Comparison of Neuropathy-Inducing Effects of Eribulin Mesylate, Paclitaxel, and Ixabepilone in Mice

Published OnlineFirst April 15, 2011; DOI: 10.1158/0008-5472.CAN-10-4184

Cancer Therapeutics, Targets, and Chemical Biology Research

Krystyna M. Wozniak1, Kenichi Nomoto3, Rena G. Lapidus2, Ying Wu1, Valentina Carozzi5, Guido Cavaletti5, Kazuhiro Hayakawa6, Satoru Hosokawa6, Murray J. Towle3, Bruce A. Littlefield4, and Barbara S. Slusher1

Abstract Chemotherapy-induced neurotoxicity is a significant problem associated with successful treatment of many cancers. Tubulin is a well-established target of antineoplastic therapy; however, tubulin-targeting agents, such as paclitaxel and the newer epothilones, induce significant neurotoxicity. Eribulin mesylate, a novel microtubule- targeting analogue of the marine natural product halichondrin B, has recently shown antineoplastic activity, with relatively low incidence and severity of neuropathy, in metastatic breast cancer patients. The mechanism of chemotherapy-induced neuropathy is not well understood. One of the main underlying reasons is incomplete characterization of pathology of peripheral nerves from treated subjects, either from patients or preclinically from animals. The current study was conducted to directly compare, in mice, the neuropathy-inducing propensity of three drugs: paclitaxel, ixabepilone, and eribulin mesylate. Because these drugs have different potencies and pharmacokinetics, we compared them on the basis of a maximum tolerated dose (MTD). Effects of each drug on caudal and digital nerve conduction velocity, nerve amplitude, and sciatic nerve and dorsal root ganglion morphology at 0.25 Â MTD, 0.5 Â MTD, 0.75 Â MTD, and MTD were compared. Paclitaxel and ixabepilone, at their respective MTDs, produced significant deficits in caudal nerve conduction velocity, caudal amplitude and digital nerve amplitudes, as well as moderate to severe degenerative pathologic changes in dorsal root ganglia and sciatic nerve. In contrast, eribulin mesylate produced no significant deleterious effects on any nerve conduction parameter measured and caused milder, less frequent effects on morphology. Overall, our findings indicate that eribulin mesylate induces less neuropathy in mice than paclitaxel or ixabepilone at equivalent MTD-based doses. Cancer Res; 71(11); 1–11. 2011 AACR.

Introduction Breast cancer is the most common cause of cancer-related death among women in the Unites States. Many patients Peripheral neuropathy is a common dose-limiting toxicity progress from early-stage breast cancer to metastatic disease of many chemotherapeutic regimens. Chemotherapy-induced within short periods of time. Although several chemothera- peripheral neuropathy (CIPN) causes numerous debilitating peutic agents exist for metastatic disease, the overall prognosis symptoms, impairs functional capacity, and results in dose remains poor, with the 5-year survival rate approximating to reductions or possible cessation of chemotherapy. As a con- only 23% (1). sequence, effective chemotherapeutic regimens with a lower Paclitaxel is often administered as first-line therapy in propensity to induce neuropathy would be favored. breast cancer patients with metastatic disease, achieving overall response rates in the range of 30% to 60%, or 20% to 40% when used as second-line or salvage therapy (2). Taxanes are among the most effective antineoplastic agents Authors' Affiliations: 1Johns Hopkins Brain Science Institute; 2University of Maryland Greenebaum Cancer Center, Baltimore, Maryland; 3Eisai Inc., against many cancers, but their side effect of peripheral Andover, Massachusetts; 4Department of Biological Chemistry and Mole- neurotoxicity has limited their use (3, 4). Chronic neuropathic cular Pharmacology, Harvard Medical School, Boston, Massachusetts; pain affects between 20% and 50% of women after their breast 5University of Milan-Bicocca, Italy; and 6Eisai Co., Ltd., Tsukuba, Japan cancer treatment (5). Another chemotherapeutic agent admi- Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). nistered to metastatic breast cancer patients whose cancer is resistant or no longer responding to paclitaxel is the recently Corresponding Author: Krystyna M. Wozniak, Johns Hopkins Brain Science Institute, NeuroTranslational Drug Discovery Program, 855 North approved ixabepilone, which is a semisynthetic analogue of Wolfe Street, Baltimore, MD 21205. Phone: 443-287-0178; Fax: 410-614- epothilone B with antineoplastic activity against taxane-resis- 0659. E-mail: [email protected] or Barbara S. Slusher, Johns Hopkins – Brain Science Institute, NeuroTranslational Drug Discovery Program, 855 tant cell lines (6 8). Although ixabepilone may be useful for North Wolfe Street, Baltimore, MD 21205. Phone: 410-614-0662; E-mail: the treatment of locally advanced or metastatic breast cancer, [email protected] it induces neuropathy in up to 72% of patients (9). Both doi: 10.1158/0008-5472.CAN-10-4184 ixabepilone and paclitaxel are microtubule-stabilizing agents 2011 American Association for Cancer Research. that promote polymerization of microtubules (10). However,

www.aacrjournals.org OF1

Wozniak et al.

the precise mechanisms of paclitaxel- and ixabepilone- a final volume (as 80% of final) and the solution was main- induced neurotoxicities are not yet completely understood tained on ice during dosing. Dosing solutions were made (11–13). One of the reasons for this is the incomplete char- freshly every day and dosed in an administration volume of acterization of function and pathology of peripheral nerves in 10 mL/kg. chemotherapy-treated subjects, either from patients or pre- Ixabepilone was purchased as part of an IXEMPRA kit clinically from animals. Further studies in this area are, thus, (Bristol-Myers Squibb) for clinical administration. The ixabe- warranted. pilone solution was prepared according to instructions pro- Eribulin mesylate (previously E7389), another promising vided in the package insert. Basically, the kit consists of 2 vials, microtubule-targeting agent, is currently undergoing Food 1 containing 47 mg ixabepilone powder and the other contain- and Drug Administration (FDA) fast-track review. Recently, ing 23.5 mL diluent, which were stored in the refrigerator at positive results from randomized phase II and III trials of 4C. The total volume of the diluent was added to the total eribulin mesylate in patients with advanced metastatic breast amount of powder; therefore, after reconstitution, the con- cancer have been reported (14, 15). Eribulin mesylate is a centration of ixabepilone in the solution was 2 mg/mL. [The structurally simplified, macrocyclic, ketone analogue of hali- diluent supplied consists of a sterile nonpyrogenic solution of chondrin B, and it inhibits microtubule dynamics through a 52.8% (w/v) purified polyoxyethylated castor oil and 39.8% novel mechanism relative to other tubulin-targeting agents, (w/v) dehydrated alcohol.] The formulated ixabepilone stock including the taxanes, Vinca alkaloids, and epothilones. It solution (2 mg/mL) was immediately aliquoted and stored at binds with high affinity and specificity to positively charged À80C until use. On each experimental day, the stock solution ends of microtubules, thereby suppressing microtubule was diluted by adding 50% ethanol/50% cremophor with dynamics (16). Preclinically, eribulin mesylate has shown subsequent vortexing to yield a resultant solution that potent anticancer activity in vitro and in vivo (17). Preliminary was 5 times the required dosing concentration. Finally, reports from clinical observations of therapy with eribulin 4Â volumes of PBS were added, while vortexing, to achieve mesylate suggest that, at efficacious exposures, there is rela- a final dosing concentration of 10 mL/kg. tively low incidence and severity of neuropathy (18). Based on this finding, preclinical studies were conducted to directly Animals compare maximum tolerated dose (MTD) regimens (and Female BALB/c mice (approximately 7–8 weeks of age at equivalent fractions thereof) of paclitaxel, eribulin mesylate, onset of dosing) were used for all experiments. Mice were and ixabepilone for neuropathy induction in mice, which was obtained from the Harlan Laboratories Inc. and maintained evaluated by parameters including nerve conduction velocity with ad libitum access to water and a standardized synthetic (NCV), amplitude, and sciatic nerve and dorsal root ganglia diet (Harlan Teklab), both before and during the course of the (DRG) morphologic endpoints. In this article, we describe our study. findings from this study. Animal housing and procedure Materials and Methods Room temperature and humidity were maintained at 20 Æ 2C and 55 Æ 10%, respectively. Artificial lighting provided a Test materials 12-hour light/12-hour dark cycle (light 7 AM–7 PM). All The following chemicals were used in this study: eribulin experimental protocols were approved by the Institutional mesylate (synthesized at Eisai Research Institute, Andover, Animal Care and Use Committee of Eisai, Baltimore, MD, and MA) was stored at À80 C in the dark. Paclitaxel (C47H51NO14) conformed to all of the applicable institutional and govern- was purchased from LC Laboratories and stored at À20Cin mental guidelines for the humane treatment of laboratory the dark until required for use. Ixabepilone (C27H42N2O5S) was animals. purchased from Myoderm Medical and stored at 4C in the dark. MTD determination For each experimental compound, an MTD on a (Q2D Â 3) Formulations Â 2 weeks schedule (every other day for 3 injections with a 2- Eribulin mesylate was dissolved in 100% anhydrous day rest between weekly cycles for a total of 6 injections) was dimethyl sulfoxide (DMSO; Sigma-Aldrich; catalogue no. determined using groups of 10 mice each. MTD was defined as D2650) to produce a 10-mg/mL stock solution, which was the highest dose level at which no more than 10% deaths separated into aliquots and stored in the dark at À80C until occurred and/or at which no mice displayed more than 20% the day of administration. On each day of administration, the individual weight loss and/or overt clinical signs of distress stock solution was thawed and diluted with saline to a final and/or inability to eat and drink, thus requiring euthanasia. concentration of 0.25 mg/mL in 2.5% DMSO/97.5% saline to Doses chosen for the MTD studies were based on antitumor yield dosing solutions in a 10-mL/kg volume. activity and/or neuropathy-inducing doses in mice as pre- Paclitaxel was dissolved in ethanol (100%) at 10% of the final viously described (12, 17, 19, 20). Doses used for eribulin desired volume and vortexed for 2 to 3 minutes. An equal mesylate were 0.5, 0.75, 1.0, 1.25, 1.5, and 1.75 mg/kg per dose volume of cremophor (10% of final volume) was then added administration. Doses used for paclitaxel were 20, 25, 30, 35, and the mixture was re-vortexed for about 10 minutes. Imme- 40, and 45 mg/kg per dose administration. Doses used for diately prior to injection, ice-cold saline was added to make up ixabepilone were 2.0, 2.25, 2.5, 2.75, 3.0, 3.5, and 4.0 mg/kg per

OF2 Cancer Res; 71(11) June 1, 2011 Cancer Research

Neuropathy Induced by Eribulin, Paclitaxel, and Ixabepilone

dose administration. All injections were administered into the the electrodes used for recording and stimulating provides a caudal vein at a volume of 10 mL/kg. measure of the conduction velocity in the nerve segment Mice were weighed 3 times a week during the 2-week under investigation. Amplitudes measure the baseline-to-peak treatment period and for 2 weeks following completion of amplitude of the neural response. dosing. day 1 refers to the first administration of experimental compound or vehicle. day 12 refers to the day when all 6 doses Sciatic nerve and DRG histology of chemotherapeutic/vehicle had been administered. day 26 Immediately after recording of final nerve conduction refers to the final experimental day (2 weeks after completion parameters, 5 randomly selected mice from each group were of dosing). Clinical signs and survival were monitored daily. chosen for nerve/DRG excision. Sciatic nerve segments were dissected from the region immediately before the trifurcation NCV and amplitude measurement of the sciatic nerve (into the common peroneal, tibial, and Baseline NCV was measured 1 week prior to initiation of sural branches); L4 and L5 DRG were dissected from the spinal dosing in 50 mice. Mice were subsequently randomized into 5 cord. Samples were then fixed in 3% glutaraldehyde and 4% treatment groups such that mean digital NCVs (as well as all paraformaldehyde/2% glutaraldehyde, respectively. Samples other parameters) for each group of 10 mice were equivalent. were processed and resin-embedded according to previously Each group then received vehicle or chemotherapy (at MTD, published protocols (21–24) and used for light microscopic 0.75 Â MTD, 0.5 Â MTD, or 0.25 Â MTD) on a (Q2D Â 3) Â determinations. After embedding, samples were cut in 1-mm 2 week schedule. Posttreatment nerve conduction measure- semi-thin sections with a microtome RM2265 (Leica Micro- ments were made 24 hours after last chemotherapy dose. systems GmbH), and sections were stained with toluidine blue During all recording sessions, mice were anesthetized with for light microscopic examination under a Nikon Eclipse E200 2% isoflurane (by inhalation, for induction and maintenance) light microscope (Nikon). Section codes were masked before and placed in a prone position. Throughout testing, animals examination such that 2 independent examiners were una- were positioned on a warm heating pad with rectal tempera- ware of the treatment administered to the examined animal. ture monitored and maintained between 37.0C and 41.0C. Platinum subdermal needle electrodes (Grass Technologies) Results were used for both recording and stimulation. NCV and the peak compound action potential amplitude were assessed in Eribulin mesylate both caudal and digital nerves. Caudal and digital NCVs were As expected in this MTD-finding experiment, mice treated recorded orthodromically from recording sites at the proximal with 1.5 and 1.75 mg/kg eribulin mesylate on a (Q2D Â 3) Â 2- tail and the lateral malleolus, respectively. Supramaximal week schedule displayed significant body weight losses aver- stimulation was achieved using a constant-voltage square- aging between 14% and 17% (Fig. 1A) as well as clinical signs wave pulse (0.02- to 0.05-millisecond duration) isolated from characterized by piloerection, general unkempt condition, and ground and produced by the MP100 (BIOPAC Systems Inc.). decreased motor activities. Average weight losses in groups Each nerve segment stimulation was repeated for at least 3 treated with eribulin mesylate at doses 1.25 mg/kg or less were times, up to a maximum of 6 times, with increasing voltage minimal (<10%), and mice in these groups displayed no overt until the maximal response had been achieved, as evidenced clinical signs. Weight lost was completely recovered following by no further increase or a reduction in amplitude in spite of cessation of dosing. The MTD of eribulin mesylate was deter- increase in voltage. Neuroelectric signals were impedance mined to be 1.75 mg/kg when administered according to this matched and differentially amplified with a gain of 20,000 regimen, with an average group weight loss of 17%. and a frequency band of 20 to 3 kHz using a BIOPAC MP100 After 2 weeks of dosing with eribulin mesylate, there were and AcqKnowledge software version 3.7.3 (BIOPAC Systems no observed deficits in either caudal or digital NCV or Inc.). Latencies were scored from stimulus onset, and ampli- amplitude (Fig. 2). Surprisingly, dosing eribulin mesylate at tudes from baseline, using computer cursors. Latency mea- MTD (1.75 mg/kg) produced a significant increase in caudal surements were scored to the nearest 0.01 milliseconds, and amplitude compared with the control, and a similar trend was amplitude measurements were scored to the nearest 0.01 mV. noted for 0.75 Â MTD (1.31 mg/kg). A possible explanation for Differences in measurements across groups were deter- this paradoxical finding may be dehydration of the tail in these mined by direct statistical comparisons (ANOVA, followed high-dose groups. Because the amplitude of the response to by Tukey's post hoc comparisons; Prism Graphpad Software, nerve stimulation is a measure of the density of the respond- version 4.03) of the amplitude and conduction velocity data. ing fibers, decreased tissue volume secondary to dehydration Statistical significance was defined as P < 0.05. may present as an artifactual increase in amplitude. However, Caudal NCV was recorded from electrodes in a bipolar as high-dose paclitaxel- and ixabepilone-treated mice also configuration at the base of the tail (at the hairline); the exhibited similar weight loss without accompanying augmen- stimulating cathode was positioned 35 mm further distally. tation of caudal amplitude, this may not be the only explana- Digital NCV was recorded using stimulation at the base of the tion of this finding. The underlying reason for this observation second toe and recording at the level of the lateral malleolus. will require further study. The distance traveled was measured for each mouse and, in A dose-dependent effect on sciatic nerve morphology was general, was between 9 and 14 mm. The response latency at evident with 0.5 Â MTD (0.875 mg/kg) to MTD (1.75 mg/kg) supramaximal stimulation divided by the distance between doses of eribulin mesylate. At these doses, eribulin mesylate

www.aacrjournals.org Cancer Res; 71(11) June 1, 2011 OF3

Wozniak et al.

A B Vehicle

Paclitaxel 30 mg/kg i.v. Vehicle Eribulin 0.75 mg/kg i.v. Paclitaxel 35 mg/kg i.v. Eribulin 1.0 mg/kg i.v. Paclitaxel 40 mg/kg i.v. Eribulin 1.25 mg/kg i.v. Paclitaxel 45 mg/kg i.v. Eribulin 1.5 mg/kg i.v. Eribulin 1.75 mg/kg i.v. 110 110 105

105 100

100 95

90 95

85 90 80

85 % Body weight change % Body weight change 75

80 70

75 65

26 3 5 8 9 4 15 17 19 y 12 15 22 ay ay 1 y Day 1 Day 3 Day 5 Day 8 ay ay 10 y y y 17 Day 10 ay ay Day 22 Day 24 D Day 1 D Da D a ay Day 12 D Day D D D Da Da D Da Day 2 Day 26

C 110

105 Vehicle 100 Ixabepilone 2 mg/kg i.v. 95 Ixabepilone 2.25 mg/kg i.v.

90 Ixabepilone 2.5 mg/kg i.v.

85 Ixabepilone 2.75 mg/kg i.v.

80 Ixabepilone 3 mg/kg i.v. % Body weight change

75 Ixabepilone 3.5 mg/kg i.v.

70 Ixabepilone 4 mg/kg i.v.

5 7 2 10 15 16 19 2 24 ay 1 ay y D Day 3 D Day 8 ay ay ay ay 26 D Day 12Day D Day 1D Day Da D

Figure 1. Body weight of mice. Body weight of mice expressed as a percentage of their starting weight following treatment with vehicle versus eribulin mesylate (A), paclitaxel (B), and ixabepilone (C) on a (Q2D Â 3) for 2 weeks’ schedule.

induced mild to moderate pathologic changes consistent with Clear cytoplasmic vacuolation of DRG neurons appeared axonal degeneration (Fig. 5B). Axonal degeneration at MTD between 0.75 Â MTD to MTD of eribulin mesylate doses, seemed to affect both large- and small- diameter fibers; no whereas dark inclusions were only very rarely observed at the regenerative figures (e.g., thin myelinated fibers) were present. MTD (Fig. 6B).

OF4 Cancer Res; 71(11) June 1, 2011 Cancer Research

Neuropathy Induced by Eribulin, Paclitaxel, and Ixabepilone

AB Caudal nerve conduction velocity Caudal nerve amplitude

35.0 120.0 *P < 0.05 vs. vehicle *

30.0 90.0 µ V ** *** Amplitude, Amplitude,

Velocity, m/s Velocity, 25.0 60.0

20.0 30.0 Vehicle MTD 0.75 MTD 0.5 MTD 0.25 MTD Vehicle MTD 0.75 MTD 0.5 MTD 0.25 MTD (1.75 mg/kg) (1.31 mg/kg) (0.875 mg/kg) (0.44 mg/kg) (1.75 mg/kg) (1.31 mg/kg) (0.875 mg/kg) (0.44 mg/kg)

Eribulin Eribulin CD Digital nerve conduction velocity Digital nerve amplitude

25.0 45.0

40.0

20.0 ** 35.0 *

30.0 Velocity, m/s Velocity, 15.0 µ V Amplitude, 25.0

10.0 20.0 Vehicle MTD 0.75 MTD 0.5 MTD 0.25 MTD Vehicle MTD 0.75 MTD 0.5 MTD 0.25 MTD (1.75 mg/kg) (1.31 mg/kg) (0.875 mg/kg) (0.44 mg/kg) (1.75 mg/kg) (1.31 mg/kg) (0.875 mg/kg) (0.44 mg/kg)

Eribulin Eribulin

Figure 2. Eribulin mesylate effect on NCV and amplitude. A, eribulin mesylate at 0.25 Â MTD, 0.5 Â MTD, 0.75 Â MTD, and MTD had no inhibitory effect on caudal NCV. B, eribulin at MTD increased caudal amplitude with no significant effects at other doses. Eribulin did not affect digital nerve conduction velocity (C) or amplitude (D) at any dose. Figure depicts mean Æ SEM.

Paclitaxel group over the duration of the treatment regimen are shown As shown in Figure 1B, all mice receiving more than 30 mg/ as a percentage of starting weight (day 1) in Figure 1B. The kg paclitaxel on a (Q2D Â 3) Â 2 week schedule suffered MTD of paclitaxel was determined to be 30 mg/kg when appreciable weight loss (>20% of original starting weight). administered according to this regimen. Furthermore, mice in these groups displayed significant hin- After 2 weeks of dosing with paclitaxel, clear dose-depen- dlimb nerve malfunction; these animals were sacrificed before dent decreases in caudal NCV and amplitude were evident receiving all planned doses in the study. The remaining mice (Fig. 3). The MTD of paclitaxel (30 mg/kg) caused significant in the 35-mg/kg group (2/10) recovered their weight loss after reduction in caudal NCV compared with control (P < 0.01), dosing. In contrast, doses of 30 mg/kg paclitaxel or less were whereas MTD and 0.75 Â MTD (30 and 22.5 mg/kg, respec- well tolerated, with mice in these groups completing the entire tively) caused significant reduction in caudal amplitude com- chemotherapy regimen. Average body weights per treatment pared with the control (P < 0.001). Paclitaxel doses of

www.aacrjournals.org Cancer Res; 71(11) June 1, 2011 OF5

Wozniak et al.

AB Caudal nerve conduction velocity Caudal nerve amplitude

**P < 0.01 vs. vehicle **P < 0.01 vs. vehicle 35.0 9.0

30.0 6.0 ** ** ** Amplitude, µ V 25.0 3.0 Velocity, m/s

20.0 0.0 Vehicle MTD 0.75 MTD 0.5 MTD 0.25 MTD Vehicle MTD 0.75 MTD 0.5 MTD 0.25 MTD (30 mg/kg) (22.5 mg/kg) (15 mg/kg) (7.5 mg/kg) (30 mg/kg) (22.5 mg/kg) (15 mg/kg) (7.5 mg/kg)

Paclitaxel Paclitaxel C D Digital nerve conduction velocity Digital nerve amplitude

**P < 0.01 vs. vehicle *P < 0.05 vs. vehicle 25.0 45

30 ** * 20.0 25 ** Amplitude, µ V 20 Velocity, m/s 15

10 15.0 Vehicle MTD 0.75 MTD 0.5 MTD 0.25 MTD 5 (30 mg/kg) (22.5 mg/kg) (15 mg/kg) (7.5 mg/kg) Vehicle MTD 0.75 MTD 0.5 MTD 0.25 MTD (30 mg/kg) (22.5 mg/kg) (15 mg/kg) (7.5 mg/kg)

Paclitaxel Paclitaxel

Figure 3. Paclitaxel effect on NCV and amplitude. Paclitaxel at its MTD reduced caudal NCV (A) and caudal amplitude (B). C, paclitaxel had no significant affect on digital nerve conduction velocity. D, paclitaxel at its MTD and 0.75 MTD reduced digital nerve amplitude. Figure depicts mean Æ SEM.

0.5 Â MTD and lower had no effect on caudal NCV and A dose-dependent effect of paclitaxel on sciatic nerve amplitude. Furthermore, MTD and 0.75 Â MTD of paclitaxel morphology was observed. At MTD (30 mg/kg), paclitaxel (30 and 22.5 mg/kg, respectively) caused significant reductions caused severe pathologic changes consistent with axonal in digital nerve amplitude (Fig. 3D; P < 0.01 and P < 0.05, degeneration (Fig. 5C). Axonal degeneration seemed to affect respectively). In contrast, although all paclitaxel doses tested both large- and small-diameter fibers, and no regenerative tended to adversely affect digital NCV, no dose effect attained figures (e.g., thin myelinated fibers) were present. The statistical significance (Fig. 3C). severity of axonopathy was milder but still clearly evident

OF6 Cancer Res; 71(11) June 1, 2011 Cancer Research

Neuropathy Induced by Eribulin, Paclitaxel, and Ixabepilone

AB Caudal nerve conduction velocity Caudal nerve amplitude

***P < 0.001 vs. vehicle ***P < 0.001 vs. vehicle **P < 0.01 vs. vehicle 35.0 80.0 **P < 0.01 vs. vehicle *P < 0.05 vs. vehicle

30.0 ** * 60.0 ** *** *** Amplitude, µ V

Velocity, m/s 25.0 40.0

20.0 Vehicle MTD 0.75 MTD 0.5 MTD 0.25 MTD 20.0 (3.0 mg/kg) (2.25 mg/kg) (1.5 mg/kg)(0.75 mg/kg) Vehicle MTD 0.75 MTD 0.5 MTD 0.25 MTD (3.0 mg/kg) (2.25 mg/kg) (1.5 mg/kg)(0.75 mg/kg) Ixabepilone Ixabepilone CD Digital nerve conduction velocity Digital nerve amplitude

*P < 0.05 vs. vehicle 25.0 45.0 **P < 0.01 vs. vehicle

35.0 ** 20.0 * Amplitude, µ V Velocity, m/s

25.0

15.0 15.0 Vehicle MTD 0.75 MTD 0.5 MTD 0.25 MTD Vehicle MTD 0.75 MTD 0.5 MTD 0.25 MTD (3.0 mg/kg) (2.25 mg/kg) (1.5 mg/kg) (0.75 mg/kg) (3.0 mg/kg) (2.25 mg/kg) (1.5 mg/kg) (0.75 mg/kg) Ixabepilone Ixabepilone

Figure 4. Ixabepilone effect on NCV and amplitude. A, ixabepilone at 0.5 Â MTD, 0.75 Â MTD, and MTD decreased caudal NCV. B, ixabepilone at 0.75 Â MTD and MTD produced dose-dependent decreases in caudal nerve amplitude. Ixabepilone at MTD caused a significant reduction in digital NCV (C) and amplitude (D). Figure depicts mean Æ SEM. ingroupstreatedwith0.75Â MTD (22.5 mg/kg) or 0.5 Â sions in the cytoplasm. In addition, DRG from the 0.75 Â MTD (15 mg/kg) paclitaxel (Supplementary Fig. S1A and B). MTD (22.5 mg/kg) group had clear vacuolations in the Furthermore, in DRG, dose-dependent adverse effects of cytoplasm of neurons as well as in satellite cells (Supple- paclitaxel were evident. Paclitaxel at 0.5 Â MTD (15 mg/ mentary Fig. 2SB). At the MTD of paclitaxel (30 mg/kg), a kg; Supplementary Fig. S2A) caused formation of dark inclu- proportion of neurons in the DRG were degenerating and

www.aacrjournals.org Cancer Res; 71(11) June 1, 2011 OF7

Wozniak et al.

A B

Figure 5. Effect of MTDs of eribulin mesylate, paclitaxel, and ixabepilone on sciatic nerve morphology. Morphologic changes were most severe with paclitaxel (C) and ixabepilone (B), with both agents causing frequent and severe pathologic changes consistent with axonal degeneration of both large and small fibers (highlighted with arrows and arrowheads, C D respectively). No regenerative figures (e.g., thin myelinated fibers) were evident with either treatment. Eribulin mesylate at its MTD induced some mild pathology (B), but it was less frequent than with paclitaxel or ixabepilone, more closely resembling vehicle-treated mice (A). Scale bar, 20 mm.

their cytoplasm appeared much darker than normal neurons 3 mg/kg) caused significant reductions in caudal NCV, (Fig. 6C). Dose-dependent axonal degeneration was present whereas doses of 0.75 Â MTD to MTD (2.25–3 mg/kg) caused in the proximal axons at 0.5 Â MTD and greater (Fig. 6C; significant reduction in caudal amplitude. At MTD (3.0 mg/ Supplementary Fig. 2S). kg), ixabepilone significantly reduced both digital NCV and amplitude (Fig. 4C and D). Ixabepilone Ixabepilone at doses below the MTD (0.75–2.25 mg/kg) All mice receiving ixabepilone doses greater than 3.0 mg/kg generally had minimal effect on the morphology of nerve on a (Q2D Â 3) Â 2-week schedule suffered appreciable weight fibers, although rare and sporadic alterations were present loss (more than 20% of original weight) and displayed overt in sciatic nerves (Supplementary Fig. S1C). Sciatic nerves clinical signs of non–well-being (including piloerection and from mice receiving MTD of ixabepilone (3 mg/kg) showed unkempt coats); these animals were euthanized before the end severe and frequent pathologic changes, represented by of the planned doses. Moderate weight losses (between 11% different stages of axonal degeneration. As shown in Figure and 15%) were observed in the middle-dose groups (2.75 and 5D, axonal degeneration seemed to affect both small- and 3.0 mg/kg ixabepilone), which reversed during the 14-day large-diameter fibers, although no regenerative figures (e.g., recovery/observation period after cessation of dosing. In thin myelinated fibers) were present. Ixabepilone, at its contrast, 2.5 mg/kg or lower dose of ixabepilone caused MTD, caused very severe, frequent, morphologic alterations, minimal weight loss (10%), no overt clinical signs, and both in neuronal and in glial compartments (Fig. 6D) of the was well tolerated by mice. Minimal clinical signs were DRG. The neurons had dark cytoplasmic inclusions, often observed in the 2.75- and 3.0-mg/kg dose groups. The MTD localized to the perinuclear area. Moreover, clear vacuola- of ixapebilone on a (Q2D Â 3) Â 2-week schedule was tions and swelling phenomena were evident in the cyto- determined to be 3 mg/kg (Fig. 1C). plasm of satellite cells. Changes were indicative of severely After 2 weeks of dosing with ixabepilone, a clear dose– injured and degenerating cells. Furthermore, adverse effects response for decreasing caudal NCV and amplitude was of ixabepilone administration were evident with other doses. evident (Fig. 4A and B). Doses of 0.5 Â MTD to MTD (1.5– DRG from mice receiving ixabepilone at 0.25 Â MTD and

OF8 Cancer Res; 71(11) June 1, 2011 Cancer Research

Neuropathy Induced by Eribulin, Paclitaxel, and Ixabepilone

Figure 6. Effect of MTDs of eribulin mesylate, paclitaxel, and ixabepilone on DRG morphology. A B DRG morphology at the light microscopic level showed changes after each chemotherapy (at its MTD). Ixabepilone (D) caused most severe and frequent morphologic changes both in neuronal and glial compartments. Severely injured, degenerating neurons are outlined (red). Dark cytoplasmic inclusions were evident and were often localized in the perinuclear area. Vacuolations (white arrows) and swelling phenomena (red arrowheads) were evident in the cytoplasm of satellite cells (D). DRG from paclitaxel-treated mice (C) also C D displayed degenerating nerve cells (outlined by red circle) with dark cytoplasm (red arrows) and clear vacuolations in cytoplasm of satellite cells (white arrow). Alterations in the proximal axons of DRG were also observed (white circle). DRG from eribulin mesylate-treated mice (B) showed mild pathologic changes evidenced by some cytoplasmic vacuolation (white arrow) and degenerating nerve cells (red arrow), as compared to vehicle- treated mice (A). Scale bar, 20 mm.

0.75 Â MTD displayed occasional cytoplasmic swelling in the neurotoxicity symptoms is often incomplete, and a long satellite cells (Supplementary Fig. S2C and D). Cytoplasmic period of regeneration is sometimes required to restore dark inclusions and degenerating neurons, as well as function. vacuoles of satellite cells, were also evident in the 0.75 Â Microtubule-targeting agents were first introduced into MTD and MTD ixabepilone groups (Fig. 6D; Supplementary clinical oncology in the 1960s and are essential components Fig. S2D). Moreover, rare episodes of clear cytoplasmic in the therapy of many cancers, including lymphoma as well as vacuolation of neurons were also observed. breast, ovarian, lung, and head and neck cancers (31). In spite of their associated neurotoxic side effects, these agents, in Discussion general, remain the most effective treatment options for survival prolongation in advanced disease (32). The anticancer Chemotherapy-induced peripheral neurotoxicity is a major effects of tubulin-targeting agents generally derive from their clinical problem representing the dose-limiting side effect of ability to bind to microtubules, interfere with mitotic spindle many antineoplastic drugs, and it has a significant negative formation, and ultimately to block mitosis, resulting in cell impact on the quality of life and often results in treatment death. However, because somatic neurons do not divide, the delays or discontinuation (25–27). Indeed, neurotoxic side neurotoxic effects of tubulin-targeting agents could derive effects associated with chemotherapies are second in fre- from effects on interphase microtubules not involved in quency only to hematologic toxicities. Unlike the hematologic mitosis. Although some information exists, our understanding side effects for which there are known effective treatments, of the mechanisms behind the neurotoxic effects of tubulin- neurotoxicity side effects cannot be effectively treated or targeting agents is far from complete. Further studies of the prevented (28, 29). Neurotoxicity may develop as a conse- neurotoxic side effects of tubulin-targeting chemotherapies quence of treatment with platinum analogues (e.g., cisplatin, are thus warranted (33). oxaliplatin, carboplatin), taxanes (e.g., paclitaxel, docetaxel), For the past 20 years, several chemotherapy-induced Vinca alkaloids (vincristine), and, more recently, thalidomide neuropathy animal models have been published, most and bortezomib as well as epothilones (9, 30). The degree and commonly using rodents. Similar to the human situation, type of neuropathy depend on the chemotherapeutic drug, chemotherapy-induced neuropathy rodent models are char- dose intensity, and cumulative dose. Recovery from peripheral acterized by neurophysiologic deficits and morphologic

www.aacrjournals.org Cancer Res; 71(11) June 1, 2011 OF9

Wozniak et al.

alterations in DRG and myelinated nerve fibers (23, 34–36). mesylate were noticeably fewer and less severe than those Somewhat surprisingly, in spite of the existence of such seen with paclitaxel or ixabepilone at their respective models, to our knowledge, there has never been a systematic MTDs. Because all 3 agents target tubulin, the question comparison of the various microtubule-targeting agents in arises as to why eribulin mesylate is less damaging to preclinical models. neurons in this mouse model. Eribulin mesylate, a nontaxane microtubule dynamics inhi- Paclitaxel and ixabepilone areconsideredmicrotubule- bitor belonging to the halichondrin class of antineoplastic stabilizing agents because they promote polymerization of agents, is a structurally simplified macrocyclic ketone analo- microtubules and increase polymer mass in cells. In con- gue of halichondrin B, a natural product originally isolated trast, eribulin mesylate is a microtubule-destabilizing agent from the marine sponge Halichondria okadai (17). Eribulin because it binds to the Vinca-binding domain of tubulin and mesylate is currently undergoing development for the treat- inhibits tubulin polymerization, thereby inducing cell-cycle ment of advanced breast cancer and other solid tumors. Other arrest and apoptosis (10, 39). At low concentrations, both microtubule-targeting agents used to treat breast cancer stabilizers and destabilizers suppress microtubule dynamics patients, including paclitaxel and ixabepilone, display a com- without changing polymer mass (40, 41). As the understand- mon dose-limiting toxicity of peripheral neuropathy. The ing of the interactions of tubulin-targeting agents at the purpose of this study was to investigate neuropathy-inducing molecular level increases, differences in activity and side effects of eribulin mesylate, compared with those of paclitaxel effects may become clearer. Differences in microtubule- or ixabepilone, in mice. binding properties may have significant effects on the toxi- All 3 chemotherapeutic drugs have antineoplastic activity in city profile of each agent (10, 36). various animal models (6, 7, 17) and human cancer conditions In this context, Jordan and colleagues found that eribulin (2, 37, 38). Importantly, all 3 compounds have shown efficacy mesylate affects microtubule growth phases, as opposed to in multiple in vivo cancer efficacy models at doses often less the shortening phase, in association with sequestration of than the MTD described in this study in the case of eribulin tubulin into aggregates (42). Eribulin mesylate binding mesylate (17), or generally comparable with the MTD, in the along microtubule sides has at least 10-fold lower affinity case of paclitaxel (17) and ixabepilone (6, 7). than at the positive ends (16). This is in contrast to the For example, eribulin mesylate has shown efficacy in a binding pattern of paclitaxel and ixabepilone, which each human breast (MDA-MB-435) cancer xenograft model in affect both the growth and shortening phases of micro- athymic mice at doses between 0.25 and 1.0 mg/kg using tubule dynamics, and each of which are thought to target a(Q2DÂ 3) Â4 weeks schedule (17). These efficacious doses the same, or close to the same, binding site on b-tubulin are lower than the MTD described in this study (1.75 mg/kg). (10, 43). As neuronal microtubules seem responsible for Paclitaxel has shown antineoplastic effects in the same intracellular transport of essential molecules along the breast xenograft model at 25 mg/kg, similar to the MTD axon, it is possible that binding along the microtubule described in this study (30 mg/kg; ref. 17). Ixabepilone has sides may inhibit transport of essential molecules with shown antitumor activity in mouse models, including for resulting peripheral neurotoxicity (44). These theories breast cancer [KPL4, Pat-21], at a dose range of 3.2 to 6.0 mg/ require further investigation. kg administered Q2D Â 5orQ4DÂ 3 (6, 7), which is also In spite of the limitations of this study, such as its relatively similar to the ixabepilone MTD (3 mg/kg) employed in this short duration and small sample size, the findings suggest study. Therefore, it is reasonabletoconcludethatthemice that, if these animal studies are corroborated in the clinical used in this study received similar (paclitaxel, ixabepilone) setting, eribulin mesylate could have a positive impact on the or greater (eribulin mesylate) exposures than those required side effect profile of cancer patients receiving microtubule- for efficacy. based therapies. Because these drugs have different potencies and pharmacokinetics, we compared them on the basis of MTD by Disclosure of Potential Conflicts of Interest using a neuropathy-inducing dosing regimen (Q2D Â 3) for

2 weeks. Once MTDs were defined as 1.75 mg/kg eribulin K.M. Wozniak, B.S. Slusher, R.G. Lapidus, B.A. Littlefield, and Y. Wu were mesylate, 30 mg/kg paclitaxel, and 3.0 mg/kg ixabepilone, employees of Eisai at the time of this work; K.M. Wozniak, Y. Wu and B.S. Slusher we administered 0.25 Â MTD, 0.5 Â MTD, 0.75 Â MTD, and currently have a sponsored research agreement with Eisai. No other conflicts of interest were disclosed. MTD and evaluated their effects on NCV, amplitude, and morphology of sciatic nerves and DRG. Using these comparisons, ixabepilone and paclitaxel dosed on a (Q2D Â 3) Grant Support for 2 weeks schedule were found to produce significant deficits in conduction parameters of caudal and digital This work was supported by Eisai Inc.; K.M. Wozniak, K. Nomoto, R.G. Â Lapidus, Y. Wu, K. Hayakawa, S. Hosokawa, M.J. Towle, B.A. Littlefield, and B.S. nerves at doses of 0.5 MTD or more. In contrast, eribulin Slusher were Eisai employees at the time this study was conducted. mesylate induced no deficits in any monitored nerve con- The costs of publication of this article were defrayed in part by the payment duction parameter (velocity or amplitude) at any dose of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. tested, including MTD (1.75 mg/kg). Similarly, although all compounds produced morphologic changes in DRG and Received November 18, 2010; revised February 24, 2011; accepted March 20, sciatic nerves at MTD, the changes induced by eribulin 2011; published OnlineFirst April 15, 2011.

OF10 Cancer Res; 71(11) June 1, 2011 Cancer Research

Neuropathy Induced by Eribulin, Paclitaxel, and Ixabepilone

References 1. Altekruse SF, Kosary CL, Krapcho M, Neyman N, Aminou R, Waldron 24. Meregalli C, Canta A, Carozzi VA, Chiorazzi A, Oggioni N, Gilardini A, W, et al., (eds). SEER Cancer Statistics Review, 1975–2007. Bethesda, et al. Bortezomib-induced painful neuropathy in rats: a behavioral, MD: National Cancer Institute; 2009. [http://seer.cancer.gov/csr/ neurophysiological and pathological study in rats. Eur J Pain 2010; 1975_2007/] 14:343–50. 2. Perez EA. Paclitaxel in breast cancer. Oncologist 1998;3:373–89. 25. Windebank AJ. Chemotherapeutic neuropathy. Curr Opin Neurol 3. Kuroi K, Shimozuma K. Neurotoxicity of taxanes: symptoms and 1999;12:565–71. quality of life assessment. Breast Cancer 2004;11:92–9. 26. Quastthoff S, Hartung HP. Chemotherapy-induced peripheral neuro- 4. Guastalla JP 3rd, Dieras V. The taxanes: toxicity and quality of life pathy. J Neurol 2004;249:9–17. considerations in advanced ovarian cancer. Br J Cancer 2003;89 27. Ocean AJ, Vadhat LT. Chemotherapy-induced peripheral neuropathy: Suppl 3S16–22. pathogenesis and emerging therapies. Support Care Cancer 5. Bokhari F, Sawatzky JV. Chronic neuropathic pain in women after 2004;12:619–25. breast cancer treatment. Pain Manag Nurs 2009;10:197–205. 28. Windebank AJ, Grisold W. Chemotherapy-induced neuropathy. J 6. Lee FY, Covello KL, Castaneda S, Hawken DR, Kan D, Lewin A, et al. Periph Nerv Syst 2008;13:27–46. Synergistic antitumor activity of ixabepilone (BMS-227550) plus bev- 29. Sioka C, Kyritis AP. Central and peripheral nervous system toxicity of acizumab in multiple in vivo tumor models. Clin Cancer Res 2008;14: common chemotherapeutic agents. Cancer Chemother Pharmacol 8123–31. 2009;63:761–7. 7. Lee FY, Borzilleri R, Fairchild CR, Kamath A, Smykla R, Kramer R, et al. 30. Park SB, Krishnan AV, Lin CS, Goldstein D, Friedlander M, Kiernan Preclinical discovery of ixabepilone, a highly active antineoplastic MC. Mechanisms underlying chemotherapy-induced neurotoxicity agent. Cancer Chemother Pharmacol 2008;63:157–66. and the potential for neuroprotective strategies. Curr Med Chem 8. Hunt JT. Discovery of ixabepilone. Mol Cancer Ther 2009;8:275–81. 2008;15:3081–94. 9. Steinberg M. Ixabepilone: a novel microtubule inhibitor for the treat- 31. Jordan MA, Kamath K, Manna T, Okouneva T, Miller HP, Davis C, et al. ment of locally advanced or metastatic breast cancer. Clin Ther The primary antimitotic mechanism of action of the synthetic hali- 2008;30:1590–617. chondrin E7389 is suppression of microtubule growth. Mol Cancer 10. Perez EAMicrotubule inhibitors: differentiating tubulin-inhibiting Ther 2005;4:1086–95. agents based on mechanisms of action, clinical activity and resis- 32. Giannakakou P, Nakano M, Nicolaou KC, O'Brate A, Yu J, Blagosk- tance. Mol Cancer Ther 2009;8:2086–95. lonny MV, et al. Enhanced microtubule-dependent trafficking and p53 11. Hagiwara H, Sunada Y. Mechanism of taxane neurotoxicity. Breast nuclear accumulation by suppression of microtubule dynamics. Proc Cancer 2004;11:82–5. Natl Acad Sci U S A 2002;99:10855–60. 12. Lee JJ, Swain SM. Peripheral neuropathy induced my microtubule 33. McLeod JG, Penny R. Vincristine neuropathy: an electrophysiological stabilizing agents. J Clin Oncol 2006;24:1633–42. and histological study. J Neurol Neurosurg Psychiatry 1969;32: 13. Canta A, Chiorazzi A, Cavaletti G. Tubulin: a target for antineoplastic 297–304. drugs into the cancer cells but also in the peripheral nervous system. 34. Jamieson SM, Liu JJ, Connor B, Dragunov M, McKeage MJ. Nucleolar Curr Med Chem 2009;16:1315–24. enlargement, nuclear eccentricity and altered cell body immunostain- 14. Vahdat LT, Pruitt B, Fabian CJ, Rivera RR, Smith DA, Tan-Chiu E, et al. ing characteristics of large sized sensory neurons following treatment Phase II study of eribulin mesylate, a halichondrin B analog, in patients of rats with paclitaxel. Neurotoxicology 2007;28:1092–8. with metastatic breast cancer previously treated with an anthracycline 35. Cavaletti G, Gilardini A, Canta A, Rigamonti L, Rodriguez-Menendez and a taxane. J Clin Oncol 2009;27:2954–61. V, Ceresa C, et al. Bortezomib-induced peripheral neurotoxicity: a 15. Twelves C, Cortes J, Vahdat LT, Wanders J, Akerele C, Kaufman PA. neurophysiological and pathological study in the rat. Exp Neurol Phase III trials of eribulin mesylate (E7389) in extensively pretreated 2007;204:317–25. patients with locally recurrent or metastatic breast cancer. Clin Breast 36. Authier N, Balayssac D, Marchand F, Ling B, Zangarelli A, Descoeur J, Cancer 2010;10:160–3. et al. Animal models of chemotherapy-evoked painful peripheral 16. Smith JA, Wilson L, Azarenko O, Zhu X, Lewis BM, Littlefield BA, et al. neuropathies. J Am Soc Exp Neurother 2009;6:620–9. Eribulin binds at microtubule ends to a single site on tubulin to 37. Jimeno A. Eribulin: rediscovering tubulin as an anticancer agent. Clin suppress dynamic instability. Biochemistry 2010;49:1331–7. Cancer Res 2009;15:3903–5. 17. Towle MJ, Salvato KA, Budrow J, Wels BF, Kuznetsov G, Aalfs KK, 38. Goel S, Mita AC, Mita M, Rowinsky EK, Chu QS, Wong N, et al. A et al. In vitro and in vivo anticancer activities of synthetic macrocyclic phase 1 study of eribulin mesylate (E7389), a mechanistically novel ketone analogues of halichondrin B. Cancer Res 2001;61:1013–21. inhibitor of microtubule dynamics, in patients with advanced solid 18. Cigler T, Vahdat LT. Eribulin mesylate for the treatment of breast malignancies. Clin Cancer Res 2009;15:4207–12. cancer. Expert Opin Pharmacother 2010;11:1587–93. 39. Kuznetsov G, Towle MJ, Cheng H, Kawamura T, TenDyke K, Liu D, 19. Mimura Y, Kato H, Eguchi K, Ogawa T. Schedule dependency of et al. Induction of morphological and biochemical apoptosis following paclitaxel-induced neuropathy in mice: a morphological study. Neu- prolonged mitotic blockage by halichondrin B macrocyclic ketone rotoxicology 2000;1:513–20. analog E7389. Cancer Res 2004;64:5760–6. 20. Lee FY, Smykla R, Johnston K, Menard K, McGlinchey K, Peterson 40. Zhou J, Giannakakou P. Targeting microtubules for cancer che- RW, et al. Preclinical efficacy spectrum and pharmacokinetics of motherapyCurr Med Chem Anticancer Drugs 2005;5:65–71. ixabepilone. Cancer Chemother Pharmacol 2009;63:201–12. 41. Jordan MA, Wilson L. Microtubules as a target for anticancer drugs. 21. Cavaletti G, Petruccioli MG, Tredici G, Marmiroli P, Barajon I, Fabbrica Nat Rev Cancer 2004;4:253–65. D, et al. Effects of repeated administration of low doses of cisplatin on 42. Jordan MA, Kamath K, Manna T, Okouneva T, Miller HP, Davis C, et al. the rat nervous system. Int J Tissue React 1991;13:151–7. The primary antimitotic mechanism of action of the synthetic hali- 22. Cavaletti G, Tredici G, Marmirioli P, Petruccioli MG, Barajon I, Fab- chondrin E7389 is suppression of microtubule growth. Mol Cancer brica D. Morphometric study of the sensory neuron and peripheral Ther 2005;4:1086–95. nerve changes induced by chronic cisplatin (DDP) administration in 43. Yue Q-X, Liu X, Guo D-AMicrotubule-binding natural products for rats. Acta Neuropathol 1992;84:364–71. cancer therapy. Planta Med 2010;76:1037–44. 23. Cavaletti G, Tredici T, Braga M, Tazarri S. Experimental peripheral 44. Dumontet C, Sikic BI. Mechanisms of action and resistance to anti- neuropathy induced in adult rats by repeated intraperitoneal admin- tubulin agents: microtubule dynamics, drug transport and cell death. J istration of taxol. Exp Neurol 1995;133:64–72. Clin Oncol 1999;17:1061–70.

www.aacrjournals.org Cancer Res; 71(11) June 1, 2011 OF11

Comparison of Neuropathy-Inducing Effects of Eribulin Mesylate, Paclitaxel, and Ixabepilone in Mice

Krystyna M. Wozniak, Kenichi Nomoto, Rena G. Lapidus, et al.

Cancer Res Published OnlineFirst April 15, 2011.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-10-4184

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2011/04/15/0008-5472.CAN-10-4184.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2011/05/13/0008-5472.CAN-10-4184. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.