular and intrathecal (IT) therapies. The greatest proportion of CSF is actively secreted by the choroid plexuses of the lateral, third, and fourth ventricles. CSF is also produced, to a much lesser extent, from extrachoroidal sources that gain access to the CSF through the brain extracellular fluid (ECF). The CSF and the ECF are in dynamic equilibrium, with the ependymal cells serving as a sieve rather than as a barrier to exchange between the two. The ependymal cell cilia may have the further functions of reducing the impact of an unstirred layer at the interface of the CSF and the cell surface and of expediting the mixing of metabolites, catabolites, and drugs between the CSF and the ECF.

The hydrostatically driven flow of CSF is from the ventricles to the basal cisterns to the subarachnoid space and, ultimately, to the arachnoid granulations and into venous blood. This flow provides a strong diffusion gradient to lower ECF metabolites and catabolites being produced by the brain parenchymal cells. This capacity to lower catabolites produced by the cells has been referred to as the sink effect of the CSF. In addition to this CSF sink action, other functions that the CSF serves are

regulation of the brain extracellular environment,
movement of neuroendocrine hormones, and
buoyancy and protection of the brain from jarring associated with body movement.

PHARMACOKINETICS

The CSF compartments and the adjacent CNS ECF constitute a regional cavity for the purposes of drug administration. This cavity is not closed, and a drug administered into the CSF can leave through

bulk absorption through the arachnoid granulations,
diffusion into the ECF and surrounding cells,
diffusion into the ECF and then across capillaries into the systemic circulation, and
biotransformation.

Figure 8.1 K_out = CL_csf/V_csf (3) and was plotted against the [log (octanol/water partition coefficient) M_r^-1/2]. The data were fit by a nonlinear least-squared technique. These data were obtained from the literature.³^,⁹^,³⁷^,³⁸^,³⁹^,⁴⁰ ACNU, nimustine; Ara-C, cytarabine; AZQ, aziridinylbenzoquinone; DFMO, difluoromethylornithine DL-α-difluoromethylonithine; MGBG, methylglyoxal-bis-guanylhydrazone.

Aside from biotransformation rates, which vary depending on the structure of the drug, the clearance of drugs from the CSF will be somewhat predictable, based on molecular size and partitioning into lipid (e.g., octanol/water partition coefficient). Figure 8.1 shows the observed relationship for K_out versus log(octanol/water partition coefficient) M_r^-1/2 as described by Levin and Landahl.³⁶

Just as the relation between physical characteristics and K_out from the CSF is predictable, so is the distance that the drug in the CSF can penetrate into adjacent brain parenchyma. The factors that determine the effective diffusing distance from the CSF-brain interface are

the tortuosity of the ECF,
the molecular size of the permeant drug,
the cell/ECF partitioning of the drug, and
the stability of the drug.

Figure 8.2 shows the hypothetical relation between drug half-life (a term that in this context includes biotransformation and the loss of drug from the ECF to the blood through permeation across capillaries), diffusion coefficient in cm² per second, and the distance into brain parenchyma that 10% of the CSF drug level will permeate. This bar graph was computed assuming diffusion from a planar surface.

A number of therapeutic points emerge for consideration. Drugs administered intrathecally (spinal subarachnoid) will not reflux into the ventricular system unless a reversal of the normal hydrostatic gradient or hydrokinetic flow has occurred. Reversal of the hydrokinetic flow is observed in the hydrocephalus. This is important because it means that the IT route is inferior when the drug is required to reach the choroid plexuses or the intraventricular ependymal surface. Other reasons to avoid the IT route include
inadvertent drug injection into the subdural space instead of the CSF and the risk of local damage to the spinal cord or nerve roots because of high concentrations of drug at the injection site. Intraventricular drug administration with gentle barbotage four to six times quickly dilutes the drug in a large ventricular fluid volume and enhances drug egress from the ventricles to a half-time in the order of minutes,³⁶ although for most relatively small molecules such as those that are currently used, transit from the ventricles is quite rapid.

Figure 8.2 This bar graph shows the relations among free water diffusion coefficient (D), drug half-life, and the distance into brain parenchyma attainable by 10% of the CSF drug level.

Diffusion

Because drug diffusion into brain parenchyma is slow relative to its clearance from the CSF, the depth of effective drug concentration is small, and a therapeutic benefit for a deep parenchymal lesion, and by extrapolation, a nodular lesion on the meningeal surface, is unlikely (Fig. 8.2). Further compromising drug delivery to a nodular tumor in the CSF pathway is the fact that when drug diffuses into a nodular lesion, as soon as it comes into contact with a tumor capillary, it crosses into the systemic circulation and reduces the potential tumor drug level. Pharmacokinetic calculations suggest that nodules greater than 5 mm in diameter are inadequately treated by regional therapies.⁴¹^,⁴² Administration of drug into the CSF often fails because the drug does not make contact with tumor cells long enough to kill them. This is because

the tumor obliterates the subarachnoid space and the drug cannot reach many of the cells;
the tumor is nodular and only the most superficial cells make contact with the drug; and
the drug is too unstable or is small and nonionized, being cleared from the CSF before reaching distal sites in the CSF pathway.

Sometimes these shortcomings can only be overcome by increasing the administered dose at the risk of CNS toxicity or by providing a carrier, such as a monoclonal antibody, to slow the CSF clearance of the drug and to increase its targeting to tumor cells. To fully evaluate the potential of this last approach, further experimental study is required to ascertain optimal conditions for carrier-drug targeting.

Regional Delivery

Regional delivery of chemotherapeutic agents to the CNS had its earliest use in the treatment for infectious meningitis. Antibiotics were injected through a spinal needle through frontal burr holes into the frontal lobes and ventricles. In an effort to diminish the morbidity associated with this form of therapy, Ayub Ommaya developed an implantable closed delivery system while working at the National Institutes of Health.⁴³ Using this system, the ease with which therapeutic agents could circumvent the BBB opened new strategies for the treatment of CNS infections and malignancies. Although great advances have been made since the introduction of the indwelling closed ventricular delivery system, major problems still face clinicians who utilize these systems.

Refinements, such as valves and shunt attachments, have permitted greater flexibility in the use of reservoirs. Lumbar reservoirs have been used more frequently for the delivery of analgesics than for chemotherapy. Although some studies have reported good results in using lumbar reservoirs,⁴⁴ their high rate of malfunction⁴⁵ and the better distribution of drug when delivered into a ventricle⁴⁶ have made the intraventricular route favored for chemotherapy for CNS malignancy.

Following percutaneous lumbar administration of methotrexate (MTX), the drug at times fails to reach the head and leaks into the epidural or subdural space; this complication is more likely after repeated lumbar administration.⁴⁷ Even with successful lumbar administration, the intraventricular levels of MTX are highly variable, as indicated by a study measuring drug concentrations at both levels⁴⁶ (Fig. 8.3A). In that study, serum MTX levels were approximately 1 × 10^-3 those achieved in the CSF (Fig. 8.3B). More recent data suggest that, with respect to progression-free survival, intraventricular administration may be more important for agents with short, as compared with long, half-lives.⁴⁸

Given the cell cycle-specific action of MTX, administering that drug by a pulse technique may limit its effectiveness while increasing the chances of neurotoxicity, as toxicity seems to correlate with peak levels. One randomized study of children with leptomeningeal acute lymphocytic leukemia (ALL) compared standard 12 mg per m² therapy twice weekly with concentration time therapy.⁴⁹ By ventricular reservoir, 1 mg of MTX was administered every 12 hours for 6 doses. The dosage
was adjusted to achieve an MTX level of 5 ± 2 × 10^-7 in the lumbar spinal fluid before administration of the next dose of MTX. This course was repeated every 7 to 10 days until remission occurred. As with standard pulse therapy, the course was repeated during the consolidation and maintenance periods. Less neurotoxicity was seen in patients who received the “concentration × time” therapy than in those who received the standard pulse therapy, with no difference in efficacy between the two approaches.

Figure 8.3 A: Nine studies of intraventricular MTX concentration after lumbar administration. Solid circles represent a dose of 6.25 mg per m². B: MTX distribution (mean ± range) in five studies after intraventricular administration of 6.25 mg per m². (Reprinted from Shapiro W, Young D, Mehta B. Methotrexate distribution in cerebrospinal fluid after intravenous, ventricular and lumbar injections. N Engl J Med. 1975;293:161-166, with permission.)

Intraventricular catheters are relatively safe. The most important complications include catheter malposition or obstruction (3.8% to 6.6%)⁴⁵^,⁵⁰ or reservoir exposure (1.6%),⁵⁰ and catheter-related infections (3.8% to 7.5%).⁴⁵^,⁵⁰ Most infections are controllable with antibiotics. Infected reservoirs can be successfully managed with antibiotic therapy without removal of the hardware,⁵¹ although in some cases, reservoir removal is required just as it is in other foreign body infections. Skin bacteria are the most frequent cause of infection, but scrupulous attention to technique will lessen the incidence of this complication.⁵¹ An additional benefit of the reservoirs is the avoidance of repeated lumbar punctures that are painful and inconvenient.

Whether the intralumbar or intraventricular reservoir administration route is used, it is critical to establish the patency of the CSF pathways, as blockages are common in leptomeningeal malignancy. Local compromise of CSF flow may leave some areas untouched by the administered therapeutic agents and may increase local toxicity if CSF circulation is impaired. Radionuclide imaging with ¹¹¹indium-albumin accurately assesses spinal fluid flow. Focal radiation therapy can often ease areas of blockage once they have been identified. Focal CSF blocks that are unrelieved by radiotherapy portend a worse prognosis.⁵²^,⁵³

INTRAVENTRICULAR CHEMOTHERAPY— BACKGROUND

Intraventricular therapy is used to treat established leptomeningeal malignancy as well as for prophylaxis of tumors with a high incidence of CNS invasion, such as ALL. The earliest widespread use of direct CNS therapy was for childhood acute lymphocytic leukemia (c-ALL). In the late 1960s, when advances in chemotherapy for c-ALL increased life expectancy in those patients from weeks to years, the prospects of long-term survival were diminished by an incidence of c-ALL relapse in the CNS as high as 75%.⁵⁴ Often following this relapse, these patients would also develop a c-ALL relapse in bone marrow, which frequently proved fatal.⁵⁵

The BBB and the BCSFB prevent adequate levels of most systemically administered cytotoxic drugs from reaching the leukemic cells present in the CNS. Leukemia can invade the CNS from the systemic circulation and may reside in a quiescent state and become evident once treatment for systemic tumor is discontinued.⁵⁶ Strategies to treat the CNS have used irradiation and more recently intensified systemic chemotherapy and intraventricular therapy for high-risk patients and have succeeded in decreasing the rate of relapse in the CNS to less than 5% of patients treated as well as in prolonging systemic remission.⁵⁷^,⁵⁸

Prophylaxis: Childhood Acute Lymphocytic Leukemia

Based on the first successful CNS prophylactic program developed by St. Jude’s Cancer Research Hospital using cranial irradiation and IT MTX (dosing based on childhood body surface that lags behind CSF volume [Fig. 8.4]), the CNS relapse rate in average-risk c-ALL patients fell to between 2% and 5% and the relapse rate in high-risk patients to 6%.⁵⁹ To reduce radiation-related effects (cognitive and endocrine impairments and secondary malignancies), XRT doses were reduced from 24 to 18 and then to 12 Gy without loss in tumor control.³⁷^,⁶⁰^,⁶¹ A meta-analysis of 2,848 patients revealed that XRT could be replaced by long-term IT therapy without detriment to event free survival or overall survival.³⁸ Another review of 5,564 children with c-ALL enrolled in trials from 1985 through 2001 confirmed that front-line radiotherapy no longer has a role in preventing CNS relapses.⁵⁸ Furthermore, single-agent IT MTX is as effective as, and possibly less neurotoxic than, multidrug IT regimens.³⁹^,⁴⁰^,⁶¹

Risk factors for meningeal relapse are high WBC count, T-cell ALL, Ph-positive ALL, a traumatic lumbar puncture with blasts seen, CSF with ≥5 WBC per µl and blasts present, and infant age.⁶²^,⁶³^,⁶⁴^,⁶⁵ These factors allow for some tailoring of therapy. Recently, a risk-stratified approach where intensification of the systemic chemotherapy and the addition of more frequent IT chemotherapy are administered in a graded fashion, and cranial irradiation is excluded from prophylaxis in contrast to prior standards.⁵⁸^,⁶⁶^,⁶⁷^,⁶⁸ Further supporting this approach, a recent analysis of T-cell ALL (high-risk) children revealed that high-dose systemic MTX (8 g per m²) versus cranial radiation demonstrated no compromise in survival and avoidance of the sequelae of cranial radiotherapy.⁶⁹

Only gold members can continue reading. Log In or Register to continue