Purpose
The purpose of this research was to evaluate the incidence, risk factors, and complications of ocular graft-versus-host disease (GVHD) in a large single-center study.
Design
Retrospective observational case series.
Methods
This study included 283 patients who underwent hematopoietic stem cell transplantation (HSCT) between 2005 and 2020. Ocular GVHD was diagnosed according to International Chronic Ocular GVHD Consensus Group criteria. Potential risk factors for ocular GVHD were evaluated using the Cox proportional hazards model.
Results
The cumulative incidence of ocular GVHD was 19.7% at 1 year, 29.3% at 2 years, 40.7% at 3 years, 47.2% at 4 years, and 49.7% at 5 years. Ocular GVHD was significantly associated with recipient age (hazard ratio [HR]: 1.228; 95% confidence interval [CI]: 1.033–1.459; P = .020); female sex (HR: 1.797; 95% CI: 1.195–2.703; P = .005); peripheral blood stem cell use (PBSC) (HR: 2.079; 95% CI: 1.268–3.411; P = .004); and previous acute GVHD (HR: 1.276; 95% CI: 1.073–1.518; P = .006). Ocular complications after HSCT included cataract, corneal ulcer, corneal perforation, lacrimal obstruction, herpetic keratitis, and cytomegalovirus retinitis.
Conclusions
Half of patients developed ocular GVHD in the 5 years following HSCT. Older age, female sex, use of PBSC, and acute GVHD disease were significant predictors of ocular GVHD. Hematologists and ophthalmologists should be aware of its vision threating complications.
Hematopoietic stem cell transplantation (HSCT) is the only definitive therapeutic strategy for a large group of hematological, autoimmune, and hereditary disorders. Graft-versus-host disease (GVHD) continues to be a leading cause of morbidity and mortality after allogeneic HSCT, limiting its chances of success. The condition is the result of a highly complex immune process, involving donor T-cell responses to host antigens and the dysregulation of pro-inflammatory cytokines followed by the development of immunity-mediated inflammation and fibrosis of target tissues and organs.
In the past, the distinction of acute versus chronic GVHD was based on the time of its onset. Acute GVHD was defined as disease occurring in the first 100 days after transplantation, whereas GVHD occurring after 100 days was referred to as chronic. However, this arbitrary distinction did not account for the differences in pathogenesis and clinical manifestations of the 2 forms. Thus, the US National Institutes of Health (NIH) consensus development project defined new criteria for the diagnosis, recommending that acute and chronic GVHD should be distinguished based on clinical manifestations. Although acute GVHD is characterized by maculopapular erythematous rash, cholestatic hepatitis, and gastrointestinal symptoms, chronic GVHD is a pleiotropic multiorgan syndrome whose diagnosis requires at least 1 diagnostic manifestation or 1 distinctive manifestation confirmed by biopsy or other testing.
Ocular GVHD is a frequent manifestation of chronic GVHD, occurring in 30%–60% of patients after HSCT. Dry eye disease associated with fibrosis of lacrimal and meibomian glands, superficial punctate keratopathy, and conjunctival scarring represent the hallmark of the disease. In more severe cases, the disease may be complicated by corneal neovascularization, infectious keratitis, and sterile corneal perforation leading to melting and perforation. , Previously reported risk factors for ocular GVHD include acute GVHD, diabetes mellitus, and non-white ethnicity. However, results of the available studies are inconsistent, and the incidence of other complications than dry eye disease remains largely undetermined.
The objective of the present study was to evaluate the incidence, risk factors, and complications of ocular GVHD. For this purpose, a retrospective study was conducted using data involving 283 hematological patients who underwent allogeneic HSCT in a single Italian center.
MATERIALS AND METHODS
This single-center retrospective study included adult patients who received a first allogeneic HSCT between January 2005 and January 2020 at the Hematology Units of the S.Orsola-Malpighi University Hospital (Bologna, Italy) and underwent subsequent ocular surface examinations at the Ophthalmology Unit of the same hospital. Exclusion criteria included survival <100 days after transplantation; presence of other ocular surface disorders or any systemic disease potentially affecting the ocular surface at the time of HSCT; and missing ophthalmological data after HSCT. The study was performed in accordance with the principles of the Declaration of Helsinki and was approved by the local Ethics Committee (Comitato Etico di Area Vasta Emilia Centro della Regione Emilia-Romagna).
The source of stem cells was bone marrow (BM), peripheral blood, or cord blood. All donors were matched related or matched unrelated donors. Human leukocyte antigen (HLA) compatibility was based on the best available typing results at the time of the analysis. The conditioning was myeloablative or reduced-intensity regimen based on patient age, previous treatments, comorbidities, and status of malignancy. All patients underwent GVHD prophylaxis using either cyclosporine and methotrexate or cyclosporine and mycophenolate mofetil. Demographic and hematological data, including age, sex, primary hematological disorder; type of donor and source of hematopoietic stem cells; intensity of conditioning; age and sex of donor; presence of sex; HLA and ABO type mismatch; and cytomegalovirus (CMV) donor positivity were recorded for each patient.
We defined standard-risk diseases as acute myeloid leukemia and acute lymphoblastic leukemia in first or second remission; chronic myeloid leukemia in the first or second chronic phase or in the accelerated phase; myelodysplastic syndrome with refractory anemia or refractory anemia with ringed sideroblasts; and aplastic anemia. All other conditions were defined as high risk. Grading of acute GVHD was performed in on a scale of 0–IV according to the Glucksberg classification system. The severity of chronic GVHD was scored using 2014 NIH criteria, which uses a scale of 0–3 scale for each organ and a global score of mild, moderate, or severe.
Ophthalmic examinations after HSCT were performed at months 3, 6, and 12, and every year thereafter. A subgroup of patients was also examined before HSCT 7 to 9 days before the beginning of the conditioning regimen. Subjective ocular discomfort symptoms were scored by using the Ocular Surface Disease Index (OSDI) validated questionnaire. Subsequently, all patients underwent a comprehensive ocular surface examination including tear film break-up time (TBUT), corneal staining, and the Schirmer test. TBUT was evaluated after administration of 2 µL of 2% fluorescein dye and measurement of the time interval between the last complete blink and the first appearance of a dry spot or disruption in the tear film. Corneal staining was graded using the National Eye Institute score. The Schirmer test was performed without anesthesia, using test strips kept in the temporal lower conjunctival sac for 5 minutes with closed eyes. Dry eye disease before HSCT was ascertained according to Tear Film and Ocular Surface Society Dry Eye Workshop II Criteria.
The diagnosis of ocular GVHD was based on the International Consensus Criteria on Chronic Ocular GVHD Group, which assigns a scoring point of 0–3 to the Schirmer test, corneal fluorescein staining, and OSDI, and a scoring point of 0–2 to conjunctival injection. However, these criteria were introduced only in 2013, and we did not routinely score conjunctival injection before then. Thus, modified criteria were used without conjunctival injection, and the aggregate was reduced by 1 score point, which was required for reaching the diagnosis: in the presence of systemic GVHD, a score ≥5 indicated ocular GVHD; in the absence of systemic GVHD, a score ≥7 indicated ocular GVHD. The number of complications were recorded.
The incidence of ocular GVHD was estimated on the basis of cumulative incidence curves, with death following HSCT as a competing risk. The Cox proportional hazards model was used to evaluate the effect of confounding variables on the likelihood of developing ocular GVHD. The following variables evaluated for association included: patient and donor ages at transplantation; hematological diagnosis; donor type; source of HSCT; presence of donor-recipient sex mismatch and HLA mismatch; ABO mismatch; intensity of conditioning regimen (myeloablative vs. reduced intensity); total body irradiation; use of anti-T-lymphocyte globulin; acute GVHD grades 1–4; and donor CMV immunoglobulin G serostatus. Factors having a P value <0.1 for association with ocular GVHD by univariate testing were added sequentially to a multivariate Cox regression model. Mann-Whitney U test was used to compare the ocular surface parameters in patients with and without systemic GVHD. Continuous variables are reported as mean ± SD, unless otherwise stated. All analyses were conducted using R version 4.0.0 software and RStudio version 1.2.5042 software (R Project, Vienna, Austria).
RESULTS
A total of 283 patients (162 male and 121 female subjects) were included in the study. Mean age at the time of transplant was 45.8 ± 12.2 years (range: 18–72 years). Baseline hematological characteristics are shown in Table 1 . A pretransplantation baseline ophthalmic examination was available in 144 patients (50.9%). In those patients, mean OSDI before HSCT was 10.0 ± 11.3, corneal staining was 1.7 ± 2.3, the Schirmer test value was 20.3 ± 13.0 mm, and the TBUT was 8.6 ± 4.6 s.
| Characteristic | Value |
|---|---|
| Diagnosis | |
| ALL | 50.0 (17.7) |
| AML | 90.0 (31.8) |
| CLL | 7.0 (2.5) |
| CML | 22.0 (7.8) |
| HL | 26.0 (9.2) |
| MDS | 26.0 (9.2) |
| MM | 20.0 (7.1) |
| NHL | 26.0 (9.2) |
| Other | 16.0 (5.7) |
| Disease risk | |
| Standard risk | 157.0 (55.5) |
| High risk | 126.0 (44.5) |
| Source of stem cells | |
| Bone marrow | 87.0 (30.7) |
| Peripheral blood | 182.0 (64.2) |
| Cord blood | 14.0 (4.9) |
| Type of donor | |
| Matched unrelated donor | 205.0 (72.4) |
| Matched related donor | 78.0 (27.6) |
| Sex match | |
| Male-to-male (match) | 117.0 (41.3) |
| Female-to-female (match) | 38.0 (13.4) |
| Female-to-male | 45.0 (15.9) |
| Male-to-female | 83.0 (29.3) |
| HLA compatibility | |
| Matched | 167.0 (59.0) |
| Mismatched (≥1 antigen) | 116.0 (41.0) |
| ABO compatibility | |
| Matched | 117.0 (41.3) |
| Mismatched | 166.0 (58.7) |
| Type of conditioning | |
| Myeloablative | 200.0 (70.7) |
| Reduced intensity | 83.0 (29.3) |
| Total body irradiation | 28.0 (9.9) |
| Anti T-lymphocyte globulin | 224.0 (79.2) |
| Donor CMV positivity | 134.0 (47.3) |
Following HSCT, 101 patients (35.7%) developed acute GVHD, and 67 patients (23.7%) developed grades II–IV acute GVHD. Among the patients with acute GVHD, 16 of 101 (15.1%) had documented conjunctival involvement in the form of hyperemia, chemosis, or discharge. Conversely, 96 patients (33.9%) developed extraocular chronic GVHD involving the skin in 85 of them (30.0%); in the mouth in 66 (23.3%); the liver in 26 (9.2%); the lung in 37 (13.1%); the joints in 20 (7.1%); the gastrointestinal tract in 15 (5.3%); and the genitalia in 16 (5.7%). Five years following HSCT, mean OSDI was 25.6 ± 20.6, corneal staining was 2.4 ± 3.1, the Schirmer test value was 17.3 ± 12.7 mm, and TBUT was 6.5 ± 4.5 s. As shown in Figure 1 , all ocular surface parameters turned out to be significantly worse in patients with systemic GVHD (all P < .05).
Figure 2 shows the cumulative incidence curve of ocular GVHD. The cumulative incidence of ocular GVHD was 19.7% at 1 year, 29.3% at 2 years, 40.7% at 3 years, 47.2% at 4 years, and 49.7% at 5 years. The median time from HSCT to the onset of ocular GVHD was 397 days. Among the patients with ocular GVHD, 77.1% had ocular GVHD in the context of systemic GVHD, whereas 22.9% had isolated ocular GVHD.
The factors examined for an association with ocular GVHD by univariate analysis are shown in Table 2 . Recipient age and female sex were identified as significant risk factors for ocular GVHD (respectively, P = .032, P = .024). The use of peripheral blood stem cells (PBSC) was associated with an increased hazard of ocular GVHD ( P = .006). Acute GVHD (grades I–IV) was a significant predictor of ocular GVHD ( P < .001). The other factors showed no significant association with the hazard of ocular GVHD (always, P >.05).
