Quantitative Stain-Free Conjunctival Collagen Imaging in Cicatrizing Conjunctivitis Using Second Harmonic Generation-Two Photon Excitation Technology

Abstract

Purpose: We aim to study the structure of collagen in cicatrizing conjunctivitis (CC), a disease with significant morbidity, to find sensitive and quantitative measures of severity of scarring as current progression of conjunctival scarring is reliant only on clinical assessment.

Methods: We used two-photon excitation and second harmonic generation to scan conjunctival tissues from patients with CC from Stevens-Johnson syndrome and its relation to disease severity by correlation with a validated clinical severity assessment tool. Collagen morphometry in region of interest was analyzed with FibroIndex software (HistoIndex Pte Ltd.) in conjunctival biopsies.

Results: Eighteen patients (seven CC and 11 controls) were included. Mean age was 60.7 ± 14.4 years old, with no difference between groups (P = 0.89) Compared to controls, diseased group has significantly smaller collagen area ratio (CAR) (P < 0.01), collagen fiber number (CFN)/mm² (P < 0.01) and larger collagen fiber density (P = 0.03) In diseased groups, CAR correlated with inflammation (R = −0.55, P = 0.011), scarring (R = 0.61, P = 0.0034), morbidity (R = 0.46, P = 0.035) and overall composite score (R = 0.52, P = 0.017). In all groups, CFN/mm² negatively correlated with inflammation (R = −0.50, P < 0.01), scarring (R = −0.25, P = 0.07) morbidity (R = −0.37, P < 0.01), and overall composite score (R = −0.33, P = 0.02).

Conclusions: We have further characterized the defining features in CC. CAR has significant correlation to all scores in diseased groups and, hence, may be a reliable marker for diagnosis. CFN/mm², as the only parameter with significant negative correlation with all scores, can potentially be a predictor for severity.

Cicatrizing conjunctivitis (CC) is a heterogeneous and potentially sight-threatening group of disorders characterized by conjunctival inflammation and scarring.¹ CC can lead to significant ocular morbidity and debilitating chronic pain. Bilateral visual loss, caused by the complications of chronic ocular surface inflammation and scarring, affects as many as one in five patients who have CC.^1–4 Because CC can affect individuals of all ages, they can result in significant socio-economic burden.⁴ 

Ocular mucous membrane pemphigoid (OcMMP), or ocular cicatricial pemphigoid, is the most common underlying etiology of CC in Caucasian populations.⁴ Other significant causes of CC include Stevens-Johnson syndrome (SJS), the most common cause in Asian populations, and trachoma, which predominantly affects lower socioeconomic class countries.⁵ Additional causes of CC include atopic keratoconjunctivitis, mechanical or chemical injuries, ocular rosacea, and Sjogren's syndrome.¹ 

Conditions associated with CC have been shown to have altered collagen structure. In OcMMP, autoantibodies have been shown to be predominantly directed against BP180 (also known as BPAG2, type XVII collagen), BP230, laminin 332 and type VII collagen, all of which are components of junctional adhesion complexes promoting epithelial-stromal attachment in the stratified epithelia.⁶ Binding of autoantibodies to basement membrane zone (BMZ) antigens results in a cascade of events characterized by inflammatory cell chemotaxis to the region of BMZ and secretion of proinflammatory cytokines, leading to damage to epithelial BMZ, fibroblast activation and proliferation, and type III collagen production.⁶ In particular, for OcMMP, there is a subsequent attempt at repair with aberrant type IV and VII collagen production.⁷ In CC, the contraction of the subepithelial fibrous tissue formed by collagen fibers is one of the main factors contributing to chronic cicatrization and entropion formation.⁸ Increased deposition of type IV collagen and new type V collagen formation have also been shown to contribute to the development of conjunctival fibrosis in scarred trachoma.⁹ However, the in-depth study of collagen and its structure in CC has yet to be explored. Existing methods for disease detection and severity assessment rely solely on clinical examination techniques, which are often imprecise and qualitative or, at best, semiquantitative.^10,11 There is thus a need for more sensitive and quantitative measures of disease severity in conjunctival scarring both for research and clinical practice. Such measures could also be used to assess the effectiveness of novel drugs for scarring, disease monitoring, or prognostication. 

Second harmonic generation/two-photon excitation (SHG-TPE) technology represents an innovative, cutting-edge tissue imaging system using nonlinear optical microscopy. This technology allows for the observation of natural tissue signals in unstained samples. 

SHG involves a nonlinear optical phenomenon where two photons of the same frequency interact with a nonlinear material, resulting in the emission of one photon with double the frequency (half the wavelength) of the initial photons. This technique can be used to visualize certain structures within tissue samples, particularly those that exhibit non-centrosymmetric properties, like collagen fibers, enabling the quantification of collagen structure at a finer level of detail.¹² SHG has been used to evaluate the abundance and structure of fibrillar collagen with high resolution and specificity and is highly sensitive to changes in collagen fibril and fiber structure, as well as connective tissue remodeling.^12,13 TPE is another nonlinear optical process wherein a fluorophore simultaneously absorbs two photons, leading to its excitation to a higher energy state. Combining SHG and TPE principles in SHG-TPE microscopy enables high-resolution imaging of ex vivo biological samples, particularly those containing natural nonlinear optical properties such as collagen. SHG-TPE imaging has proven to be valuable in staging fibrosis in liver cirrhosis,¹⁴ pulmonary fibrosis,¹⁵ and renal fibrosis,¹⁶ among various other tissues. 

In this study, we used a semiquantitative validated clinical severity assessment tool to grade control and diseased subjects with CC.¹⁰ We then used SHG-TPE imaging techniques to examine conjunctival tissues from these subjects, aiming to establish correlations between the SHG-TPE imaging findings and disease severity. 

Methods

Study Design and Patient Recruitment

This cross-sectional study was conducted on a cohort of adult patients diagnosed with SJS with ocular involvement under the clinical care of the Singapore National Eye Centre, identified from existing patient databases. Controls did not have a history of conjunctival pathologies or used any ophthalmic medications other than ocular lubricants. These controls were undergoing routine cataract surgery and biopsy was done at the same time as cataract surgery. The study protocol was approved by the SingHealth Centralized Institutional Review Board (CIRB Reference 2018/2738). Informed consent was obtained from all participants. The study adhered to the tenets of the Declaration of Helsinki. Patients were independently examined in the clinic by an experienced ophthalmologist (HSO) using a validated clinical assessment tool and conjunctival biopsy specimens were obtained from both the study and control groups. 

Clinical Assessment Tool

We previously developed and introduced a semiquantitative validated clinical severity assessment tool for CC.¹⁰ This is the only validated clinical tool available for CC. This tool comprises 12 components in three functional categories: (a) grading of inflammation (bulbar conjunctival hyperemia, limbitis), (b) grading of scarring (subconjunctival fibrosis, limitation in ocular motility, upper and lower fornix symblephara, upper and lower fornix depth measurement), and (c) grading of ocular morbidity, which encompasses the effects of inflammation and scarring (distichiasis, conjunctival and corneal keratinization, corneal vascularization, corneal opacity).¹³ Composite scores were derived from two main categories: (a) grading of scarring and (b) grading of morbidity (effects of inflammation and scarring). Composite scores were calculated by summing the scores of different individual components and then dividing by the number of components within each category. 

Conjunctival Biopsy Specimens

Conjunctival biopsy specimens were obtained from recruited patients using the following technique: 

• • The conjunctiva was anesthetized using preservative-free tetracaine eye drops
• • 2.0 × 2.0 mm biopsy specimens were taken from the superior bulbar conjunctiva.
• • For diseased subjects, they were taken over clinically visible diseased (scarred) areas.

Fixation and Cryosectioning

Conjunctival biopsy specimens were promptly fixed in 4% paraformaldehyde (Sigma-Aldrich, St. Louis, MO, USA) overnight. After a series of washings with 1 × PBS (1st BASE, Singapore), the tissues were embedded in the OCT cryo-compound (Leica Microsystems, Wetzlar, Germany). Serial cryosections of 8 µm in thickness of the tissues were cut with a Microm HM525 cryostat (Thermo Fisher Scientific, Waltham, MA, USA) and then mounted on Superfrost Plus glass slides (Thermo Fisher Scientific). 

Immunofluorescence Staining

Sections on the slides were initially washed in 1× PBS (1st BASE), permeabilized with 0.15% Triton X-100 (Sigma-Aldrich) for 30 minutes, blocked with 2% bovine serum albumin (Sigma-Aldrich) and 5% normal goat serum (Thermo Fisher Scientific) for one hour, and double-stained with a mouse monoclonal antibody against collagen type I (Sigma-Aldrich) and a goat polyclonal antibody against collagen type III (Southern Biotech, Birmingham, AL, USA) at 4°C overnight. The antibodies were diluted in the blocking serum. After washing with copious amounts of 1× PBS, the sections were incubated with rat anti-mouse Alexa Fluor 488 secondary antibody (Thermo Fisher Scientific) and rabbit anti-goat rhodamine Red-X (Jackson ImmunoResearch, West Grove, PA, USA) at room temperature for one hour. Slides were then washed and mounted with UltraCruz Mounting Medium containing DAPI (Santa Cruz Biotechnology, Dallas, TX, USA). A negative staining was also performed by substituting the incubation of primary antibodies with blocking serum on separate tissue sections. Sections were observed and imaged using an AxioImager Z1 microscope (Carl Zeiss, Oberkochen, Germany). All samples were taken at the same light intensity. Quantification of the fluorescence signal was carried out with ImageJ (NIH, Bethesda, MD, USA). This involved converting the image in green (collagen I) or red (collagen III) channels to eight-bit image, creating a region of interest (ROI) by hand-drawing around the conjunctival stroma, auto-thresholding, and measuring the signal intensity from the mean gray value. The mean gray value of the positive signal was subtracted from the mean gray value of the ROI in the negative staining sample. 

Multiphoton Microscopy

Some unstained cryosections were scanned using a Genesis 200 multiphoton imaging platform (HistoIndex Pte Ltd, Singapore). The samples were stimulated with a 780 nm multi-photon laser. The TPE signal was captured using photomultiplier tubes set at 550 nm channel and using a 450DCLP dichroic mirror, the TPE signal was separated from SHG signal at 390 nm. Signals were acquired with 2× frame averaging setting. Images at 20× magnification with dimensions of 200 × 200 µm² and generated at 512 × 512 pixel resolution. Photo montages of the whole biopsy specimens were automatically stitched and presented. (Fig. 1) Collagen morphometry in the ROI was then analyzed using FibroIndex software (HistoIndex Pte Ltd., Singapore). Further details of the procedure can be found in the FibroIndex user manual.¹⁷ Distribution of collagen fiber morphometric traits, including area, area ratio and fiber density, were described by normalized quantitative fibrosis parameters (Supplementary Table S1). Collagen area (CA) is the area of collagen within the area of region of interest (AROI). Collagen area ratio (CAR) is the relative amount of collagen compared to the total AROI. The higher this number, the more collagen there is relative to the total AROI. CAR in tissue is the relative amount of collagen compared to the total tissue area. The higher this number, the more collagen there is relative to the total tissue area. Collagen fiber density (CFD) is a measure of how closely packed collagen fibers are to each other within the CA. Collagen fibers nodes ratio (CFNR) is the ratio expression of the number of branching points relative to the number of collagen fibers. This is a quantitative measure of the inter-connectedness of the collagen network. A higher number indicates that there is more branches relative to the number of fibers within the collagen network. Collagen fibers N (CFN) is the number of fibers per square millimeter. Collagen fiber mean thickness (CFMT) is the average value of the collagen fibers thickness distribution. Collagen fiber mean length is the average value of the collagen fibers length distribution. Collagen area reticulation density (CARD) is the ratio expression of the number of branch points relative to the area of collagen. The higher the ratio, the more branches there are per unit area within the area of collagen. Collagen reticulation index (CRI) is the measure of how many collagen fiber branches there are within the entire length of the collagen network. Tissue area ratio (TAR) is the relative amount of tissue compared to the total region of interest. The higher this number, the more tissue there is relative to the total region of interest. 

Figure 1.

 

Immunohistochemistry and unstained SHG images of conjunctival tissues from healthy controls participants and patients with chronic ocular SJS. Left and middle panels showing representative raw immunohistology and corresponding processed grayscale images highlighting collagen Type 1 and Type 3 in the conjunctiva of control and diseased subjects; the green represents antibody positive signals and blue represents cell nuclei; Right panels show representative SHG images of the same control patient and diseased patients; red demarcation line represents the region of interest; bar graph showing higher collagen III / I ratio in diseased tissues compared to normal controls.

Statistical Analysis

The statistical analysis was performed using SPSS version 29.0 (IBM, Chicago, IL, USA). All data were expressed as mean ± standard deviation (SD). When comparing diseased and control values, the significance of differences was determined using the two-tailed Student t-test. The correlation of collagen parameters with clinical severity was analyzed using the Pearson correlation coefficient. Correlation analyses were initially conducted within the diseased groups; further analyses were extended to all groups, combining both diseased and control subjects, to demonstrate the applicability of SHG collagen parameters across all eyes. Since control eyes exhibited minimal clinical severity and very little variability, the first set of analyses was restricted to diseased eyes only to mitigate the impact of highly skewed data seen in the control group. We defined the strength of correlation as such: at least 0.8 = very strong, 0.6 up to 0.8 = moderately strong, 0.3–0.5 = fair, less than 0.3 = poor.¹⁸ Statistical significance was defined as P < 0.05. 

Results

Eighteen patients were included: seven patients with chronic ocular SJS and 11 control patients (Table 1). The causes of SJS include sulfamethoxazole-trimethoprim (two cases [28.6%]), anti-epileptics (lamotrigine, phenytoin) (two cases [28.6%]), allopurinol (one case [14.3%]), and cisplatin (one case [14.3%]). The cause of SJS in one case (14.3%) was unknown. The mean age was 60.7 ± 14.4 years old. There were no significant differences in mean age between diseased and control subjects (P = 0.89). Ten of 18 (55.6%) subjects were females. There were 14 (77.7%) Chinese, one (5.6%) Indian, and three (16.7%) Malay subjects. 

Table 1.

 

Characteristics of Study Population According to Presence of Cicatrizing Conjunctivitis

 

Collagen Parameters

Immunohistochemistry Data

The mean area of collagen I across diseased groups was 21.0 and across control groups was 55.9 (P < 0.01). The mean area of collagen III across diseased groups was 32.3 and across control groups is 35.1 (P = 0.94). The mean collagen III/I ratio in control groups was 0.66 while in the diseased groups was 1.92 (P = 0.05) (Fig. 1). There was no statistically significant correlation between collagen III/I ratio and clinical severity score (R = 0.476, P = 0.062). 

Multiphoton Data

 Figure 2 illustrates a comparison of collagen parameters between the diseased and control subjects using multiphoton data. Compared to controls, the diseased group had smaller CAR (23.20 vs. 34.78, P < 0.01), larger CFD (37.7 vs. 31.1, P = 0.03), and a lower CFN/mm² (1467.41 vs. 2347.65, P < 0.01). 

Figure 2.

 

Collagen morphometric parameters analyzed by the FibroIndex software (HistoIndex, Singapore) from SHG images of conjunctival tissues; CAR, CFD, and CFN/mm² had statistically significant differences between control and patient groups.

We correlated the three components (inflammation, scarring, and morbidity) of the clinical severity assessment tool to the collagen parameters. Table 2 shows the correlation of clinical severity scores with various collagen parameters in the diseased groups. In the diseased groups, CAR, CFD, CRI, CARD, CFNR, and CFMT were significantly correlated with inflammation score. In diseased groups, CAR (R = −0.55, P = 0.011), CFD (R = −0.61, P = 0.0031), CFMT (R = −0.46, P = 0.036) were negatively correlated with inflammation score; whereas CRI (R = 0.54, P = 0.012), CARD (R = 0.51, P = 0.018), and CFNR (R = 0.8, P < 0.001) exhibited positive correlations with the inflammation score. In diseased groups, significantly positively correlated with scarring score were the parameters, CAR (R = 0.61, P = 0.0034), TAR (R = 0.44, P = 0.046), and CFN/mm² (R = 0.71, P < 0.01). In diseased groups, significantly positively correlated with morbidity score were the parameters, CAR (R = 0.46, P = 0.035), CRI (R = 0.44, P = 0.046), and CFN/mm² (R = 0.66, P = 0.0012). However, collagen fiber mean length was negatively correlated with morbidity score (R = −0.5, P = 0.023). In diseased groups, CAR (R = 0.52, P = 0.017), TAR (R = 0.49, P = 0.024), CFN/mm² (R = 0.67, P < 0.001) were all significantly positively correlated with the overall composite score. 

Table 2.

 

Correlation of Clinical Severity Score, Graded Using the CCAT,¹⁰ With Various Collagen Morphometric Parameters in Diseased Groups

 

Table 3 shows the correlation of clinical severity score with various collagen parameters in all groups (diseased and control). In all groups, CAR was negatively correlated with the inflammation score (R = −0.63, P < 0.001), the morbidity score (R = −0.52, P < 0.001), and the overall composite score (R = −0.32, P = 0.024). On the other hand, the CFD was positively correlated to the scarring score (R = 0.36, P = 0.01) and the overall composite score (R = 0.31, P = 0.03). CFN/mm² was negatively correlated with the inflammation score (R = −0.50, P < 0.01), the morbidity score (R = −0.37, P < 0.01), and the overall composite score (R = −0.33, P = 0.02). The summary of significant positive and negative correlations between collagen morphometric parameters and clinical severity scores is found in Supplementary Table S2. 

Table 3.

 

Correlation With Clinical Severity Score, Graded Using the CCAT,¹⁰ and Various Collagen Morphometric Parameters in All Groups (Diseased and Controls)

 

Discussion

We used quantitative stain-free multiphoton microscopy to image conjunctival samples of patients with conjunctival scarring as a result of chronic ocular SJS. SHG data was used to compare the fibrillar nature of the collagen matrix and the features of collagen bundles, in the conjunctiva of diseased patients and compare them to controls. Multiphoton imaging can produce similar images to immunohistology images as seen in Figure 1. Multiphoton imaging, compared to standard stained histology techniques, uses longer wavelengths allowing for deeper penetration into tissues, providing a more comprehensive view of tissue structure and produce high-resolution 3D images with detailed molecular and structural information that can be difficult to achieve with traditional histology.¹⁹ By harnessing intrinsic contrast mechanisms like SHG, external staining is not required hence preserving the integrity of samples for further analysis.²⁰ 

Through immunostaining techniques, mean areas of collagen I and III across diseased groups was found to be also lower compared to control groups (P < 0.001 and P = 0.762, respectively). In the diseased group, we postulate that the mean area is smaller due to the contraction of the scar tissue, which may be denser. Furthermore, the mean collagen III/I ratio in diseased groups was higher compared to control groups (P = 0.05). Limited studies have explored the micro-alterations of collagen in ocular diseases, specifically focusing on the quantity and ratio of collagen. Collagen I and collagen III are types of fibrillar collagen and mainly located in the dermis. Collagen type III is generally associated with scar tissue and injury.²¹ Furthermore, the ratio of type III/type I collagen has been shown to increase with aging and increased pathology.²² Hence, the changes we observed in our study could mirror such age- and injury-related changes. In normal skin, type I collagen is predominant. However, in skin studies, increase in type III collagen expression exceeds that of type I collagen during the early stages of wound healing, with normal ratio restored only when the scar matures.²² Also, type I and III collagen have been shown to increase in active inflammation but decrease in chronic stages of trachoma³⁰ and autoimmune uveoretinitis.³¹ Our findings indicate that the mean collagen III/I ratio in diseased groups was higher. This could indicate that the conjunctival changes in diseased groups are still undergoing wound healing and are experiencing ongoing remodeling in the extracellular matrix. 

However, SHG cannot be used for imaging of nonfibrous or symmetric fibrous collagen samples.²⁶ Among the fibril-forming collagens, collagen I, II and III can result in SHG signals. Collagen IV has a nonfibrous structure and, hence, does not produce SHG signals.²⁷ Collagen V by itself does not give good SHG contrast as it forms the core of the fiber and is wrapped around by collagen.²⁸ However, the collagen fibers implicated in fibrotic diseases are usually collagen I and III,²⁹ which was the focus of our study. 

Through multi-photon imaging, we observed that compared to controls, the diseased group had significantly smaller CAR (P < 0.01), larger CFD (P = 0.03), and smaller CFN/mm² (P < 0.01). Both CAR and CFN/mm² were negatively correlated with inflammation, morbidity and overall composite scores in all groups. On the other hand, CFD was positively correlated with scarring and overall composite scores in all groups. CAR showed statistically significant correlation with inflammation, scarring, morbidity and overall composite scores in diseased groups, as well as correlation with inflammation, morbidity and overall composite scores in all groups. With significant correlation to all clinical scores in the diseased groups, CAR may thus be a reliable marker for diagnosis. In all groups, CFN/mm² could be used to accurately predict disease severity, because this parameter has a negative correlation with three out of four clinical scores. 

Scarring is known to change the dimensions of collagen fiber. In wounded ligament, the persistence of small collagen fibril diameters in the scar matrix has been described.²³ Thinner collagen fibers have been detected in conjunctiva that has undergone postoperative scarring, specifically glaucoma filtration surgery, indicating a long-term change as this structural phenotype persisted for 21 days in mature scars.¹⁷ Significantly shorter fibers have also been detected.¹⁷ This corresponds to the changes in our diseased group, which had significantly smaller CAR and smaller CFN/mm². In all groups, also found that CFN/mm² was negatively correlated with the inflammation, morbidity and overall composite scores. The lower the CFN/mm², the more likely the groups are to be diseased. 

We have also observed that the orientation of corneal collagen fibril was known to be altered secondary to pathological changes and injuries, such as corneal exposure to alkali inducing irregular arrangement of a large number of fibroblasts and collagen fibers, combined with inflammatory cell infiltration.²³ Hence, we can infer that the inflammatory changes of CC can result in higher density similar to that seen in other corneal injuries. 

Our study highlights the value of multiphoton technology, which helps to define the dynamic alterations that occur in the collagen structure in CC over time through ex vivo biopsy specimens. To further understand the structural phenotype of collagen in CC, we can utilize polarization-resolved SHG imaging to facilitate the extraction of more precise structural data about the macrostructure of the fiber bundles in fibrotic collagen, which is not possible with standard histology techniques.^24,25 This provides an objective method for assessing early signs of CC and developing drug targets for CC. We have also correlated collagen indices with the parameters of our validated CC assessment tool. CAR has significant correlation to all clinical scores in disease groups and, hence, may be a reliable marker for diagnosis. CFN/mm² is the only parameter with significantly negative correlation with all clinical scores in all groups (diseased and controls) and hence can potentially predict the severity of disease. This can be beneficial in clinical practice because it can help to analyze the collagen in conjunctival tissue, which may help clinicians assess the severity of CC. It can also help us achieve more accurate staging and prognosis, because a decrease in CFN/mm² may indicate advanced stages of CC. We can also use it to guide treatment response. Improvement in CFN/mm² may suggest a positive response to therapeutic interventions, guiding treatment strategies. Progression of scarring takes time to become clinically evident and clinical methods are not sensitive enough to detect subtle changes. Obtaining SHG-TPE images requires biopsy specimens but can potentially be used in research or clinical trials to assess more subtle earlier changes in fibrosis within the conjunctiva and as a tool in screening and evaluation of putative novel therapies. 

We acknowledge limitations to our study. A cross-sectional design facilitated the exploration of novel associations between collagen morphometric parameters derived from SHG, immunopathological findings, and clinical features of patients with SJS without waiting for disease progression, which was the primary focus of the study. As a proof-of-concept study, this design enabled the collection of sufficient data within a shorter timeframe, than would have been possible in a prospective longitudinal study in what is a rare disease. Nevertheless, the cross-sectional nature of the study did not allow for assessment of how disease progresses over time or how treatment impacts on disease progression. Although this would have been of interest, it was not an aim of this study. Using the findings from this study, we plan to incorporate such data in future longitudinal studies, which will allow us to evaluate the reliability and reproducibility across different time points for individual patients. Conducting longitudinal studies would also allow for the tracking of collagen changes using multiphoton microscopy over time in patients with CC. This could help in understanding the progression of fibrosis and the effectiveness of various therapeutic interventions. By monitoring patients over extended periods, researchers can identify early markers of disease progression and assess how collagen alterations correlate with clinical outcomes. Another limitation is that we did not include other parameters such as collagen fiber straightness and further studies can be used to explore this. We also had a relatively small sample size although there was sufficient power to detect significant differences when samples from diseased subjects were compared to those obtained controls. Further validation with a larger number of cases would strengthen our findings. Race could have been a confounding factor although there has been no studies to show any racial differences in collagen structures. In addition, although designed for use in any disease causing CC, our clinical assessment tool has only been evaluated in patients with MMP and SJS.¹¹ Further studies are required to evaluate its utility in other causes of CC. For future studies, we can consider inducing conjunctival scarring in an animal model and correlate this with SHG with regards to our severity. Future studies can also aim to investigate other causes of CC to enhance the generalizability of the findings. Last, SHG does not have the ability to discriminate between types of collagen within tissue samples. However, it is not always crucial that the types of collagens are known, especially in the context of clinical evaluation. 

Conclusions

In conclusion, our study harnessed the potential of intrinsic SHG signals to extract quantitative biomorphologic features on collagen-related changes, thereby characterizing the defining features of scar development in CC, which can be summarized as a more dense assembly of thinner and shorter collagen fibers. By correlating these parameters with our validated clinical severity assessment tool, we can predict which features will have the most impact clinically. These collagen characteristics may have potential applications as more sensitive measures for evaluating putative antifibrotic therapies, as well as the development of antifibrotic therapeutics to modulate conjunctival collagen morphology. Our discovery of collagen structural deviations may lead to the identification of new targets for anti-fibrotic drugs, as we can, for example, specifically target tissues with lower CFN/mm². 

Acknowledgments

The authors thank John Dart for his invaluable time, expertise, and constructive feedback on this study. 

Supported by the following three grants held by Hon Shing Ong: Sing-Health Foundation Grant, Singapore (Ref: SHF/FG707S/2017), SingHealth Duke-NUS Ophthalmology & Visual Sciences Academic Clinical Programme Award (Ref: 05/FY2019/P1/18-A33), and the National Medical Research Council (NMRC) Transition Award (Ref: MOH-TA24jul-0007). None of the funders had any role in the design or conduct of this research. 

Author Contributions: Conceptualization and supervision: HSO, JSM; Data curation: HSO, AR, NSHbH; Specimen provision: HSO; Formal analysis, investigation & methodology: RS, HSO, AR, NSHbH; Funding acquisition: HSO, JSM; Writing draft, review & editing: RS, HSO, AR, JKD, JSM. 

Disclosure: R. Sim, None; A.K. Riau, None; N. Shahinda Humaira binte Halim, None; J.S. Mehta, None; H.S. Ong, None