Abstract
The azole antifungal drug posaconazole caused phospholipidosis in neurons of the central nervous system, dorsal root ganglia of the spinal cord, and myenteric plexus in chronic toxicity studies in dogs. The time of onset, light and electron microscopic features, neurologic and electrophysiologic effects on the central and peripheral nervous systems, and potential for regression were investigated in a series of studies with a duration of up to one year. Nuclei of the medulla oblongata were the prominently affected areas of the brain. Neurons contained cytoplasmic vacuoles with concentrically whorled plasma membrane-like material (i.e., multilamellar bodies) morphologically identical to that commonly caused in other tissues by cationic amphiphilic drugs. Some axons in the brain and spinal cord were swollen and contained granular eosinophilic, electron-dense lysosomes. There were no features suggesting degeneration or necrosis of neurons or any associated elements of nervous tissue. The earliest and most consistent onset was in neurons of dorsal root ganglia. The observed neural phospholipidosis did not result in any alteration in the amplitude or latency of the auditory, visual, or somatosensory evoked potentials. The histopathologic changes did not progress or regress within the three-month postdose period. The results indicate that phospholipidosis can be induced in central and peripheral neurons of dogs by administration of posaconazole, but this change is not associated with functional effects in the systems evaluated.
Introduction
Posaconazole is an azole antifungal drug that is active against a wide spectrum of fungal species, including Aspergillus, Candida, Coccidioides, and many other unusual fungal pathogens. The structure is given in Figure 1. Although the bulk of the molecule is hydrophobic, the OH-moiety at the 2-hydroxypropyl on the right side of the 1,2,4-triazol makes it amphiphilic. In common with at least fifty other cationic amphiphilic drugs (CADs), posaconazole causes phospholipidosis in animals. This effect is common to the class of azole antifungal drugs, which includes ketoconazole, fluconazole, itraconazole, and voriconazole.
Phospholipidosis is characterized by the formation of intra-cytoplasmic lysosomal vacuoles containing whorls of material derived from cell membrane. The most commonly affected cell types include monocytes, histiocytes of lymphoreticular organs, and hepatocytes. Phospholipidosis in the brain and peripheral ganglia is unusual and has been reported only for azithromycin (Summary Basis of Approval, 2001), chlorphentermine, iprindole, 1-chloro-amitriptyline (Lüllmann-Rauch 1976), perhexiline maleate (Jung and Suzuki 1978), and fen-fluramine (Dastur et al. 1985), and in some sites in the brain, chloroquine and quinacrine (Frisch and Lüllmann-Rauch 1980). Phospholipidosis in the brains of rats treated with chloroquine and quinacrine occurs in areas where the blood–brain barrier is incomplete (Frisch and Lüllmann-Rauch 1980), suggesting that cells of the nervous system may not have an intrinsic resistance to CAD-induced phospholipidosis, but rather are protected from drug concentrations sufficient to induce it.
Cationic amphiphilic drugs are lipophilic and typically accumulate in high concentration in the affected lysosomes. Proposed mechanisms for the production of phospholipidosis include inhibition of the lysosomal phospholipases that break down cell membrane material during the normal process of membrane turnover, or interference with the process of membrane breakdown caused by the presence of drug molecules in the membrane (Jung and Suzuki 1978; Reasor and Kacew 2001).
Deficiencies of organ or cell function have not been commonly associated with the occurrence of phospholipidosis. Function has been most commonly studied with respect to lung and the phospholipid-laden macrophages that accumulate in alveoli, but studies of other leukocyte types and liver show no consistent or convincing functional deficit (Reasor and Kacew 2001). It has been suggested that aminoglycoside-induced renal phospholipidosis is responsible for the nephrotoxicity caused by these drugs (Mingeot-LeClercq and Tulkens 1999), but this relationship is disputed (Reasor and Kacew 2001).
Perhaps because the drugs that cause neuronal phospholipidosis are few, and most of the reported studies were done in rats, there are no reports of detailed assessments of nervous system function in animals with drug-induced phospholipidosis. We herein describe phospholipidosis, caused by posaconazole, of neurons in the brain, dorsal root ganglia (DRG), and myenteric plexus in dogs. The occurrence of this change was dose and time dependent and was associated with no change in neurologic function as measured by an extensive battery of physical and electrophysiologic techniques.
Methods
In two separate studies, beagle dogs, four per sex per group, were dosed daily for six months (Study No. 1) or one year (Study No. 2) with 0, 3, 10, or 30 mg/kg posaconazole, suspended in 0.4% w/v hydroxypropylmethylcellulose, by oral gavage. Dogs were housed in individual cages in temperature-and humidity-controlled rooms. Water was available ad libitum. Food was given one hour before dosing, to take advantage of the positive food effect on absorption of posaconazole. In the one-year study, the treatment schedule included initial dosing at 3 mg/kg for one week, followed by escalation to 10 mg/kg for one week for the mid- and high-dose groups, followed by escalation to 30 mg/kg after one additional week for the high-dose group. This dose-escalation sequence prevents the onset of a potentially fatal coagulopathy syndrome that occurs in dogs if doses of 30 mg/kg or higher are given without previous administration of lower doses (data not presented).
In Study No. 3, beagle dogs, initially fifteen per sex per group, were similarly dosed with 0 or 30 mg/kg for up to one year. The dose-escalation sequence described above was used to reach the dose of 30 mg/kg. The dogs were housed in individual runs. Twice prior to the start of the dosing period; after three, six, nine, and twelve months of dosing; and after a one-or three-month postdose recovery period, detailed neurologic examinations and electrophysiologic measurements were conducted on all survivors. Two dogs per sex per group were necropsied at each measurement interval, except at the end of the recovery period, when five per sex were necropsied.
Dogs were euthanized at the end of the appropriate dosing or recovery periods by an overdose of intravenous pentobarbital. In all studies, brain, spinal cord, and samples of major organs were fixed in 10% neutral buffered formalin. In the third study, the entire spinal cord with DRG was preserved. Eyes were fixed in Davidson’s solution. Tissues were embedded in paraffin, sectioned at 5–6 μm, and stained with hematoxylin and eosin (H&E) by standard techniques. Selected sections of neural tissues were stained with luxol fast blue–periodic acid Schiff, Holmes silver stain, or histochemical stain for glial fibrillary acidic protein (GFAP). Formalin-fixed samples of the thalamus and cuneate nucleus of the medulla were processed by standard methods for electron microscopy. Micrographs from selected treated and control dogs were evaluated from the first one-year study.
Neurologic and Electrophysiologic Measurements
Detailed neurologic assessments were conducted in the third study (one year, doses of 0 or 30 mg/kg). Neurologic examinations were done on all dogs before the start of the dosing period, and on all remaining dogs during weeks 14, 27, 40, 53, 58, and 67 (i.e., before each sacrifice interval). Neurologic observations included mentation, behavior, posture, attitude, menace response, pupillary light reflexes, oculocephalic reflexes, globe position, facial sensation, facial movement, masticatory muscle mass and tone, gag reflex, tongue movement, flexion and reflexion of limbs, and anal and tail tone.
A battery of electrophysiologic measurements was recorded in all dogs on two occasions prior to the onset of dosing and in all surviving dogs during weeks 12/13, 27/28, 39/40, 53, 59, and 67. Data were recorded using subdermal, platinum-needle electrodes (Grass-Telefactor, West Warwick, RI) positioned with respect to bony landmarks, and a dedicated MP100 system, version 3.5.7 (BioPac Systems, Inc., Santa Barbara, CA). The system includes all necessary amplifiers and stimulating and processing modules. Throughout the recording period, dogs were lightly anesthetized (Telazol, 13 mg/kg) and maintained in a prone position on a water-perfused heating pad.
Electrophysiologic measures of central nervous system (CNS) function included:
Brainstem auditory evoked potentials (BAEP), driven by monaural 80 db SPL click stimuli, latency of components 1 and 5. Visual evoked potentials (VEP), recorded at the cortex and the b-wave of the electroretinogram (ERG) evoked by a binocular high-intensity, full-field stroboscopic stimulation. Somatosensory evoked potentials (SEP), recorded overlying the spinal cord and cortex driven by supra-maximal stimulation of the contralateral median nerve. Nerve conduction in the distal segment of the peroneal nerve. Amplitude of the compound muscle action potential (CMAP) of the extensor digitorum brevis muscle of the foot. Minimal onset latency of the long-loop peroneal F-wave response.
Electrophysiologic measures of peripheral nervous system (PNS) function included:
The principal analysis of electrophysiological data was a cross-sectional analysis of variance (ANOVA) at each time point. Additional analysis included assessment of possible treatment-related differences in longitudinal trends. Alpha level was set at 0.05.
The use of animals in these studies followed the guidelines in the National Research Council Guide for the Care and Use of Laboratory Animals (1996) and the Animal Welfare Act. The protocols were approved by the Institutional Animal Care and Use Committee of the Schering-Plough Research Institute.
Results
Morphologic Findings
In all three studies, foamy cytoplasmic vacuoles morphologically typical of phospholipidosis were present in neurons of the brain, DRG, and myenteric plexus of the small intestine. Neurons in the myenteric plexus were affected at all doses (3, 10, and 30 mg/kg), DRG at ≥10 mg/kg, and brain at 30 mg/kg. In the diencephalon, vacuolated neurons were present in the dorsal and ventral thalamus, as well as the hypothalamus (Figure 2). Vacuolated neurons were also observed in the gracilis nucleus and the medial and lateral cuneate nucleus of the medulla oblongata. The distribution of affected areas was bilaterally symmetrical. In H&E-stained sections, the cytoplasm of neurons contained prominent clear vacuoles. Cell nuclei were not remarkable but were occasionally displaced peripherally toward the cell membrane by the vacuoles. By electron microscopy (Figure 3), the cytoplasmic vacuoles were characterized by electron-dense, laminated whorls within a monolayer lysosomal membrane (i.e., multilamellar bodies). The variably sized lysosomes were often filled and expanded by the accumulated membrane material. Some lysosomes contained a dark core of membrane-bound vacuoles indicative of lipofuscin granules. Other neuronal cellular structures and organelles were normal by light and electron microscopy.
Occasional axons were swollen by finely granular eosinophilic material (Figure 4), which, upon ultrastructural examination (Figure 5), contained electron dense membrane-bound bodies interpreted to be lysosomes. Affected axons were most common in the cuneate nucleus region of the medulla oblongata.
Swollen axons were seen, but infrequently, in the spinal cord; the most commonly affected locations were the dorsal nucleus gracilis, medial cuneate nucleus, and fasciculus cuneatus. The axons of the fasciculus cuneatus terminate in the cuneate nucleus of the caudal medulla oblongata. Therefore, the presence of swollen axons in the fasciculus cuneatus may reflect the changes in neuronal cell bodies of the cuneate nucleus.
No changes indicative of neuronal degeneration or necrosis occurred. There were no changes in the surrounding neuropil or associated structures to indicate cellular degeneration or necrosis of any cell type. Inflammatory changes were not observed. Occasional small foci of increased GFAP staining occurred in association with clusters of swollen axons.
If sensitivity is estimated by time of onset, the neuronal cell bodies in the DRG were more sensitive than brain or myenteric plexus to the posaconazole-induced neuronal phospholipidosis, as indicated by the uniform occurrence of phospholipidosis in these ganglia after three months of treatment (Figure 6a and 6b), before neurons in other sites were affected. However, if sensitivity is estimated according to dose, phospholipidosis occurred at the lowest dose, 3 mg/kg, in the myenteric plexus. The incidence of neuronal phospholipidosis in the brain, DRG, and myenteric plexus by duration of dosing at 30 mg/kg is given in Table 1.
The phospholipidosis in tissues of the nervous system did not regress during the three-month recovery period in Study No. 3.
Phospholipidosis in Other Organs
Neurons were not uniquely affected with phospholipidosis after administration of posaconazole. Cytoplasmic vacuolation typical of phospholipidosis was present in the tissues commonly affected following exposure to CADs, including histiocytes in the lymph nodes (Figure 6c), spleen, thymus, and pulmonary alveoli; and Kupffer cells of the liver. No changes occurred in the eye.
Neurologic Examinations
All CNS and PNS electrophysiologic measures were within normal limits. There was no evidence of significant differences in mean values for any measure in the treated versus control animals at each time point, and there were no significant differences in longitudinal patterns across treatment groups. The results for selected representative parameters are shown in Table 2.
Premature Decedents
A total of four dogs died in Study No. 3: one control (gavage error) and three treated dogs. In dogs given posaconazole, the most common cause of death was focal cerebral vasculitis with associated hemorrhage and inflammation. The vasculitis may have been related to the coagulopathy syndrome that necessitated staged escalation of the dose in the first two weeks of the study. However, the pathogenesis of the focal vasculitis observed only in the brain is unknown. Of particular interest, there was no evidence of phospholipidosis in the cerebral endothelium, and the pattern of neuronal phospholipidosis in dogs that died was similar to that in survivors. There was no evidence that phospholipidosis contributed to the cause of death. The premature decedents are not included in Table 1 so as to keep the time of reported onset of the phospholipidosis-related changes consistent.
In multiple-dose short- and long-term studies in monkeys and mice, posaconazole caused phospholipidosis in the lung and lymphoid organs, but not in the brain, DRG, myenteric plexus, or other nervous tissue. In the long-term carcinogenicity study of posaconazole in rats, enlarged axons were present in the gracilis and cuneate nuclei, but no neuronal vacuolation occurred (data not shown).
Discussion
Phospholipidosis is commonly encountered in animal toxicity studies of pharmaceutical agents (Halliwell 1997), but there is little or no evidence that it causes deleterious effects in the organ systems in which function has been studied (Reasor and Kacew 2001). The occurrence in dogs of posaconazole-induced phospholipidosis in neurons of the brain and DRG has provided a model in which possible functional effects in the nervous system could be assessed. Consistent with the lack of effect in other organs, we detected no functional impairment in the systems assessed related to the chronic occurrence of phospholipidosis in neurons of the brain, DRG, and myenteric plexus. Electrophysiologic measurements of the central and peripheral nervous system revealed no effect on function at any interval up to three months after the cessation of one year of drug treatment, even though phospholipidosis in the DRG was seen after three months, and phospholipidosis in neurons of the brain was seen within six months of the beginning of treatment.
A key finding in this study is the lack of change in the evaluated functional measures (i.e., sensory evoked potentials, peripheral nerve conduction velocity) in spite of the clear presence of cytoplasmic vacuoles typical of phospholipidosis in several regions of the brain and DRG. The battery of electrophysiologic measures in the present study was selected because of its demonstrated reliability, validity, and sensitivity to compound-induced changes in neural function (Arezzo, Simpson et al. 1985; Boyes et al. 2009; Herr et al. 2007; Spencer 2000). Sensory evoked potentials and peripheral nerve conduction studies have been used by the authors of the present study over the past three decades to demonstrated significant changes in CNS and PNS function associated with a variety of neurotoxic insults (Arezzo et al. 1980; Arezzo et al. 1982; Arezzo, Schaumburg et al., 1985; Kingman et al. 2005; Michalek et al. 2001; Pettersen et al. 2008; Schaumburg et al. 1989; Sweeney et al. 1993); exposure to chemotherapy agents and antiviral compounds (Anderson et al., 1994; Reser et al. 1999; Swain and Arezzo 2008; Warner 1995); induced inflammation (Brosnan et al. 1989; Martiney et al. 1990); induced vacuolation (Arezzo et al. 1989; Schroeder et al. 1992), demyelination (Arezzo et al. 1988; Giesser et al. 1987), vascular compromise (Horner et al. 2004; Portenoy et al. 1985); and genetic manipulations (Maycox et al. 1997; Michaelson et al. 1996). Specifically, the SEP and VEP latency measures reflect transmission within the heavily myelinated afferent fiber tracks of the CNS, including the dorsal columns of the spinal cord, the lemniscal pathways, the optic tracts, and the thalamocortical radiations. These measures are also sensitive to the integrity of intracortical processing within two distinct primary cortical regions. The subcomponents of the BAEP are a standard index of changes in the multisynaptic relay nuclei of the brainstem. We did not directly measure NCV in distal sensory nerves, which would be altered by functional pathology in the peripheral projection of the bipolar DRG cells. However, the latency of the cortical SEP, which is dependent on the central projection of the DRG cells, was unaffected by posaconazole.
There are brain regions and functions that are not measured by this battery, but the selected physiologic measures sample activity in the principal sensory pathways of the CNS and in several of the brain regions with evidence of neuronal phospholipidosis associated with posaconazole, including the thalamus dorsal column nuclei. The most reasonable explanation for the observed findings is that although present, the level and distribution of neural phospholipidosis induced by posaconazole is not sufficient to interfere with key aspects of normal function, such as axoplasmic transport, excitation thresholds, or transmembrane and paranodal ion segregation. Further, the observed swelling and accumulation of lysosomes in some cells may not have induced structural deficits such as demyelination or change in mean cross-sectional diameter of axons. It is noteworthy that there was no evidence of neural degeneration, apoptosis, or necrosis in any CNS or PNS site evaluated.
Although phospholipidosis in lymphoreticular tissues and liver are caused by many drugs, phospholipidosis in neurons is unusual, probably the result of exclusion of the drug from the nervous tissue by the blood–brain barrier. Other amphiphilic drugs are known to cause phospholipidosis in regions of the central nervous system that have an incomplete blood–brain barrier (Drenkhahn and Lüllmann-Rauch 1979). In experimental studies with rats, amiodarone-induced inclusions were found only in areas of the nervous system lacking a blood–brain or blood–nerve barrier (Santoro et al. 1992). The present investigations are also consistent with this pattern. In dogs treated with posaconazole, the neuronal phospholipidosis developed in the DRG before the brain. This vulnerability likely exists because many of the capillaries in the vicinity of the DRG lack specialized endothelial tight junctions, and thus the blood–nerve barrier is relatively ineffective in this region (Arezzo, Schaumburg et al. 1985). Thus, the bipolar sensory neurons in the DRG are subjected to higher concentrations of drug than other parts of the nervous system, which suggests that the sensitivity of the DRG in our studies may be a reflection of the relatively permeable blood–brain barrier in these ganglia.
The program of toxicity evaluation of posaconazole also included studies of up to two years duration in rats and mice, and one year duration in monkeys. Only dogs develop neuronal phospholipidosis. The reasons for the species specificity are not understood. It cannot be explained by plasma pharmacokinetics. Compared to dogs, higher plasma levels of posaconazole were achieved in studies in rats and mice, without the production of neuronal phospholipidosis (data not presented). As assessed by whole-body autoradiography, brain levels of drug in rats are low but no quantitative data are available.
The results of our studies extend the body of evidence indicating phospholipidosis, at the level observed in the present study, has little or no functional toxicologic importance.
Footnotes
Acknowledgements
The authors wish to thank Mr. Kennard Lee for the outstanding histology and Mr. Richard Geissler for the excellent transmission electron microscopy support.