Small incision lenticule extraction (SMILE) is an all-femtosecond laser technique that has recently been approved by the U.S. Food and Drug Administration for the surgical correction of myopia. It has been hypothesized that by obviating the need for excimer laser photoablation, SMILE may elicit less chemokine and cytokine release with reduced inflammatory response and wound healing reaction of the corneal stroma.1–3 Evidence derived from rabbit models has accumulated that SMILE4,5 (and its predecessor technique femtosecond lenticule extraction [FLEx]2) induces low levels of inflammatory cell infiltration, keratocyte proliferation, and keratocyte apoptosis. To our knowledge, however, evidence from human tissue is confined to a single study that showed a minimum of keratocyte apoptosis and inflammatory reaction in the extracted lenticules of 50 eyes of 50 patients that had undergone SMILE.3 These observations were made in human lenticules extracted and processed directly after femtosecond laser application and, hence, these findings cannot be directly transferred to the wound healing reaction that actually takes place in the stromal lenticule bed. This is mainly due to the complex and cascade-like nature of the corneal wound healing response after refractive surgery, which takes time to commence because it involves migration of inflammatory cells and keratocytes that are attracted to the wound site by chemotaxis with subsequent remodeling of the extracellular matrix.4
The purpose of this ex vivo laboratory study was to assess the wound healing and inflammatory response of the human corneal tissue in SMILE and compare it to femtosecond laser–assisted LASIK (FS-LASIK).
Materials and Methods
Source of Tissue
A total of 16 eyes of 16 human donors aged 69 ± 11 years (range: 40 to 83 years) were obtained from the corneal tissue bank of the Ludwig-Maximilians-University, Munich, Germany. All donor corneas had contraindication for transplantation according to the German Paul Ehrlich Institute for vaccines and biomedicines. Informed specific consent for research purposes was obtained from all donors or their relatives according to the national legal requirements. All experiments were conducted in agreement with the tenets outlined in the Declaration of Helsinki and were approved by the Ethics Committee of the Medical Faculty of the Ludwig-Maximilians-University; Munich, Germany under the approval identification number 73416.
Corneoscleral discs were excised from donor eyes using a 15-mm trephine and brought to organ culture at standard organ culture conditions (37°C and 5% CO2) using MEM Earls cell culture medium (Merck; Berlin, Germany) containing FCS (Merck) and penicillin (100,000 IE/mL) / streptomycin (100 mg/mL) and amphotericin B (5 mg/mL) (MilliporeSigma; St. Louis, MO). Before corneal surgical intervention, the corneoscleral discs were placed in the same medium, but containing 6% of dextran for at least 24 hours (MilliporeSigma) to ensure corneal clarity and absence of corneal edema. For corneal surgery, the corneoscleral discs were mounted epithelial side up on a Barron artificial anterior chamber (Katena Products, Inc., Denville, NJ).
A total of 5 corneoscleral discs (SMILE group) underwent conventional SMILE using the VisuMax 500-kHz femtosecond laser (Carl Zeiss Meditec AG; Jena, Germany). The technical principles of the SMILE procedure have been described in detail previously.6 The surgical refractive correction was −5.00 diopters of sphere (corresponding to 91 μm of central lenticule thickness) in all eyes. Programmed cap thickness was 120 μm, optical zone size was 6.5 mm, and a 3-mm incision was created at the lenticule border. A laser spot spacing of 4.5 μm was applied for the lenticule and of 2 μm for its border. The laser cut energy was set to level 32 (corresponding to 160 nJ). A further 3 corneoscleral discs (SMIL-W/O-E group) underwent the identical SMILE procedure, but after application of the femtosecond laser the preformed lenticule was not extracted from the corneal stroma nor was the incision opened. Five corneoscleral discs received FS-LASIK (FS-LASIK group). The VisuMax 500-kHz femtosecond laser was employed for creating a 120-μm flap followed by excimer laser ablation of −5.00 diopters of sphere (corresponding to an ablation depth of 70 μm) using the Wavelight Allegretto Eye-Q laser (Alcon Laboratories, Inc., Fort Worth, TX). The optical zone was programmed to 6.5 mm with a total ablation zone diameter of 8.3 mm. This Wavelight Allegretto Eye-Q system delivers a pulse frequency of 400 Hz and uses an ablation diameter of 0.95 mm. Three corneoscleral discs served as the control group without any surgical intervention.
Postoperatively, all corneoscleral discs were returned into the same organ culture medium comprising 6% of dextran for 72 hours under the conditions outlined above. Thereafter, the corneoscleral discs were manually dissected into two sagittal halves using a surgical blade: a single linear dissection was performed through the center of the SMILE incision or, respectively, through the center of the FS-LASIK flap hinge. The three control corneoscleral discs were halved along the sagittal plane at a random axis. One half of each corneoscleral disc underwent histological fixation for immunofluorescence analysis and the other half was subjected to SEM.
For SEM, specimens were dehydrated through an ethanol series from 20% to 100% ethanol following critical point drying (Critical point dryer K850; Quorum Technologies, Lewes, United Kingdom). To generate a conductive surface, samples were sputter-coated with gold (Sputter Coater 108auto; Cressington Scientific Instruments, Watford, United Kingdom) and examined with a scanning electron microscope (Microscope Leo 1550; Carl Zeiss Meditec AG) at magnifications of up to 4,000×. Only corneas of the SMILE (n = 5) and FS-LASIK (n = 5) groups were evaluated using SEM: the central posterior stromal bed was evaluated in both groups and the incision was also evaluated in the SMILE group. To enable SEM analysis of the posterior stromal bed, the LASIK flap and SMILE cap were carefully removed using fine surgical forceps and a blade.
After 72 hours of incubation in cell culture conditions as described above, corneoscleral specimens were subjected to fixation by immersion in 10% formalin for 24 hours. Subsequently, specimens were dehydrated using an ethanol series from 20% to 100% ethanol and, ultimately, transferred into paraffin wax and sliced in 4-μm sections. Afterward, the slides were dried at 37°C overnight and dewaxed in increasing concentrations of ethanol and, finally, distilled water. Antigen unmasking was performed enzymatically with 0.1% pepsin (Sigma-Aldrich) and unspecific antibody staining was blocked by 5% BSA and 0.1% Tween (Sigma-Aldrich).
To detect fragmentation of DNA as an indicator of keratocyte apoptosis, corneal specimens were assessed with a terminal deoxynucleotidyl transferase-mediated dUTP-digoxigenin nick-end labelling (TUNEL) assay (Roche Applied Science; Penzberg, Germany) as detailed by the manufacturer. Furthermore, indirect immunocytochemistry was performed on all specimens. Primary antibodies were used for Ki67 (Ki67 [C-20]: sc-7844 [goat]; Santa Cruz Biotechnology, Inc., Dallas, TX) as a marker of keratocyte proliferation, for CD11b (Purified Mouse Anti-Human CD11b/MAC-1: [ICRF44]; BD Biosciences, Franklin Lakes, NJ) as an indicator of inflammation and for fibronectin (Fibronectin [2755-8]: sc-69681; Santa Cruz Biotechnology, Inc.) as a marker for reactive fibrosis. The specimens were labeled overnight with one of the three primary antibodies according to the manufacturers' instructions with a dilution of 1:50. After washing in PBS three times, secondary antibodies (donkey IgG anti-mouse IgG [H+L] conjugated with Cy2 [Order No. 715-225-150]; donkey IgG anti-rabbit IgG [H+L] conjugated with Cy3 [Order No. 711-165-152]; donkey IgG anti-goat IgG [H+L] conjugated with Cy5 [Order No. 705-175-003]; Jackson Immunoresearch, West Grove, PA) were applied at dilutions of 1:100 for 2 hours.
Thereafter, corneal sections were observed and imaged centrally (at the corneal apex) using a fluorescence microscope (DM 2500; Leica, Wetzlar, Germany). Immunofluorescence staining intensities were subjectively graded by two blinded evaluators as “no staining” (0), “weak staining” (+), “moderate staining” (++), and “intense staining” (+++). For this purpose, the digital fluorescence photographs were displayed on a 13.3-inch computer screen (MacBook Pro Retina display; Apple Inc., Cupertino, CA) in randomized order.
The current study is the first to analyze the wound healing and inflammatory reaction of the corneal stroma after SMILE and FS-LASIK in human ex vivo corneas. Furthermore, we employed high-magnification SEM to study the wound ultrastructure of the SMILE incision and the surface texture of the stromal interface after both refractive procedures.
As this study's key finding on a cellular level, we found generally mild and comparable levels of keratocyte proliferation and apoptosis in the stromal bed after SMILE and FS-LASIK, respectively. With regard to postoperative keratocyte apoptosis and proliferation, our data are in concordance with a previous rabbit model experiment by Riau et al.2 that indicated low and comparable counts of TUNEL- and Ki67-positive cells 1 day after FS-LASIK and femtosecond lenticule extraction, respectively. Similarly, a second animal study by Dong et al.4 comprising 72 rabbits that underwent FS-LASIK or SMILE found limited keratocyte proliferation marker expression during the first month after both procedures. Furthermore, the body of evidence on keratocyte death after SMILE comprises an experiment by Mastropasqua et al.3 that found minimal apoptosis immediately after surgery in the extracted lenticules of human patients. As far as the postoperative corneal inflammatory response is concerned, our human donor cornea study was able to confirm a range of previous rabbit experiments2,4,5,7 and the human lenticule-based analysis of Mastropasqua et al.3 In agreement with our data, these previous works concordantly hinted toward a negligible to undetectable presence of immune cells in the residual stroma bed after FS-LASIK, the corneal stromal bed after SMILE, or the extracted lenticule, respectively.
As this study's main finding on the extracellular level, fibronectin expression (as a marker for fibrosis) at the site of laser injury appeared more pronounced after FS-LASIK than after SMILE. Fibronectin, an extracellular matrix glycoprotein that is produced by activated fibroblasts in the corneal stroma,5 plays a key part in corneal stromal wound healing because it provides a provisional extracellular matrix, thereby facilitating the migration of fibroblasts.8,9 Our study substantiates older animal experiment data,2 which suggested a relative excess of fibronectin expression after FS-LASIK compared with femtosecond lenticule extraction. In this animal model, the discrepancy was more pronounced with greater surgical refractive correction. The authors attribute this observation to the fact that, in FS-LASIK, the amount of laser energy delivered to the corneal stroma is to a large extent determined by the surgical refractive correction: the greater the surgical refractive correction, the more tissue is to be ablated by the excimer laser, which entails more energy being delivered to the corneal stromal bed. Accordingly, in FS-LASIK, a correction of −9.00 diopters raises the total amount of energy delivered to the corneal stroma (approximately 11.9 J) by a factor of almost three as compared with a correction of −3.00 diopters (approximately 4.1 J).2 In contrast, SMILE achieves different amounts of surgical correction by altering the shape of the lenticule that is created by the femtosecond laser. Thus, by obviating the need for excimer laser photoablation energy levels delivered to the stroma are much lower (approximately 0.58 J) and essentially independent from the amount surgical refractive correction.2 Although the current study was not designed to elucidate the effect of different surgical refractive corrections on postoperative corneal behavior, Mastropasqua et al.3 found no association between the amount of myopia corrected by SMILE (ranging between −3.75 and −10.00 diopters) and the extent of keratocyte apoptosis and inflammatory cell infiltration. Nevertheless, it needs to be considered that their study assessed extracted lenticules from human patients, which may not necessarily represent the actual cellular reaction in the stromal lenticule bed.
With regard to the effect of surgical lenticule extraction, remarkably, we observed no obvious disparity in cellular or extracellular corneal behavior between SMILE specimens with and without extraction of the lenticule. Although the number of cadaver eyes that underwent SMILE without lenticule extraction was limited, this finding may suggest that the (early) postoperative keratocyte and extracellular matrix reaction after SMILE is more likely to be triggered by the femtosecond laser itself than by the manual extraction of the lenticule. In support of this hypothesis, Riau et al.2 reported comparable levels of fibronectin expression, keratocyte death, and extracellular matrix remodeling after femtosecond lenticule extraction with and without lenticule extraction in the rabbit model. On the contrary, Liu et al.5 reported increased fibronectin expression due to surgical extraction of hyperopic SMILE lenticules from rabbit corneas. However, this finding may not be directly transferable to myopic SMILE procedures because the lenticule profile differs significantly between hyperopic SMILE (lenticule thinnest at the center) and myopic SMILE (lenticule thickest at the center), which may alter the tissue strain caused during surgical manipulation.
In addition to immunohistological analyses, the scope of this study encompassed high-magnification SEM imaging of postoperative corneal ultrastructure. Previously, our knowledge and understanding of the wound ultrastructure after SMILE was mainly derived from studying the surface quality of extracted human lenticules10–12 and from animal models.13,14 To our knowledge, the current study is the first to image the SMILE incision architecture and the surface texture of the lenticule bed after SMILE in human donor corneas by means of SEM. Although the SMILE incisions displayed sharp and uniform stromal edges, the surface texture of the stromal lenticule bed had a more irregular appearance with more fringed collagen lamellae when compared to stromal beds after FS-LASIK. On the one hand, this may be due to a smoothing effect of the excimer laser ablation that is implicit to the FSLASIK technique. Remarkably, SEM images of 500 to 4,000× magnification revealed 1.5 to 2 μm surface concavities in the SMILE lenticule bed, which were arranged in a grid-like pattern. Because the spacing of this grid closely corresponded to the programmed laser spot spacing of 4.5 μm, we hypothesize that the observed concavities represent the sites of femtosecond laser photodisruption (ie, the sites of cavitation gas bubble formation). The fact that no concavities were detectable in the FS-LASIK stroma beds may also be due to the smoothing effect of the excimer laser ablation. On the other hand, the less straightforward surgical dissection of the lenticule plane in SMILE through a 3-mm incision compared with the flap-lifting maneuver in FS-LASIK must not be overlooked as a contributing factor to the increased roughness of the lenticule bed. Arguably, these findings should be regarded as highly dependent on, and specific for, the SMILE treatment parameters (eg, laser energy level and spot spacing) and the excimer laser platform employed in this study.
Aside from the ex vivo nature of this study, further limitations exist. First and foremost, the limited number of eyes included in this post-mortem study, owing to the generally limited availability of human donor eyes for research purposes, hampered sensible statistical analysis of data. Furthermore, cultivation of corneal specimens was performed for 72 hours following surgery to allow corneal wound healing, inflammation, and tissue reaction to commence. Hence, it must be taken into account that this analysis was confined to the early postoperative period. However, it has been established in animal model experiments that keratocyte death and proliferation, inflammatory infiltration, and extracellular matrix remodeling predominantly take place during the first postoperative days after corneal refractive procedures.2,4,7,15
This first human donor eye study of SMILE and FS-LASIK endorses previous animal model experiments and studies of extracted human lenticules, suggesting that both refractive procedures elicit virtually identical and minimal keratocyte activation, cell death, and inflammation. In addition, our findings indicate more pronounced reactive stromal fibrosis following FS-LASIK, which could be due to this technique's use of two different laser wavelengths and the resulting difference in the applied laser energy level. This relatively subtle difference in the wound healing pattern between SMILE and FS-LASIK might be more accentuated at higher surgical refractive corrections because the laser energy delivered to the cornea by the former technique is constant and independent from the refractive error corrected. On the contrary, the smoothing effect of the excimer laser produced a more regular ultrastructural appearance of the FS-LASIK stromal bed as compared with the SMILE lenticule bed.