Meibomian gland dysfunction: a route of ocular graft-versus-host disease progression that drives a vicious cycle of ocular surface inflammatory damage.

Abstract

Purpose:

To investigate the role of aggressive Meibomian gland dysfunction (MGD) in the immune-pathogenesis of ocular graft-versus-host disease (GVHD).

Methods:

In mice, an allogeneic GVHD model was established by transferring bone marrow (BM) and purified splenic T cells from C57BL/6J mice into irradiated C3-SW.H2b mice (BM+T). Control groups received BM-only (BMO). Mice were scored clinically across the post-transplantation period. MGD severity was categorized using degree of atrophy on harvested lids. Immune disease was analyzed using flow cytometry of tissues along with fluorescent tracking of BM cells onto the ocular surface. In humans, parameters from 57 patients with ocular GVHD presenting to the Duke Eye Center were retrospectively reviewed. MGD was categorized using the degree of atrophy on meibographs. Immune analysis was done using high-parameter flow cytometry on tear samples.

Results:

Compared to BMO, BM+T mice had higher systemic disease scores that correlated with tear fluid loss and eyelid edema. BM+T had higher immune cell infiltration in the ocular tissues and higher CD4+-cell cytokine expression in draining lymph nodes. BM+T mice with worse MGD scores had significantly worse corneal staining. In ocular GVHD patients, 96% had other organs affected. Ocular GVHD patients had abnormal parameters on dry eye testing, high matrix metalloproteinase-9 positivity (92%), and abundance of immune cells in tear samples.

Ocular surface disease signs were worse in patients with higher MGD severity scores.

Conclusions:

Ocular GVHD is driven by a systemic, T cell-dependent process that causes Meibomian gland damage and induces a robust form of ocular surface disease that correlates with MGD severity.

1. INTRODUCTION

Hematopoietic Stem Cell Transplant (HSCT) is a therapeutic, and potentially curative, option for the treatment of both benign and malignant hematologic diseases ^1,2. One of the common complications of allogeneic-hematopoietic stem cell transplant (aHSCT) is an inflammatory attack on the host’s tissues by the donor immune cells which is referred to as graft-versus-host disease (GVHD) ^1–5. oGVHD is a T cell-mediated response, whereby donor T cells target histocompatibility antigens on the host cells, thus triggering the release of inflammatory cytokines ^6–10. This, in turn, leads to a cascade of events that initiate the activation, proliferation, and differentiation of an array of immune components that drive the disease process. This robust inflammatory condition can manifest in a variety of organ systems and commonly presents in the gut, skin, lungs, and eye ¹.

Ocular GVHD (oGVHD) occurs when GVHD involves the ocular tissues and is estimated to affect 40 to 60% of patients undergoing aHSCT ^8,11,12. The hallmark of oGVHD is an immune-driven cascade that can affect any tissue of the ocular adnexa, particularly the Meibomian glands, thereby contributing to an aggressive form of inflammatory dry eye disease (DED) ^6–8,13,14. One of the risk factors associated with obstructive MGD is the presence of chronic eyelid inflammation as seen in oGVHD ^1,15–18. In such situations, extensive inflammatory damage to the Meibomian glands disrupts the quantity and quality of the gland’s secretions resulting in a severe form of Meibomian gland dysfunction (MGD), the leading cause of evaporative DED ^19–22. This is accompanied by a myriad of manifestations resulting in troubling symptoms, persistent ocular surface damage, and disturbed visual acuity, thereby considerably affecting the quality of life ^19,23–25.

Currently, our grasp of the pathogenesis of MGD in relation to oGVHD and its disease manifestations remains limited, which fundamentally restricts the therapeutic options available. Notably, there is no gold-standard pharmaceutical treatment for MGD, and therapy is limited to conservative, palliative measures that lack a wide repertoire of options ^1,8,21,26,27. As such, there is significant interest in improving the therapeutic arsenal for patients with severe MGD and DED such as those present in oGVHD. To optimize the identification and management of patients with oGVHD, and potentially generalizable to other forms inflammatory DED, it is crucial to improve understanding of MGD pathophysiology in these patients, including the signs and symptoms which arise in both mice and humans.

In this work, we exploit the inflammatory milieu of an established murine model of oGVHD, in addition to the ample population of patients with oGVHD at the Foster Center for Ocular Immunology at Duke Eye Center, to investigate the pathophysiology underlying the progression of MGD in oGVHD and its downstream manifestations from both a clinical and immune disease standpoint.

We hypothesize that the early recruitment of donor T cells into the ocular adnexa after aHSCT orchestrates inflammatory responses that induce the progression of MGD, which contributes to the development of keratopathy, poor vision, and decreased quality of life. Understanding the disease processes which underpin the signs and symptoms that manifest at the level of the ocular surface would allow promoting an approach that transcends symptomatic treatment and targets the cause of such manifestations.

2. METHODS:

This study was approved by the Duke University Hospital Institutional Review Board (protocol # Pro00095121) with informed consent obtained for the prospective collection of tear wash samples from consenting patients and a waiver of informed consent for chart review of patients presenting to the Foster Center for Ocular Immunology.

All animal studies were conducted according to protocols approved by Duke University and the University of Miami’s Animal Care and Use Committees and in accordance with the Association for Research in Vision and Ophthalmology’s (ARVO) Statement for the Use of Animals in Ophthalmic and Vision Research.

2.1. PRE-CLINICAL MURINE oGVHD MODEL OF HSCT

2.1.1. Animals:

C57BL/6J (B6) (H2b), C3H.SW (H2b), enhanced green fluorescent protein B6-EGFP transgenic (H2b), were purchased from Jackson Laboratory and maintained at the university facilities in pathogen-free conditions. All mice used in experiments were 8–10 weeks old, exhibited no obvious ocular surface and eyelid disease at baseline, and were fed a standard caloric diet for their age. The animals were routinely monitored prior to all procedures and until the experiment ended.

2.1.2. Induction of ocular and systemic GVHD:

An MHC I and II-matched with minor mismatch allo-HSCT mouse model was used. Recipient mice (C3H.SW, H2b, Ly9.1+) were split into a diseased group receiving T cell-depleted (TCD) bone marrow and purified T cells (BM+T), and a control group receiving TCD bone marrow only (BMO). All recipient mice received ablative conditioning with a single dose of 10.5 cGy total body irradiation (TBI, x-ray, and Cesium) on the day of the transplant. Mice were provided antibiotic water (gentamycin, 25 mg/gallon) from day-3 till day 14 post-transplant for prophylaxis against bacterial infection.

In the standard GVHD experiments, the donor cells were obtained from wild-type mice. Donor B6 mice (H2b, Ly9.1+) were euthanized by cervical dislocation, and tissues were processed as previously described ^28–30. In short, femurs and tibiae were removed from donor mice. From the collected long bones, bone marrow cells were flushed with cold RPMI-1640 using a syringe fitted with a 26-gauge needle. Donor marrow inoculum (TCD-BM) was prepared using anti-Thy-1.2 Miltenyi MACS (San Diego, CA, USA) magnetic beads and negative selection to remove T cells, washed, and adjusted before transplant to 1 × 10⁷/mL. To prepare donor T cells, spleen cells were mixed with CD90.2 microbeads (Miltenyi Biotec) and passed through MACS separation columns attached to a magnetic apparatus. Selected cells were collected, and a small aliquot was stained with anti-CD4 and anti-CD8 mAb (BD Pharmingen, San Diego, CA, USA) to determine precise percentage compositions. Cell suspensions for the BM+T group were attained by mixing a solution with 2.3 × 10⁷/ml purified T cells with a solution with 5 × 10⁷/ml TCD-BM cells. Cell suspensions for the BM+T group were attained by mixing a solution with 5 × 10⁷/ml TCD-BM cells with an equal volume of PBS. For the immune tracking experiments, B6-EGFP transgenic (H2b) with fluorescently labeled bone marrow cells were used as donors. For the remainder of the experiments, bone marrow cells were obtained from wild-type B6 mice.

Through tail vein injection, BM+T group recipient mice received 5×10⁶ TCD bone marrow cells and 2.3×10⁶ purified T cells in 200 uL whereas BMO group recipient mice received 5×10⁶ TCD bone marrow cells in 200 uL of sterile saline.

2.1.3. GVHD clinical disease assessment:

Mice were clinically assessed and followed up 3 times a week for 21 days as previously described ^28–32. In short, for systemic disease, a composite systemic GVHD score (total of 14) was used and consisted of the following scores ranging from 0–2 each: Weight loss, Activity, Posture, Fur Texture, Alopecia, Skin Integrity, Diarrhea. For ocular disease, eyelid edema was assessed using a predetermined scale and was scored from 0–4. Tear production was measured in millimeters using phenol red-impregnated cotton threads. Epitheliopathy was assessed by determining corneal mean green fluorescence intensity (MGI) using fluorescein-stained photographs under a blue filter.

Meibomian gland dysfunction (MGD) was followed up using a pre-determined scale based on plugging quantity and quality at the lid margin (no plugging: score 0; [1–4] plugs and no cysts: score 1; >4 plugs or presence of cysts: score 2; >4 plugs and presence of cysts: score 3). In addition, MGD was assessed at the endpoint using photographs of the interior of the harvested lids of each mouse and quantified based on the degree of atrophy/dropout (no atrophy: score 0; 1–24% atrophy: score 1; 25–49% atrophy: score 2; 50–74% atrophy: score 3; 75–100% atrophy: score 4).

Ocular and systemic clinical parameters measured were graphed versus time using GraphPad Prism version 9.1.1. for Microsoft Windows.

2.1.4. Immune disease assessment of Systemic and Ocular GVHD

Tissues were harvested on day 21 and assessed using flow cytometry as previously described ^28–30,33. Submandibular lymph nodes (SLN) were harvested at the end-point and single-cell suspensions were prepared in PBS+2% fetal calf serum + sodium azide. Flow cytometry for oGVHD associated inflammatory cytokines, specifically IFNγ, IL-2, and IL-17 expressed in T cells was done using intracellular staining with cell marker assessment for immune/effector cell phenotyping. The antibody panels used for the lymph nodes: CD4: PE Dazzle (BioLegend); CD8: BV785 (BioLegend); IFN-g: PE (BioLegend); IL-17: FITC (BD Horizon) Live/Dead: Zombie Yellow.

Conjunctival tissue was harvested at the endpoint and incubated with collagenase (Sigma-Aldrich Corp., St. Louis, MO, USA) for 60 minutes at 37^oC. Afterward, the samples were strained to obtain a single cell suspension using cell strainers (BD Falcon, Franklin Lakes, NJ, USA). Cells were stained with the following panel: F4/80: PE (BioLegend); CD64: PerCPCy5.5 (BioLegend); CD11b: BV510 (BD horizon); CD45: BV650 (BioLegend); Ly6G: Alx700 (BioLegend); CD11c: BV711 (BD horizon); IA/IE: APC Cy7 (BioLegend); CD8a: BV785 (BioLegend); CD4: PE Dazzle (BioLegend); Ly6C: PE Cy7 (BioLegend); Siglec F: APC (BD Horizon); Live/Dead: Zombie Yellow.

Flow cytometry results were analyzed using FACS Diva software (version 10.8.1, BD, San Jose, CA, USA) and FlowJo software (version 9.0.1, Tree Star, Ashland, OR, USA). The gating strategy used is represented in Supplementary Figure 1 ³⁴.

In Vivo Ocular Fluorescent Stereomicroscopy was performed in recipients of EGFP-labeled cell populations as previously described ³⁰. Mice were evaluated at different time-points using fluorescent microscopy (Leica MZ16FA, EGFP 2 filter [excitation 480/40 nm, barrier 510 nm]), allowing measurement of EGFP expression in the cornea, which was quantified as mean green intensity (MGI) as described previously ³⁵. Photographs of the cornea were taken using the bio-microscope in a standardized fashion for all eyes. The image was analyzed using Adobe Photoshop v21.2.4.323 for Microsoft Windows, and the MGI index was calculated for each cornea.

2.2. CLINICAL STUDY OF oGVHD IN PATIENTS THAT ARE RECIPIENTS OF HSCT

2.2.1. Human subjects:

A single-center retrospective chart review was conducted on a database of patients with the diagnosis code Graft- versus-host disease, unspecified (CMS-HCC), Chronic graft-versus-host disease (CMS-HCC), or Acute graft-versus-host disease (CMS-HCC) presenting to the Foster Center for Ocular Immunology at the Duke Eye center from January 1, 2017, to December 31, 2021. In addition, prospective unilateral tear collection was performed on consenting participants from a subset of oGVHD patients visiting the Foster Center for Ocular Immunology.

Inclusion criteria were patients over 18 years of age who had one of the diagnosis codes listed above and presented within the aforementioned timeframe.

2.2.2. Clinical study variables and characteristics:

A combination of demographic and clinical characteristics was noted for this study. Parameters noted at first presentation included: age, self-identified race, sex assigned at birth, primary malignancy, documentation of GVHD in non-ocular tissue. In addition, live/dead status on January 1, 2022, was noted. Eye exam parameters of both eyes noted at presentation were included in the following order: Ocular Surface Disease Index (OSDI, Allergan) questionnaire score ³⁶, central corneal sensitivity, measured with the Cochet-Bonnet esthesiometer, Meibomian gland atrophy documented with the Keratograph 5M meibography (Oculus, Germany) graded using the Meiboscale (score 0: no atrophy, score 1: [1–25]% atrophy, score 2: ]25–50]% atrophy, score 3: ]50–75]% atrophy, score 4: ]75–100]% atrophy) in accordance with Pult et al ^37,38 , matrix metalloproteinase-9 (MMP-9) positivity, testing with the Inflammadry device (Quidel Corporation, USA), tear break up time (TBUT), corneal staining sodium fluorescein staining (NaFl) graded using the NEI score ³⁹, conjunctival lissamine green staining using the NEI grading score and tear production using Schirmer’s test without anesthesia measured at 5 min

2.2.3. Immune tear cytometry analysis:

The immune assessment of the collected tear samples was performed using high parameter 36-panel flow cytometry on unilaterally collected tear samples from consenting patients. The panel used is: CCR7: BV 421^™ (BioLegend); CCR6: BV 711^™ (BioLegend); CXCR5: BV750 (BD Biosciences); CCR5: BUV563 (BD Biosciences); CXCR3: PE-Cyanine7 (Thermo); TCR g/d: PerCP-eFluor 710 (Thermo); CD38: APC-Fire810 (BioLegend); CD57: FITC (BioLegend); CD95: PE-Cyanine5 (Thermo); CD19: Spark NIR^™ 685 (BioLegend); CD14: Spark Blue^™ 550 (BioLegend); CD45: PerCP (Thermo);CD25: PE (Thermo); CD27: APC (Thermo); HLA-DR: APC-eFluor 780 (Thermo); CD11b: PerCP/Cyanine5.5 (BioLegend); CD24: PE-eFluor 610 (Thermo); CD1c: Alexa Fluor^® 647 (BioLegend); CD127: APC-R700 (BD Biosciences); CD16: BUV496 (BD Biosciences); IgD: BV480 (BD Biosciences); CD45RA: BUV395 (BD Biosciences); CD56: BUV737 (BD Biosciences); CD8: BUV805 (BD Biosciences); CD11c: BUV661 (BD Biosciences); CD123: Super Bright 436; IgM: BV 570^™ (BioLegend); CD28: BV 650^™ (BioLegend); CD141: BB515 (BD Biosciences); CD161: eFluor 450 (BD Biosciences); CD3: BV 510^™ (BioLegend); CD20: Pacific Orange (Thermo); IgG: BV605 (BD Biosciences); PD-1: BV 785^™ (BioLegend); CD4: BUV615 (BD Biosciences). Immune cells were analyzed by using unbiased cluster analysis to determine differences in the population of immune cells in patients with a diagnosis of oGVHD in comparison to those with other forms of dry eye disease (DED) as described previously ⁴⁰. Samples were acquired using a four-laser Cytek Aurora Spectral Flow Cytometry System. Single color controls for spectral unmixing were acquired with PBMCs from healthy control blood and UltraComp eBeads (ThermoFisher). Raw data were unmixed and further analyzed using either FlowJo for manual gating or Omiq (https://www.omiq.ai) for clustering visualization and analysis.

2.3. STATISTICS AND ANALYSIS

Comparison between continuous variables of the control and experimental groups at the endpoint was done using t- tests for the following variables in (1) mice: systemic score, weight loss %, tear fluid production, eyelid edema, conjunctival cell counts and ratios, SMLN CD4 % expression of cytokines, MGD plugging scores, fluorescein staining epitheliopathy (MGI) and (2) humans: Tear breakup time (TBUT), and corneal fluorescein staining scores. Shapiro-Wilk test was used to assess for normality and non-parametric t-test were used when values were not normally distributed. For categorical variable comparison, a Chi-square test was used for the matrix metalloproteinase-9 (MMP-9) positivity in the MGD severity groups. In addition, systemic disease scores were analyzed for correlation tear production and/or eyelid edema scores using Spearman’s correlation measure. In all statistical tests, a P<.05 was used as a cutoff for significance.

3. RESULTS:

1. MURINE GVHD

3.1.1. An MHC I and II-matched unrelated (“MUD”) minor mismatch allo-HSCT mouse model demonstrates the development of a significant systemic GVHD that correlated with the severity of ocular disease:

The MHC-matched oGVHD mouse model (C57Bl/6 → C3H.SW) consisting of two recipient groups: BMO (n=20) and BM+T (n=20). Baseline weight for the BM+T and BMO groups showed no significant differences at 20.6 +/− 2.0 g and 20.1 +/− 1.4 g, respectively (p>0.1). The BM+T mice developed progressive systemic disease, as shown through the elevated composite score assessment and a higher percentage of weight loss throughout the 3-week assessment compared to the BMO group that did not receive T cells. On day 21, the BM+T mice had a significantly higher systemic disease score (6.75 vs 0.5, P<.0001, Figure 1A) and a significantly lower weight percentage of baseline (74% vs 91%, P<.0001, Figure 1B) compared to the BMO group that did not receive T cells. Survival rates on day 21 were 85% and 100% for the BM+T and BMO groups, respectively.

Figure 1. Assessment of systemic disease in the murine oGVHD mouse model.

(A) composite systemic score and (B) weight percent of baseline demonstrated increasing systemic disease development. (C) Survival >80% enabled assessment of the ocular adnexa. (D) Progression of ocular scores in the oGVHD model was concomitantly carried out. The endpoint ocular findings are showcased in photographs representative of BMO vs BM+T eyes. Ocular scores were also graphed across time for (E) tear production and (F) eyelid edema and were significantly worse in the BM+T group compared to the BMO. When correlation assessment was carried out, the systemic disease severity score approached significant correlation with tear production score (G) and was significantly correlated with eyelid edema score (H). (n=20 mice/group). (BMO: bone marrow only recipients; BM+T: recipients of bone marrow and T cells)

BMO and BM+T mice were also assessed using an established ocular scoring system using a predetermined scale for eyelid edema and Meibomian gland plugging ³¹. BM+T mice developed progressively decreased tear production and increased eyelid edema (Figures 1D). At the day 21 end-point, BM+T transplanted mice had significantly lower tear production (4 mm vs 2 mm, P<.0001, Figure E) and higher eyelid edema score (2 vs 0, P<.0001, Figure 1F) compared to the BMO group. In addition, systemic GVHD score was significantly correlated with the eyelid edema score (r=0.38, P=.01) (Figures 1G and 1H).

3.1.2. Immune analyses of the ocular compartment:

On day 21 flow cytometry, conjunctiva from the BM+T group had a higher CD45 cell count compared to BMO controls without T cells (2187 vs 1260, P=.021, Figure 2A). The BM+T mice had a significantly higher PMN count (202 vs 53, P=.001) and CD8 cell count (867 vs 426, P=.04) (figure 2B). There were no significant differences between the BM+T and BMO groups with regards to monocyte count (254 vs 124, P=.1) and CD4 cell count (222 vs 179, P=.52) (Figure 2B). Within the CD45 population, the BM+T group demonstrated a higher composition of the 4 immune cells of interest (67% vs 50%, P<.0001) and overall higher composition of CD8 cells (37% vs 27%, P=.04) (Figure 2C). Upon testing the CD4/CD8 ratio, the BM+T mice had a significantly lower ratio compared to BMO in which T cells were absent (0.45 vs 0.23, P=.0004, Figure 2D)

Figure 2. Flow cytometry analysis of the conjunctiva of the BM+T vs BMO groups.

Flow cytometry analysis demonstrated higher CD45⁺ immune cell counts and infiltration in the BM+T groups compared to that of the BMO mice (A). When the subpopulations were assessed the BM+T group had a significantly higher PMN and CD8 cell count per conjunctiva compared and a higher CD4 and mononuclear count to the BMO (B). In addition, percent composition of CD45⁺s in each group demonstrated higher overall composition of the 4 immune cells tested and a higher CD8% composition in the BM+T group (C). Moreover, when compared, the CD4/CD8 ratio is significantly higher in the BM+T group compared to the BMO (D). (n=10 mice/group) (BMO: bone marrow only recipients; BM+T: recipients of bone marrow and T cells)

Moreover, CD4 T cells of the SMLN demonstrated significantly higher expression of IFN-g (1.6 vs 0.22, P=.015), and IL-17 (2.09 vs 0.44, P=.017, Figure 3) in the BM+T group compared to the BMO controls.

Figure 3. Analysis of immunological phenotypes of the CD4⁺ T-cell population in the SMLN.

SMLN were harvested on day 21 and cells were stained for CD4+ and intracellular IFN-γ and IL-17 expression. Results demonstrated significantly higher expression of IFN-γ and IL-17 cytokines in the CD4⁺ T-cells of the BM+T group compared to the BMO on day 21. (n=7 mice/group) (BMO: bone marrow only recipients; BM+T: recipients of bone marrow and T cells)

For the kinetics of bone marrow-derived cell recruitment into the ocular tissues, particularly the cornea, , the BM+T group exhibited fluorescent signal in both the ocular surface and the lid margin as early as 2 weeks post-transplant, and this progressively increased until the endpoint at week 8 compared to the BMO mice that did not receive T cells during the transplant (Figure 4).

Figure 4. Progressive infiltration of fluorescently-labeled BM-derived cells as detected by in-vivo microscopy.

C3H.SW^H2B recipients were transplanted with EGFP labeled BM cells and in vivo images were obtained at different time-points. Infiltration in the BM+T group starts as early as 2 weeks post-transplantation in the lids and continues to increase weekly subsequently involving the ocular surface. Such infiltration was marginally progressive in the BM+T group compared to the BMO (no T-cells) group. (BMO: bone marrow only recipients; BM+T: recipients of bone marrow and T cells)

3.1.3. Significant MGD development in the GVHD group:

To investigate the role of MGD in the progression of oGVHD, we evaluated Meibomian gland changes in the GVHD mouse model. To do so, subjective, and objective measures were obtained to quantify the disease progression. BMO (n=15) and BM+T (n=15) mice were assessed using the predetermined Meibomian gland plugging score across the 21-day post-transplant period. At the end-point on day 21, the BM+T group had a significantly higher MGD plugging score per mouse compared to the BMO group (1.5 vs 0, P<.0001, Figure 5A).

Figure 5. Progression of MGD in the oGVHD model.

(A) Progression of Meibomian gland plugging scores in the BM+T vs BMO) during first 3 weeks’ post-transplant period. (B) Imaging of the dissected lids showcased the presence of Meibomian gland disruption in the BM+T group (red arrows) compared to BMO. (C) Quantification of the degree of atrophy using lid photographs (n=30 eyes/group): 93.3% of eyes of the BM+T group exhibited some form of atrophy, specifically Score 1: 13.3%; score 2: 33.3%; score 3: 26.7%; score 4: 20%. (n=15 mice/group) (BMO: bone marrow only recipients; BM+T: recipients of bone marrow and T cells)

Moreover, upon imaging the dissected lids at day 21 post-transplant, BM+T mice had noticeable MG atrophy compared to BMO mice (Figure 5B). Upon analyzing degree of atrophy, all the BMO eyes remained at score 0 translating to no noticeable atrophy. In the BM+T mice, 28/30 (93%) of the eyes had some degree of noticeable atrophy at the score 1 category or worse, with included 14/30 (47%) of the eyes falling in the more severe categories (scores 3 and 4) (Figure 5C).

3.1.4. GVHD induced ocular surface disease manifestations were more robust in mice with more severe MGD:

Fluorescein staining of the ocular surface was performed in both groups of recipient mice (n=30 eyes/group). Progressively increased staining was identified in BM+T mice compared to the BMO group animals (Figures 6A). Analysis on day 21 found that the BM+T group had a significantly higher fluorescein staining mean green fluorescence index compared to the BMO group (25 vs 4, P<.0001, Figure 6B).

Figure 6. Fluorescein staining scores increase over time post-transplant in the cornea of recipients of BM + T cells in the oGVHD model.

(A) Stained representative fluorescein photographs of BMO vs BMT eyes at day 21 showcasing epitheliopathy in the experimental group compared to the controls. Mean green fluorescence was calculated by selecting the cornea and excluding the fluorescein pooling on the lids. (B) Fluorescein staining scores are presented over time post-transplant as fluorescence staining index (see Methods). (C) On day 21 post-transplant, mice with a higher MGD atrophy score determined using lid photographs exhibited more staining compared to mice with milder MGD atrophy on lid photographs (n=15 mice/group). (Markers for CD45+ cells: CD4+ for CD4 T cells; CD8+ for CD8 T cells; CD11b+/Ly6G+/Ly6C+ for neutrophils; CD11b+/Ly6C+/Ly6C- for monocytes.) (BMO: bone marrow only recipients; BM+T: recipients of bone marrow and T cells)

Mice were then assessed based on the severity of their MGD atrophy score and we found that the mean green fluorescence index was significantly higher in the group with a severe MGD score compared to the mice with mild- moderate scores (26 vs 16, P=.023, Figure 6C).

3.2. PATIENT GVHD:

3.2.1. Patient demographics:

Similar to the analyses performed in the GVHD experimental mouse model, investigations into the oGVHD patient population at the Foster Center for Ocular Immunology at the Duke Eye Center were also performed to examine the parallel inflammatory processes that drive MGD and clinical ocular surface disease findings. This would enable us to determine if some of the observations identified in mice were also present in patients.

A total 57 patients with oGVHD presenting to the Foster Center for Ocular Immunology at the Duke Eye Center were included in this study. The demographics of this population (Table 1) included 32 males (56%) and 25 females (44%) with a male/female ratio of 1.28. The average age at presentation was 53 +/− 15.3 years with a range of 18–75. 47 (83%) of the population were Caucasian in addition to 6 (11%) African-Americans, 1 (2%) Asian, and 4 (5%) without a specified racial group. Age at the time of diagnosis of the primary hematologic disorder was 46.1 years ± 16.8. The time from primary diagnosis until the transplant was done was 2.4 years ± 3.8. Among the transplant patients reviewed and out of those with the reported characteristics, 28/40 (70%) were sex-matched, 19/53 (36%) were from a related donor, and 51/52 (98%) were major HLA matched. As of January 1^st, 2022, 7/57 (12%) of the population were deceased. The primary hematologic disease indications for transplant in these patients are summarized in Figure 7.

Table 1.

oGVHD patient demographics (n=57)

Gender (n)	Male= 32, Female= 25 M:F ratio= 1.28
Age at diagnosis   Average [(n +/− SD)]:   Range:	53 years ± 15.3 Range= 18–75
Race [n(%)]   Caucasian:   African-American:   Asian:   Other:	47 (82.4%) 6 (10.5%) 1 (1.7%) 4 (5.3%)
Age at Diagnosis of Hematologic Disorder (n=55)   Average [(n +/− SD)]:   Range:	46.1 years ± 16.8 0–72 years
Time from Diagnosis to Transplant (n=55)   Average [(n +/− SD)]:   Range:	2.4 years ± 3.8 0–18 years
Time from Transplant to First visit (n=55)   Average [(n +/− SD)]:   Range:	7.42 years ± 14.8 0–13 years
Donor Sex Match (n=40)   Matched:   Unmatched:	28 (70%) 12 (30%)
Donor Relation (n=53)   Related:   Unrelated:	19 (35.8%) 34 (64.1%)
HLA Major Match (n=52)   Matched:   Unmatched:	51 (98%) 1 (2%)
Live/Dead status on January 1st, 2022 n(%)	Alive= 50 (87.7%)

Figure 7.

Primary disease indication for transplant in the 57 oGVHD patients.

3.2.2. Presence of systemic GVHD tissue involvement in patients with oGVHD:

Of 56 patients with ocular GVHD, 54 (96%) had one or more forms of GVHD in other organs (Figure 8). The most common concomitant GVHD presence was skin involvement which was observed in 48/56 (86%) followed by gastrointestinal/Oral GVHD found in 34/56 (61%) of patients with oGVHD.

Figure 8. Organs reported being affected by GVHD in the patients with oGVHD that presented to the Foster Center for Ocular Immunology at the Duke Eye Center.

The highest co-existing GVHD with ocular disease was skin involvement in 48/57 (85.7%) followed by gastrointestinal (GI) or oral involvement in 34/56 (60.7%) of patients with oGVHD.

3.2.3. Substantial ocular disease was noted in oGVHD with a high MMP-9 positivity:

Ocular and dry eye disease workup in the oGVHD patients was carried out at the time of presentation (Table 2). On average, oGVHD patients demonstrated an ocular surface disease index (OSDI) score of 45.7 +/− 24.4, tear production on Schirmer’s strip of 4.0 +/− 4.9, corneal surface staining using NaFl index at 4.1 +/− 5.6, and a conjunctival staining using Oxford grading scheme at 2.86 +/− 2.0.

Table 2.

Ocular workup in oGVHD patients.

Test (N)	Average n +/− SD
OSDI (N=45)	45.7 +/− 24.4
Schirmer’s Test (N=31)	4.0 +/− 4.9
Total Corneal Stain (N=42)	4.1 +/− 5.6
Total Conjunctival Stain (N=29)	2.86 +/− 2.0
MMP-9 (N=49)
Positive (+)	41 (83.6%)
Single Positive (+/−)	4 (8.1%)
Negative (−)	4 (8.1%)

Notably, a high matrix metalloproteinase-9 (MMP-9) positivity in one (8%) or both eyes (84%) was observed in the majority of patients with oGVHD (Table 2). MMP-9 is a well-documented mediator of inflammation across multiple organs and the ocular point-of-care test was emphasized in this sample to investigate the inflammatory process in the ocular surface ^41–43.

3.2.4. Tears from GVHD contained a high number of immune cells:

In the studies, we employed an emerging technique of spectral flow cytometry to enable investigations into the immune phenotype of the ocular surface in 7 patients with oGVHD. Patients with oGVHD demonstrated a substantial cell population consisting of mainly neutrophils (PMNs) along with other subsets of mononuclear cells and CD4 cells. The cell populations of the oGVHD patients were distinct from those of non-GVHD dry eye patients (Figure 9). Compared to controls that had no identifiable immune cells in their tears, 4/7 (57%) of oGVHD patients had at least 100 CD45+ cells and 3/7 (43%) had neutrophils (PMNs) in their tear wash samples. On average, neutrophils made up 40 +/− 29 % of the CD45+ population in the oGVHD patients (Table 3).

Figure 9. TSNE plots of oGVHD vs non-GVHD dry eye disease patients.

(A) UMAP of normal blood 36c panel. Expression level indicated (Blue = lowest; Maroon = highest). Respective markers are shown above the plot. Fluorophores, clones, and gating strategy published in Yu et al, Med, 2021. (B) UMAP of normal blood and patient tears, as indicated. Respective immune cell type annotated by color in the figure legend. non-GVHD n=21; GVHD n=4. Results identified an abundance of neutrophils (PMNs) in the tear samples of oGVHD patients along with another population of mononuclear cells (MNPs) and a set of CD4+ T-cells.

Table 3.

Characteristics and tear fluid analysis of the 7 oGVHD patients in which tear collection was performed

Parameter		GVHD (n=7)
Age		57 ± 9.6
Race	Caucasian	5 (71.5%)
	African-American	2 (28.5%)
	Asian	0
Gender	Male	2 (28.5%)
	Female	5 (71.5%)
	F:M	2.5:1
MGD Severity	Score 0	0
	Score 1	0
	Score 2	0
	Score 3	0
	Score 4	7 (100%)
Tear fluid CD45 Count	Present > 100 cells	4 (80%)
	Average	1288 ± 1000
Tear fluid PMNs	Present > 50 cells	3 (43%)
	Average Count	799 ± 768
	Average Percentage	39.7% ± 28.8

3.2.5. Patients with oGVHD demonstrated substantial MGD and gland atrophy:

To assess for MGD presence in clinical GVHD, the morphology of the Meibomian glands was investigated in our GVHD patient population. The objective was to note the severity of MGD in this population and use this information to categorize and correlate ocular surface disease parameters in relation to MGD grading. To do so, review of the meibographs of these patients was carried out and scored for severity based on the Meiboscale ^37,38,44. Meiboscale was performed on 89 eyes of the original oGVHD sample. Of these, 86/89 (97%) showed some degree of Meibomian gland atrophy with 41 eyes (46%) in the moderate/severe category (i.e. scores 3 and 4) (Figure 10A and Figure 10B).

Figure 10. Imaging of eyes in patients with oGVHD illustrating the degree of Meibomian gland atrophy on Meibography.

The majority of patients (97%) in our sample had at least some form of Meibomian gland atrophy and almost half (46%) of the sample was in the scores 3 and 4 categories (A). This demonstrates the high prevalence of disrupted Meibomian gland morphology in oGVHD. (B) Highlights a representative photograph of a normal eye of a non-GVHD control compared to a Meibography representative of the Meibomian gland atrophy seen in oGVHD.

When the diseased eyes were grouped based on the Meiboscale score, eyes that had a higher degree of MGD atrophy also had a significantly higher proportion of MMP-9 positivity (X²= 4.05, P=.044) and a significantly lower tear breakup time (TBUT) (2.3 vs 4.2, P=.0176) compared to samples that had a lower score on the Meiboscale (Figures 11A and 11B). Corneal staining did not differ between the MGD groups (Figure 11C). In addition, the more severe MGD group did not significantly differ in Schirmer’s tear production compared to the milder group (2.4 vs 4.9, P=.682) (Figure 11D).

Figure 11. Ocular workup within the oGVHD patients was performed in the context of MGD severity to test for severity-related correlations.

(A) Significantly increased MMP-9 positivity and (B) Significantly decreased tear breakup time (TBUT) were found in patients with more severe MGD compared to those with milder forms. Interestingly, both groups had similar disease levels of corneal staining (C).

4. DISCUSSION:

4.1. SYNOPSIS:

Since its inception as a curative therapeutic option for multiple forms of malignancies and hematologic disorders, aHSCT and its success is often complicated by the extent of GVHD in the recipients’ tissues. Unfortunately, this inflammatory attack arises from the same immune cells that were transplanted to protect these patients against invasive and malignant antigens ^1,2,13,14. Despite the unique immune environment of some parts of the eye, the ocular tissues are not spared and a cascade of inflammatory progressions can drive immune-mediated damage to the visual system in the form of oGVHD ^8,9,30,45. oGVHD is co-present in 60–90% of patients with other forms of GVHD with a correlation in severity, highlighting the systemic immune etiology of the ocular disease process ^33,46. This process entails a substantial inflammatory process involving virtually any of the tissues of the ocular adnexa, including the Meibomian glands, that results in a variety of downstream consequences most commonly manifesting in the form of a robust DED and keratopathy ^7,16–18,20. Because visual function and ocular symptoms substantially impact a patient’s quality of life, oGVHD can frequently be a significant burden on a patient’s post-transplant well-being^12,23,47,48.

4.2. THE GAP IN KNOWLEDGE:

Much of our knowledge of oGVHD has been obtained from the use of animal models and investigations on patient populations ^{8,12,24,26,28,29,48–50}. There has been a growing emphasis in the past decade on identifying the immune pathways governing the development of oGVHD and inflammatory DED in general ^48,51. Despite the recent discoveries, our understanding and therapeutic arsenals remain limited, and the role of the progressive Meibomian gland damage in the advancement of ocular manifestations in oGVHD is poorly understood. Currently, there is a limited therapeutic arsenal for the direct or indirect treatment of MGD in oGVHD ^{8,15,27,46,52}. Nonetheless, a variety of topical immune-modulators, such as cyclosporine or tacrolimus, have been proven to be safe and effective in the local treatment of dry eye and lid disease in oGVHD ^53–55. Despite this, a better understanding of the different immune processes involved and uncovering the role of MGD in disease progression will undoubtedly help reveal the inflammatory basis of oGVHD. Such conclusions would, subsequently, assist in establishing new frontlines for therapeutic targets which may be generalizable to other forms of inflammatory DED.

4.3. BRIDGING THE GAP:

In the present work, we rely on two different perspectives to uncover the progression of MGD post-aHSCT and its role in the development of subsequent ocular surface manifestations. First, we rely on our established mismatch aHSCT mouse model in which the kinetics of induction of GVHD can be controlled allowing for disease development and follow-up of the associated damage at the ocular adnexa, including the Meibomian glands, along with the subsequent clinical and immune manifestations ^28–31. Second, this study investigated the ocular and systemic findings of a sizable sample of patients with oGVHD that present to our tertiary referral center at the Foster Center for Ocular Immunology at the Duke Eye Center.

These investigations enabled us to establish clinical disease findings and, using the emerging spectral flow cytometry, uncover immune patterns. Together, utilizing the clinical and immune disease findings from both mice and humans allows for the identification of common patterns to establish novel pathways of Meibomian gland involvement starting with the upstream systemic processes that are involved in MGD development in addition to the downstream manifestations of MGD in which the ocular surface suffers.

4.4. GVHD DRIVES A SYSTEMIC INFLAMMATORY DISEASE:

In our investigations, we first aimed to establish the systemic background in the subjects and how it related to the ocular manifestations of interest.

In the pre-clinical mouse studies, the success of the GVHD model is determined by the development of significantly worse systemic GVHD scores in the experimental BM+T group compared to the group not receiving T cells (BMO). Because this is not a highly lethal aHSCT model, adequate survival allows for analysis of mice during the course of progressive disease. In our experiments, the BM+T group had a significantly worse systemic GVHD score at the end-point with an 85% survival of the diseased mice establishing adequate circumstances for in-depth analysis of the ocular manifestations (Figure 1).

In humans, our sample size consisted of 57 aHSCT patients with an average age of 53 +/− 15.3 years and, in this cohort, the male-to-female ratio was 1.28. The epidemiology of oGVHD is not well described, and there are conflicting reports about the role of differences in sex in the development of this condition. However, a higher number of males in our population is consistent with other studies reported in the literature and can shed some light on the sex demographics of this condition ⁵⁶. In addition, our sample included a diversity of racial groups with the majority being Caucasian (82%) (Table 1). On average, patients were 46.1 ± 16.8 years old at the time of diagnosis with a range of 18–75 highlighting the wide range of age groups requiring primary malignancies and aHSCT. It took an average of 2.4 years ± 3.8 from diagnosis to transplant in this cohort. In addition, it took, on average, 7.4 years from transplant to referral to our tertiary care center. The majority of patients were sex-matched 28/40 (70%) and from an unrelated donor 34/53 (64%). Full major HLA matching was reported in 51/52 (98%) patients, indicating the role of minor antigens in the progression of the GVHD process which resembles that of our mouse model. In our cohort, 7/57 (12%) of the population were deceased highlighting the lethality of this condition despite the novel developments in management of hematologic disease and complications of aHSCT (Table 1). Our cohort also consisted of patients with different primary hematologic backgrounds (Figure 7). Although further studies will be required, the consistency with other reports strengthens the notion of the potential generalizability of our findings to other oGVHD patient populations.

4.5. THE SYSTEMIC DISEASE PROCESS OF GVHD MANIFESTS ACROSS DIFFERENT TISSUES INCLUDING THE EYE:

In the mouse model, the BM+T experimental mice developed significantly worse eyelid edema and decreased tear production compared to the BMO controls, with eyelid edema severity significantly correlated with that of systemic disease (Figure 1). The correlation of systemic GVHD, particularly that of the skin, with oGVHD has been reported in the literature and is further supported by our prior investigations ³³. Furthermore, we previously demonstrated in experimental GVHD models that the amelioration of systemic GVHD via treatment with prophylactic medications such as cyclophosphamide decreases the incidence of oGVHD ⁵⁷. The finding of significantly worsened systemic and ocular disease in mice that received T cells with their transplant, compared to the BMO group that didn’t, emphasizes the role of T cells in the disease process of GVHD which goes in line with the literature surrounding this condition ^8,13,14,29.

Based on this mouse data, it is notable that a similar trend was observed in human recipients of aHSCT whereby the majority (96%) of oGVHD patients had a co-manifestation of at least one other form GVHD in another organ with the skin being the most commonly involved (86%) (Figure 8). In addition, the human oGVHD sample demonstrated ocular disease findings across different parameters which included abnormal ocular symptomatology (OSDI 45.7 +/− 24.4), tear production (Schirmer’s 4.0 +/− 4.9), corneal staining (NaFl 4.1 +/− 5.6), and conjunctival staining (NaFl 2.86 +/− 2.0) (Table 2).

Both mouse and human findings suggest systemic GVHD to be the main driver of the disease process reflected by the development of pathology in multiple sites in which the eye and ocular tissues are not spared resulting in substantial ocular signs and symptoms. This is consistent with literature findings which showcase the substantial burden that is carried by patients that suffer from this condition which includes poor quality of life and ocular surface complications ^19,23,24.

4.6. oGVHD IS DRIVEN BY A T CELL INFLAMMATORY CYTOKINE PROCESS THAT INITIATES EFFECTOR IMMUNE CELL INFILTRATION OF THE OCULAR TISSUES:

The goal of these experiments was to assess how the immune milieu in the ocular adnexa alters the progression of oGVHD. In the mouse studies, investigation of the immune cells of the ocular tissues demonstrated increased CD45 immune cell infiltration of multiple hematopoietic cell populations, including CD4 T cells, CD8 T cells, PMNs, and macrophages, in the conjunctiva of the BM+T transplanted mice compared to animals in the BMO group (Figure 2A, 2B, and 2C). Moreover, the conjunctiva of the BM+T group had a lower CD4/CD8 ratio compared to that of the BMO paralleling the systemic immune phenotype observed in hemato-lymphoid tissues (Figure 2D) ³⁰. The finding of the role of Th17 cells and their release of IL-17 in the pathogenesis of DED and oGVHD goes in line with what has been reported and highlights the multifactorial inflammatory process underlying the manifestations (Figure 3)^58–60. Finally, tracking the fluorescently-labeled BM-derived cells demonstrated the infiltration of these cells into the ocular surface as well as the adnexal tissue (i.e. lid margin) as soon as 2-weeks post-transplant (Figure 4). Through prior experiments from our group, we have shown that these cells are bone marrow-derived correlated with CD11b which is mainly expressed in immune effector cells such as monocytes, macrophages, neutrophils, and natural killer cells. Therefore, we hypothesize that CD45+ bone marrow-derived macrophages and other effector cells are recruited early into the lid margin ³⁰. Considering their substantial presence in the BM+T group compared to the T cell-deprived BMO group, we propose this is orchestrated by donor T cells and contributes to the aggressive MGD development in oGVHD.

The immune changes in the ocular compartment of mice undergoing oGVHD were also analogously observed in humans. MMP-9 is a well-studied collagenase and has been known to be a mediator of inflammation across multiple organs ^41,42. In the eye, a positive finding on ocular point-of-care testing has been found to be associated with decreased tear production (Table 2) ⁴³. The finding of increased matrix metalloproteinase in the ocular surface of patients with oGVHD is also consistent with findings in the literature and validates the immune process arising in this condition ⁵⁰. Immune cell infiltration of the ocular surface was also demonstrated in oGVHD patients through high-parameter flow cytometry of their tear samples (Table 3 and Figure 9). In addition, neutrophil elastase was found to be elevated in patients with oGVHD and this elevation was correlated with other inflammatory markers further elucidating the role of neutrophils in this condition ⁵⁰. The finding of immune cell infiltration, particularly neutrophils, in the tear samples of oGVHD subjects has also been reported in the GVHD mice by our group and was noted to drive Meibomian gland obstruction and disease ⁶¹. This leads us to posit the presence of an axis involving Meibomian glands and the ocular surface in oGVHD with neutrophil infiltration playing the role of a key mediator.

These findings in both experimental and clinical GVHD demonstrate the systemic inflammatory process led by donor T cells that initiates a cytokine-driven cascade of events resulting in effector immune cell infiltration of the ocular tissues ultimately resulting in substantial disease including the ocular surface. The involvement, kinetics, and role of donor T cells in this process have been observed and previously noted in several studies ^1,6,8,61. This inflammatory cascade is noted to affect the ocular-associated tissues, with the lacrimal gland damage being the classically noted driver of ocular surface disease and is thought to be the result of interstitial fibrosis and inflammation mediated by CD34+ fibroblasts ^62,63. We posit that the lacrimal gland damage is likely related to the epithelial (squamous) nature of cells in this gland which is shared with other target tissues including the GI tract and skin. However, the extensive damage suggests that concomitant damage to other ocular adnexal tissues is also involved. f to the lacrimal glands, cornea, and conjunctiva has been described in the oGVHD process, literature on the role and specifics of the Meibomian glands (MG) remains scarce. This tissue is important because damage to its function leads to evaporative dry eye, which is already a common finding in dry eye patients, and eventual corneal damage. Our work aims to highlight and investigate the important role played by the Meibomian glands as a target of the systemic inflammatory process that links, and possibly instigates, the resultant damage seen in the ocular surface.

4.7. MGD IS PREVALENT IN OGVHD AND SEVERITY IS ASSOCIATED WITH WORSENING OCULAR SURFACE PARAMETERS:

The primary aim of this work is to elucidate the lesser-investigated role of the Meibomian glands in the development of ocular surface disease in oGVHD. A novel aspect of this work is the investigation of the Meibomian glands within the development of oGVHD in both mice and humans. Potentially, another benefit of these findings could be their consideration in the context of other forms inflammatory DED. In the preclinical models, BM+T mice had progressive worse MGD plugging and epitheliopathy compared to BMO (Figure 6 and Figure 8). In addition, imaging of the Meibomian glands demonstrated at least some level of atrophy in >90% of the eyes examined with ~45% in the most severe categories (Figure 5). The significant association between epitheliopathy and increased MGD severity highlights the importance of MGD in the ocular surface damage in oGVHD (Figure 6C).

A similar investigation on MGD and oGVHD was carried out in the human population. On meibography of these patients, 97% of the eyes had at least some degree of atrophy on the Meiboscale with 46% being in the more severe categories (scores 3 and 4) (Figure 10A and 10B). The prevalence of MGD in oGVHD is noted in the literature ⁶⁴, but the role played by this condition in the inflammatory process that results in ocular surface dysfunction is poorly reported. When assessed using Meiboscale severity, we found the more severe group had a higher percentage of MMP-9 positivity (92% vs 80%, p=.0359) and lower TBUT (2.3 vs 4.2, p=.0176) compared to the eyes exhibiting milder scores (Figure 11).

4.8. FINDINGS SUPPORT THE HYPOTHESIS THAT THE T CELL-MEDIATED SYSTEMIC INFLAMMATORY PROCESS IN GVHD DRIVES OCULAR ADNEXAL INFLAMMATION CAUSING MGD THAT PERPETUATES A VICIOUS CYCLE OF OCULAR SURFACE IMMUNE DAMAGE INSTIGATING SUBSTANTIAL DRY EYE DISEASE:

The results between the mouse and human populations highlight the prevalence of MGD in patients with oGVHD and how the severity of Meibomian gland involvement associates with ocular surface disease findings. Our findings support our original hypothesis and involving the relationship between the Meibomian glands and the ocular surface in coordinating and instigating the oGVHD process. The role of T cells as the main instigator of oGVHD goes in line with other forms of GVHD ^28,29,65. This role was further made apparent in the mouse model whereby mice receiving T cells (BM+T) achieved higher T cell cytokine expression and developed substantial disease compared to those that received TCD-BM (BMO). Moreover, the T cell-dependent inflammatory infiltration into the ocular adnexa was highlighted in the increased inflammatory cell infiltration of the conjunctival of BM+T mice. This, in turn, led to MGD which was well noted in both the experimental mouse model and the diseased human population. MGD was then shown to be associated with different ocular surface disease parameters highlighting the important role this route plays in oGVHD. Immune cell infiltration of the ocular surface as a result of this process was further supported in both mice and humans and showcases the inflammatory basis of the ocular surface seen in this condition.

The Meibomian glands are essential orchestrators of a healthy tear film, and MGD seen in oGVHD disrupts this process leading to ocular surface disease. Although experimental studies need to be performed to precisely elucidate a causality between MGD and the ocular surface, it is likely that bidirectional interactions may lead to a vicious cycle of progressive disease that ultimately results in total Meibomian gland atrophy and severe keratopathy. On one hand, inflammation of the ocular adnexa results in immune cell infiltration resulting in Meibomian gland damage causing evaporative DED that disrupts the tear film leading to damage to the ocular surface. In return, the damage to the ocular surface as a result of MGD as well as other causes (i.e. lacrimal gland damage) activates further inflammation and immune cell infiltration in the eyelids and adnexa that obstructs and instigates added Meibomian gland damage. This, in turn, results in an exaggerated form of multi-factorial DED seen in these patients leading to a substantial encumbrance on a patient’s vision and quality of life.

4.9. STUDY UTILITY AND LIMITATIONS:

Such an investigation comes with a variety of implications, which have led us to consider several hypothesis and speculations. As previously highlighted, the specific role of the Meibomian glands in the disease process is poorly described, and understanding a different route involving bidirectional signaling and damage between MGD and keratopathy in the oGVHD process can allow for and identifying and understanding a different route involving bidirectional signaling and damage between MGD and keratopathy (Figure 12). More specifically here we envision that understanding how inflammatory MGD may drives oGVHD could help towards identifying novel reagents and targets regulating this pathway. As the current therapeutic arsenal for MGD is limited, novel agents are genuinely needed to improve quality of life in these patients. While oGVHD is one of many forms of inflammatory DED, this disease process is one in which we are able to establish the precise initiating time point of onset. While the work here is specific to GVHD, it is possible that the inflammatory processes involved may be shared with different forms of ocular surface immune diseases including inflammatory DED. Future studies will likely provide insight into any common pathways and mechanisms.

Figure 12. (Created with BioRender.com). The vicious cycle of ocular inflammation in ocular GVHD.

Systemic inflammation from the donor-derived immune cells makes its way into the eye and drives an inflammatory response that damages the ocular surface and virtually any orbital adnexal tissue including the Meibomian glands, lacrimal glands, and ocular surface. The resultant downstream damage instigates a vicious cycle that drives bidirectional continuous damage to the ocular adnexa and ocular surface. On one hand, Meibomian gland damage disrupts the tear film leading to dry eye disease and its subsequent complications (i.e. keratopathy, infections, etc.). On the other hand, ocular surface disease, resulting from direct inflammatory damage as well as secondary damage (i.e. lacrimal gland atrophy, MGD, etc.), drives further immune cell infiltration into the ocular surface which, in turn, drives Meibomian gland plugging and subsequent atrophy.

It is important to note that this study was accompanied by several limitations. Firstly, despite many similarities between the different GVHD mouse models, it is well appreciated that different tissues can be more or less involved dependent on the donor/recipient strains, and therefore the generalizability of our findings to other models will need to be determined. Regardless, similar investigations in the different models may reveal novel observations that could ultimately further our understanding of this disease pathway. Although Meibomian gland atrophy is one of the indications of MGD severity, there are other determining factors involved that limits the categorization of severity in the present analysis. It should also be noted that the single-center non-controlled retrospective design of the patient investigation will need to be increased to provide additional corroboration and rigor. Because these findings stem from a single-center population, the conclusions in this manuscript may not be transmittable to other centers in which patient conditions vary. In addition, the retrospective design introduces a variety of biases that limit the collection and interpretation of the data. Accordingly, multi-center investigations are required before broad generalizations can be considered. In total, we posit that the comparison of findings and the identification of similar disease patterns between the mouse models employed and patient populations examined in this study, will help in providing impetus and insight for future investigations towards understanding the key role of inflammation in MGD.

5. CONCLUSION:

oGVHD is a debilitating subtype of inflammatory DED that burdens the patients suffering from it after aHSCT and substantially reduces the quality of life in these patients who, in many cases, had just overcome a life-threatening hematologic condition. Although recent advances are promising towards unmasking multiple processes involved in the progression of this condition, the involvement and role played by the Meibomian glands in this condition remains to be precisely characterized. Our investigations in animal models and human patients highlighted the systemic inflammatory process driving an early T cell-driven bidirectional vicious cycle of disease progression that ultimately manifests in a robust form of inflammatory keratopathy. Thus, future studies will be designed to test the role of MGD progression in ocular surface disease severity in oGVHD. We are optimistic that such research will result in identifying new preventive and therapeutic targets to improve the quality of life in patients with GVHD and inflammatory eye disorders.