Assessment of Corneal Epithelial Changes and Related Factors in Ocular Chronic Graft-Versus-Host Disease (GVHD) by in Vivo Confocal Microscopy

ABSTRACT

Purpose

To evaluate corneal epithelial changes and related factors in chronic ocular graft-versus-host disease (oGVHD) patients.

Methods

21 patients (35 eyes) with chronic oGVHD and 8 patients (12 eyes) without oGVHD after bone marrow transplantation were recruited for assessment involving in vivo confocal microscopy (IVCM) analysis, ocular surface parameter determination and tear cytokine level analysis. The IVCM corneal epithelial scoring system was used to evaluate corneal epithelial changes.

Results

There was a significant difference in the corneal epithelial score (p = .001) between the two groups. The corneal epithelial scores were significantly correlated with the corneal fluorescein staining scores (CFS, r = 0.463, p < .001), Schirmer’s test (r = −0.389, p = .009) and tear cytokine levels of EGF (r = −0.491, p < .001) and APRIL (r = −0.318, p = .030).

Conclusions

The depth of corneal epithelial defects can be estimated by the CFS. Corneal epithelial changes of chronic oGVHD are considered to be associated with lacrimal deficiency and a lack of EGF.

KEYWORDS:

• Corneal epithelial defect
• hematopoietic stem cell transplant
• in vivo confocal microscopy
• ocular graft-versus-host disease
• tear cytokines

Chronic graft-versus-host disease (cGVHD) is a major and severe complication after allogeneic hematopoietic cell transplantation (HCT) and has a significant impact on patients’ quality of life.^1,2 Its clinical manifestations are similar to those of an autoimmune disease, and it may affect many organs, including skin, nails, mouth, eyes, muscles, fascia, joints, female genitalia, gastrointestinal tract, liver and lungs.^1,3 A total of 60–90% of patients who receive hematopoietic stem cell transplantation present with ocular manifestations.⁴ Up to 90% of chronic ocular GVHD (oGVHD) patients have symptoms similar to severe dry eye, such as irritation, dry eyes, photophobia, itching, redness, pain and foreign body sensation.^5–7 Due to tear deficiency, goblet cell loss and meibomian gland dysfunction (MGD), the tear film is unstable, resulting in ocular surface damage, including punctate keratitis, persistent epithelial defects, corneal ulceration, perforation and angiogenesis.^8–10 The pathophysiological process of ocular GVHD involves the participation of various cytokines. Previous studies have revealed significant differences in the levels of several tear cytokines between ocular GVHD patients and healthy individuals, dry eye patients and non-GVHD patients, which is expected to assist the screening and diagnosis of GVHD.^11–13 However, no study has been conducted to explore the correlation between inflammatory factors and epithelial injury in patients with GVHD.

Epithelial defect is a common lesion associated with GVHD. A previous study revealed that the prevalence of persistent epithelial defects (PEDs) in patients with chronic ocular GVHD was 8.1%.¹⁴ In vivo confocal microscopy (IVCM) is a noninvasive examination tool that can be used to evaluate corneal epithelial morphological changes and defect depths. Comparing the central corneal images of GVHD patients with normal controls using IVCM, it was found that the density of superficial cells, wing cells and basal cells of the corneal epithelium was significantly decreased in patients with ocular GVHD.¹⁵ In addition, changes in epithelial cells in patients with GVHD are similar to those in patients with dry eye, but more severe. Enlarged, hyperreflective epithelial cells in patients with ocular GVHD were observed in the IVCM images.¹⁵ Pathological sections of the GVHD animal model showed atrophy of the corneal epithelium and vacuolization of epithelial cells in the basal layer and wing cell layer.¹⁶ However, besides describing morphological changes and calculating epithelial cell density, there is a lack of clinical research that has comprehensively evaluated the epithelial changes in ocular GVHD patients.

Therefore, the purpose of this study was to use IVCM to explore the corneal epithelial cell changes in patients with ocular cGVHD after bone marrow transplantation and their correlation with ocular surface parameters and the level of tear inflammatory factors. It is hoped that this study will contribute to monitoring and assessing the severity of cGVHD and further elucidate the factors related to epithelial changes in patients with ocular cGVHD.

Material and methods

Patients

This study was a single-center, case-control study involving 35 eyes of 21 patients with oGVHD and 12 eyes of 8 patients without oGVHD (non-oGVHD) after bone marrow transplantation. The participants were recruited from the Cornea and Ocular Surface Disease Specialist Clinic.

Ocular chronic GVHD was diagnosed by several experts from Peking University Third Hospital based on the 2014 National Institutes of Health (NIH) Consensus Criteria,² the 2013 International Chronic Ocular GVHD Group Consensus¹⁷ and rich clinical experience.

Patients who met the following criteria were included in the experimental group: (1) diagnosed with ocular cGVHD; (2) first visit to the ophthalmology clinic; and (3) had not received topical immunosuppressant treatment. The inclusion criteria for the control group were as follows: (1) had undergone allogeneic HSCT; (2) no ocular rejection occurred; and (3) not diagnosed with ocular GVHD. Patients presenting with the following were excluded: (1) had ocular surgery in the past six months; (2) had other ocular diseases (autoimmune diseases, glaucoma, infection, retinopathy, allergy, cataract, ocular injury and so on); (3) were unable to undergo an IVCM examination because of poor physical and/or mental states; and (4) had incomplete medical records.

This study was approved by the Peking University Third Hospital Medical Science Research Ethics Committee. The study complied with the Health Insurance Portability and Accountability Act of 1996 and adhered to the tenets of the Declaration of Helsinki. All participants provided written informed consent.

Clinical examination

The clinical examination process was essentially the same as that described previously.¹¹ First, the patients’ demographic data, type of primary hematological disease, HLA matching, other organ rejections, and previous treatment were collected. The OSDI (Ocular Surface Disease Index) questionnaire was used to assess patients’ symptoms. In addition, the following data were collected for objective ocular examination: best visual acuity, intraocular pressure, corneal fluorescein staining (CFS) score, fluorescein tear film break-up time (T-BUT), and Schirmer’s test score without anesthesia.

The environment was controlled when samples and measurements were taken. Clinical examination was performed and tear samples were collected every Wednesday afternoon in the same consulting room, where the ambient temperature was controlled at 16–28 degrees centigrade, the humidity was controlled at 30–60%, and the light source is indoor and constant.

Tear collection and analysis

After obtaining the patient’s consent, 30 ml of sterile saline was instilled into the conjunctival sac of each eye, and a total of 15 ml tears were subsequently collected at the medial and lateral canthus with a capillary tear collector. The samples were then immediately stored at −80°C for later use. The concentrations of the following nine cytokines in the tears were examined: IL-6, IL-10, TNF-a, EGF, MMP-2, MMP-3, MMP-7, BAFF and APRIL. Microsphere-based immunoassay analysis was used to measure the cytokine concentrations.^11,18,19

In vivo confocal microscopy

The central cornea of all patients was scanned using in vivo confocal laser scanning microscopy (Heidelberg Retina Tomograph III [RCM/HRT III]; Heidelberg Engineering GmbH). The examination steps were as follows: First, the examiners applied an anesthetic to the patient’s ocular surface twice. The ophthalmic gel was dropped onto the surface of a corneal microscope objective, and a sterile corneal contact cap was then placed. The patient placed his jaw and forehead into the bracket and fixed his gaze at a point. The examiner adjusted the position of the laser scanning camera so that the laser beam was in the corneal lesion area. The camera was slowly advanced forward so that the contact cap was in slight contact with the central cornea. Corneal images at different depths were obtained by adjusting the focal plane. Each collected image was 400 × 400 μm in size.

Image analysis

Images from the epithelial layer of the patients’ cornea were selected for analysis. The two most representative images of the wing and basal cell layers of the corneal epithelium were selected for the quantitative analysis of each eye. ImageJ software (http://imagej.nih.gov/ij/; provided in the public domain by the National Institutes of Health, Bethesda, MD, USA) was used to count the number of cells, measure the area of the pictures and calculate the density of the wing cells and basal cells. The rectangular tool was used to intercept the clearest section of the cell in each picture for counting. The area of the region ranged from 17000 μm² to 22000 μm², and more than 100 cells were counted per graph. All intact epithelial cells and truncated cells at the top and right borders of the selected region were counted.

Based on the previous in vivo confocal microscopic scoring system established by Pauly et al,^20,21 a corneal epithelial scoring system was established to comprehensively evaluate the changes in the corneal epithelium (Table 1). For the construction and specific operation of the scoring system, please refer to.²⁰ Briefly, the corneal epithelium were divided into three layers, the superficial, wing and basal cell layers, and each layer was observed and scored. The experimenter observed IVCM images of the epithelium to determine whether the epithelium was defective and whether the cells could be counted and subsequently chose the corresponding level of the presence (P) parameter. Then, it was determined whether the shape/size (S/S), reflectivity (R), and inflammation (I) parameters were abnormal. The clinical signs corresponding to the abnormal types of S/S, R and I are shown in Table 1. Finally, all the parameters of all three layers were combined to obtain the total score (max 42).

Table 1. IVCM scoring system for the evaluation of corneal epithelium.

			Single level Parameters			Final scores
			S/S*	R*	I*
			+1pt	+1pt	+1pt
I.SUPERFICIAL EPITHELIUM
Multilevel parameter:PRESENCE (P)	Level A. totally present & cells can be counted	0pt				0/1/2/3
	Level B. totally present & cells cannot be counted	4pt				4/5/6
	Level C. partially present & cells can be counted in present areas	7pt				7/8/9/10
	Level D. partially present & cells cannot be counted in present	11pt				11/12/13
	Level E. mainly absent (>3/4 of the entire corneal surface)	14pt				14
Score I = (P)+(S/S)+(R)+(I)
II. WING EPITHELIUM
Multilevel parameter:PRESENCE (P)	Level A. totally present & cells can be counted	0pt				0/1/2/3
	Level B. totally present & cells cannot be counted	4pt				4/5/6
	Level C. partially present & cells can be counted in present areas	7pt				7/8/9/10
	Level D. partially present & cells cannot be counted in present	11pt				11/12/13
	Level E. mainly absent (>3/4 of the entire corneal surface)	14pt				14
Score II = (P)+(S/S)+(R)+(I)
III. BASAL EPITHELIUM
Multilevel parameter:PRESENCE (P)	Level A. totally present & cells can be counted	0pt				0/1/2/3
	Level B. totally present & cells cannot be counted	4pt				4/5/6
	Level C. partially present & cells can be counted in present areas	7pt				7/8/9/10
	Level D. partially present & cells cannot be counted in present	11pt				11/12/13
	Level E. mainly absent (>3/4 of the entire corneal surface)	14pt				14
Score III = (P)+(S/S)+(R)+(I)
FINAL SCORE (max 42) = (I) + (II) + (III)
For items I, II and III, please select one P severity level (A, B, C, D or E), then evaluate S/S, R and I on the same line: for each of them, give one point if at least one sign of abnormality in the corresponding list^a is detected. Add P, S/S, R and I scores and select the total score in the right column. ^aLists of detailed criteria:Shape/Size (S/S) (only when cells can be counted): anisocytosis, microcytosis, macrocytosis, cells irregular in shape, edematous cells, swollen cells, loss of cell bordersReflectivity (R): abnormal reflectivity patterns, i.e., hyperreflective cells, nuclei not visible for superficial epithelium, distinct heterogeneously reflective areas/microcysts for basal epitheliumInflammation (I): Presence of inflammatory infiltrates (bright points or dendritic-like structures)

This scoring standard referred to the article of Pauly et al. ²⁰ and had been partially modified.

Sample size calculation and statistical analysis

Before the experiment, we calculated the required sample size based on previous literature to have enough statistical power. Since non-oGVHD patients after bone marrow transplantation were relatively difficult to be enrolled, we adopted a 3:1 enrollment method between experimental group (N1) and control group (N2). According to the data reported by Ahmad Kheirkhah et al²² (CFS, 7.1 ± 2.4 in GVHD group and 5.4 ± 1.8 in DED group), the calculated sample size was N1 = 26, N2 = 8 (α = 0.05, power (1-β) = 0.9). According to the data reported by Jingliang He et al²³ (CFS, 5.17 ± 2.29 in GVHD group and 0.20 ± 0.63 in non-GVHD group), the calculated sample size was N1 = 6, N2 = 2 (α = 0.05, power (1-β) = 0.9). According to the data reported by Tudor C. Tepelus et al¹⁵ (basal cell density, 7851 ± 724 cells/mm² in oGVHD group and 8570 ± 913 cells/mm² in DED group), the calculated sample size was N1 = 35, N2 = 11 (α = 0.05, power (1-β) = 0.8). Finally, 35 eyes were included in the experimental group and 12 eyes were included in the control group.

Statistical analysis was performed using SPSS for Windows version 24 (SPSS, Inc., Chicago, IL, USA). The χ2 test was used for the comparison of qualitative variables. The Shapiro–Wilk test was conducted to determine whether the continuous variables were normally distributed. Normally distributed variables are expressed as the mean ± standard deviation (SD), nonnormally distributed variables are expressed as the median ± interquartile range, and categorical variables are expressed as frequencies and percentages. For normally distributed variables, a t test was used, and for nonnormally distributed variables, the Mann–Whitney U test was used to compare whether there were significant differences between the two groups. In addition, Spearman’s rank-order correlation test was conducted to verify the correlations between variables. p values were two-sided, and less than 0.05 was considered statistically significant.

Results

Demographic details and clinical characteristics of the study subjects

Thirty-five eyes from 21 chronic oGVHD patients (19 males and 16 females) and twelve eyes from 8 non-oGVHD patients after HSCT (9 males and 3 females) were included in this study. The demographic details of the patients in both groups are shown in Table 2. All patients had received allogeneic hematopoietic stem cell transplantation. There were no significant differences in age (p = .282), sex (p = .207), donor sex (p = .190), HLA matching (p = .195), topical eye drops or systemic drugs used in the past month (p > .05) between the oGVHD and non-oGVHD groups. The primary hematological diseases of the oGVHD group were acute myeloid leukemia (AML, n = 8), acute lymphocytic leukemia (ALL, n = 6), lymphoma (n = 1), plasma cell leukemia (PCL, n = 1), myelodysplastic syndrome (MDS, n = 3) and aplastic anemia (AA, n = 2), and those in the non-oGVHD group were AML (n = 2), ALL (n = 2), chronic myeloid leukemia (CML, n = 1), acute lymphoblastic transformation of CML (n = 1) and MDS (n = 2).

Table 2. Demographic details of the study subjects.

	Total (N = 47)	oGVHD Group (N = 35)	Non-oGVHD Group (N = 12)	p Value
Age ^a, y	35 ± 28	38 ± 28	35 ± 33	0.282
Sex, %				0.207
Male	28 (60)	19 (54)	9 (75)
Female	19 (40)	16 (46)	3 (25)
Donor Sex, %				0.190
Male	30 (64)	22 (63)	8 (67)
Female	17 (36)	13 (37)	4 (33)
HLA Matching, %				0.195
Haploidentical	26 (55)	19 (54)	7 (58)
Nonidentical	1 (2)	0 (0)	1 (8)
Identical	20 (43)	16 (46)	4 (33)
Eye drops, used/not used in the past month
Artificial tears	36/11	29/6	7/5	0.083
Immunosuppressant	0/47	0/35	0/12	/
Glucocorticoid	15/32	12/23	3/9	0.552
Eye drops that promote corneal epithelial growth	17/30	10/25	7/5	0.064
Antibiotics	9/38	9/26	0/12	0.051
NSAIDs	4/43	2/33	2/10	0.241
Oral or intravenous drugs, used/not used in the past month
Steroids	17/30	10/25	7/5	0.133
Antibiotics	7/40	3/32	4/8	0.108
	TotalN = 29	oGVHD GroupN = 21	Non-oGVHD GroupN = 8	p Value
Primary hematological diseases, %				0.381
Acute myeloid leukemia (AML)	10 (34)	8 (38)	2 (25)
Acute lymphocytic leukemia (ALL)	8 (28)	6 (29)	2 (25)
Chronic myeloid leukemia (CML)	1 (3)	0 (0)	1 (13)
Acute lymphoblastic transformation of CML	1 (3)	0 (0)	1 (13)
Lymphoma	1 (3)	1 (5)	0 (0)
Plasma cell leukemia (PCL)	1 (3)	1 (5)	0 (0)
Myelodysplastic syndrome (MDS)	5 (17)	3 (14)	2 (25)
Aplastic anemia (AA)	2 (7)	2 (10)	0 (0)

^aMedian ± Interquartile Range, Mann–Whitney U test

Other variables, χ2 test

The patients’ clinical characteristics and details of the tear analysis in both groups are shown in Table 3 [median ± interquartile range]. There was no significant difference in visual acuity [(Log MAR), p = .595], intraocular pressure [mmHg, (p = .864)] or OSDI scores (p = .415) between the two groups. The CFS scores in the oGVHD group were significantly higher than those in the non-oGVHD group (p < .001***). In contrast, the opposite was observed in the tear film break-up time [BUT, secs, (p = .031*)] and Schirmer tear secretion scores [mm, (p < .001***)] between the two groups. In addition, nine cytokines were analyzedL IL-6, IL-10, TNF-a, EGF, MMP2, MMP3, MMP7, BAFF and APRIL. The results showed that EGF (p = .001**), MMP7 (p = .032*) and APRIL (p = .002**) levels were significantly different between the two groups. Detailed information on the levels of tear cytokines is shown in Table 3.

Table 3. Clinical characteristics of the study subjects.

	oGVHD Group (N = 35)	Non-oGVHD Group (N = 12)	p
Visual acuity	0.60 ± 0.43	0.50 ± 0.60	0.595
Intraocular pressure (mmHg)	13.60 ± 4.90	13.60 ± 5.30	0.864
OSDI ^a	31.00 ± 20.167	25.27 ± 17.26	0.415
Corneal fluorescein staining scores (CFS)	6.0 ± 5.0	0.0 ± 2.0	<0.001***
Central CFS	1.0 ± 2.0	0.0 ± 0.0	<0.001***
Tear film break-up time (BUT, secs)	2.0 ± 1.0	3.0 ± 1.0	0.031*
Schirmer tear secretion scores (mm)	1.0 ± 1.0	5.0 ± 3.0	<0.001***
IL-6	89.27 ± 208.81	47.11 ± 320.26	0.845
IL-10	32.40 ± 29.49	28.06 ± 9.39	0.903
TNF-a	10.83 ± 16.68	14.23 ± 83.15	0.961
EGF	57.63 ± 120.22	275.45 ± 438.61	0.001**
MMP2	2151.71 ± 5238.42	4414.99 ± 27242.64	0.845
MMP3	8529.32 ± 22065.45	3081.67 ± 36708.43	0.137
MMP7	38500.36 ± 106174.38	179400.77 ± 330707.68	0.032*
BAFF	146.28 ± 228.79	143.68 ± 410.15	0.807
APRIL	1375.12 ± 1887.65	10690.68 ± 42428.27	0.002**

^aMean ± standard deviation, t test

Other variables, median ± interquartile range, Mann–Whitney U test

*Statistically significant (p < 0.05) compared with controls

Details of confocal analysis and the relationship between clinical and IVCM parameters

Figure 1 shows representative IVCM pictures of ocular cGVHD patients’ corneal epithelia. Epithelial defects, morphological changes, abnormal reflectivity or inflammatory cell infiltration were observed in the corneal epithelium of some patients.

Figure 1. Representative IVCM images of superficial, wing and basal corneal epithelial layers in two ocular cGVHD patients. a Wing epithelium of patient 1 with fuzzy cell boundaries and enlarged, hyperreflective nuclei. b Basal epithelium of patient 1 with fuzzy cell boundaries and visible, hyperreflective nuclei. c Superficial epithelium of patient 2 showing enhanced desquamation with invisible nuclei. d Wing epithelium of patient 2 with enlarged and irregularly shaped cells. e Basal epithelium of patient 2 with mostly normal cells.

The results of the analysis of the corneal epithelial score and cellular density are shown in Table 4. In vivo confocal microscopy showed that there was a significant difference in the corneal epithelial score (p = .001**) and the basal cell density (p = .049*) between the two groups. However, epithelial cell density in the wing cell layer (p = .804) did not differ between the two groups. The wing cell density and the basal cell density were 4973.86 ± 1106.72 cells/mm² and 6416.89 ± 1478.93 cells/mm² in the oGVHD group and 5066.55 ± 1129.66 cells/mm² and 7734.29 ± 1607.35 cells/mm² in the non-oGVHD group, respectively [median ± interquartile range].

Table 4. Results of the IVCM analysis.

	oGVHD Group (N = 35)	Non-oGVHD Group (N = 12)	p
Corneal epithelial score	14.54 ± 4.70	8.75 ± 4.45	0.001**
Wing cell density (cells/mm^2)	4973.86 ± 1106.72	5066.55 ± 1129.66	0.804
Basal cell density (cells/mm^2) ^a	6416.89 ± 1478.93	7734.29 ± 1607.35	0.049*

^aMedian ± Interquartile Range, Mann–Whitney U test

Other variables, mean ± standard deviation, t test

*Statistically significant (p < 0.05) compared with controls

The correlations between ocular surface parameters and the confocal analysis are shown in Table 5. The corneal epithelial scores significantly correlated with corneal fluorescein staining scores (p = .001**, r = 0.463), central corneal fluorescein staining scores (p < .001***, r = 0.491) and Schirmer tear secretion scores (p = .009**, r = −0.389). We also analyzed the correlation between the corneal epithelial score and tear cytokines and found that the corneal epithelial score was significantly correlated with EGF (p < .001***, r = −0.491) and APRIL (p = .030*, r = −0.318).

Table 5. Correlations between ocular surface parameters, tear cytokines and confocal analysis.

	OSDI		CFS		Central CFS		BUT			Schirmer tear secretion scores
Corneal epithelial score
r	0.156		0.463		0.491		−0.271			−0.389
p	0.348		<0.001***		0.001**		0.069			0.009**
	IL-6	IL-10	TNF-a	EGF			MMP-2	MMP-3	MMP7	BAFF	APRIL
Corneal epithelial score
r	−0.097	−0.163	0.028	−0.491			−0.105	0.166	−0.145	0.086	−0.318
p	0.517	0.275	0.853	<0.001***			0.482	0.264	0.330	0.563	0.030*

Spearman’s test was used for the correlation analysis

Discussion

Our study recruited 21 patients (35 eyes) with chronic ocular GVHD and 8 patients (12 eyes) without ocular GVHD after HSCT. The patients underwent clinical examination, tear collection for cytokine analysis and in vivo confocal microscopy. We evaluated the corneal epithelium of the two groups through IVCM and explored the correlations between these data and clinical observations and tear cytokine levels, aiming to develop methods for assessing the severity and treatment of GVHD and to explore the relative factors of corneal epithelial defects and morphological changes in patients with chronic oGVHD.

Morphological changes of epithelial cells under IVCM

IVCM is an important noninvasive tool for observing the morphology of the corneal layer.²⁴ Under an in vivo confocal microscope, the three layers of the corneal epithelia are clearly visible, from shallow areas to deep areas. Normal superficial cells are large and polygonal with a bright cytoplasm, small, centered, highly reflective nuclei and perinuclear dark halos. Wing cells, which form the middle layers of epithelia, are uniformly polygonal, closely arranged and smaller than superficial cells, with highly reflective cell boundaries, low reflective cell bodies, and nuclei that are generally difficult to see. Basal cells are light and dark polygonal cells with the smallest cell bodies, closely arranged, highly reflective cell boundaries and invisible nuclei.²⁵

Profound changes were observed in the partial corneal epithelium of oGVHD, consistent with the findings of previous studies.¹⁵ The superficial epithelial cells showed fuzzy cell boundaries and invisible nuclei. Wing cells and basal cells exhibited morphological changes, nuclear hyperreflexia and inflammatory infiltration (Figure 1). These microstructure changes were similar to those in severe DED patients¹⁵ and may be related to tear film hyperosmolality, enhanced tear evaporation, ocular surface inflammation and ocular surface homeostasis disorder.²⁶

Changes in epithelial cell density under IVCM

Previous researchers have used IVCM to study corneal epithelial cell density, sub-basal nerves, globular immune cells (GICs) and dendritic cells (DCs) in patients with oGVHD.^15,22,23,27 They found that the density of corneal epithelium in patients with ocular GVHD was significantly lower than that in normal subjects but similar to that in dry eye patients with the same disease severity and patients with hematological malignancies without cGVHD.^15,22,27

By observing IVCM images, researchers counted the cell density of the wing and basal epithelial layer and performed a statistical analysis. We demonstrated that the density of wing cell layers was not significantly different between the two groups, but the density of basal cell layers in patients with ocular cGVHD was significantly lower than that in patients without GVHD after HSCT. This result shows that the density of basal cells is decreased in patients with ocular GVHD, which is consistent with the relatively more serious clinical examination indexes of cGVHD patients.

Corneal epithelial score

An analysis of the existing literature shows that most researchers have focused on describing epithelial images and calculating cell density; there is a lack of indicators for the comprehensive evaluation of epithelia and epithelial defects. In this study, the corneal epithelial score was used to evaluate epithelial defects and cell morphological changes for the first time. Our study referred to the corneal evaluation system established in the previous literature^20,21 and made some modifications to obtain an epithelial evaluation system suitable for GVHD patients. Compared with simply counting epithelial cell density, this method can more comprehensively quantify the changes in epithelial cells. In addition, based on this quantitative parameter, this study permitted analysis of the degree of epithelial changes and highlighted the relationship between clinical indicators and morphological changes in a statistically significant manner.

Our results revealed a significant difference in the corneal epithelial score between the experimental group and the control group, indicating that epithelial defects and morphological changes are more serious in patients with ocular GVHD.

The relationship between clinical parameters and IVCM corneal epithelial score

The corneal epithelial scores were compared with the clinical examination data, and the results showed that the corneal epithelial score was significantly correlated with the corneal fluorescence staining score and Schirmer tear secretion score.

Clinically, we typically evaluate the degree of epithelial defects through CFS; however, this method can only reveal whether there is an epithelial defect and the area of the defect and cannot demonstrate the depth of the defect and cell microscopic changes. Moreover, in most cases, a large area of shallow defects can still quickly restore transparency after treatment, which has little impact on patient prognosis. The depth of infiltration of the defect lesions, on the other hand, substantially impacts prognosis. When the defect reaches a certain depth or causes a certain level of damage to the deep epithelium, healing is challenging, and corneal cloudiness that affects vision can be present. Our results revealed a significant correlation between the epithelial score and corneal fluorescence staining, suggesting that the depth of epithelial defects can be roughly understood from the corneal fluorescein staining score. This result shows that the histological changes in the corneal epithelium obtained by CFS can reflect the degree of cytological changes in the corneal epithelium. It would be more convenient for clinicians to evaluate the severity and prognosis of epithelial lesions for ocular GVHD patients and is of clinical significance for primary hospitals without IVCM.

In addition, the finding that the corneal epithelial score is significantly correlated with the Schirmer tear secretion score indicates that epithelial changes in patients with ocular GVHD may be related to lacrimal deficiency.

The relationship between tear cytokines and IVCM corneal epithelial score

Tear cytokines related to GVHD and the corneal epithelia were selected for detection and analysis, including IL-6, IL-10, TNF-a, EGF, MMP-2, MMP-3, MMP-7, BAFF and APRIL. IL-6, IL-10 and TNF-a are all tear cytokines that have been mentioned in the previous literature as having diagnostic ability for chronic GVHD and are correlated with ocular surface parameters.^11–13,28 EGF and MMPs are tear cytokines closely related to corneal epithelium.^29,30 APRIL and BAFF are both important regulators of B cell survival and proliferation.^31,32 There have been few reports concerning changes of APRIL and BAFF in tears of patients with chronic oGVHD.

The results showed that EGF, APRIL and MMP-7 were significantly decreased in ocular cGVHD patients compared to in non-cGVHD patients after HSCT, and EGF (r = −0.491, p < .001) and APRIL (r = −0.318, p = .030) were significantly correlated with the corneal epithelial score.

EGF (epidermal growth factor) is mainly secreted by the lacrimal gland.^33,34 Studies have shown that the lacrimal gland is damaged and tear secretion is reduced in patients with GVHD.³⁵ This could explain why the level of tear EGF in patients with chronic oGVHD decreased significantly compared with patients without oGVHD after HSCT. In addition, EGF has been shown to stimulate the proliferation and migration of corneal epithelial cells, thus playing important roles in promoting corneal epithelial self-renewal and wound healing.^29,36–41 Our study revealed that EGF levels were significantly negatively correlated with the corneal epithelial score. Therefore, we speculate that the significant decrease in EGF levels in tears is related to corneal epithelial defects and morphological changes in patients with chronic oGVHD. The decrease in EGF levels might slow the repair of epithelial cells, resulting in more serious corneal epithelial changes in patients with oGVHD. Exogenous EGF supplementation may be helpful for the treatment of epithelial defects in chronic oGVHD patients.

APRIL (a proliferation-inducing ligand), belonging to the TNF (tumor necrosis factor) family, is an important regulator of B cell survival and proliferation.^31,32,42 It is mainly secreted by immune cells, such as monocytes, neutrophils, dendritic cells, macrophages and T cells.^43–45 Previous studies mostly focused on its role in cancer and autoimmune diseases such as systemic lupus erythematosus and Sjögren’s syndrome.^32,46,47 Several studies have shown that B cells also take part in the pathogenesis of acute and chronic GVHDs.^48,49 One study revealed a strong correlation between acute GVHD and the serum concentration of APRIL, which further suggested that B cells are involved in acute GVHD and that the serum APRIL concentration may be an important predictor of the severity of acute GVHD.⁵⁰ However, there have been few reports concerning changes in APRIL in the tears of patients with chronic GVHD. Recently, our research group discovered for the first time that APRIL levels in tears was lower in patients with chronic oGVHD than DED patients. Besides, APRIL/BAFF shows satisfactory diagnostic capabilities.⁵¹ This study improve that the level of APRIL in tears of ocular cGVHD patients is significantly lower than that in the tears of non-GVHD patients. It also revealed that the corneal epithelial score was significantly correlated with APRIL levels. Thus, we speculate that APRIL is a protective factor that might play a role in inhibiting the local immune response or epithelial repair. Further studies should be performed to explore the role of APRIL and B cells in ocular GVHD.

MMPs (matrix metalloproteinases) are enzymes that degrade the structural components of the ECM (extracellular matrix).⁵² In the cornea, MMPs are involved in epithelial wound healing and interstitial remodeling.^30,53,54 Interestingly, our study revealed that the levels of EGF and MMP-7, tear inflammatory factors that are upregulated in corneal wound healing, were significantly lower in patients with ocular cGVHD than in non-GVHD patients after HSCT, suggesting that patients with cGVHD might have adverse reactions to injury, which might affect the healing of epithelial defects.

In addition, our study revealed that there is no significant difference in specific inflammatory factors, such as IL-6 and IL-10, in ocular cGVHD patients compared with the control group. Based on our overall results, we speculate that the cause of epithelial damage is not an acute specific inflammatory reaction but lacrimal gland damage, which leads to tear deficiency, resulting in nonspecific inflammation caused by ocular surface dryness and poor repair of epithelial defects caused by the decrease in EGF levels in tears. However, neither this nor previous studies could confirm whether the changes in tear inflammatory factors and corneal epithelia were due to the GVH disease process itself or to a consequence of dry eye, which requires further and more rigorous experiments to explore.

This study was limited by the relatively small sample size. Relatively few patients experience GVHD, and even fewer met our inclusion and exclusion criteria. Further studies with a larger sample size are needed to verify our results. Besides, we didn’t test all the inflammatory factors related to chronic oGVHD because of our limited funds and the difficulty in purchasing some antibodies due to the outbreak of COVID-19.

Conclusion

In summary, using the IVCM corneal epithelial score, our study revealed that cGVHD patients have more severe epithelial changes than non-GVHD patients after HSCT. In addition, the corneal epithelial score was correlated with the corneal fluorescence staining score, Schirmer tear secretion score and level of tear cytokines EGF and APRIL. These results suggest that the depth of corneal epithelial defects can be estimated by the corneal fluorescence staining score. In addition, the corneal epithelial changes in ocular cGVHD patients are presumed to be related to the lack of tear and repair factors rather than a specific inflammatory response.

Author contributions

Design of the study (SW Liu, RM Peng, J Hong); conduct of the study (SW Liu, RM Peng); collection and management of data (J Ma, Z Shen, BH Hu, YH Zhao); analysis and interpretation of data (SW Liu, RM Peng); writing of manuscript (SW Liu); and review or approval of manuscript (J Hong).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants but are available from the corresponding author Jing Hong.

Additional information

Funding

This study was approved by the Peking University Third Hospital Medical Science Research Ethics Committee, approval number [M2021526]. The study complied with the Health Insurance Portability and Accountability Act of 1996 and adhered to the tenets of the Declaration of Helsinki. All participants provided written informed consent.