Laser vision correction is the most frequently performed elective surgical procedure in the world today. Ectasia or corneal weakening that develops following laser vision correction is a complication that has been previously reported for laser in situ keratomileusis (LASIK), small incision lenticule extraction (SMILE),1 and photorefractive keratectomy (PRK).2 SMILE is a relatively new refractive technique using a flap-free intrastromal laser for correcting myopia and astigmatism.3 SMILE is gaining preference over LASIK for its safety and effectiveness,4 revealing only minimum damage to the subbasal nerve plexus5 and greater retention of the anterior collagen network, which is attributed to approximately 60% of total corneal tensile strength.6 Randleman et al.7 created a quantitative scoring system to screen patients at risk for developing ectasia after refractive surgery (LASIK/PRK). Hypothetically, this same risk parameter scoring system could also be applied to stratify patients who undergo SMILE. A recent review of ectasia after SMILE reported 5 cases8–12 between 2011 and March 2018.1 Of these, 3 had either a high Randleman's ectasia risk score preoperatively or forme fruste keratoconus, giving sufficient evidence of risk factors preoperatively.8–10
The molecular signature of diseased tissue adds a new dimension to the existing knowledge of disease pathogenesis and its mechanism. We previously reported increased corneal dendritic cell density and an aberrant inflammatory state in the tear cytokine profile of eyes with ectasia after LASIK when compared with normal matched controls.13 In another study, we observed that the local tissue-specific molecular factors (lysyl oxidase [LOX], matrix metalloproteinase 9 [MMP9], interleukin 6 and 10 [IL-6, IL-10], tumor necrosis factor alpha [TNF-alpha], and tissue inhibitor of metalloproteinase-1 [TIMP1]) at the ectatic region of the corneal epithelium and stroma are responsible for the focal corneal weakening in keratoconus.14 Human diseased corneas undergoing active tissue remodeling have higher cluster of differentiation 68 (CD68) (lysosomal membrane marker) positive cells.15 Mackiewicz et al.16 reported a high presence of cathepsin K in the keratoconic corneal buttons compared to the control group. However, levels of cathepsins B and L were unchanged around corneas with ectasia after LASIK and keratoconus compared with normal corneas.17 In mouse keratocytes, the conditionally deleted integrin beta-1 (ITGbeta-1) gene leads to corneal thinning and edema during the stromal maturation phase, revealing its role in the maintenance of corneal structural integrity.18 In the mouse liver fibrosis model,19 the signal transducer and activator of transcription 3 (STAT3) pathway plays a pivotal role by inducing leptin and thus advances the mechanism of liver fibrosis. Therefore, we have chosen this specific set of molecular markers in our current study.
To understand the tissue molecular profile associated with ectasia after SMILE, we analyzed a case of markedly asymmetric ectasia, where the affected eye was compared to the “suspect” eye of the same patient and control eyes with optimal outcomes postoperatively.
Patients and Methods
This is a retrospective observational study of patients who had SMILE that was approved by the institutional ethics committee of Narayana Nethralaya Multispecialty Eye Hospital, Bangalore, India, in accordance with the tenets of the Declaration of Helsinki. Patients were recruited for this study after obtaining informed written consent.
A total of 178 patients were followed up over a period of 2 to 3 years, after having their intraoperatively extracted corneal lenticules preserved (with informed consent) postoperatively. The follow-up included manifest refraction, detailed slit-lamp examination, tomography (Pentacam HR, v1.20r98; OCULUS Optikgeräte GmbH, Wetzlar, Germany), and air-puff deformation videography (Corvis ST, v1.3r1476; OCULUS Optikgeräte GmbH) at each visit.
One of the patients from this cohort, a 25-year-old highly myopic woman with normal corneal thickness (central corneal thickness [CCT] of approximately 540 µm in both eye), normal Pentacam tomography (Belin/Ambrósio total D [BAD-D] score was 0.76 in the right eye and 0.94 in the left eye), and normal Corvis biomechanical indices (Corvis Biomechanical Index [CBI]/Total Biomechanical Index [TBI] score was 0.12/0.38 in the right eye and 0.0/0.13 in the left eye), reported progressively worsening vision 2 years after bilateral SMILE surgery. The preoperative refraction was −8.00 −1.50 × 100 in the right eye and −8.00 −1.00 × 90 in the left eye. At the time of ectatic presentation, the left eye had corrected distance visual acuity (CDVA) of 0.17 logMAR (Snellen 20/30), subjective spherical refraction of −2.25 D, subjective cylindrical refraction of −2.75 D, CCT of 418 µm, and a Pentacam BAD-D score of 5.29. The right eye had CDVA of 0.0 logMAR (Snellen 20/20), subjective spherical refraction of −0.50 D, subjective cylindrical refraction of −1.00 D, CCT of 439 µm, and a BAD-D score of 4.17. The refraction and topography (Figure A, available in the online version of this article) revealed that the patient had developed bilaterally asymmetric ectasia after surgery. The left cornea that went on to develop an aggressive ectasia after SMILE surgery was designated the “ectatic eye” and the contralateral right eye was designated the “suspect-ectatic eye.”
Tangential (A, C, and E) and axial (B, D, and F) anterior curvature maps from Pentacam HR (OCULUS Optikgeräte, Wetzlar, Germany) (absolute American style color bar in diopters) for before surgery, after surgery, and 2-year follow-up time points, respectively. (A and B) Normal eye curvature maps are shown in the first row, (C and D) suspect-ectatic eye curvature maps are shown in the second row, and (E and F) ectatic eye curvature maps are shown in the third row.
Control eyes matched for age, gender, and duration of follow-up with clear vision were referred to as “normal eyes.” The five eyes of three control patients had CCT of 563, 552, 508, 515, and 485 µm and were corrected for −8.00 −0.75 × 10, −8.50 −0.75 × 10, −6.25 −1.50 × 175, −6.00 −0.75 × 170, and −4.50 −0.75 × 180, respectively.
SMILE surgery was performed using the VisuMax femtosecond laser system (Carl Zeiss Meditec AG) with a 500-kHz repetition rate. The spot spacing was 4.5 µm for creation of the lenticule and 2 µm for creation of the lenticule side-cut with a preset laser cut energy of 170 nJ and track distance of 3 mm. The cap thickness was kept at 110 µm with a lenticule diameter of 6 mm (optical zone) and a cap diameter of 7.7 mm. After its creation, the lenticule was separated with a dissector through the 3-mm entry incision located superiorly, and then removed manually. The lenticule thickness resected was 136 µm in the right eye and 131 µm in the left eye of the ectatic case. The cornea was remoistened with a wet sponge at the end of the procedure, and one drop of moxifloxacin hydrochloride 0.5% (Vigamox; Alcon Laboratories, Inc., Fort Worth, TX) was instilled in each eye. The routine postoperative regimen given in each eye included moxifloxacin hydrochloride 0.5% eye drops (Vigamox) four times a day for 1 week, tapering doses of topical 1% fluorometholone eye drops (Flarex; Alcon Laboratories, Inc.) over 12 days, and topical lubricants (Optive; Allergan, Inc.) four times a day for 3 months. Intraoperatively collected lenticules were immediately transferred to −80°C for storage and later used for RNA extraction.
Air-puff deformational videography (Corvis ST) was used to measure the intraocular pressure (IOP), deformation amplitude, deflection amplitude, CBI,20 and TBI20 before and after surgery. Pentacam HR anterior curvature maps (tangential and axial) were primarily used in diagnosing the surgery-induced ectasia. Pre-operative CCT reported by the Pentacam HR was also recorded, and the mean keratometry calculated before surgery was plotted against CCT and IOP to understand their association with the population. Further, five lenticule samples from five eyes of three control patients matched for age, sex, and postoperative duration were randomly chosen from the larger cohort for in-depth analysis of the biomechanical and molecular profiles.
Assessment of Gene Expression in Human Corneal Lenticule
Stromal lenticules were finely chopped using a sterile surgical blade (Lister; Sterilized by Gamma Radiation, ISO 9001:2000 Co.) and total RNA was extracted using TRIZOL reagent according to the manufacturer's instructions (Invitrogen, Carlsbad, CA) followed by quantification and quality assessment. Briefly, the quantitative real-time polymerase chain reaction (PCR) cycle includes preincubation at 95°C for 5 minutes, 40 amplification cycles at 95°C for 10 seconds, 58°C for 15 seconds, and 72°C for 30 seconds using a CFX Connect real-time PCR detection system (Bio-Rad, Philadelphia, PA). Total levels of LOX, MMP9, IL-6, IL-10, bone morphogenetic protein 7 (BMP7), TNF-alpha, TIMP1, collagen type I alpha 1 (COLIA1), collagen type IV alpha 1 chain (COLIVA1), CD68, cathepsin K, ITG-beta-1, signal transducer and activator of transcription 3 (STAT3), transforming growth factor beta (TGF-beta), and transforming growth factor beta receptor 2 (TGF-betaR2) were measured after normalization to actin. Total RNA was also isolated from primary human corneal fibroblast cells transduced for LOX overexpression and matching controls for subsequent measurement of LOX expression. Primer sequences are available on request.
Tear samples were collected from both eyes of the patient who presented with ectasia after SMILE using Schirmer's strips (Whatman filter paper, 5 × 35 mm2; ContaCare Ophthalmics and Diagnostics, Vadodara, Gujarat, India) and stored at −80°C until further use. Extraction of tear fluid from Schirmer's strips was performed as described previously.21
Cytometric Bead Array
The levels of cytokines, chemokines, cell adhesion molecules, and soluble receptors in the tears were measured using cytometric bead array (BD CBA Human Soluble Protein Flex Set System; BD Biosciences, San Jose, CA) on a flow cytometer (BD FACSCalibur; BD Biosciences), as described previously.13 The assay was a performed as per the manufacturer's instruction using a BD Human Soluble Protein Master buffer kit.
Human Corneal Fibroblast Cell Culture
Human corneal fibroblast cells were cultured from human lenticule tissue obtained during SMILE. The corneal lenticule was washed with antibiotic-antimycotic solution (Gibco; ThermoFischer Scientific, Carlsbad, CA), and then was cut into small pieces with the scalpel blade and incubated in a humidified carbon dioxide incubator at 37°C using a 1:1 mixture of Dulbecco's modified Eagle medium (DMEM) and Nutrient Mixture F12 (Gibco) supplemented with 10% fetal bovine serum (FBS, certified USA Origin), 0.3 mg/mL L-glutamine (Invitrogen, Gibco), 0.1 mg/mL streptomycin, and 1,000 IU/mL penicillin (Gibco) to obtain the primary human corneal fibroblast cells. The fibroblasts were used between the third to fifth passages.
Plasmid, Lentivirus Preparation, and Transduction
A plasmid for overexpressing lysyl oxidase (LV. pLX304-LOX; TransOMIC Technologies, Huntsville, AL) was used to generate LOX expressing lentiviruses using protocols described previously. A total of 60% to 70% confluent cultures of primary human corneal fibroblast cells in 12 well plates were used for the transduction experiments. Lentiviral transduction was done in serum-free media, and fresh serum-containing media was added after 3 hours, making the final 10% serum media.22 After 48 hours of transduction, primary human corneal fibroblast cells were harvested, validated for LOX expression, and used for gel contraction assay.
Collagen Gel Contraction Assay
Primary human corneal fibroblast cells transduced with LOX overexpression were detached by 0.25% trypsin in 0.02% EDTA and suspended in media. The cells were counted using a counting chamber. Collagen I gels (Rat, Tail Protein; ThermoFisher Scientific) were prepared by mixing 1M NaOH, DMEM/F12 media and cells. The final concentration was 1 mg/mL for collagen and 1 × 105 cells/mL for the fibroblast cells. A total of 600 µL of the mixture was casted into each well of 24 wells culture plates (Eppendorf). For polymerization, the plate was kept in the incubator at 37°C humidified 5% carbon dioxide for 30 minutes. After polymerization, the gels were gently released from the wells to allow contraction to happen and incubated for 48 hours. The area of each gel was imaged and the contraction was measured by placing the gels on graph paper and further analyzed using ImageJ software ( https://imagej.nih.gov/ij/). Each image was processed for the total area of measurement, and this value was then converted to a total pixel value.
All statistical analyses were performed with Med-Calc v12.5.0 (MedCalc, Inc., Ostend, Belgium) and GraphPad Prizm 6.0 (GraphPad Software, Inc., La Jolla, CA) software. The Welch test, assuming unequal variance, was run for analysis. All values were reported as mean and 95% confidence interval of the mean. Significance was defined as a P value less than .05.
Figures AA–AB show the tangential and axial curvature maps for a matched control “normal eye” having undergone surgery at the same calendar date, whereas Figures AC–AD show the same for the “suspect-ectatic” eye and Figures AE–AF for the “ectatic” eye. In all preoperative eyes, the tangential and axial maps appeared normal, with the early postoperative maps all showing central myopic flattening without obvious abnormality. Among the later postoperative maps, the normal eye and even the suspect-ectatic eye showed a relatively stable shape, whereas the tangential and axial curvature maps of the ectatic eye showed a clear central ectatic progression. The Pentacam HR preoperative BAD-D score for the matched control normal eye, right suspect-ectatic eye, and left ectatic eye was within normal limits at 1.19 ± 0.36 (mean ± standard deviation for 5 eyes), 0.77, and 0.43 (normal < 1.60), respectively. Meanwhile, the postoperative values were sequentially each more elevated at 4.26 ± 0.96 (mean ± standard deviation for 5 eyes), 4.83, and 5.32, respectively.
Table 1 shows the CBI for the randomly chosen normal eyes (5 eyes), the suspect-ectatic eye, and the ectatic eye. The preoperative CBI, TBI, peak deformation amplitude, and peak deflection amplitude for the left ectatic and right suspect-ectatic eyes were all normal at 0.0, 0.13, 1.19, and 1.01 mm, respectively, for the left eye and 0.12, 0.38, 1.28, and 1.11 mm, respectively, for the right eye (CBI and TBI values ranged from 0.0 to 1.0 with relative normality at < 0.5). Postoperatively, the same parameters were 1.0, 1.0, 1.38, and 1.2 mm, respectively, in the left eye and 1.0, 1.0, 1.39, and 1.26 mm in the right eye. This observation is significant because preoperative deformation and deflection amplitude, CBI, and TBI were within the acceptable limits.
Demographics and Corneal Stiffness Parameters
Gene Expression Profiling of Patient Lenticule
Table A (available in the online version of this article) contains the gene expression profile for the ectatic, suspect-ectatic, and matched normal eyes. Section 1 of the table has mean and 95% confidence intervals of the mean for normal eyes and the four repeat experiments done on the ectatic and suspect-ectatic eyes. For genes listed in section 2, the ectatic and suspect-ectatic sample was only repeated four times; hence statistical analyses were not performed. Figure 1 shows the mRNA expression of LOX is significantly reduced (P = .02) in the ectatic eye (mean expression of 2.01×10−3) compared to the normal (mean expression of 6.73×10−3) and suspect-ectatic (mean expression of 5.08×10−3) eyes (Figure 1A, Table A). However, COLIA1 (Figure 1B) and MMP9 (Figure 1C) showed no change among the three groups. COLIA1 expression (fold change normalized to housekeeping gene Actin) was 1.85×10−2 [8.16×10−3 to 2.87×10−2] for the ectatic eye, 1.79×10−2 [5.08×10−3 to 3.08×10−2] for the suspect-ectatic eye, and 2.24×10−2 [1.38×10−2 to 3.09×10−2] for the normal eyes. MMP9 expression was unaltered for the ectatic (1.69×10−2 [1.04×10−2 to 2.35×10−2]) and suspect-ectatic (1.70×10−2 [1.15×10−2 to 2.24×10−2]) eyes, respectively, and with a slight reduction of 1.57×10−2 [1.62×10−3 to 2.98×10−2] in normal eyes. In the ectatic eye, mRNA expression of TGF-beta (Figure 1D) was 1.00×10−2 [−2.58×10−4 to 2.03×10−2], but there was a decreasing trend for the suspect-ectatic (8.08×10−3 [−1.34×10−3 to 1.75×10−2]) and normal (3.76×10−3 [1.65×10−3 to 5.87×10−3]) eyes. A similar pattern was observed for BMP7 (7.11×10−2 in the ectatic eye, 4.17×10−2 in the suspect-ectatic eye, and 4.96×10−2 [3.02×10−2 to 6.90×10−2] in the normal eyes) (Figure 1E). Interestingly, TGFbetaR2 expression (Figure 1F) was unchanged across the three groups at 2.14×10−2, 2.06×10−2, and 2.19×10−2 [1.34×10−2 to 3.04×10−2], respectively.
Gene Expression Data for Normal Eyes (n = 5 Eyes), Suspect-Ectatic Eye (4 Replicates), and Ectatic Eye (4 Replicates)
Normalized gene expression profile of patients who developed ectasia (black bar), suspect-ectatic eye (gray bar), and normal eye (white bar). (A) Lysyl oxidase (LOX) was significantly reduced in the ectatic eye. (B and C) Collagen type I alpha 1 (COLIA1) and matrix metalloproteinase 9 (MMP9) expression was unchanged. (D and E) Transforming growth factor-beta (TGF-beta) and bone morphogenetic protein 7 (BMP7) exhibited decreasing trends and (F) transforming growth factor-beta receptor 2 (TGFbetaR2) did not change.
Table B (available in the online version of this article) reports the ratio of tear cytokines, chemokines, cell adhesion molecules, and soluble receptors profile of the patient with ectasia after SMILE. Ratio values greater than 1 indicate a higher tear cytokine profile in the ectatic eye when compared with the contralateral suspect-ectatic eye of the patient. However, there was no significant difference observed between the eyes.
Ratio of Tear Cytokines, Chemokines, Cell Adhesion Molecules, and Soluble Receptors Profile Between the Ectatic and Suspect-Ectatic Eye of a Patient With Ectasia After SMILE
Assessment of LOX Functional Activity
LOX was ectopically overexpressed in primary human corneal fibroblast cells via transduction of a lentiviral construct with the gene expression levels being measured. Figure 2A shows increased mRNA expression of LOX, confirming successful transduction. Figure 2B represents images of the polymerized collagen gel, layered with human corneal fibroblast cells transduced with LOX overexpression constructs at three different concentrations (5×, 20×, and 30×, respectively) in the 24 well plate after 48 hours. For measuring the contracted area, gels were placed on the graph paper (Figure 2C) and were further analyzed using ImageJ software. LOX overexpression in human corneal fibroblast cells induced significantly more collagen gel contraction (as shown by reduction in gel size), indicating its role in strengthening the corneal stroma.
Gel contraction assay using primary human cornea fibroblast cells. (A) Validation of lysyl oxidase (LOX) over-expression by reverse transcription polymerase chain reaction. (B) Gel image for each experimental condition showing contraction. (C) Gels placed on graph paper for the area measurement.
A variety of anatomical (eg, Pentacam BAD-D)23 and biomechanical (eg, CBI and TBI20 and the deformation amplitude and deflection amplitude form) factors affect our current state-of-the-art ectasia prevention assessment criteria. However, ectasia developed within 2 years after surgery in one of our patients, despite meeting all of the mentioned criteria. An abnormality in corneal biomechanics has been attributed to ectasia in the past.24 Although both the Corvis ST and the Ocular Response Analyzer (ORA; Reichert Inc., Depew, NY)25 provide a realistic estimation of corneal biomechanics in the absence of anatomically linked metrics, there are additional factors that may make this estimation incomplete. We observed that in our uniquely featured patient, this bilaterally asymmetric ectasia was associated with an altered postoperative biomechanics and preoperative tissue gene expression profile compared to patients who did not develop postoperative ectasia.
It is possible that the Corvis ST profile is not sensitive enough to predict an ectatic cornea from a “sub-clinical ectasia” and that a differential biological behavior may exist between both eyes. The eye with significantly reduced levels of LOX exhibited ectasia. The contralateral eye of the same patient with a relatively higher LOX level did not present with ectasia, although biomechanically both eyes were weaker than matched controls. However, TGF-beta expression level was greater in the ectatic eye compared with the suspect-ectatic and matched control eyes. This significant relative difference in the LOX and TGF-beta level between the eyes, despite similar biomechanics, suggests that the ectatic behavior could be the result of inherent tissue-specific changes in molecular profile that appear to be more sensitive than the Corvis ST parameters in differentiating a clear ectasia from a mild one.
LOX is a natural collagen cross-linking enzyme produced in the extracellular matrix in the cornea and known to be reduced in ectatic eye diseases.14 LOX is a copper amine oxidase that catalyzes the formation of cross-links between collagen bundles in the cornea, providing them with strength and stiffness. We recently demonstrated that a reduced preoperative level of LOX was associated with a less favorable clinical outcome of riboflavin/ultraviolet-A corneal cross-linking in keratoconic eyes.26 Reduced LOX levels in corneal tissue have also been associated with disease severity of keratoconus.27 This same study further demonstrated a similar reduction in the enzymatic activity of LOX in the tear fluid of patients with keratoconus27 in correlation with an increasing grade of the disease. Furthermore, we have also shown reduced LOX levels in the ectatic cone area of keratoconus are also associated with microtrauma to Bowman's layer, possibly due to eye rubbing or other corneal insult.28 A meta-analysis by Zhang et al.29 has given insight about the genetic variations in LOX and its association with keratoconus, underpinning its role in maintaining corneal strength.
In our study, ectopic expression of LOX in primary corneal fibroblast cells leads to a dose-dependent increase in the tensile strength of the collagen matrix observed in the form of an increased contraction of the collagen gel. These data and the known literature of LOX support a scenario where inherently lower levels of LOX could be a critical factor in predisposing corneas to ectasia after a surgical insult. We have previously shown that the tear cytokine profile of ectasia after LASIK has an active inflammatory signature.13 Therefore, in the current case of our patient with ectasia after SMILE, we compared the tear cytokine profile between the ectatic and suspect-ectatic eyes to determine the potential inflammatory role in the disease process. We observed no significant difference in the tear cytokine profile between the ectatic and suspect-ectatic eyes. This suggests that the ocular surface inflammatory status may not have been a contributing factor to the process of ectasia in this patient.
In a 5-year outcomes study, Blum et al.30 reported SMILE as an effective, stable, and safe procedure for treating myopia and myopic astigmatism. To date, only 4 cases of postoperative ectasia after SMILE have been reported. One of these by Sachdev et al.11 reported an eye that developed ectasia after SMILE despite presenting with normal preoperative clinical indices. Although no explanation of the possible underlying mechanism was offered, our study provides a plausible reason for the development of ectasia after SMILE by uncovering an altered, inherent tissue-specific molecular profile, in particular a reduced LOX level. Our observed reduction of LOX, the asymmetric expression between two eyes with asymmetric disease, could explain why an otherwise clinically healthy patient with normal corneal indices undergoing SMILE surgery can be predisposed to develop postoperative ectasia. Tissue-specific molecular factors and their response to healing postoperatively have been shown to be important determinants of subsequent clinical outcomes. In the patient who developed ectasia after SMILE, despite normal preoperative clinical indices, inherently low levels of LOX may have prevented the cornea from regaining its tensile strength. Hence, determining LOX status in tears might aid in screening postoperative risk for the development of ectasia in refractive surgery.