:Nerve Growth Factor Prevents Experimental Cisplatin Neuropathy Stuart C. Apfel, MD," Joseph C. Arezzo, PhD,'? LeeAnne Lipson,* and John A. Kessler, MD't
Cisplatin is a widely used antitumor agent, the dose-limiting toxicity of which is predominantly large-fiber sensory neuropathy. Prevention of such a neuropathy would extend the usefulness of this agent, allowing higher doses and longer periods of treatment. We report here that we have successfully established cisplatin neuropathy in mice measured by using behavioral, biochemical, and electrophysiological techniques, and that subcutaneous administration of human recombinant nerve growth factor (NGF)prevents or delays the neuropathy. Cisplatin administration reduced sensory ganglion levels of the peptide transmitter, calcitonin gene-related peptide, slowed nerve conduction in the tail and impaired proprioception as measured by the ability to balance on a rotating dowel. NGF coadministration appeared to prevent all these abnormalities. Treatment of the human toxic neuropathy with its well-established time of onset, simple clinical course, and the accessibility of nerve to NGF administered systemically may provide the best clinical setting for the first human trials of NGF. Apfel SC, Arezzo JC, Lipson LA, Kessler JA. Nerve growth factor prevents experimental cisplatin neuropathy. Ann Neurol 1992;31:76-80
Few specific treatments are available to prevent neuronal dysfunction or death after injury to the nervous system. The recognition that neuronotrophic growth factors maintain survival and function of damaged neurons, however, suggests that such trophic agents may be useful in the treatment of a variety of neurological diseases. Because administration of growth factors to the central nervous system presents a number of technical problems, the first clinical applications may be in the treatment of peripheral neuropathy. Some of the most devastating peripheral neuropathies occur as a complication of chemotherapeutic and antiviral drugs, and the development of neuropathy may limit the dose of such agents. For example, the widely used chemotherapeutic agent., cisplatin, is a drug of choice for a number of neoplasms including bladder, testicular, and ovarian tumors I1 1. The dose of cisplatin, however, is often limited by the development of a largely irreversible toxic neuropathy that affects predominantly largefiber sensory neurons and that impairs vibratory and proprioceptive perception 111. The defined time of onset, rapid time course, and severity of this neuropathy make it a good candidate for the initial evaluation of growth factor therapy. Establishment of a reliable animal model was therefore essential to test the possible therapeutic benefits of such treatments. Studies of experimental toxic neuropathies ha.ve traditionally used primarily morphological and electro-
physiological techniques to evaluate peripheral nervous system function. Recently, however, we reported an animal model of the toxic sensory neuropathy induced by taxol, another antitumor drug, for which biochemical and behavioral parameters proved sensitive for detecting early neurotoxicity {2]. Taxol causes a painful, small-fiber peripheral polyneuropathy in humans. To evaluate that neuropathy, levels of the nociceptive peptide neurotransmitter, substance P, were measured in sensory ganglia, and thermal pain threshold was determined using a tail-flick test. Because cisplatin involves large-fiber proprioceptive neurons, other indices would be expected to be more sensitive. For example, the putative peptide transmitter, calcitonin gene-related peptide, is widely distributed in dorsal root ganglion (DRG) sensory neurons and is therefore likely present in proprioceptive neurons. In this study, we have used this biochemical measurement in association with a behavioral measure of proprioception and electrophysiological evaluation to examine the effects of nerve growth factor (NGF) treatment on cisplatin neuropathy in mice. N G F is a polypeptide growth factor, necessary for the development of D R G sensory neurons and for the maintenance of normal ganglion function in adults. The factor regulates the function of a large number of subpopulations of D R G neurons but not all the sensory neurons. This study reports that N G F treatment also
From thc Departments of "Neurology and tNeuroscienct , Albert Einstein College of Medicine. Bronx, NY 10461.
Address correspondence to Dr Apfel, Room 329, Kennedy Center, 1410 Pelham Parkway South, Albert Einstein College of Medicine, Bronx' NY 104G1'
Received Apr 10, 1991, and in revised form Jun 20. Accepted for publication Jun 26, 1091.
76 Copyright 0 I992 by the American Neurological Association
prevents t h e biochemical, behavioral, and electrophysiological manifestations of cisplatin-induced sensory neuropathy in mice.
Methods Drug Administration Three groups of 10 C D 1 male mice (10-15 gm at the onset) were injected with cisplatin (Sigma Chemical, St Louis, MO) (10 mgikg of body weight intraperitoneally) once a week for 8 consecutive weeks. T h e cisplatin was formulated at 0.2 mg/ ml in normal saline and was prepared immediately before injecting the mice. Two of these groups also received subcutaneous injections of recombinant human N G F (Genentech, South San Francisco, CA) at doses of 5 pg/gm or 1 pg/gm, respectively, three times a week. The third group received saline instead of NGF. A fourth group of control mice received intraperitoneal and subcutaneous saline injections at the same volumes proportionate to weight.
Behavioral Testing One week after the final cisplatin dose, the mice underwent behavioral testing. Two behavioral tests were conducted to detect abnormalities of both thermal pain threshold and proprioception. T o measure thermal pain threshold, we used the modified tail-flick test that we described in detail in a prior publication [2]. In brief, the mouse was restrained with its tail immersed in a beaker of water. The water temperature was gradually increased, and the threshold temperature at which the mouse flicked its tail was recorded. Proprioceptive testing was performed using a modified rotating dowel test. Our apparatus consisted of a rod, 45 cm long and 1 cm in diameter, which was turned by a motor at a steady rate of 12 rpm. The mice were initially tested in the light for their ability to balance on the rotating bar for a maximum of 15 seconds. In all cases, the mice were able to stand on the bar for the full 15 seconds of the test, thus establishing the absence of weakness that would prevent them from standing on the bar. The test was then repeated in the dark with the mice placed on the rod immediately after the lights were turned off. Each mouse was given two trials with a maximum time of 15 seconds per trial. The trial with the longest time on the bar was recorded. Initially, these tests were performed unblinded. Subsequently, they were repeated blinded with identical results.
E lectrophysiological Testing At this time, the mice also underwent electrophysiological testing. Compound sensory amplitudes and distal latencies were measured in the caudal nerve of the mouse's tail as previously reported 121. Each mouse was anesthetized with halothane before recording. The tails were restrained and surface electrodes (platinum-iridium needles) were placed along the distal section of the caudal nerve. The active recording electrode was positioned at a fixed distance of 40 mm distal to the stimulating cathode. Brief pulses of constant voltage stimulation were delivered through an anode-cathode pair positioned overlying a proximal section of the caudal nerve. Five to ten stimuli were averaged, and the procedure was repeated. Latency was determined from the onset of the initial depolarizarion and was measured to the nearest 0.1
msec. Amplitude was measured from the baseline to peak and was measured to the nearest 0.1 pV.Rectal temperature of each mouse was monitored and maintained within 0.5"C; there were no group differences in core temperature at baseline or after treatment.
Biochemical Measurementj O n the day after the behavioral and electrophysiological testing, the mice were killed, and the C-6 cervical D R G were removed. The ganglia were placed individually in polypropylene microfuge tubes with 125 p1 of 0.01 N HCI, and homogenized with a mechanical homogenizer. The tubes were then placed in a heating block at 100°C for 10 minutes. The tissue was vacuum dried and assayed for levels of calcitonin generelated peptide (CGRP) by radioimmunoassay. The assay buffer consisted of 50 mM N a P 0 4 buffer, p H 7.4, 10 mM ethylenediamine tetraacetic acid (EDTA), and 0.3% bovine serum albumin. A standard curve was set up using commercial CGRP (Sigma). CGRP antibody (kindly provided by D r Ronald Lindsay, Regeneron Pharmaceuticals, Tarrytown, NY) was used at a 1:200,000 dilution. Dried samples were reconstituted in 300 ~1 of assay buffer and one-third was aliquoted for each sample tube. '"I-Labeled CGRP (Amersham, Arlington Heights, IL) was added at a concentration of 40,000 dpm/ml to each tube. The assay was allowed to incubate at 4°C for 5 days. Separation was performed using activated charcoal and dextran suspended in a separation buffer consisting of 50 mM NaPO,, p H 7.4, 0.25% gelatin, and 10 mM EDTA. The supernatant and pellets were separated and counted using a y-counter. Percent bound was calculated as counts bound divided by total counts times 100. All biochemical assays were done blinded.
Statistics Data were analyzed using an analysis of variance (ANOVA) or Student's t-test when applicable.
Results Behavioral Testing Clinically, cisplatin causes a predominantly large-fiber sensory neuropathy affecting t h e perception of vibration and position sense more than pain and temperature. To determine if our mouse model of cisplatin neuropathy was analogous to t h e clinical situation, we tested mice for behavioral evidence of both abnormal thermal pain sensation and abnormal proprioceptive sensation. To test thermal sensation, we measured t h e tail-flick temperature threshold. The control group of mice had a mean temperature threshold of 45.0 k 0.38"C (SEM) (Fig 1). Cisplatin administration resulted i n a small b u t significant elevation of the mean temperature threshold t o 46.2 -+ 0.31"C (p < 0.03, by A N O V A ) . Although this increase was statistically significant, it was m u c h smaller than t h e 10°C elevation seen with taxol administration [2). T r e a t m e n t with NGF ( 5 kg/gm) resulted in a mean temperature threshold of 45.2 -+ 0.43"C, thus preventing the cisplatin-induced elevation. To test proprioception, we measured t h e ability of
Apfel et al: Experimental Cisplatin Neuropathy
77
1.850
I*
V >-
w z t-
4
1.750
K t
0
In
W z In
2
47 7-
J
1.650
I CONTROL
CISPLATIN
CISPLATIN NGF
+
CONTROL
Fig I . Efects qf cisplatin administration on thl, tail-flick response. Afrer 8 ujeeks of risplatin administratioa, the temperatuw at ichiih the taiL-flick response u'as elicited lexpressed as mean tenipeiwture "C t SEMI was dightl3J bnt signific-antb elemted. Coadniinijtration of nerve growth factor ( N G F ) 15 pgI p i prerentcd this change. n = I0 per group. "Diyferrfrom both other groups d t p < 0.0.3, by ana1yri.r- of iuriunce.
T
CONTROL
CISPLATIN iNGF
Fig 2. Effkts (if cispbtiri admini.rtration on peifornzance on the d w d test. After 8 u'eeb of cispiatin ardm'mini.rtration, the treated mil-e had a significantly reduced ability t o balance on the doulel (expvtwed as mean time on the dowel in seconds & SEMI. Coadniinistratiomof neme grou~thfactor (NGFi ( 5 pglgm) prevented this rhange. n = 10 per group. "Differ.\ from both other gr0up.r at p < O.005. bji analysis of variance.
mice to maintain their balance on a rotating rod, both in the presence and the absence of visual cues. All mice were able to maintain their balance on the rod for the full 15 seconds with the lights on. In the dark, control mice maintained their balance for a mean 13.6 ? 0.88 seconds (SEM) (Fig 2). The mean time for cisplatin1.39 treated mice was significantly reduced to 8.0 seconds @ < 0.005, by ANOVA). Coadministration of 5 pg/gm of N G F prevented this reduction with a mean time of 12.4 1.29 seconds, which did not differ significantly from control mice but was significantly different from the cisplatin group by the Duncan procedure (p < 0.05). In contrast, coadministration of 1 pg/ gm of N G F resulted in a mean time of 8.0 ? 2.07
*
*
78 Annals o f Neurology Vol 31
CISPLATIN NGF
+
Fig 3. Effects o j - ci-rplatin administration on the distd senso y latency in the caudal newe of the tail.Values are expressed a.r mean msec & S E M . Cisplatin administration signi$i.antly eleouted the distal latency. but roadminijtration of newe growth factor tNGF,i 15 pg/gmf pretmted this change. n = 10 per group. "0ifler.r fkim both other group,! at p
Cisplatin is a widely used antitumor agent, the dose-limiting toxicity of which is predominantly large-fiber sensory neuropathy. Prevention of such a neuropathy would extend the usefulness of this agent, allowing higher doses and longer periods of treatment. We report here that we have successfully established cisplatin neuropathy in mice measured by using behavioral, biochemical, and electrophysiological techniques, and that subcutaneous administration of human recombinant nerve growth factor (NGF)prevents or delays the neuropathy. Cisplatin administration reduced sensory ganglion levels of the peptide transmitter, calcitonin gene-related peptide, slowed nerve conduction in the tail and impaired proprioception as measured by the ability to balance on a rotating dowel. NGF coadministration appeared to prevent all these abnormalities. Treatment of the human toxic neuropathy with its well-established time of onset, simple clinical course, and the accessibility of nerve to NGF administered systemically may provide the best clinical setting for the first human trials of NGF. Apfel SC, Arezzo JC, Lipson LA, Kessler JA. Nerve growth factor prevents experimental cisplatin neuropathy. Ann Neurol 1992;31:76-80
Few specific treatments are available to prevent neuronal dysfunction or death after injury to the nervous system. The recognition that neuronotrophic growth factors maintain survival and function of damaged neurons, however, suggests that such trophic agents may be useful in the treatment of a variety of neurological diseases. Because administration of growth factors to the central nervous system presents a number of technical problems, the first clinical applications may be in the treatment of peripheral neuropathy. Some of the most devastating peripheral neuropathies occur as a complication of chemotherapeutic and antiviral drugs, and the development of neuropathy may limit the dose of such agents. For example, the widely used chemotherapeutic agent., cisplatin, is a drug of choice for a number of neoplasms including bladder, testicular, and ovarian tumors I1 1. The dose of cisplatin, however, is often limited by the development of a largely irreversible toxic neuropathy that affects predominantly largefiber sensory neurons and that impairs vibratory and proprioceptive perception 111. The defined time of onset, rapid time course, and severity of this neuropathy make it a good candidate for the initial evaluation of growth factor therapy. Establishment of a reliable animal model was therefore essential to test the possible therapeutic benefits of such treatments. Studies of experimental toxic neuropathies ha.ve traditionally used primarily morphological and electro-
physiological techniques to evaluate peripheral nervous system function. Recently, however, we reported an animal model of the toxic sensory neuropathy induced by taxol, another antitumor drug, for which biochemical and behavioral parameters proved sensitive for detecting early neurotoxicity {2]. Taxol causes a painful, small-fiber peripheral polyneuropathy in humans. To evaluate that neuropathy, levels of the nociceptive peptide neurotransmitter, substance P, were measured in sensory ganglia, and thermal pain threshold was determined using a tail-flick test. Because cisplatin involves large-fiber proprioceptive neurons, other indices would be expected to be more sensitive. For example, the putative peptide transmitter, calcitonin gene-related peptide, is widely distributed in dorsal root ganglion (DRG) sensory neurons and is therefore likely present in proprioceptive neurons. In this study, we have used this biochemical measurement in association with a behavioral measure of proprioception and electrophysiological evaluation to examine the effects of nerve growth factor (NGF) treatment on cisplatin neuropathy in mice. N G F is a polypeptide growth factor, necessary for the development of D R G sensory neurons and for the maintenance of normal ganglion function in adults. The factor regulates the function of a large number of subpopulations of D R G neurons but not all the sensory neurons. This study reports that N G F treatment also
From thc Departments of "Neurology and tNeuroscienct , Albert Einstein College of Medicine. Bronx, NY 10461.
Address correspondence to Dr Apfel, Room 329, Kennedy Center, 1410 Pelham Parkway South, Albert Einstein College of Medicine, Bronx' NY 104G1'
Received Apr 10, 1991, and in revised form Jun 20. Accepted for publication Jun 26, 1091.
76 Copyright 0 I992 by the American Neurological Association
prevents t h e biochemical, behavioral, and electrophysiological manifestations of cisplatin-induced sensory neuropathy in mice.
Methods Drug Administration Three groups of 10 C D 1 male mice (10-15 gm at the onset) were injected with cisplatin (Sigma Chemical, St Louis, MO) (10 mgikg of body weight intraperitoneally) once a week for 8 consecutive weeks. T h e cisplatin was formulated at 0.2 mg/ ml in normal saline and was prepared immediately before injecting the mice. Two of these groups also received subcutaneous injections of recombinant human N G F (Genentech, South San Francisco, CA) at doses of 5 pg/gm or 1 pg/gm, respectively, three times a week. The third group received saline instead of NGF. A fourth group of control mice received intraperitoneal and subcutaneous saline injections at the same volumes proportionate to weight.
Behavioral Testing One week after the final cisplatin dose, the mice underwent behavioral testing. Two behavioral tests were conducted to detect abnormalities of both thermal pain threshold and proprioception. T o measure thermal pain threshold, we used the modified tail-flick test that we described in detail in a prior publication [2]. In brief, the mouse was restrained with its tail immersed in a beaker of water. The water temperature was gradually increased, and the threshold temperature at which the mouse flicked its tail was recorded. Proprioceptive testing was performed using a modified rotating dowel test. Our apparatus consisted of a rod, 45 cm long and 1 cm in diameter, which was turned by a motor at a steady rate of 12 rpm. The mice were initially tested in the light for their ability to balance on the rotating bar for a maximum of 15 seconds. In all cases, the mice were able to stand on the bar for the full 15 seconds of the test, thus establishing the absence of weakness that would prevent them from standing on the bar. The test was then repeated in the dark with the mice placed on the rod immediately after the lights were turned off. Each mouse was given two trials with a maximum time of 15 seconds per trial. The trial with the longest time on the bar was recorded. Initially, these tests were performed unblinded. Subsequently, they were repeated blinded with identical results.
E lectrophysiological Testing At this time, the mice also underwent electrophysiological testing. Compound sensory amplitudes and distal latencies were measured in the caudal nerve of the mouse's tail as previously reported 121. Each mouse was anesthetized with halothane before recording. The tails were restrained and surface electrodes (platinum-iridium needles) were placed along the distal section of the caudal nerve. The active recording electrode was positioned at a fixed distance of 40 mm distal to the stimulating cathode. Brief pulses of constant voltage stimulation were delivered through an anode-cathode pair positioned overlying a proximal section of the caudal nerve. Five to ten stimuli were averaged, and the procedure was repeated. Latency was determined from the onset of the initial depolarizarion and was measured to the nearest 0.1
msec. Amplitude was measured from the baseline to peak and was measured to the nearest 0.1 pV.Rectal temperature of each mouse was monitored and maintained within 0.5"C; there were no group differences in core temperature at baseline or after treatment.
Biochemical Measurementj O n the day after the behavioral and electrophysiological testing, the mice were killed, and the C-6 cervical D R G were removed. The ganglia were placed individually in polypropylene microfuge tubes with 125 p1 of 0.01 N HCI, and homogenized with a mechanical homogenizer. The tubes were then placed in a heating block at 100°C for 10 minutes. The tissue was vacuum dried and assayed for levels of calcitonin generelated peptide (CGRP) by radioimmunoassay. The assay buffer consisted of 50 mM N a P 0 4 buffer, p H 7.4, 10 mM ethylenediamine tetraacetic acid (EDTA), and 0.3% bovine serum albumin. A standard curve was set up using commercial CGRP (Sigma). CGRP antibody (kindly provided by D r Ronald Lindsay, Regeneron Pharmaceuticals, Tarrytown, NY) was used at a 1:200,000 dilution. Dried samples were reconstituted in 300 ~1 of assay buffer and one-third was aliquoted for each sample tube. '"I-Labeled CGRP (Amersham, Arlington Heights, IL) was added at a concentration of 40,000 dpm/ml to each tube. The assay was allowed to incubate at 4°C for 5 days. Separation was performed using activated charcoal and dextran suspended in a separation buffer consisting of 50 mM NaPO,, p H 7.4, 0.25% gelatin, and 10 mM EDTA. The supernatant and pellets were separated and counted using a y-counter. Percent bound was calculated as counts bound divided by total counts times 100. All biochemical assays were done blinded.
Statistics Data were analyzed using an analysis of variance (ANOVA) or Student's t-test when applicable.
Results Behavioral Testing Clinically, cisplatin causes a predominantly large-fiber sensory neuropathy affecting t h e perception of vibration and position sense more than pain and temperature. To determine if our mouse model of cisplatin neuropathy was analogous to t h e clinical situation, we tested mice for behavioral evidence of both abnormal thermal pain sensation and abnormal proprioceptive sensation. To test thermal sensation, we measured t h e tail-flick temperature threshold. The control group of mice had a mean temperature threshold of 45.0 k 0.38"C (SEM) (Fig 1). Cisplatin administration resulted i n a small b u t significant elevation of the mean temperature threshold t o 46.2 -+ 0.31"C (p < 0.03, by A N O V A ) . Although this increase was statistically significant, it was m u c h smaller than t h e 10°C elevation seen with taxol administration [2). T r e a t m e n t with NGF ( 5 kg/gm) resulted in a mean temperature threshold of 45.2 -+ 0.43"C, thus preventing the cisplatin-induced elevation. To test proprioception, we measured t h e ability of
Apfel et al: Experimental Cisplatin Neuropathy
77
1.850
I*
V >-
w z t-
4
1.750
K t
0
In
W z In
2
47 7-
J
1.650
I CONTROL
CISPLATIN
CISPLATIN NGF
+
CONTROL
Fig I . Efects qf cisplatin administration on thl, tail-flick response. Afrer 8 ujeeks of risplatin administratioa, the temperatuw at ichiih the taiL-flick response u'as elicited lexpressed as mean tenipeiwture "C t SEMI was dightl3J bnt signific-antb elemted. Coadniinijtration of nerve growth factor ( N G F ) 15 pgI p i prerentcd this change. n = I0 per group. "Diyferrfrom both other groups d t p < 0.0.3, by ana1yri.r- of iuriunce.
T
CONTROL
CISPLATIN iNGF
Fig 2. Effkts (if cispbtiri admini.rtration on peifornzance on the d w d test. After 8 u'eeb of cispiatin ardm'mini.rtration, the treated mil-e had a significantly reduced ability t o balance on the doulel (expvtwed as mean time on the dowel in seconds & SEMI. Coadniinistratiomof neme grou~thfactor (NGFi ( 5 pglgm) prevented this rhange. n = 10 per group. "Differ.\ from both other gr0up.r at p < O.005. bji analysis of variance.
mice to maintain their balance on a rotating rod, both in the presence and the absence of visual cues. All mice were able to maintain their balance on the rod for the full 15 seconds with the lights on. In the dark, control mice maintained their balance for a mean 13.6 ? 0.88 seconds (SEM) (Fig 2). The mean time for cisplatin1.39 treated mice was significantly reduced to 8.0 seconds @ < 0.005, by ANOVA). Coadministration of 5 pg/gm of N G F prevented this reduction with a mean time of 12.4 1.29 seconds, which did not differ significantly from control mice but was significantly different from the cisplatin group by the Duncan procedure (p < 0.05). In contrast, coadministration of 1 pg/ gm of N G F resulted in a mean time of 8.0 ? 2.07
*
*
78 Annals o f Neurology Vol 31
CISPLATIN NGF
+
Fig 3. Effects o j - ci-rplatin administration on the distd senso y latency in the caudal newe of the tail.Values are expressed a.r mean msec & S E M . Cisplatin administration signi$i.antly eleouted the distal latency. but roadminijtration of newe growth factor tNGF,i 15 pg/gmf pretmted this change. n = 10 per group. "0ifler.r fkim both other group,! at p
