Ototoxicity of Carboplatin Delivered Locally in a Monkey Brainstem

Abstract

Ototoxicity is a common side effect of platinum-based chemotherapy. Intratumoral drug delivery theoretically could reduce the ototoxic effects of systemic drug infusions. However, local delivery to central nervous system (CNS) tumors might promote ototoxicity through drug release into cerebrospinal fluid (CSF). This report describes an examination of the cytoarchitecture of vestibular cells of cynomolgus monkeys that had chronic brainstem infusions with the maximum tolerated dose (MTD) of carboplatin. The brainstems of adult monkeys were infused for 30 days at 0.42 μl/h with 0.025 to 0.25 mg/kg (MTD) of carboplatin. The vestibular sensory epithelia of eight drug-treated animals were isolated for microscopic examination of vestibular hair cells and support cells. Local infusions produced chronic elevated CSF levels of platinum, neurological symptoms, and radiographic evidence of pontine injury. Histology revealed significant cell damage at the infusion sites. Microscopic examinations of vestibular support cells and hair cells demonstrate a small reduction in cell counts in the drug-treated monkeys compared to a noninfused control animal. Parametric and nonparametric tests show no effect of dose in predicting the vestibular cell counts. In this single study of eight monkeys, a dose-dependent reduction of vestibular hair cells or support cells was not observed in animals infused with brainstem infusions of 0.025 to 0.25 mg/kg of carboplatin.

Systemic chemotherapy with cisplatin and dose-intensive carboplatin is associated with a high incidence of irreversible hearing loss (Freilich et al. 1996). Hearing loss also has been noted in patients treated with a regimen of carboplatin (200 mg/m²) with blood-brain barrier disruption (Williams et al. 1995). Several strategies are being investigated to mitigate ototoxicity and other side effects. These include the use of targeted drugs, formula-based dosing, the simultaneous administration of chemoprotective drugs (Muldoon et al. 2001), and local delivery. Local methods, which include bolus and chronic intratumoral drug infusions, have demonstrated safety and efficacy in model studies and clinical trials (Olivi et al. 1996). There is a clinical impression that local delivery has reduced nontarget effects such as myelosuppression and renal failure. However, this has not been systematically evaluated, and there is little information about infusion-associated neurotoxicity or ototoxicity.

Neuro-oncologists at our Institution treated 13 cynomolgus monkeys with brainstem infusions of carboplatin to obtain pre-clinical data leading to new therapies for pontine gliomas (Storm et al. 2003). Clinical symptoms of neurotoxicity were dose-dependent. Magnetic resonance imaging (MRI) demonstrated dose-dependent edema at the point of the infusion. Postmortem histology revealed large fields of tissue necrosis consistent with the edema seen on MRI (Storm et al. 2003). Hematology parameters were normal at all dose levels tested, including doses that produced rapidly progressive neurotoxicity (Strege et al. 2004).

Eight of the animals in the safety study were available for examination of vestibular support cells and hair cells. Although the number of animals was small, we hypothesized that increasing carboplatin doses would lead to a significant decreased count of support and hair cells related to increased cerebrospinal (CSF) levels of drug. CSF levels of platinum consistently were 10-fold greater than blood levels, and by week 2 of the 4-week infusion, CSF platinum levels measured 15 to 89 μg/L. Yet, there appeared to be little evidence of a relationship between vestibular hair cell and support cell counts and dose. Although infusions into normal parenchyma may not reliably predict the patterns of drug distribution in abnormal tissue (tumors), the present study suggests that local therapy with carboplatin may mitigate the ototoxicity commonly found with dose-intensive systemic therapy.

MATERIALS AND METHODS

Animals

Adult, 4- to 10-year-old male cynomolgus monkeys (Macaca fascicularis), weighing 3 to 9 kg, were obtained from Biological Research Farms (Houston, TX). Animals were quarantined until found to be in good health by the veterinary staff. Monkeys were housed in individual cages with enrichment devices and free access to primate pelleted chow and Baltimore City water. Procedures were conducted under Johns Hopkins Animal Care and Use Committee regulations and in accord with Federal and Maryland codes. The number of animals in each dose group is shown in Table 1.

Infusions

Five percent dextrose solutions of carboplatin (Paraplatin) were delivered with 1 ml DUROS infusion pumps calibrated to deliver solution at 10 μl/day (0.42 μl/h) when fully hydrated at 37°C. Nickel-titanium infusion catheters (outer diameter, 0.3 mm; inner diameter, 0.2 mm) were connected to the pump by 30 cm of silicon tubing. A 2.25 cm long blunt tipped catheter having 10 pores (0.025 mm diameter) 0.5 to 3 mm from the distal tip was used to infuse animals. Pumps, tubing, and catheters were obtained from DURECT (Cupertino, CA). Tubing and catheter were empty on an implant to allow the tissue to seal 24 to 30 h before the drug solution was infused (Storm et al. 2003).

Surgery

The surgical process has been described (Storm et al. 2003). Briefly, after induction of anesthesia, a burr hole was made in the midline on the occipital bone 2.5 cm below the inion. The dura was opened and the edges were coagulated by bipolar cautery. The infusion catheter was then inserted free hand to a pontine target of 1.75 mm anteroposterior, −12.5 mm dorsoventral, and 0 mm mediolateral in a stereotaxic atlas (Szabo and Cowan 1984). The tubing was secured to the bone with cyanoacrylate. The body of the pump was placed in a subcutaneous pocket in the low-cervical/high-thoracic region and connected to the catheter by the silicon tubing.

Drug Levels

The MTD was estimated from studies of local infusions in Fischer rats (Storm et al. 2003). Two animals received 10% (0.025 mg/kg) of the MTD; three received 30% (0.075 mg/kg); five received 100% (0.25 mg/kg); and three received 300% (0.75 mg/kg) of the estimated MTD. Two monkeys in the 100% MTD group were euthanized on completion of the 30-day infusion. Three monkeys in the 100% group were monitored for 13 to 14 months after the infusion was stopped. One monkey in the 10% group, three in the 30% group, and four in the 100% MTD group were available for this study. The animals in the latter group included the three monkeys who had undergone long-term surveillance and one monkey that was euthanized after drug delivery. The control monkey was from a placebo-control group in a drug trial.

Clinical and Laboratory Evaluation

Animals were observed twice daily for signs of postoperative morbidity, including gait, interaction with care takers, grooming habits, and food intake. Temperature and weight were recorded at 2 to 4-week intervals. Serial complete blood counts and laboratory tests were obtained from serum and EDTA anticoagulated blood under ketamine anesthesia. CSF was obtained under ketamine anesthesia for platinum assays and for cultures to rule out infection as a source of neurotoxicity. Blood and CSF platinum levels were measured by inductively coupled plasma/mass spectrometry by National Medical Services (Willow Grove, PA). The lower limit of platinum detection was 2 μg/L. The low calibrator was 0.05 μg/L; the coefficient of variation for control plasma containing 0.4 μg/L was 7.8% measured over 4 weeks. Blood in two monkeys and CSF in one were analyzed to verify the absence of platinum before infusions.

Radiographic Studies

A postoperative noncontrast computed tomography (CT) scan (Toshiba Aquilion), 1-mm slices, was used to determine placement of the catheter. The animals had MRI scans (GE Signa 1.5 Cardiac LX): sagittal T1 pre and post contrast, axial T1 pre and post contrast (gadolinium), and axial T2 approximately 2 weeks after surgery and at intervals thereafter as described in Results.

Histology

Carboplatin-infused monkeys were sedated with ketamine and pentobarbital to achieve a lethal dose. The monkeys were perfused through the left ventricle into the aorta after clamping the descending aorta. The perfusion solutions consisted of 1 L of ice cold saline followed by 500 ml of freshly prepared, ice cold, 4% paraformaldehyde solution containing 1% acrolein, 1% picric acid, and 5% sucrose in 0.1 M phosphate buffer, pH, 7.4. The control specimens were obtained from a non-infused monkey from a placebo group in an unrelated vaccination trial. NeuroScience Associates (Knoxville, TN) cut serial coronal sections 5-μ thick at 960-μ intervals through the brainstem and cerebellum. Adjacent sections were stained for histopathology and analyzed for tissue platinum levels. Histology included hematoxylin and eosin (HE), silver stain, glial fibrillary acidic protein, and Perls stain. Tissue platinum levels were measured by atomic absorption spectroscopy (Strege et al. 2004).

For ototoxicity evaluations, the semicircular canalsensory epithelia were dissected, washed in phosphate-buffered saline (PBS), osmicated, dehydrated in graded ethanol solutions, and embedded in Epon (Polysciences, Inc. Warrenton, PA) resin. Two micron thick cross sections were taken for quantification of hair cell nuclei and support cell nuclei. Sections were mounted and stained with Richardson’s stain, photomicrographs were made, and the nuclei of types I and II hair cells and nuclei of support cells were identified and tabulated. The analysts were blinded to the amount of drug the animal received.

Two adjacent sections from the mid-portion of each semicircular canal crista were selected for examination because the central zone of the crista is most susceptible to ototoxic damage (Lindeman 1969). The number of type I and type II vestibular hair cells within these sections were estimated using the dissector technique in order to avoid over counting hair cells in sections that were thinner than the hair cell nuclei diameter (Gundersen 1986; Lysakowski and Goldberg 1997). Briefly, light photomicrographs from two consecutive sections from the midportion of the crista were superimposed using Adobe Photoshop 6.0 (Adobe Systems, Inc., San Jose, CA). Only the hair cell nuclei present in the first section but not seen in the second section were counted. The process was then repeated with the sections in reverse order. Using the dissector technique on serial sections of 2 μm thickness ensures that no double-counting of hair cells occurs because their nuclei have a diameter greater than 2 μm.

Linear densities of cells were calculated by dividing these counts by the length of the basement membrane of the neuroepithelium. Support cells were identified by their polygonal nuclei that stained darkly and homogeneously except for one to several clumps of chromatin. Support cell nuclei were located low in the epithelium, adjacent to the basement membrane. Hair cells were identified by their lighter-staining elliptical nuclei with only one nucleolus per nucleus. Hair cell nuclei were located well above the basement membrane. Type I hair cells had a flask shape, constricted neck, and were surrounded by minimally stained calyceal endings. Their nuclei tended to be round and had several clumps of chromatin. Type II hair cells assumed a cylindrical or barrel shape without a constricted neck and were contacted by bouton endings or the outer faces of calyceal endings. Their nuclei had a more homogeneous distribution of chromatin than did those of type I hair cells, and type II nuclei tended to be close to the surface of the neuroepithelium because these hair cells were shorter. Sections from the central crista of primates are predominantly populated with type I hair cells (Fernandez, Lysakowski, Goldberg 1995).

Statistics

To determine if there were differences between the densities of type I and II hair cells or support cells among the controls and treatment-dose groups, the nonparametric Kruskal-Wallis test was used (Minitab version 11.21; Minitab, Inc. State College, PA).

RESULTS

Clinical reports from 13 monkeys in the preclinical trial of brainstem carboplatin infusions have been published (Strege et al. 2004). They are briefly summarized in Table 1. Although the trial was designed to infuse increasing doses of carboplatin for 30 days, the infusions at 300% MDT group were stopped at days 4 to 11 due to rapidly progressive neurotoxicity and loss of consciousness in two of three monkeys. Neurotoxicity was the predominant toxicity observed in animals receiving more than 100 μg of drug. Signs of myelosuppression and renal toxicity were absent with the exception of an elevated blood urea nitrogen test in one animal in the 300% group (Strege et al. 2004). Five monkeys received the estimated MTD of 0.25 mg/kg. Three animals in this group were given 1 mg/kg of oral dexamethasone daily. Steroids had no effect on the development of ataxia, which was apparent by day 25 in all animals of this group. The three steroid-treated animals were held for 1 year to monitor long-term effects. Two of these animals recovered normal behavior approximately 2 months after the infusion was stopped. The third animal in the group improved markedly but still retained a blunted effect 1 year off-drug.

Postmortem histology revealed little or no tissue damage in the control animals (no drug infusion) and the monkeys infused with 10% (0.025 mg/kg) of the MTD, Figure 1A and B, respectively. The black arrows illustrates the extensive injury in the HE stained serial sections surrounding the catheter in the animals receiving more than 100 μg of drug, e.g., the animals in the 30% (Figure 1C) and 100% dose groups (Figure 1D).

The eight drug-treated monkeys available for the present analysis of vestibular hair cells and support cells consisted of one animal in the 10% group, three in the 30% group, and four from the 100% MTD group. The incident adverse event and tissue levels of platinum are listed in Table 2. With the exception of the monkey in the 10% dose group, the animals in this group displayed notable clinical and radiographic evidence of neurotoxicity. By the approximate midpoint in the 30-day infusion cycle, CSF platinum levels of six of the seven monkeys displaying neurotoxicity ranged from 30 to 89 μg/L of platinum. The day-16 CSF sample for animal 5 in Table 2 was lost. The day-30 CSF platinum level for this animal was 64 μg/L. Thus, seven of the eight animals in this group had significant CSF drug exposures for at least 2 weeks. Monkeys 6 to 8 in Table 2 received 1 mg/kg of steroids during the infusion and were autopsied 1 year after their infusions were withdrawn. The remaining five animals were euthanized at end of the drug infusion period.

Table 3 summarizes the densities of vestibular hair cells and support cells in the semicircular canal cristae from the eight monkeys in Table 2. For a control sample, the right and left ear sections were isolated from a noninfused cynomolgus monkey. The panels in Figure 2 show representative sections from the midportions of semicircular canal cristae for the control animal (Figure 2A, no infusion), animal 1 (Figure 2B, 10% MTD), animal 2 (Figure 2C, 0.3 × MTD), and animal 5 (Figure 2D, 1.0 × MTD), respectively. The relatively intact morphology shown in Figure 2A and 2B are consonant with the intact brainstem histology shown in Figure 1A and B. However, the relative normal morphology seen in Figure 2C and 2D are remarkable in comparison to the companion Figure 1C and 1D from the same monkeys. The two control cristae had average (± standard deviation, SD) densities of 5.98 ± 0.09 hair cells per 100 μm² and 10.20 ± 1.89 support cells per 100 μm². The treated epithelia had somewhat lower hair cell densities, ranging from 2.82 per 100 μm² in one of the 10% MTD animals to 4.33 in one of the 30% MTD animals. These hair cell densities in the treated epithelia were significantly lower than in the controls (p = .045; Mann-Whitney test). However, because of the large variability in the hair cell densities in the treated specimens, there was not a significant effect of drug dose in predicting hair cell density among the specimens from treated animals (p = .19; Kruskal-Wallis test). There was no significant difference in the density of support cells between control and treated cristae (p = .19; Mann-Whitney test).

DISCUSSION

Recent pediatric oncology trials of carboplatin have demonstrated that ototoxicity does not appear to be a significant risk factor in patients treated systemically with standard doses of carboplatin (Bertolini et al. 2004). However, efforts to maximize therapeutic response via dose intensification remain challenged by significant non-target effects including ototoxicity (Dubs et al. 2004). The results shown in Table 3 suggest that chronic intratumoral infusions may provide a mechanism for dose-intensification with limited side effects.

The present study found that despite marked, irreversible clinical neurotoxicity and large areas of brainstem tissue necrosis in animals dosed at 0.075 and 0.25 mg/kg of carboplatin, vestibular hair cell, and support cell loss was relatively limited. Figures 1 and 2 illustrate the contrast between brainstem tissue damage and the relatively normal vestibular cell structure in the animals examined in this report. The necrosis evident in Figure 1C and D was coincident with the tissue-bound platinum levels, which may have been responsible for some of the tissue loss during the preparation of the sections (Strege et al. 2004). Thus, direct comparison of hair and support cell loss with tissue damage is not possible.

The results in the present study are consonant with our report that central auditory neurons appeared to be intact after carboplatin was infused into the brainstem. The average hearing threshold was elevated 8.8 dB (SD = 7.3 dB) in monkeys 6 to 8 (Table 3) 6 months after the infusion was stopped. The average threshold elevation increased to 10.7 dB 1 year after drug treatment, a loss that became statistically significant (p < .05) (May et al. 2005).

Several factors limit the interpretation of this study and its application to clinical trials. The number of animals studied was small and the analysis was performed retrospectively. The control animal was not infused, which might be expected to cause us to underestimate control values of vestibular and support cells. However, comparison with published values for hair cell density in squirrel monkey shows that our control values from cynomolgous are quite comparable (Lysakowski and Goldberg 1997). Cell counts in monkeys 6 to 8 were included in the analyses even though these animals were off-drug for more than 12 months. These animals were included because there is no evidence for regeneration of vestibular hair cells lost via platinum-based chemotherapy. The average hair and support cell counts from these three animals in the 0.25 mg/kg was lower compared to the counts from the animal in the same dose group that was examined immediately after the infusion was stopped.

Although the CSF levels of platinum demonstrate significant interstitial flow of locally infused drug, platinum is a surrogate marker and over estimates the concentration of the intact drug. It is not possible from this data to accurately determine CSF drug levels that lead to ototoxicity. Model studies have shown that carboplatin-induced ototoxicity appears to be related to oxidative stress to the cochlea and inner hair cell loss (Husain et al. 2003). Carboplatin is a selective inner hair cell toxin whereas cisplatin is selective outer hair cell toxin (Bauer and Brozoski 2005). It may be valuable to study the effects of locally infused carboplatin and cisplatin in rats and chinchilla models for which there is significant background information.

Studies of drug interactions in normal tissues may not predict outcomes in abnormal (tumor) tissues. For example, studies of chronically infused small and large molecular weight drugs in rat brains have shown notable differences between normal and tumor models (Khan et al. 2005). Chemotherapy infusions in tumors with abnormal membranes may lead to greater CSF levels of drug and greater exposures of the drug to the vestibular cells.