How does ocular graft-versus-host disease fit under the dry eye umbrella? A review

Abstract

Graft-versus-host disease (GVHD) is a systemic disease that can affect multiple organs as a consequence of allogeneic hematopoietic stem cell transplant. One organ system that is often affected in GVHD is the eyes. Ocular GVHD (oGVHD) may involve various structures within the eye including the lacrimal glands, eyelids, conjunctiva, cornea, and nasolacrimal ducts, and is a source of morbidity in patients with GVHD. Common presenting features of GVHD overlap with dry eye disease (DED), including decreased tear production, epithelial disruption, and Meibomian gland dysfunction (MGD). In this review, we aim to compare oGVHD and DED to better understand similarities and differences between the conditions, with a focus on pathophysiology, risk factors, clinical features, and treatments.

1. INTRODUCTION

Graft-versus-host disease (GVHD) occurs due to complications of allogeneic hematopoietic stem cell transplant (HSCT). GVHD is the most common cause of non-relapse morbidity in stem cell transplant recipients. [1] Despite this, the pathophysiology of GVHD is incompletely understood. One target organ that is commonly affected in GVHD is the eye, with features that overlap with dry eye disease (DED). While ocular GVHD (oGVHD) and DED have distinct etiologies, some clinical and pathophysiological features are shared across the conditions. oGVHD arises as a complication of HSCT and oGVHD related-DED can be caused by multiple factors, including intrinsic (e.g., older age, auto-immune disease) and extrinsic (e.g., adverse environmental conditions) influences. Despite differing etiologies, symptoms (e.g., dryness, burning, visual disturbances), signs (e.g., decreased tear production, epithelial disruption), and pathophysiological mechanisms (e.g., inflammation) are shared between the two conditions. This paper examines our current understanding of oGVHD and compares and contrasts aspects of the disease to DED, pointing out similarities and differences. An improved understanding of disparate and shared features between oGVHD and DED can improve our understanding of mechanisms and management, with the goal of reducing patient morbidity.

2. CLASSIFICATION OF GVHD

Historically, GVHD was categorized into two phases based on time of presentation following HSCT: the acute phase (within 100 days) and the chronic phase (greater than 100 days). [2] However, based on the National Institutes of Health (NIH) 2014 criteria, the classification system changed to reflect symptoms and clinical findings and not only focus on time to presentation. [3]

Acute GVHD (aGVHD) is defined as “the appearance of an allogeneic inflammatory response in exclusively three organs: the skin (e.g., inflammatory maculopapular erythematous skin rash), the liver (e.g., hyperbilirubinemia due to cholestatic jaundice), and/or the gastrointestinal (GI) tract (e.g., upper and/or lower GI tract manifestations: anorexia with weight loss, nausea, vomiting, diarrhea, severe pain, GI bleeding and/or ileus)”. [4] These findings must occur in the absence of chronic GVHD (cGVHD) diagnostic and distinctive features (Table 1). [5] aGVHD grade is defined by the severity of skin (e.g., body surface area percent affected and generalized erythroderma with bullous formation), liver (e.g., serum bilirubin concentration), and GI tract (e.g., nausea, diarrhea, and severe abdominal pain) findings. [6–8] aGVHD is further divided into sub-types based on the time after HSCT at which these symptoms present. Classic aGVHD occurs within 100 days of HSCT and persistent/recurrent/ late-onset aGVHD occurs when an individual displays clinical features of aGVHD beyond 100 days. [5]

Table 1.

National Institutes of Health (NIH) 2014 criteria: Diagnostic and distinctive features of cGVHD. [5]

Organ/system	Diagnostic features (Found exclusively in cGVHD)	Distinctive features (Frequently found in cGVHD, but not enough to establish a diagnosis)
Skin and adnexa	- Poikiloderma - Lichen-planus-like features - Lichen-sclerosis-like features - Sclerotic features - Morphea-like features	- Skin depigmentation - New onset of scarring or nonscarring scalp alopecia - Nails: dystrophy, longitudinal ridging, splitting, or brittle features, onycholysis, pterygium unguis, nail loss
Mouth/oral mucosa	- Lichen-type features - Hyperkeratotic plaques - Restriction of mouth opening from sclerosis	- None
Genitalia	- Lichen planus like features - Vaginal scarring or stenosis	- Erosions - Fissures - Ulcers
Gastrointestinal tract	- Esophageal webs - Strictures or stenosis in the upper to mid third of esophagus	- None
Lungs	- Bronchiolitis obliterans diagnosed with lung biopsy	- Bronchiolitis obliterans diagnosed with pulmonary function tests (PFTs) and radiology
Musculoskeletal	- Fasciitis - Joint stiffness or contractures secondary to sclerosis	- Myositis or polymyositis
Eyes	- None	- Symptoms: New onset dry, gritty, or painful eyes - Signs: Cicatricial conjunctivitis, confluent areas of punctate epithelial keratopathy

To make a diagnosis of cGVHD there must be the presence of one diagnostic feature. If no diagnostic features are present, there must be at least 1 distinctive feature along with support by histologic, radiologic, or laboratory evidence of GVHD from any site.

The diagnosis of cGVHD requires one diagnostic manifestation or at least one distinctive manifestation confirmed by biopsy/ testing in that organ or other organs (Table 1). Additionally, other possible diagnoses explaining the findings must be excluded. Unlike aGVHD, cGVHD does not have a set time limit for diagnosis but usually occurs after the first 100 days. The severity of cGVHD is determined by the number of organs involved and the severity within each organ. Like aGVHD, cGVHD is divided into subtypes including classic cGVHD (no features of aGVHD present) and overlap syndrome (features of aGVHD and cGVHD co-occur) in which active aGVHD and features of cGVHD are present concomitantly. [5]

3. DIAGNOSTIC CRITERIA FOR OCULAR GVHD (OGVHD) AND DED

Various organizations have implemented diagnostic criteria for oGVHD over the years; however, the two more commonly used grading scales were created by the International Chronic Ocular GVHD Consensus Group (ICCGVHD) and the NIH. The ICCGVHD diagnostic criteria are based on the presence and severity of ocular signs, symptoms, and systemic GVHD involvement (Table 2). [9]

Table 2.

International Chronic Ocular GVHD Consensus Group (ICCGVHD) diagnostic criteria for ocular graft-versus-host disease (oGVHD). [9]

Ocular symptoms		Points (0-11)
OSDI (points)	<13	0
	13-22	1
	23-32	2
	>33	3
Ocular signs
Schirmer’s (mm/5min)	>15	0
	11-15	1
	6-10	2
	<5	3
CFS	No staining	0
	Trace staining	1
	Mild-Moderate staining	2
	Severe staining	3
Conjunctival injection	No injection	0
	Mild-Moderate injection	1
	Severe injection	2

OSDI=Ocular Surface Disease Index; CFS=Corneal Fluorescein Staining

Diagnosis: 6-7 points without systemic GVHD or 4-5 points with systemic GVHD = Probable oGVHD; ≥8 points without systemic GVHD or ≥6 points with systemic GVHD = Definite GVHD

The NIH 2014 defines oGVHD as new ocular sicca (starting after HSCT) (Schirmer’s test ≤5 mm/5 minutes or 6-10 mm/5 minutes with new onset of keratoconjunctivitis sicca (KCS) (post-HSCT) by slit lamp not due to other causes). [3] oGVHD severity is determined by symptom intensity, artificial tear use, and impact on activities of daily living (ADL (Table 3). [3, 10] Both criteria consider ocular symptoms and signs, while the NIH criteria also consider eye drop use and impact on ADLs.

Table 3.

oGVHD severity classification according to the National Institutes of Health (NIH) 2014 consensus

Severity	ADL	ATs	Visual Impairment
Mild	Not affected	>3/day	None
Moderate	Partially affected	>3/day or punctal plugs	None
Severe	Affected or unable to work		Yes

ADL= Activities of daily living; ATs = Artificial tears

In addition to the metrics noted above, other aspects of the DED assessment include grading of Meibomian gland (MG) and eyelid abnormalities (e.g., laxity, anterior blepharitis, telangiectasias, plugging, atrophy, meibum quality), tear instability (e.g., tear break-up time (TBUT)), ocular surface inflammation (e.g., metalloproteinase (MMP-9)), and anatomic features (e.g., conjunctivochalasis, pterygium, Salzman nodular degeneration). [11] In addition, beyond the OSDI, various questionnaires are used to assess aspects of DED symptoms, including the 5 Item Dry Eye Questionnaire (DEQ-5) [12], and the Standard Patient Evaluation of Eye Dryness (SPEED). [13] Other questionnaires focus on vision-related quality (e.g., National Eye Institute Visual Function Questionnaire-25) [14] and pain (e.g., Ocular Pain Assessment Survey (OPAS) [15] and National Pain Symptom Inventory-Eye (NPSI-Eye)). [16] Various abnormalities are noted with respect to these other metrics both in oGVHD and DED outside the purview of GVHD. Similar to oGVHD, there is not one accepted definition for DED or DED sub-types such as aqueous tear deficiency (ATD), evaporative dry eye (EDE), or Sjögren Syndrome (SS) associated dry eye, pointing to the need for consensus and improved definitions.

4. PATHOPHYSIOLOGY OF GVHD

4.1. Pathophysiology of Acute GVHD

An estimated 30-50% of patients who undergo HSCT develop aGVHD. [17] The most commonly affected organs include the skin, liver, and gastrointestinal (GI) tract. [18] aGVHD is broken up into three phases: 1) conditioning-mediated tissue damage 2) donor T cell activation and 3) target organ destruction.

4.1.1. Conditioning-mediated Tissue Damage

When HSCT is conducted in individuals with malignant conditions, conditioning regimens are used to ensure adequate immunosuppression to prevent graft rejection and reduce tumor burden (reduce tumor size/activity). [19] However, this conditioning regimen causes damage to epithelial cells, disrupts the diverse gut microbiome, and is linked to the development of aGVHD. [20] In one Spanish study, individuals who received a more intensive conditioning regimen (n=88) had a higher incidence of aGVHD than those who received a less intense conditioning regimen (n=150) (67% vs 44%; p<0.001). [21] Overall, higher-intensity conditioning is associated with more severe aGVHD. Damaged tissues from conditioning release antigens such as pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs) that are engulfed by donor antigen-presenting cells (APCs) and presented via human leukocyte antigens (HLAs) encoded by major histocompatibility complexes (MHCs) to donor T cells (after transplantation). [22] These HLA proteins have a high degree of genetic variation, and the level of HLA disparity between HSCT donor and recipient is linked to the likelihood of developing aGVHD. [23] Therefore, an important aspect of HSCT is HLA matching the recipient and donor.

4.1.2. Donor T Cell Activation

As T cell receptors bind to the antigen-MHC complex, co-stimulation occurs, which activates donor T cells. [24] T cells then differentiate into T helper (Th) 1 cells and Th17 cells which produce cluster of differentiation (CD4+) T cells, CD8+ cytotoxic T cells, and Natural Killer (NK) cells which can all produce inflammatory cytokines (tumor necrosis factor-alpha (TNF-α) and interleukin-1 (IL-1)) and cause tissue damage. [22] [25] In the gut, damaged epithelium and mucosa compromise the gut barrier, allowing lipopolysaccharide (LPS), an endotoxin, to leak through the mucosa into the systemic circulation. LPS binds to toll-like receptor-4 (TLR-4) on innate immune cells (such as macrophages and dendritic cells) leading to the secretion of inflammatory cytokines (IL-1β, TNF-α, and IL-6). [26] These cells can establish residence in various organ tissues.

4.1.3. Target Organ Destruction

During the final phase of aGVHD, T cells and cytokines that migrated into organs cause targeted destruction, primarily in the skin, GI tract, and liver. CD8+ T cells and NK cells activate the perforin/granzyme pathway leading to cell lysis and death [22]; TNF-α induces apoptosis by acting as a death ligand, and higher serum concentrations have been associated with more severe aGVHD. [27] Overall, CD8+ T cells, NK cells, and TNF- ɑ are major players in cell death in target organs.

4.2. Pathophysiology of Chronic GVHD (cGVHD)

cGVHD is the most common cause of non-relapse mortality in individuals who received HSCT. In a US study of 937 individuals who underwent HSCT, ~38% of non-relapse mortality was due to cGVHD. [28] The incidence of cGVHD is similar to that of aGVHD. One US study evaluated 784 adult and pediatric HSCT recipients and found that the 3-year cumulative incidence of cGVHD was 25%. [29] Four theories have been created to explain the pathophysiology of cGVHD: 1) thymic damage 2) CD4+ regulatory T cell (Treg) deficiency 3) production of autoantibodies by aberrant B cells and 4) development of profibrotic lesions. [30]

One contributor to cGVHD is impaired central tolerance which disrupts immune tolerance to self-antigens as a result of thymic epithelial damage during conditioning and donor T cell target organ destruction during aGVHD. [30, 31] An imbalance in Treg and CD4+ effector T cells are also to blame. CD4+ T cells promote inflammation and are thought to outcompete CD4+ Tregs which are responsible for immunological self-tolerance. [32–34] In addition to T cell dysregulation, B cell homeostasis is also disrupted. In cGVHD, high levels of B-cell activating factors have been noted, which are thought to promote the survival of alloreactive and autoreactive B cells. [35] Finally, fibrosis is another pathological component of cGVHD that is thought to be driven by macrophages in damaged tissues. Local release of TNF- ɑ stimulates fibroblast proliferation and release of platelet-derived growth factor-alpha (PDGF-ɑ), leading to fibrosis. [36, 37] To conclude, various pathways and cell types contribute to cGVHD phenotypes.

5. GVHD HISTOLOGY AND PATHOPHYSIOLOGY

5.1. The Lacrimal Gland: Acute oGVHD, CoGVHD, and DED

Lacrimal gland histology in acute ocular graft-versus-host disease (aoGVHD), chronic ocular graft-versus-host disease (coGVHD), and dry eye disease (DED) have been examined in human and animal models (Figure 1). A US study biopsied lacrimal glands of 18 individuals post-HSCT upon autopsy and found that 72% (n=13) of individuals with GVHD (11 aGVHD; 1 overlap a/cGVHD) had lacrimal gland stasis defined as an accumulation of cellular debris in the lumina of the acini and the ductules. Of the 6 individuals without GVHD, only 1 had “minimal” lacrimal gland stasis. [38] This postmortem study demonstrates early abnormalities in the lacrimal gland of individuals with GVHD. Animal studies have also examined this question. In mice, cellular infiltration was noted in the lacrimal gland of post-HSCT mice at 4 weeks, with predominantly CD3+ and CD8+ T cells, followed by CD4+ T cells, CD68+ macrophages, and a negligible number of CD20+ B cells. In controls (syngeneic mice), CD3+ and CD8+ T cells were noted to a lesser extent in the lacrimal gland. [39] These findings demonstrate inflammation in the lacrimal glands of oGVHD mice within 6 weeks following HSCT. While similar studies have not been conducted in humans, these findings support the underlying mechanism of targeted organ destruction by activated T cells, especially CD8+ T cells, in aGVHD. These data suggest that lacrimal gland abnormalities occur early in the disease, with subsequent impact on tear production in the chronic phase.

Figure 1: Similarities and differences in ocular graft-versus-host disease and dry eye disease pathophysiology and presentation.

oGVHD=ocular graft-versus-host disease; DED=dry eye disease; MMP-9=matrix metalloproteinase-9

In addition to inflammatory cell infiltration, fibrosis, and cell loss have been noted in the lacrimal glands of individuals with coGVHD. One Japanese study performed lacrimal gland biopsies in 9 individuals with coGVHD (reason for biopsy not given). Fibrosis was noted in the lacrimal gland interstitium along with CD34+ staining indicative of nonhematopoietic fibroblasts. [40] In addition, in 7 female recipients with male donors, 13% to 27% of CD34+ fibroblasts had a Y-chromosome by fluorescein in situ hybridization (FISH), indicating that the fibroblasts were from the male donors. [40] This suggests that donor fibroblasts play a role in the development of lacrimal gland fibrosis in HSCT recipients. In addition to fibrosis, patchy acini cell loss and mild lymphocytic infiltration of the lacrimal gland were noted. [40] The acini cell loss may be a result of CD8+T cell infiltration into the lacrimal acini during the aoGVHD phase leading to apoptosis and ongoing acinar cell loss in coGVHD. The same group performed immunohistochemistry on the lacrimal gland biopsies in 9 individuals with cGVHD and found that CD4+ T cells predominated in the periductal area whereas CD8+ T cells predominated within the ductules and acinar epithelium. Additionally, in one sample CD4+ and CD8+ T cells were found in close proximity to CD34+ fibroblasts that expressed HLA-DR and other co-stimulatory molecules such as CD40, CD80, and CD86 around a medium-sized duct. [41] This suggests that CD8+ T cells can activate co-residing fibroblasts, leading to fibrosis of the ductules in coGVHD. Moreover, T cells and fibroblast may co-stimulate each other to produce an ongoing inflammatory environment in coGVHD.

Similarities and differences are noted with respect to lacrimal gland histology in DED subtypes and oGVHD. A similarity between the two conditions is the presence of lacrimal gland inflammation; however, one difference between the two diseases is with respect to the sites where the inflammatory cells reside. The Japanese study noted above also biopsied the lacrimal glands of 5 individuals with SS-DED along with the 9 individuals with coGVHD and found higher levels of CD4+ T cell in SS vs coGVHD glands (3.8±0.5 vs 2.2±1.1, p= 0.01; scale 0-4; 0: no cells; 1: 1–9 cells; 2: 10–49 cells; 3: 50–99 cells; 4: 100+ cells/ 4mm²). However, CD8+ T cells were less common in the epithelium of the gland in SS vs coGVHD glands (1.0±1.0 vs 2.7±0.5, p=0.001). Furthermore, CD34+ fibroblasts were not seen in SS glands but were found in all coGVHD specimens. [41] IFN-γ distribution has also been found to differ in DED vs coGVHD. The same Japanese group biopsied the lacrimal glands of 3 individuals with SS and 3 with coGVHD and found IFN- γ predominantly on capillary endothelium in SS and predominantly on fibroblasts in coGVHD. [42] While IFN- γ may play a role in lymphocytic cell recruitment based on its presence on capillary endothelium in SS, its presence on fibroblasts in coGVHD suggests it can influence fibrotic processes in coGVHD.

5.2. The Conjunctiva: Acute oGVHD, CoGVHD, and DED

The conjunctiva can be affected in both oGVHD and DED (Figure 1). First, the conjunctiva can be involved in aoGVHD. One Japanese study described a case of a 26-year-old female with bilateral pseudomembrane formation and mucin deposition as part of her aGVHD presentation 38 days following HSCT using a male donor. On staining, a lymphocytic infiltrate (CD3+, CD56−, CD79a−) was noted upon biopsy of the conjunctival pseudomembrane reflective of T cell infiltration (CD3+), but the absence of Natural Killer cells (CD56−) and B lymphocytes (CD79a−). [43] These findings conform to accepted aGVHD pathophysiology in which activated donor T cells migrate into host tissues and cause damage. [22] Fluorescence in situ hybridization (FISH) analysis further supported this pathophysiology as a Y chromosome was identified in the infiltrated cells, indicating their donor source. [43] Findings in the conjunctiva in aoGVHD resemble those of skin aGVHD. For example, one study found that of 263 individuals who underwent allogeneic HSCT, 19 developed conjunctival aGVHD. Disease presentations varied, ranging from chemosis alone to pseudomembranous conjunctivitis. Histologic evaluation of conjunctival biopsies in two individuals with pseudomembranous conjunctivitis revealed a mononuclear inflammatory infiltrate of the basal epithelium and dyskeratosis, findings seen in skin aGVHD. [44] Conjunctival keratinization can also be seen as part of GVHD. One US study performed conjunctival biopsies in 40 individuals after HSCT and found that keratinization was noted in the conjunctiva of 25% (8 with aGVHD, 1 with mixed a/cGVHD, 1 with cGVHD). [38] Therefore, histologic findings in the conjunctiva resemble that of the skin in aGVHD; however, further investigation is needed regarding T-cell subtypes and the cytokines present.

Conjunctival involvement has also been examined in coGVHD and involves both inflammatory and fibrotic components. [45] One Brazilian study performed immunostaining in cryopreserved conjunctiva and found elevated inflammatory infiltrate cell counts in the epithelium and stroma of 20 individuals following allogeneic HSCT with cGVHD in comparison to 13 individuals pre-HSCT (some unique individuals and some with biopsies pre and post-HSCT). In the epithelium, cell counts of CD3+ T cells and CD14+ macrophages were higher post versus pre-HSCT (CD3+ T cells: 1.42±0.37 vs 1.23±0.51, p<0.05; CD14+ macrophages: 0.35±0.11 vs 0.09±0.09, p<0.02). Similar findings were noted in the stroma, with higher levels of CD3+, CD4+, and CD8+ T cells in post versus pre-HSCT specimens (CD3+: 8.84±2.13 vs 2.16±0.69, p<0.05; CD4+: 7.78±1.44 vs 1.06±0.44, p<0.01; CD8+: 5.67±1.56 vs 1.13±0.41, p<0.05; CD14+: 4.06±0.84 vs 0.30±0.18, p<0.001). In the same study, conjunctival specimens from allogeneic HSCT recipients with cGVHD (n=20) were compared to those of autologous HSCT recipients (n=8) without cGVHD. CD8+ T cells numbers were higher in the stroma of allogeneic versus autologous recipients (5.67±1.56 vs 1.95±0.84, p<0.001). In fact, the CD4+/CD8+ T cell ratio was higher in autologous vs allogeneic post-HSCT biopsies (epithelium: OR: 1.63, SD:1.01-2.62, p=0.01; stroma: OR: 3.09, SD: 2.39-3.99, p=0.00) [46], suggesting that while CD4+ T cells are seen at greater frequencies than CD8+ T cells, CD8+ T cells in the conjunctiva contribute to coGVHD pathogenesis. In addition to T cells, increased numbers of CD14+ macrophages were noted in the epithelium (0.57±0.17 vs 0.08±0.06, p=0.01) and stroma (5.13±1.26 vs 2.75±1.01, p<0.05) of allogeneic vs autologous recipients, further highlighting their role in coGVHD. [46] Therefore, while CD4+T cells are predominant in both groups, in the individuals who underwent allogeneic HSCT, there was a higher ratio of CD8+ T cells in the conjunctival stroma and more CD14+ macrophages in the conjunctival epithelium.

Beyond cellular composition, leukocyte function associated antigen-1 (LFA-1) expression has also been examined post versus pre-HSCT. LFA-1 is normally found on conjunctival vascular endothelium as well as lymphocytes and monocytes. The LFA-1 pathway governs the adhesion of activated CD8+ T cells to its targets. [47] In the study above, higher LFA-1 expression was also noted in the stroma of post-allogeneic HSCT cGVHD individuals (n=9) versus pre-HSCT (n=8) specimens (1.3 vs 0.4, p<0.05; scale: 0-4; 0= none; 4= marked immunoreactivity) which likely plays a role in CD8+ T cell recruitment into the conjunctival stroma in cGVHD. [46] CD8+ T cells can release inflammatory cytokines which damage local structures such as conjunctival goblet cells. One Japanese study found that goblet cell density was lower in individuals with coGVHD than post-HSCT individuals without oGVHD (396±381 vs 1030±433 cells/mm², p<0.05). [48] Reduced goblet cell density can contribute to tear instability often noted in GVHD. Fibrosis is also seen in cGVHD, driven by donor and recipient conjunctival myofibroblasts. One Swedish study examined conjunctival specimens of 25 females with male donor recipients and found the ratio of donor myofibroblasts to total myofibroblasts correlated with coGVHD severity. Additionally, the ratio of donor myofibroblasts (XY) to recipient myofibroblasts increased with time post-transplant. [49] Donor fibroblasts and subsequent fibrosis also contribute to signs of tear dysfunction and abnormal anatomy in coGVHD.

Both similarities and differences have been noted between conjunctival findings in DED subtypes and oGVHD. For example, T cells have been found in the conjunctiva in DED subtypes and coGVHD. A US study found that CD4+T cells predominated in conjunctival specimens of individuals with SS-DED and ATD-DED (staining and Schirmer’s ≤8 mm/5 min or staining and Schirmer’s ≤ 10 mm/5 min), with no differences in number of T cells between the two DED subtypes. B cells were seen to a lesser extent. [50] CD4+ T cells are also the predominant infiltrating lymphocytes in coGVHD [46], with minimal B cells noted in aoGVHD. [43] One unique aspect of coGVHD compared to DED is the presence of CD8+ T cells as drivers of disease. Similar to coGVHD, goblet cell numbers can be reduced in SS-DED and ATD-DED. In a US study, goblet cell density (cells/0.1mm²) was lower in SS-DED (n=12) and ATD-DED (n=16) (OSDI, Schirmer’s ≤ 5 mm/5 min and staining) compared to controls (both DED groups ~75 cells/0.1mm² vs controls ~115 cells/0.1mm²). [51] Although difficult to compare across studies, goblet cell density appears to be even lower in coGVHD [48, 51], suggesting that goblet cells are impacted to a greater degree in coGVHD than SS-DED and ATD-DED.

5.3. The Cornea: Acute oGVHD, CoGVHD, and DED

The cornea can also be affected in oGVHD (Figure 1). One US study performed corneal biopsies upon autopsy and noted keratinization (hematoxylin-eosin staining) in 6 of 40 individuals post-HSCT (13 with aGVHD, 1 overlap a/cGVHD). [44] Additionally, epithelial thinning has been found in 75% (n=30) of post-HSCT individuals in one study; noted in 22 with aGVHD, 1 with mixed a/cGVHD, as well as 7 individuals post-HSCT without GVHD. [52] Corneal changes in aGVHD have also been examined in animal models. In mice, corneal fluorescein staining increased by 2 weeks after transplantation in mice who received HSCT with donor T cells compared to controls HSCT without donor T cells (HSCT+T: 14.13±1.77; HSCT alone: 3.98±0.70, p<0.0001). At 30 days, the neurokinin-1 receptor (NK1R) was also found to be upregulated in the cornea. [53] NK1R is a receptor for substance P, a pro-inflammatory molecule found in the tears of humans with ocular surface inflammatory conditions, including GVHD. [54] In the same model, CD3+ T cells and CXCL10 (an inflammatory cytokine) were also expressed in the cornea. [53] Interestingly, the administration of an NK1R antagonist (rosaprepitant) decreased the percentage of area occupied by CD3+ T cells and CXCL10 on the cornea. In addition to CD4+ and CD8+, donor macrophages and inflammatory cytokines (e.g., IFN-γ, TNF-α, and IL-6) have all been found in the cornea in oGVHD. These data highlight the myriad inflammatory pathways involved in GVHD, and the potential for multiple targets to address the inflammatory response. [55]

Limited histological data are available on corneal specimens in coGVHD. One Japanese study examined corneal specimens obtained during corneal transplantation of 4 individuals with coGVHD who experienced a corneal perforation. Innate immune cells and matrix-degrading proteases were noted in all specimens. Specifically, CD68+ macrophages surrounded the borders of the perforation along with matrix metalloprotease-9 (MMP-9). Interestingly, CD4+ and CD8+ T cells were not found in the cornea. [52] While data are limited, the cellular makeup within the cornea may be different than that of the lacrimal gland and conjunctiva in cGVHD, where CD4+ and CD8+ T cells predominate. Angiogenesis has also been noted as part of cGVHD in mice models [56] and humans. In a Korean study of 94 individuals with coGVHD, 8.5% (n=8) had clinically apparent corneal neovascularization. [57, 58] Therefore immune cell deposition varies in the cornea in comparison to other ocular structures with GVHD and the role of angiogenesis warrants further investigation.

Corneal histology in DED subtypes shares features but also differs from coGVHD, with the majority of information gleaned from animal studies. In C57BL/6 mice, substance P mRNA (copies/10⁶ GAPDH copies) was higher in cases compared to controls 4 days following DED induction via desiccating stress (1.84 times higher, p=0.001). [59] MMP-9 has also been noted in DED and oGVHD. For example, a Korean study found that of 63 individuals with SS and symptoms of dryness, Schirmer’s was <5mm/min in 80% (n=32) of individuals with positive MMP-9 (> 40 ng/mL) and 57% (n=13) in individuals with negative MMP-9 (InflammaDry, Quidel). [60] Therefore, various cytokines, including MMP-9, are shared between DED and oGVHD.

5.4. Ocular surface microbiome

The ocular microbiome may be abnormal in oGVHD although it is not known if this is a contributor or consequence of disease. One Japanese study examined the ocular surface microbiome using culture in 32 individuals post-HSCT with cGVHD, 10 individuals post-HSCT without GVHD, and 5 controls. Several species were more commonly recovered in cases compared to both control groups including Staphylococcus epidermidis(75% vs 20%), alpha-hemolytic Streptococcus (29% vs 0%), and aerobic gram-positive cocci (13% vs 0%). Additionally, multiple species were detected in 46% of individuals with GVHD, 4% of post-HSCT non-oGVHD individuals, and 0% in controls. The number of species positively correlated with GVHD severity score (r=0.50, p=0.0002), corneal staining severity (r = 0.41, p=0.003), and fibrosis severity (r=0.31, p=0.03). [61] These findings suggest ocular surface microbiome alterations in oGVHD but their role in disease is not clear. This topic warrants further investigation.

DED subtypes can also present with alterations in the ocular surface microbiome. One Chinese study evaluated the ocular surface microbiome using 16S rRNA technology and noted differences between individuals with autoimmune-related (e.g., rheumatoid arthritis, systemic lupus erythematosus, Sjögren’s syndrome, systemic sclerosis, and Grave’s disease) (n=38) and isolated ATD (OSDI ≥13; Schirmer’s ≤5 mm/5 min) (n=49). In this study, increases in Actinobacteria, Firmicutes, and Bacteroides and decreases in Proteobacteria were noted in autoimmune versus isolated ATD. [62] Microbiome differences have also been noted when individuals with DED were compared to coGVHD. In one US study, the ocular surface microbiome of 8 individuals with DED (DEWS II criteria) had lower α-diversity and a higher relative abundance of Corynebacterium compared to 4 individuals with coGVHD. [63] The impact of the noted microbiome composition on disease presentation and progression needs further study.

5.5. Tear fluid

Tear cytokines abnormalities have been noted in coGVHD. In one South Korean study, tear cytokine levels were evaluated in individuals post-allogeneic HSCT with (n=30) and without (n=14) cGVHD. IL-6, IL-10, and TNF-α were found to correlate with the NIH (r=0.42, p=0.008; r=0.51, p=0.001; r=0.44, p=0.005; respectively) and ICCGVHD (r=0.60, p< 0.001; r=0.46, p=0.003; r=0.57, p<0.001; respectively) disease severity scores (Figure 1). [64] Thus, tear fluid biomarkers could serve as potential diagnostic and severity biomarkers in cGVHD.

Tear cytokines abnormalities have also been noted in DED subtypes. In one systematic review, individuals with various DED subtypes (n=~300 individuals per cytokine analysis) had higher levels of some cytokines compared to healthy eyes (n=~200 per cytokine analysis). Across studies, higher levels of IL-1 β, IL-6, IL-8, IL-10, IFN- γ, and TNF- ɑ were noted in DED compared to controls (Figure 1). [65] Therefore, as with oGVHD, cytokines in tear fluid may be useful markers for DED subtypes as well.

Various molecules are shared across the two conditions. For example, intercellular adhesion molecule 1 (ICAM-1), a protein found on the surface of APCs that allows their binding to T lymphocytes via LFA-1, was noted to be elevated in DED and oGVHD. [66] A Chinese study found elevated levels of ICAM-1 in the tear fluid of individuals with coGVHD (n=18) and to a lesser extent EDE-DED (n=11) (OSDI≥13 and TBUT<10 or ocular surface) (37315±33373 vs 1349±1937 pg/mL, p<0.0001). [67] Other shared tear cytokines in DED and coGVHD include IL-2, IL-8, IL-17, with higher values in coGVHD compared to DED (IL-2: 299 ±141 pg/mL vs 2.1±7.1, p<0.0001; IL-6: 96±257 vs 3.6±5.1 pg/mL, p<0.0001; IL-17: 35±48 vs 4.8±2.9 pg/mL, p<0.0001). Epidermal growth factor (EGF) levels, a cell growth regulatory, on the other hand, were higher in DED compared to coGVHD (496±488 pg/mL vs 70±197, p<0.0001). [67] Therefore, while shared cellular and soluble inflammatory components have been noted across DED and oGVHD, their frequencies and levels differ between the two disease states.

6. SYMPTOMS OF OGVHD VS DED

Symptoms of acute and coGVHD differ, both with respect to symptom descriptions and time course. One retrospective Chinese study evaluated 143 individuals with oGVHD, 21 with acute and 127 with chronic disease (based on NIH 2014 criteria). All individuals with aoGVHD reported eye discharge (described as “clear and thick mucus”), 90.5% (n=19) reported a red eye, 52% (n=11) reported excess tearing, and 43% (n=9) reported sensations of dryness. On average, symptoms began 8.0 ± 5.0 months after transplantation [68], supporting changing opinions on the timing of late-onset aGVHD. In those with coGVHD, symptoms included sensations of dryness in 91% (n=115), eye discharge in 53% (n=68), light sensitivity in 39% (n=50), and lack of tearing in 35% (n=45). Only a minority of individuals 22% (n=15) reported “mucus” eye secretions but 78% (n=53) described “fibrous and stringy” secretions, highlighting different symptom profiles between acute and coGVHD. On average, symptoms in those with coGVHD began 10.7 ± 9.1 months after transplantation. [68] Another US study evaluated 306 individuals post-HSCT with oGVHD (acute/chronic not specified) and found other symptoms (not previously mentioned) present including burning/stinging in 59% (n=181), vision changes in 53% (n=162), pain in 52% (n=159), and wind sensitivity in 46% (n=141). [69] One American study also found that sensations of grittiness were reported by 42% of individuals with coGVHD. [70] Therefore, further investigation is warranted to distinguish the symptom profile in aoGVHD vs coGVHD.

DED and oGVHD share similarities with respect to symptom profiles. A US population-based study was performed by providing a questionnaire regarding symptoms such as grittiness, burning, dryness, redness, crusting, or stuck shut as well as the frequency of these symptoms (always, frequently, sometimes, or rarely). 2482 individuals completed the survey and had the following symptoms always, frequently, or sometimes: ~33% reported grittiness, ~25% reported burning, ~22% reported dryness, and ~22% reported redness. [71] In an Australian population-based study, 1172 individuals completed a survey asking about itchiness, discomfort, grittiness, dryness, and in the last 12 months. They found that ~30% reported discomfort, ~25% reported grittiness, and ~20% reported dryness. [72] In another Australian population-based study of approximately 900 individuals, ~50% reported photophobia, ~35% reported tearing, ~32% reported discomfort, and ~17% reported dryness (with severities of mild, moderate, or severe). [73] Therefore, shared symptoms have been reported, with varying prevalence and severity, in oGVHD and various general populations around the world.

7. SIGNS OF OGVHD VS DED

While aoGVHD is less frequent (and less well studied) than coGVHD, data suggest differing signs between the two conditions. Both acute and coGVHD can affect various ocular structures, most commonly the conjunctiva and cornea. Conjunctival findings unique to aoGVHD include chemosis [74] and pseudomembrane formation. [44] For example, one US study evaluated 263 individuals after HSCT, and 79% of individuals (n=19) with aoGVHD had pseudomembrane formation. [44] Other corneal findings in aoGVHD (but not unique to aoGVHD) include epithelial erosions (with epithelial sloughing a more severe form), filamentary keratitis, and corneal ulcerations. [44, 75] Further investigation is needed to fully describe disease phenotypes in aoGVHD.

More data is available on clinical signs of coGVHD. Periorbital changes in coGVHD include vitiligo and hyperpigmentation. One French study found that 3 of 270 patients who underwent HSCT developed hyperpigmentation secondary to a lichenoid periorbital eruption. [76] Eyelid findings reported in cases of coGVHD include eyebrow and eyelash loss (madarosis), depigmentation (poliosis), and trichiasis (ingrowth/introversion of the eyelashes). [75, 77]

Meibomian gland dysfunction (MGD) is a common occurrence in coGVHD [78, 79] and upper MG area has been found reduced in individuals with coGVHD compared to post-HSCT individuals with no oGVHD. A German study used meibography to evaluate MG in 54 individuals with coGVHD and 8 individuals post-HSCT. Upper eyelid MG area (% MG over total tarsus area) was lower in coGVHD (right eye: 18.6%±14.7%; left eye: 20.5±13.2%) compared to post-HSCT non-oGVHD individuals (right eye: ~35%±8%, p=0.01; left eye: ~32%±10%, p=0.06). [80] Additionally, this study also found that overall coGVHD severity (NIH 2014 Criteria) did not correlate with upper MG area, highlighting that the severity of various coGVHD abnormalities do not necessarily correlate with one another.

Conjunctival findings in coGVHD include subtarsal conjunctival inflammation and cicatricial changes which can lead to lagophthalmos, entropion/ectropion, and symblepharon. [75, 81] Cicatrizing conjunctivitis (CC) is relatively common in coGVHD. One US study found that in 20 individuals with oGVHD, 50% (n=10) had subtarsal fibrosis, indicating the presence of CC. On the other hand, CC was not noted in any of the 40 patients with DED (OSDI > 22; Schirmer’s < 10 mm/5 min and/or TBUT < 10 seconds). Furthermore, subtarsal fibrosis extent positively correlated with fluorescein-assessed corneal epithelial disruption (r=0.32, p<0.01). [82] The hypothesis is that fibrosis in oGVHD leads to corneal epithelial damage. However, not all studies have found such high frequencies of CC. In a US-based study of 62 individuals with coGVHD, only 6% (n=4) had CC (n=3 with symblepharon). [81] Therefore, CC is one feature of coGVHD but its frequency varies by oGVHD study.

Corneal findings in coGVHD include inflammation which has been evaluated with IVCM. One review on the topic concluded that while central dendritic cell (DC) density could not distinguish eyes with and without oGVHD post-HSCT, some eyes with oGVHD had very high DC density in the central cornea (>200 cells mm²). [65] Corneal ulcers, both due to inflammation and infection, can also be seen in coGVHD, with perforation as a potential visually threatening complication. [44, 83, 84] coGVHD can also impact the corneal nerves. While disparate findings have been noted with respect to nerve density in oGVHD [85], the majority of studies found higher tortuosity (graded 0-4) in eyes with oGVHD (range across studies 2.2±0.7 to 3.1±0.7) compared to non-oGVHD post-HSCT (range across studies 1.3±0.5 to 2.2±0.5). [85]

Corneal sensation has also been examined in coGVHD. A Japanese study found that individuals with coGVHD (n=10) (Schirmer with nasal stimulation ≤10 mm/5 min, CFS ≥3 points) had a lower corneal sensation (measured with Cochet-Bonnet) compared to controls (n=14) (54.98±7.75 vs 60 mm, p<0.05). Individuals post-HSCT without coGVHD (n=5) also had decreased sensation as compared to controls but the difference was not significant (57.5±4.63 mm). [48] These studies suggest that nerve abnormalities can be seen as part of coGVHD. In summary, a variety of findings can be seen in coGVHD beyond those captured as part of the disease definition. These include periorbital changes (skin and hair), MGD, conjunctival fibrosis, and corneal nerve changes, at varying levels of severities and frequencies.

coGVHD and DED subtypes can present with overlapping signs, but some features are more unique to oGVHD, specifically CC in chronic disease. Low tear production, tear instability, and epithelial disruption are all shared features of oGVHD and some DED sub-types. For example, one Chinese study evaluated 27 individuals with coGVHD and 22 individuals with EDE-related DED (OSDI ≥13 and TBUT <10mm/5 min or CFS) and found tear production (Schirmer’s) and stability (TBUT) were lower in coGVHD than the DED group, but that the values were abnormal in both groups (TBUT: 2.3±1.7 vs 3.6±1.2 seconds, p<0.0006; Schirmer’s: 2.4±2.4 vs 12.9±7.2 mm, p<0.0001). [67] Nerve abnormalities are also seen in DED subtypes. A US study found that SS-DED had decreased nerve density and elevated dendritic cell density in comparison to controls (nerves: 156±996 vs 2725±687 μm/frame, p<0.001; dendritic cells: 71.65±72.54 vs 27.53±5.58 cells/mm², p<0.001). [86] Nerve tortuosity can also be seen as part of DED, both in SS-DED and non-SS DED. [86] MGD is also a common finding in DED with the various aspects of MGD (e.g., plugging, keratinization of the eyelid margin, abnormal meibum quality, MG atrophy) noted in a large proportion of the population. [87] A French study found that lower lid MG presence (% MG over total tarsus area) was lower in individuals with MGD (n=91) (MGD international workshop criteria) compared to those without MGD (n=55) (~1.5 vs ~1.0, p=0.003; scale 0-4). [88] Thus, while many features are shared between coGVHD and DED sub-types, unique features of GVHD include pseudomembrane formation in acute disease and cicatricial changes in chronic disease.

8. RISK FACTORS FOR OGVHD AND DED

Several risk factors have been identified for the development of chronic oGVHD. These include pre-transplant, transplant, and post-transplant factors. Regarding pre-transplant risk factors, older age, female gender, non-White race, pre-existing medical conditions, and low humidity have all been identified as risk factors for oGVHD. One Italian study examined 283 patients who underwent HSCT and found that oGVHD development was significantly associated with older age of the transplant recipient (hazard ratio (HR): 1.23, 95% CI: 1.03-1.46, p=0.02). [89] With regard to gender, a US study of 172 individuals noted that male recipients from female donors (n=38) had a significantly higher likelihood of developing oGVHD compared with other donor-recipient combinations (OR: 2.57, 95% CI: 1.09-6.06, p=0.03). [90] However, disparate results have been found across the literature. An Italian study of 283 individuals who underwent HSCT found that female recipients were at higher risk of developing oGVHD than male recipients (HR:1.59, 95% CI:1.06-2.37, p<0.02). [89] With regard to ethnicity, a Canadian study of 146 individuals found that being White lowered the odds of developing oGVHD (OR: 0.29, 95% CI: not given, p=0.01) compared to non-White individuals. [91] Several co-morbidities have also been linked to oGVHD risk. A Korean study of 635 individuals found that diabetes mellitus (DM) prior to transplant increased the risk of oGVHD development (OR: 4.22, 95% CI: 1.66–10.72). [92] Malignancy type has also been found to impact oGVHD severity. A Korean study reported that severe oGVHD (NIH grade of 3) was more frequent in individuals with myelodysplastic syndrome compared to other malignancies (OR: 7.27, 95% CI: not given, p=0.01). [57] In addition to pre-existing conditions, environmental factors during the time of HSCT may also play a role in oGVHD risk. One German study found that coGVHD incidence increased if HSCT occurred in a low-humidity setting. For example, in those who developed coGVHD, median humidity (32±7%) was lower compared to those without coGVHD (39±9%, r=−0.17, p = 0.03). [93] Thus showing a negative correlation between ward humidity and the development of coGVHD. Taken together, pre-transplant demographics, comorbidities, and environmental considerations at the time of transplantation may impact oGVHD risk.

Regarding transplant factors, stem cell source, human leukocyte antigen (HLA) match status, and donor viral seropositivity have all been found to play a role in oGVHD risk. An Italian study found that 283 individuals who received peripheral blood stem cell transplants had an increased risk of oGVHD development than those who received bone marrow transplants (HR: 2.08, 95% CI: 1.27-3.41, p=0.004). [89] A US study of 200 individuals found that individuals with HLA mismatch (5/6 matched) (n=29) had a significantly shorter median time to oGVHD onset (219 days, interquartile range (IQR): 153-348) compared to individuals who were fully matched to their donor (n=150; 294 days, IQR: 223-506, p=0.03). [94] Additionally, time to onset was shorter in matched-related compared to matched-unrelated transplants (250 days (IQR:113-512) versus 120 days (IQR: 41-279; p<0.004). [94] Viral presence has also been related to oGVHD. One Canadian study examined 146 individuals and discovered that recipients of EBV seropositive donors were more likely to develop oGVHD (OR: 4.39, CI: not given, p=0.01). [91] In consideration of all these factors, full HLA-matching, lack of EBV seropositivity, and bone marrow source may reduce the risk of developing oGVHD.

Post-transplant risk factors for oGVHD can be split into those occurring acutely, within 100 days following transplantation, and those occurring in the chronic phase, more than 100 days after transplantation. aGVHD has been found to be a risk factor for oGVHD. One Dutch study evaluated 101 individuals and found that individuals who developed aGVHD (compared to those who did not) within 100 days post-HSCT had a higher risk of developing oGVHD (n=64; OR: 5.22, 95% CI: 2.06-13.20, p=0.00). [95] In addition to having aGVHD, specific organ involvement in aGVHD such as skin (OR: 2.57, 95% CI: 1.17-5.64, p=0.02) [90] has been linked to oGVHD risk. Other studies have examined relationships between GVHD in specific organs and the severity of eye involvement in cGVHD. For example, 56% of severe (grade 3) oGVHD (NIH 2014 criteria) had cGVHD lung manifestations. [57] Therefore, skin involvement during aGVHD and lung involvement in cGVHD have been linked to oGVHD risk and severity.

Some risk factors in DED are shared with the noted risk factors of oGVHD. For example, demographics, such as female sex and increased age in both genders have been identified as DED risk factors. [96] One New Zealand study examined 322 individuals using TFOS DEWS II diagnostic criteria and found that age and female sex were significant risk factors for DED development (combination of symptoms and various signs; age: OR: 1.21, 95% CI: 1.06-1.44, p=0.01; female sex: OR: 1.83, 95% CI: 1.06-3.15, p< 0.03). [96] Regarding co-morbidities, diabetes mellitus has also been found to be a risk factor for aspects of DED. One German study evaluated 86 insulin-dependent diabetics with retinopathy and 84 non-diabetic patients. Tear production was significantly lower in diabetic patients versus controls (Schirmer’s: 10±3 vs 18±5 mm/5 min, p<0.001), [97] although mean values were normal in both groups. Viruses have also been found to contribute to both oGVHD and ATD-DED. One US study surveyed 42 men with human immunodeficiency virus (HIV)-1 and found 21% (n=9, 18 eyes) had sensations of dryness, and 15 of these eyes had low tear production (Schirmer’s<10mm /5 min), whereas 10 controls (HIV positive but no dry eye symptoms) all had Schirmer’s >10mm/5 min. [98] Like oGVHD, systemic immune disorders, such as primary and secondary SS, have been linked to aspects of DED (most closely ATD), linking inflammation in the body with inflammation in the eyes. [99] Overall, demographics, co-morbidities, viral history, and systemic inflammation are shared risk factors between oGVHD and aspects of DED.

9. TREATMENT OF GVHD, OGVHD, AND DED

The approach to the management of both oGVHD and DED is dependent on underlying disease contributors, symptom severity, and clinical signs. The goal of treatment is to manage symptoms and signs by improving tear function and nerve health. [78] Treatment options for oGVHD can be considered in terms of systemic prophylactic therapy as well as local and systemic treatments.

9.1. Prophylaxis of GVHD

Prophylaxis for GVHD is a critical component of HSCT and is integrated into the conditioning regimen chosen for each patient. The prophylaxis regimen is determined based on the underlying hematologic disease, recipient comorbidities and age, donor source, and various other factors. [19] The goal of an ideal GVHD prophylactic regimen is to prevent acute and/or cGVHD while preserving graft-versus-malignancy effect. Classically, a calcineurin inhibitor (e.g., tacrolimus) and methotrexate are given and typically continued for at least 100 days after HSCT prior to taper initiation. [100] In individuals with HLA mismatch, including haploidentical and mismatched unrelated donor HSCT, post-transplant cyclophosphamide (PTCy)-based regimens have gained acceptance as a standard of care. [101, 102] The PTCy approach is also now being adopted in matched donor HSCT. [103] Remarkably, in a retrospective registry study, PTCy-treated patients (n=206) were compared to those who received calcineurin prophylaxis alone (n=764) and were found to have a lower frequency of coGVHD (41% (n=80) versus 60% (n=450), p<0.001). [104] Therefore, PTCy should be considered in both HSCT recipients with matched or unmatched donors to lower the risk of coGVHD development.

9.2. Systemic Treatment of GVHD

Systemic treatment for cGVHD is indicated when there is at least an overall grade of moderate cGVHD. Therapy is directed at stabilizing and improving the symptoms of cGVHD to improve quality of life and reduce organ damage while balancing the toxicities of therapy. The first-line treatment for cGVHD includes systemic corticosteroids, with or without calcineurin inhibitors. [105] In cases of refractory cGHVD, including cases where features of cGVHD do not improve or recur after corticosteroid taper, further therapy is indicated. In the second and later lines of therapy, several drugs have been FDA-approved for the treatment of cGVHD including ruxolitinib [106], belumosudil [107], and ibrutinib. [108] Other commonly used cGVHD treatment options include extracorporeal photopheresis [109–111], rituximab [112], and mycophenolate mofetil. [79] The choice of therapy is driven largely by local practice patterns, as treatment sequencing and/or combinations are not well supported by the literature. [113, 114]

When possible, systemic therapies should be avoided in order to spare potentially significant toxicities such as immune suppression. If topical therapies may be effective, they are strongly considered, particularly in the setting of oGVHD as detailed below.

9.3. Targeting ocular surface inflammation

Donor T cell migration and activation are part of oGVHD pathophysiology. Corticosteroids are frequently used in the treatment of acute inflammation, but caution should be used for chronic use given the potential cataracts and glaucoma development. [78] T cell modulators are thus often used chronically in individuals with oGVHD. Cyclosporine and tacrolimus are the two most commonly used topical therapies and have been shown to reduce T cell activation and improve corneal epithelial health and goblet cell density. [115, 116] One US study treated 8 patients with cGVHD and dry eye symptoms refractory to artificial tears with cyclosporine 0.05% drops twice a day. Tear production (baseline: 7.2±4.0 mm/5 min; 3 months: 1.3±2.2 mm/5 min, p=0.003) and stability improved (baseline: 3.4 seconds; 3 months: 6.6 seconds, p=0.002) after 3 months of use. [115] Topical tacrolimus has similarly been shown to have benefits in coGVHD. In one US study, 24 individuals with coGVHD were treated with topical tacrolimus 0.05% for 10 weeks. Dry eye symptoms improved (58±24→42±20 on OSDI score, p<0.02), tear stability increased (TBUT) (0.7±1.5→2.6±3.0 seconds, p=0.003), CFS decreased (8.2±3.1→3.7±2.0, p<0.001; scale: 0-15), the frequency of HLA-DR positive cells (8.7→4.7%, p=0.03) and ICAM-1 positive cells (41→24.8%, p=0.003) also decreased. [117] However, a limitation of these therapies is that many individuals cannot tolerate them due to instillation site burning and a possible increased risk of ocular herpes keratitis. [117, 118]

9.4. Local Therapy that Targets Tear Film and Epithelial Health

Artificial tears are generally used in all HSCT patients with DED symptoms. [79, 119] Artificial tears may dilute inflammatory cytokines in tear fluid and may thus protect epithelial health. A Korean study found that preservative-free artificial tears (PFATs) decreased the tear cytokine concentrations in patients with ATD-DED (Schirmer’s < 5 mm/5 min, TBUT:<5 seconds, and CFS ≥ 1) after one month (IL-1β: before: 31.57±1.56 pg/mL→ after: 28.48±1.44 pg/mL; IL-6: 56.85±2.37 pg/mL→ 39.74±2.25 pg/mL; IL-12: 140.95±8.68 pg/mL→ 94.97±4.57 pg/mL; TNF-α: 15.63±1.20 pg/mL→ 10.16±0.93 pg/mL, all p<0.05). [120] The impact of artificial tears on inflammation needs further study specifically in individuals with coGVHD.

The use of antibiotics is also frequently used in individuals with oGVHD to target MGD. Their effect may be through changes in the ocular surface microbiome although more information is needed on this topic. [61] While there are minimal studies detailing the effect of antibiotics on oGVHD in humans, they have been evaluated in animal models. For example, in one mouse model, bone marrow from B10.D2 mice was transplanted in BALB/c mice to induce cGVHD and found that with the use of gentamicin, there was a decrease in coGVHD manifestations such as fibrosis which was decreased in comparison to controls (no antibiotics) (fibrotic area/field) (~10 vs ~100; p<0.0001). Supporting this finding, myofibroblast numbers (α- SMA/field) were lower in mice treated with gentamicin vs controls with coGVHD (~20 vs ~60; p<0.05). [121] Further studies are warranted in humans on this topic.

Autologous serum tears (AST) have been shown to improve symptoms and corneal surface health in oGVHD. One German study administered 100% AST eye drops to 17 patients with coGVHD and found improvement in symptoms and signs. Specifically, OSDI (baseline: 88±21; 6 months: 63±23, p=0.02) and CFS (OD: baseline: 3±1; 6 months: 2±1, p=0.004) scores improved after 6 months. [122] AST has also been shown to improve other aspects of tear health. One Korean study treated 16 patients with coGVHD with 20% AST and found that after 4 weeks, impression cytology (IC) (squamous metaplasia; graded 0-6 using Tseng classification) (before: 3.4±0.5; after: 2.4±0.6, p<0.01), and tear osmolarity (before:311±16; after:285±6 mOSM, p<0.01) values decreased. In addition, goblet cell density (before: 90.6±44.1; after:122.6±49.2 cells/mm², p<0.01) increased. [123] In practice, AST and other blood-derived products are often used in individuals who have persistent symptoms and signs despite artificial tear use. AST and other blood products are believed in part to exert their effect via growth factors in the serum. As such, data are needed on the impact of recombinant growth factors, such as nerve growth factor, on coGVHD symptoms and signs. [124]

Mucin-secretagogue eye drops have also been used in oGVHD given that conjunctival goblet cells may be impacted by disease. [79] Diquafosol ophthalmic solution is a mucin secretagogue that activates P2Y2 receptors on the conjunctiva, thus enhancing mucin production. A Japanese study evaluated 10 individuals with coGVHD who used diquafosol 3%. Improved CFS (before: 5.9±0.6; after: 1.3±1.1 points, p=0.00002; scale: 0-9) and tear stability (TBUT: before: 2.6±0.9; after: 4.6±1.6 seconds, p<0.01) were noted after a mean of 12 months of use. [125] Diquafosol 3.0% ophthalmic solution is only approved in Japan, South Korea, Thailand, Vietnam, and China. [126]

Punctal occlusion has also been used to increase the time tears remain on the ocular surface. [78, 79] One Swiss study evaluated 16 men with coGVHD who underwent punctal occlusion. One year following occlusion, patients reported less symptoms (before:1.10; after: 0.59, p<0.001; scale 0-2) and less CFS (before:1.07±0.42; after:0.75±0.43, p<0.001; scale: 0-2) compared to pre-treatment. [127] Therefore, punctal occlusion can be considered in individuals with symptoms of dry eye and signs of coGVHD post-HSCT, who have failed other therapies.

Hormonal modulation may also be considered as a treatment for oGVHD. 1% progesterone gel administered on the forehead was found to reduce the frequency and severity of symptoms in individuals with coGVHD after 6 weeks compared to placebo (SANDE; scale: 0- 100) (progesterone: mean reduction: −26.3, standard error (SE): 4.1 vs placebo:−5.6, SE:5.8) as well as signs after 2 weeks compared to placebo (CFS) (progesterone: mean reduction: −2.5, SE:0.5 vs placebo: −0.2, SE: 0.8, p<0.05). [128] In rats with intact or excised lacrimal glands, 1% progesterone was found to provide corneal antinociception. [129] More data are needed on the impact of hormones as a treatment for GVHD.

9.5. Devices

Scleral contact lenses and bandage soft contact lenses (BSCL) have also been used in oGVHD to improve symptoms and surface health. One US study of 19 individuals with coGVHD evaluated the impact of extended-wear BSCLs replaced every 2 to 4 weeks and topical antibiotic prophylaxis. OSDI scores decreased 2 weeks after BSCL placement (54.5±6.19→36.8 ±5.32, p=0.002) and continued to decrease with BSCLs 3 months later (35.6±6.5 units, p=0.001). [79, 130] Amniotic membranes (AM) have also been used in individuals with oGVHD. [78] In one US study, a self-retained cryopreserved AM was placed for three days in the right eye of an individual with aoGVHD complicated by severe corneal epithelial disruption refractory to artificial tears, corticosteroids, and bandage contact lens, while this conventional therapy was continued in the left eye. [131] After three days, the AM was removed from the right eye and BSCL was removed from the left eye. In the right eye, visual acuity improved from 20/500 to 20/70 in three days, with complete epithelialization of the cornea, a decrease in conjunctival inflammation, and symptomatic improvement. However, in the left eye, visual acuity improved from 20/300 to 20/150, with residual CFS and persistent symptoms. [131] In general, contact lenses and AMT are reserved for more refractory coGVHD cases, after artificial tears, anti-inflammatory, and blood products (if available) have achieved insufficient results.

9.6. Surgeries

Although rare, there are reports of limbal and conjunctival stem cell allografts being performed in coGVHD-associated limbal stem cell deficiency using the same donor as the HSCT. In one US study, 2 patients were treated with living-related conjunctival limbal allografts (lr-CLAL) due to limbal stem cell deficiency in the setting of coGVHD. In both cases, visual acuity improved following lr-CLAL with successful restoration of the ocular surface by re-epithelialization. [132]

9.7. Future Therapies

Several studies are examining agents that target the other pathophysiological mechanisms leading to the symptoms and signs of oGVHD. [78, 79] One such compound is a janus kinase (JAK) inhibitor with spleen tyrosine kinase (SYK) which is being studied in a phase 2 study of 30 individuals with coGVHD (treated with R348 or vehicle topically, twice daily, for 12 weeks). While R348 showed no difference in symptoms (OSDI) or tear production (Schirmer’s), it was associated with decreased CFS at 12 weeks (−6.0±3.9) compared with vehicle (−2.1±2.6, NEI scoring; p<0.05). [133] Transilast, a TGF-β inhibitor, is another compound undergoing investigation. In a Japanese non-randomized study of 18 individuals with coGVHD, 8 were given topical transilast and 10 topical artificial tears, sodium hyaluronate, and vitamin A. Individuals treated with transilast were assumed to have less lacrimal gland fibrosis given preserved Schirmer scores (transilast:15.8±12.1→22.0±10.1 mm; vehicle: 13.0±9.6→9.9± 10.6 mm, p< 0.05) after 3 months of use. [134] Exosome technology (membrane-bound extracellular vesicles) from mesenchymal stromal cells has also been examined in eye drop form as a treatment of coGVHD, with reduction of inflammation noted and improvement in corneal staining. [135] There is excitement about the potential for improved treatments in coGVHD as these new therapies are being explored.

9.8. Therapies shared with DED

Given overlaps in clinical features and pathophysiology, all the local therapies mentioned above have also been used to target different aspects of DED. For example, short courses of topical corticosteroids and long-term treatment with various concentrations of cyclosporine are often used to treat various DED sub-types. [136] Similarly, devices such as punctal plugs, scleral contact lenses, and BSCL, used in oGVHD, are also used in patients with more severe DED. [137, 138]

10. CONCLUSION

The studies above highlight the similarities and differences between the presentation, pathophysiology, clinical manifestations, and treatments of oGVHD and DED. There are different diagnostic criteria for oGVHD and DED, but, similarly, there is a lack of consensus on the optimal diagnostic criteria for both entities. While oGVHD and DED have different underlying causes, both diseases share cellular and soluble inflammatory mediators that enter the lacrimal gland, conjunctiva, and cornea and drive disease manifestations. While these cellular and soluble mediators have been explored more thoroughly in DED, there are many gaps in the literature in oGVHD that are pivotal to better understanding targetable pathophysiological mechanisms. Both conditions can present with symptoms of ocular surface pain and visual disturbances and signs of decreased tear production, MGD, and epithelial disruption. Fibrosis is a distinctive feature of coGVHD that is not as commonly seen in DED. In both oGVHD and DED, many individuals chronically suffer from symptoms. Therefore, better treatment algorithms are needed to reduce disease morbidity, and in the case of oGVHD, mortality. Future therapies for oGVHD targeting the underlying pro-inflammatory and pro-fibrotic mechanisms are currently underway and require further investigation. Both oGVHD and DED require a better understanding of underlying mechanisms to improve treatment regimens in the individual patient.

Table 4.

Therapies for GVHD

Treatment type	Medications/interventions		Mechanism of action
GVHD prophylaxis	Tacrolimus		Calcineurin inhibitor [100]
	Methotrexate		Dihydrofolate reductase inhibitor [139]
	Cyclophosphamide		Alkylating agent [104]
GVHD treatment	First-line therapies:	Systemic corticosteroids ± calcineurin inhibitor	Inhibition of prostaglandin synthesis [105]; calcineurin inhibitor [100]
	Second-line (FDA-approved) therapies:	Ruxolitinib	Tyrosine kinase inhibitor [106]
		Belumosudil	Serine/threonine kinase inhibitor [107]
		Ibrutinib	Bruton’s tyrosine kinase inhibitor [108]
	Other commonly used therapies:	Extracorporeal photopheresis	Lymphocyte irradiation using photoactivating 8-methoxypsoralen with ultraviolet A light [109–111]
		Rituximab	Anti-CD20 monoclonal antibody [112]
		Mycophenolate mofetil	Inosine monophosphate dehydrogenase inhibitor [79]
Targeting ocular surface inflammation	Topical corticosteroids		Suppress local inflammation [78]
	Topical cyclosporine and tacrolimus		Reduce T cell activation; improves corneal epithelial health and goblet cell density [115, 116]
Local Therapy that Targets Tear Film and Epithelial Health	Artificial tears		Dilute inflammatory cytokines in tear fluid; preserve epithelial health [120]
	Antibiotics		Target MGD in oGVHD with azithromycin & doxycycline; animal studies have shown decreased ocular fibrosis in cGVHD with gentamicin [121]
	Autologous serum tears		Provide lubrication, epithelial and nerve growth factors, and vitamins to ocular surface [79, 122]
	Punctal occlusion		Increase the time tears remain on ocular surface [78, 79] [127]
	1% progesterone gel (on forehead)		Hormonal modulation [128]
Devices	Bandage contact lens		Stabilize tear film [79, 130]
	Amniotic membrane transplantation		Suppress inflammation and provide epithelial growth factors [131]
Surgeries	Limbal and conjunctival stem cell transplantation		Replace insufficient limbal stem cells [132]
Future therapies	R348		Janus kinase inhibitor with spleen tyrosine kinase [133]
	Transilast		TGF-β inhibitor [134]
	Exosome technology		Reduce local inflammation [135]

GVHD= Graft-versus-host disease; HSCT= hematopoietic stem cell transplantation; FDA= Food and Drug Administration; CD20= cluster of differentiation 20; MGD= meibomian gland disorder; oGVHD= ocular graft-versus-host disease; cGVHD= chronic graft-versus-host disease; TGF-β= transforming growth factor- beta