Corneal and Epidermal Nerve Quantification in Chemotherapy Induced Peripheral Neuropathy.

Introduction

Chemotherapy-induced neurotoxicity is one of the most common side effects of cancer treatment. It has been recognized as a major determinant of functional impairment, increased pain, and reduced quality of life ( 1 , 2 ). Its prevalence reaches 68% in the first month following chemotherapy, and persists after more than 6 months in one-third of them ( 2 ). Chemotherapeutic agents that are routinely used to treat breast, lung, ovarian, bladder, and colon cancers are associated with neurotoxicity. Paclitaxel, for instance, can induce mild to moderate neuropathy in 93% of treated patients, which progresses to severe in 25% of them ( 3 – 5 ). Likewise, platinum compounds lead to chemotherapy-induced peripheral neuropathy (CIPN) in up to 50% of the patients ( 6 – 8 ). CIPN has been reported also following Bortezomib, with a prevalence ranging from 38 to 53%, and after thalidomide, with a frequency of 63% ( 8 , 9 ). Similarly, cyclophosphamide has been reported to have neurotoxic effects on both younger and elderly patients ( 7 , 10 ).

The diagnosis of CIPN is currently based on clinical testing, supported by neurophysiological studies and/or skin biopsy to evaluate epidermal nerve density. Although skin biopsy is a minimally invasive technique, it is not well accepted by some patients and it is not an option when frequent and/or repeated evaluations are needed ( 11 ).

In vivo corneal confocal microscopy (IVCM) has been successfully used to diagnose and stage peripheral neuropathies -specifically diabetic- showing higher sensitivity than skin biopsy ( 12 – 14 ).

Proper assessment and early diagnosis of peripheral neuropathy is crucial to allow prompt treatment modulation or interruption in most severe cases ( 1 ). In this paper, we propose that the combination of corneal nerve density and morphology assessed by IVCM is useful to diagnose and monitor the neurotoxic effects of chemotherapy.

Methods

Patients

Ninety five patients, 63 males and 32 females, were enrolled for the study. The average age was 59 years, with the youngest patient being 31 and the oldest 82 years old. Inclusion criteria were as follows: age > 18 years; exposure to a regimen of neurotoxic chemotherapeutic agents (Paclitaxel, platinum compounds, Bortezomib-Thalidomide-Dexamethasone (VTD), or Cyclophosphamide-combined treatments); expected survival of at least 6 months. Exclusion criteria were: prior neurotoxic therapy; diabetes; pre-existing peripheral neuropathy; corneal disease/surgery; contact lens wearing or eye drop application. Patients who met inclusion/exclusion criteria and signed the informed consent were recruited from the Oncology and Hematology Departments of the San Raffaele Hospital between 2011 and 2016. The study was carried out in accordance with the guidelines established by the Declaration of Helsinki, and the Institutional Review Board/Ethics committee (Comitato Etico Istituto Scientifico Ospedale San Raffaele) approval was obtained.

Clinical and Imaging Parameters

Neurological examination, including clinical Total Neuropathy Score (cTNS) ( 15 ) and IVCM evaluation, were performed before and within 4 weeks after completion of chemotherapy. Briefly, cTNS is a comprehensive parameter that includes signs, symptoms and objective testings. In particular, scores are attributed according to the patient's motor, sensory and autonomic abilities, and neurological signs of muscle power, reflexes and pain sensitivity are also evaluated. Then, the change in cTNS (ΔcTNS) observed after chemotherapy was calculated as follows: ΔcTNS = cTNS after – cTNS before.

Cumulative dose was obtained by multiplying the selected dose by the number of cycles received from the patients.

Corneal confocal images were taken using a Heidelberg Retina Tomograph III, equipped with the Rostock Cornea Module (Heidelberg Engineering, Heidelberg, Germany) as previously described ( 16 ). Corneal confocal images (total number: 14.716 images) were normalized and analyzed by applying the deep learning technique Convolutional Neural Network (CNN), which includes Gabor filtering to enhance nerve visibility ( 17 ). Corneal Nerve Fiber length was determined as cumulative length of the corneal nerves per unit of area of the cornea (mm/mm 2 ). Corneal nerve tortuosity was calculated by using the short-range tortuosity algorithm, as previously described ( 18 , 19 ). Short-range tortuosity is defined as frequent, small-amplitude directional changes in nerves ( 20 , 21 ). A ratio was calculated, as previously described ( 22 ), by dividing CNF length with corneal nerve tortuosity, to have a more comprehensive vision on patient's parameters.

Skin biopsy was collected from the lateral-malleolar region 10 cm upper from the malleolus in 14 random-selected patients who signed the informed consent before and after treatment. Epidermal nerve fibers were counted manually, and density was obtained by dividing the number of nerve fibers over the epidermal length.

Statistics

To test differences in nerve parameters before and after treatment, Mann-Whitney and Spearman's non-parametric tests were used for continuous and ordinal variables. One-sample t -test was used to compare ΔTNS among the experimental groups. Two-sided p < 0.05 was considered statistically significant for all comparisons. Results are presented as mean ± standard error of the mean (SEM). The statistical analysis was performed using GraphPad Prism 8.0 (GraphPad Software, La Jolla, CA).

Results

Clinical and IVCM Neuropathy Assessment

Ninety-five patients, including the 14 that underwent epidermal nerve examination, were recruited. Details of cancer types are listed in Table 1 . Of them, 22 were lost at the post-treatment follow-up; therefore, 73 patients were considered for the analysis. Thirty two patients fully completed the cTNS examination before and after chemotherapy. All the patients received a complete corneal examination. In those patients, cTNS significantly worsened from the first to last visit (+62%, p < 0.0001, Figure 1A ). cTNS worsened significantly after treatment with Paclitaxel ( p = 0.0029), Platinum compounds ( p = 0.0138), and in combination with Cyclophosphamide ( p = 0.0428), while no significant difference was found in the VTD group ( p = 0.1872, Figure 1B ). CNF length ( p = 0.57, Figure 1C ) and corneal nerve short range tortuosity ( p = 0.1350, Figure 1D ) showed no significant difference. Intriguingly, however, combining the two parameters by calculating their ratio (CNF length/tortuosity), resulted in a significant decrease after chemotherapy ( p = 0.0333, Figure 1E ).

Further analysis revealed that cumulative dose of patients receiving paclitaxel, which was the only group having an adequate number of patients, was correlated with CNF length/tortuosity reduction ( r = −4174, p = 0.0156, Figure 1F ). Similarly, the number of chemotherapy cycles was correlated with CNF length/tortuosity reduction ( r = −0.04790, p = 0.0009, Figure 1G ). Subgroup correlation analysis for paclitaxel confirmed the previous result, indicating that an increasing number of cycles corresponded to worse corneal parameters ( r = 0.2024, p = 0.0076, Figure 1H ).

Dissection of Corneal Parameters in Single Drug Groups

Following initial analysis on the entire group of patients, we considered for statistics the single drug groups of most neurotoxic agents. Even though patients receiving paclitaxel, platinum compounds and cyclophosphamide in combination with other drugs showed a decrease in CNFL, data did not reach statistical significance ( Figure 2A ). Similarly, tortuosity was increased in all groups, even though not significantly due to high variability ( Figure 2B ). CNFL/tortuosity ratio values were reduced in all drug groups, but it was only significant in paclitaxel group ( p = 0.0023, Figure 2C ).

Nerve Morphology Quantification in the Epidermis and the Cornea

A subgroup of 14 patients underwent both epidermal and corneal nerve examination. Patients' demographics are shown in Table 2 . Clinical grading of neuropathy worsened significantly after chemotherapy initiation (+69% cTNS; p = 0.0018, Figure 3A ). Yet, no changes in epidermal nerve density were detected ( p = 0.4618, Figures 3B,F ). Interestingly, however, corneal nerve density was significantly reduced (−22%, p = 0.0457, Figures 3C,G ). No significant differences were observed in corneal nerve tortuosity before and after chemotherapy ( p = 0.6347, Figure 3D ). A reduction was observed, although not significant, in the CNF density/tortuosity ratio following chemotherapy ( p = 0.1251, Figure 3E ).

Discussion

The cornea is the most densely innervated tissue of the body; therefore, it is ideally suited to study nerve pathology. We previously showed that corneal and epidermal nerve density show good correlation in an animal model of paclitaxel-induced neuropathy ( 23 ). Recently, in vivo corneal confocal microscopy (IVCM) has gained popularity because it is an in vivo , non-invasive, repeatable imaging technique that allows detailed examination of corneal nerves ( 24 , 25 ). In fact, IVCM has been used to evaluate peripheral nerve damage following chemotherapy with neurotoxic agents such as oxaliplatin, although no association with epidermal nerve density was provided ( 26 – 28 ), and paclitaxel ( 28 ).

Emerging evidence demonstrates the potential diagnostic and prognostic power of IVCM in peripheral neuropathies ( 12 – 14 ); still the clinical utility of this technique is matter of debate. For instance, it has been suggested that IVCM is not useful when CIPN is mild to moderate, at least when corneal nerve length and density are measured ( 29 ).

Here, we provide for the first time a comprehensive study, analyzing patients with several different types of cancer and undergoing different pharmacological treatments (paclitaxel, platin, VTD, and cyclophosphamide-combined). We show that neuropathy clinical grading (cTNS) worsens following chemotherapy, supporting that these medications are neurotoxic. In particular, ΔTNS was significantly affected after treatment with paclitaxel, platinum compounds and cyclophosphamide in combination with other drugs. Unexpectedly, the peripheral neuropathy severity of cyclophosphamide group was similar to the other more neurotoxic drugs. This could be a consequence of the combined therapy with other neurotoxic compound (i.e., paclitaxel or platinum compounds).

Interestingly, epidermal nerve length was not reduced in the subgroup where it was measured. This was unexpected and may be a consequence of the limited sample of patients who accepted to undergo biopsy, and/or to the inter- and intra-subject variability of such a marker ( 30 ). It is worth noting that IVCM showed an increased sensitivity in detecting peripheral neuropathy than quantification of epithelial nerves in our cohort. Indeed, tortuosity was increased in all drug groups, even though not significantly, and similarly CNFL was reduced in paclitaxel, platinum compounds and cyclophosphamide groups. Moreover, IVCM has the advantage of being non-invasive and repeatable, which makes it a more suitable option in CIPN patients.

Since reduced density and increased tortuosity of corneal nerves are hallmarks of peripheral neuropathy ( 12 – 14 , 21 ), we decided to combine both parameters by calculating their ratio. Interestingly, the CNF density/tortuosity ratio, previously used to successfully diagnose neuropathy in diabetic patients ( 22 ), was more effective in detecting the alterations in corneal nerves induced by chemotherapy than the two parameters separately. In particular, paclitaxel group, which was the therapy administered more frequently, showed more severe nerve loss, and correlated with the cumulative dose received from the patient. One cross sectional study ( 29 ) did not find nerve parameters useful to detect peripheral neuropathy. Our prospective study analyzed corneal nerve morphology before and after treatment, which we believe brings advantages over cross-sectional design, because high corneal nerve density variability in humans could impact the ability to detect between group differences. Other authors have shown that CNF length was increased following chemotherapy in a small cohort of patients. However, the study only included upper gastrointestinal cancer patients, a small number of which attending the follow-up, and considered a small sample (3 pictures per eye), which were manually counted ( 31 ).

Finally, we observed the CNF density/tortuosity ratio was inversely correlated with the number of chemotherapy cycles, which suggests that it can capture cumulative drug toxicity.

While further studies are needed to elucidate which morphological parameters better parallel clinical signs and symptoms, our results suggest that measurement of corneal nerve density and tortuosity by IVCM shows promise for the detection of peripheral neuropathy in patients undergoing neurotoxic chemotherapy.

Data Availability Statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethics Statement

The studies involving human participants were reviewed and approved by the Comitato Etico Istituto Scientifico Ospedale San Raffaele. The patients/participants provided their written informed consent to participate in this study.

Author Contributions

NR, FB, RL, and GF: drafting the work and revising it critically for important intellectual content. FB, RL, PF, MB, FS, AR, and IL: organized the database, analysis of data, and performed the statistical analysis. GF, MR, TD, IL, AQ, NR, FA, PR, and MM: acquisition of data. GF: provide final approval for publication of the content. All authors contributed to manuscript revision, read, and approved the submitted version.

Funding

This study has been funded by the Italian Department of Health (grant code: GR-2010-2319274).

Conflict of Interest

AR was employed by Resono Ophthalmic srl. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher's Note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

References

1. Banach M Juranek JK Zygulska AL . Chemotherapy-induced neuropathies—a growing problem for patients and health care providers . Brain Behav. ( 2017 ) 7 : e00558 . 10.1002/brb3.558 28127506 PMC5256170

2. Seretny M Currie GL Sena ES Ramnarine S Grant R Macleod MR . Comprehensive review incidence, prevalence, and predictors of chemotherapy-induced peripheral neuropathy: a systematic review and meta-analysis . Pain. ( 2014 ) 155 : 2461 - 70 . 10.1016/j.pain.2014.09.020 25261162

3. Gradishar WJ Tjulandin S Davidson N Shaw H Desai N Bhar P . Phase III trial of nanoparticle albumin-bound paclitaxel compared with polyethylated castor oil-based paclitaxel in women with breast cancer . J Clin Oncol. ( 2005 ) 23 : 7794 – 803 . 10.1200/JCO.2005.04.937 16172456

4. Argyriou AA Koltzenburg M Polychronopoulos P Papapetropoulos S Kalofonos HP . Peripheral nerve damage associated with administration of taxanes in patients with cancer . Crit Rev Oncol/Hematol. ( 2008 ) 66 : 218 – 28 . 10.1016/j.critrevonc.2008.01.008 18329278

5. Antoine JC Camdessanché JP . Peripheral nervous system involvement in patients with cancer . Lancet Neurol. ( 2007 ) 6 : 75 – 86 . 10.1016/S1474-4422(06)70679-2 17166804

6. Lazić A Popović J Paunesku T Woloschak GE Stevanović M . Insights into platinum-induced peripheral neuropathy–current perspective . Neural Regener Res. ( 2020 ) 15 : 1623 . 10.4103/1673-5374.276321 32209761 PMC7437596

7. Nagarajan R Peters C Orchard P Rydholm N . Report of severe neurotoxicity with cyclophosphamide . J Pediatr Hematol Oncol. ( 2000 ) 22 : 544 – 6 . 10.1097/00043426-200011000-00016 11132226

8. Chan A Hertz DL Morales M Adams EJ Gordon S Tan CJ . Biological predictors of chemotherapy-induced peripheral neuropathy (CIPN): MASCC neurological complications working group overview . Support Care Cancer. ( 2019 ) 27 : 3729 – 37 . 10.1007/s00520-019-04987-8 31363906 PMC6728179

9. Areti A Yerra VG Naidu VGM Kumar A . Oxidative stress and nerve damage: Role in chemotherapy induced peripheral neuropathy . Redox Biol. ( 2014 ) 2 : 289 – 95 . 10.1016/j.redox.2014.01.006 24494204 PMC3909836

10. Kuhlen M Bleckmann K Möricke A Schrappe M Vieth S Escherich G . Neurotoxic side effects in children with refractory or relapsed T-cell malignancies treated with nelarabine based therapy . Br J Haematol. ( 2017 ) 179 : 272 – 83 . 10.1111/bjh.14877 28771662

11. Lauria G . Small fibre neuropathies . Curr Opin Neurol. ( 2005 ) 18 : 591 – 7 . 10.1097/01.wco.0000177330.35147.70 16155446

12. Alam U Jeziorska M Petropoulos IN Asghar O Fadavi H Ponirakis G . Diagnostic utility of corneal confocal microscopy and intra-epidermal nerve fibre density in diabetic neuropathy . PLoS ONE. ( 2017 ) 12 : e0180175 . 10.1371/journal.pone.0180175 28719619 PMC5515394

13. Chen X Graham J Dabbah MA Petropoulos IN Ponirakis G Asghar O . Small nerve fiber quantification in the diagnosis of diabetic sensorimotor polyneuropathy: comparing corneal confocal microscopy with intraepidermal nerve fiber density . Diab Care. ( 2015 ) 38 : 1138 – 44 . 10.2337/dc14-2422 25795415 PMC4439535

14. Dhage S Ferdousi M Adam S Ho JH Kalteniece A Azmi S . Corneal confocal microscopy identifies small fibre damage and progression of diabetic neuropathy . Sci Rep. ( 2021 ) 11 : 1 – 9 . 10.1038/s41598-021-81302-8 33479291 PMC7820596

15. Cavaletti G Frigeni B Lanzani F Piatti M Rota S Briani C . The total neuropathy score as an assessment tool for grading the course of chemotherapy-induced peripheral neurotoxicity: comparison with the National Cancer Institute-Common Toxicity Scale . J Peripher Nerv Syst. ( 2007 ) 12 : 210 – 5 . 10.1111/j.1529-8027.2007.00141.x 17868248

16. Ferrari G Grisan E Scarpa F Fazio R Comola M Quattrini A . Corneal confocal microscopy reveals trigeminal small sensory fiber neuropathy in amyotrophic lateral sclerosis . Front Aging Neurosci. ( 2014 ) 6 : 278 . 10.3389/fnagi.2014.00278 25360111 PMC4199282

17. Colonna A Scarpa F Ruggeri A . Segmentation of corneal nerves using a U-Net-based convolutional neural network . In: Stoyanov D Taylor Z Ciompi F Xu Y Martel A Maier-Hein L editors. Computational Pathology and Ophthalmic Medical Image Analysis. Cham : Springer International Publishing ( 2018 ). p. 185 – 92 .

18. Scarpa F Colonna A Ruggeri A . Multiple-image deep learning analysis for neuropathy detection in corneal nerve images . Cornea. ( 2020 ) 39 : 342 – 7 . 10.1097/ICO.0000000000002181 31658167

19. Scarpa F Ruggeri A . Development of clinically based corneal nerves tortuosity indexes . In: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Cham : Springer Science and Business Media Deutschland GmbH ( 2017 ). p. 219 – 26 .

20. Lagali N Poletti E Patel DV McGhee CNJ Hamrah P Kheirkhah A . Focused tortuosity definitions based on expert clinical assessment of corneal subbasal nerves . Investigat Ophthalmol Vis Sci. ( 2015 ) 56 : 5102 – 9 . 10.1167/iovs.15-17284 26241397

21. Kallinikos P Berhanu M O'Donnell C Boulton AJM Efron N Malik RA . Corneal nerve tortuosity in diabetic patients with neuropathy . Invest Ophthalmol Vis Sci. ( 2004 ) 45 : 418 – 22 . 10.1167/iovs.03-0637 14744880

22. Edwards K Pritchard N Vagenas D Russell A Malik RA Efron N . Standardizing corneal nerve fibre length for nerve tortuosity increases its association with measures of diabetic neuropathy . Diab Med. ( 2014 ) 31 : 1205 – 9 . 10.1111/dme.12466 24750318

23. Ferrari G Nalassamy N Downs H Dana R Oaklander AL . Corneal innervation as a window to peripheral neuropathies . Exp Eye Res. ( 2013 ) 113 : 148 . 10.1016/j.exer.2013.05.016 23769950 PMC3737766

24. Smith AG Kim G Porzio M Allen B Koach M Mifflin M . Corneal confocal microscopy is efficient, well-tolerated, and reproducible . J Peripher Nerv Syst. ( 2013 ) 18 : 54 – 8 . 10.1111/jns5.12008 23521645

25. Kalteniece A Ferdousi M Adam S Schofield J Azmi S Petropoulos I . Corneal confocal microscopy is a rapid reproducible ophthalmic technique for quantifying corneal nerve abnormalities . PLoS ONE. ( 2017 ) 12 : e0183040 . 10.1371/journal.pone.0183040 28817609 PMC5560560

26. Campagnolo M Lazzarini D Fregona I Cacciavillani M Bergamo F Parrozzani R . Corneal confocal microscopy in patients with oxaliplatin-induced peripheral neuropathy . J Peripher Nerv Syst. ( 2013 ) 18 : 269 – 71 . 10.1111/jns5.12036 24028197

27. Maddaloni E Sabatino F . In vivo corneal confocal microscopy in diabetes: where we are and where we can get . World J Diab. ( 2016 ) 7 : 406 . 10.4239/wjd.v7.i17.406 27660697 PMC5027004

28. Chiang JCB Goldstein D Trinh T Au K Mizrahi D Muhlmann M . A cross-sectional study of sub-basal corneal nerve reduction following neurotoxic chemotherapy . Transl Vis Sci Technol. ( 2021 ) 10 : 1 – 11 . 10.1167/tvst.10.1.24 33510963 PMC7804570

29. Bennedsgaard K Ventzel L Andersen NT Themistocleous AC Bennett DL Jensen TS . Oxaliplatin-and docetaxel-induced polyneuropathy: clinical and neurophysiological characteristics . J Peripher Nerv Syst. ( 2020 ) 25 : 377 - 87 . 10.1111/jns.12413 32902058 PMC7756561

30. Selim MM Wendelschafer-Crabb G Hodges JS Simone DA Foster SXYL Vanhove GF . Variation in quantitative sensory testing and epidermal nerve fiber density in repeated measurements . Pain. ( 2010 ) 151 : 575 – 81 . 10.1016/j.pain.2010.06.034 20851518

31. Ferdousi M Azmi S Petropoulos IN Fadavi H Ponirakis G Marshall A . Corneal confocal microscopy detects small fibre neuropathy in patients with upper gastrointestinal cancer and nerve regeneration in chemotherapy induced peripheral neuropathy . PLos ONE. ( 2015 ) 10 : e0139394 . 10.1371/journal.pone.0139394 26430773 PMC4592260