Corneal Nerve Regeneration After Collagen
Cross-Linking Treatment of Keratoconus A
5-Year Longitudinal Study
Marlen Parissi, Stefan Randjelovic, Enea Poletti, Pedro Guimaraes, Alfredo Ruggeri, Sofia
Fragkiskou, Thu Ba Wihlmark, Tor Paaske Utheim and Neil Lagali
Linköping University Post Print
N.B.: When citing this work, cite the original article.
Original Publication:
Marlen Parissi, Stefan Randjelovic, Enea Poletti, Pedro Guimaraes, Alfredo Ruggeri, Sofia
Fragkiskou, Thu Ba Wihlmark, Tor Paaske Utheim and Neil Lagali, Corneal Nerve
Regeneration After Collagen Cross-Linking Treatment of Keratoconus A 5-Year Longitudinal
Study, 2016, JAMA ophthalmology, (134), 1, 70-78.
http://dx.doi.org/10.1001/jamaophthalmol.2015.4518
Copyright: American Medical Association (AMA)
http://jama.jamanetwork.com/journal.aspx
Postprint available at: Linköping University Electronic Press
http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-127289
Corneal Nerve Regeneration after Collagen Cross-linking for Keratoconus: a
1
Five Year Longitudinal Study
2
Marlen Parissi MSc1,2, Stefan Randjelovic MSc2, Enea Poletti PhD3, Pedro Guimarães PhD3, 3
Alfredo Ruggeri PhD3, Sofia Fragkiskou MD4, Thu Ba Wihlmark MD4, Tor Paaske Utheim MD 4
PhD1,2,5, and Neil Lagali PhD4 5
6
1Department of Medical Biochemistry, Oslo University Hospital; and University of Oslo, Oslo,
7
Norway 8
2The Norwegian Dry Eye Clinic, Oslo, Norway
9
3Department of Information Engineering, University of Padova, Padova, Italy
10
4Department of Ophthalmology, Institute for Clinical and Experimental Medicine, Linköping
11
University, Linköping, Sweden 12
5Department of Oral Biology, Faculty of Dentistry, University of Oslo, Norway
13 14 15
Abstract word count: 343 16
Word count (excl. abstract): 2998 17
18
None of the authors have any proprietary/financial interest to disclose. 19
No conflict-of-interest relationship exists for any author. 20
21 22
Running head: Subbasal nerve architecture in keratoconus 23
24
Sources of Funding: Funding from the Swedish Research Council and Princess Margareta’s 25
Foundation for the Visually Impaired to NL, and funding from the Norwegian Research 26
Council to MP. The funding organizations had no role in the design or conduct of this 27
research. 28
29 30
Corresponding author and address for reprints: 31
32
Neil Lagali, PhD 33
Department of Clinical and Experimental Medicine – Ophthalmology 34
Faculty of Health Sciences 35 Linköping University 36 581 85 Linköping, Sweden 37 neil.lagali@liu.se 38 Tel +46 10 1034658 39 Fax +46 10 1033065 40 41
ABSTRACT
42Importance: It is unknown whether a neurotrophic deficit or pathologic nerve morphology
43
persists in keratoconus in the long term after corneal collagen cross-linking (CXL) treatment. 44
Nerve pathology could impact long-term corneal status in keratoconus patients. 45
Objective: To determine whether CXL treatment for keratoconus results in normalization of
46
subbasal nerve density and architecture up to five years after treatment. 47
Design: Observational study of keratoconus patients with longitudinal follow-up to five
48
years postoperatively (2009-2015), including healthy comparator group. 49
Setting: Primary care center, university hospital ophthalmology department.
50
Participants: Nineteen consecutive patients with early-stage keratoconus indicated for a
51
first CXL treatment and nineteen age-matched healthy volunteers. 52
Exposures: Keratoconus patients underwent standard epithelial-off UVA-riboflavin CXL
53
treatment with 30 min UVA exposure at 3 mW/cm2 irradiance. 54
Main Outcome Measures: Central corneal subbasal nerve density and subbasal nerve
55
architecture by laser-scanning in vivo confocal microscopy. Subbasal nerve analysis by two 56
masked observers, a fully-automated method, and wide-field mosaics of subbasal nerve 57
architecture by an automated method. Ocular surface touch sensitivity by contact 58
esthesiometry. 59
Results: Relative to healthy corneas, subbasal nerve density in stage I-II keratoconus was
60
reduced 51-56% (mean difference 10.7 mm/mm2,95% CI: 6.8 to 14.6 mm/mm2, t-test, P < 61
0.001). After CXL, nerves continued to regenerate up to five years, but nerve density 62
remained significantly reduced relative to healthy corneas at final follow-up (mean 63
reduction of 8.5 mm/mm2,95% CI: 4.7 to 12.4 mm/mm2, t-test, P < 0.001) despite recovery 64
of touch sensitivity to normal levels by 6 months. Preoperatively, more frequent nerve loops 65
(P < 0.001), crossings (P = 0.03), and greater crossing angles (P = 0.02) were observed, 66
relative to healthy corneas. Postoperatively, nerve looping frequency increased, crossings 67
were more frequent, and nerve tortuosity increased. Wide-field mosaics indicated 68
persistent disrupted orientation of the regenerating subbasal nerves five years after CXL. 69
Conclusions and Relevance: Keratoconus is characterized by a neurotrophic deficit and
70
altered nerve morphology that CXL treatment does not address, despite providing a positive 71
biomechanical effect in the stroma. Given the widespread use of CXL in keratoconus 72
management, progression of abnormal innervation post-CXL should be recognized. 73
74
Introduction
75
Corneal collagen cross-linking (CXL) has emerged as a promising treatment to strengthen 76
the cornea in conditions such as corneal ectasia and keratoconus.1 Results from longer-term 77
clinical studies2-7 suggest a lasting benefit of CXL treatment in halting the progression of 78
keratoconus, thereby avoiding the need for transplantation. At the tissue level, knowledge 79
of the effect of cross-linking has been gained from rabbit studies,8-11 and use of in vivo 80
confocal microscopy (IVCM) in patients.3,12-21 Patient investigations have revealed not only a 81
cross-linking effect in the corneal stroma, but also an effect of the procedure on corneal 82
epithelial nerves.12, 14-17, 19-21 In epithelium-off CXL, epithelial nerves are completely removed 83
in the treatment zone, typically an 8-9 mm diameter region of the central cornea. Analysis 84
of the subbasal nerve plexus by IVCM has indicated gradual regeneration of these nerves 85
postoperatively.15,19-21 Nerve regeneration is important for re-establishment of a healthy 86
epithelium, protective blink reflex and trophic effects on the corneal stroma.22 Corneal 87
nerves have also been postulated to have a role in the development of keratoconus.23 88
Regeneration of subbasal nerves after CXL has been shown to occur within the first 89
postoperative year,12, 14, 19-21 but the long-term effect of CXL on corneal nerves has not been 90
reported. It is unknown if corneal nerves reach equilibrium after the first year, whether they 91
continue to regenerate over time, or if the reduced nerve density in keratoconus23-28 can 92
improve after CXL. It is therefore of interest to investigate whether CXL can restore a 93
healthy subbasal nerve density to the keratoconic cornea in the long term, or if a nerve 94
deficit persists despite clinical success of the treatment. CXL is a relatively new treatment 95
often given to young patients, whereas long-term clinical consequences such as a potential 96
neurotrophic deficit may take decades to manifest. 97
In addition to reduced subbasal nerve density, several reports have indicated disrupted 98
subbasal nerve patterns in keratoconus, including tortuous, branching, and looping 99
patterns.20, 23, 25-27, 29 It is not known, however, how prevalent such patterns are in healthy 100
corneas or if CXL can influence these patterns (and by proxy the neurotrophic status) in the 101
regenerated nerve plexus. Because subbasal nerve guidance is closely linked with epithelial 102
cell migration,30-32 subbasal nerves can mirror the epithelial status, which has been shown to 103
be pathologic in keratoconus.23, 26, 27 104
To better understand the regenerative capacity of subbasal nerves in keratoconus and in 105
response to CXL treatment, a prospective study was conducted in a young patient 106
population with early stage keratoconus undergoing CXL treatment. 107
Methods
108
Subjects and Examinations
109
Prior to recruitment, ethical approval was obtained from the Linköping Regional Human 110
Ethics Review Committee. All study subjects gave voluntary informed consent to participate, 111
and the study followed the tenets of the Declaration of Helsinki. Patients were included if 112
they had documented progressive keratoconus over at least two clinic visits within a 12-113
month period, defined by decrease in uncorrected visual acuity ≥ 0.1 (decimal), increase in 114
astigmatism ≥ 1D, increase in max K reading ≥ 1D, decrease in minimum corneal thickness 115
(MCT) ≥ 20 µm or combination thereof, in sequential examinations made by an 116
ophthalmologist and/or optometrist. Those with preoperative MCT below 400 µm were 117
excluded. Persons under 18 years of age, those with other ocular pathology or prior ocular 118
surgery, dry eye symptoms, diabetics, and pregnant women were also excluded from the 119
study. 120
Preoperative examination included determination of uncorrected and best spectacle-121
corrected visual acuity (BSCVA), measurement of MCT by ultrasound pachymetry (UP; 122
Tomey SP-2000, Japan) and anterior segment optical coherence tomography 123
(ASOCT;Visante®, Carl Zeiss Meditec, Jena, Germany), topographic measurement (Orbscan 124
II; Bausch & Lomb, Rochester, NY, USA), and in vivo confocal microscopy (IVCM; HRT3-RCM, 125
Heidelberg Engineering, Heidelberg, Germany). 126
Study visits were conducted on seven separate occasions: preoperative, 1-6 m, 7-12 m, 13-127
24 m, 25-36 m, 37-48 m, and 49-60 m postoperative. Postoperatively, IVCM, ASOCT, 128
topography, and refraction were performed. Additionally at the final postoperative visit, 129
Schirmer’s test for tear production (without anesthesia) and the tear break-up time test 130
were performed. Ocular surface sensitivity was measured by contact esthesiometry (Cochet-131
Bonnet; Luneau Ophthalmlogie, Chartres, France) preoperatively and at the 3, 6, and 12 132
month postoperative visits and at the final study visit. 133
Additionally a comparison group of age-matched healthy subjects was recruited. After 134
obtaining informed consent, general medical status was taken and a full ophthalmic 135
examination (including refraction, slit lamp biomicroscopy, ASOCT, and intraocular pressure 136
measurement) was conducted to exclude systemic or ocular pathology. Only asymptomatic, 137
healthy subjects with a clear cornea on slit lamp examination were included. IVCM and 138
ASOCT examinations were performed for this group. 139
UVA-Riboflavin Collagen Cross-linking Treatment (Epithelium-off Method)
140
Standard epithelium-off CXL was performed as follows. The epithelium was removed in an 141
8-9 mm diameter central zone using alcohol. Riboflavin 0.1% with 20% dextran or a 142
hypotonic riboflavin 0.1% solution was given topically, one drop every three minutes for 30 143
min (hypotonic solution for MCT < 430 µm). After confirming penetration of riboflavin into 144
the anterior chamber, UVA irradiation was applied at 5 cm distance from the corneal surface 145
with a 9 mm aperture for 30 minutes, during which time one drop of riboflavin was 146
administered every three minutes. Preoperatively, the UVA source (with potentiometric 147
voltage regulator; UV-X, IROC AG, Zürich, Switzerland) was calibrated (UV Light Meter, 148
Model: YK-34UV, Lutron Electronic Enterprise Co., Ltd. Taipei, Taiwan) to give 3.0 mW/cm2 149
at the corneal surface at 365 nm wavelength. 150
After treatment, patients received topical antibiotics (Oftaquix 5mg/ml, SantenPharma AB, 151
Solna, Sweden) 4 times daily for 7 days. Starting day 5 postoperatively, dexamethasone 152
(Maxidex 0.1%, Alcon, Stockholm, Sweden) was applied 3 times daily for 3 weeks. Patients 153
were also given analgesics (e.g., acetaminophen and diclofenac) and tear substitutes (e.g., 154
Viscotears, Laboratoires Thea, Clermont-Ferrand, France). 155
In Vivo Confocal Microscopy
156
IVCM was performed according to an established protocol.33 A motorized joystick was used 157
to locate the subbasal nerve plexus layer, and images were acquired in sequence scan mode 158
as the field of view was scanned over the subbasal nerve plexus. Two experienced observers 159
selected images of subbasal nerves based on an earlier protocol33 taking into account 160
contrast, absence of artifacts, no overlap and central location. Three images meeting these 161
criteria were selected randomly for each subject and time point, and were coded to mask 162
subject group and postoperative time. The resulting set of images was used for manual and 163
automated nerve tracing analysis. For manual analysis, nerves were traced independently by 164
the observers using NeuronJ,34 and main subbasal nerve crossings (excluding thinner 165
secondary branches) were defined as two nerve branches continuing in an unaltered path 166
after intersection. The narrowest crossing angle was measured using the angle tool in the 167
software Fiji.35 Presence of nerve loops was noted, defined as main subbasal nerves with at 168
least 180° change in path direction within a single image frame. 169
Automated analysis consisted of fully automatic image pre-processing, nerve recognition 170
and tracing, and post-processing to remove false recognitions, all without human 171
intervention.33,36 Automated analysis yielded subbasal nerve density and tortuosity using a 172
previously reported index.37 173
Generation of Subbasal Nerve Mosaics
174
At final follow-up, IVCM data from six patients was used for wide-field mosaic 175
reconstruction. Mosaicking was performed by a fast, fully-automated algorithm described 176
previously.38 Briefly, the algorithm iteratively compared pairs of images to determine image 177
positioning in the mosaic space, and were registered by translation, rotation, and affinity 178
transformations. Blending based on pixel intensity weighting provided a merged mosaic 179
with homogeneous luminosity and contrast. 180
Statistical Analysis
181
The 95% limits of agreement for inter-observer and inter-method differences in subbasal 182
nerve density were determined by the Bland-Altman method.39 Frequency of nerve loops 183
across groups were tested with the z-test for proportions. MCT, nerve crossings and angles, 184
and nerve density between specific groups were compared with independent t-tests, and 185
the Mann-Whitney test for non-normal data. Tortuosity and time-dependence of nerve 186
density were assessed using one-way ANOVA on ranks with Dunn’s method for post-hoc 187
comparison. For longitudinal corneal sensitivity, one-way repeated measures ANOVA was 188
used with the Holm-Sidak post-hoc method. With the exception of post-hoc tests, a two-189
tailed alpha level of < 0.05 was considered significant. Statistics were performed using 190
SigmaStat for Windows (Systat Inc, Chicago, IL, USA). 191
Results
192
Patient Characteristics
193
Patient characteristics (eTable 1) indicated thinner corneas in the keratoconus group (P < 194
0.001) while astigmatism, MCT, and K readings in the patient cohort represented early-stage 195
keratoconus. Twelve keratoconus patients (63%) were classified as stage I, while seven 196
(37%) were stage II, according to the Amsler-Krumeich classification.40 197
Comparison of Subbasal Nerve Density
198
Preoperative subbasal nerve density in the keratoconus cohort was compared to healthy 199
age-matched subjects (Figure 1). Subbasal nerve density in early-stage keratoconus (10.3 ± 200
5.6 mm/mm2, mean ± SD) was reduced (by 51%) relative to the healthy, age-matched group 201
(21.0 ± 4.2 mm/mm2), yielding a mean difference of 10.7 mm/mm2 (95% CI: 6.8 to 14.6 202
mm/mm2, t-test, P < 0.001). Automated analysis similarly indicated reduced nerve density in 203
keratoconus, by 56% (8.9 ± 4.1 vs. 20.2 ± 3.6 mm/mm2; mean difference 11.3 mm/mm2, 204
95% CI: 8.3 to 14.3 mm/mm2 t-test, P < 0.001). 205
Inter-observer and inter-method comparisons of nerve density (eTable 2) revealed 206
over/underestimation of nerve density by the manual/automated method, which was more 207
pronounced in keratoconus subjects. Agreement between manual observers was stronger 208
(narrower limits of agreement) than between methods. 209
Regeneration of Subbasal Nerves after CXL Treatment
210
CXL procedures were completed without intra-operative complications. Each patient 211
attended a mean of 5.5 visits during the 0-66 month study period (attendance rate of 79%). 212
Longitudinal analysis of nerve regeneration corresponded to study visits arranged by 213
interval: preoperative, 1-6 m, 7-12 m, 13-24 m, 25-36 m, 37-48 m, and 49-66 m 214
postoperative. Nerve regeneration by manual and automated methods of analysis was time-215
dependent (P < 0.001 for both, Figure 2). Regardless of method, nerve density was reduced 216
up to 6 months, followed by an increase at 7-12 m (ANOVA, P < 0.001). At 7-12 m, nerve 217
density did not differ from preoperative; however, median nerve density increased up to 4 -218
5 years postoperative. By both analysis methods, final nerve density did not differ from 219
preoperative but remained reduced relative to healthy corneas (Manual: mean reduction 220
8.5 mm/mm2,95% CI: 4.7 to 12.4 mm/mm2, t-test, P < 0.001; Automated: 8.4 mm/mm2,95% 221
CI: 5.0 to 11.8 mm/mm2, t-test, P < 0.001). 222
Ocular surface sensitivity (Figure 2) was normal preoperatively (59 ± 3 mm), declined to 52 ± 223
13 mm at 3 months (P = 0.017), and recovered to preoperative, healthy levels at 6 months 224
(60 ± 0 mm), with no further change at 12 months or at five years relative to preoperative. 225
At final follow-up, tear production by the Schirmer test was 21 ± 6 mm in 5 min (range: 12 – 226
30 mm), and tear break-up time was 14 ± 4 sec (7 – 20 sec). 227
Subbasal Nerve Morphology
228
Preoperatively, reduced nerve density, nerve loops and crossings were evident (Figure 3). 229
Regenerated nerves also exhibited loops and crossings, some following tortuous paths. No 230
looping nerves and rare crossings were observed in healthy corneas, where dense nerves 231
had mainly parallel orientations (Figure 3). Nerve loops were present in 0% of images from 232
healthy subjects, 30% of preoperative images, and in 56% of images at final follow-up. A 233
greater proportion of looping nerves was present in the keratoconus corneas compared to 234
healthy corneas (P < 0.001, z-test). Crossings of main subbasal nerves were observed three 235
times more frequently in keratoconus than in healthy subjects (Figure 4). The mean number 236
of crossings per image frame was 0.27 for healthy subjects, 0.76 preoperatively (P = 0.03 237
relative to healthy), and 0.89 one year or longer post-CXL (P = 0.002). The mean crossing 238
angle of subbasal nerve trunks was 57° ± 18° in healthy subjects, 70° ± 15° preoperatively (P 239
= 0.02), and 65° ± 16° postoperatively. Tortuosity differed among healthy, preoperative, and 240
final follow-up (ANOVA P = 0.008; Figure 4) with an increase after the first year post-CXL 241
relative to healthy corneas. 242
Architecture of Regenerated Subbasal Nerves
243
At the five year follow-up, wide-field mosaics of the subbasal nerve plexus were constructed 244
in six patients (Figure 5). As standard epi-off CXL removes the subbasal nerve plexus while 245
leaving intact the nerve fiber bundles within and underneath Bowman’s layer, patterns of 246
nerve regeneration were examined by observing subbasal nerve paths starting at the 247
penetration points (Figure 5A, E, and F, black arrows) into the subbasal layer. Nerves 248
adopted radial, circumferential, or mixed orientations as they regenerated. Predominantly 249
circumferential paths were observed in Figure 5A and F while Figure 5B, C, D, and E depicted 250
all orientation types. Radial paths originated in the central cornea and were directed 251
towards the periphery in straight lines. Mixed paths alternated between radial and 252
circumferential orientations. Different orientation types appeared to give rise to the nerve 253
patterns observed in single-image analysis. Crossings (black arrowheads in Figure 5B and E) 254
were intersection points between radial and circumferential paths. Likewise, nerve loops 255
appeared as paths alternating between circumferential and radial (white arrowheads in 256
Figure 5B, C, E, and F). The dominance of one orientation over another appeared to give rise 257
to abrupt or more gradual directional changes, resulting in sharp (Figure 5F) or smooth 258
(Figure 5C and E) looping structures. Highly tortuous regenerated nerves were also 259
apparent, representing frequent path alternations on a smaller scale than those giving rise 260
to nerve loops (white arrows in Figure 5A, D, and E). 261
Effect of Contact Lens Wear on Nerve Parameters
262
Four and six patients had a history of pre- and postoperative contact lens wear, respectively. 263
When stratified by contact lens wear, no difference in subbasal nerve density or the number 264
of nerve crossings, respectively, was found preoperatively (P = 0.82, P= 0.62) or 265
postoperatively (P = 0.77, P = 0.79). 266
Stromal Status Five Years Post-CXL
267
The full stromal thickness was scanned by IVCM in patients at final follow-up. Isolated zones 268
devoid of keratocytes were evident, with apparent cellular debris and linear needle-like 269
structures indicative of keratocyte apoptosis (eFigure 1). Outside these narrow zones 270
(typically spanning a depth range of 10 – 20µm), normal-appearing keratocytes were visible. 271
Discussion
272
This study reports subbasal nerve regeneration after CXL over the longest follow-up period 273
to date. Nerve density in the long-term remained reduced (by over 50%) relative to age-274
matched healthy corneas. Despite clinical success of CXL in halting keratoconus progression 275
and recovery of touch sensitivity,19,41 subbasal nerves did not regenerate beyond the 276
original level even five years after CXL. Earlier studies have highlighted a poor correlation of 277
subbasal nerve density and mechanical touch sensitivity;42,43 however, the root cause of 278
abnormally sparse innervation of the subbasal plexus in keratoconus is clearly not addressed 279
by the CXL treatment. 280
Another major finding was impaired nerve guidance resulting in loops, crossings, and 281
tortuous paths seldom observed in healthy corneas. Moreover, abnormal nerve migration 282
tended to progress after CXL treatment. Subbasal nerves forming open or closed loops have 283
been noted qualitatively in keratoconus,23, 25, 29 and images indicating nerve path crossings 284
are visible in several studies,20, 23, 25, 26 but were not specifically noted or recognized as 285
pathologic or characteristic of keratoconus. Additionally, subbasal nerve tortuosity has been 286
noted to be subjectively increased in keratoconus.23, 29 Quantifying these features for the 287
first time and comparing to a healthy age-matched group, we report an increased frequency 288
of nerve loops, crossings, right-angled crossings, and elevated tortuosity in early-stage 289
keratoconus. Imaging these nerve features by IVCM could aid in the detection of early-stage 290
keratoconus. 291
Besides analysis at the single-image level, reconstructed wide-field mosaics provided striking 292
evidence that the CXL-treated cornea does not possess normal subbasal nerve architecture. 293
While the normal spiraling architecture of corneal subbasal nerves44 has been shown to be 294
perturbed in keratoconus,25 examination of mosaics after removal of the plexus during CXL 295
presents a unique opportunity to examine subbasal nerve guidance. Balanced 296
circumferential and radial forces resulting in a spiral pattern in the healthy cornea are 297
dramatically disrupted in keratoconus. Preoperatively and long after clinical halting of 298
progression, some nerves migrate only radially while others migrate only circumferentially 299
(leading to inevitable right-angled crossings). Still other nerves receive mixed signals, 300
changing orientation to form loops and tortuous paths. 301
Recent clinical studies indicate that the gross morphology of the corneal stroma is stabilized 302
for at least 4-5 years after CXL,2-6 but it is unlikely that the pathologic expression of proteins 303
and enzymes in the keratoconic eye is altered by the treatment. Corneal subbasal nerves 304
(axons originating outside the stroma) may instead reflect the underlying disease process in 305
the long term. Clinical signs of a neurotrophic deficit (such as inflammation, modified tear 306
film, or development of dry eye) were absent in this study; however, accumulation of 307
dendritic cells was noted in several patients and a detailed investigation of the epithelium 308
was not undertaken. Additional long-term study of these parameters is warranted. Within 309
the stroma, persistent zones devoid of keratocytes, accompanied by features indicative of 310
earlier apoptosis45 was an unexpected secondary finding also requiring further investigation. 311
Fully-automated nerve analysis led to the same conclusions as manual analysis, despite 312
wider limits of agreement and a tendency to underestimate nerve density when fewer 313
nerves were present (such nerves were often thinner with reduced contrast). Nevertheless, 314
automation minimizes human bias and could enable near real-time analysis in the clinic. 315
It is pertinent to highlight limitations of the present study. The proportion of patients 316
wearing contact lenses was low, which could mask a possible impact of contact lens wear on 317
nerve regeneration after the CXL treatment in this small subset of subjects. Also, the cohort 318
size was relatively small; larger prospective, long-term studies are warranted to confirm the 319
present findings and establish more precise estimates of subbasal nerve parameters after 320
CXL treatment. Finally, the present study focused only on early-stage keratoconus and not 321
severe, advanced cases - including these patients could yield additional insight into 322
progressive changes in corneal nerve parameters and morphology in keratoconus. 323
In summary, CXL treatment did not improve the nerve deficit in keratoconus and nerve 324
disorientation persisted, reflecting the progressive condition. In CXL treatment for 325
keratoconus and other corneal pathologies, the unlikelihood of improving neurotrophic 326
status should be recognized. 327
328
Acknowledgements
329 330
None of the authors have any relevant relationships or conflicts of interest to disclose. The 331
authors wish to acknowledge the following sources of Funding: the Swedish Research 332
Council (Grant No. 2012-2472) and Princess Margareta’s Foundation for the Visually 333
Impaired to NL, and funding from the Norwegian Research Council to MP. The funding 334
organizations had no role in design and conduct of the study; collection, management, 335
analysis, and interpretation of the data; or preparation, review, or approval of the 336
manuscript ordecision to submit the manuscript for publication.NL had full access to all of 337
the data in the study and takes responsibility for the integrity of the data and the accuracy 338
of the data analysis.Author Contributions: conception or design: NL, TBW, TPU; acquisition, 339
analysis, or interpretation of data: MP, SR, EP, PG, AR, SF, TBW, TPU, NL; drafting the 340
manuscript: NL, MP, AR; critical review of manuscript: MP, SR, EP, PG, AR, SF, TBW, TPU, NL; 341
statistical analysis: NL, MP; obtaining funding: NL, MP; administrative, technical or material 342
support: MP, SR, EP, PG, AR, SF, TBW, TPU, NL; supervision: AR, TPU, NL. 343
References
345
1. Wollensak G, Spoerl E, Seiler T. Riboflavin/ultraviolet-a-induced collagen cross-linking for the 346
treatment of keratoconus. Am J Ophthalmol 2003;135:620-627. 347
2. Raiskup F, Theuring A, Pillunat LE, Spoerl E. Corneal collagen crosslinking with riboflavin and 348
ultraviolet-A light in progressive keratoconus: ten-year results. J Cataract Refract Surg 349
2015;41:41-6. 350
3. Caporossi A, Mazzotta C, Baiocchi S, Caporossi T. Long-term results of riboflavin ultraviolet a 351
corneal collagen cross-linking for keratoconus in Italy: the Siena eye cross study. Am J 352
Ophthalmol 2010;149:585-593. 353
4. O'Brart DP, Kwong TQ, Patel P, McDonald RJ, O'Brart NA. Long-term follow-up of 354
riboflavin/ultraviolet A (370 nm) corneal collagen cross-linking to halt the progression of 355
keratoconus. Br J Ophthalmol 2013;97:433-437. 356
5. Hashemi H, Seyedian MA, Miraftab M, Fotouhi A, Asgari S. Corneal collagen cross-linking 357
with riboflavin and ultraviolet a irradiation for keratoconus: long-term results. 358
Ophthalmology 2013;120:1515-1520. 359
6. Vinciguerra R, Romano MR, Camesasca FI, et al. Corneal cross-linking as a treatment for 360
keratoconus: four-year morphologic and clinical outcomes with respect to patient age. 361
Ophthalmology 2013;120:908-916. 362
7. Wittig-Silva C, Chan E, Islam FM, Wu T, Whiting M, Snibson GR. A randomized, controlled 363
trial of corneal collagen cross-linking in progressive keratoconus: three-year results. 364
Ophthalmology 2014;121:812-821. 365
8. Wollensak G, Spoerl E, Wilsch M, Seiler T. Keratocyte apoptosis after corneal collagen cross-366
linking using riboflavin/UVA treatment. Cornea 2004;23:43-49. 367
9. Wollensak G, Wilsch M, Spoerl E, Seiler T. Collagen fiber diameter in the rabbit cornea after 368
collagen cross-linking by riboflavin/UVA. Cornea 2004;23:503-507. 369
10. Wollensak G, Iomdina E, Dittert DD, Herbst H. Wound healing in the rabbit cornea after 370
corneal collagen cross-linking with riboflavin and UVA. Cornea 2007;26:600-605. 371
11. Wollensak G, Iomdina E. Long-term biomechanical properties of rabbit cornea after 372
photodynamic collagen cross-linking. Acta Ophthalmol 2009;87:48-51. 373
12. Mazzotta C, Traversi C, Baiocchi S, et al. Corneal healing after riboflavin ultraviolet-A 374
collagen cross-linking determined by confocal laser scanning microscopy in vivo: early and 375
late modifications. Am J Ophthalmol 2008;146:527-533. 376
13. Mazzotta C, Balestrazzi A, Traversi C, et al. Treatment of progressive keratoconus by 377
riboflavin-UVA-induced cross-linking of corneal collagen: ultrastructural analysis by 378
Heidelberg Retinal Tomograph II in vivo confocal microscopy in humans. Cornea 379
2007;26:390-397. 380
14. Kymionis GD, Diakonis VF, Kalyvianaki M, et al. One-year follow-up of corneal confocal 381
microscopy after corneal cross-linking in patients with post laser in situ keratosmileusis 382
ectasia and keratoconus. Am J Ophthalmol 2009;147:774-778, 778 e771. 383
15. Croxatto JO, Tytiun AE, Argento CJ. Sequential in vivo confocal microscopy study of corneal 384
wound healing after cross-linking in patients with keratoconus. J Refract Surg 2010;26:638-385
645. 386
16. Knappe S, Stachs O, Zhivov A, Hovakimyan M, Guthoff R. Results of confocal microscopy 387
examinations after collagen cross-linking with riboflavin and UVA light in patients with 388
progressive keratoconus. Ophthalmologica 2011;225:95-104. 389
17. Touboul D, Efron N, Smadja D, Praud D, Malet F, Colin J. Corneal confocal microscopy 390
following conventional, transepithelial, and accelerated corneal collagen cross-linking 391
procedures for keratoconus. J Refract Surg 2012;28:769-776. 392
18. Mastropasqua L, Nubile M, Lanzini M, et al. Morphological modification of the cornea after 393
standard and transepithelial corneal cross-linking as imaged by anterior segment optical 394
coherence tomography and laser scanning in vivo confocal microscopy. Cornea 2013;32:855-395
861. 396
19. Kontadakis GA, Kymionis GD, Kankariya VP, Pallikaris AI. Effect of corneal collagen cross-397
linking on corneal innervation, corneal sensitivity, and tear function of patients with 398
keratoconus. Ophthalmology 2013;120:917-922. 399
20. Jordan C, Patel DV, Abeysekera N, McGhee CN. In vivo confocal microscopy analyses of 400
corneal microstructural changes in a prospective study of collagen cross-linking in 401
keratoconus. Ophthalmology 2014;121:469-474. 402
21. Sehra SV, Titiyal JS, Sharma N, Tandon R, Sinha R. Change in corneal microstructure with 403
rigid gas permeable contact lens use following collagen cross-linking: an in vivo confocal 404
microscopy study. Br J Ophthalmol 2014;98:442-447. 405
22. Muller LJ, Marfurt CF, Kruse F, Tervo TM. Corneal nerves: structure, contents and function. 406
Exp Eye Res 2003;76:521-542. 407
23. Patel DV, Ku JY, Johnson R, McGhee CN. Laser scanning in vivo confocal microscopy and 408
quantitative aesthesiometry reveal decreased corneal innervation and sensation in 409
keratoconus. Eye (Lond) 2009;23:586-592. 410
24. Simo Mannion L, Tromans C, O'Donnell C. An evaluation of corneal nerve morphology and 411
function in moderate keratoconus. Cont Lens Anterior Eye 2005;28:185-192. 412
25. Patel DV, McGhee CN. Mapping the corneal sub-basal nerve plexus in keratoconus by in vivo 413
laser scanning confocal microscopy. Invest Ophthalmol Vis Sci 2006;47:1348-1351. 414
26. Niederer RL, Perumal D, Sherwin T, McGhee CN. Laser scanning in vivo confocal microscopy 415
reveals reduced innervation and reduction in cell density in all layers of the keratoconic 416
cornea. Invest Ophthalmol Vis Sci 2008;49:2964-2970. 417
27. Mocan MC, Yilmaz PT, Irkec M, Orhan M. In vivo confocal microscopy for the evaluation of 418
corneal microstructure in keratoconus. Curr Eye Res 2008;33:933-939. 419
28. Ozgurhan EB, Kara N, Yildirim A, Bozkurt E, Uslu H, Demirok A. Evaluation of corneal 420
microstructure in keratoconus: a confocal microscopy study. Am J Ophthalmol 421
2013;156:885-893 e882. 422
29. Al-Aqaba MA, Faraj L, Fares U, Otri AM, Dua HS. The morphologic characteristics of corneal 423
nerves in advanced keratoconus as evaluated by acetylcholinesterase technique. Am J 424
Ophthalmol 2011;152:364-376 e361. 425
30. Auran JD, Koester CJ, Kleiman NJ, et al. Scanning slit confocal microscopic observation of cell 426
morphology and movement within the normal human anterior cornea. Ophthalmology 427
1995;102:33-41. 428
31. Patel DV, McGhee CN. In vivo laser scanning confocal microscopy confirms that the human 429
corneal sub-basal nerve plexus is a highly dynamic structure. Invest Ophthalmol Vis Sci 430
2008;49:3409-3412. 431
32. Eden U, Fagerholm P, Danyali R, Lagali N. Pathologic epithelial and anterior corneal nerve 432
morphology in early-stage congenital aniridic keratopathy. Ophthalmology 2012;119:1803-433
1810. 434
33. Parissi M, Karanis G, Randjelovic S, et al. Standardized baseline human corneal subbasal 435
nerve density for clinical investigations with laser-scanning in vivo confocal microscopy. 436
Invest Ophthalmol Vis Sci 2013;54:7091-7102. 437
34. Meijering E, Jacob M, Sarria JC, Steiner P, Hirling H, Unser M. Design and validation of a tool 438
for neurite tracing and analysis in fluorescence microscopy images. Cytometry A 439
2004;58:167-176. 440
35. Schindelin J, Arganda-Carreras I, Frise E, et al. Fiji: an open-source platform for biological-441
image analysis. Nat Methods 2012;9:676-682. 442
36. Scarpa F, Grisan E, Ruggeri A. Automatic recognition of corneal nerve structures in images 443
from confocal microscopy. Invest Ophthalmol Vis Sci 2008;49:4801-4807. 444
37. Scarpa F, Zheng X, Ohashi Y, Ruggeri A. Automatic evaluation of corneal nerve tortuosity in 445
images from in vivo confocal microscopy. Invest Ophthalmol Vis Sci 2011;52:6404-6408. 446
38. Poletti E, Wigdahl J, Guimaraes P, Ruggeri A. Automatic Montaging of Corneal Sub-Basal 447
Nerve Images for the Composition of a Wide-Range Mosaic. Proc. 36th Annual International 448
Conference of IEEE-EMBS, pp. 5426-9, IEEE, New York, 2014. 449
39. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of 450
clinical measurement. Lancet 1986;1:307-310. 451
40. Alio JL, Shabayek MH. Corneal higher order aberrations: a method to grade keratoconus. J 452
Refract Surg 2006;22:539-545. 453
41. Wasilewski D, Mello GHR, Moreira H. Impact of collagen crosslinking on corneal sensitivity in 454
keratoconus patients. Cornea 2013;32:899-902. 455
42. Patel DV, Tavakoli M, Craig JP, Efron N, McGhee CN. Corneal sensitivity and slit scanning in 456
vivo confocal microscopy of the subbasal nerve plexus of the normal central and peripheral 457
human cornea. Cornea 2009;28:735-40. 458
43. Hamrah P, Cruzat A, Dastjerdi MH, Zheng L, Shahatit BM, Bayhan HA, Dana R, Pavan-459
Langston D. Corneal sensation and subbasal nerve alterations in patients with herpes 460
simplex keratitis: an in vivo confocal microscopy study. Ophthalmology 2010;117:1930-6. 461
44. Patel DV, McGhee CN. Mapping of the normal human corneal sub-Basal nerve plexus by in 462
vivo laser scanning confocal microscopy. Invest Ophthalmol Vis Sci 2005;46:4485-4488. 463
45. Sharma N1, Suri K, Sehra SV, Titiyal JS, Sinha R, Tandon R, Vajpayee RB. Collagen cross-464
linking in keratoconus in Asian eyes: visual, refractive and confocal microscopy outcomes in 465
a prospective randomized controlled trial. Int Ophthalmol 2015; Feb 24. [Epub ahead of 466 print]. 467 468 469 470
Figure Legends
471 472
Figure 1. Subbasal nerve density in the central cornea in early-stage keratoconus
473
(preoperative) versus healthy age-matched subjects (19 subjects per group).. 474
475
Figure 2. Subbasal nerve regeneration up to five years after CXL treatment for progressive
476
keratoconus.. Top: manual analysis of subbasal nerve density indicated a significant 477
reduction in the early postoperative period (asterisk).. Center: Automated analysis yielded a 478
similar pattern of nerve regeneration as manual analysis.Bottom: Mean and 95% confidence 479
interval for corneal sensitivity, indicating a significant but minor reduction in sensitivity at 3 480
months postoperative (P = 0.017).. 481
482
Figure 3. Nerve architecture in keratoconus. Left column: subbasal nerve plexus with
483
roughly parallel nerve fiber bundles, low tortuosity, and rare crossings (black arrows). 484
Centre column: looping nerves (white arrows) and increased crossings (black arrows). Right 485
column: persistent looping nerves (white arrow), crossings (black arrows) and tortuous 486
nerve paths (white arrowheads). All images are 400 µm x 400 µm. 487
488
Figure 4. Quantitative analysis of subbasal nerve morphology. Top: the number of subbasal
489
nerve crossings per image frame.. Center: the minimum crossing angle of subbasal nerves in 490
cases of crossings.. Bottom: nerve tortuosity. 491
492 493
Figure 5. Nerve plexus mosaics in 6 different patients five years after corneal collagen
cross-494
linking treatment for keratoconus. (A) Circumferential nerve paths emerging from 495
penetration points (black arrows), and tortuous paths (white arrows). (B) Crossings (black 496
arrowheads) at intersections of radial and circumferential nerves, and loops (white 497
arrowheads) . (C) Loops (white arrowheads) varying between radial and circumferential 498
orientations. (D) Tortuous paths (white arrows). (E) Nerves penetrate (black arrows) and 499
orient radially. Crossings (black arrowheads) where radial and circumferential nerves 500
intersect. Also, tortuousity (white arrow) and loops (white arrowheads). (F) After 501
penetration (black arrows), abrupt orientation changes (white arrowheads) form loops. All 502
images, bar = 400µm. 503
Plot 1
keratoconus
healthy
keratoconus
healthy
subbasal nerv
e density
(mm/mm
2 )
0
5
10
15
20
25
30
35
40
P < 0.001 P < 0.001manual
automated
Months post-CXL Preop 0-6 7-12 13-24 25-36 37-48 49-66 S ubbasal nerv e dens ity , auto (mm/mm 2) 0,0 5,0 10,0 15,0 20,0 25,0 * Months post-CXL Preop 0-6 7-12 13-24 25-36 37-48 49-66 S ubbasal nerv e density , manual (mm/mm 2 ) 0,0 5,0 10,0 15,0 20,0 25,0 * Months post-CXL Preop 3 6 12 60 Corneal sensitiv ity (thread length) 0 10 20 30 40 50 60 *
Pre-CXL Post-CXL
Healthy
healthy preop >12m post CXL
main subbasal nerv
e crossings per frame
0,0 0,5 1,0 1,5 2,0 2,5 3,0 P = 0.03 P = 0.002
healthy preop >12m post CXL
subbasal nerv
e crossing angle (degrees)
0 10 20 30 40 50 60 70 80 90 P = 0.02 Plot 1
Healthy Preop >12m Post-CXL
Tortuosity Index 0,0 0,5 1,0 1,5 2,0 P < 0.05
A
B
C
D
E
A
F
Online Only Material
1
The Online Only material consists of the following elements:
2
eTable 1. Subject Characteristics
3
eTable 2. Inter-observer and inter-method comparison of subbasal nerve density.
4
eFigure 1. IVCM images of the corneal stroma in 4 different patients taken 58 months after
5
standard epithelium-off collagen cross-linking treatment.
6 7
eTable 1. Subject Characteristics. 8 Healthy corneas Keratoconus n = 19 n = 19 Sex, n (%) Male 12 (63) 17 (89) Female 7 (37) 2 (11) Age (y) 29.9 ± 6.8 27.5 ± 7.1 Range (y) 20 - 45 19 - 44 MCT (µm) 529 ± 23 428 ± 36 Range (µm) 487 - 559 372 – 497 Astigmatism (D) 5.6 ± 3.1 Range (D) 0.4 - 15.2 Max K (D) 50.5 ± 4.9 Range (D) 42.2 - 60.4 Min K (D) 44.9 ± 4.9 Range (D) 38.3 - 59.8
n, number of subjects. Values for the keratoconus group are preoperative. 9
10 11
eTable 2. Inter-observer and inter-method comparison of subbasal nerve density.
12
Healthy Keratoconus
n = 57 n = 226
Inter-observer (Obs 1 - Obs 2)
Mean density difference (mm/mm2) -0.30 0.06
95% LOA (mm/mm2) ± 1.58 ± 1.22
Inter-method (Automated – Manual)
Mean density difference (mm/mm2) -0.26 -1.74
95% LOA (mm/mm2) ± 3.13 ± 4.41
n refers to the number of confocal microscope images analyzed from each group. Three images were 13
analyzed for each healthy reference subject and for each keratoconus subject at preoperative and 14
postoperative time points. Obs 1, Obs 2 refer to the two trained human observer values from 15
manual nerve tracing of images. 95% LOA are the 95% limits of agreement according to the Bland-16
Altman method.38
17 18
19
eFigure 1. IVCM images of the corneal stroma in 4 different patients taken 58 months after
20
standard epithelium-off collagen cross-linking treatment. In certain depth zones, the central
21
corneal stroma was devoid of keratocytes and populated by apparent cellular debris (white
22
arrows) and linear needle-like structures (black arrows), features indicative of keratocyte
23
apoptosis. Depths of the images from the corneal surface (0µm) are 91, 185, 214, and
24
391µm for A-D, respectively. All images are 400 × 400µm.
25 26 27 28
