Tear Fluid Progranulin as a Noninvasive Biomarker for the Monitoring of Corneal Innervation Changes in Patients With Type 2 Diabetes Mellitus

Abstract

Purpose: This study aimed to investigate the expression levels of progranulin (PGRN) in the tears of patients with diabetic retinopathy (DR) versus healthy controls. Additionally, we sought to explore the correlation between PGRN levels and the severity of ocular surface complications in patients with diabetes.

Methods: In this prospective, single-visit, cross-sectional study, patients with DR (n = 48) and age-matched healthy controls (n = 22) were included and underwent dry eye examinations. Tear fluid was collected, and its components were analyzed using the Luminex assay. The subbasal nerve plexus of all participants was evaluated by in vivo confocal microscopy.

Results: Patients with DR exhibited more severe dry eye symptoms, along with a reduction in nerve fiber density, length, and branch density within the subbasal nerve plexus, accompanied by an increase in the number of dendritic cells. Tear PGRN levels were also significantly lower in patients with diabetes than in normal controls, and the levels of some inflammatory factors (TNF-α, IL-6, and MMP-9) were higher in patients with DR. Remarkably, the PGRN level significantly correlated with nerve fiber density (R = 0.48, P < 0.001), nerve fiber length (R = 0.65, P < 0.001), and nerve branch density (R = 0.69, P < 0.001).

Conclusions: Tear PGRN levels might reflect morphological changes in the corneal nerve plexus under diabetic conditions, suggesting that PGRN itself is a reliable indicator for predicting the advancement of neurotrophic keratopathy in patients with diabetes.

Translational Relevance: PGRN insufficiency on the ocular surface under diabetic conditions was found to be closely associated with nerve impairment, providing a novel perspective to discover the pathogenesis of diabetic complications, which could help in developing innovative therapeutic strategies.

Introduction

As the most common ocular complication of diabetes mellitus (DM), diabetic retinopathy (DR) is the primary cause of blindness and visual impairment,¹ with a global prevalence of approximately 22% among individuals diagnosed with type 2 diabetes mellitus (T2DM).² Besides the retina, DM can also affect the anterior segment, particularly the cornea. Diabetic neurotrophic keratopathy (DNK) is characterized as reduced corneal innervation, delayed corneal wound healing, persistent epithelial defects and ulcerations, and corneal edema,^3,4 and it affects approximately 46% to 64% of patients with diabetes.⁵ Although sometimes neglected for insignificancy, DNK can have profound consequences on the ocular surface, even sight-threatening,⁶ and requires further attention. Currently, the clinical evaluation of DNK includes corneal sensitivity assessment, slit-lamp biomicroscopy examination of the ocular surface, tear-film function, and in vivo confocal microscopy (IVCM) imaging of corneal nerves.⁷ Treatment strategies depend on the severity of keratopathy, ranging from the topical application of artificial tears and antibiotics to surgical intervention.^4,7 Despite current management protocols, their efficacy can sometimes be limited. Several studies have suggested that diabetic keratopathy is associated with corneal basement membrane alterations,⁸ reduced tear secretion, impaired innervation,⁹ advanced glycation end product accumulation, and oxidative stress.¹⁰ Nevertheless, the mechanisms underlying the disease have not yet been completely clarified. 

The human tear fluid contains several complex components, including proteins, lipids, electrolytes, mucins, metabolites, hormones, and desquamated epithelial cells.¹¹ Because tears come in direct contact with the ocular surface and the molecules in tears may originate from the conjunctival vessels, they could provide novel insights into the exploration of promising biomarkers in both ocular and systemic diseases.¹² Due to the minimally invasive nature of tear collection compared with the collection of blood serum or vitreous humor, the tear is now emerging as a significant source of biomarkers because of the advancing characterization strategies. Numerous studies have revealed tear biomarkers for common ocular diseases, including dry eye disease and Sjögren's syndrome, with a remarkable focus on pro-inflammatory cytokines and tear proteins.^13,14 Inflammatory mediators, including interleukin-6 (IL-6), tumor necrosis factor alpha (TNF-α),^15,16 interleukin-1 alpha (IL-1α), and matrix metalloproteinase-9 (MMP-9),^17,18 show elevated levels in ocular inflammatory conditions and correlate with dry eye disease severity. Although numerous proteins in tears have been identified in patients with DR to improve the screening methods for DR detection,^19,20 there has been limited focus on DNK. Several tear markers associated with diabetic nerve damage were recently discovered, including insulin-like growth factor binding protein 3 (IGFBP-3)²¹ and substance P.²² 

Progranulin (PGRN) is a glycoprotein with a mass of 68 to 88 kDa and plays a vital role in various physiological and pathological processes. PGRN binds to several receptors to exert anti-inflammatory effects in immune responses and promote cell proliferation and wound healing in different tissues.^23,24 Recent studies have demonstrated the neuroprotective role of PGRN in neurodegenerative diseases, including frontotemporal lobar degeneration and Alzheimer's disease, based on evidence that PGRN regulates lysosomal degradative processes and alleviates neuroinflammation.^25,26 PGRN overexpression has frequently been detected in cancer tissues, where it plays a pro-tumorigenic role by promoting cancer cell proliferation, migration, invasiveness, and immune escape.^27,28 It is also one of the adipokines that induce insulin resistance.²⁹ As PGRN can be secreted by different types of cells and detected in biofluids, it has been considered a promising diagnostic and prognostic marker for neurodegenerative diseases, cancer, and metabolic diseases.³⁰ Several studies have linked serum PGRN levels to obesity, glucose metabolism, and diabetic complications, including diabetic microangiopathy,^31,32 suggesting that circulating PGRN can be a biomarker for the early diagnosis and monitoring of disease progression in various conditions. 

To our knowledge, no previous studies have investigated the level of PGRN in tear fluid among healthy individuals or those with diabetes. Therefore, we conducted this prospective, cross-sectional study to investigate the expression level of PGRN in the tears of patients with DR and normal controls, as well as to explore the correlation between PGRN and the severity of dry eye and corneal innervation changes caused by DM. 

Materials and Methods

Study Design and Population

This was a prospective, single-visit, cross-sectional study conducted at a single medical center, wherein we investigated the tear levels of PGRN in relation to the severity of DNK. This study was approved by the Ethics Committee of Shanghai Ninth People's Hospital (2016-212-T161) and complied with the tenets of the Declaration of Helsinki. Written informed consent was signed by each patient before participating in the study. 

A total of 70 participants referred to the Department of Ophthalmology of Shanghai Ninth People's Hospital were recruited from November 3, 2022, to April 28, 2023, including patients with DR and normal controls. The inclusion criteria for the diabetic group were patients with T2DM newly diagnosed with DR according to post-dilated fundus examination and fundus fluorescence angiography based on the 2019 U.S. Diabetic Retinopathy Preferred Practice Pattern.³³ For healthy controls, the inclusion criteria were healthy volunteers of similar age. The exclusion criteria included a recent history of ocular surgery, ocular trauma, infectious keratitis and/or other ocular-associated diseases, recent use of topical medications and contact lenses, type 1 DM, cancer, or any uncontrolled systemic disease that may adversely impact study results. 

All participants underwent fasting glucose testing and eye examination, which included dry eye testing, tear collection, and IVCM examination of corneal nerves. Patients were also requested to fill in the Ocular Surface Disease Index (OSDI) questionnaire, a validated questionnaire to subjectively evaluate dry eye. 

Clinical Evaluations

First, random blood glucose levels of each participant were evaluated. Then, each patient was thoroughly examined by slit-lamp biomicroscopy to rule out corneal pathology. Corneal fluorescence staining scores were obtained according to the National Eye Institute (NEI) grading scheme³⁴ using 2% fluorescein sodium. Tear functions were evaluated using the tear-film break-up time (TBUT) and Schirmer I test for basal tear production as described by Ibrahim et al.³⁵ All tests were conducted on both eyes, and the values for the right and left eyes were averaged to achieve a final measurement. All types of examinations were conducted by similarly trained investigators (TZ and YC). 

Tear Collection

Tear fluid was collected from the inferior conjunctival fornix of both eyes (the left and right eyes combined in a tube) by a flushing method with saline solution as described by Glinská et al.³⁶ Briefly, 10 µL of sterile saline (sodium chloride injection 0.9%) was added to the inferior palpebral sulcus, and then the flushed eye content was immediately collected using 2-µL glass capillary tubes. These samples were centrifuged at 4000 rpm for 20 minutes at 4°C to remove cellular debris and then frozen and maintained at –80°C until use. 

In Vivo Confocal Microscopy

IVCM examination was performed using the Rostock Cornea Module of the Heidelberg Retina Tomograph II (Heidelberg Engineering, Heidelberg, Germany) on the right eye of each participant to acquire images of the corneal subbasal nerve plexus. Eyes were anesthetized with one drop of oxybuprocaine hydrochloride (Benoxil; Santen Pharmaceutical, Osaka, Japan). A drop of 0.3% Vidisic (carbopolymer gel; Chem.-Pharm. Fabrik GmbH, Berlin, Germany) was applied centrally on the objective lens as an optical coupling agent. The lens was manually controlled to make contact with the central surface of the cornea when the patients gazed at a fixed target. A drop of antibiotic (0.5% levofloxacin; Santen Pharmaceutical) was instilled into the conjunctival sac at the end of the examination. Each image consisted of 384 × 384 pixels covering an area of 400 × 400 µm² with an optical resolution of 0.96 pixel/µm. Four representative images of the subbasal nerve plexus at the central cornea from each participant were selected for image analysis, adhering to specific criteria, including <20% overlap between images, located between 40 and 60 µm in depth, located in the same layer, presenting excellent contrast and quality, and containing the highest value of nerve fiber density (NFD).^37,38 

Image Analysis

Images of the subbasal nerve plexus were analyzed using ImageJ software with the Neuron J plugin (National Institutes of Health, Bethesda, MD) as described previously.³⁹ The image selection and analysis were conducted by a single, masked investigator (ZD) who was kept blinded to the patient's identity. NFD was determined by calculating the number of major nerve trunks per frame. The nerve fiber length (NFL) was obtained by measuring the total length (in micrometers) of all nerve fibers in the 400 × 400-µm² confocal microscopy image. Nerve branch density (NBD) was defined as the total number of branch points per frame. Dendritic cell (DC) density was defined as the total number of counted DCs per image. The highly reflective cells with dendriform structures were first counted using the ImageJ cell counter plugin and then double-checked manually.³⁷ Data obtained from the four representative images of each participant were then averaged for statistical analysis. 

Analysis of Tear Components

The levels of PGRN, TNF-α, IL-6, MMP-9, and IL-1α in tear samples were evaluated by an immune bead–based array using a Luminex 200 instrument (Luminex Corporation, Austin, TX) according to the manufacturer's instructions. Data acquisition and analysis were performed using the Luminex xPONENT software at Novogene Bioinformatics Technology (Beijing, China). 

Statistical Analysis

R 4.3.0 (R Foundation for Statistical Computing, Vienna, Austria) was used for statistical analysis. The normality of each variable was determined using the Shapiro–Wilk test. For variables with normal distributions, the data are expressed as the mean ± standard deviation (SD), and Student's t-test was performed for analyses between two groups. For variables that followed a non-normal distribution, data are expressed as the median with interquartile range (IQR) or median with range (minimum–maximum), and the Mann–Whitney U test was performed for between-group comparisons. Furthermore, nominal variables were analyzed using the χ² test. Linear regression analysis was performed to test for relationships between tear PGRN levels and clinical parameters. The linear relationship between variables was evaluated using Pearson’s correlation coefficient (r). Multiple variable regression analysis was applied to minimize the effects of several confounding factors. For statistical significance against the null hypothesis, P < 0.001 indicated very strong evidence, P < 0.01 indicated strong evidence, P < 0.05 indicated moderate evidence, P < 0.1 showed weak evidence or a trend, and P ≥ 0.1 indicated insufficient evidence.⁴⁰ 

Results

Characterization of the Study Population

A total of 70 participants were included in this study and categorized as the DR group (n = 48 participants, 96 eyes) and control group (n = 22 participants, 44 eyes) according to the inclusion criteria. A posterior power analysis was conducted to ensure that the statistical power of the study was sufficient. Table 1 shows the demographic and clinical features of the study participants. The mean age of patients with DR was 59.15 ± 9.51 years, and the mean age of normal controls was 57.32 ± 9.66 years, with no sufficient evidence of difference (P = 0.460). Among the diabetic group, 29 patients were men (60.4%) compared to 10 men (45.5%) in the control group, demonstrating consistency in gender composition (P = 0.362) to a certain degree. Random blood glucose levels were first measured to determine each patient's glycemic condition, and they indicated that the DR group generally had higher blood glucose levels than the control group (9.32 ± 4.10 mmol/L vs. 6.95 ± 1.71 mmol/L; P = 0.011). 

Table 1.

 

Characteristics of Study Participants

 

Dry Eye Testing

According to the dry eye evaluation, the DR group demonstrated a more severe tendency in terms of dry eye parameters (Table 2). The mean TBUT was 5.27 ± 1.77 seconds in the DR group, which was significantly shorter than the 7.35 ± 3.81 seconds in the control group (P = 0.003). Results from Schirmer's test showed that the diabetic group had a much lower tear secretion volume than the control group (6.98 ± 6.62 mm compared to 12.41 ± 6.37 mm; P = 0.002). Another significant difference was observed in OSDI scores, with the DR group showing a mean score of 14.56 ± 5.69 and the control group 6.50 ± 3.13 (P < 0.001). Moreover, although there was insufficient evidence to differentiate between groups, the range of NEI staining scores was larger in the diabetic group (median, 3.00; IQR, 0.00–10.00) than in the control group (median, 2.00; IQR, 0.00–6.00; P = 0.244). 

Table 2.

 

Dry Eye Evaluations

 

Corneal Subbasal Nerve Evaluation

Figure 1 shows the representative IVCM images of the subbasal nerve plexus of controls and patients with DR. The median values of the variables regarding the corneal nerve are shown in Table 3. A significant reduction in NFD was detected in patients with DR (median, 3.00 per frame; range, 1.75–4.25) compared to the control group (median, 5.50 per frame; range, 4.00–7.67; P < 0.001) (Fig. 2A). The NFL in patients with diabetes was reduced to half (median, 1216.85 µm; range, 450.97–2037.84) compared with that in normal controls (median, 2971.53 µm; range, 1882.64–3928.15; P < 0.001) (Fig. 2B). As nerve branching positively correlated with NFL,²¹ we also observed a significant decrease in NBD in the DR group (median, 1.50 per frame; range, 0.25–4.50) compared to the control group (median, 8.67 per frame; range, 3.67–12.33; P < 0.001) (Fig. 2C). Correspondingly, the average number of DCs in patients with DR (median, 4.00 cells/frame; range, 0.25–12.50) increased compared with that in the control group (median, 2.00 cells/frame; range, 0.00–6.33; P = 0.001) (Fig. 2D). 

Figure 1.

 

Representative IVCM images of the subbasal nerve plexus. (A) Confocal image of a healthy individual. (B) Confocal image of a patient with DR with fewer nerve fibers. Scale bar: 100 µm.

Table 3.

 

Corneal Nerve Evaluations

 

Figure 2.

 

Comparison of nerve morphological changes between the diabetic group and normal control group. (A) NFD was reduced in patients with diabetes (P < 0.001). (B) A significant reduction was observed in the NFL of patients with diabetes compared with that of normal controls (P < 0.001). (C) The NBD decreased in the diabetic group (P < 0.001). (D) The number of DCs increased among the patients with diabetes (P = 0.001).

Composition Changes in Tear Fluid

We measured the concentrations of PGRN and pro-inflammatory factors (TNF-α, IL-6, MMP-9, and IL-1α) in patients with DR and normal controls. Tear PGRN levels were significantly downregulated in patients with diabetes (174,346 ± 116,994 pg/mL) compared with those in normal controls (388,802 ± 169,485 pg/mL; P < 0.001) (Fig. 3A). As anticipated, the levels of several inflammatory cytokines increased in the tears of patients with DR, including TNF-α (DR group, 5.38 ± 6.64 pg/mL; control group, 3.47 ± 0.813 pg/mL; P = 0.055, showing weak evidence), IL-6 (DR group, 68.2 ± 106 pg/mL; control group, 6.77 ± 4.38 pg/mL; P < 0.001), and MMP-9 (DR group, 18,605 ± 41,974 pg/mL; control group, 1332 ± 1568 pg/mL; P = 0.007) (Figs. 3B–3D). However, there was no significant difference in IL-1α levels between the two groups (DR group, 49.98 ± 41.07 pg/mL; control group, 56.03 ± 41.44 pg/mL; P = 0.57) (Fig. 3E). 

Figure 3.

 

Level of different components in tears quantified using Luminex assay. (A) Tear PGRN levels were significantly reduced in patients with diabetes compared with that in normal controls (P < 0.001). (B–D) The levels of three of the pro-inflammatory cytokines increased in the DR group, including TNF-α (P = 0.055) (B), IL-6 (P < 0.001) (C), and MMP-9 (P = 0.007) (D). (E) No significant difference was observed in IL-1α levels between the two groups (P = 0.57).

Clinical Correlations With Tear PGRN

A linear regression analysis was conducted to determine the relationship between tear PGRN levels and clinical parameters. Comparison of the tear PGRN level against Schirmer's score revealed a significantly positive correlation (R = 0.37; P = 0.0015) (Fig. 4B). However, there was no significant difference between PGRN level and the other two dry eye parameters, including TBUT (R = 0.17; P = 0.16) (Fig. 4A) and staining scores (R = –0.18; P = 0.13) (Fig. 4C). We further examined the correlation between PGRN levels and nerve morphological changes, which indicated that tear PGRN level positively correlated with NFD (R = 0.48; P < 0.001) (Fig. 5A), NFL (R = 0.65; P < 0.001) (Fig. 5B), and NBD (R = 0.69; P < 0.001) (Fig. 5C), but no strong correlation was detected between tear PGRN level and DC density (R = –0.17; P = 0.18) (Fig. 5D). 

Figure 4.

 

Scatter diagram with regression lines and 95% CI ranges indicating the relationship between tear PGRN level and dry eye parameters. (A) No significant correlation was observed between PGRN level and TBUT (R = 0.17; P = 0.16). (B) PGRN level was positively correlated with Schirmer's score with significance (R = 0.37; P = 0.0015). (C) No significant correlation was observed between PGRN level and staining scores (R = –0.18; P = 0.13).

Figure 5.

 

The correlation between tear PGRN level and subbasal nerve plexus changes. (A) The positive correlation between PGRN level and NFD was established (R = 0.48; P < 0.001). (B) PGRN level was positively correlated with NFL (R = 0.65; P < 0.001). (C) An excellent correlation was observed between PGRN level and NBD (R = 0.69; P < 0.001). (D) No strong correlation was detected between tear PGRN level and DC density (R = –0.17; P = 0.18).

As DR is associated with impaired tear film function and subbasal nerve damage⁴¹ and may be related to PGRN level, we considered DR a possible confounder. To ensure the validity of the results, we further conducted an analysis to adjust for confounding variables such as age and gender, which are commonly known to affect study outcomes. As depicted in Figure 6, after adding the effects of possible confounding factors (gender, age, and DR), the model showed no evidence of correlation between tear PGRN level and Schirmer's test results (P = 0.15) (Fig. 6A). Although tear PGRN level weakly correlated with NFD (P = 0.051) (Fig. 6B) and NFL (P = 0.069, showing weak evidence) (Fig. 6C), a strong correlation was detected between tear PGRN level and NBD (P = 0.0059) (Fig. 6D). Furthermore, DR was identified as a solid confounder affecting both nerve variables and PGRN level (P < 0.001). As shown in Figure 7, a differentially expressed protein, PGRN, was identified as a potential biomarker in the tears of patients with DR and healthy individuals. 

Figure 6.

 

Further analysis for adjusting confounders (gender, age, and DR). In the scatter diagram, dots indicate female, triangles indicate male, and blue and red represent the DR and control groups, respectively. The size of the legend increases with age. (A) The correlation between tear PGRN level and Schirmer's test result was not significant considering the three covariates (P = 0.15). (B) There was a weak association between PGRN levels and NFD (P = 0.051). (C) PGRN level was weakly correlated with NFL (P = 0.069). (D) Comparison of tear PGRN level with NBD revealed significance (P = 0.0059).

Figure 7.

 

Graphical abstract of the study findings.

Discussion

Over half a billion adults worldwide are living with diabetes, placing a significant burden on individuals and health systems. Microvascular complications such as retinopathy, nephropathy, and peripheral neuropathy are commonly associated with DM. Up to 54% of patients with diabetes suffer from peripheral neuropathy, resulting in morbidity and reduced quality of life.⁴² DNK occurs as a consequence of the neuropathy of the trigeminal nerve in the ophthalmic division, which is characterized by corneal nerve damage, superficial punctate keratitis, and persistent epithelial defects. Although the levels of PGRN have been investigated in the serum and vitreous humor of patients with diabetes,^43–45 its expression in tears has not been investigated. In this study, we compared tear PGRN levels between patients with DR and healthy controls and found that the diabetic group had a significantly reduced concentration of tear PGRN. We also evaluated the correlations between PGRN level and keratopathy. As reported by a previous study that DM would aggravate the risk of dry eye,⁴⁶ we also found that participants with diabetes exhibited more severe dry eye signs and symptoms. Furthermore, we performed IVCM examinations to observe changes in the subbasal nerve plexus in patients with DR and observed a significant reduction in NFD, NFL, and NBD, as well as an elevated number of DCs. A strong positive correlation was observed between PGRN level and Schirmer's score, NFD, NFL, and NBD. After adjusting for multiple covariates (including gender, age, and DR), the PGRN level remained highly associated with nerve parameters, especially NBD. 

Several methods are used to examine corneal nerve morphology and physiological alterations in ocular and systemic diseases, such as corneal esthesiometry, histological examination, and IVCM.⁴⁷ Currently, corneal confocal microscopy is considered a minimally invasive technique for evaluating corneal innervation with a high-quality resolution.⁴⁸ However, it is relatively time consuming and requires the patient's lengthy cooperation for accuracy. Our study proposed a novel non-invasive and cost-effective method for screening corneal nerve changes. Through simple tear collection and biomarker detection, it would be possible to predict nerve changes and personalize prevention and treatment. A previous study reported that the degree of corneal nerve damage is strongly associated with the severity of somatic neuropathy in patients with diabetes.⁴⁹ Therefore, interpretation of corneal nerve plexus impairment has been acknowledged as an effective method for detecting, staging the severity of, and monitoring the progression of diabetic polyneuropathy.^7,50,51 Another previous study demonstrated that NFL and NBD in patients with DR reduced significantly, showing a progressive decrease with the worsening of DR.⁵² Considering that common DR therapies such as argon laser photocoagulation, anti-VEGF treatment, and vitrectomies may impair the regeneration of the corneal subbasal nerve plexus,^53,54 which may delay or even worsen the restoration of the ocular surface, it is useful to evaluate corneal nerve damage for clinical prediction and precaution. 

PGRN is a multifunctional protein expressed in various cell types, with high levels in the retina.^55,56 According to previous research, PGRN can bind to multiple receptors, activate downstream signaling pathways, and exert therapeutic effects.⁵⁷ PGRN competitively inhibits the TNF-α–activated nuclear factor kappa B (NF-κB) inflammatory pathway by binding with tumor necrosis factor receptor 1 (TNFR1).^58,59 PGRN can act extracellularly with notch receptors⁶⁰ and functions as a neuronal survival and axonal growth factor.⁶¹ Through interactions with sortilin⁶² or prosaposin,⁶³ PGRN can also enter cell lysosomes and act as a chaperone for lysosomal enzymes, such as cathepsin D, which plays a role in the degradation of misfolded proteins.⁶⁴ In the central nervous system, PGRN deficiency promotes microglial neurotoxicity⁶⁵ and lysosomal dysfunction.⁶⁶ Our results suggest that PGRN is detectable in human tears and its level remains relatively high in the normal state. Moreover, its insufficiency under diabetic conditions is closely associated with limited tear production and nerve impairment to a greater extent. Although its exact origin is not completely understood, we hypothesized that tear PGRN is secreted by corneal epithelial cells, neurons, and the lacrimal gland. As no studies conducted to date have demonstrated the expression of PGRN on the ocular surface, further research is necessary to prove our hypothesis. 

Although serum levels of PGRN were found to be increased in patients with T2DM,^43,44 and PGRN levels increased during wound healing⁶⁷ or inflammatory processes,⁶⁸ which may contrast with our results, we speculate that the reduction of tear PGRN level in patients with DR was reasonable. First, serum PGRN level could be affected by multiple factors, such as the overproduction by adipose tissue and epithelial tissue and reduced renal elimination due to nephropathy,³² whereas tear PGRN has a limited origin and more of a narrow reflection of the ocular microenvironment. Moreover, besides its anti-inflammatory effect, PGRN is a neurotrophic factor. In our study, PGRN level was found to be more associated with corneal nerve morphological variations than with Schirmer's test results, which to some degree suggested that PGRN deficiency in the ocular surface is closely associated with corneal nerve damage. Finally, the intact form of PGRN can be cleaved into small granulin peptides by proteases⁶⁹ such as MMP-9,⁷⁰ MMP-12,⁷¹ proteinase 3, and neutrophil elastase.⁷² Because we detected the level of full-length PGRN in tears and observed an increased concentration of MMP-9 in the diabetic group, our results suggest that the downregulation of tear PGRN level in patients with diabetes could partially be due to the cleavage by activated proteases. 

A recent clinical study by Bilgin et al.⁴⁵ revealed no significant difference in vitreous PGRN levels between patients with or without DR. Considering the different sample sources (vitreous humor vs. tear), ethnic origins (Caucasian vs. Asian), and methods of measurement (ELISA vs. Luminex), the unmatched conclusions were not quite unanticipated. We observed that the level of PGRN in tears (174.346 ± 116.994 ng/mL in the DR group and 388.802 ± 169.485 ng/mL in the control group) might be much higher than that in the vitreous humor (13.78 ± 3.87 ng/mL in the DR group and 15.85 ± 5.93 ng/mL in the control group), which could be due to the smaller tear sample volumes we collected with higher enrichment. 

To our knowledge, this is the first study to explore the relationship between tear PGRN and diabetic keratopathy, and it had some limitations. First, the size of the study population was small. As it was conducted at a single medical center, the regional distribution of participants may be restricted. Second, it would be beneficial to complement the study with several additional clinical parameters such as corneal esthesiometry. It may also be useful to measure serum PGRN levels simultaneously to completely elucidate the role of PGRN in the pathology of T2DM complications. Third, this was a cross-sectional study; therefore, we could not determine whether the relationship between tear PGRN levels and DNK was causative or correlational, and longitudinal follow-up research is necessary to observe dynamic changes. As reported by Semeraro et al.,³ DNK can be classified into three stages. Stage 1 is characterized by superficial punctate keratitis, stage 2 consists of persistent epithelial defects and stroma swelling, and stage 3 includes deep ulcerations and thinning of the cornea. Although we observed several cases with moderate area of superficial epithelial defects without stroma swelling, we could not determine whether they were persistent, and we also could not perform further grading during patients’ single visits. Finally, when examining the subbasal nerve plexus, we selected IVCM images with substantial NFD to ensure uniform criteria for comparison. This selection methodology may not align with other published studies, which may prevent direct comparison of our study results. 

To summarize, PGRN in the tears of patients with DR and healthy individuals was identified as a potential biomarker. Moreover, PGRN level correlated with tear secretion volume, NFD, NFL, and especially NBD with significance. These data provide evidence of the potential of using tear fluid–based PGRN as a biomarker for the diagnosis and staging of DNK; however, further research is necessary to validate these findings. 

Conclusions

This prospective, cross-sectional study investigated tear PGRN levels in patients with newly diagnosed DR and normal individuals. PGRN levels were significantly downregulated in the tears of patients with diabetes and showed a strong correlation with corneal subbasal nerve changes, including NFD, NFL, and NBD. Therefore, PGRN is a promising indicator for diagnosing DNK non-invasively and predicting its progression. Moreover, it may provide a novel perspective to discover the pathogenesis of diabetic complications. Additional studies are required to evaluate PGRN levels at different stages of the disease to gain a deeper understanding of the underlying mechanisms. 

Acknowledgments

The authors express their sincere gratitude to the Department of Ophthalmology, Shanghai Ninth People's Hospital, and Shanghai Key Laboratory of Orbital Diseases and Ocular Oncology for providing the research platform. 

Supported by grants from the National Natural Science Foundation of China (82271041, 82070919), the Program of Shanghai Academic/Technology Research Leader (22XD1401800), and the Biomaterials and Regenerative Medicine Institute Cooperative Research Project, Shanghai Jiao Tong University School of Medicine (2022LHA06). 

Disclosure: T. Zhou, None; Z. Dou, None; Y. Cai, None; D. Zhu, None; Y. Fu, None