Porcine studies were conducted using fresh porcine cadaveric adenexal tissue retrieved within 24 h postmortem from a slaughterhouse. The slaughterhouse was aware of the intended use of the porcine orbit specimens for medical research but was not asked to give specific consent for the study parameters nor was this required by the IRB who reviewed the study protocol. For human specimens, a prospective study was conducted of patients undergoing wedge excision for floppy eyelid syndrome between October 1, 2016 and April 1, 2018 in a single surgeon’s practice at the Wilmer Eye Institute. After giving written informed consent, patients undergoing unilateral or bilateral wedge excisions for floppy eyelid syndrome were included. The removal of eyelid tissue was in keeping with standards of clinical care, and the tissue would have otherwise been discarded. Patients with a history of prior upper eyelid surgery or with other indications for wedge excision such as neoplasm or trichiasis were excluded. The Johns Hopkins Institutional Review Board prospectively granted approval for this study (No 00116226), which followed the tenets of the Declaration of Helsinki. Informed consent was obtained from each patient for inclusion in the study, which was HIPAA-compliant.
Porcine collagen crosslinking
A total of 73 fresh porcine adnexa were used in this study. For each experiment, porcine upper eyelid tarsus was dissected free of surrounding tissues. The conjunctiva of porcine eyelids is thicker and less adherent to the tarsal plate than in human eyelids, and was dissected off the tarsal plate completely using blunt and sharp dissection with Stevens scissors. Each porcine adnexa contributed 1 upper eyelid tarsal plate to either the treatment or control group. As porcine adnexa were obtained from the supplier as a unit already removed from the animal, it was not possible to select a contralateral eyelid from the same animal as an internal control. A 30-min riboflavin pre-treatment was then completed for both treatment and control groups, with 1 drop of riboflavin solution applied every 5 min to the anterior surface of the dissected specimens. For the treatment samples, this was followed by collagen crosslinking under UVA light application (370 nm) at irradiance of 8–9 mW/cm2 as previously described in prior experiments for corneal tissue for either 30 min or 60 min [22]. The control samples were covered with aluminum foil to block ambient UV light during this time period. During the entire irradiance period, 1 drop of riboflavin solution was applied every 5 min to both specimen groups. Specimens were then stored at 4 °C in 200 uL of balanced salt solution, and either tensile testing or imaging was then performed as below.
In the conventional Dresden protocol for corneal crosslinking, the corneal epithelium is removed and 0.1% riboflavin in isotonic 20% dextran-T-500 is applied as the photosensitizing solution [18]. A preliminary porcine experiment was performed to determine if variations of the corneal crosslinking protocols would have differing effects on tarsal plate tissue. The Dresden protocol riboflavin solution as well as several variations were tested, including 0.1% riboflavin in hypotonic 20% dextran (Sigma-Aldrich, St. Louis, MO, USA), 0.1% riboflavin in hypotonic 20% dextran with 0.05% benzalkonium chloride (BAK) (Sigma-Aldrich, St. Louis, MO, USA), and 0.1% riboflavin in isotonic 20% dextran with 0.05% BAK (Fig. 1). BAK has been shown to increase riboflavin penetration and absorption through corneal tissue, and hypotonic riboflavin solution has been described for crosslinking in thin corneas [23, 24]. In addition, corneal crosslinking has been described with both epithelium intact or with epithelial debridement (“epithelium-off”), with a greater effect seen for epithelium-off crosslinking [25]. In the preliminary porcine experiment, conjunctiva was left in place for one group (“conjunctiva-on”) and was dissected off for a second group (“conjunctiva-off”) (Fig. 1). The 0.1% riboflavin in isotonic 20% dextran solution and conjunctiva-off procedure was selected for all further porcine experiments and for all human experiments.
Human collagen crosslinking
A total of 9 patients (16 eyelids) who underwent wedge excision were included in the study. Collagen crosslinking was performed on the same day as the patient’s surgery within 8 h of the time of specimen collection, with specimens stored in balanced salt solution at 4 °C until crosslinking was performed. The tarsal plate was dissected free from the surrounding tissues, and the skin and orbicularis were completely removed from the anterior surface. The tarsal conjunctiva of human eyelids is thinner and more adherent, so the posterior surface was debrided with a #15 blade to remove the epithelium. When bilateral specimens were collected, the contralateral side was used as the experimental control. Treatment was applied to the anterior surface of the tarsal plate specimens. The riboflavin solution used was 0.1% riboflavin in isotonic 20% dextran. Pre-treatment with riboflavin was performed for both control and treatment specimens with 1 drop riboflavin administered every 5 min for 30 min. The treatment specimens were then placed under the same UVA light as for porcine specimens for 30 min, and the control specimens were covered with aluminum foil. During the 30-min irradiation period, 1 drop of riboflavin was applied to both groups every 5 min. Specimens were stored at 4 °C in 200 uL of balanced salt solution, and either tensile testing or imaging was then performed within 48 h as below.
Imaging and “flop” analysis
Porcine tarsal plate specimens were dissected in the manner described above and cut to a standard size of 7 mm wide by 30 mm long. All specimens were stored at 4 °C until the time of dissection, and were kept at room temperature from the time of dissection to photographing which was approximately 1 h. Gross photographs of porcine tissue were taken using a tripod-mounted Nikon D7100 DSLR and 105 mm lens (Nikon, Melville, NY, USA). Standardization of gross photographs for “flop” measurement was established using a horizontal metal clasp mounted on a wood frame with the axis of the clamp aligned to 180 degrees (Fig. 2a). The clasp end was positioned 2 mm from the end of the specimen. The 60 min treatment protocol was selected as this demonstrated the most significant changes (wider hyperreflective band) on OCT analysis. Photographs were taken before and immediately after treatment for 10 control specimens and 10 extended (60-min) treatment specimens. Angle of flop as degrees below 180 before and after treatment was measured accordingly from the photographs. The change in angle from before treatment to after treatment, or “delta flop,” was compared between control and 60-min treatment groups (Fig. 2b).
Anterior segment optical computed tomography (AS-OCT) of porcine and human tarsal plate specimens was performed within 48 h of crosslinking with the Zeiss Cirrus spectral domain HD-OCT device (Carl Zeiss Meditec, Dublin, CA, USA) using an anterior chamber lens. For porcine specimens, Adobe Photoshop (Adobe Systems, San Jose, CA, USA) was used to measure the area of the observed hyperreflective band with the following protocol (Fig. 3): The image coloring and tonality was standardized and compressed in two steps identically for all images: 1) the negative space of the scan was set to the black point of the image using the Levels tool and 2) the midtone point was set to 0.25. The image was then inverted and the Find Edges filter was applied to highlight the boundary of the hyperreflective band. In the central 150 pixel wide area of each scan, the Cluster Pixel Selection tool (tolerance 20) was used to highlight the areas of the hyperreflective band (Fig. 3, red) and the homogenous layer of the scan (Fig. 3, yellow). The area edges were smoothed with the Selection Smooth function with a factor of 2 and then the Measurement panel was used to calculate the area, average height, greyscale brightness, mean, minimum and maximum for each band. Comparison was then made between the 30- and 60-min treatment groups for the height and area of the hyperreflective band and the greyscale differences within specimens for the two measured areas (red and yellow). For human specimen AS-OCT, no measurements or adjustments were made (Fig. 4).
Tensile testing
Porcine and human specimens underwent collagen crosslinking with the 30-min treatment protocol as above. Uniaxial tensile testing was then performed using a MTS Criterion 43 Tensile Tester (MTS, Eden Prairie, MN, USA). Pig and human tarsal samples were each cut into uniform 5 mm strips. Tissue thickness and central width for each strip was measured using a Vernier caliper (Mitutoyo, Japan) and entered into the software. The ends of individual tarsal sample strips were then placed between sandpaper and affixed using super glue to minimize slipping during the tensile test. Strips were placed in the metal clamps of the tensile tester with a sample test length of 5 mm. Pre-loading was done at 5% as is standard for soft tissue. Samples were uniaxially stretched at 0.2 mm/s until failure using a 100 N load cell. Load/displacement measurements were recorded and analyzed. Tensile strength as indicated by peak stress and Young’s elastic modulus were measured.
Statistical analyses
Statistical analyses were performed using SAS software. Statistical differences of means were calculated using student’s t-test with P-values less than 0.05 considered significant.