at SciVerse ScienceDirect

Experimental Eye Research 113 (2013) 148e150

Contents lists available

Experimental Eye Research

journal homepage: www.elsevier .com/locate/yexer

Short communication

Corneal innervation as a window to peripheral neuropathies

Giulio Ferrari a,b,*, Nambi Nalassamy a, Heather Downs c, Reza Dana a,Anne Louise Oaklander c,d

a Schepens Eye Research Institute, Massachusetts Eye and Ear Infirmary, Harvard Medical School, 20 Staniford St., Boston, MA 02114, USAb San Raffaele Scientific Institute, Via Olgettina 60, 20132 Milan, ItalycDepartments of Neurology and Pathology (Neuropathology), Massachusetts General Hospital, Harvard Medical School,275 Charles St., Boston, MA 02114, USAdDepartment of Neurology, Massachusetts General Hospital, 275 Charles St., Boston, MA 02114, USA

a r t i c l e i n f o

Article history:Received 21 March 2013Accepted in revised form 17 May 2013Available online 14 June 2013

Keywords:corneal nervesperipheral neuropathy

Abbreviations: CIPN, chemotherapy-induced perintraepidermal nerve-fiber density; CCM, in vivo cCNFD, corneal nerve-fiber density.* Corresponding author. Cornea and Ocular Surfa

Raffaele Hospital, Via Olgettina 60, 20122 Milan, Itfax: þ39 02 26436164.

E-mail addresses: giulio.ferrari@schepens.harv(G. Ferrari).

0014-4835/$ e see front matter � 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.exer.2013.05.016

a b s t r a c t

The cornea receives the densest sensory innervation of the body, which is exclusively from small-fibernociceptive (pain-sensing) neurons. These are similar to those in the skin of the legs, the standardlocation for neurodiagnostic skin biopsies used to diagnose small-fiber peripheral polyneuropathies.Many cancer chemotherapy agents cause dose-related, therapy-limiting, sensory-predominant poly-neuropathy. Because corneal innervation can be detected non-invasively, it is a potential surrogatebiomarker for skin biopsy measurements. Therefore, we compared hindpaw-skin and cornea innervationin mice treated with neurotoxic chemotherapy. Paclitaxel (0, 5, 10, or 20 mg/kg) was administered to C57/Bl6 mice and peri-mortem cornea and skin biopsies were immunolabeled to reveal and permit quan-titation of innervation. Both tissues demonstrated dose-dependent, highly correlated (r ¼ 0.66) nervefiber damage. These findings suggest that the quantification of corneal nerves may provide a usefulsurrogate marker for skin peripheral innervation.

� 2013 Elsevier Ltd. All rights reserved.

The cornea receives the densest sensory innervations of anyanatomical site (Bonini et al., 2003) and nerves can be easily imagedin the cornea, even in vivo, given its accessibility and transparence.Corneal innervation has been studied extensively in a number ofocular diseases; however, little information is available in thesetting of systemic peripheral nerve damage.

Use of cancer chemotherapy is increasing due to improvedsurvival rates and new drug discovery, but many patients are un-able to complete optimal chemotherapy regimens due to adverseevents. In fact, painful, distal sensory chemotherapy-induced pe-ripheral neuropathy (CIPN) is dose-limiting for many antineoplas-tics, including taxanes, vinca alkaloids, platinum-complex, andproteasome-inhibitors. Small-diameter pain fibers are the mostoften involved because their inefficient non-saltatory conductionplus their minimal capacity for axon transport make them exqui-sitely vulnerable to chemotherapy-evoked mitochondrial toxicity

ipheral neuropathy; IENFD,orneal confocal microscopy;

ce Unit, Eye Repair Lab, Sanaly. Tel.: þ39 02 26436186;

ard.edu, ferrari.giulio@hsr.it

All rights reserved.

(Flatters and Bennett, 2006). As a consequence, neuropathic pain iscommon in CIPN, and this can be debilitating, long-lasting andtreatment-resistant, forcing physicians to curtail chemotherapy,even at the risk of compromising survival. The effect of systemicchemotherapy on corneal nerves is unknown; however, we hy-pothesized that they should be similarly affected as longer axonsinnervating other anatomical areas. If this hypothesis were true, itcould have clinical implications, specifically the use of in vivocorneal confocal microscopy (CCM), as a potential non-invasivesurrogate marker for CIPN. CCM allows repeatable visualizationand quantitation of sub-basal corneal nerves.

For early CIPN diagnosis, distal-leg skin biopsy to measureintraepidermal nerve-fiber density (IENFD) is most recommendedalthough biopsies are slightly invasive and technically demandingto process and interpret (England et al., 2009). Small series suggestthat CCM can detect corneal nerve damage in sensory poly-neuropathy (Gemignani et al., 2010; Mimura et al., 2008); however,sensitivity is a concern given the short length of corneal afferentscompared to the long axons innervating the feet. Thus, corneal anddistal-limb measurements deserve comparison. The few studies ofdiabetic polyneuropathy are encouraging, but diabetics withoutpolyneuropathy also had reduced corneal nerve-fiber density(CNFD) and distal-leg IENFD, and correlation was modest(rs ¼ 0.385) (Quattrini et al., 2007). CCM has been studied in a

G. Ferrari et al. / Experimental Eye Research 113 (2013) 148e150 149

limited number of patients with non-diabetic polyneuropathies.Our single-case report of capecitabine-CIPN prompted the currentprospective, controlled, doseeresponse comparison of skin andcorneal measurements in a mouse model (Ferrari et al., 2010). Wedecided to study CIPN induced by paclitaxel, a compound of thetaxane family, because taxanes are globally used for commoncancers, and because persistent painful CIPN develops in almost 1/4of patients receiving standard doses and nearly all aggressivelytreated patients (Gradishar et al., 2005).

Using approved methods, six-week-old male C57/BL6 micereceived one intraperitoneal injection of paclitaxel dissolved inCremophor-EL at doses spanning the experimental range; 5 mg/kg

Fig. 1. Reductions in corneal and hindpaw small-fiber innervation following paclitaxel admianti-beta 3 tubulin immunoreactive mouse corneas (upper row) and anti-PGP9.5-immunorlabeled innervation (P < 0.0001) (magnification 200�). Bar length: 300 mm (cornea), 50 mm ((C) Epidermal nerve density is inversely related with paclitaxel dosage (P < 0.0001). (D) Cowith standard deviation.

(n ¼ 8), 10 mg/kg (n ¼ 8), or 20 mg/kg (n ¼ 13). Fourteen untreatedmice provided controls. At sacrifice two weeks later, corneas and2 mm punch-biopsies from one plantar hindpaw were removed,fixed in lysine-periodate-paraformaldehyde, vertically sectioned(50 mm) and immunohistochemically labeled against PGP9.5 (skin)or anti-beta 3 tubulin (corneas). Secondary antibody comprisedfluorophore-conjugated donkey anti rabbit IgG FITC. IENFD wasquantitated by a single blinded morphometrist using standardclinical diagnostic methods (England et al., 2009). Total lengths oflabeled corneal axons were digitally measured through segmen-tation, binarization and skeletonization (The Mathworks, Inc.,Natick, MA). Between-group differences were assessed by ANOVA

nistration. (A) Immunohistochemical visualization of sensory nerve endings (arrows) ineactive hindpaw-skin biopsies. Both tissues had highly significant dose-related loss ofskin). (B) Corneal nerve density is inversely related with paclitaxel dosage (P < 0.0001).rneal and epidermal nerve density is highly correlated (r ¼ 0.66). Bars represent mean

G. Ferrari et al. / Experimental Eye Research 113 (2013) 148e150150

plus KruskaleWallis post-hoc tests; Pearson correlation co-efficients were used.

In vivo corneal examination revealed no keratitis or lesions intreated or untreated mice. Immunohistochemistry revealed highlysignificant dose-related reductions in corneas and hindpaws(Fig.1A). CNFD averaged 27,131�2446 mm/image in untreatedmice,17,924 � 848 mm/image among 5 mg/kg mice, 11,634 � 783 mm/image among 10 mg/kg mice, and 4259 � 523 mm/image among20 mg/kg mice (all P < 0.0001) (Fig. 1B); consistent with dose-related neurotoxicity. Biopsies from hindpaw-skin biopsiesrevealed similar dose-related reductions; 527 � 53 neurites/mm2

skin surface area in untreated mice; 226 � 19 neurites/mm2 skinsurface area among 5 mg/kg mice, 107 � 14 neurites/mm2 skinsurface area among 10mg/kgmice, and 135�17 neurites/mm2 skinsurface area among 20 mg/kg mice (all P < 0.0001) (Fig. 1C).

In summary, nerves in the skin and in the cornea were pro-gressively reduced following increasing doses of paclitaxel. Indeed,correlation between CNFD and IENF measurements was good(r ¼ 0.66) (Fig. 1D).

Interestingly, the cornea did not show any obvious pathologicalchange even in the group of animals with the most severe nervedepletion. This could be due to the fact that even a limited amountof spared fibers may be sufficient to maintain corneal integrity.Alternatively, the relatively short observation time (two weeks)may not be sufficient for epithelial damage to develop. Although itis believed that extensive nerve damage is required to developneurotrophic keratitis, the exact threshold is still unknown. In thisvein, we observed that neurotrophic keratitis developed in miceafter complete ablation of corneal nerves (Ferrari et al., 2011).

These findings support our hypothesis that the cornea may be auseful “window” to image peripheral nerves. In other words,corneal axons might provide surrogate biomarkers for distal-limbskin biopsy in detecting early paclitaxel axonopathy. Additionally,doseeresponse relationships were even more linear in the corneathan the hindpaw, suggesting that corneal imaging may effectivelycapture peripheral nerve damage. The chief limitation of this studyis the use of ex vivo rather than in vivo imaging as CCM was notavailable for this animal study. In fact, given the increasing avail-ability of repeatable, non-invasive, CCM imaging of patients, ourdata suggest that this technique be considered for prospective

comparison with distal-limb skin biopsy measurements in patientswith sensory polyneuropathies.

Author contributions

All authors concurred in the experimental, data analysis,manuscript writing and approved the final version.

Funding

Supported in part by the Public Health Service (NINDS K24-NS059892 and NEI RO1-EY20889). Presented in abstract form atthe annual meetings of the Association for Research in Vision andOphthalmology and the American Neurologic Association.

References

Bonini, S., Rama, P., Olzi, D., Lambiase, A., 2003. Neurotrophic keratitis. Eye (Lond.)17 (8), 989e995.

England, J.D., Gronseth, G.S., Franklin, G., et al., 2009. Practice parameter: evaluationof distal symmetric polyneuropathy: role of autonomic testing, nerve biopsy,and skin biopsy (an evidence-based review). Report of the American Academyof Neurology, American Association of Neuromuscular and ElectrodiagnosticMedicine, and American Academy of Physical Medicine and Rehabilitation.Neurology 72, 177e184.

Ferrari, G., Gemignani, F., Macaluso, C., 2010. Chemotherapy-associated peripheralsensory neuropathy assessed using in vivo corneal confocal microscopy. Arch.Neurol. 67, 364e365.

Ferrari, G., Chauhan, S.K., Ueno, H., Nallasamy, N., Gandolfi, S., Borges, L., Dana, R.,2011. A novel mouse model for neurotrophic keratopathy: trigeminal nervestereotactic electrolysis through the brain. Invest. Ophthalmol. Vis. Sci. 52 (5),2532e2539.

Flatters, S.J., Bennett, G.J., 2006. Studies of peripheral sensory nerves in paclitaxel-induced painful peripheral neuropathy: evidence for mitochondrial dysfunc-tion. Pain 122, 245e257.

Gemignani, F., Ferrari, G., Vitetta, F., Giovanelli, M., Macaluso, C., Marbini, A., 2010.Non-length-dependent small fibre neuropathy. Confocal microscopy study ofthe corneal innervation. J. Neurol. Neurosurg. Psychiatry 81, 731e733.

Gradishar, W.J., Tjulandin, S., Davidson, N., et al., 2005. Phase III trial of nanoparticlealbumin-bound paclitaxel compared with polyethylated castor oil-basedpaclitaxel in women with breast cancer. J. Clin. Oncol. 23, 7794e7803.

Mimura, T., Amano, S., Fukuoka, S., et al., 2008. In vivo confocal microscopy ofhereditary sensory and autonomic neuropathy. Curr. Eye Res. 33, 940e945.

Quattrini, C., Tavakoli, M., Jeziorska, M., et al., 2007. Surrogate markers of small fiberdamage in human diabetic neuropathy. Diabetes 56, 2148e2154.